A Theory of Global Biodiversity PDF
A Theory of Global Biodiversity PDF
A Theory of Global Biodiversity PDF
Global Biodiversity
MONOGRAPHS IN POPULATION BIOLOGY
SIMON A. LEVIN AND HENRY S. HORN, SERIES EDITORS
bo ris wo rm
and
derek p. tittens or
press.princeton.edu
10 9 8 7 6 5 4 3 2 1
This work is dedicated to Sylvie Moe, and all the other children
who will inherit the fragile beauty that graces our planet.
Contents
Acknowledgments ix
1. Introduction 1
1.1. Integrating Land and Sea 2
1.2. A Brief History of Biodiversity Research 5
1.3. Goals and Structure of This Book 8
7. Conclusions 171
7.1. Summary of Major Findings 171
7.2. Ecological Theory 174
7.3. A Niche for Neutrality? 176
7.4. Spatial Scale 177
7.5. Ecological versus Evolutionary Time 178
7.6. Applications 179
7.7. Limitations 180
7.8. Final Outlook 182
References 185
Index 207
Acknowledgments
indebted to our partners, Heike Lotze and Andrea Moe, who were incredibly sup-
portive over the period of writing this volume. Finally, we are grateful to the Tri-
dent coffee shop and The Henry House, Halifax, for providing inspiring writing
environments and appropriate hydration.
A Theory of
Global Biodiversity
CHAPTER ONE
Introduction
0.00000900 Diversity
0.00000800 Plant
0.00000700
Frequency (%)
0.00000600
0.00000500
0.00000400
0.00000300 Fish
Bird
0.00000200 Insect
Tree
0.00000100 Invertebrate
0.00000000 Coral
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Year published
Figure 1.1. Mentioning of species diversity in published books. Plant diversity is referenced
for around three times longer than and at least eight times as often as any other group. Note the
increasing mention of other species groups only after around 1970, when spatial biodiversity
science further expanded to other realms. Data from Google ngram project (https://books.google
.com/ngrams).
long separated the study of biodiversity on land and in the sea. Since humans are
land animals, we have often been primarily concerned with the terrestrial organ-
isms that we can more easily observe. This is also reflected in the history of eco-
logical study. Many scientists who shaped the foundations of our field, such as
von Humboldt, Darwin, Wallace, Elton, Hutchinson, and MacArthur, had a ter-
restrial or sometimes freshwater focus, but they rarely considered the oceans in
detail (Darwin’s beautiful monographs on corals and barnacles being among the
notable exceptions). These pioneers also shared a keen interest in the forces that
shape patterns of species richness. Hence, published work on this topic is heavily
biased toward the land. A quick online search confirms that there are about 30%
more scientific articles and books devoted to the biodiversity of a single terrestrial
habitat type—forests—alone than to marine biodiversity. Likewise, the mention of
“plant diversity” in all published books, scientific or otherwise, goes back at least
to the beginning of the twentieth century, whereas marine species groups appeared
much later in this context and have only slowly risen to more prominence (fig. 1.1).
In focusing on the oceans in addition to the land in this book, we treat different
marine habitats as further “replicates” of global biodiversity patterns: different
from terrestrial, but potentially organized (at least at some fundamental level) by
similar principles (fig. 1.2). At the very least, this assertion is a null-hypothesis to
be confronted with data. At the very best, it increases the “degrees of freedom”
when we test our ability to understand and predict the first-order patterns of bio-
diversity on our planet. As the marine environment is less familiar to most, it
tends to challenge our assumptions about how the world works, and allows us to
critically examine ecological concepts that have been developed with a largely
terrestrial focus. Are these patterns, ideas, hypotheses, and theories truly general
4 CHAPTER 1
A B
C D
Figure 1.2. Biodiversity across marine and terrestrial habitats provides four largely independent
“replicates” with which to explore patterns of biodiversity. Examples of (A) coastal; (B) pelagic;
(C) deep sea; and (D) terrestrial diversity are shown. Credits: (A) Ami Gur; (B) Samuel Blum;
(C) NOAA; (D) Frank Brodrecht.
to all systems? It is our assertion that ecology, and its theoretical foundations,
should be general, no matter whether its subjects happen to be wet or dry.
At the same time, our—and indeed society’s—current interest in biodiversity
is much more than academic. The rapid erosion of biodiversity—both on land
and in the ocean—is cause for much concern, as it threatens individual species
with extinction (Dirzo et al. 2014; McCauley et al. 2015), and ecosystems with
loss of functionality, essential services, and resilience (Worm et al. 2006; Hooper
et al. 2012). There is growing awareness of this biodiversity crisis as a defining
problem of our time (Tickell 1997), with international policies being shaped in an
attempt to slow and ultimately reverse the rate of loss (Tittensor et al. 2014). This
has also resulted in renewed interest in the fundamental processes that give rise to,
maintain, or threaten, biodiversity at local, regional, and global scales.
Some of these fundamental processes can be further unveiled by contrasting and
comparing current patterns of and changes in diversity to those that have unfolded
INTRODUCTION 5
through deep time (Jablonski et al. 2006; Valentine and Jablonski 2015; Yasuhara et
al. 2015). By carefully dissecting the paleontological record, researchers have been
able to trace patterns of species richness throughout Earth’s history, learning how
they have changed as the environment around them became altered by geological,
biological, and climatic forces (Renema et al. 2008; Krug et al. 2009; Yasuhara et
al. 2012; Jablonski et al. 2013; Mannion et al. 2014). At present, human beings are
exerting pressures and inducing changes as pronounced as many of these deep-time
processes, and are certainly a dominant force shaping life on Earth (Waters et al.
2016). As an example, CO2 is now emitted at a rate ~20 times greater than the fastest
CO2 emission rates in recorded geological history (Zachos et al. 2008), and is warm-
ing the planet at record speed. Yet, our ability to predict the future of biodiversity is
both more limited and more uncertain than our understanding of its past (Sala et al.
2000). We posit that a more general understanding of global biodiversity patterns,
and in particular the ecological and evolutionary forces that shape them, may help
us to forecast biological reconfigurations in both the short and the long term under
a given rate and pattern of environmental change.
been anticipated, and that much of it was concentrated in tropical forests on land,
or contained in structured habitats such as kelp forests and coral reefs in the
ocean. In the marine realm, however, the pace of discovery was generally slower,
particularly in the vast reaches of the open ocean and the deep sea, the latter long
presumed to be devoid of life.
Darwin’s theory of evolution (Darwin 1859) provided another milestone. The
theory was elegant in that it required only two fundamental processes—specifi-
cally, the generation of variation in species’ traits and the forces of natural selec-
tion acting upon that variation—to explain the emergence of new species, and by
extension all of biodiversity. Although the Modern Synthesis suggests that other
processes, such as drift, are also important, the ramifications of Darwin’s theory
as a structural foundation for anchoring an understanding of the spatial distribu-
tion of biodiversity are obvious. Yet, there was little theory available or developed
at the time to understand the processes shaping the distribution of those species
across regions and around the globe.
Modern biodiversity science was arguably born out of Gene Evelyn Hutchin-
son’s “Homage to Santa Rosalia” (Hutchinson 1959), in which he posed the
question of how the great variety of observed species could coexist in a given
environment while competing for a few limiting resources. Examples include the
astounding richness of plankton species that can coexist in a single drop of water,
or the large number of tree species found in a patch of tropical forest. Hutchin-
son’s student Howard Sanders took this question to the deep sea, where he and
Robert Hessler discovered surprising levels of macrofaunal diversity that were
at the time perceived to perhaps match those of hyperdiverse tropical forests and
coral reefs (Hessler and Sanders 1967). Soon Sanders, among others, was for-
mulating theory to explain these observations (Sanders 1968). The stability-time
hypothesis, which was borne out of these observations, was one of the first theo-
ries specifically developed to explain observed biodiversity patterns; it related
large-scale differences in species richness to the severity and frequency of dis-
turbances. This work partially inspired several decades’ worth of research into
marine biodiversity patterns, both from biologists, focusing on coral reef and
deep-sea macrofauna (McCoy and Heck 1975; Grassle and Maciolek 1992; Rex
et al. 1993; Roberts et al. 2002; Brandt et al. 2007), and geologists, focusing on
bivalves, foraminifera, and other microfossils (Ruddiman 1969; Stehli et al. 1969;
Rutherford et al. 1999; Valentine and Jablonski 2015).
In a parallel development, MacArthur and Wilson published their Theory of
Island Biogeography, arguably the first mathematical theory of biodiversity pat-
terns (MacArthur and Wilson 1967). While originally focusing on the immigration
and extinction of terrestrial species on oceanic islands, it was subsequently applied
to other isolated habitats such as lakes, mountaintops, and forest fragments. A
INTRODUCTION 7
later extension of this theory, The Unified Neutral Theory of Biodiversity and Bio-
geography (Hubbell 2001), which built on MacArthur and Wilson’s work, was
heavily informed by observations in tropical forests. When its predictions were
tested in other habitats both marine (Dornelas et al. 2006) and terrestrial (McGill
et al. 2006), however, results were mixed, and its generality remained unclear. The
Metabolic Theory of Ecology is another, more recent theory that makes general
predictions about changes in community structure and consequently biodiversity
in relation to temperature (Brown et al. 2004). On its own, it also falls short of
capturing a majority of diversity gradients on land (Hawkins et al. 2007), but it has
not been evaluated in comprehensive detail for marine environments.
The new millennium brought about an era of large-scale data integration and
synthesis, spurred by the growth of macroecology as a prominent subfield in
ecology (Brown 1995; Gaston 2000). The First Global Census of Marine Life
(2000–2010), for example, built on Fred Grassle’s and Jesse Ausubel’s bold
vision to systematically chart and understand marine biodiversity patterns across
all the many different habitats, from shallow reefs to abyssal plains (Ausubel
1999; Grassle and Stocks 1999). Grassle was a student of Sanders, and thus it is
conceivable that the vision of a systematic and global marine census may ulti-
mately be traced back through a direct academic lineage to Hutchinson’s seminal
thoughts on patterns of species diversity. The Census of Marine Life was by far
the largest initiative of its kind and certainly invigorated interest in marine bio-
diversity, both in academia and in the public eye through regular media coverage
of its many spectacular and photogenic discoveries. Scientifically, it fostered an
interdisciplinary, comparative, and highly collaborative approach that culminated
in a series of synthetic papers on the large-scale patterns of species distribution,
abundance, and richness across different marine habitats (Tittensor et al. 2010;
Block et al. 2011; Mora et al. 2011; Ramirez-Llodra et al. 2011). At the same
time, parallel synthetic developments in terrestrial ecology led to similar efforts at
empirical synthesis, focusing largely on patterns in plant (Kier et al. 2005; Kreft
and Jetz 2007) and vertebrate species richness (Jetz and Rahbek 2002; Grenyer et
al. 2006; Jetz and Fine 2012). Finally, the creation of dedicated institutions such
as the National Center of Ecological Analysis of Synthesis (NCEAS) spurred the
search for general ecological principles to be extracted from these newly docu-
mented global patterns. Most recently, online databases such as the Encyclopedia
of Life (http://eol.org/); the Map of Life (https://www.mol.org/); the International
Union for Conservation of Nature (IUCN) Red List (http://www.iucnredlist.org/);
the Global Biodiversity Information Facility (http://www.gbif.org/); and the
Ocean Biogeographic Information System (http://www.iobis.org/) have enabled
even better integration of biological and environmental data, facilitating further
syntheses of observed patterns of species richness and community structure.
8 CHAPTER 1
Despite these exciting advances, however, biodiversity research has not yet
produced a testable body of theory that can coherently explain the manifold pat-
terns of global biodiversity on land and in the sea. Yet, the time seems ripe to
attempt such a synthesis. On the empirical side, field ecologists have compiled an
unprecedented number of observations across multiple taxonomic groups along
with associated environmental predictors that might be related to underlying pro-
cesses. There also exists an extensive list of hypothesized drivers and mecha-
nisms (Rohde 1992), but no cohesive model to reconstruct observed patterns of
biodiversity from such mechanisms at the global scale. Especially challenging is
the fact that some of the hypothesized drivers, such as temperature, can operate
through both ecological (for example, niche constraints) and evolutionary (for
example, speciation rates) mechanisms, and correlative models provide no way to
separate these effects. Another obstacle is that much relevant theory (for example,
island biogeography theory, metabolic theory, or indeed neutral theory) has been
developed with a terrestrial focus, and its applicability to marine habitats has been
less extensively explored in terms of spatial distributions of biodiversity. Integrat-
ing empirical observations with mechanistic models both on land and in the ocean
may provide a clearer picture of both their generality and their limitations.
theory. From this emerge a number of predictions on species richness, but also
range size, niche breadth, and other macroecological patterns that are explored
in some detail. When confronting our model with empirical data (chapter 5), we
derive surprisingly realistic predictions of species richness for known marine and
terrestrial species groups on a global grid that includes both the continents and
the oceans. On the more applied side (chapter 6), we use our empirical synthesis
to map out possible priority areas for conserving biodiversity at large scales, and
employ our theoretical model to explore scenarios of long-term changes in future
biodiversity patterns in response to projected global warming. In the long term, it
is our hope that our thoughts will in some way contribute to a broader, unifying
synthesis in biodiversity science (chapter 7).
The ideas presented in this book are necessarily circumscribed by inherent biases
in our collective knowledge of different species groups. We tend to be familiar with
larger-bodied taxa that are easily observable, or that we find beautiful, nutritious,
or fearsome. Much less information is available, particularly when seeking global
coverage, for those taxa that are smaller, less visually interesting, or less obviously
useful for humankind. Microbial organisms, for example, are critically important in
regulating biogeochemical processes at the planetary scale (Falkowski 2012), yet
our knowledge of their biodiversity remains very limited (Sunagawa et al. 2015).
Less well researched taxa may display different spatial patterns of biodiversity
from those with which we are more familiar. We are unable, unfortunately, to over-
come this limitation, and can only work with the empirical information that we
have available at this point in time; future discoveries and empirical compilations
can be used to test, revise, or overturn the ideas in this book. In addition, our
work maintains a consistent focus on species richness, and treats species evenness,
functional diversity, species turnover, and community composition only in passing.
This is most certainly not for a lack of interest but due to the inherent limitations
involved in covering a broad range of taxa and patterns in a single book.
We also focus deliberately on large scales, for the pragmatic reason that our
knowledge at large spatial grains (hundreds to thousands of km) tends to be more
complete, on average, than at finer grains (Mora et al. 2008). The idea is to capture
the first-order latitudinal and longitudinal patterns of species richness, its peaks
and troughs, rather than its detailed regional variation, which might be structured
by different driving forces. We realize that this approach may disappoint some,
who would like to see a more detailed representation of localized marine and
terrestrial biodiversity patterns, as well as a detailed explanation of smaller scale
variation, but this would necessitate a very different focus for this volume. We
fully acknowledge that we present only generalities and that potentially important
exceptions and scale-related differences exist, and we direct the reader toward
more regional studies wherever possible.
CHAPTER TWO
Throughout this chapter, we also retain our focus on the simplest measure of
biodiversity—namely, species richness. Other aspects such as species turnover,
functional diversity, evenness, and genetic diversity are acknowledged as clearly
important, but are not sufficiently well known at the scales that we examine to
enable a major synthesis across taxonomic groups on land and in the sea. We sepa-
rate coastal, pelagic, deep-sea, and terrestrial richness patterns, since these major
realms are sufficiently discrete as to necessitate independent treatment (Gaston
2000; Tittensor et al. 2010; Woolley et al. 2016), and can provide informative
contrasts, shedding insight on generalities and differences. Within these broad
habitats, we separate by major taxonomic groups, discussing and comparing pat-
terns for plants, invertebrates, and vertebrates.
Coastal habitats, here defined as the shallow waters stretching from the intertidal
zone to the edge of the continental shelf (0–200 m depth), represent the most
accessible marine environments for humans and hence the best known. As a tran-
sition zone between land and sea, they are strongly influenced by a combination
of terrestrial and marine processes. Land runoff and rivers increase the availabil-
ity of nutrients, and on average result in a much higher (10- to 100-fold) net pri-
mary productivity (NPP) compared with most open-ocean environments. Coastal
NPP frequently ranges in the order of 500 to 2500 g C m−2 a−1, which is similar to
productive grasslands and forests (Lieth and Whittaker 2012). Also similar to the
land, we find high habitat diversity and structure here, often created or enhanced
by foundation species such as corals, macroalgae, seagrasses, or mangroves.
Many coastal environments show sharp physical gradients both vertically and
horizontally in temperature, salinity, productivity, and depth that enhance the
heterogeneity provided by biogenic habitats. These sharp environmental gradi-
ents and complex habitats can affect both ecological and evolutionary processes
that in turn structure biodiversity, resulting in complex local gradients and het-
erogeneity. Aside from a dynamic physical environment, coastal habitats also
experience significant impacts from a growing human population that has chosen
to settle preferentially near the world’s seashores. The effects of exploitation,
pollution, and habitat conversion are all concentrated in coastal environments,
particularly in the northern hemisphere (Halpern et al. 2008), and over time have
profoundly altered patterns of biological diversity, habitat structure, and produc-
tivity (Lotze et al. 2006). In summary, the most outstanding features of coastal
environments when compared to the pelagic ocean and the deep sea, and akin to
those on land, are complex habitat structure, high nutrient input and productivity,
O B S E RV E D PAT T E R N S 13
and growing human impacts, all of which influence the distribution of coastal
biodiversity.
Large-scale patterns of coastal biodiversity were first described for bivalves (Mol-
lusca; Class: Bivalvia), initially at a regional scale (Thorson 1952, 1957), and then
globally in the foundational studies by Stehli and Sanders (Stehli et al. 1967; San-
ders 1968). Bivalves are particularly useful specimens, as they are found in all
coastal habitats, are easily sampled, and are reasonably taxonomically tractable
(~10,000 known extant species). Moreover, this group covers much of the geo-
logical history of metazoans, having emerged during the Cambrian explosion of
animal diversity about 540 Ma ago. Bivalves fossilize well, which sheds light on
their evolutionary history, and helps to interpret the emergence of biogeographical
patterns over time. It is worthwhile to read these classic studies closely, as they
reveal luminous insights into some of the first-order patterns, and likely drivers,
of marine biodiversity. By fitting diversity contours around samples of bivalve
diversity from just 36 locations (fig. 2.1A), Stehli and coauthors first identified
two prominent centers of diversity (later called hotspots): one in the tropical Indo-
Pacific centered on the Indonesian archipelago, and a secondary one in the tropical
Eastern Pacific off the northern coast of Central America (Stehli et al. 1967). These
species-rich features were almost perfectly centered on the equator, and were also
seen at the genus and family level. Other primary features of the overall patterns
were (1) steep latitudinal gradients with maximum diversity in the tropics, (2) less
steep longitudinal gradients away from the Indonesian and Central American
hotspots, and (3) low diversity in the Arctic and Southern Oceans (fig. 2.1).
Progressive updates and the inclusion of new data have largely confirmed these
earlier results (Crame 2000; Valentine and Jablonski 2015), with the exception
that richness appears to peak slightly north of the equator, but still in tropical
waters (fig. 2.1B). These results prefigure the canonical pattern of biodiversity
later identified for a variety of other coastal marine taxa (Tittensor et al. 2010).
This consistency may hint at some generalities in spatial biodiversity patterns
that can be readily gleaned from incomplete and limited data at a coarse spatial
grain (see fig. 2.1). Another interesting observation from this early work relates
to the fact that steeper latitudinal gradients were seen in more recently evolved
bivalve taxa; this may suggest a tropical origination for most bivalves and a long
dispersal period (106 to 108 years) by which newly formed species groups, or
clades, reached the poles via adaptive radiation (Crame 2000; Jablonski et al.
2006; Jablonski et al. 2013).
14 CHAPTER 2
A
48
54
50 46
36
38 76
57
100
93
90
200
158
158 180
100 206 238
300
176 135
400
131 200 415
357 152
157 500
213 451
259 1037
388
502
500 254
400 524
500
319
300 338 400 471
328
300 279
178
200
86
100
1579
1189
798
408
18
Figure 2.1. Diversity of recent bivalves. The first global map of marine biodiversity was pub-
lished in 1967 for bivalves ((A) sampled richness of marine bivalves, after data in Stehli et al.
1967). Based on a limited number of samples from 36 locales, Stehli and colleagues synthesized
a first-order global diversity pattern in coastal and continental shelf ecosystems. Remarkably,
the general patterns have not changed significantly, although data availability has improved con-
siderably since then ((B) total richness of marine bivalves, after data in Valentine and Jablonski
2015). Likewise, the pattern appears consistent when changing taxonomic resolution; near iden-
tical gradients were seen at the genus and family level (Stehli et al. 1967). Finally, it has emerged
that very similar patterns are seen across most known coastal species groups, implying marked
generality within and across taxa (figs. 2.2–2.4).
O B S E RV E D PAT T E R N S 15
Apart from their groundbreaking work on bivalve species richness, Stehli and
colleagues also analyzed the richness of reef-building corals (Stehli and Wells
1971), which form the most species-rich habitats in the coastal realm. Coral reefs
are biogenic coastal habitats created by stony corals (Cnidaria; Order Scleractinia)
that first evolved in the Triassic, about 240 Ma ago. Coral reefs are often likened
to rainforests due to their astounding local diversity, although their spatial extent
is much smaller, at about 260,000 to 600,000 km2 worldwide, approximately 5%
of total rainforest area, and only 0.2% of ocean surface area. Yet these reefs have
among the highest diversity per unit area of any known habitat, much of it remain-
ing undescribed (Bouchet et al. 2002; Bouchet 2006). As in rainforests, the largest
number of species is not found among the structural species themselves (reef-
building corals contain ~800 known species) but in the diverse fauna that inhabits
these complex habitats. Birds and insects in a rainforest find an equivalent in the
fishes and invertebrates on a coral reef.
In a similar manner to the bivalves, the diversity pattern for reef-building corals
is centered on a global biodiversity hotspot in the tropical Indo-Pacific (fig. 2.2).
Species richness declines with increasing distance from this center, the so-called
coral triangle between the Philippines, Indonesia, and Australia that supports most
known species of reef-building corals. A secondary hotspot in the Caribbean har-
bors only ~50 or so species of hard corals. Even in the earliest works on the topic,
it was already noted that the coral triangle and Caribbean hotspots appear to have
emerged independently of one another, as faunal overlap is very small (Stehli and
Wells 1971). Moreover, the authors remarked that both hotspots were associated
with the warmest sea surface temperature (SST) anomalies in the Indo-Pacific and
Atlantic, respectively; in both instances, these occur in the western part of the
basin, and somewhat north of the equator (Stehli and Wells 1971). The authors
argued that water temperature may play a leading role in enabling high richness, as
opposed to solar radiation, which peaks at the equator. The role of habitat area was
also discussed, as both regions harbor large archipelagos with long coastlines, and
complex spatial structure, but the relative importance of this was not resolved ana-
lytically. Furthermore, the authors noted that the age of surveyed genera increased
with increasing distance from the centers, and young genera were predominantly
found in both the Caribbean and western Pacific centers of diversity. This supported
the hypothesis that these centers of high diversity were also centers of evolutionary
innovation. Interestingly, the mean age of coral genera increases much faster with
latitude than with longitude (Stehli and Wells 1971), implying slower evolutionary
speeds at higher latitudes (and at lower temperatures), and maybe a slower spread
of newly evolved genera across latitude when compared to longitudinal expansion.
We will return to these environmental drivers in chapter 3.
Corals
584
438
Latitude
292
147
1
0 162 324
Mean richness
Shelf ophiuroids
137
110
Latitude
84
57
30
47 80 114
Mean richness
Cone snails
205
154
Latitude
103
52
1
0 36 72
Mean richness
Non−squid cephalopds
35
26
Latitude
18
10
1
2 7 12
Mean richness
O B S E RV E D PAT T E R N S 17
0.70
Latitude
0.47
0.24
Figure 2.2. Coastal invertebrates. Shown are global species richness patterns and corresponding
latitudinal gradients for corals, ophiuroids (brittle stars), cone snails, and coastal cephalopods in
coastal and continental shelf waters <200 m depth. Data from Tittensor et al. (2010); IUCN (2016);
Woolley et al. (2016). Note that the cephalopod pattern includes only commercial species. The
bottom panel displays mean normalized richness, in which richness values for each taxon were
scaled from zero to one and then averaged to standardize patterns irrespective of differences in
total richness between taxa; normalized richness also includes the bivalves data from fig. 2.1B.
Mangroves
17
13
Latitude
9
1
0 4 7
Mean richness
Seagrasses
18
Latitude
14
10
1
0 4 8
Mean richness
Mean normalized richness
0.89
0.67
Latitude
0.46
0.24
0.03
0.0 0.2 0.4
Mean richness
Figure 2.3. Coastal plants. Shown are the global species richness patterns and correspond-
ing latitudinal gradients for the two groups of flowering plants that colonize shallow marine
waters—namely, seagrasses and mangroves. Mean normalized richness pattern averaged across
these groups is shown at the bottom. After data from Tittensor et al. (2010).
corals are sometimes used consecutively by the same reef fish species as larval,
juvenile, and adult habitat, respectively (Mumby et al. 2004). Yet this potential
ecological relationship and successional sequence does not explain why spatial
patterns of species richness (as opposed to presence or biomass) are so strongly
overlapping, and again “diversity begets diversity” cannot be the whole expla-
nation. Possibly these species groups may relate in a similar way to the envi-
ronmental factors that shape the habitat in which they co-occur. These include
the extent and complexity of coastal habitats found in the Indonesian-Australian
20 CHAPTER 2
and Caribbean archipelagos and the fact that these regions experience the highest
average sea surface temperatures in their respective ocean basins. These factors
will be analyzed and discussed in detail in chapter 3.
In temperate coastal environments, most shallow corals and mangroves can-
not survive due to their limited tolerance of cold water (McCoy and Heck 1975;
Hutchings and Saenger 1987), though see Cairns (2007) for a discussion of cold-
water corals. Instead, macroalgae (on rocky shores) and seagrasses (on soft sedi-
ments) provide shallow temperate biogenic habitats. Reflecting this pattern of
distribution, seagrass species richness is skewed more poleward when compared
to mangroves (see fig. 2.3) or corals (see fig. 2.2). In contrast to seagrasses and
mangroves, which are less speciose, we do not presently have a complete pic-
ture of macroalgae species richness patterns on a global scale; this is partly due
to higher richness, problematic species identification, and fluid taxonomy at the
species level. Genus richness, however, has been assessed, and peaks at progres-
sively higher latitudes for the three main clades: green algae tend to center in
the tropics (Archaeplastida; Division: Chlorophyta), red algae (Archaeplastida;
Division: Rhodophyta) at intermediate latitudes, and brown algae (Ochrophyta;
Class: Phaeophyceae) at temperate latitudes, such as off South Australia, Europe,
and Japan (Kerswell 2006; Keith et al. 2014). At the species level, the Bryopsi-
dales, a speciose order of mostly turf-forming green algae, has been analyzed in
more detail; like other Chlorophyta, this group shows highest species diversity
in the tropics, and specifically in the tropical Indo-Pacific. A secondary hotspot
is found in the Caribbean, but with much lower diversity in the Atlantic overall
(Kerswell 2006). As such, the distribution of species richness does closely follow
that of corals and mangroves. The reason might be that Bryopsidales tend to be
largely reef-associated, whereas many other macroalgae thrive on rocky surfaces
not occupied by reef-forming corals, and hence tend to be distributed in a more
poleward manner (Kerswell 2006; Keith et al. 2014). In summary, macroalgae
seem to show a variety of richness patterns, ranging from tropical distributions
similar to corals, to temperate distributions similar to seagrasses (see fig. 2.3).
Expanding the scope from largely sessile invertebrates and plants to more mobile
vertebrates (Chordata; Subphylum: Vertebrata) provokes the question of whether
observed diversity patterns may blur or weaken due to increased capacity for
migration and dispersal and the lack of hard boundaries in a continuous fluid
marine habitat. Vertebrates have evolved as mobile consumers, with bony fishes
representing the most speciose taxon, containing half of all known vertebrate
O B S E RV E D PAT T E R N S 21
species. The modern bony fishes (Class: Osteichthyes) probably evolved in the
late Silurian, about 416 Ma ago. They inhabit surface to abyssal depths and open-
ocean, coastal and estuarine, as well as freshwater habitats. About 16,000 to
17,000 species have been described in the oceans (Mora et al. 2008; Eschmeyer
et al. 2010). According to an analysis of species discovery curves, this comprises
an estimated 79% of total marine fish richness (Mora et al. 2008), suggesting
that fish diversity is much more completely known than most other marine taxa
(Mora et al. 2011).
Somewhat surprisingly, given their contrasting dispersal ability and evolution-
ary history, coastal bony fishes again yield a similar diversity pattern as corals,
mangroves, and seagrasses, with a pronounced diversity peak in the tropical Indo-
Pacific, although with small bimodal peaks north and south of the equator (fig.
2.4). However, latitudinal and longitudinal gradients of species richness appear
less steep in fish, possibly relating to enhanced dispersal, and to the fact that
coastal fishes cover a wider latitudinal range than either corals or mangroves.
Secondary hotspots for coastal fish diversity occur in the western Indian Ocean,
the Caribbean, and the eastern tropical Pacific (see fig. 2.4). These general pat-
terns are seen across a large number of regional to global studies looking at fish
diversity (see review by MacPherson et al. 2009). Note that the Eastern Tropical
Pacific and Caribbean hotspots have a shared evolutionary history (prior to the
closure of the Isthmus of Panama 3 Ma ago), and still show large overlaps in spe-
cies composition; about 35% of Eastern Pacific fish genera are shared with, and
only with, the Caribbean (Rosenblatt 1967).
There is an interesting contrast, however, between coastal bony fishes and
coastal sharks (Class: Chondrichthyes; Superorder: Selachimorpha; see fig. 2.4),
which evolved in the Silurian (440 Ma ago). The majority of the 508 or so known
shark species are associated with coastal habitats at least for part of their life
cycle (Lucifora et al. 2011), but these coastal sharks reach their highest diversity
at somewhat higher latitudes than coastal teleost fishes: distinct peaks of shark
species richness are found at around 20 degrees latitude North or South in the
Western Pacific and Eastern Indian Ocean (see fig. 2.4). As such, their longitudi-
nal diversity pattern is similar to other coastal species, but the latitudinal pattern
is spread out farther toward the subtropics. This might be related either to the
larger average size of sharks compared with bony fishes, which may translate
into enhanced dispersal and larger ranges, or the longer evolutionary age, which
allows more time for dispersal and adaptation to cooler climates (MacPherson et
al. 2009). Indeed, several species of lamnid sharks, such as white (Carcharodon
carcharias), porbeagle (Lamna nasus), and salmon sharks (Lamna ditropis), have
evolved partial endothermy, which allows for a more poleward distribution, at
least for part of the year (Block et al. 2011).
Coastal fishes
2555
1919
Latitude
1284
648
13
35 453 869
Mean richness
Non−oceanic sharks
104
78
Latitude
52
27
1
0 16 31
Mean richness
Pinnipeds
10
8
Latitude
1
1 3 6
Mean richness
Mean normalized richness
0.67
0.50
Latitude
0.33
0.17
Figure 2.4. Coastal vertebrates. Shown are the global richness patterns and corresponding lati-
tudinal gradients for coastal bony fish, sharks, and pinnipeds (seals, sea lions, and walrus). Mean
normalized richness pattern averaged across these groups is shown at the bottom. After data
from Tittensor et al. (2010).
O B S E RV E D PAT T E R N S 23
2.1.4. Synthesis
In summary, the coastal realm is, in terms of marine environments, the closest to
us spatially, and also offers the closest comparison to the land, as habitat structure
and nutrient supply are more similar to terrestrial analogues than other marine
environments. Empirically, we observe substantial overlap in the distribution of
diversity among different species groups, including invertebrates, plants, and ver-
tebrates. A majority of these groups show maximum diversity in the Indonesian-
Australian (Indo-Pacific) and Caribbean (Atlantic) archipelagos, respectively.
Latitudinal peaks in richness are generally tropical (though not necessarily equa-
torial) or subtropical, except for cold-water adapted subgroups. These generaliza-
tions are supported when averaging normalized diversity across groups (see figs.
2.2 to 2.4). For all groups examined here, total richness in the Atlantic is lower on
average than in the Indo-Pacific, possibly reflecting the younger geological age of
the Atlantic, which formed only 130 Ma ago. The large cluster of species richness
24 CHAPTER 2
in the central Indo-Pacific is striking and has no equivalent in the western Indian
Ocean or eastern Pacific, for example. This pronounced longitudinal gradient is
also markedly different from pelagic, deep-sea, and terrestrial taxa, which are
discussed in the next sections.
The pelagic realm, here defined as the open waters that extend beyond the conti-
nental shelf, is a defining feature of our planet, yet few people ever get the chance
to truly explore it. Pelagic waters encompass about two-thirds of the planet’s sur-
face and are characterized by great depth ranges (up to ~11,000 m). As such, this
represents by far the largest volume of any habitat on Earth, yet our knowledge
lags far behind what we know about biodiversity in coastal waters. In the pelagic
ocean, most data pertain to the euphotic zone—that is, the top 200 m layer of the
water column that receives sufficient sunlight for photosynthesis. Deeper pelagic
waters, particularly those beyond 2000 m depth, are much less well studied, and
are poorly understood in terms of their biodiversity. Even the largest known deep-
water pelagics, such as giant squid (Architeuthis sp.) have only recently been
documented for the first time in their native habitat (Roper and Shea 2013).
For this reason, we largely focus on the euphotic (sunlit) pelagic zone here (top
200 m), with some discussion of mesopelagic taxa (200–2000 m). Deep-water
taxa (>2000 m depth) will be discussed in the next section.
The pelagic environment experiences relatively little direct influence from the
land and the sea floor, and consequently receives low input of macronutrients. The
availability of limiting nutrients—particularly, nitrogen, phosphorus, and iron—
is generally dependent on the physical structure of the water column and can be
enhanced by local upwelling of nutrient-rich deep water—for example, in the
equatorial upwelling regions—or around sharp frontal zones (Moore et al. 2013).
Biological productivity is often low (typically in the order of 100 g C m−2 a−1),
and some oceanic regions like the large oceanic gyres have been described as
marine deserts, largely due to severe nutrient limitation (Polovina et al. 2008).
The absence of biogenic habitats is another defining feature, with the interest-
ing exception of pelagic macroalgae (Sargassum sp.) floating in large mats in
the Sargasso Sea (tropical Atlantic). Human influences on the pelagic ocean are
certainly lower when compared to coastal regions, but still significant through the
effects of fishing, pollution, shipping, and climate change (Halpern et al. 2008).
Major groups of pelagic organisms include both planktonic (passively drifting)
organisms and actively swimming nekton. Primary production is performed by
single-celled phytoplankton, ranging from submicrometer cyanobacteria to larger
diatoms and dinoflagellates, with the community composition largely determined
O B S E RV E D PAT T E R N S 25
30
Latitude
23
16
10
3
6 14 21
Mean richness
Euphausiids
39
Latitude
30
20
10
1
5 17 28
Mean richness
S quids
22
Latitude
17
12
1
5 9 13
Mean richness
Mean normalized richness
0.88
Latitude
0.67
0.46
0.25
0.05
0.1 0.4 0.6
Mean richness
Figure 2.5. Pelagic invertebrates. Shown are the global species richness patterns and corre-
sponding latitudinal gradients for pelagic zooplankton (Foraminifera and Euphausiidae), and
invertebrate predators (squid). Mean normalized richness pattern averaged across these groups
is shown at the bottom. After data from Tittensor et al. (2010). Note that the pattern for squid
includes only commercial species.
O B S E RV E D PAT T E R N S 27
organisms (Amend et al. 2013), and most authors described a mid-latitudinal peak
in diversity (Fuhrman et al. 2008; Sul et al. 2013a; Sunagawa et al. 2015) simi-
lar to other pelagic taxa (Tittensor et al. 2010). For example, in a comprehen-
sive metagenomics study that collected standardized microbial samples across all
oceans except the Arctic, it was found that diversity of operational taxonomic units
(OTUs) peaked at intermediate latitudes around 30 to 40 degrees North or South
(Sunagawa et al. 2015). A geographically more limited study suggested a peak at
20 degrees North along North and Central American coastlines (Raes et al. 2011).
Yet another data set suggested that microbial richness peaks seasonally at high
latitudes in winter (Ladau et al. 2013). Clearly, these results are still in flux, and
probably influenced in part by different sampling techniques (a common challenge
for global biodiversity studies), yet all suggest a global biodiversity pattern that
is more similar to other pelagic species and less similar to coastal or land species.
2.6) again show similar patterns of mid- to high-latitude diversity, peaking around
30 to 40 degrees latitude North or South (Schipper et al. 2008; Tittensor et al. 2010;
Kaschner et al. 2011; Lucifora et al. 2011), with sharks being clearly bimodal and
cetaceans showing an asymmetric peak in the southern hemisphere. The pattern
for sharks tends to be skewed toward the coast, whereas cetaceans appear more
widely distributed throughout the oceans, and generally occur at higher latitudes
than sharks. Yet it is remarkable how the most species-rich cells for both groups
cluster around California, Argentina, South Africa, Japan, and Australia (see fig.
2.6), suggesting common environmental or evolutionary drivers.
Pelagic birds (Order: Procellariiformes) include the albatrosses, shearwaters,
and petrels, which are uniquely adapted wide-ranging seabirds that visit the land
only to nest. They show an interesting pattern of high-latitude richness, again with
a distinct unimodal peak in the southern hemisphere, at around 40 degrees South
(see fig. 2.6; Davies et al. 2010). The most species rich cells are found south of
Australia. Davies et al. (2010) explain this pattern through the reliance of these
seabirds on high and continuous wind speeds allowing long-distance travel to
patchy food sources in the open ocean. Such winds are particularly found in the
circumpolar waters of the Southern Ocean, where no land masses restrict the flow
of air. It is in that region that pelagic seabird richness is highest. It is also notewor-
thy that, like the pinnipeds and cetaceans, seabirds are endotherms, which pro-
vides them with a unique competitive advantage in cold-water environments, and
may in part explain their higher-than-average latitudinal distribution pattern. We
further note that endothermy comes at a cost of much higher metabolic rates and
calorific requirements, which may make these species seek out high-productivity
regions that are typically found in higher-latitude oceans.
2.2.3. Synthesis
The combined patterns of species richness for pelagic invertebrates (see fig. 2.5)
and vertebrates (see fig. 2.6) differ significantly from the patterns seen in coastal
species, as well as those on land. Species richness tends to be more uniformly
distributed latitudinally and longitudinally. Biodiversity hotspots are rarely found
near the equator, and are typically broader and less peaked than in coastal taxa.
Areas of highest richness tend to be located at intermediate or high latitudes,
most commonly between 20 to 40 degrees North or South (figs. 2.5 and 2.6), with
about half of the species groups being particularly diverse in the southern hemi-
sphere (euphausiids, tunas and billfish, cetaceans, and seabirds). It is conceivable
that the much larger pelagic ocean area in the southern hemisphere has promoted
radiation of pelagic taxa there. Despite these hemispheric differences, there is rea-
sonably strong overlap among spatial biodiversity patterns in the pelagic realm,
Tunas and billfishes
11
Latitude
6
1
1 5 8
Mean richness
Oceanic sharks
15
12
Latitude
8
1
0 4 7
Mean richness
Cetaceans
42
Latitude
32
22
11
1
12 22 33
Mean richness
Seabirds
50
38
Latitude
26
14
2
11 22 33
Mean richness
O B S E RV E D PAT T E R N S 31
0.66
Latitude
0.44
0.23
0.01
0.2 0.4 0.6
Mean richness
Figure 2.6. Pelagic vertebrates. Shown are the global species richness patterns and corre-
sponding latitudinal gradients for tunas and billfishes, pelagic sharks, cetaceans (whales and
dolphins), and seabirds (Procellariiformes). Mean normalized richness pattern averaged across
these groups is shown at the bottom. After data from Davies et al. (2010); Tittensor et al. (2010);
Lucifora et al. (2011).
and volume, yet scientists have surveyed only a small fraction of them, due to
their size and inaccessibility. The deep sea features large, seemingly monotonous
plains of fine sediment dotted by intriguing anomalies like seamounts, hydro-
thermal vents, and cold seeps, many of which have not even been discovered or
put on a map. Primary productivity is close to zero, due to the perennial darkness
of the deep sea, where only some chemosynthetic organisms are able to derive
energy from reducing minerals such as sulfides. Most organisms are wholly
dependent on a sparse and patchy rain of detritus that slowly sinks from the sunlit
surface waters to the deeper reaches of the ocean, the so-called export productiv-
ity (or “marine snow”) that is not retained within the surface layers. As much of
this matter is consumed on its way down, total carbon input tends to be very low,
and often occurs in short and variable pulses, linked, for example to episodic
plankton blooms or events such as whale-falls (Rex and Etter 2010). Another
unique feature is the uniformly low temperature found in the deep sea, which
typically ranges between ~2 and 4°C, with the exception of the Mediterranean
deep basin. Due to both low temperature and carbon input, standing biomass and
secondary production also tend to be very low, the latter often in the order of
1 g C m−2 a−1. Exceptions are the unique chemosynthetic organisms and associ-
ated fauna that colonize deep-sea hydrothermal vents and seeps, many of which
are extremophiles (Rex and Etter 2010). Such habitats form productive “oases”
that can sustain a high biomass of a specially adapted and unique fauna. Overall,
however, the low temperature and the variable and sporadic availability of food
are likely to be key organizing factors.
At the time of the Challenger expedition in the late nineteenth century,
the deep sea was presumed lifeless due to the crushing pressure, frigid cold,
absence of light, and seeming lack of productivity. Since then, this view has
been overturned by the discovery of large numbers of morphologically bizarre
animals and microorganisms in these remote habitats. Initially, an early extrap-
olation of species numbers derived from nine stations taken off the US East
Coast suggested there could be many millions of undiscovered species in the
deep sea (Grassle and Maciolek 1992). Although likely an overestimate (May
1992; Mora et al. 2011), that study highlighted the extreme undersampling of
these habitats, and raised significant public interest in the deep sea. Since then,
modest progress has been made in resolving the enigmatic patterns of deep-sea
diversity. Reasonably comprehensive spatial patterns are so far available for
only a few abundant and taxonomically tractable groups, and most of the pat-
terns still rely on few samples, making it difficult to generalize (Rex and Etter
2010). Moreover, with the exception of a recent global study on brittle stars
(Woolley et al. 2016), virtually all previous large-scale work has focused on the
O B S E RV E D PAT T E R N S 33
Atlantic Ocean and has frequently examined only latitudinal gradients rather
than global spatial patterns.
The first fully global study of deep-sea spatial richness patterns focused on brit-
tle stars (Echinodermata; Class: Ophiuroidea), which are one of the numerically
dominant and most widely distributed macrofaunal taxa in the deep sea. The study
used a new database comprising 165,000 species distribution records (Woolley
et al. 2016). It found that deep-sea (2000–6500 m) species richness patterns fun-
damentally differed from those in continental shelf (0–200 m) and slope (200–
2000 m) waters. While ophiuroid richness on the continental shelf and slope still
matched the coastal patterns described earlier (see fig. 2.2), deep-water ophiuroid
diversity peaked at higher latitudes—specifically, the highly productive waters
of the northern North Atlantic, around Japan and New Zealand, and the South
American and South African upwelling regions (fig. 2.7).
Specifically, as seen in fig. 2.2, shallow-water ophiuroid richness peaked in the
tropics (0 to 20 degrees latitude North and South, 20–200 m; Woolley et al. 2016).
A strong latitudinal biodiversity gradient exists at these depths with reduced rich-
ness at higher latitudes (>45 degrees South and >55 degrees North). However,
at mid-slope to abyssal depths (200–4500 m), diversity maxima gradually shift
toward temperate latitudes (30 to 40 degrees South and 40 to 50 degrees North;
Woolley et al. 2016), a unique global pattern that is broadly supported by regional
analyses, discussed later. Southern ocean peaks in diversity are as pronounced as
Abyssal ophiuroids
39
Latitude
30
21
13
4
13 19 26
Mean richness
Figure 2.7. Deep-sea ophiuroids. Shown are the global species richness patterns and corre-
sponding latitudinal gradients for ophiuroids (brittle stars) in waters >2000 m depth and cor-
responding latitudinal gradients for this species group. After data from Woolley et al. (2016).
34 CHAPTER 2
northern hemisphere ones, but less numerous, resulting in lower average richness
across a latitudinal band (see fig. 2.7). The fact that this taxon shows a very differ-
ent global richness pattern in coastal and shelf seas suggests that there are unique
processes structuring biodiversity in the deep sea, likely related to the patterns
in the availability of energy provided by the thin rain of potential food particles
descending from surface waters (Tittensor et al. 2011; McClain et al. 2012; Wool-
ley et al. 2016). However, we caution that the general undersampling of the deep
ocean and the fact that true global spatial patterns have been described only for a
single taxon render our knowledge of the deep sea far more tenuous than for other
major habitats; an exciting opportunity remains to further test models and theories
as new data become available.
Because of the scarcity of global richness data for deep-sea taxa, we briefly dis-
cuss relevant regional studies in addition. Latitudinal patterns in deep-sea benthic
fauna (Mollusca; Classes: Gastropoda, Bivalvia) were first reported from Atlantic
Ocean samples collected between 500 and 4000 m depth (Rex et al. 1993). A
monotonic equator-to-poles latitudinal gradient was observed in the North Atlan-
tic for some taxa, particularly after accounting for different sampling depths.
However, this was not as clear in the South Atlantic, where highest diversity is
commonly observed at high latitudes (Rex et al. 1993; Brey et al. 1994; Brandt
et al. 2007). Another taxon, the sea lice (Arthropoda; Class: Isopoda), showed
high diversity at both tropical and temperate latitudes in the North Atlantic, and at
temperate latitudes in the South Atlantic (Rex et al. 1993). North Atlantic patterns
were also influenced by anomalously low diversity data from the Norwegian Sea
that probably reflect the regional effects of glaciation more than generalizable
gradients of diversity (Rex and Etter 2010). No oceanwide study of these taxa has
been published for the Pacific or Indian Ocean.
Small-bodied meiofauna have also been studied along regional gradients.
Deep-sea benthic nematodes (Phylum: Nematoda) showed no clear patterns
when plotted against latitude in the Atlantic (Lambshead et al. 2000; Rex et al.
2001), but a negative relationship appeared between 0 and 23 degrees latitude in
the Pacific, possibly reflecting the positive effects of enrichment by equatorial
upwelling (Lambshead et al. 2002). Likewise, benthic foraminifera (2000–4000 m
depth) from the Atlantic showed a broad decline in species richness toward the
poles (Culver and Buzas 2000). Peak diversity, however was observed at about 25
degrees South and 45 degrees North, possibly reflecting increased productivity at
these stations. In summary, regional studies on other taxa are somewhat variable
O B S E RV E D PAT T E R N S 35
but partly correspond with the global results on ophiuroids, notwithstanding pro-
nounced uncertainty considering limited sampling.
2.3.3. Synthesis
The vast environment of the deep sea is the most removed from the land, and
has yielded the least scientific data of the four major environmental realms dis-
cussed in this book. Yet it covers 60% of the Earth’s surface and appears to fea-
ture distinct patterns of biodiversity. Although most of the deep sea is perennially
dark, cold, and food-limited, there are unique environments of the deep ocean—
including whale-falls, seeps, and vents—that are sites of high and often unique
biodiversity (Rex and Etter 2010). A global-scale study on brittle stars, supported
by some regional data for other invertebrate and protozoan taxa, suggests peak
diversity at higher latitudes. These zones of relatively high diversity in the abyss
appear associated with areas of high surface productivity and carbon export, as
well as proximity to coastlines (Woolley et al. 2016). We conclude that the light-
less and cold deep sea appears to harbor unique patterns of diversity that we are
only beginning to unveil.
As far as we know, life in the ocean diversified for >2 billion years before the
land was first colonized by plant-like “Ediacaran” life forms 635–542 Ma ago
(Retallack 2013). The lack of an aqueous medium is the most distinguishing
feature of the land, and has a number of important consequences for all life
forms attempting to thrive there. The main obstacle, undoubtedly, is the risk
of desiccation, which requires a series of unique adaptations. These include
protection derived from thicker cell walls and specialized cuticular structures.
Thicker cell walls also play a role in the evolution of body support that needs to
be strengthened when not suspended in water. Furthermore, while algae simply
absorb nutrients from the surrounding environment, a land plant must absorb
both water and nutrients from the soil and transport it to its limbs. After plants
began to use rigid cell walls to grow taller, a system of tubular vessels evolved
to transport water and nutrients. As plants spread farther and diversified, they
also began forming complex habitat structures above and below ground. These
structures transformed the land surface from a largely two-dimensional to a
three-dimensional habitat with many opportunities for coevolution among
plants and animals. Diversification by coevolution, as well as the presence of
36 CHAPTER 2
Vascular plants
6229
4773
Latitude
3316
1860
404
819 2209 3598
Mean richness
Figure 2.8. Land plants. Shown are the global species richness patterns and latitudinal gradients
for vascular plants. After data from Kier et al. (2005); Kreft and Jetz (2007).
Comprehensive information has been compiled for vascular land plants (Tra-
cheophyta) at a global scale (Kier et al. 2005). Generally, and with few exceptions,
plants on land reach their highest diversity in the wet tropics (fig. 2.8), particularly
in Southwest Asia and South America. The highest observed plant species richness
is found in the Borneo lowlands (10,000 species), followed by several regions
located in Central and South America with ~8000 species each (Kier et al. 2005).
Diversity declines monotonically from the equator to the poles on all continents
(fig. 2.8), with the exception of an extraordinary diversity of plants in a few mid-
latitude hotspots such as the Mediterranean Basin and the South African Cape Flo-
ristic Region (Kreft and Jetz 2007). Dry and cold regions, such as the Gobi Desert
or Greenland, show low plant diversity of <500 species per region.
Below-ground diversity on land, particularly of soil and arbuscular mycorrhizal
fungi has recently been studied at some detail (Tedersoo et al. 2014; Davison et al.
2015). Broadly, these groups appear to show similar latitudinal patterns as plants,
but weaker gradients toward the poles (Tedersoo et al. 2014) and much stronger
overlap of species composition between continents, pointing toward their apparent
ease of dispersal via miniscule spores rather than seeds (Davison et al. 2015).
Land vertebrates are likely more comprehensively sampled than any other group
either on land or in the sea. This detailed knowledge, and the strong overlap that
Amphibians
300
Latitude
225
150
76
1
0 71 142
Mean richness
Reptiles
200
150
Latitude
100
51
1
0 33 67
Mean richness
Birds
1664
1249
Latitude
834
420
5
18 277 536
Mean richness
Mammals
350
263
Latitude
176
88
1
2 85 167
Mean richness
O B S E RV E D PAT T E R N S 39
0.59
Latitude
0.39
0.20
0.00
0.00 0.13 0.26
Mean richness
Figure 2.9. Land vertebrates. Shown are the global species richness patterns and corresponding
latitudinal gradients for amphibians, reptiles, birds (excluding Procellariiformes, fig. 2.6), and
land mammals. Mean normalized richness pattern averaged across these groups is shown at the
bottom. After data from IUCN (2016). Note that all groups have >90% species coverage, except
reptilians, which have about 50% coverage.
is observed between species groups, has likely contributed to the common percep-
tion of a uniform latitudinal pattern. Amphibians (Class: Amphibia), for example,
show a pattern of global diversity that is exceedingly similar to plants, with major
hotspots in tropical South America and Southeast Asia and rapidly declining
diversity toward higher latitudes (fig. 2.9). Of course, plants and amphibians are
both ectotherms and are prone to the effects of freezing and desiccation. Hence,
it is maybe unsurprising that they both thrive in similarly warm and wet climates.
Among the vertebrates, amphibians are likely the most susceptible to dry condi-
tions, and uniquely among the land species groups their richness drops to zero in
some of the world’s driest places. Reptiles (Class: Reptilia; excluding sea turtles,
sea snakes, and marine iguanas) are similar to plants and amphibians in their
ectothermy, but more readily able to colonize dry places, and even deserts. Yet
their diversity pattern also centers on the wet tropics, particularly Southeast Asia,
but with some subtropical hotspots in Central America and Madagascar (fig. 2.9).
Their latitudinal pattern peaks slightly north of the equator, reflecting the regional
hotspots in Asia and Central America.
Terrestrial birds and mammals are endothermic and hence regulate their inter-
nal temperature. At the same time, they can be very mobile, in contrast to, say,
plants and amphibians, and some species engage in long-distance migration.
Despite these unique traits, the global richness patterns of birds and mammals
are very similar to those of other vertebrates and plants (see fig. 2.9), with major
peaks in the wet tropics of South America, Africa, and Asia. One major difference
is the much higher diversity in Sub-Saharan Africa for birds, and particularly for
mammals, when compared to plants and amphibians. It is interesting that a major
40 CHAPTER 2
contrast between endotherm and ectotherm diversity patterns is seen not on land,
but in the sea. Specifically, land mammals and land birds still peak in the tropics
(see fig. 2.9), whereas marine mammals and birds show their highest diversity at
higher latitudes (see figs. 2.4 and 2.6).
Some habitats on land are moist year-round and support unique freshwater spe-
cies, including plants and algae, invertebrates, fish, amphibians, birds, and mam-
mals. Perhaps unsurprisingly, amphibians, birds, and mammals show similar
geographic patterns of species richness in freshwater as they do on land, with
richness patterns peaking close to the equator on all continents (Tisseuil et al.
2013). But the same is true for lake and river fishes, which may be more surpris-
ing, because one would suspect the pattern for freshwater fishes to be closer to
that of their marine counterparts. Fish thus provide an interesting case study,
with contrasting gradients from terrestrially influenced closed habitats (lakes)
to those that are fully marine (open ocean). The similarities between freshwater
fish diversity patterns and those of other terrestrial taxa may suggest common
evolutionary and ecological drivers across land and freshwater habitats. Estua-
rine fish are partly influenced by the land, partly by the ocean. Again, perhaps
unsurprisingly, their gradients in global richness appear somewhat intermediate
between freshwater and coastal marine fish, but with additional variation intro-
duced by the individual characteristics of each estuary (Vasconcelos et al. 2015).
Fish diversity in coastal (see fig. 2.4) and pelagic habitats (see fig. 2.6) is again
quite different from freshwater and estuarine fish, each conforming more closely
to patterns seen in other, taxonomically unrelated, groups that share their habitat.
This exemplifies that patterns of diversity appear much more strongly driven by
the features of particular habitats (land, coast, pelagic, and deep sea) than by
taxonomic affiliation.
2.4.4. Synthesis
Spatial patterns of species richness on land appear strikingly similar across plants
and four different vertebrate groups (see figs. 2.8 and 2.9). This similarity occurs
largely irrespective of mobility, endothermy, or trophic position (Jetz et al. 2009),
and likely extends to invertebrate taxa such as insects (Eggleton 2000). Spatial
correlation between most land groups is therefore high, pointing toward common
environmental or evolutionary drivers. This point is reinforced by the fact that
O B S E RV E D PAT T E R N S 41
patterns of freshwater richness tend to follow those on land, at least at the broad
scales examined here. The monotonic latitudinal gradient in diversity from the
tropics to the poles is also clearest on land, but as we saw earlier, tends to break
down progressively from coastal to pelagic to deep-sea environments, an impor-
tant point that we will return to in chapter 3. One important question in paleo-
ecology is whether these gradients are specific to our particular epoch, or whether
they are found throughout the history of life on Earth. The next section will briefly
highlight the evidence available to answer that question.
Like all aspects of nature, large-scale patterns of species richness are certainly
not static through time. Considerable changes in species richness may be forced
by dynamic changes in climate, geological activity, plate tectonics, ocean cir-
culation, and habitat features (Renema et al. 2008). Hotspots of benthic forami-
niferan (Foraminifera living in sediments) diversity, for example, have shifted
progressively over the last 50 million years from southern Europe to Southeast
Asia, likely in response to tectonic changes that altered the availability of coastal
shallow-water habitats on a continental scale (fig. 2.10). During the Eocene, the
number of fossilized foraminiferan genera peaked in southwest Europe, northwest
Africa, and along the eastern shore of the Arabian Peninsula. By the end of the
Eocene, peak diversity had shifted to the Arabian Sea, and by the end of the Mio-
cene Epoch this had shifted again toward Southeast Asia (Renema et al. 2008).
This large change in the distribution of peak species richness has been discussed
primarily as a consequence of coastal habitat loss largely due to regional uplift
during the Arabia-Eurasia collision, likely resulting in faunal depletion and the
demise of the Arabian hotspot. Fossil and molecular evidence for mollusks sup-
ports a similar pattern of “hopping hotspots” through time (Renema et al. 2008).
This, and related work on corals (Leprieur et al. 2016), highlights how changes in
environmental conditions and geological forces can rearrange large-scale diver-
sity patterns over very long timescales, and emphasizes the importance of habitat
availability for maintaining high regional species richness.
The observed rearrangement in benthic foraminiferan diversity over geological
time entailed a shift in peak diversity from intermediate latitudes (Mediterranean)
to tropical regions as the global climate cooled (note that benthic foraminifera
are coastal species and as such more closely aligned with patterns in fig. 2.2 than
with the pelagic foraminifera in fig. 2.5). Recently, it has been suggested that such
latitudinal changes may be a more general pattern throughout the Earth’s history
(Mannion et al. 2014): tropical peaks in diversity tend to occur in cold “icehouse”
Late middle Eocene
Early Miocene
Recent
C
1–4 5–8 9–12 13–16 >16
Figure 2.10. Changes in biodiversity hotspots through time. Shown is the generic richness of
large benthic foraminifera in samples from (A) the late Middle Eocene (42 to 39 Ma); (B) the
Early Miocene (23 to 16 Ma); and (C) recent. Solid lines delimit the changing nature of regions
that support peak diversity from the Tethyan Sea (now Mediterranean), to the Arabian Sea, to
the Indonesian-Australian Archipelago (IAA). Redrawn after data from Renema et al. (2008).
O B S E RV E D PAT T E R N S 43
ecological and evolutionary factors that shaped their diversity patterns. Indeed, a
spatial effect that partly captures such historic differences was significant for all
land plants (Kreft and Jetz 2007), and both past and present environmental condi-
tions have been invoked to explain contemporary patterns in terrestrial vertebrate
diversity (Jetz and Fine 2012). Furthermore, in a comprehensive statistical analy-
sis of marine richness patterns it appeared that most coastal taxa, but not a single
pelagic taxon, showed a statistically significant relationship to historical differ-
ences between ocean basins (Tittensor et al. 2010). This supports the idea that
dispersal constraints preserve some evolutionary footprint in sessile, or relatively
slow-moving, taxa on land and in the coastal ocean.
In this context, there is considerable debate in the literature about whether pres-
ent diversity hotspots for land and coastal taxa arise as centers of species origina-
tion (“cradles”) or centers of species accumulation (“museums” of biodiversity;
Jablonski et al. 2006). The present consensus appears to converge toward the
realization that both might be true—that is, high species origination rates lead to
accumulation of novel taxa in the tropics, which are largely retained there but can
slowly spread to higher latitudes. This combined hypothesis has been termed the
“out-of-the-tropics” model (Jablonski et al. 2006), and is well-supported by the
bivalve fossil record in particular (Roy and Witman 2009). This model of species
evolution assumes both higher evolutionary rates as well as lower extinction rates
in the tropics (Jablonski et al. 2006; Brown 2014).
In summary, we conclude that spatial diversity patterns are not necessarily
stable through time, but are influenced by historical changes in climate and habi-
tat availability, among other factors. Such an evolutionary imprint is likely to be
more visible in dispersal-limited taxa, particularly sessile coastal and land spe-
cies, than in more mobile pelagic species. Clearly, it appears that the latitudinal
gradient that is so often discussed in the literature is not driven by latitude per se,
but by environmental factors or drivers that covary with latitude, and that these
processes may vary among different habitats and throughout different periods in
Earth’s history. These spatial and temporal contrasts will be used to shed more
light on hypothesized environmental drivers in chapter 3.
The global patterns of biodiversity discussed in this section (table 2.1, later) have
been derived from data collected at various levels of taxonomic resolution and
using a variety of sampling methods. Here, we examine how sensitive docu-
mented diversity patterns are to changes in these two parameters. We also briefly
O B S E RV E D PAT T E R N S 45
discuss how our chosen metric of biodiversity (total species richness) relates to
other metrics, which may or may not yield similar patterns at a global scale.
Possibly, some of the variation seen between groups and habitats could be driven
by differences in sampling and approaches to producing data, rather than environ-
mental or evolutionary drivers. Broadly, there are three fundamentally different
methods with slightly different limitations and subsequent inference: (1) Bio-
diversity patterns may be constructed by overlaying individual species ranges.
Vertebrates
2715
2037
Latitude
1359
681
3
17 304 591
Mean richness
Mammals
426
320
Latitude
214
107
1
2 85 167
Mean richness
Carnivores
50
38
Latitude
26
13
1
1 14 27
Mean richness
Bears
3
Latitude
1
0 1 2
Mean richness
O B S E RV E D PAT T E R N S 47
Cats
13
10
Latitude
7
1
0.0 3.6 7.1
Mean richness
These ranges are typically based on observations of species presence and some
degree of interpolation between sampling stations, often informed by expert opin-
ion, to estimate the total extent of occurrence. This method is commonly used for
paleontological data, but also to map the range of total presence in extant species.
It provides a cumulative view of biodiversity, and can overestimate actual ranges
(as there are likely to be gaps within bounding polygons), and hence local biodi-
versity. (2) A second approach is to construct sample-based biodiversity patterns
based on actual biodiversity surveys where the occurrences of all species in the
community are recorded together, and then spatially interpolated by geostatistical
methods such as kriging or relating to environmental parameters. This method is
likely to underestimate biodiversity if sampling is incomplete or accumulation
curves uncorrected (though they can be and sometimes are extrapolated toward
asymptotes using various methods). This approach is also unlikely to fully rep-
resent rare species, and does not provide species identities at interpolated sites.
(3) A third approach is to model individual species ranges using environmental
(and possibly other) predictors—that is, to use habitat or niche models, and then
overlay these species ranges to reconstruct aggregate biodiversity patterns. This
method may overestimate range sizes, and hence biodiversity, if other factors
(such as biotic interactions or historical disturbances) cause marked differences
between realized and fundamental niches and hence restrict actual species ranges
within a possible environmental envelope. Potentially, it may also underestimate
range sizes if the niche is not adequately sampled and characterized.
48 CHAPTER 2
Coastal fishes are one group for which all of these approaches have been imple-
mented at a global scale. Reassuringly, the resulting patterns are very similar,
suggesting that the basic features of large-scale biodiversity patterns are captured
independently of sampling method (fig. 2.12). Whether modeling global occur-
rence records (Tittensor et al. 2010) or compiling transect-based samples (Edgar et
al. 2014), expert-derived range maps (Roberts et al. 2002), or habitat models (Selig
et al. 2014), all observed patterns centered around a primary diversity hotspot in
the western tropical Pacific, with richness decreasing latitudinally and to a lesser
degree longitudinally from there. Likewise, fish richness in the Atlantic centered
on the Caribbean, irrespective of sampling method (fig. 2.12). This suggests some
degree of robustness to differences in sampling methods and data processing for
derived measures of species richness, at least at the coarse scales we choose to
examine here. However, how does species richness relate to other measures of
diversity, and is it a reasonable proxy for functional diversity, evenness, or phylo-
genetic diversity? We explore this question briefly in the following section.
433
292
150
8
B
3856
2895
1934
974
13
C
2555
1919
1284
648
13
D
55
42
29
16
Figure 2.12. Sensitivity of fish diversity patterns to different sampling methods. Shown are
patterns of species richness based on (A) expert-derived range maps for reef fishes (after data in
Roberts et al. 2002), (B) habitat-model derived range maps for all fishes (after data from Selig et
al. 2014, updated October 2017 using www.aquamaps.org), (C) extrapolated occurrence records
for coastal fishes (after data from Tittensor et al. 2010), and (D) transect sampling of reef fishes
(mean richness per transect) by divers (after data from Edgar et al. 2014). Data sources and
methods for estimating species richness were completely independent, yet show very consistent
patterns at the global scale. Data have been restricted to coastal regions where necessary.
50 CHAPTER 2
richness and genus richness that has also been described for land taxa such as
mammals (Safi et al. 2011). Likewise, the pattern of functional richness in coastal
fishes closely resembles the overall pattern of species richness at global (Stuart-
Smith et al. 2013) and regional scales (Micheli and Halpern 2005). These results
suggest a direct although nonlinear link between the number of species and func-
tions in an ecosystem.
In contrast to functional richness, bivalve functional evenness displays a
reverse pattern, peaking at higher latitudes (Berke et al. 2014). Similarly, species
evenness of reef fishes tended to be higher at higher latitudes, and lowest in the
tropics (Stuart-Smith et al. 2013). At least for bivalves, which have a detailed
fossil record available, this contrasting pattern of richness and evenness was
explained by uneven origination rates of different functional groups in the tropics,
and the movement of a small, random subset to higher latitudes (Jablonski et al.
2006). Simulation models showed that such processes could lead to higher even-
ness at high latitudes, indirectly supporting the out-of-the-tropics model for the
analyzed bivalve communities (Berke et al. 2014). Data were broadly consistent
with the hypothesis that high-latitude fauna is to a first approximation an attenu-
ated sample of the global species pool.
Finally, there has been considerable and growing interest in patterns of phylo-
genetic richness, which describes the number of unique evolutionary lineages in
a species group. This metric reflects the recognition that species are not indepen-
dent entities, but rather their functional and ecological similarities are shaped by
patterns of common ancestry (Harvey and Pagel 1991); it also has conservation
implications in terms of preserving evolutionary distinctness. The idea is that it
is not necessarily the most species-rich taxa that harbor the greatest diversity of
lineages, particularly if one or a few groups have radiated profusely but with mini-
mal evolutionary novelty. As with functional richness, there is a problem with
lumping or splitting lineages, and no simple “standard” measure is available at
this point. Still it appears that at large scales, phylogenetic richness does follow
the pattern of species richness surprisingly well, for example, in mammals (Safi et
al. 2011). Taken together, these results suggest some support for the use of species
richness as a surrogate of both functional and evolutionary complexity for some
taxa, though it may not be reflective of other measures such as evenness.
2.7. SYNTHESIS
peaks (n = 16 each; see table 2.1 and fig. 2.13), whereas species groups cresting
in temperate (n = 6) or polar latitudes (n = 2) were more exceptional. Thus the
oft-cited unimodal pattern of biodiversity appears frequently, particularly on land,
but there is also evidence that supports a newly emerging paradigm of asym-
metric unimodal or bimodal peaks, often in the subtropics, and particularly in the
marine realm (Chaudhary et al. 2016). We will return to this difference in marine
and terrestrial gradients in the next chapter, on drivers. Longitudinally, a clear
global maximum in the eastern hemisphere (mostly in South Asia) was seen in
10 groups, whereas global richness peaked in the western hemisphere (Americas)
in only 2 taxa. Most groups (n = 20), however, showed maxima in species rich-
ness in both hemispheres (see table 2.1). These broad patterns across habitats
were robust to differences in taxonomic resolution across species group: when all
taxa below the class level were excluded, the overall latitudinal richness pattern
remained similar (fig. 2.13).
The present synthesis (see table 2.1 and fig. 2.13), confirms earlier studies of
contemporary and fossil-derived patterns of species richness in the sea, indicat-
ing flatter latitudinal gradients and bimodal latitudinal richness patterns, espe-
cially for pelagic taxa (Tittensor et al. 2010; Powell et al. 2012; Chaudhary et
al. 2016). Yet this is the first globally comprehensive comparison of diversity
patterns on land and in all three major marine realms. While diversity patterns
on land (including freshwater habitats) tended to converge on one broad uni-
modal pattern with a strong tropical peak, marine biodiversity patterns were dis-
similar from the land and from each other—coastal, pelagic, and deep-sea taxa
showed a progressively greater tendency for peak diversity at higher latitudes
(see fig. 2.13), and also displayed dissimilar longitudinal patterns (see table 2.1).
As the examples of fishes, mammals, and birds indicate, processes affecting
diversity tend to vary more by realm (land, coastal, pelagic, or deep sea) and less
by taxonomic group—that is, diversity patterns are more similar between differ-
ent taxa in the same environmental realm than within a single taxon that occurs
across different realms.
Total known eukaryotic species richness for just those taxa sampled and pre-
sented in table 2.1 is likely highest on land (~325,000), intermediate in the coastal
ocean (~22,000 species plus ~1000 macroalgal genera), and lowest in the pelagic
realm (~500 species)—though this of course does not reflect total richness across
these taxa, as it is biased toward species we sample well, and also ignores the
raft of taxa for which global patterns remain undescribed. Species richness in the
deep sea is too poorly sampled (a single taxon) to directly compare at a global
scale. This contrast in absolute richness has not been explained comprehensively,
but probably has to do with the presence of structured habitats on land and in
coastal regions. Such features pose constraints on dispersal and favor speciation
52 CHAPTER 2
Latitudinal Longitudinal
Habitat Species group Taxonomy Species peak peak Source
Latitudinal Longitudinal
Habitat Species group Taxonomy Species peak peak Source
Tropical 16 Western 2
Hemisphere
Subtropical 16 Eastern 10
Hemisphere
Temperate 6 Both 20
Polar 2
Note: Shown are the approximate climatic zones where species richness peaks for different species
groups. Most groups have comprehensive global coverage; those with more limited global sampling or
limited species coverage are marked with an asterisk (*). Most studies reported richness of species, except
for macroalgae (richness of genera), and bacterioplankton and soil fungi (richness of operational taxonomic
units, OTU). Latitudinal and longitudinal patterns of peak richness are summarized at the bottom; see also
fig. 2.13 for a graphical summary.
54 CHAPTER 2
20
15 All habitats
10
5
0
10
Land
10
Number of taxa
Coastal
10
Pelagic
2
Deep sea
0
Tropical Subtropical Temperate Polar
Latitude
Figure 2.13. Synthesis of latitudinal biodiversity patterns. Shown are the approximate loca-
tions of latitudinal peaks in species richness in all habitats combined, and separated. Solid line:
all taxa with global data; dashed line: all taxa below Class level removed. Data from table 2.1.
by isolation, seen for example around mountain chains on land and in the coastal
maze of the Indonesian-Australian Archipelago (Bellwood et al. 2012). Land
and coastal habitats also tend to be the most productive, with significant nutrient
sources and frequent vertical mixing. Productivity drops off with increasing dis-
tance from shore and increasing water depth, and is lowest in the deep sea, which
receives only a fraction of the surface productivity from sunlit pelagic waters. At
O B S E RV E D PAT T E R N S 55
the same time, the impacts that people have on marine biodiversity tend to attenu-
ate with increasing distance from land (Halpern et al. 2008) and increasing depth
(Pauly et al. 2003)—though an increasing human footprint is now visible even in
the deep sea (Ramirez-Llodra et al. 2011; Levin and Le Bris 2015; Jamieson et
al. 2017).
Some species move between habitats during their life history, but most are
primarily associated with one of the four major realms discussed here, render-
ing the unique patterns of richness found within each realm largely independent.
Moreover, as we will see in the next chapter, the environmental predictors and
potential drivers of diversity are often spatially distinct between major environ-
mental realms. As such, the contrasting biodiversity patterns among land, coastal,
pelagic, and deep-sea habitats offer a powerful contrast and rich testing ground for
ideas about the fundamental structuring forces of diversity at global scales, and
how they may play out over evolutionary and ecological time. We will examine
empirical evidence for those possible structuring forces in detail in chapter 3, and
then develop a general theory of the processes that may connect diverging pat-
terns and drivers in chapter 4.
CHAPTER THREE
significantly more contrast to separate their relative influence (Kreft and Jetz
2007; Tittensor et al. 2010; Jetz and Fine 2012). Such an analysis has the potential
to be particularly powerful when analyzing multiple terrestrial and marine taxa, as
they can show contrasting patterns of diversity (see chapter 2), and feature differ-
ent environmental gradients and drivers, hopefully shedding light on underlying
unifying principles. The goal of this chapter is to confront published hypotheses
about putative drivers of diversity with comprehensive empirical information on
the environmental predictors of diversity on land and in the oceans. Ultimately,
we hope to identify common drivers and mechanisms that could form the basis for
a synthetic theory of global biodiversity patterns.
Many hypotheses have been formulated to identify drivers and describe associ-
ated mechanisms that may influence large-scale patterns of species richness (table
3.1). Conceptually, these hypotheses fall into three broad categories: (1) those
concerned with environmental factors that are thought to promote diversity (such
as solar radiation, thermal energy, productivity, and environmental stability);
(2) those concerned with factors that may limit diversity (such as environmental
stress or disturbance); and (3) those that focus on size and quality of the habi-
tat, within which observed diversity patterns emerge (table 3.1). It is remarkable,
however, that as yet there is no body of theory that combines these three dimen-
sions of environmental variation to explain, in a testable framework, biodiversity
patterns at the planetary scale.
We note here that hypothesized drivers from the three preceding categories
can operate to promote or limit diversity via the same mechanism—for exam-
ple, increased energy availability and increased habitat area can both result in a
larger community, which, all else being equal, can increase total diversity (the
more individuals hypothesis; Wright 1983; Hurlburt and Jetz 2010). This hypoth-
esized relationship between a higher number of individuals and greater diversity
can itself operate through multiple pathways, such as increased sampling of the
regional species pool, or reduced extinction rates due to larger population sizes.
In this chapter, we focus primarily on the hypothesized drivers of diversity and
their relationship to environmental correlates, but we discuss potential mecha-
nisms wherever possible, to explore them in more detail in our model simula-
tions in chapter 4. Note also that one might separate out potential drivers, as they
may scale differently even if they operate through the same mechanism (Hurlburt
and Jetz 2010). We explore such scaling relationships when fitting our models to
observed data in chapter 5.
58 CHAPTER 3
well play a role. A comprehensive test of all alternative mechanisms within our
analytical framework will be possible, but will not be attempted here.
Life on Earth requires external energy input from the sun, as well as a set of
essential resources such as carbon, water, and nutrients. A number of authors have
proposed that large-scale patterns of species diversity and the latitudinal gradient
on land, in particular, may be generated and maintained by greater energy avail-
ability toward the equator. The concept of biologically available energy promot-
ing diversity, whether such energy is in the form of photosynthetically active solar
radiation (PAR), thermal energy, or chemical energy (Gibbs free energy in the
tissues of organisms), has been formalized in the species-energy theory (Wright
1983; Clarke and Gaston 2006). Hypothesized mechanisms by which energy
may affect diversity fall into two broad classes: (1) Increased energy availability
leads to greater diversification due to increased metabolic rates, faster commu-
nity turnover, and/or more rapid evolutionary processes; this is the evolutionary-
speed hypothesis (Stehli et al. 1969; Rohde 1992; Allen et al. 2002). (2) Increased
energy availability may support higher biological productivity and sustain more
individuals per unit area, which allows more species to coexist; this is the more-
individuals hypothesis (Hutchinson 1959; Wright 1983; Clarke and Gaston 2006).
Note that these mechanisms are of course nonexclusive; they could both operate
simultaneously, the former on evolutionary and the latter on ecological times-
cales (see table 3.1). Here, we examine the three forms of energy that may pro-
mote diversity—solar, thermal, and chemical—followed by an examination of the
influence of environmental stability over time in promoting diversity.
Solar Energy
There is marked latitudinal variation in the seasonality and intensity of solar
energy input, the latter due to the shallower angle of incidence and greater scatter-
ing in the longer atmosphere path length toward higher latitudes. When averaged
over the year, the difference between received solar energy (in W m−2 a−1) at the
tropics and poles is about fourfold (Öpik and Rolfe 2005). Where PAR, water,
and sources of nutrients are abundant, plant productivity may be elevated, which
channels more available energy to higher trophic levels.
Conversely, lower levels of PAR might lead to lower plant diversity because,
all else being equal, fewer individuals can coexist on a given amount of incom-
ing energy; this is the more-individuals hypothesis described earlier (Hutchinson
1959; Wright 1983; Evans et al. 2005). However, only a small fraction of the
60 CHAPTER 3
incident PAR, typically less than 1%, is used by photosynthetic organisms (Öpik
and Rolfe 2005); thus, incident PAR itself may not typically be a strong limiting
factor that drives diversity gradients. Furthermore, such differences in PAR can-
not explain the altitudinal diversity gradient. As is well established on land, diver-
sity decreases for most species groups with increasing altitude (Gaston 2000),
and microevolutionary rates appear lower at higher altitudes (Bleiweiss 1998;
Gillman et al. 2009). Yet solar energy input is constant, or even increasing due to
shorter atmospheric pathlength, with increasing altitude. Finally, a direct effect of
solar irradiance on species richness is not compatible with the extratropical rich-
ness peaks commonly seen in pelagic organisms (see chapter 2). A mechanism
by which solar energy could affect diversity indirectly, however, is through its
effects on (1) surface temperature or (2) productivity. These mechanisms will be
discussed in the two following sections.
Surface Temperature
Elevated solar energy input into tropical environments not only affects PAR but
also elevates average surface temperature compared to temperate or polar areas.
On first glance, it appears obvious for most taxa that more species prefer warm
conditions to cold ones; hence temperature tolerances could explain gradients of
biodiversity for many species groups along both latitudinal and altitudinal clines
(Boyce et al. 2008; Sunday et al. 2011; Beaugrand et al. 2013). Yet, like other
explanations based on the concept of perceived environmental harshness and tol-
erance, this is ultimately circular reasoning, as “thinking of the tropics as benign
and the polar region as harsh is only a habit of thought; it results from the fact
that life is more abundant in the tropics” (Pielou 1979; see also the discussion in
section 3.1.2). More specifically, it raises the key question of why more species
evolved in warmer environments in the first place, and the mechanisms underly-
ing this relationship.
A number of hypotheses have been proposed that link higher temperature to
a faster speed of evolution (the aforementioned evolutionary-speed hypothesis),
hence providing an evolutionary mechanism linking (thermal) energy to diversity
(Clarke and Gaston 2006). Prominent among these hypotheses is the metabolic
theory of ecology (Brown et al. 2004), which posits that the metabolic rate is
the fundamental biological rate that governs many macroecological patterns in
nature. The theory is based on empirically observed relationships between body
size, temperature, and metabolic rate across all organisms. Specifically, small-
bodied organisms tend to have higher mass-specific metabolic rates than larger-
bodied ones. This relationship is empirically described by Kleiber’s law,
B = B0 M3/4,
DRIVERS AND PREDICTORS 61
where B is the whole organism metabolic rate (in watts or another unit of power),
M is organism mass (in kg), and Bo is a mass-independent normalization constant.
The constant exponent of 0.75 (3/4) is thought to be explained by fractal scaling
of resource distribution networks (West et al. 1997).
Furthermore, organisms that operate at higher temperature (either through
endothermy or by living in warm environments) have higher metabolic rates than
those at colder temperatures. This fundamental temperature dependence of bio-
logical reactions is described by the Arrhenius function,
I = ae−E/kT,
where I is a physiological rate, a is an intercept, E is the activation energy in
electron volts (typically around 0.6–0.7 eV for respiration), k is the Boltzmann
constant (8.62 × 10−5 eV K−1), and T is the internal temperature of the organism
in degrees Kelvin (K). Note, however, that the assumed log-linear temperature
dependence holds up only to a species-specific maximum temperature, above
which metabolic rates decline quickly due to the denaturation of proteins at high
temperature (Corkrey et al. 2014). Kleiber’s law and the Arrhenius function were
unified into a metabolic theory of ecology (Gillooly et al. 2001), which modeled
metabolic rates as a function of both body size and temperature, following
B = B0M3/4e−E/kT.
The main implication is that metabolic rate, as influenced by body size and tem-
perature, provides the fundamental constraint by which ecological processes are
governed. If this holds true, many ecological patterns from the level of the indi-
vidual up to ecosystems might be explained, at least in part, by the relationship
between metabolic rate, body size, and temperature.
The theory is supported by the finding that underlying physiological mecha-
nisms are very general. Recently, it was shown that the same metabolic temperature
dependence function applies to prokaryotes as well as unicellular and multicellu-
lar eukaryotes, from extreme-cold-water-adapted to hyperthermophilic forms (−2
to 120°C) (Corkrey et al. 2014). Apparently, the fundamental kinetics remain the
same across the whole range of known thermal adaptations and can actually be
used to predict protein thermodynamics directly from growth rate data. This means
that all known life forms, which evolved at different times over at least 3 bil-
lion years, can be at least broadly described by the same temperature-dependence
model, implying a single, highly conserved reaction that may well trace back to
the last common ancestor of prokaryotes and eukaryotes. The generality of such
molecular mechanisms supports, at least in principle, the metabolic theory (Cor-
krey et al. 2014), and may in part explain why temperature effects on many ecolog-
ical patterns and processes including biodiversity are as reproducible as they are.
62 CHAPTER 3
In light of such generality, the domain of metabolic theory has been gradu-
ally extended beyond the original focus on organismal biology. Most relevant in
the context of this book is the metabolic theory of biodiversity (Allen and Gil-
looly 2006), which is an extension of the metabolic theory of ecology. It assumes
that both the number of generations per unit time (community turnover) and the
number of DNA mutations increase with increasing metabolic rate, and thus with
temperature. It further assumes that the rate of speciation is proportional to muta-
tion rate. Combining these assumptions leads to the prediction that speciation
rate, and (by inference) species richness, increase exponentially with temperature
(Allen et al. 2006).
Assumptions and predictions generated by the metabolic theory of biodiver-
sity have been tested empirically, with varying results. On the one hand, there is
some good evidence that the “molecular clock” of nucleotide substitution and
DNA evolution (Gillooly et al. 2005), as well as rates of genetic divergence and
speciation over geological time (Allen et al. 2006), are positively related to tem-
perature (fig. 3.1A). There is also some detailed experimental evidence that shows
increased mutation rates at higher temperature (Muller 1928; Lindgren 1972),
particularly when ambient temperature approaches the upper tolerance limit of a
species, inducing physiological stress (Matsuba et al. 2013). Temperature stress
may directly interfere with DNA repair mechanisms, which are more error-prone
when individuals are in poor condition (Agrawal and Wang 2008). Alternatively,
the metabolic-rate hypothesis posits that most mutations are caused by genetic
damage from free radicals produced as by-products of metabolism; therefore,
mutation rate should be related to mass-specific metabolic rate and hence body
temperature (Martin and Palumbi 1993).
Species richness, on the other hand, does not always display the straightfor-
ward relationship with temperature predicted by metabolic theory (Algar et al.
2007; Hawkins et al. 2007). It has been hypothesized that other factors, like water
availability (Hawkins et al. 2003), may confound or override the influence of
temperature on species richness. Note, however, that this specific factor would not
apply to the marine environment. Generally speaking, while increased mutation
rate under elevated temperature has been demonstrated many times, the link to
speciation and macroevolutionary rates has not been established experimentally,
and the mechanisms remain unclear.
Empirically, however, the fossil record provides compelling correlative evi-
dence that such a link may exist. Long before the metabolic theory of biodiversity
was proposed, a relationship between temperature and speciation rate had already
been inferred from the marine fossil record, particularly for mollusks and corals,
where evolutionary rates for warm-water fauna seemed to exceed those of cos-
mopolitan or cold-water species, leading to the typical equator-to-poles richness
gradient over latitude. This gradient was seen in fossil marine assemblages at least
0
A
y = –0.67x + 22.96
–1
R2 = 0.34, P= 0.003
(% substitutions nucleotide -1 Ma-1 g1/4)
–2
ln(rate of evolution)
–3
–4
–5
–6
–7
39 39.5 40 40.5 41
Ocean temperature (1/kt)
4
B
3
2
Residuals (sd)
–1
–2
–3
–500 –400 –300 –200 –100 0
Time (Ma)
Figure 3.1. Environmental drivers of diversity over deep time. Shown are (A) effects of inverse
ocean temperature on the rate of evolution in foraminifera; (B) corresponding patterns of tem-
perature variability (black circles) and marine invertebrate species origination rates (white
circles) over the last 500 Ma. Gray circles denote the five major mass extinctions. The residu-
als plotted are mean-standardized after detrending. Redrawn after data in Allen et al. (2006);
Mayhew et al. (2012).
64 CHAPTER 3
270 Ma ago, and persists into the present for many well-studied groups (Stehli et
al. 1969). The authors hypothesized that there might be thermodynamic effects
on reaction and mutation rates, or, alternatively, a direct effect of incoming solar
energy on evolutionary rates. Further supporting this assertion was the observa-
tion that the age of coral genera was negatively correlated with water temperature
in both the Pacific and the Atlantic Ocean, such that genera that recently evolved
were most abundant in the warmest regions of these two oceans (Stehli and Wells
1971). Later, detailed analyses of the fossil record across marine and terrestrial
taxa yielded contrasting results and much discussion regarding the effects of tem-
perature on diversity (Mayhew et al. 2007). This controversy was resolved by
applying new methods that correct for temporal sampling bias in the global fossil
record (Mayhew et al. 2012). The corrected data show a clear positive associa-
tion between temperature and the rate of evolution over time, and also a positive
relationship between temperature and origination rates for marine taxa (see fig.
3.1B). Likewise, on land a latitudinal difference in the rates of molecular evolu-
tion (faster in the tropics) has been described in a range of organisms using sister
species comparisons of plants, frogs, and mammals (Dowle et al. 2013).
Note, however, that such predictable relationships are not always found in
endotherms, since their internal temperature and metabolic rate is largely decou-
pled from external temperature. Indeed, the rate of molecular evolution of land
birds (Bromham and Cardillo 2003) did not show a latitudinal gradient. A compre-
hensive study of sister taxa of land mammals, however, gave strong evidence of
microevolutionary rates being elevated both at low latitudes and at low altitudes,
indicating some effect (direct or indirect) of environmental temperature on the
speed of evolution (Gillman et al. 2009). Results could not be attributed to other
factors thought to influence rates of microevolution, such as body mass or genetic
drift. Instead, the results indicated that the tempo of microevolution among mam-
mals is either responding directly to the thermal environment or indirectly via a
mechanism such as the “Red Queen” effect of increased evolution driven by rap-
idly evolving disease agents (Gillman et al. 2009). Similar results were obtained
when contrasting low-altitude and high-altitude hummingbird taxa that exist along
a smaller-scale temperature gradient (Bleiweiss 1998). In summary, it appears that
the evidence for temperature effects on evolutionary rates is strong, but not uni-
versal, especially when considering both ectothermic and endothermic organisms.
Productivity
All else being equal, greater energy availability will lead to greater production
of organic matter over time (g C m−2 a−1), and support a greater density of indi-
viduals in both plants and animals that feed on them. This productivity hypothesis
can act through multiple mechanisms that link increased abundance to increased
DRIVERS AND PREDICTORS 65
A second class of hypotheses emphasizes the factors that may limit the prolif-
eration of species in challenging environmental regimes—for example, at high
latitudes or high altitudes. These include the effects of chronic, seasonal, or epi-
sodic environmental stress or disturbance (see table 3.1). Environmental stress
is defined as “an action, agent, or condition that impairs the structure or func-
tion of a biological system” (Cairns Jr. 2013). Typically, this relates to harsh
DRIVERS AND PREDICTORS 67
Environmental Stress
With respect to environmental stress, there is little doubt that stressful condi-
tions limit the survival of organisms not adapted to these conditions, and hence
may impact both community abundance and diversity (Thiery 1982). Few
environmental stressors, however, vary consistently by latitude, and dominant
stressors vary between land and ocean. Desiccation or water stress, for example,
is a major issue on land, but not in the oceans, except for some intertidal envi-
ronments (Baker 1910). Oxygen depletion, by contrast, strongly affects the dis-
tribution and diversity of some marine taxa (Brill 1994; Worm et al. 2005), but
is not an issue in most terrestrial habitats. As such, most hypotheses that invoke
environmental stress are specific to a particular species group and do not easily
scale across taxa and habitats. One universal factor that scales with latitude in
the oceans and on land, however, is temperature. Several published hypotheses
on the geographical patterns in species diversity emphasize temperature stress
(Rohde 1992; Beaugrand et al. 2013; Beaugrand 2014), and the fact that species
are limited in their distribution by their widely varying tolerance to suboptimal
temperature. These tolerances represent the lower, upper, and optimum temper-
atures at which species can metabolize, grow, and reproduce. Such physiologi-
cal limits tend to be fairly stable, a phenomenon that has been described as niche
conservatism (Wiens and Graham 2005). Experimental evidence for thermal
niche conservatism comes from a strain of the bacterium Escherichia coli that
was maintained over 2000 generations (Bennett and Lenski 1993). Selection
for low, medium, and high temperature produced changes in the physiological
temperature optimum, but no change at all in the upper and lower limit of the
temperature niche. This suggests that within their temperature niches, at least
some organisms are able to adapt to fluctuations or directed changes in tem-
perature, but that this does not affect their absolute thermal niche. However,
68 CHAPTER 3
the same authors observed slight (1–2 degree) changes in niche width when
selection pressures operated at the lower thermal range boundary (Mongold et
al. 1996). Likewise, the paleontological record also suggests that clades that
originated in the tropics can evolve tolerance to cooler temperatures; however,
substantial time lags (~5 Ma) between the origins of tropical clades and their
expansion into the temperate zone suggest that this process occurs rather slowly,
if at all (Jablonski et al. 2013).
From this arises a question about the physiological mechanisms that underlie
the observed conservatism of thermal niches. Or as Hutchinson (1959) famously
asked: If one species can adapt to colder environments, why don’t they all do it?
(Note that this question applies to other stressors as well.) In a book devoted to
the topic, Johnston and Bennett (1996) summarize the evidence for temperature
adaptation in animals from a physiological and evolutionary perspective. Adap-
tation to low temperature emerges as a multifaceted problem, involving a large
number of specialized mechanisms at the molecular (for example, changes to key
proteins and enzymes), cellular (changes to cytoskeleton and membranes), and
physiological (changes to circulatory system) levels. This may suggest that the
basic metabolic “machinery” of life evolved at warm temperatures and that colo-
nization of cold environments (that is, high latitudes, high altitudes, and the deep
sea) involves a suite of “evolutionary fixes” that make life at low temperature pos-
sible. There also seem to be strong trade-offs involved, resulting in narrower tem-
perature niches in polar species than in warm-adapted ones. An example are the
Antarctic cod icefishes (Nothotheniidae), which feature many specialized adapta-
tions to survive subzero temperatures, but cannot survive at ambient temperatures
greater than 6 to 8°C.
The absolute width of thermal niches may also be related to the variability
experienced in a given environment. This is Rapoport’s rule, which states that
species at higher latitudes should have larger latitudinal ranges (Taylor and Gaines
1999), because seasonal temperature variability is larger at higher latitudes, and
thus species need to be tolerant of a range of conditions, or perform seasonal
migrations, both of which could result in larger ranges. This is particularly true
on land, where absolute fluctuations are much higher than in the sea, particularly
near the poles.
Given these findings about the importance of temperature niches, it may not
be surprising that the latitudinal range limits of many species can be predicted
well from their temperature tolerances (Boyce et al. 2008; Sunday et al. 2012),
and observed temperature changes tend to go in lockstep with changing species
ranges at seasonal (Whitehead et al. 2008) as well as interannual scales (Pinsky
et al. 2013). There appear to be some interesting contrasts between terrestrial
and marine species, however: Sunday et al. (2011) found that upper and lower
DRIVERS AND PREDICTORS 69
Disturbance
Disturbances are, by definition, discrete and episodic events that disrupt commu-
nities and influence diversity in many ways. Large disturbances, such as ice ages,
could have shaped geographic patterns of diversity by driving up extinction rates
at high latitudes and resetting the evolutionary trajectory (Fischer 1960). There is
good evidence that the ice ages have had a lasting effect on communities at higher
70 CHAPTER 3
latitudes, particularly on land (Comes and Kadereit 1998). However, it is not clear
that the legacy is one of reduced species richness; indeed, these disturbances can
lead to reproductive isolation of populations, genetic divergence, and speciation
(Hewitt 1996). Moreover, few studies have found general support for elevated
extinction rates at high latitudes, and there is a general sense that differences in
extinction rates alone cannot sufficiently explain observed geographic gradients
in species richness (Hawkins et al. 2006; Jablonski et al. 2006; Dowle et al. 2013).
Only marginally related to the question of global gradients in biodiversity, the
intermediate disturbance hypothesis does not consider speciation and extinction
as much as ecological interactions within a community. It predicts high diversity
under “intermediate” rates of disturbance, where competitive exclusion by domi-
nant species is prevented. While Connell (1978) used this hypothesis to explain
high richness in coral reefs and rainforests, later tests of the hypothesis found
that it has reasonable predictive power in rocky shore and plankton communi-
ties, albeit at a local scale (Sousa 1979; Petraitis et al. 1989; Floeder and Sommer
2000). Like many hypotheses that focus on species interactions, the intermedi-
ate disturbance hypothesis is more concerned with mechanisms that explain the
maintenance of species diversity in a particular community, rather than the gen-
eration of diversity gradients at global scales on land or in the sea.
Seasonality
One of the obvious differences between tropical and temperate environments is
the pronounced seasonality found in the latter, which necessities specialized adap-
tations such as hibernation, migration, and larger thermal niches. Larger variation
in environmental conditions may cause mortality and possibly local extinction—
for example, during harsh winters—thus limiting diversity (seasonality hypoth-
esis; Fischer 1960). Another aspect is food supply, which is much more seasonal
and episodic at high latitudes, requiring a shift from specialist to generalist feed-
ing strategies. Taylor and Taylor (1977) observed such a shift in feeding patterns
among predatory gastropods in the North Atlantic, concomitant with a large
decrease in diversity. This decrease was most marked around 40 degrees North
latitude, a point at which primary production changes from a less seasonal to a
strongly seasonal regime (Taylor and Taylor 1977). Later authors, however, did
not find a consistent correlation with seasonality across species groups (Currie
1991), and a similar breakpoint in diversity around 40 degrees latitude is not gen-
erally seen in the maps in chapter 2. Like environmental stress and disturbance,
it appears that seasonality may explain aspects of individual patterns in diversity,
but might not present the level of generality across taxa that is evident in the
global patterns documented in chapter 2.
DRIVERS AND PREDICTORS 71
Finally, it is intuitive that the size and quality of the available habitat area may
play a role in setting at least an upper limit on the number of species that can
coexist there (Connor and McCoy 1979). Larger areas can support larger popula-
tions, all else being equal, similarly to areas of higher productivity, potentially
leading to increased diversity through mechanisms such as reduced extinction
rates or sampling from the regional species pool (the more-individuals hypoth-
esis). Larger areas also often feature a greater diversity of habitat types, poten-
tially providing a greater variety of niches for species to inhabit. Finally, the
physical complexity and shape of a habitat (for example, a range of elevations,
or structural complexity of a forest versus a grassland), and the trophic complex-
ity of associated food webs, may further elevate observed richness of species
occupying those habitats.
Habitat Area
The simple observation that larger areas typically yield greater species counts
is formalized in the species-area relationship (SAR)—without doubt one of the
most general ecological laws (Preston 1962; MacArthur and Wilson 1967; Con-
nor and McCoy 1979; Hubbell 2001; Drakare et al. 2006). The SAR captures the
relationship between the area of a habitat (A), and the number of species (S) found
within that habitat. Empirically, the relationship is most often fitted by a power
function of the form
S = cAz.
Here, z denotes the linear slope of the relationship in log-log space, and c the
intercept. A meta-analysis of 794 SARs confirmed the relationship’s generality,
but also found that its parameters can vary systematically with latitude and body
size (steeper slopes in the tropics and for large organisms), habitat, as well as
sampling scale and design (Drakare et al. 2006).
How can this pattern be explained mechanistically? MacArthur and Wilson
explored the SAR from first principles in their theory of island biogeography
(MacArthur and Wilson 1967), modeling the balance of immigration and extinc-
tion on islands of different size and isolation. Larger islands support larger popu-
lations on average, which are predicted to result in lower extinction rates, and
hence more species will coexist on a larger island, all else being equal. Building
on MacArthur and Wilson’s work, the neutral theory of biodiversity and bioge-
ography generalizes to communities not located on islands by constructing a neu-
tral metacommunity with constant per capita rates of dispersal (immigration and
72 CHAPTER 3
emigration), speciation, and extinction (Hubbell 2001). The model also yields
a species-area relationship that closely follows empirically observed patterns.
Hence, there is both a solid theoretical as well as an empirical foundation for the
effects of increasing habitat area on species richness. Note that the SAR and the
productivity hypothesis share at least one common mechanism—specifically, that
both larger areas and more productive ones will support larger populations that
are less likely to go locally extinct (see table 3.1). Hence, Wright (1983) treated
the effects of productivity and area on species richness as additive. In his empiri-
cal analysis, the best predictor for bird species richness on islands was total island
area multiplied by the rate of primary productivity per unit area. More recent work
on birds, however has suggested more complex scaling relationships, in which
both area and productivity scale with richness, but to different degrees in different
regions (Hurlburt and Jetz 2010).
Habitat Complexity
Another aspect of the environment that may affect diversity is the number of dif-
ferent habitat types, which also tends to increase with sampling area, and corre-
lates well with species richness. Both effects (larger populations and more habitat
types) may contribute to the observed rise in species richness with habitat area
described by the SAR, but these mechanisms are not explicitly captured by the
theory of island biogeography or the neutral theory of biodiversity. Their rela-
tive importance can be analyzed by quantifying the effects of surface area, the
number of habitats types (for example, different vegetation types), and the struc-
tural complexity of the habitats (for example, elevational range). Some aspects
of habitat complexity are physical (for example, topographic complexity) while
others are biological (for example, structural complexity of biogenic habitats such
as forests and reefs). There is a clear sense that more complex habitats allow for
a greater range of species to coexist, owing to a larger variety of environmen-
tal conditions and food sources. Likewise, trophic complexity may beget greater
diversity by allowing for more niches, such as specialized predators, parasites,
and the like. Some aspects of trophic complexity may also act to maintain high
diversity, such as the action of a keystone predator maintaining diversity of prey
items by preventing competitive exclusion; this is the predation hypothesis (Paine
1966). These two aspects, increasing specialization and trophic complexity, point
at potential mechanisms by which “diversity may beget diversity” (Brown 2014).
Such mechanisms help to explain the extraordinary diversity found in some com-
plex habitats, but cannot explain the emergence of a geographic gradient in the
first place, as they do not identify the causal mechanisms behind initial diversifi-
cation (Rohde 1992).
DRIVERS AND PREDICTORS 73
Habitat Domain
The spatial domain within which habitat and species assemble geographically
may influence the spatial pattern of diversity. A much-discussed example is the
mid-domain hypothesis. This is different from the previously discussed theo-
ries in that no environmental driver is assumed to operate here, but purely an
interaction of geometry and species ranges. The mid-domain hypothesis is a
null-model that assumes random placement of species ranges across a habitat
area, or domain. This will result in maximum richness near the center of that
spatial domain, such as a continent or an ocean basin. The argument is that the
observed maximum richness in the central Indo-Pacific, for example, could be
predicted by such a null-model (Colwell and Lees 2000). Empirical support for
this hypothesis has been quite weak at biogeographic scales, however (Hawkins
and Diniz-Filho 2002; Zapata et al. 2005; Currie and Kerr 2008), indicative of
nonrandom placement of species ranges. Yet this hypothesis has invigorated an
interest in null-models in ecology to be confronted with available data. Recently,
Beaugrand et al. (2013) extended the null-expectation of a spatial mid-domain
effect to niche space, assuming random placement of temperature niches within
the bounded range of observed sea surface temperatures in the global ocean.
As discussed earlier, this randomly generated pattern fit reasonably well to the
observed richness gradients in pelagic zooplankton, but required some additional
modifications due to the assumed effects of ice ages reducing richness at high
latitudes (Beaugrand et al. 2013). We will return to null-models of species rich-
ness in chapter 4.
0.9
0.8 B Birds
0.7
0.6 Total
Temp
0.5 NPP
R2
0.4 ElRange
HabVar
0.3 Area
0.2
0.1
0
0.9
0.8 C Amphibians
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Local 100 200 400 1000 2000
Spatial scale
DRIVERS AND PREDICTORS 75
Such results reflect an important reason why we are focusing our discussion
on large spatial scales and the general ecological and evolutionary mechanisms
operating at those scales. As discussed in the previous chapter, this is also why we
map all data to the same coarse grid, avoiding the scale dependence and reduced
explanatory power mentioned earlier, and also minimizing uncertainty and spuri-
ous overprecision associated with some spatial data. Local diversity patterns are
likely shaped by an alternative set of mechanisms (see fig. 3.2), including but not
limited to local species interactions, environmental constraints, and even human
impacts on local habitat structure and biodiversity (Sommer and Worm 2002).
These factors may determine which species from the regional species pool will be
able to colonize, reproduce, and maintain a viable population at a local site. While
these are clearly important questions—for example, in the context of local-scale
management in the face of environmental change—we are primarily concerned
with the factors that have determined the size and distribution of the regional spe-
cies pool in the first place.
We also have to be aware of different processes operating at different temporal
scales. First, there is the geological timescale (10–100 Ma). At this scale, major
tectonic events are changing the face of the planet, affecting the size and geog-
raphy of ocean basins and continental margins. The fossil record suggests that
hotspots of marine biodiversity wax and wane in the wake of such major tectonic
events: they persist for some time (tens of Ma), and then fade as new geological
events reshape these habitats (Bellwood et al. 2012). These are the previously
discussed “hopping hotspots” (Renema et al. 2008), which provide a dynamic
deep-time perspective on the spatial organization of marine biodiversity (see fig.
2.10). Second, there is the evolutionary timescale (0.01–10 Ma). At this scale,
macroevolutionary processes play out within a particular geological setting. For
example, repeated changes in sea level can bring about isolation of islands, atolls,
and semi-enclosed seas, thereby creating spatial barriers that lead to speciation
(the so-called species pump; Bellwood et al. 2012). Third, at the ecological tim-
escale (<0.01 Ma), shorter term climatic and biological processes may prevail,
Figure 3.2. Analyzing environmental predictors of vertebrate species richness across scales.
Spatial scale represents the approximate diameter (km) of the spatial units that were used to ana-
lyze global relationships between hypothesized environmental correlates and species richness.
While the effects of annual net primary productivity (NPP) dominate at smaller scales, tempera-
ture (Temp) becomes the dominant environmental predictor at larger scales. Total predictive
power and variance explained (Total R2) also increased at larger spatial scale. Elevation range
(ElRange), habitat variety (HabVar), and area (Area) are less important, particularly at the larger
spatial scales. Redrawn after data in Belmaker and Jetz (2011).
76 CHAPTER 3
Variable Fishes Sharks Pinnipeds Bivalves Corals Snails Cephalopods Seagrasses Mangroves
SST 10.7*** 7.1*** −10.0*** 7.3*** 7.7*** 8.6*** 7.1*** 4.4*** 9.3***
SST slope 2.4* 4.3**
Coastline length 7.9*** 13.0*** 4.5*** 5.7*** 3.1** 4.1* 6.5*** 4.3** 2.0*
Primary productivity 3.6** 5.5*** −2.2**
SST range −2.5* −3.2** 3.7**
Oxygen stress
Indian Ocean 3.7** 3.8** 7.0*** −1.8* 2.6*
Pacific Ocean 4.5*** 3.5** 6.9*** −2.8** 2.0*
Pseudo R 2 0.71 0.75 0.88 0.97 0.73 0.78 0.89 0.75 0.85
Note: Shown are results of minimal-adequate spatial linear model for environmental correlates (SST = sea surface temperature). Numbers are z-values; asterisks
represent significance levels at P < 0.05 (*), 0.01 (**), or 0.00001 (***). Ocean column z-values represent contrast against Atlantic Ocean. Species groups and
scale of analysis correspond to patterns shown in figs. 2.2 and 2.3, with the exception of ophiuroids, which can be found in table 3.2. Individual drivers are shown
in fig. 3.4.
Source: After data in Tittensor et al. (2010), Valentine and Jablonski 2015, and IUCN 2016.
Table 3.3. Environmental Predictors of Species Richness in the Pelagic Realm
22
14
−1
B
Sea surface temperature slope
°C / 100km
2.5
1.9
1.3
C Coastline length
km
30000
22501
15002
7503
D
Net primary productivity
gC/ m2 / year
500
380
259
139
19
Figure 3.3. Present-day environmental predictors of species richness in: the surface ocean
(A–D), the deep-ocean (E–F) and on land (G–I). These predictors empirically represent the best
correlates of species richness overall, see also tables 3.2 to 3.4 and chapter 5. A–D after data
in Tittensor et al. (2010); E–F after data in Woolley et al. (2016); G–I after data in Kreft & Jetz
(2007) and Climatic Research Unit Global Climate Dataset (http://www.ipcc-data.org/observ
/clim/cru_climatologies.html).
E
Annual mean seafloor temperature
°C
10.0
7.3
4.5
1.8
F
Seafloor particulate organic carbon flux
g C / m2 / year
5.0
3.8
2.6
1.4
G
Temperature
°C
28
16
−8
−20
H
Wet days
/ month
27
20
13
1120
747
374
1
20
13
7
82 CHAPTER 3
0
I
Net primary productivity
gC / m2 / year
1493
1120
747
374
Environmental predictors that are commonly analyzed in the oceans are displayed
in fig. 3.4 at the same resolution as the species richness data in chapter 2. These
relate to the hypotheses mentioned earlier as follows: Sea surface temperature
(SST) and net primary productivity (NPP) relate to the effects of thermal energy
(evolutionary-speed hypothesis and thermal-niche hypothesis) and productiv-
ity (more-individuals hypothesis). Productivity in the deep sea, however, is not
directly fueled by NPP, but by export production raining down from surface
waters, and measured as particulate organic carbon (POC) flux (Tittensor et al.
2011; McClain et al. 2012; Woolley et al. 2016), thus replacing NPP as a predic-
tor of species richness (more-inidviduals hypothesis). Moreover, a major stress
factor unique to the ocean is low oxygen concentration (<2 mL L−1), found regu-
larly in upwelling areas, and expanding elsewhere (stress hypothesis; Stramma
et al. 2010). The length of coastline (km) relates to the area of coastal habitat
(habitat-area hypothesis; Tittensor et al. 2010). The steepness of the SST gradient
approximates the availability of thermal fronts, a critical habitat feature in pelagic
waters (habitat-area and habitat complexity hypotheses; Olson et al. 1994). The
standard deviation in SST and seasonal variability in NPP can both be used to test
the seasonality hypothesis.
When analyzing statistical relationships between these predictors and richness
of coastal and pelagic species groups, the spatial pattern of species richness was
most strongly and consistently related to sea surface temperature (see tables 3.2
and 3.3), and the univariate relationship between temperature and species richness
was either near-linear (in most coastal taxa) or saturating (in most pelagic taxa;
see fig. 3.4). This finding corresponds to patterns of coastal species richness peak-
ing equatorially or in the tropics (see figs. 2.1 to 2.4) and pelagic species peaking
in the subtropics (see figs. 2.5 and 2.6), respectively.
DRIVERS AND PREDICTORS 83
Cetaceans Pinnipeds
10
40
8
Species richness
30
6
20
4
10
2
0
0 5 10 15 20 25 30 0 5 10 15 20 25
14 100
12 80
Species richness
10
60
8
6 40
4 20
2
0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
SST SST
Figure 3.4. Univariate relationship of species richness and sea surface temperature in coastal
and pelagic taxa. Note distinct relationships for endotherms (pinnipeds, cetaceans, and sea-
birds). Each point is a cell in the global equal-area grid. Trends indicated by smoothed fit from
generalized additive model with basis dimension 3. Gray shading indicates 95% confidence
limits. Updated and expanded; after data in Tittensor et al. (2010), Davies et al. (2010), and
Valentine and Jablonski (2015).
8
Species richness
1500 6
4
500
2
0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
20
30
Species richness
15
20
10
10
5
5
0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
SST SST
Likewise, in pelagic taxa, the second most influential predictor of diversity was
sea surface temperature (SST) slope, which indicates the steepness of oceano-
graphic fronts and transition zones that are commonly used by pelagic organisms
as feeding habitats and migratory routes (Haney 1986; Olson et al. 1994; Polovina
et al. 2001; Ferraroli et al. 2004). Steep frontal areas add habitat complexity, com-
mingle individuals through advection, and aggregate species across diverse taxa.
As such, these pelagic habitat features may act similarly to the more solid coastline
habitats for near-shore organisms, or topographic complexity for terrestrial taxa.
It is interesting that the primacy of temperature as a positive predictor of diver-
sity was not seen as clearly for endothermic groups such as pinnipeds, cetaceans,
and seabirds (see tables 3.2 and 3.3). These endothermic species groups may be
partly decoupled from the constraining effects of ambient temperature in terms
of their distribution but also their evolutionary speed: no significant difference
DRIVERS AND PREDICTORS 85
Corals Mangroves
500 15
Species richness
300 10
5
100
15 20 25 15 20 25
Seagrasses Foraminifera
30
15 25
Species richness
20
10
15
5 10
0 5 10 15 20 25 30 5 10 15 20 25 30
SST SST
was seen in diversification rates of birds across a global latitudinal gradient (Jetz
et al. 2012b). Yet this partial independence from ambient temperature comes at
a cost of high metabolic demands, which require high ambient productivity to
satisfy (Lavigne et al. 1986). Hence, many endothermic species may converge
around areas of high productivity, irrespective of water temperature. Indeed, pro-
ductivity is a strong predictor for marine mammal species richness, which tends
to peak in high-latitude convergence zones or low-latitude upwelling areas, both
of which are cool and nutrient-rich (Schipper et al. 2008); similar patterns are seen
in seabirds (Davies et al. 2010). Where sufficient productivity is available to sup-
port high-energy lifestyles in cooler waters, the enhanced metabolic rate found in
endotherms may provide competitive advantages over ectothermic species.
The second group that was not primarily predicted by ambient temperature was
deep-sea species (see table 3.4). While the richness of shelf and slope ophiuroids
86 CHAPTER 3
30 150
Species richness
20 100
10 50
0 0
0 5 10 15 20 25 30 10 15 20 25 30
Seabirds Bivalves
50 1500
40
Species richness
1000
30
20
500
10
0 5 10 15 20 25 30 0 5 10 15 20 25 30
SST SST
was still strongly and positively related to ambient temperature, deep-sea richness
was much more strongly related to patterns in export production, and the sea-
sonality in NPP, which drives large pulses of export production during seasonal
blooms (see table 3.4; Woolley et al. 2016). Similarly, for gastropods, bivalves,
and nematodes analyzed in the Atlantic Ocean, export productivity from surface
waters was consistently the best predictor of species richness (Lambshead et al.
2000; Lambshead et al. 2002; Tittensor et al. 2011). This likely relates to the fact
that, except for some isolated basins such as the Mediterranean, temperature is
near-uniformly cold across the deep sea, typically ranging between 2 and 4°C.
Thus, metabolic rates will not vary dramatically in these habitats and are more
likely to respond to local changes in food supply than temperature. Hence, while
thermal energy (proxied by temperature) clearly emerged as a primary environ-
mental predictor for most groups, productivity was also important for species
DRIVERS AND PREDICTORS 87
that experience little temperature variation, either because they are endothermic
(near-constant warm body temperature) or live in the deep sea (near-constant cold
body temperature).
When considering such statistical associations between present-day environ-
mental variables and diversity, there is of course always the question of causation.
A static pattern provides weaker evidence for causal relationships than a dynamic
pattern, especially as the modern-day environment may not reflect past influences.
Paleo-ecological studies can be useful in this regard, as the association between
the environment and diversity can be tracked across changing conditions. Taxa
that form abundant microfossils are particularly well-suited for this work, because
they can be retrieved from sediment cores across all oceans in large numbers, pro-
viding a comprehensive image of global diversity patterns through time (Yasuhara
et al. 2015).
Foraminifera, for example, form small exoskeletons that fossilize well and are
readily identified to species. As for most other marine groups, the present richness
pattern of pelagic foraminifera correlates closely with sea surface temperature,
which explained 90% of the observed variation in foraminiferan richness (Ruth-
erford et al. 1999). Other variables, such as chlorophyll a concentration, salinity,
nitrate, or mean solar irradiation, were much less powerful predictors of richness
for this taxon. Subsequent analysis of temporal changes in diversity patterns for
pelagic foraminifera supported the overriding influence of temperature variation
as a driving force (Yasuhara et al. 2012). While the strength of latitudinal gradi-
ents adjusted dynamically to changes in average water temperature across gla-
cial and interglacial cycles over the last 3 million years, the diversity-temperature
relationship remained remarkably constant (Yasuhara et al. 2012). This suggests a
causal relationship that is driving biogeographic responses to variation in climate
at long timescales. Interestingly, a similar adjustment is often observed at much
shorter timescales, corresponding to seasonal and interannual variation in water
temperature (Whitehead et al. 2008; Pinsky et al. 2013), again suggesting a causal
relationship that tracks temperature variation over space and time.
Of course, such a relationship may not apply equally to each taxonomic
group, or at all spatial scales. While species richness of most coastal inverte-
brates shows a strong relationship to temperature at a global scale (Tittensor et
al. 2010; Valentine and Jablonski 2015), for example, Fernandez et al. (2009)
found reverse latitudinal gradients (and relationships to SST) among benthic
invertebrates with different life histories along the coast of Chile. Species with
planktonic larvae showed a strong positive relationship with temperature, but
those with direct development showed the opposite pattern. This effect of life
history held true both in crustaceans and mollusks (Fernandez et al. 2009). The
authors hypothesized that such a pattern might be explained by the effect of
88 CHAPTER 3
This finding may apply more broadly to terrestrial taxa. A comprehensive liter-
ature review concluded that measures of ambient thermal energy, water, or water-
energy balance explained spatial variation in richness better than other climatic
and nonclimatic variables in 82 of 85 cases (Hawkins et al. 2003). Even when
considered in isolation, water and energy variables explained on average over
60% of the variation in the richness of a wide range of plant and animal groups
(Hawkins et al. 2003). As discussed earlier, however, this strong link to climatic
variables is most prominent at large scales, particularly for mammals and birds,
for which diversity is more likely to relate to local productivity at smaller spatial
scales (Belmaker and Jetz 2011). This suggests that temperature is more important
in driving regional, but less so local, richness in these taxa (Belmaker and Jetz
2011; see fig. 3.2).
The strong convergence between the richness patterns of vertebrate consumers
and plants could be causal as well, such that plant richness is driven by climatic
variables, but animal richness relates to the number of plant species (that is, diver-
sity begets more diversity; Brown 2014). While there is good evidence for such
a direct effect at small scales (in individual habitats), it has been shown that at
large scale vertebrates and plants are both driven by climatic variables, rather than
causally related to each other (Jetz et al. 2009).
While much of the terrestrial research has focused on vertebrates and plants,
newer studies include less conspicuous taxa such as microorganisms and fungi.
A study of soil samples from North and South America concluded that micro-
bial diversity was mostly related to soil pH, with maximum diversity at neutral
pH (7 to 8), and declining diversity at higher or lower pH (Fierer and Jackson
2006). Fungi are another poorly known group, despite the fact that they may be
one of the most species-rich taxa on land. When analyzing the diversity of soil
fungi across all continents, it appeared that species richness was strongly linked
to moisture and soil chemistry, particularly the availability of calcium (Tedersoo
et al. 2014). In contrast to animal taxa, richness of all fungi and functional groups
was unrelated to plant diversity, with the exception of ectomycorrhizal root sym-
bionts. This suggests that plant-soil feedbacks do not influence the diversity of
soil fungi at the global scale (Tedersoo et al. 2014).
Freshwater taxa offer another interesting contrast because their habitats are
embedded in the terrestrial sphere but are not moisture-limited. Across freshwater
vertebrates (mammals, birds, amphibians, and fish) and crayfish, ambient energy
(temperature and solar radiation) consistently explained a majority of variation
in total species richness and endemism (Tisseuil et al. 2013). Habitat area and
historical factors related to dispersal explained much of the remaining varia-
tion. The modeled response to changes in ambient energy was stronger than for
other factors and near-linear (Tisseuil et al. 2013). This shows that the diversity
90 CHAPTER 3
of freshwater taxa appears to be driven by similar factors as land taxa, with the
exception of moisture.
In conclusion, it appears that the diversity of life on land is primarily related
to the availability of ambient energy, measured as temperature or PET, in com-
bination with sufficient moisture, habitat area, and complexity. Thus, there is a
fundamental similarity between primary correlates of diversity on land and in the
sea, considering similar primacy of ambient energy and habitat as leading predic-
tors of species richness. Beyond this, there is the unique dependence on ambient
moisture that characterizes life on land, where water is often limiting, and the
greater oxygen constraints in the ocean. Of course, these factors have all changed
over time—for example, between glacial and interglacial cycles. Analyzing both
historic and present-day conditions, it was shown that a combination of current
temperature and historically integrated habitat area and productivity explained
up to 87% of observed variation in terrestrial vertebrate richness (Jetz and Fine
2012). Similar conclusions were reached for marine species, particularly the well-
studied mollusks (Valentine and Jablonski 2015), and for vascular plant richness
on islands, which carries a large signal of historical climate and area changes,
especially for endemic species (Weigelt et al. 2016).
3.4. SYNTHESIS
Three major classes of hypotheses seek to explain species richness patterns on our
planet. These relate to (1) factors that may promote high diversity by increasing
speciation rates or the density of individuals and coexisting species; (2) factors
that may limit diversity by increasing extinction rates, or by limiting individual
density or niche space; and (3) factors that describe the role played by habitat area
and complexity. Of these, theories relating diversity to environmental gradients
in temperature, productivity, and habitat area have over time developed the most
explicit mechanistic foundations.
Empirically, we also find that these three environmental predictors gather the
most support across realms and species groups (table 3.6). When we combined
the available information on primary and secondary environmental predictors
analyzed at global or near-global scales (see table 3.6), some interesting patterns
emerged. Across all 36 studied species groups discussed here, ambient tempera-
ture (or PET, which is closely related to temperature) was by far the most con-
sistent primary predictor of species richness (72% of studies), whereas habitat
area (6%) and productivity (14%) were less commonly identified as primary pre-
dictors. Habitat area emerged as the most important secondary predictor in 39%
of cases, with temperature (14%) and productivity (14%) gathering less support.
Table 3.6. Environmental Predictors of Species Richness
Secondary
Habitat Species group Primary predictor predictor Source
Coastal Stony corals Temperature Habitat area Tittensor et al. 2010
Coastal Brittle stars Temperature Productivity Woolley et al. 2016
Coastal Bivalves Temperature Productivity Valentine and Jablonski 2015
Coastal Cone snails Temperature Habitat area This volume, table 3.1
Coastal Cephalopods Temperature Habitat area Tittensor et al. 2010
Coastal Benthic Temperature Chlorophyll (−) Macpherson 2002
macrofauna
Coastal Seagrasses Temperature Habitat area Tittensor et al. 2010
Coastal Mangroves Temperature Habitat area Tittensor et al. 2010
Coastal Macroalgae Temperature Nutrients Keith et al. 2014
Coastal Fish Temperature Habitat area Tittensor et al. 2010
Coastal Sharks Habitat area Temperature Tittensor et al. 2010
Coastal Pinnipeds Temperature (−) Productivity Tittensor et al. 2010
Pelagic Foraminifera Temperature Habitat area Rutherford et al. 1999;
Tittensor et al. 2010
Pelagic Copepods Temperature Salinity Rombouts et al. 2009
Pelagic Euphausiids Temperature Habitat area Tittensor et al. 2010
Pelagic Squids Temperature Habitat area Tittensor et al. 2010
Pelagic Macrofauna Nitrate Temperature Macpherson 2002
Pelagic Tuna and Temperature Habitat area Worm et al. 2005; Tittensor
billfish et al. 2010
Pelagic Sharks Temperature Habitat area Tittensor et al. 2010
Pelagic Cetaceans Productivity Temperature Tittensor et al. 2010
Pelagic Seabirds Habitat area Temperature (+/−) Davies et al. 2010
Pelagic Bacteria Temperature — Pommier et al. 2007;
Fuhrman et al. 2008
Pelagic Microbes Temperature — Sunagawa et al. 2015
Deep sea Brittle stars Export productivity (+/−) Oxygen stress Woolley et al. 2016
Deep sea Gastropods Export productivity (+/−) Depth Tittensor et al. 2011
Deep sea Bivalves Export productivity (+/−) Temperature Tittensor et al. 2011
Deep sea Foraminifera Export productivity — Culver and Buzas 2000
Land Vascular Potential Wet days Kreft and Jetz 2007
plants evapotranspiration
Land Soil fungi Moisture Soil calcium Tedersoo et al. 2014
Land Soil microbes pH (+/−) — Fierer and Jackson 2006
Land Amphibians Temperature Productivity Belmaker and Jetz 2011
Land Birds Temperature Productivity Belmaker and Jetz 2011
Land Mammals Temperature Elevational range Belmaker and Jetz 2011
Land Vertebrates Temperature Habitat area Jetz and Fine 2012
Freshwater Vertebrates Temperature Habitat area Tisseuil et al. 2013
Freshwater Crayfish Temperature Habitat area Tisseuil et al. 2013
Secondary
Predictor Main predictor predictor
Temperature 26 5
Habitat area 2 14
Productivity 5 5
Other 3 8
Note: All effects are positive, except where indicated: (−) denotes a negative correlation with richness, (+/−) a
unimodal one. Cumulative evidence for the three most common predictors is summarized at the bottom of the
table. All publications included here were global in scope and data coverage, except for five studies that were
at the scale of ocean basins (Culver and Buzas 2000; MacPherson 2002; Rombouts et al. 2009; Tittensor et al.
2011) or continents (Fierer and Jackson 2006).
92 CHAPTER 3
Developing a Theory
of Global Biodiversity
for an “efficient theory,” as defined by Marquet et al. (2014)—that is, a theory that
is grounded in first principles and mathematical expressions, makes few assump-
tions, and generates a large number of predictions per free parameter, enabling the
testing of model predictions against multiple empirical datasets.
The building blocks for our theory of global biodiversity are based on the empir-
ical evidence assembled and analyzed in the previous chapters. From our review
of the existing literature on marine and terrestrial biodiversity patterns, we surmise
that the primary forces that structure large-scale patterns of species richness are
most likely related to some combination of ambient thermal energy, habitat area,
or productivity. The effects of thermal energy can be captured through metabolic
equations relating rates of community turnover and evolution (that is, evolution-
ary speed) to temperature. The effects of habitat area and productivity can both be
captured through equations linking species richness to community size (that is, the
number of individuals in a community). The equations describing these theoretical
building blocks and resultant models are outlined in the rest of this chapter.
We begin with a simple, spatially explicit null-model built on Hubbell’s well-
known metacommunity model that implements his neutral theory of biodiversity
and biogeography (Hubbell 2001). This model was chosen as a base because it
incorporates both evolutionary processes of speciation and extinction and eco-
logical processes of dispersal and disturbance, all of which influence species rich-
ness (Vellend 2010; Tittensor and Worm 2016). Like its predecessor, the theory of
island biogeography (MacArthur and Wilson 1967), this neutral model assumes
ecological equivalence among species. This assumption is of course an unreal-
istic abstraction of nature. Yet neutral theories have performed reasonably well
in predicting some major macroecological properties such as patterns of relative
abundance, range size, and diversity, although many open questions remain (Bell
2001; Rosindell et al. 2012; Marquet et al. 2014). This suggests at least the possi-
bility that species-specific differences may not always be critical in driving major
macroecological patterns. We evaluate this proposition with respect to observed
global patterns of species richness across taxa. Subsequently, we relax that
assumption by departing from neutrality and allowing species within our spatially
explicit model to differ in their niche width and thermal tolerances, accounting
for observed differences among species in nature (Sunday et al. 2011; Beaugrand
2014). In this way, we are combining elements of neutral theory, metabolic theory,
and niche theory into a synthetic theory of global biodiversity.
BOX 4.1
Analytical Solution for a Spatially Explicit Neutral Model
where J is the size of the local community, and Ni is the number of individu-
als in species i (Hubbell 2001).
96 CHAPTER 4
Two fundamental principles underlie the basic model implementation. The first
principle states that all processes that are parameterized (disturbance, dispersal,
and speciation) are modeled as per capita rates, such that all of these processes
have a given probability per individual. Note that species extinction is not param-
eterized, but emerges from the model due to the random dynamics of births and
deaths. All individuals regardless of species are assumed ecological equivalents—
that is, none are more likely to be disturbed, disperse, or speciate than any other,
and are selected at random to undergo these processes. We also assume zero-sum
ecological dynamics, meaning that there are a fixed number of individuals that can
be supported by a given environment, and while there is constant disturbance rate
and associated death and turnover of such individuals, their total number does not
change. This type of behavior applies only to ecologically similar species at one
trophic level—for example, metacommunities of trees, corals, herbivorous mam-
mals, or planktivorous fish—which are likely to be competing with one another for
resources such as space (Hubbell 2001). However, it is possible to envision over-
laying multiple models for different trophic groups atop one another, in an attempt
to determine broader biodiversity patterns, and perhaps to examine interactions
between these groups. We will return to this point later in this book.
The second fundamental principle is that environmental parameters thought
to influence and shape diversity patterns, such as ambient temperature, or habitat
area, are added in stepwise fashion to the base model. The expanded model is still
purely neutral within each local community, with per capita rates being identical
among all individuals, but now includes spatially heterogeneous environmental
factors across the metacommunity.
Our initial model begins from a local community that is composed of a number
of sites (fig. 4.1), and has no connection to a metacommunity. Each of these sites
is occupied by a single individual of a designated species. Local community size J
is constant through time. The local community is initially seeded at random from
a pool of S species, where each species has an equal chance of occupying each
site, in accordance with the principle of neutrality. It was previously shown that
the equilibrium community of this type of model was not sensitive to the initial
number of species (Hubbell 2001). At each time step, one randomly selected site
is disturbed, and the individual occupying that site dies and is replaced at random,
where the probability of a particular species filling that site is proportional to
its local abundance in that same community. Hence, species that attain a higher
abundance in the community are more likely to recolonize disturbed sites in the
next time step. Because of this positive feedback, the community converges to a
monoculture (S = 1) over time. This result is well established for neutral models
without immigration (Hubbell 2001).
Next, we add immigration to the model by linking the local community to a
global, spatially explicit, metacommunity M, allowing for dispersal between local
98 CHAPTER 4
Figure 4.1. Basic mechanics of a neutral model community. A local community is occupied by
J individuals belonging to S different species. A randomly selected individual dies at each time
step (square symbol) and is replaced at random with an individual born in the same community
(arrow symbol) or with an individual immigrating from a neighboring community (lightning
symbol). The probability of immigration is given by the dispersal parameter m. The full model
also allows for dispersing individuals to speciate as they colonize a neighboring community
(presumably by point mutation and reproductive isolation). The probability is given by the spe-
ciation rate v. In a neutral community, the per capita rates of birth, death, dispersal, and specia-
tion are assumed to be equal among all individuals, and subject only to random variation or drift.
Redrawn after Rosindell et al. (2011).
Small area
Cold
Figure 4.2. Global metacommunity model. Our spatially explicit global metacommunity con-
sists of 21 × 21 local communities, each occupied by J individuals (with local community size
J = 16 and metacommunity size JM = 7056 as the base case). Shades in local communities sym-
bolize different species. Communities are connected through dispersal (at rate m) to neighboring
cells (arrows). Although the grid is depicted as flat, the left-hand and right-hand borders are con-
nected to form a cylindrical shape. The dashed line in the center represents the equator, and the
top and bottom of the main grid represent the poles. Disturbance and replacement of individuals
on the grid is a per capita function that is constant for all local communities (in the neutral base
model) or can become a function of temperature (in the combined neutral-metabolic model),
with increased turnover rates or speciation rates at higher temperature. The effect of increasing
habitat area or productivity can be simulated by scaling the number of individuals J in any spe-
cific local community up or down. Redrawn after Tittensor and Worm (2016).
in the local community in the second. At each time step, an individual can move
only to immediately neighboring communities. Ultimately, however, it can spread
through the global community via successive Moore neighborhoods (Bell 2001),
with faster dispersal through the metacommunity at higher values of m.
In this simple metacommunity model, species richness declines over time to
monodominance, due to the random processes outlined earlier, albeit more slowly
than for a single community. We find that even a low dispersal rate (m = 0.01)
introduces a spatial rescue effect that slows the spread of local monocultures (and
decelerates global extinction). Higher dispersal rates initially lead to higher aver-
age local species richness in each community, but lower global richness across the
metacommunity, as found by Hubbell (2001, 218). Ultimately, however, without
the input of any new species to the system, it will eventually settle to a monodom-
inant system. Because species are ecologically identical, the species that comes to
dominate cannot be predicted a priori.
Next, we introduce speciation into the model, which has a chance of occur-
ring when a new site is colonized by a dispersing propagule. Thus, every specia-
tion event corresponds to a dispersal event: dispersal occurs and then speciation is
determined independently. Speciation is assumed to occur as a point mutation with
a per capita probability of v. Although other speciation mechanisms have been pro-
posed (Rosindell et al. 2010), we use this simplest possible form in our first attempt
at a simple process-based model. Experimenting with different forms of speciation
remains one of many aspects that could be tested within the model framework.
Neutral theory characterizes the (sampling) distribution of species abundances
in a metacommunity of size JM and speciation rate v by a composite parameter
known as the fundamental biodiversity number,
ν
θ = JM
1− ν
1. Dispersal, mortality, and speciation rates are constant and individuals are
ecologically equivalent (identical per capita rates regardless of species
identity).
2. The number of vacant sites opening up in a community is proportional to
the disturbance rate, which is equivalent for each local community, as one
random site in the metacommunity is disturbed per time step.
3. The probability of a species filling a single vacancy is equal to its relative
abundance in the local community (if local dispersal) or Moore neighbor-
hood (if metacommunity dispersal).
4. A finite amount of space is assumed in each local community (J) with a
constant number of individuals in the metacommunity, JM, competing for
the space across all local communities.
This model differs from the spatial model described in Hubbell (2001) and else-
where only in the implementation, in which a single community is disturbed each
time step, rather than all communities being disturbed in each time step. This
produces identical results to these other model implementations, but will become
important later, when additional components are added to the model, and assump-
tions of neutrality are relaxed.
A B
100 100
Log 10 (% relative abundance)
0.1 0.1
0.01 0.01
Dispersal = 0.5 Dispersal = 0.5
0.001 0.001
50 100 150 50 100 150
Rank Rank
Figure 4.3. Comparison of the global metacommunity model with original results by Hubbell
(2001). Rank-abundance plots are depicted. (A) Model results for a metacommunity of 7056
individuals with fundamental biodiversity number = 10 simulated in coalescence mode (see
text). Solid lines indicate mean values, and the gray-shaded area indicates 95% range of 100 rep-
licate runs. Dashed black lines indicate values from fig. 7.7 of Hubbell (2001, 219) at the same
parameter settings. (B) Same, but solid lines represent results from forward-simulation mode,
run to 25,000 community turnovers, for approximately 1.7 billion time steps. Both forward and
coalescence modes produce results identical to Hubbel (2001).
as a function of internal organism temperature and body size (see chapter 3).
In the context of our neutral model, we can ignore the effect of body mass, as
all organisms are assumed ecological equivalents. The temperature dependence
is described by the Boltzmann factor e−E/kT, where E is the empirically derived
activation energy (~0.65 eV for ectotherms, as stated by Brown et al. 2004), k is
Boltzmann’s constant (8.617 × 10 −5 eV K−1), and T is absolute internal temperature
in degrees Kelvin (K). This relationship predicts about a doubling of metabolic
rate with every 10 K increase in temperature. We assumed a 30 K gradient in aver-
age surface temperature between the equator and each pole, approximating the
observed gradients in sea surface temperature in today’s oceans (fig. 4.4A). This
realistic temperature gradient results in an approximately eightfold difference
between average metabolic rates from the equator to the poles for ectotherms.
This process was simulated in the model by making the probability of disturbance
at each site in a local community proportional to the rate difference (Tittensor and
Worm 2016). Therefore, a site located in an equatorial community had around an
eightfold higher chance of being disturbed in any individual time step than one at
the poles, with intermediate latitude cells following a gradient between the two.
We always normalized the total probabilities to sum to one, to enable the simple
random selection of a cell for disturbance.
When we implemented the specified 30 K latitudinal temperature gradient
along with its predicted effect on community turnover in the forward model,
we observed an unstable latitudinal gradient in mean local community species
richness (fig. 4.5). Faster turnover and shorter generation times in the tropics
resulted in a more rapid approach toward the dynamic equilibrium richness state
A B
25
Ocean area (million km2)
1.5
Mean temperature (°C)
20
15 1.0
10
0.5
5
0
0.0
−50 0 50 −50 0 50
Latitude Latitude
Figure 4.4. Observed global gradients in sea surface temperature and total ocean area by latitude.
(A) Surface temperature and (B) ocean area both peak around the equator and decrease sharply
toward the poles. The regional peak of ocean area in the Arctic partly reflects areas permanently
covered by ice. Redrawn after data in Tittensor et al. (2010); Allen and Gillooly (2006).
A B I J
10 10
8 8
Local richness
Local richness
Latitude
Latitude
6 6
4 4
2 2
0 5 10 0 5 10
Local richness Local richness
C D K L
10 10
8 8
Local richness
Local richness
Latitude
Latitude
6 6
4 4
2 2
0 5 10 0 5 10
Local richness Local richness
E F M N
10 10
8 8
Local richness
Latitude
Local richness
Latitude
6 6
4 4
2 2
0 5 10 0 5 10
Local richness Local richness
G H O P
10 10
8 8
Local richness
Latitude
Local richness
Latitude
6 6
4 4
2 2
0 5 10 0 5 10
Local richness Local richness
Figure 4.5. A transient biodiversity gradient resulting from increased community turnover at
higher ambient temperature. Left-hand column depicts a global metacommunity grid of 21 × 21
communities with shading indicating their species richness S. Right-hand column depicts the
resulting latitudinal gradient, with the solid line indicating the mean, and shading representing
the 95% range of 100 model runs. Shown is the appearance of a transient latitudinal gradient
in species richness from a forward-mode neutral-metabolic model after (A, B) 1 community
turnover (7056 time steps); (C, D) 10 community turnovers; (E, F) 50 community turnovers; (G,
H) 100 community turnovers; (I, J) 125 community turnovers; (K, L) 250 community turnovers;
and (M, N) its disappearance within 1000 community turnovers. Coalescence model results
(O, P) are shown for comparison. Dispersal parameter m = 0.1; speciation rate v = 0.01; global
metacommunity size Jm = 7056 individuals. After Tittensor and Worm (2016).
D E V E L O P I N G A T H E O RY 107
there, elevating tropical relative to polar richness. Eventually, however, the polar
regions did catch up and reached the same asymptotic richness and relative spe-
cies abundance distribution as equatorial areas (fig. 4.5M,N). The coalescence-
mode simulation (fig. 4.5O,P) confirmed that no gradient existed at equilibrium.
This partly resulted from the lack of any stabilizing force to maintain the asymp-
totic gradient of equilibrium richness—it is simply a matter of time—and partly
from the initial richness being lower than equilibrium richness. This latter point
can be demonstrated by starting a forward-mode model with local richness values
set to be higher than the mean equilibrium value, in which case the more rapid
approach of equatorial regions toward the equilibrium value produces a transient
inverse latitudinal gradient (results not shown). We note that it actually remains an
open question as to whether Earth has reached an equilibrium in terms of global
biodiversity, a fact that is deemed unlikely given the continuing increase of fossil
biodiversity up to the present, as well as the large-scale disturbance events that
“reset” biodiversity at regional and sometimes global scales (Raup and Sepkoski
1982; Benton and Twitchett 2003; Alroy et al. 2008).
Metabolic theory also invokes an effect of temperature on speciation rates
independent of the previously described effect on turnover times. As discussed
in chapter 3, there is empirical evidence that mutation rates increase as a function
of temperature (Muller 1928; Lindgren 1972; Matsuba et al. 2013). This may be
mechanistically explained by the increased production of free oxygen radicals at
higher metabolic rates, which increases nucleotide substitution rates, and may
ultimately drive up speciation rates (Allen et al. 2006). However, we note that the
mechanisms that link nucleotide substitution to speciation are not entirely clear.
Despite this gap in mechanistic understanding, increased speciation rates during
periods of climate change and increased temperature have also been evident in
the fossil record (Stehli et al. 1967; Sepkoski 1998; Mayhew et al. 2012). We
simulated such thermal effects on speciation rates in our model by setting the
speciation rate, v, to be a normalized function of temperature:
νbase e −E/kT j
νj = E ,
min (e − )kTj
where base is the “basic” speciation rate at the lowest temperature, and the meta-
bolic effect of temperature, e E/kT , is normalized by dividing by its minimum
j
value. Thus, at the poles the speciation probability equals base, and it increases
exponentially toward the equator as predicted by metabolic theory (Brown et al.
2004). This effect can be simulated independently of or in combination with the
effect of temperature on turnover rates.
We implemented this effect to explore under which conditions a permanent
richness gradient emerges (fig. 4.6). As previously discussed, no effect was seen
in the base model (fig. 4.6A,B), or when modeling the effects of temperature on
A B
20
15
Local richness
Latitude
10
0 20 40
Local richness
C D
20
15
Local richness
Latitude
10
0 20 40
Local richness
E F
20
15
Local richness
Latitude
10
0 20 40
Local richness
G H
20
15
Local richness
Latitude
10
0 20 40
Local richness
Figure 4.6. Emergence of permanent latitudinal biodiversity gradient. Shown are results from
100 coalescence model runs each: (A, B) in a neutral base model; (C, D) including a thermal
effect on community turnover; (E, F) including thermal effect on speciation rates; (G, H) includ-
ing a gradient of increasing habitat area from the tropics to the poles. Symbols and base param-
eter settings as in fig. 4.5. After Tittensor and Worm (2016).
D E V E L O P I N G A T H E O RY 109
15
Local richness
Latitude
10
0 20 40
Local richness
C D
20
15
Local richness
Latitude
10
0 20 40
Local richness
E F
20
15
Local richness
Latitude
10
0 20 40
Local richness
G H
20
15
Local richness
Latitude
10
0 20 40
Local richness
Figure 4.7. Combined effects of area and temperature gradients on latitudinal richness: (A, B)
including effect of temperature on both community turnover rate and speciation rate; (C, D) includ-
ing an area effect, and a temperature effect on community turnover rate; (E, F) including an area
effect, and a temperature effect on speciation rate; (G, H) including an area effect, and a temperature
effect on both community turnover rate and speciation rate. Symbols and base parameter settings as
in fig. 4.5. Thermal effects are parameterized as per the metabolic theory of ecology. Area and tem-
perature gradients approximate those found in the global oceans. After Tittensor and Worm (2016).
D E V E L O P I N G A T H E O RY 111
combining turnover and speciation and area effects (fig. 4.7G,H). We conclude
that gradients in speciation rate and community size, but not turnover, produce
persistent gradients in species richness. The effects of area and speciation can act
synergistically, with model results roughly equally sensitive to variation in both
parameters. Note that we assume here that increases in area-specific productivity
may produce similar results as increases in habitat area (Wright 1983), as both
are increasing local community size; we do not further distinguish between them
here, but contrast them when fitting to data in chapter 5, accounting for the fact
that they can show different scaling relationships with species richness in nature
(Hurlbert and Jetz 2010).
We explored the sensitivity of metacommunity properties to variation in specia-
tion rate v and dispersal parameter m for model runs with thermal gradient effects
on both speciation and turnover (fig. 4.8). Mean local community species rich-
ness increased both with increasing metacommunity dispersal and with increasing
speciation rates (fig. 4.8A). Clearly, higher speciation rates led to the evolution of
more species, whereas higher dispersal rates allowed those species to spread over
more local communities and avoid the higher extinction probability that plagues
small isolated populations. Mean global metacommunity richness showed an even
more pronounced increase with increasing speciation rate (fig. 4.8B), reflecting a
more positive net balance between the probabilities of speciation and extinction.
The effects of changes in the dispersal parameter m on global richness, however,
were minimal, acting only to slightly increase the total richness at low dispersal
values, probably owing to the greater “regionalization” of communities under low
dispersal regimes, resulting in higher beta-diversity. When assessing the effect of
both parameters on the strength of the latitudinal biodiversity gradient (fig. 4.8C),
we saw an interaction between the effects of speciation rate and dispersal. While
increasing speciation rate increased the strength of the latitudinal richness gradient
across most tested dispersal parameter values, this effect was more pronounced at
low dispersal rates than at high dispersal rates (see fig. 4.6). Dispersal had a slightly
positive effect on gradient strength at low speciation rates, but a negative effect at
the highest speciation rates. Note here that not all of these parameter combinations
might occur in nature—for example, high dispersal rates might prevent speciation
by undermining reproductive isolation, and thus high dispersal rates would not
necessarily co-occur with high speciation rates. As a final sensitivity check, we
explored the effects of changes in community size, and found that increasing com-
munity size to J = 64 (JM = 28,244) individuals increased local and global richness
at equilibrium but did not affect any of the latitudinal patterns we documented for
smaller community sizes. While such parameter space exploration proved insight-
ful, we acknowledge that the relationship to actual rates of dispersal and specia-
tion, as well as actual community sizes found in nature, is not fully resolved and
should be explored in future work.
Local species richness
16
14
15 12
Richness
10 10
8
5
6
1 0.1
Di 0.1 0.01 es 4
sp 0.01 rat
ers 0.001
al 0.001 1e−04 tion 2
rat ia
es ec
Sp
4000
4000
3000
3000
Richness
2000
1000 2000
1 0.1
Di 0.1 0.01 es 1000
sp rat
ers 0.01 0.001
al 0.001 1e−04 tion
rat ia
es ec
Sp 0
10
10 8
Delta richness
8
6
6
4
2
4
1 0.1
Di 0.1 0.01 es
sp rat 2
ers 0.01 0.001
al 0.001 1e−04 tion
ra ia
tes ec 0
Sp
Figure 4.8. Parameter space exploration of the neutral-metabolic model. Community proper-
ties shown as a function of variation in dispersal and speciation rate parameters. (A) Effects on
mean local community species richness. (B) Effects on global metacommunity species richness.
(C) Effects on the strength of the latitudinal gradient, represented as the difference between
mean equatorial and mean polar richness. Shown are mean values of 100 model runs. After Tit-
tensor and Worm (2016).
D E V E L O P I N G A T H E O RY 11 3
A final step in the development of this model framework introduces thermal toler-
ances and niches, and hence represents a departure from the strictly neutral model
in that we can now assume ecological differences among species. In a sense, we
are testing the sensitivity of previously described model results to the (very unre-
alistic) assumption of ecological equivalence by relaxing this assumption and
evaluating the effect of thermal niches on observed patterns. For some marine
species, it has been shown that we can reconstruct latitudinal ranges (Sunday et
al. 2012) and some global diversity patterns (Boyce et al. 2008; Beaugrand et al.
2013) from individual species’ thermal tolerances. Temperature niches have also
been used extensively in habitat models and were invoked in a general model
of latitudinal diversity, primarily applied to pelagic organisms (Beaugrand et al.
2013; Beaugrand 2014). Hence, we explored the effects of temperature niches on
latitudinal patterns in our model.
We assumed that all species have the same niche width (w), but different mid-
points in their thermal niche. The midpoint is given by the ambient temperature
where the species evolved. For example, assuming a ±5 K niche width, a species
that evolved in a 25°C environment is assumed to tolerate from 20 to 30°C ambi-
ent temperature. A species that evolved at 5°C may tolerate 0 to 10°C. Hence, spe-
cies are still assumed equivalent in that they all have identical thermal tolerance
ranges, but they also feature individual temperature niches in that they are adapted
to different optimal temperatures. Within each local community, species remain
equivalent, but within the metacommunity, they differ in their ability to disperse
to regions of different temperature. Species are generally assumed to be unable to
disperse outside their thermal niche. For phytoplankton, these assumptions have
detailed empirical support. As shown from an analysis of all available data on
temperature tolerance of phytoplankton strains, as well as from a mechanistic
model, the temperature for optimum growth is closely correlated with the mean
ambient temperature where the species is found (Thomas et al. 2012). Similar
results have been described for reef fish and invertebrates (Stuart-Smith et al.
2015). Furthermore, the width of the temperature niche for phytoplankton spe-
cies was independent of the mean ambient temperature of the environment where
the species occurred (Thomas et al. 2012). These empirical findings support our
simplified assumption of constant niche width. Finally, the maximum growth rate
of individual phytoplankton strains increased with mean ambient temperature,
indicating that adaptation to different temperature regimes does not override the
general metabolic relationship between temperature and growth (Thomas et al.
2012). This means that high-latitude species have similar temperature niches as
low-latitude ones, but with lower optimum temperature and lower realized growth
11 4 CHAPTER 4
50 50
5 5
1 1
5 10 15 20 5 10 15 20
50 50
5 5
1 1
5 10 15 20 5 10 15 20
50 50
5 5
1 1
5 10 15 20 5 10 15 20
Latitudinal range Latitudinal range
Figure 4.9. Neutral-metabolic-niche model under moderate dispersal. Shown are species lati-
tudinal range distributions at dispersal rate m = 0.1, speciation rate v = 0.01, and under different
assumed thermal niche width ranging from ±3 to ±15 K. Note logarithmic scale on the y-axis.
Few species (up to 5%) colonize large ranges, irrespective of their thermal niche width. Note
that a thermal niche width of ±15 K is nonrestrictive given a total 30 K temperature gradient, and
produced identical results to a model run with niche parameter turned off.
realized only when there was no dispersal limitation. These results were upheld
at larger community sizes, and whether we allowed niches to evolve over time or
forced them to remain fixed (results not shown).
When we further explored the effects of niches on diversity patterns across
parameter space, we found that niches can have a positive effect (~50% increase)
on local diversity (fig. 4.11), due to greater regionalization. However, this was
seen only at the highest dispersal rates, and not at lower dispersal rates, where
high regionalization exists regardless of niche settings. Little effect was seen on
global metacommunity richness, but there was an effect on the strength of the lati-
tudinal gradient at high dispersal rates (see fig. 4.11). Niches more than doubled
the steepness of the latitudinal gradient relative to the non-niche case at high
dispersal (m = 1), but not at low or moderate dispersal rates. These effects on lati-
tudinal gradient strength were seen particularly at smaller niche width (w = ±3 K),
but less so at larger niche widths (w = ±9 K, w = ±12 K) (fig. 4.12).
These results make sense when considering latitudinal range size distributions
at different niche widths w and dispersal rates m (see figs. 4.9 and 4.10). Wide-
ranging species can “flatten” the latitudinal gradient, but only when they contrib-
ute a substantial proportion of overall species richness, which occurs only at high
dispersal rates (see fig. 4.10). Under these circumstances, niches are effective
in limiting the latitudinal spread of species, and hence in maintaining a steeper
11 6 CHAPTER 4
50 50
5 5
1 1
5 10 15 20 5 10 15 20
50 50
5 5
1 1
5 10 15 20 5 10 15 20
50 50
5 5
1 1
5 10 15 20 5 10 15 20
Latitudinal range Latitudinal range
Figure 4.10. Neutral-metabolic-niche model under high dispersal. Shown are species latitudinal
range distributions at maximum dispersal rate m = 1, speciation rate v = 0.01, and under different
assumed thermal niche width ranging from ±3 to ±15 K. Under assumed maximum dispersal,
a larger fraction of species (up to 30%) colonizes large ranges, with higher fractions at larger
thermal niche width.
richness gradient. It is striking however, how little effect niches had at medium to
low dispersal rates (figs. 4.9 to 4.12). It is interesting in this context that most of
the empirical (Boyce et al. 2008; Thomas et al. 2012) and theoretical work (Beau-
grand et al. 2013; Beaugrand 2014) on temperature niches affecting global diver-
sity patterns has been conducted for pelagic communities that are typically very
mobile, and have little dispersal limitation. Empirically, we see clear evidence
that pelagic organisms tend to have larger ranges than their coastal counterparts
(see fig. 4.13, for example). This could potentially affect their patterns of species
richness, and partly explain the broad peaks of tropical to subtropical richness we
observe in many pelagic taxa (see chapter 2).
When we check these results with our forward-simulation model, we find very
similar patterns, but in addition we observe some transient effects of niches under
most parameter combinations. Specifically, tropical richness was elevated when
(restrictive) niches were present for 100 to 250 complete metacommunity turn-
overs, and a notable gradient appears after about 500 turnovers. Yet, over time,
these effects disappeared, and patterns at equilibrium were near-identical to the
case with no niches (results not shown).
At this point, we should clarify that of course niches can have large and impor-
tant effects on individual species, their distribution, local abundance, and extinc-
tion risk (fig. 4.14). They also have clear effects on community composition and
species identity, and may keep many species bound at the latitude at which they
16
A 14
15 12
10
Richness
10
8
5
6
1 0.1
Di 0.1 0.01 s 4
sp e
ers 0.01
al 0.001 rat
rat 0.001 tion
es 1e.04 cia 2
e
Sp
B
4,000
4,000
3,000
Richness
3,000
2,000
1,000 2,000
1 0.1
Di
sp 0.1 0.01 s 1,000
ers 0.01 ate
al
rat 0.001
0.001
ionr
es 1e.04 ciat
e
Sp 0
12
C
10
10 8
Delta richness
8
6
6
4
2
4
1 0.1
Di 0.1 0.01 s
sp e 2
ers 0.01
al 0.001 rat
rat 0.001 tion
es 1e.04 cia
e
Sp 0
Figure 4.11. Parameter space exploration of the neutral-metabolic-niche model with assumed
thermal niche width ±3 K. (A) Effects of different speciation and dispersal rates on on mean local
community species richness. (B) Effects on global metacommunity species richness. (C) Effects
on the strength of the latitudinal gradient, represented as the difference between mean equatorial
and mean polar richness. Comparison with neutral-metabolic model without temperature niches
(see fig. 4.8) reveals only minor differences.
11 8 CHAPTER 4
14 14
3 degree niche
12 12
Local richness
Local richness
10 10
Latitude
Latitude
8 8
6 6
4 4
2 2
0.0 7.5 15.0 0.0 7.5 15.0
Mean local Mean local
richness richness
14 14
9 degree niche
12 12
Local richness
Local richness
10 10
Latitude
Latitude
8 8
6 6
4 4
2 2
0.0 7.5 15.0 0.0 7.5 15.0
Mean local Mean local
richness richness
14 14
12 degree niche
12
Local richness
12
Local richness
10
Latitude
Latitude
10
8 8
6 6
4 4
2 2
0.0 7.5 15.0 0.0 7.5 15.0
Mean local Mean local
richness richness
Figure 4.12. Effects of changes in thermal niche width and dispersal rate on global gradients in
species richness. Shown are metacommunities with the same 30 K temperature gradient and the
base model speciation rate v = 0.01. At moderate dispersal rate (m = 0.1), thermal niches are not
affecting gradients of species richness, whereas high dispersal leads to higher richness overall,
flatter gradients, and more pronounced latitudinal gradients at reduced niche widths.
originated, as seen in the fossil record (Jablonski et al. 2013). From the macro-
ecological perspective, however, and abstracting from species identities, niches
appeared to have limited effect on the strength of global richness gradients, except
at very high dispersal. Our model results suggest that even moderate dispersal
constraints will have a similar effect in constraining most (but certainly not all)
species to their original habitat (see fig. 4.9).
With respect to species’ ranges, we also found that the distribution patterns
that emerged from the model tended to support Rapoport’s rule, which states that
the latitudinal ranges of plants and animals are generally smaller at lower than at
Shark range sizes
8e+08
6e+08
Km2
4e+08
2e+08
0e+00
Coastal Oceanic
25
20
Degrees C
15
10
0
Coastal Oceanic
25
20
Degrees C
15
10
5
Coastal Oceanic
Figure 4.13. Realized ranges and thermal niches in coastal versus oceanic sharks. Shown are box-
plots of the ranges and temperature requirements of all known shark species, contrasting coastal
and oceanic species. Oceanic sharks have both larger ranges and realized thermal niches, but mean
temperature preference similar to coastal species. After range data from Lucifora et al. (2011).
120 CHAPTER 4
No niches Niches
16 16
Local abundance
Local abundance
12 12
1 community
turnover 8 8
6 6
4 4
2 2
16 16
Local abundance
Local abundance
12 12
10 community
turnovers 8 8
6 6
4 4
2 2
16 16
Local abundance
Local abundance
12 12
50 community
8 8
turnovers 6
6
4 4
2 2
16 16
Local abundance
Local abundance
12 12
100 community
8 8
turnovers
6 6
4 4
2 2
Figure 4.14. Effects of thermal niches on individual species. The distribution of two example
species is shown over time. It is evident how narrow (3K) temperature niches can constrain the
latitudinal spread of a species, its local and global abundance, as well as extinction risk.
No niches Niches
2.0 2.0
Grand mean latitudinal range size
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
5 10 15 20 5 10 15 20
Latitude strip Latitude strip
Figure 4.15. Rapoport’s rule emerging from a neutral model. Average latitudinal range size is
plotted versus latitude in a neutral-metabolic model (A) without niches or (B) with ±3K thermal
niches. Larger ranges consistently emerge at higher latitudes, possibly as a consequence of lower
community richness at those latitudes.
increased numbers of species able to coexist and compete for limiting resources,
and hence limiting the range sizes of those same species relative to their high-
latitude counterparts.
In conclusion, results emerged from our model that suggest that the effects of
temperature on speciation rate and the effects of habitat area and productivity on
community size matter most for the equilibrium species richness gradient. These
mechanisms correspond to the evolutionary speed and more-individuals hypoth-
eses discussed in chapter 3. Temperature effects on species distribution were also
addressed via thermal niches, which had surprisingly little effect on patterns of
species richness at the global scale. Niches appeared to have an effect mostly at
high dispersal rates, as may be found, for example, in pelagic communities and for
widely dispersing species groups such as seabirds, whales, and possibly bacteria.
Clearly, niche-width constraints become influential when species have the ability to
move about widely and could be more far-ranging were they not limited by thermal
constraints. It is under these conditions that we observed effects of niches on the
latitudinal gradient, which became steeper when small niches were imposed at high
dispersal rates (see fig. 4.12). Overall, however, temperature-driven gradients in
speciation rates appeared much more influential than dispersal constraints or niches
in determining global richness patterns, at least through the lens of our model. This
may partly explain why even species with minimal dispersal constraints can show
strong patterns and gradients of biodiversity (Sunagawa et al. 2015).
122 CHAPTER 4
The body of theory that is developed here represents only one of many possible
perspectives on global biodiversity. At one end of the modeling spectrum, purely
statistical models that capture empirical patterns typically perform best in pre-
dicting species distribution and diversity (Kreft and Jetz 2007; Tittensor et al.
2010; Belmaker and Jetz 2011; Tisseuil et al. 2013), but provide little mechanistic
understanding. At the other end, process-based models provide better understand-
ing, but tend to have poorer predictive capacity (Rohde 1992; Beaugrand et al.
2013; Brown 2014).
Here, we have attempted to merge these perspectives by reviewing statistical
associations between observed diversity and environmental drivers (chapters 2 and
3) and hence building a solid empirical basis. We then introduced environmental
factors that had gathered the most empirical support into a spatial metacommunity
model to assess their effects on the evolution of diversity gradients. Our model is
built on, and combines, three of the most prominent ecological theories—namely,
neutral, metabolic, and niche theories. The model that emerged from these efforts
captures both evolutionary (speciation, extinction) and ecological (dispersal and
individual species-environment relationships) dynamics.
Our model may also help to reconcile a long-standing dispute between sup-
porters of niche-based and neutral perspectives. This dispute has been cast as an
example of a more general clash between two philosophical perspectives: realism
and instrumentalism, respectively (Wennekes et al. 2012). In the realist’s view,
a true model will always perform well, precisely because it is true. In contrast,
instrumentalists emphasize the predictive value and other uses of a model; the
literal truth of the model is less of an issue (Wennekes et al. 2012). Hence, in
the niche-neutrality debate, supporters of niche models critique the fact that the
key assumption of ecological equivalence in neutral models is clearly unrealis-
tic, while proponents of neutral models are more interested in their predictive
capacity (Rosindell et al. 2012). Using art as a metaphor, Wennekes et al. (2012)
suggest that neutral theory can provide only broad sketches, while niche theory
provides beautiful detail, but leaves much of the canvas blank. They argue that
a combined approach would be the most useful way to go about “painting the
image,” and that this may also be the only way to end up with a reasonable pic-
ture. Our theory and model attempt to offer such a combined perspective in the
context of a global metacommunity.
In the literature, there is a continuum of purely niche to purely neutral models,
for which empirical support falls somewhere in the middle (Gravel et al. 2006;
Leibold and McPeek 2006; Adler et al. 2007; Gravel et al. 2011). Some mod-
els have attempted to incorporate aspects of both neutral and niche theory, but
D E V E L O P I N G A T H E O RY 123
typically applied at the local community scale and not spatially explicit (Gravel
et al. 2006; Leibold and McPeek 2006; Adler et al. 2007). Such models often
find that neutral dynamics become more important at high diversity and at larger
scales (Chisholm and Pacala 2010), which is relevant with respect to our ques-
tions about large-scale biodiversity patterns.
Empirical studies have likewise found evidence for both neutral processes
(ecological drift) and environmental filtering via niches to explain community
structure. For example, a large-scale study conducted along a meridional transect
of the Atlantic (50 degrees North to 50 degrees South) suggested that phytoplank-
ton communities were slightly more determined by niche segregation (24%), than
by dispersal limitation and ecological drift (17%). Notably, in 60% of the sur-
veyed communities, the assumption of neutrality in species’ abundance distribu-
tions could not be rejected (Chust et al. 2012).
Plankton communities also served as model systems in a previous attempt to
reconstruct global diversity patterns from first principles (Beaugrand et al. 2013).
In that study, hypothetical “pseudospecies” were randomly assembled along a
latitudinal temperature gradient. Purely by chance, more species assembled
toward the center of the temperature domain, in a niche analogy to the spatial
mid-domain effect (see also chapter 3). Accounting for assumed loss of species
at higher latitudes owing to long-term climatic variability (such as ice ages), a
gradient that is broadly representative of those observed in many pelagic groups
could be constructed (Beaugrand et al. 2013; Beaugrand 2014). The question
that emerges from this work is whether niches are indeed necessary to explain
observed diversity patterns. We observed that their effect is mostly small in our
model, because most species tended to have small latitudinal ranges regardless
of assumed niche width constraints (see fig. 4.9). In our fully neutral model, it
appeared that niches were not a major driver of species distributions, except
when there was no other limitation to dispersal. In other words, temperature
niches appeared here to be more a consequence rather than a root cause of the
eco-evolutionary processes that shape global biodiversity patterns. We see these
results as a starting point to dive deeper and to further explore the rich vari-
ety of diversity patterns found in different realms—namely, terrestrial, coastal,
pelagic, and deep-sea environments. This is the topic of our next chapter, where
we attempt to confront our theory with the existing data on species richness pat-
terns in the ocean and on land.
CHAPTER FIVE
(2) oceanographic fronts for pelagic groups, (3) export production for deep-sea
taxa, and (4) water availability on land, all of which influence habitat availabil-
ity or productivity, and hence community size. Empirically, we have found that
these habitat features, in conjunction with temperature, correlate with and help
to predict observed patterns of species richness (chapter 3). Mechanistically, this
is likely the case because greater habitat area or productivity allows for higher
numbers of individuals to coexist in a given habitat, and hence larger commu-
nity size, larger populations sizes, and reduced extinction rates (Wright 1983;
Gaston 2000). Empirical observations strongly support the notion that a larger
area of shoreline supports more individuals (Abele and Patton 1976; Connor et
al. 2000), frontal areas sustain higher densities of pelagic animals (Haney 1986;
Olson et al. 1994), areas of high export production support larger communities
in the deep sea (Haedrich and Rowe 1977; Rex and Etter 2010), and wetter areas
allow for more biomass to develop on land (Silvertown et al. 1994; Kreft and
Jetz 2007), all else being equal. By incorporating these habitat features into a
metacommunity model with realistic spatial structure, and scaling them to the
number of individuals in our model communities, we attempt to reproduce broad
patterns of biodiversity observed in these four major habitats. Much like in the
previous chapter, we do try to keep the model framework as simple and trac-
table as possible, or “instrumental” rather than “realistic” sensu (Wennekes et al.
2012). Time steps, community sizes, and speciation rates as implemented in our
model, for example, remain abstractions and are not directly comparable to the
real world. We also do not employ a formal statistical model selection framework
to optimize model fit to data. Rather, to aid in our understanding, we explore a set
of base parameters that match the data reasonably well, and then we allow these
parameters to vary such that we can explore the sensitivity of model predictions
and their fit to observed data.
JM was variable, and a product of the number of cells in each realm (between
205 and 526) and the number of individuals per community (between 16 and
240). As in chapter 4, we assumed that dispersal occurs only between adjacent
cells (local communities), and that cells outside that particular realm would block
dispersal. Hence, terrestrial communities could not disperse across marine cells,
coastal communities could not disperse across terrestrial or noncoastal marine
cells, and so forth. The Moore neighborhood of each local community therefore
varied based on the geography of the grid cell. Further, we assumed that there was
no dispersal through the Suez and Panama canals, which could act as dispersal
pathways between the Atlantic and Indo-Pacific. As these pathways were opened
relatively recently, we surmised that they may not yet have influenced global rich-
ness patterns in a major way. When we tested the sensitivity of predicted richness
patterns to this particular assumption (by opening or closing canals), we found
little effect (results not shown). We explored the land, coastal, pelagic, and deep-
sea communities separately in our analysis by running ensembles of simulations
for each realm.
The effects of temperature on turnover and speciation rates were parameter-
ized using metabolic theory, and were exactly analogous to chapter 4, but driven
by the observed global (surface) temperature field in both latitude and longitude,
rather than the hypothetical 0 to 30 degree gradient that was assumed in chapter 4.
Global sea and land surface temperature data were taken from published sources
as documented in chapter 3.
While the relationships between temperature, metabolic rate, community
turnover, and speciation rate are based on exact mathematical expressions, there
is no analogous expression that enables us to generally parameterize a relation-
ship between empirical measures of habitat area or productivity and the number
of individuals in a local community. While an approximately linear or log-linear
relationship is empirically founded, the slope of that relationship may be vari-
able and has so far been assessed only over small spatial scales (Connor et al.
2000). Therefore, we scaled the effects of habitat area, productivity, and moisture
(on land) in a linear fashion to the number of individuals in a local community,
analogous to chapter 4. This reflects the empirically observed linear individ-
uals-area relationship for whole faunas (Connor et al. 2000).The relationship
between the number of individuals and habitat area may alternatively be derived
theoretically through consideration of two other well-documented patterns: the
species-area relationship and the species-individuals relationship (Peet 1974).
These two relationships, when combined, give rise to an individuals-area rela-
tionship. However, given the wide range of observed slopes values for species-
area and species-individuals relationships (Hillebrand et al. 2001; Drakare et al.
2006), the slope of the relationship between individuals and area will also vary
widely. We further note that these relationships may change at large scales, as
PREDICTING GLOBAL BIODIVERSITY 127
for example in the species-area slope, which was shown to be scale dependent
(Crawley and Harral 2001).
We conclude that there is both theoretical and empirical justification for a linear
relationship between the number of individuals and habitat area, but that the slope
value of that relationship is unknown. This is even more salient for measures of
productivity, which likewise should increase the density of individuals per unit
area, possibly in a linear fashion, but with unknown slope. One study suggests
that at global scales, the relationship between species (as opposed to individuals)
and energy is steeper than that between species and area (Hurlbert and Jetz 2010).
We thus explored a range of possible scaling relationships, by varying the slope of
the linear function relating empirical habitat and productivity variables (coastline
length, frontal zones, export productivity, or moisture respectively) to the number
of individuals in a local model community, treating this slope as a parameter to
be estimated. We normalized the smallest value of each empirical habitat and
productivity variable to one, with linear scaling to higher values, and estimated
the sensitivity of model predictions to different slope values. We explored a wide
range of scaling relationships, but generally found that a two- to fivefold increase
in community size with increasing habitat area or productivity provides the best
fit to data (see figs. 5.1 to 5.7). Once we identified an approximate optimal scaling
of community sizes, we explored the sensitivity of model predictions to varying
dispersal rate (1 ≤ m ≤ 0.001), and speciation rate (0.1 ≤ v ≤ 0.0001), as in chapter
4. All parameter combinations were run 100 times in coalescence mode. We took
the mean community values across the 100 replicate runs to average across the
stochastic variation inherent in these simulations.
In the following, we separate our model prediction exercise between coastal,
pelagic, deep-sea, and land species, because these major environmental realms
each showed distinct patterns of global biodiversity (chapter 2). We aim to
explore similarities and differences between relative richness patterns (normal-
ized to maximum value), and do not attempt to estimate absolute values of species
richness. We further explore model predictions for endotherms versus ectotherms,
because their metabolic rates may have different sensitivity to ambient tempera-
ture. Importantly, we fit only to observed richness data, and do not use any envi-
ronmentally extrapolated data points, to prevent circularity of reasoning. Finally,
we assess under which conditions the inclusion of temperature niches improves
model predictions of realized biodiversity patterns.
Latitudinal Longitudinal
Habitat Species group peak Accuracy peak Accuracy r P
Coastal All ectotherms Tropical Yes East Asia Yes 0.77 < 0.0001
Coastal Stony corals Tropical Yes East Asia Yes 0.61 < 0.0001
Coastal Brittle stars Tropical Yes East Asia Yes 0.70 < 0.0001
Coastal Bivalves Tropical Yes East Asia Yes 0.71 < 0.0001
Coastal Cone snails Tropical Yes East Asia Yes 0.63 < 0.0001
Coastal Cephalopods* Subtropical No East Asia Yes 0.57 < 0.0001
Coastal Mangroves Tropical Yes East Asia Yes 0.69 < 0.0001
Coastal Seagrasses Subtropical No East Asia Yes 0.48 < 0.0001
Coastal Fish Tropical Yes East Asia Yes 0.70 < 0.0001
Coastal Sharks Subtropical No East Asia Yes 0.36 < 0.0001
Coastal Pinnipeds Polar No Various No –0.39 < 0.0001
Pelagic All ectotherms Subtropical Yes Various Yes 0.87 < 0.0001
Pelagic Foraminifera Subtropical No Various Yes 0.85 < 0.0001
Pelagic Euphausiids Subtropical No Various Yes 0.78 < 0.0001
Pelagic Squids* Temperate No Various Yes 0.20 < 0.0001
Pelagic Tuna and Tropical- Yes Various Yes 0.64 < 0.0001
billfish subtropical
Pelagic Sharks Subtropical No East Asia Yes 0.59 < 0.0001
Pelagic Cetaceans Temperate No Various Yes 0.70 < 0.0001
Pelagic Seabirds Temperate- No Various Yes −0.14 < 0.01
polar
Deep sea All ectotherms Temperate Yes Various Yes 0.24 < 0.01
Deep sea Brittle stars Temperate Yes Various Yes 0.24 < 0.01
Land All ectotherms Tropical Yes Various Yes 0.70 < 0.0001
Land Vascular plants Tropical Yes Various Yes 0.77 < 0.0001
Land Amphibians Tropical Yes South America No 0.64 < 0.0001
Land Reptiles* Subtropical No Central No 0.32 < 0.0001
America,
Africa, Asia
Land Birds Tropical Yes South No 0.59 < 0.0001
America, Asia
Land Mammals Tropical Yes South Yes 0.62 < 0.0001
America,
Africa, Asia
Success
60% 88% 92%
rate
Note: Shown are the summary results of fitting neutral-metabolic metacommunity model runs to empirical
data for aggregate groups (see also figs. 5.1 to 5.8) and fits of the same model for individual taxa. The strength
and significance of the correlation is indicated by correlation coefficient r and associated P-value. Accuracy
of predicted versus observed latitudinal and longitudinal peaks in species richness is noted. Limited global
sampling or limited species coverage is marked with an asterisk (*).
130 CHAPTER 5
A Coastal ectotherms
1.00
Latitude
0.75
0.51
0.26
0.02
0.05 0.18 0.32
Mean richness
B Model projection
1.00
Latitude
0.81
0.62
0.43
0.25
0.32 0.53 0.73
Mean richness
C Residuals
0.85
0.58
Latitude
0.31
0.03
−0.24
0.25 0.30 0.36
Mean richness
Figure 5.1. Predicting coastal biodiversity. Shown are (A) observed richness patterns for all
coastal ectotherm species combined, and (B) predicted global richness patterns derived from
a neutral-metabolic metacommunity model (m = 0.1, v = 0.01) involving the effects of sea sur-
face temperature on evolutionary speed and the effects of coastline length on community size.
(C) Residual variation plot displays spatial structure in the fit between theoretical model predic-
tions and empirical data. Note that only cells with both model prediction and empirical data
have been plotted.
gradient, but may explain why some of the gradients are steeper than expected by
metabolic theory alone (Brown 2014).
When exploring the sensitivity of theoretical model predictions to changes in
speciation and dispersal parameters (see fig. 5.2C), we found moderate sensitivity
to variation in speciation rates, but low sensitivity to changes in dispersal rates,
except at very low speciation, which produced a flat diversity pattern that did not
A Correlation
1.0
Modeled diversity
0.8
0.6
0.4
r = 0.77
0.00 0.25 0.50 0.75 1.00
Empirical diversity
B Coastal ectotherms
1.0
Coastline length + SST
NPP + SST
0.8 Coastline length only
NPP only
0.6
Correlation
0.4
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Length scaling
C
0.8
0.7
0.8 0.6
0.6
Correlation
0.5
0.4
0.4
0.2
0.3
0.0
1 0.1
Di 0.1 0.01 0.2
sp
ers 0.01 es
al 0.001
rat 0.001 0.0001 n rat 0.1
io
es iat
p ec 0.0
S
Figure 5.2. Sensitivity of predictions for coastal biodiversity. (A) The best model fit is derived
by the inclusion of metabolic effects of sea surface temperature (SST) on evolutionary speed and
the effects of coastline length on community size. (B) Sensitivity of model predictions to different
correlates of community size (coastline length or net primary productivity) is shown at a range
of scaling factors. (C) Sensitivity of model results to variation in speciation and dispersal rates.
132 CHAPTER 5
fit well to observed data for any species group. Moderate speciation and dispersal
rates (m = 0.1, v = 0.001), as used in most of our simulations, produced the best fit
to the observed data, with low sensitivity to modest changes in these parameters
(see fig. 5.2C).
Deep-sea ophiuroids represented the single deep-sea taxon with appropriate global
sampling. Their global richness pattern was predicted by the effects of export pro-
ductivity, as scaled to community size in our theoretical model (fig. 5.5), but with
a modest overall fit (r = 0.24, P < 0.01; fig. 5.6A). Including the effects of deep-
water temperature made the prediction slightly worse (fig. 5.6B), and did not give
much contrast, as temperatures are uniformly cold (2–4°C) below 2000 m depth.
PREDICTING GLOBAL BIODIVERSITY 133
A Pelagic ectotherms
1.00
0.76
Latitude
0.52
0.28
0.03
0.11 0.36 0.61
Mean richness
B Model projection
1.00
0.81
Latitude
0.62
0.43
0.24
0.41 0.57 0.74
Mean richness
C Residuals
0.50
0.29
Latitude
0.08
−0.13
−0.33
0.06 0.16 0.27
Mean richness
Figure 5.3. Predicting pelagic biodiversity. Shown are (A) observed richness patterns for all
pelagic ectotherm species combined, and (B) predicted global richness patterns derived from a
neutral-metabolic metacommunity model (m = 0.1, v = 0.01) involving the effects of sea surface
temperature (SST) on evolutionary speed and pelagic frontal habitats (measured as SST slope)
on community size. (C) Residual variation plot displays spatial structure in the fit between theo-
retical model predictions and empirical data. Note that only cells with both model prediction and
empirical data have been plotted.
Modeled diversity
0.8
0.6
0.4
r = 0.87
0.00 0.25 0.50 0.75 1.00
Empirical diversity
B Pelagic ectotherms
1.0
SST + NPP
SST + SST fronts
0.8 NPP only
SST fronts only
0.6
Correlation
0.4
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Length scaling
C
0.9
0.8
0.7
0.8
0.6
Correlation
0.6
0.4 0.5
0.2 0.4
0.0
1 0.1 0.3
Di 0.1 0.01 s
sp
ers ate 0.2
nr
0.01 0.001
al io
rat iat
es 0.001 0.0001 ec 0.1
Sp
Figure 5.4. Sensitivity of predictions for pelagic biodiversity. (A) The best model fit is derived
by the inclusion of metabolic effects of sea surface temperature (SST) on evolutionary speed and
pelagic frontal habitats (measured as SST slope) on community size. (B) Sensitivity of model pre-
dictions to different correlates of community size (SST slope or net primary productivity) at a range
of scaling factors. (C) Sensitivity of model results to variation in speciation and dispersal rates.
PREDICTING GLOBAL BIODIVERSITY 135
A Empirical data
1.00
0.76
Latitude
0.53
0.29
0.06
0.17 0.33 0.49
Mean richness
B Model projection
1.00
0.88
Latitude
0.77
0.65
0.54
0.60 0.74 0.88
Mean richness
C Residuals
0.84
0.53
Latitude
0.22
−0.10
−0.41
0.28 0.40 0.53
Mean richness
Figure 5.5. Predicting deep-sea biodiversity. Shown are (A) observed richness patterns for
deep-sea ophiuroids, and (B) predicted global richness patterns derived from a neutral-metabolic
metacommunity model (m = 0.1, v = 0.01) involving the effects of export productivity on com-
munity size. (C) Residual variation plot displays spatial structure in the fit between theoreti-
cal model predictions and empirical data. Note that only cells with both model prediction and
empirical data have been plotted.
Normalized richness patterns for vascular plants, amphibians, and reptiles were
combined to examine model fit for terrestrial ectotherm species (fig. 5.7; results
for single taxa can be seen in table 5.1). Mammals and birds were excluded due to
their endothermy and are treated separately later. Adapting our neutral-metabolic
A Correlation
1.0
Modeled diversity
0.9
0.8
0.7
0.6
r = 0.24
0.25 0.50 0.75 1.00
Empirical diversity
B Deep-sea ectotherms
1.0
SST + XPP
XPP only
0.8
0.6
Correlation
0.4
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Length scaling
C
0.25
0.25
0.20
Correlation
0.20
0.15
0.15
0.10
1 0.1
Di 0.1 0.01 0.10
sp
ers 0.01 at es
al 0.001 nr
rat tio
es 0.001 1e−04 e cia
Sp 0.05
Figure 5.6. Sensitivity of predictions for deep-sea biodiversity. (A) The best model fit is derived
by the inclusion of export production (XPP) on community size. (B) Sensitivity of model pre-
dictions to the inclusion of temperature or XPP at a range of scaling factors. (C) Sensitivity of
model results to variation in speciation and dispersal rates.
PREDICTING GLOBAL BIODIVERSITY 137
A Land ectotherms
1.00
Latitude
0.75
0.50
0.26
0.01
0.05 0.16 0.26
Mean richness
B Model projection
1.00
Latitude
0.81
0.63
0.44
0.32
Latitude
0.07
−0.18
−0.44
0.00 0.20 0.40
Mean richness
Figure 5.7. Predicting terrestrial biodiversity. Shown are (A) observed richness patterns for
land ectotherms (land plants, amphibians, reptiles), and (B) predicted global richness patterns
derived from a neutral-metabolic metacommunity model (m = 0.1, v = 0.01) involving the effects
of temperature on evolutionary speed and the effects of the number of wet days on community
size. (C) Residual variation plot displays spatial structure in the fit between theoretical model
predictions and empirical data. Note that only cells with both model prediction and empirical
data have been plotted.
Modeled diversity
0.8
0.6
0.4
r = 0.70
0.00 0.25 0.50 0.75 1.00
Empirical diversity
B Land ectotherms
1.0
Temp + wet days
Wet days only
0.8
0.6
Correlation
0.4
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Length scaling
C
0.70
0.65
0.6 0.60
Correlation
0.4 0.55
0.2
0.50
0.0
1 0.1 0.45
Di 0.1 0.01
sp
ers 0.01 0.001 rates 0.40
al
rat 0.001 0.0001 a t ion
es i
ec 0.35
Sp
Figure 5.8. Sensitivity of predictions for terrestrial biodiversity. (A) The best model fit is
derived by the inclusion of metabolic effects of surface temperature on evolutionary speed and
the effects of the number of wet days on community size. (B) Sensitivity of model predictions to
the inclusion of wet days and temperature at a range of scaling factors (C) Sensitivity of model
results to variation in speciation and dispersal rates.
A Cetaceans
1.00
0.76
Latitude
0.51
0.27
0.02
0.29 0.56 0.83
Mean richness
Model projection
1.00
0.81
Latitude
0.62
0.43
0.24
0.41 0.57 0.74
Mean richness
Residuals
0.41
0.23
Latitude
0.06
−0.12
0.8
0.6
0.4
r = 0.70
Figure 5.9. Predicting biodiversity of endotherms. Shown are observed and predicted global
richness patterns for (A) marine mammals (cetacean), and (B) land mammals from a neutral-
metabolic metacommunity model (m = 0.1, v = 0.01), including the effects of surface tem-
perature and wet days on land, and sea surface temperature and net primary productivity in the
B Land mammals
1.00
0.75
Latitude
0.51
0.26
0.02
0.01 0.24 0.48
Mean richness
Model projection
1.00
Latitude
0.81
0.63
0.44
0.30
Latitude
0.07
−0.15
−0.38
0.00 0.16 0.31
Mean richness
Correlation
1.0
Modeled diversity
0.8
0.6
0.4
r = 0.62
ocean. Residual variation plot displays spatial structure, and correlation plot displays the statisti-
cal fit between theoretical model predictions and empirical data. Note that only cells with both
model prediction and empirical data have been plotted.
142 CHAPTER 5
richness. As for land ectotherms, we surmise that the effects of topographic het-
erogeneity may be important here, although we cannot at present capture this
effect in our model. The overall fit relative to observed data (r = 0.62, P < 0.0001)
was somewhat weaker than for land ectotherms, and sensitivity to change in the
dispersal parameter was more pronounced, with low predictive power at both the
high and low end of the scale for dispersal rate (not shown).
The other endotherm taxon examined was birds, which were almost equally
well predicted using the same model as employed for mammals on land (r = 0.59,
P < 0.0001). However, pelagic seabirds (Order: Procellariformes) could not be
predicted well at all (r = −0.14, P < 0.01; see table 5.1) owing to a heavily skewed
distribution of seabird diversity toward the Southern Ocean (see fig. 2.6). Pos-
sibly, the unique dependence of this taxon on wind energy may contribute to an
outlying pattern of diversity, as suggested by Davies et al. (2010). Similarly, the
pinnipeds with their polar distribution were not captured by our metacommunity
model in the same way that cetaceans were (r = −0.39, P < 0.001; see table 5.1),
likely owing to their unique circumpolar distribution.
In conclusion, our theory performs reasonably well for several larger endo-
therm groups comprising thousands of species (land mammals and birds, ceta-
ceans) but poorly for smaller, more specialized marine groups found at higher
latitudes (procellariform seabirds and pinnipeds). The fact that major species
groups are captured reasonably well by the theory may point toward an unex-
plained mechanism that links temperature to evolutionary rates in endotherms, for
which we might otherwise not expect such a pattern. Alternatively, or in addition,
there may be biotic interactions with ectotherms such as competition, predation,
or mutualism that affect the biogeography of endotherms, resulting in an apparent
relationship with temperature. Such interactions might also in part explain the
specialized distributions of endotherm taxa that find a competitive niche in colder,
highly productive environments.
sizes tends to be larger (see fig. 4.13) and there are few constraints to dispersal,
other than unsuitable temperature regimes (Beaugrand et al. 2013).
Here, we evaluate the effects of assumed temperature niches in our expanded
neutral-metabolic-niche model. We began with a base case of moderate dispersal
and speciation rates (m = 0.1, v = 0.01) and used the best-fitting model for pelagic
biodiversity, including the metabolic effects of SST on turnover and speciation
as well as the effects of frontal habitats on community size (see also see fig. 5.3).
When we introduced nonevolving or evolving thermal niches of various width
(from w = ±3 K to w = ±12 K) into this base model, the predicted patterns of
pelagic species richness were practically indistinguishable from the case with-
out niches, and the statistical fit to observed data was very similar (r = 0.81,
P < 0.0001). In order to check this negative result in another environment, and
for less mobile species, we applied the same procedure to land plants, and again
got identical results to the base case without assumed niches, with identical fit to
observed data (r = 0.77, P < 0.0001).
Next, we expanded to a scenario of unconstrained dispersal (m = 1) in pelagic
communities, where such high dispersal rates may be more plausible. In this case,
the ability of the neutral-metabolic base model to capture observed richness pat-
terns for pelagic ectotherms was poor (r = 0.54), largely because the high dispersal
rate equalized global richness and led to a flat pattern overall (fig. 5.10A). When
introducing narrow niches (w = ±3 K) into this model, these formed a barrier to
dispersal and partially restored a gradient in species richness (fig. 5.10B), and
henceforth provided an improved fit to observed data (r = 0.81). Yet these predic-
tions were not as strong as for the non-niche case at moderate dispersal (see figs.
5.3 and 5.4), which also provided a simpler model with fewer assumed param-
eters. These results mirror and support our previous conclusions from chapter 4
that niches have little effect on equilibrium species richness patterns in our model,
except for the limiting case of unconstrained dispersal. Unconstrained dispersal,
however, did not provide a better prediction of the observed data, even for highly
mobile pelagic communities.
5.4. SYNTHESIS
First, we shall pause to emphasize the difference between our theoretical model
and other approaches to predicting global patterns of biodiversity. While the
common approach is to statistically predict observed patterns from environ-
mental correlates (chapter 3 and references therein), our approach here is very
different. Based on our combined neutral-metabolic theory, we have simulated
A Empirical data
1.00
0.76
0.52
0.28
0.03
1.00
0.80
0.61
0.41
0.21
Residuals
0.53
0.32
0.11
−0.11
−0.32
Correlation
1.0
Modeled diversity
0.8
0.6
0.4
r = 0.54
0.2
0.00 0.25 0.50 0.75 1.00
Empirical diversity
Figure 5.10. Effects of thermal niches on model predictions at high dispersal rates. Shown
are predicted global richness patterns for pelagic ectotherm species derived from a neutral-
metabolic metacommunity model, and compared to observed data. The effects of sea surface
temperature (SST) and frontal habitat are included, as in fig. 5.3, but dispersal rate is assumed to
be elevated ten-fold (m = 1, v = 0.01). (A) Empirical composite pattern of species richness for
B Empirical data
1.00
0.76
0.52
0.28
0.03
1.00
0.80
0.60
0.41
0.21
Residuals
0.52
0.33
0.14
−0.05
−0.24
Correlation
1.0
Modeled diversity
0.8
0.6
0.4
0.2 r = 0.81
pelagic ectotherms and model predictions, residuals and correlation plots assuming no niches;
and (B) assuming ±3 K thermal niches. Inclusion of niches improved the predictions under
assumed high dispersal rate, but did not produce as good a fit as the non-niche model at moderate
dispersal rate (see figs. 5.3 and 5.4).
146 CHAPTER 5
that the work reported here can serve as a starting point for others, to improve
predictive capacity by expanding on the basic theory and its model implementa-
tions. We further suggest that the empirical and theoretical insights assembled
here may have some applications in exploring possible effects of human drivers
on the current state and future prospects for biodiversity, and its conservation at a
global scale. These topics will be expanded on in chapter 6.
CHAPTER SIX
Conservation Applications
can help us prioritize mitigation of certain deleterious impacts over others. The
third area of application concerns the future of biodiversity on this planet, and
how an improved understanding of patterns, drivers, and changes in global biodi-
versity can help to project possible scenarios of long-term change, under a given
rate of environmental change. We will briefly highlight these three potential appli-
cations in the following sections.
be adapted for use at finer scales to inform local to regional conservation plan-
ning (for example, Cowling et al. 2003), while accounting for ongoing changes in
human impacts and their dynamic effects on local biodiversity patterns (Pressey
et al. 2007). Remarkably, this approach has never been applied to the planet as a
whole, but only to terrestrial (for example, Myers et al. 2000; Orme et al. 2005)
or marine (for example, Roberts et al. 2002; Selig et al. 2014) environments in
isolation.
Here, we examine global richness patterns and biodiversity hotspots on land
and in the sea together. We look at these patterns through two different lenses—
namely, (1) total species richness, and (2) relative richness across taxa. For total
richness, and combining all available data from chapter 2, fig. 6.1 shows the
synthetic global pattern of total species richness summed across all well-sam-
pled groups on land and in the sea. We observe that the land and coastal marine
areas sustain higher richness than the pelagic ocean or the deep sea (fig. 6.1A),
likely reflecting fundamental differences in temperature (surface being warmer
than deep waters), habitat complexity and productivity (higher on land and in
the coastal ocean than in pelagic and deep-sea habitats), and evolutionary his-
tory (more rapid diversification on land). On the other hand, some aspects of
this strong pattern in absolute richness might reflect the biases of sampling well-
known taxa and the effects of relative sampling effort, accessibility, and length of
study on land and in coastal regions. Moreover, some features are likely shaped to
an extent by species-rich groups such as land plants and coastal fishes (see chapter
2 for similarities and contrasts between individual taxa). Yet the observed land-
sea gradient broadly matches estimated differences in total richness across these
two realms (Mora et al. 2011), though there remains considerable uncertainty
about total deep-sea biodiversity (Rex and Etter 2010).
The most species-richness cells in our global sample are consistently found
in tropical coastal areas, where the highest observed richness of terrestrial and
marine groups intersects. This is seen most prominently in Southeast Asia and
in western South America, where global biodiversity hotspots converge between
land and sea. Here, a conservation approach that integrates terrestrial and marine
protected areas and other conservation measures may be particularly fruitful.
Conservation of species-rich hotspots becomes even more important, when
considering that global patterns in sampled biodiversity also mirror similar gradi-
ents in human population density, environmental footprint, and associated impacts
(fig. 6.1B). Such impacts do not concern the land or the sea in isolation, and con-
centrate in coastal areas, where more than half of humanity has chosen to settle.
This human focus on the coast may at least in part be motivated by the rich port-
folio of natural resources found there historically (Lotze et al. 2006). Due to this
particular settlement pattern, human impacts thus tend to decline with increasing
152 CHAPTER 6
6319
4761
3204
1647
89
B Human impact
High
Medium
Low
Figure 6.1. Total species richness and human impact. Shown is (A) the sum of species counts per
cell for all taxa in chapter 2. (B) Human impacts expressed as cumulative relative impact. Data
were derived for the land from the Human Footprint Map, accessed from NASA’s Earth Observ-
ing System Data and Information System (EOSDIS) Socioeconomic Data and Applications Cen-
ter (http://sedac.ciesin.columbia.edu/data/set/wildareas-v2-human-footprint-geographic/maps);
marine aggregate human impact data were compiled from Halpern et al. (2015).
distance from the coasts, and both toward landlocked areas and toward the open
ocean and deep sea (Vitousek et al. 1997; Imhoff et al. 2004; Halpern et al. 2008).
This means that human impacts appear to be intensified in the world’s most spe-
cies-rich habitats, land and sea, imposing large risks on biodiversity there. The
overlap between species richness and human impacts is particularly strong in East
Asia, which is a major global biodiversity hotspot both on land as well as in the
sea (fig. 6.1A), and home to 60% of the human population, with a massive foot-
print on local resource use (fig. 6.1B). Other marked hotspots of overlap in spe-
cies richness and human impact, both land and sea, include Mesoamerica and the
Caribbean, as well as East Africa including Madagascar (fig. 6.1). The available
data strongly suggest that these areas should all be treated as marine-terrestrial
conservation priorities at a global scale.
Another way of looking at global biodiversity is to describe hotspots of relative
richness that are consistent across taxa, no matter what their absolute richness is.
By normalizing species richness for each taxon and then averaging across all taxa
C O N S E RVAT I O N A P P L I C AT I O N S 153
present in each cell, we derive synthetic patterns of relative richness in which each
group contributes equally to the total estimate, and the strong influence of par-
ticularly species-richness taxa is neutralized (fig. 6.2). Hotspot areas in fig. 6.2 are
arbitrarily delineated by the top 10% of cells, in terms of average normalized rich-
ness across groups. On land, such consistent hotspots across sampled taxa emerged
in the tropical Andes, central East Africa, and Southeast Asia (fig. 6.2A), all of
which sustain tropical wet forests. We note that these cross-taxa hotspots broadly
match the originally described hotspots for vascular plants (Myers et al. 2000),
pointing again at the strong correlation between plant and vertebrate taxa. In the
ocean, highest coastal diversity was observed in hotspots around the Indonesian-
Philippines-Australian Archipelago (the “coral triangle”), southern India and Sri
Lanka, southeast Africa and Madagascar, and the Caribbean (fig. 6.2B), all of which
feature warm waters, extensive coastlines, and large islands that are located close
to continental coastlines. In contrast, smaller island chains in the central Pacific
(Micronesia and Polynesia) do not support the same number of species (fig. 6.2B).
Pelagic species (fig. 6.2C) and deep-sea species (fig. 6.2D) provided a striking con-
trast to terrestrial and coastal marine species by showing much more uniform pat-
terns of diversity across ocean basins, with hotspots at ~20 to 30 degrees latitude
North or South in all oceans. Within these latitudes, the most diverse hotspots of
pelagic species were typically located closer to the continents and along bound-
ary currents such as the Gulf Stream and Kuroshio Current. This probably relates
again to the availability of favorable habitat features of oceanic fronts often forming
alongside boundary currents and along shelf breaks (Olson et al. 1994; Etnoyer et
al. 2004; Worm et al. 2005). In the deep sea, hotspots of diversity for ophiuroids, the
single taxon with global coverage, were seen in the North Atlantic, off Japan and
New Zealand, and below coastal upwelling zones in southwest Africa, South Amer-
ica, and the California Current (fig. 6.2D). The North Atlantic hotspot was also
seen in other taxa that were more regionally sampled, such as mollusks (Tittensor
et al. 2011). When comparing relative richness patterns across all sampled realms
and taxa, the Western Pacific Rim emerged as particularly prominent for terrestrial,
coastal, and pelagic taxa, including hotspots around Japan, the Philippines, and
Australia (fig. 6.2). The North Atlantic, particularly off Europe and North America,
showed regional hotspots for both pelagic and deep-sea taxa; similar overlap was
seen in the southern hemisphere off South Africa and New Zealand. Land and deep-
sea species richness hotspots coincided around eastern South America.
This highly nonrandom distribution of species richness within and across
taxa and realms is an important feature to inform systematic conservation plan-
ning on regional to global scales (Margules and Pressey 2000; Myers et al.
2000). Specific habitat features tend to harbor the highest observed diversity
across species groups; these include tropical moist forests on land, tropical
A Normalized terrestrial richness
0.80
0.60
0.40
0.20
0.00
0.80
0.60
0.40
0.20
0.00
0.80
0.60
0.40
0.20
0.00
0.80
0.60
0.40
0.20
0.00
Figure 6.2. Hotspots of species richness across taxa. Shown are normalized patterns of species
richness across all species groups in each of the four major realms: (A) land (4 taxa), (B) coastal
(9 taxa), (C) pelagic (7 taxa), and (D) deep sea (1 taxon). The 10% most species rich cells are
outlined in bold white. Richness data were derived from sources in table 2.1.
C O N S E RVAT I O N A P P L I C AT I O N S 155
As discussed in chapters 2 and 3, biodiversity has been a dynamic entity over the
past 4 billion years. Yet, at the present time, biodiversity is changing much faster
than throughout most of Earth’s history, driven by unprecedented environmen-
tal change brought about by manifold human impacts. In this section, we briefly
review what is known about present biodiversity change, and how it relates to our
previously discussed understanding of fundamental drivers. We suggest that an
improved understanding of biodiversity drivers gained in chapter 3 can help us
to prioritize the mitigation of certain human impacts over others, and to project
future scenarios of biodiversity change from process-based models (see also the
following sections).
As biodiversity continues to change and evolve in response to environmental
change, it is paramount for conservation biologists to gain a better understand-
ing of the direction of change, and the main drivers that effect this (Sala et al.
156 CHAPTER 6
Driver of Model
diversity Direction of human impact implementation Source
Land Coast Pelagic Deep
sea
Temperature Temperature Burrows et al.
2011; Allen et al.
2014
Habitat area Community size Bender et al.
1998; Worm and
Lenihan 2013
Productivity Community size Nemani et al.
2003; Boyce and
Worm 2015
Disturbance Community Watling and
turnover Norse 1998;
Dirzo et al. 2014
Connectivity Dispersal rate Fahrig 2003;
Cowen et al.
2006
Note: Shown are major drivers of biodiversity that are affected by various human impacts; the approxi-
mate direction of global impacts is indicated by arrows ( increasing, decreasing, minor change).
Model implementations (see chapter 4) of key parameters that can be adjusted to simulate change in
human impacts are listed. References highlight documented global impacts and their direction.
2000; Thomas et al. 2004; Tittensor et al. 2014). In our empirical analyses and
theoretical models, temperature (affecting species distributions, community turn-
over rates, and speciation rates) as well as habitat area and productivity (driving
changes in local community size) emerge as strong predictors of species richness
through space and time. This is significant, because human activities increasingly
affect and drive these variables (Waters et al. 2016). There is now ample evidence
that human-induced changes in temperature, habitat area and complexity, produc-
tivity and disturbance, as well as changes in connectivity among ecosystems, are
affecting biodiversity worldwide (table 6.1). Arguably, these are among the most
severe human impacts, precisely because they affect fundamental drivers of bio-
diversity. These drivers have been much studied, and their effects have been mod-
eled using various approaches, In this section, we will very briefly explore such
processes as they unfold across various ecosystems, and assess to what degree
observed changes in species distribution and diversity may relate to predictions
made by our theory.
C O N S E RVAT I O N A P P L I C AT I O N S 157
We are living through an era of rapid global climate change (Allen et al. 2014),
which is causing both regional- and planetary-scale changes in weather patterns,
temperature, and ice cover (Barnett et al. 2001; Mora et al. 2013) and further
affects sea level, thermal stratification regimes, atmospheric and oceanic circula-
tion patterns, ocean pH, oxygen content, and productivity (Richardson and Schoe-
man 2004; Sarmiento et al. 2004; Bryden et al. 2005; Behrenfeld et al. 2006;
Polovina et al. 2008; Boyce et al. 2010; Lewandowska et al. 2014; Boyce and
Worm 2015; Waters et al. 2016). As such, global climate change is predicted to
have complex effects on species distributions and diversity (Sala et al. 2000;
Brander 2010; Sydeman et al. 2015).
Clearly, temperature change in particular directly relates to species richness
empirically (chapter 2) and in our theory and model (chapters 4 and 5). Sustained
changes in temperature under global warming will affect species’ thermal niche
distributions (ecological timescale), individual metabolic rates and community
turnover (ecological timescale), and speciation and diversification (evolutionary
timescales). Here, we briefly review what is known empirically about the effects
of temperature changes on observed diversity patterns.
In the oceans, long-term fish and plankton monitoring programs have provided
particularly compelling evidence for temperature- and climate-driven changes in
species diversity (Hays et al. 2005; Perry et al. 2005; Pinsky et al. 2013). Obser-
vations most consistently indicated an increase in species richness in temperate
regions, as warm-adapted species invade from the subtropics and more than com-
pensate for cold-adapted species that are displaced to higher latitudes. The net
effect of such temperature-induced compositional changes on species richness can
be surprisingly large: for fish, an almost 50% increase in the number of species
recorded per year in North Sea bottom trawl surveys was documented between
1985 and 2006 (Hiddink and ter Hofstede 2008). This change correlated tightly
with increasing water temperatures. Similar trends have been found in the United
Kingdom’s Bristol Channel, where fish species richness increased by 39% from
1982 to 1998 (Henderson 2007). In both cases, increases in richness were mainly
driven by invasion of small-bodied southern species. Similar changes have been
observed for terrestrial taxa, where long-term monitoring has historically been
somewhat more straightforward. For example, the species richness of British but-
terflies (Menéndez et al. 2006) and epiphytic lichen in the Netherlands (van Herk
et al. 2002) have increased with warming over time.
Some polar regions have seen a similar pattern of slow invasion by temperate
species. However, only few observations on net changes in species richness are
available (Wassmann et al. 2011)—for example, surveys of Arctic macrobenthos
158 CHAPTER 6
suggest slow increases in species numbers at sites that are accessible to larval
advection from southern locations (Węsławski et al. 2011). Large uncertain-
ties remain due in part to low sampling effort (Wassmann et al. 2011), and in
part due to the complexity of this highly seasonal environment, and the com-
pounding effect of changes in sea ice, salinity, stratification, runoff, and acidity
(Michel et al. 2012).
Under rapid warming, tropical regions become increasingly heat-stressed and
may decline in richness, particularly where there is a loss of foundation species
such as trees or corals. For example, coral reefs have experienced mass bleach-
ing where sea temperatures have exceeded long-term summer averages by more
than 1°C for several weeks (Hoegh-Guldberg 1999; Donner et al. 2005; Hughes
et al. 2017). The loss of sensitive coral species causes secondary changes of reef-
associated or obligate fauna and flora (McClanahan et al. 2001). For reef fish
specifically, studies indicate large changes in species composition after bleaching
events, and a decline in species diversity that is linearly related to disturbance
intensity (Wilson et al. 2006), but see Bellwood et al. (2006) for an exception.
Other habitat-forming species, such as seagrasses, mangroves, and some sea-
weeds, also face elevated extinction risk due to warming and sea-level rise (Poli-
doro et al. 2010; Short et al. 2011; Harley et al. 2012), with consequences for
communities dependent on these habitats.
On land, the effects of warming are also complicated by changing patterns of
rainfall, which could enhance or reduce diversity (Bellard et al. 2012). Empirical
evidence for lizards (Sinervo et al. 2010) and amphibians (Pounds et al. 2006)
suggest, however, that climate-related extinctions are already under way on land.
More generally, warming will increase metabolic rates in these and other ecto-
therm species, and this increase will be disproportionally large in tropical species
due to the nonlinear relationship between temperature and metabolic rate (Dillon
et al. 2010). Such increases likely have substantial physiological and ecological
impacts including an increased need for food, increased rates of evaporative water
loss in dry environments, behavioral change, changes in tropical soil respiration,
and altered food web dynamics (Dillon et al. 2010). In the long term, as discussed
at length in chapter 3, metabolic theory also predicts increased rates of molecular
evolution and speciation (Allen and Gillooly 2006), which could ultimately foster
biological innovation and novel adaptations.
In summary, as the planet rapidly warms, and as a first approximation, the
tropics may lose diversity, temperate regions show species turnover and increases
in net diversity, whereas polar environments so far mostly show declines in ice-
dependent species and some invasion of subpolar taxa with unclear effects on net
diversity. Constraints to range shifts and expansions appear to be less important
in the oceans than on the land. In the North Sea, for example, the average rate of
C O N S E RVAT I O N A P P L I C AT I O N S 159
northward range shift was 2.2 km a−1 (Perry et al. 2005), more than three times
faster than observed range shifts in terrestrial environments, which reportedly
average 0.6 km a−1 (Parmesan and Yohe 2003). Likewise, a meta-analysis of spe-
cies range shifts showed that marine species fill their thermal niches more fully
and move more readily at both cold and warm range boundaries compared with
than terrestrial species (Sunday et al. 2012). These findings may not be surprising,
given the absence of hard physical boundaries in marine, and particularly pelagic,
environments. There are, however other potential constraints, such as light avail-
ability for corals at higher latitudes, or lower oxygen for fish in warmer waters,
that might prove to set unexpected boundaries to dispersal (Kleypas 2015). Simi-
larly, changes in habitat area or productivity could constrain realized changes in
species richness relative to anticipated outcomes based solely on thermal change,
as discussed later.
meat” hunt on land (Milner-Gulland et al. 2003), or the estimated annual fish catch
of ~100 Mt from the oceans (Pauly and Zeller 2016) can affect the regional pat-
terns of species richness (Worm et al. 2005; Effiom et al. 2013). Indeed, observed
richness of large fish (>25 cm length) on standardized underwater visual surveys
is depressed in most fished areas, but elevated in well-enforced marine protected
areas worldwide (Edgar et al. 2014). A model of species richness for all reef-asso-
ciated fish, large and small, showed the typical pattern of maximum richness in the
western tropical Pacific (Edgar et al. 2014). In contrast, observed richness for large
fish showed a subtly different pattern, peaking at remote sites in the central Pacific,
around French Polynesia and the Line Islands (Edgar et al. 2014). Similarly, an
analysis of heavily exploited tuna and billfish species sampled by a globally oper-
ating long-line fleet indicated large-scale changes in species richness patterns from
the 1950s to 1990s, and overall declines in average richness in the Atlantic and
Indian Ocean, but not the Pacific (Worm et al. 2005; Worm et al. 2010). Changing
richness patterns were at least in part due to declining range extent in a number of
species, such as three species of bluefin tuna (Worm and Tittensor 2011). Likewise,
a historical data set of coastal biodiversity showed that 7% of the species for which
data were available went regionally extinct over the last 1000 years (Lotze et al.
2006; Worm et al. 2006). Clearly, fishing is reshaping patterns of local and regional
richness for exploited groups, as well as the spatial scaling of biodiversity (Titten-
sor et al. 2007), and this change might skew our perception of what the baseline
patterns of biodiversity are (Pauly 1995; Baum and Myers 2004).
Despite these large regional changes, global extinctions remain relatively few in
the oceans. Only two marine fish species are thought to have gone globally extinct
over recent human history, although undersampling in reef and deep-water envi-
ronments may hide unrecognized extinction events (Roberts 2002; Reynolds et al.
2005). However, four marine mammals and four mollusks are known to be lost, as
well as a number of freshwater fish and one anadromous species, the New Zealand
grayling, according to the International Union for Conservation of Nature (IUCN)
Red List (Worm and Lenihan 2013). This means that the overall global richness of
marine life is likely still similar to preindustrial times, but that local and regional
patterns of community structure, biomass, and diversity have changed, sometimes
dramatically, especially for heavily exploited species (Lotze et al. 2006; Estes et
al. 2011). This is reflected in the IUCN assessments, which listed between 9%
(for bony fish) and 38% (for marine mammals) of non-data-deficient species as
threatened by extinction in 2013 (Worm and Lenihan 2013). On land, in contrast,
more than 700 extinctions have already occurred, many of them due to exploitation
or human-mediated invasion by exotic species, with tens of thousands threatened
by extinction in the near future (Barnosky et al. 2011). While this situation does
not yet reach the magnitude of previous mass extinction events, human impacts
162 CHAPTER 6
In summary, many (but not all) of the human impacts that are affecting bio-
diversity on land and in the sea relate directly to key processes that are driv-
ing observed patterns of species richness—namely, temperature, habitat area and
connectivity, productivity, and disturbances. As such, it seems prudent to further
investigate these impacts as we attempt to mitigate biodiversity change and loss.
Where empirical observations are scarce or unavailable—for example, in poorly
sampled regions, or with respect to the future—models can help us to explore pos-
sible scenarios of biodiversity change under a given impact scenario. While there
are many other process-based models that can potentially shed light on human
impacts in biodiversity, we suggest that the theoretical framework presented in
this volume may, with some modification, be useful to further explore and project
the likely long-term consequences of these impacts, and their effects on species
richness now and in the future (see table 6.1). As discussed in chapters 4 and 5,
changes in surface temperature are readily incorporated into the model via a global
temperature grid that can be adjusted to reflect observed or projected changes.
Changes in suitable habitat area correspond directly to changes in the commu-
nity size parameter J, and could thus be potentially assessed alongside the effects of
temperature. Increasing habitat fragmentation limits connectivity among neighbor-
ing communities, and could be assessed through modification in dispersal rate m,
adjusting the connectivity between local communities in our metacommunity model
(see table 6.1). Changes in disturbance or exploitation rate can be implemented as
changes in local community turnover. As such, our metacommunity model may
provide an experimental toolbox for projecting the complex effects of single or
combined environmental impacts and changes. As an example, we will illustrate
in the following the use of our model framework for exploring scenarios of future
climate change, in order to understand possible long-term effects on biodiversity.
uses empirical data synthesis and process-based models to project possible changes
in biodiversity, and the effects it may have on conservation outcomes. As spatial
data on recent changes in species richness patterns are typically not available at a
global scale (but see Worm et al. 2010; Dornelas et al. 2014; Newbold et al. 2016
for exceptions), species distribution models have been utilized most commonly to
fill this gap. These models typically reconstruct biodiversity patterns by overlay-
ing species ranges derived from habitat requirements, such as thermal tolerances,
for example. Such models can be used to forecast future changes in species rich-
ness, given projected scenarios of climate and habitat change (Cheung et al. 2009;
Kaschner et al. 2011). Here, we briefly review existing modeling approaches and
then present a new, and possibly complementary, approach based on our global
metacommunity model presented in chapters 4 and 5.
Species distribution models (SDMs) are currently the most widely used model-
ing tool to project the possible future impacts of environmental change on bio-
diversity (Franklin et al. 2013). Terrestrially, there exists a voluminous literature
on projecting biodiversity changes using SDMs (see, for example, Urban 2015);
we don’t attempt to survey it further here. The vast majority of projections for
future biodiversity scenarios have been derived from SDMs assuming static ther-
mal tolerances and thermal niches. For example, Molinos et al. (2016) recently
used observed temperature tolerances and habitat preferences for 12,796 coastal
and pelagic species represented in Aquamaps, an online resource that maps out
predicted species ranges in the marine environment (www.aquamaps.org). The
authors projected overall increases in average biodiversity under both moderate
and high greenhouse gas emissions, as implemented in the representative con-
centration pathways RCP 4.5 and RCP 8.5 (fig. 6.3). Increases in species richness
were projected at high latitudes, matching empirical observations (see section
6.2.1, earlier). But these regional increases were partly offset by declining rich-
ness in tropical regions, especially at the higher emission scenario (fig. 6.3); this
also led to a partial flattening of latitudinal richness gradients (Molinos et al.
2016). It is important to remember that these changes were entirely based on
assumed contraction or expansion of predicted species ranges; neither speciation
nor extinction processes were included, assuming no evolutionary change. Tropi-
cal losses of biodiversity, for example, were driven by the displacement of tropi-
cal species to higher latitudes, under assumed constant temperature tolerances,
and no evolution. Parallel work for fishes (Cheung et al. 2009) and marine mam-
mals (Kaschner et al. 2011) produced similar results for individual species groups,
164 CHAPTER 6
Δ Richness
<–1,000
–1,000 to –500
ΔN° species
–500 to –250
100 –250 to –50
50 ΔRRCP8.5 –50 to 0
0
0 ΔRRCP4.5
0–50
50–250
250–500
500–1,000
–1 0
–1 0
–1 0
–1 0
00
0
0
0
0
0
20
40
60
80
0
0
0
0
0
8
6
4
2
–8
–6
–4
–2
10
12
14
16
18
–1
90
70
ΔN°RCP4.5(2006–2100)
50
Latitude (°)
30
10
–10
–30
–50
–70
–90
–200 0 200
B ΔN° species
90
70
ΔN°RCP8.5(2006–2100)
50
30
Latitude (°)
10
–10
–30
–50
–70
–90
–200 0 200
ΔN° species
Figure 6.3. Future projections for marine species richness from a species distribution model.
Shown are projected changes in absolute species richness (n = 12,796 species) from 2006 to
2100 given two representative concentration pathways (RCPs) devised by the Intergovernmen-
tal Panel on Climate Change (IPCC), (A) RCP 4.5 and (B) RCP 8.5. Black contour lines cor-
respond to countries’ exclusive economic zones (EEZs). Changes to average latitudinal and
longitudinal gradients (solid line) with their 25% and 75% quartiles (shaded areas) are also
shown. Note that the high-emission scenario results in much greater species losses, particularly
in the tropics. After data in Molinos et al. (2016).
with some more recent work including more realistic dispersal constraints into the
SDMs (Cheung et al. 2015).
Beaugrand and coworkers used a more theoretical approach of modeled distri-
butions for randomly generated pelagic “pseudospecies,” mimicking present-day
distributions of pelagic zooplankton groups (Beaugrand et al. 2015). Their results
C O N S E RVAT I O N A P P L I C AT I O N S 165
showed patterns similar to the previously mentioned SDMs. The authors quanti-
fied projected changes relative to those that might have occurred between the last
glacial maximum and the present. For example, under severe global warming
(RCP 6.0 and RCP 8.5), between 50 and 70% of the global ocean was projected to
show a magnitude of biodiversity change not seen since the last glacial maximum
(Beaugrand et al. 2015).
Other approaches that sought to model future biodiversity have been based on
observed species-environment relationships from electronic tracking data (Hazen et
al. 2013), or observed species richness-environment relationships from observational
data (Whitehead et al. 2008). What connects all of these contemporary approaches
is that they do not include evolutionary dynamics. Species distribution models may
capture the short-term rearrangement of species that are thought to move along with
their temperature niche, but do not capture the effects of changes in temperature on
the processes that generate biodiversity, nor the evolutionary plasticity that allowed
species to survive previous environmental perturbations. Microevolutionary pat-
terns of adaptation to changes in the environment are becoming increasingly obvi-
ous in empirical studies (Hoffmann and Sgrò 2011). For example, it has been shown
experimentally that stickleback evolve quickly from marine forms to freshwater
forms, with one major aspect being rapid adaptation to low temperatures (Barrett et
al. 2010). Similar patterns of rapid evolutionary adaptation have recently emerged
in marine phytoplankton (Schlüter et al. 2014) and corals (Palumbi et al. 2014).
This recent work collectively suggests that changes in ocean temperature and acid-
ity provide major evolutionary challenges along which populations diversify within
species, and most likely across species and lineages. Adaptation to very high tem-
peratures, however, appears particularly challenging for many taxa and could rep-
resent a hard evolutionary barrier. Araújo et al. (2013) analyzed thermal tolerances
of a large number of terrestrial ectotherm (n = 697), endotherm (n = 227), and plant
(n = 1816) species worldwide, and showed that tolerance to heat is largely conserved
across lineages, while tolerance to cold is more malleable and varies more widely
between and within species. It is interesting that the evolutionary processes that
may have shaped global diversity patterns over millennia are also observable in real
time (Barrett et al. 2010; Hoffmann and Sgrò 2011; Palumbi et al. 2014; Schlüter et
al. 2014). In terms of the approach in this volume, our eco-evolutionary metacom-
munity model allows for such evolutionary processes to occur, while also tracking
changes in habitat, disturbance regimes, dispersal, and thermal adaptation.
In order to experiment with the utility of our approach for understanding dynamic
changes owing to human impacts, we utilized the richness patterns predicted by our
166 CHAPTER 6
1.00
0.81
0.62
0.43
0.25
1.00
0.81
0.62
0.42
0.23
Difference (future-present)
0.10
0.05
0.00
−0.05
−0.10
Figure 6.4. Future projections for coastal marine species richness from a neutral-metabolic
metacommunity model. Shown are relative changes in species richness from 2006–2010 to
2091–2100 given projected climate change effects on surface temperature and precipitations
forced by the representative concentration pathway RCP 8.5.
1.00
0.78
0.56
0.33
0.11
1.00
0.77
0.53
0.30
0.07
Difference (future-present)
0.60
0.40
0.20
0.00
−0.20
Figure 6.5. Future projections for land plant species richness from a neutral-metabolic model.
Shown are relative changes in species richness from 2006–2010 to 2091–2100 given projected
climate change effects on surface temperature and precipitations forced by the representative
concentration pathway RCP 8.5.
We extracted mean modeled land and sea surface temperature in 2091–2100 and
proportional changes in precipitation between 2006–2010 and 2091–2100. We did
not use predicted precipitation change directly in our model, but rather translated
this information into changes in wet days, assuming linear scaling between annual
precipitation and number of wet days, which have are a good global predictor of
species richness for terrestrial plants (Kreft and Jetz 2007).
One critical aspect of this exploration is that the effects of temperature on com-
munity turnover and speciation are calculated relative to other local communities
rather than pinned to an absolute value (that is, the summed probabilities for all
individuals in all communities acting within each time step is equal to one). This
means that while the model, in its present form, can be used to examine the rela-
tive change in spatial patterns of species richness, it cannot be used to examine
absolute changes in richness values (though with further modification this may be
feasible). Note also that our metacommunity model runs to a dynamic equilibrium
state. Hence, the question we are asking is: How might the long-term ecological
and evolutionary changes effected by changes in temperature and moisture alter
global richness patterns and gradients relative to the present day?
We ran future explorations for coastal ectotherms and land plants, contrasting
marine and terrestrial patterns of response to projected climate change. Changes
in the future richness of ectothermic coastal species (see fig. 6.4) were calcu-
lated in response to projected changes in temperature only, as habitat effects via
changes in coastline length could not be parameterized. As in the species dis-
tribution models discussed earlier, normalized species richness (relative to the
observed global maximum richness) increased in most cells, particularly those
projected to see large increases in temperature, such as those in the northwest
Atlantic (see fig. 6.4). Some decreases in relative richness were observed along
European and Caribbean coastlines, though.
On land, vascular plant richness according to empirical analyses (Kreft and
Jetz 2007) and our theoretical models (chapter 5) is affected by changes in both
temperature and moisture regimes. Both of these variables are tracked by Earth
System Models and can hence be included in our projections (see fig. 6.5). Our
model projected reduced relative richness in the two present hotspots, the tropical
Andes and South America, but also in Europe and West Africa. Large increases
in regional richness were projected for the Arabian Peninsula, for example, fol-
lowing projected increases in precipitation (see fig. 6.5). Some major features of
these patterns of change were also reproduced by projections based simply on
extrapolation of the statistical relationships between temperature, moisture, and
plant species richness (Sommer et al. 2010).
These results should not be viewed as solid predictions but rather simple
explorations based on changes in only one or two factors, and ignoring many
C O N S E RVAT I O N A P P L I C AT I O N S 169
As discussed throughout this volume, biodiversity patterns are not a static fea-
ture. Both in recent decades, and certainly throughout Earth’s history, the global
magnitude and distribution of biodiversity has been dynamically changing in
response to various environmental drivers, many of which are now affected by
human activities. This means that the future of biodiversity is in a very real sense
in our own hands, and future trajectories will largely depend on how we choose
to constrain or manage the cumulative impacts that arise from our actions. This
aspect of biodiversity change is not easily captured by ecological theory, as it
is largely dependent on societal choices and interactions. Fortunately, there is
growing awareness of this simple fact, and global commitments to halt the loss of
biodiversity are being implemented as we write this.
When tracking changes in these commitments, however, it is often found that
increases in pressures on biodiversity outweigh increasing societal responses to
stem the loss of biodiversity (Butchart et al. 2004; Butchart et al. 2010; Titten-
sor et al. 2014). In a recent update on this topic, progress toward the 20 “Aichi”
targets—a set of internationally agreed goals for biodiversity conservation under
170 CHAPTER 6
2000 2005 2010 2015 2020 2000 2005 2010 2015 2020 2000 2005 2010 2015 2020
Figure 6.6. Global trends in biodiversity pressures, biodiversity states, and societal responses
to biodiversity loss. Each line represents a unique response variable and its estimated significant
(continuous) or nonsignificant (dotted) trend, extrapolated to 2020. Lines represent trends rela-
tive to 2010 value (horizontal dotted black line). Redrawn; after Tittensor et al. (2014).
Conclusions
A striking and primary feature of our planet is the bewildering diversity of species
that inhabit it. As far as we know, the geographic distribution of these species has
always been highly nonrandom (Stehli et al. 1969; Yasuhara et al. 2012; Man-
nion et al. 2014), and spatial variation in biodiversity along latitudinal, altitudinal,
and moisture gradients has long been recognized by naturalists (Gaston 2000).
Here, we approached these general patterns from an empirical and a theoretical
perspective, in an attempt to gather a more comprehensive understanding of the
processes that control the distribution of species richness at the global scale. Our
approach was driven by two lines of inquiry—specifically, (1) synthesizing, test-
ing, and contrasting observed patterns, hypothesized drivers, and environmental
predictors across all available taxa on land and in the ocean on the same global-
scale grid; and (2) integrating aspects of neutral, metabolic, and niche theories
into a synthetic neutral-metabolic-niche (NMN) theory of biodiversity (fig. 7.1A)
that generates a surprisingly accurate picture of global richness patterns from few
underlying processes.
ry
Nic
eo
l th
he
a
the
utr
or y
Ne
NMN THEORY
Metabolic theory
Speciation Environment
B
Global community
Speciation
Drift
Selection
Dispersal Dispersal
Local
community
Speciation
Drift
Selection
Figure 7.1. Synthesis in ecological theory. (A) The theory that is presented in this volume
considers these fundamental processes by combining aspects of neutral, metabolic, and niche
theory into a unified framework (NMN theory) that can be used for studying large-scale bio-
diversity patterns. While previous theories focused on dispersal and speciation (neutral theory),
or dispersal-environment relations (niche theory), or how evolution is shaped by the environ-
ment (metabolic theory), we are modeling all three processes and their relationships in a syn-
thetic, mathematically explicit framework. (B) In a previous conceptual synthesis, Vellend
(2010, 2016) identifies four fundamental processes that shape local, regional, and global com-
munity structure and biodiversity: speciation, random drift, and selection shape local, regional,
and global species pools, whereas dispersal drives the exchange of species across these pools.
CONCLUSIONS 173
energy (typically measured as average surface temperature) was the only factor
that empirically correlated with areas of high diversity across all realms we inves-
tigated (chapter 3). Notable exceptions included cold-adapted endotherms such
as pinnipeds, and deep-sea species, which occur in environments that are near-
uniformly cold, with very limited temperature variation. Within each realm, we
found that variables related to habitat area and productivity were also important
(see table 3.6), both of which constrain the number of individuals in a community.
Based on this empirical synthesis, we attempted to devise a body of theory
that might explain observed biodiversity patterns within and across taxa (chap-
ter 4). The idea was to examine the ecological and evolutionary processes by
which empirically documented predictors could affect diversity—for example,
by changing community size or turnover, influencing the speed of evolution, or
constraining dispersal. From this theoretical framework, we built a global meta-
community model that could simulate these processes, singularly or in combina-
tion, and generate global biodiversity patterns, thus enabling us to investigate their
relative influence. This work entailed a synthesis of three largely separate bodies
of ecological theory—namely, neutral theory (Bell 2001; Hubbell 2001), which
includes dispersal and speciation; metabolic theory (Allen et al. 2002; Brown
et al. 2004), which parametrizes environmental constraints on metabolic activity
and speciation rates; and niche theory (Hutchinson 1957; Pearman et al. 2008;
Beaugrand et al. 2013), which links the environment to constraints on dispersal
(fig. 7.1A). Hence, we developed a synthetic theory that attempted to reconcile
previously idiosyncratic patterns of global biodiversity on land and in the ocean,
and produced generally strong fits to the empirical data via our model implemen-
tation (chapter 5).
Somewhat surprisingly, this theory suggests that only two variables are required
to predict the majority of first-order patterns of biodiversity on our planet—
namely, ambient temperature and community size. Temperature primarily affects
the rate of community turnover and the speed of evolution, while community size
determines the number of individuals on which evolutionary processes can act.
To use a chemical analogy, a community of individuals may be equivalent to the
number of molecules reacting with each other in a test tube: the more molecules
are interacting, and the higher the temperature, the higher the probability that
new compounds will form purely by chance. Or, by similar analogy, life emerged
on a primitive Earth almost 4 billion years ago (Tashiro et al. 2017), presumably
through random chemical reactions at elevated temperature, whereas more com-
plex organic compounds formed from simpler inorganic precursors (Miller and
Orgel 1974). This analogy also helps us to highlight a significant weakness of the
theory—namely, that the step from random mutations and molecular evolution to
speciation rate is not mechanistically resolved, much like the step from randomly
174 CHAPTER 7
formed organic compounds to functioning cells is not clear. Even though our
theory is not fully mechanistic due to this gap in understanding, it is encouraging
that with an assumed link between temperature and speciation rate, and realistic
spatial structure, the global patterns of biodiversity are captured surprisingly well
(chapter 5).
The empirical and theoretical synthesis offered here may therefore help in some
way to resolve a long-standing debate on the causes and mechanisms that underlie
observed patterns of species richness at large scales (Rohde 1992; Gaston 2000;
Gotelli et al. 2009; Brown 2014). From a theoretical perspective, this is significant
because it sheds light on some of the processes that may drive first-order bio-
geographic patterns on our planet. From an applied perspective, this knowledge
has relevance for understanding and conserving global biodiversity in the face
of human impacts and global environmental change (chapter 6). From a methods
perspective, our theory provides a flexible “modeling toolbox” to test alternative
hypotheses on the relative importance of changes in environmental factors (for
example, temperature, productivity), metacommunity processes (for example, dis-
persal), ecological attributes (for example, niche width), or even sampling effects
(for example, the mid-domain effect) in driving the spatial organization of bio-
diversity. Of course, this toolbox could be extended, as there are many other pro-
cesses (for example, the effects of large-scale disturbances, or habitat complexity)
that could be specified and tested within this modeling framework. Ultimately, we
hope that others will take our approach as a starting point and push it further, by
improving the realism, confronting it with more data, specifying missing mecha-
nisms, or invoking alternative hypotheses and approaches. We consider our theo-
retical and modeling framework as complementary to the statistical modeling more
commonly used in macroecology and biodiversity science (Gotelli et al. 2009).
Notwithstanding some outstanding questions about specific mechanisms, we hope
that our work will help spur the development and growth of mechanistic macro-
ecology, a more process-oriented approach to understanding large-scale patterns of
biological variation (Purves et al. 2013; Harfoot et al. 2014).
Our work is founded on the Unified Neutral Theory of Biodiversity and Bio-
geography (Hubbell 2001), which itself built on the Theory of Island Biogeography
(MacArthur and Wilson 1967). Both are neutral models, in which no inherent eco-
logical differences among species need to be invoked to explain observed patterns
of species richness. While MacArthur and Wilson’s theory focused on local ecolog-
ical processes, specifically local community size and dispersal, Hubbell extended
CONCLUSIONS 175
dynamics (see fig. 7.1B). This theory was later implemented in a neutral two-
species community model that served to illustrate some of the underlying pro-
cesses (Vellend 2016), and hypotheses generated by the theory were discussed
and compared against empirical data. Generally, there are conceptual similarities
between Vellend (2016) and our volume, the major differences being the inclusion
of metabolic theory, and the focus on large-scale biogeographic patterns herein
(see fig. 7.1A). We believe that our theory is the first of global biodiversity to inte-
grate all of Velland’s key processes, re-create them in a model, and translate their
effects from local communities to the global scale. Our theory might also be con-
sidered “efficient” sensu Marquet et al. (2014), as it is grounded in first principles
and mathematical expressions, makes few assumptions, generates a large number
of predictions per free parameter, and enables the testing of model predictions
against empirical data, as demonstrated in chapter 5. We are, of course, aware
of its limitations, and of the trade-offs between simplicity and realism. Thus, we
make no claims of providing a final answer to some of these questions; this is
merely a first attempt at reconciling some fundamental processes in a relatively
coarse, abstract, and simple approach.
Our theory might perhaps help to inform an ongoing debate on the importance
of neutral (independent of species traits) versus niche (species-specific adapta-
tions) processes in shaping macroecological patterns. We find that on the large
scales that we consider here, thermal niches do not emerge as a major factor in
the generation of global species richness patterns, though they do have a clear
influence on both species composition and community turnover (chapter 4).
Indeed, if we were concerned with the identity of the species that make up our
local communities, or with metrics that took such identities into account, then
niches would clearly play a large role in constraining those. However, reason-
ably realistic latitudinal distributions along temperature gradients consistently
emerged in a neutral-metabolic metacommunity whether niches were assumed
or not. Our model results suggested that the average species will not establish a
large range, not necessarily because of limited niche width, but because of the
difficulty of colonizing new habitats and establishing a larger range across suc-
cessive neighboring communities (Note that the model assumption of zero-sum
ecological dynamics is important in this context.) The model results indicated that
it is far more likely for a new species to become extinct than it is for it to grow in
abundance and to spread to other regions. Thermal niches did significantly con-
strain realized niches only in cases where we assumed every disturbance led to a
CONCLUSIONS 177
One of the fundamental premises of this book was the focus on large scales, spe-
cifically on regional to global patterns and drivers of biodiversity. Patterns and
drivers of biodiversity do change across scales, however, likely becoming more
idiosyncratic at smaller grain sizes, as local variations in ecological interactions
and processes become more important in selecting from the regional species pool.
178 CHAPTER 7
For land vertebrates, for example, it has been shown that the predictability of spe-
cies richness patterns decreases sharply at smaller scales and that temperature and
moisture as major drivers of species richness are replaced by other factors, such
as productivity and environmental heterogeneity (Belmaker and Jetz 2011). This
means that our theory at present captures some global-scale processes at coarse
resolution quite well, but would need to be modified to apply at smaller scales.
One advantage of our metacommunity model is that it is easily scalable. While
we considered only a tractable number of local communities (<1000), each of
which was unrealistically small, there is no reason, other than computational
cost, why this could not be scaled up to include many thousands of local com-
munities, each of which could include large numbers of individuals. Our limited
explorations in chapter 4 suggested that increasing the number of individuals in
each local community does not qualitatively change the observed patterns (see
also Tittensor and Worm 2016). Fitting a conceptually similar model to local or
regional data and using realistic dispersal constraints would allow a closer look
at processes operating at a variety of scales. As such, this approach could perhaps
help to illuminate and test hypotheses about cross-scale variation in the processes
and drivers of species richness.
Several authors have recently called upon ecologists to better integrate ecological
and evolutionary processes in the study of communities and ecosystems (Fritz
et al. 2013; Mannion et al. 2014; Weigelt et al. 2016). Our theory attempts such
integration in an effort to explain global biodiversity patterns (see fig. 7.1). Within
this framework, and over evolutionary timescales, assumed thermal effects on
speciation emerge as the most general process that determines the speed at which
new species evolve. At ecological timescales, habitat area and productivity adjust
the number of individuals that can be supported, and thereby affect the number of
species in a community. This also relates to spatial scale: at local scales, contem-
porary ecological drivers appear important in determining community size and
selecting from the regional species pool. At larger scales, the size of the species
pool, as determined by evolutionary drivers, becomes more important (Vellend
2016). This change from ecological to evolutionary drivers with increasing spatial
scale might help to explain the empirically observed shift from productivity and
habitat at smaller scales to temperature as the major environmental predictor at
larger scales (see fig. 3.2 and data in Belmaker and Jetz 2011).
Biogeographically, the global distribution of thermal energy from a hot equa-
tor to cold poles is determined by the angle of the planet toward the sun, which
CONCLUSIONS 179
7.6. APPLICATIONS
7.7. LIMITATIONS
resulting in zero-sum dynamics (Hubbell 2001). Thus, our approach does not
explicitly include trophic interactions such as predation or competition (though
mortality events may implicitly be caused by either). Other effects that are miss-
ing, but are known to be important, include the effects of habitat complexity and
body size. We propose that these are reasonable assumptions and limitations for a
first iteration of our model framework applied to very large scales.
Fully neutral models also have recognized limitations or assumptions in terms
of their simplified representation of speciation, assumptions of species equiva-
lence, evolutionary rates, and variable fit to empirical patterns (Bell 2001; Gravel
et al. 2006; Rosindell et al. 2011; Marquet et al. 2014). In part, such limitations
have provided incentive for us to relax the assumption of neutrality and to include
potentially important differences among species—for example, with respect to
their thermal tolerances. At the same time, we agree with other commentators
who have pointed out that grossly simplified models can be very useful tools even
when some of their assumptions are “wrong” (as with any model), by helping to
inform us which processes may be important (Marquet et al. 2014).
Other inherent limitations of our model framework relate to abstractions in time
and space. In particular, we do not attempt to define a specific temporal period
that corresponds to each model time step. This is especially true for our modi-
fied version of Hubbell’s model, which features probabilistic selection of single
local communities for disturbance, rather than simultaneous or probabilistic dis-
turbance of all communities. Our representation of time and time-specific rates is
thus abstracted, and not easily comparable to empirically measured time steps or
rates. Our model is also running on a coarse spatial grid, but using very small local
communities, mostly for reasons of computational efficiency. At such scales, and
particularly when fitting to a global grid, the meaning of a model “individual” does
not equate to an individual in the real world, but more likely to a population. While
we have tested both finer resolution metacommunities and larger local communi-
ties, we recognize that neither of these scenarios is strictly “realistic.” The same
goes for our parameter values of speciation rate and dispersal rate. Because both
are scaled relative to a theoretical time step, of unspecified length, these rates do
not yet have solid empirical grounding, but serve as tools for exploring fundamen-
tal processes. Again, we emphasize that this is a theoretical model that strives more
for simplicity, tractability, and understanding rather than realism.
In summary, the degree of valid inference is limited by the simple nature of the
mechanistic model that we derived. The emphasis here is on first-order patterns
and processes, and model assumptions remain to be scrutinized or challenged by
future work. Nonetheless, at present, no equivalent model exists, so we under-
stand this as a “first step” rather than a finish line. The model in its simplicity and
generality can be extended in many ways—for example, to include other factors
182 CHAPTER 7
How does species richness vary across the planet, and can we predict some of the
observed patterns from simple ecological theory and first principles? In this vol-
ume, we have developed a synthetic empirical understanding of global patterns
in species richness and developed a theory that can predict first-order patterns of
species richness on land and in the sea. Much work remains to be done in testing,
revising, and extending this theory. In closing, we point out some of the further
research questions we see emerging from this work.
Clearly, more detailed analysis should be conducted to explain spatial patterns
of biodiversity, especially at finer spatial grains. Given the greater availability
of data on land and recent integration in global databases, it may be feasible to
understand patterns and mechanisms in finer detail than in the ocean. This might
also provide new insights about changing drivers and correlates of diversity across
scales (Belmaker and Jetz 2011) and the factors that shape the exchange between
local, regional, and global species pools (Vellend 2010, 2016).
Another open question is how the speed of evolution in endotherms and ecto-
therms is related to environmental conditions. How do gradients in temperature
or productivity affect these groups similarly or differently? It is interesting, for
example, that vascular plants (see fig. 2.8) and mammals on land (see fig. 2.9)
show very similar geographic patterns of biodiversity, whereas those patterns
diverge in the oceans for the same species groups (compare seagrasses and man-
groves [see fig. 2.3] to cetaceans and pinnipeds [see figs. 2.4 and 2.6]). Yet in both
environments, these groups strongly relate to gradients in surface temperature,
albeit in different ways (see fig. 3.4). Clearly, the mechanistic basis of these rela-
tionships is not well resolved, especially for endotherms, but is an area of active
research (Gillman and Wright 2014).
More broadly, our model could also be used to explore other macroecologi-
cal patterns and biodiversity attributes in greater detail, such as species-area or
range-abundance relationships. We have touched on this in our investigations of
Rapoport’s rule (see fig. 4.15), but clearly there is much more that can be done.
Our strong focus on global richness has not allowed for much detailed exploration
of other facets of biodiversity, but such work could potentially be accomplished
with the tools and data at hand.
It would also be fascinating to examine past patterns of species richness under
previous environmental regimes, and to test some of the ideas that are emerging
CONCLUSIONS 183
Abele, L. G., and W. K. Patton. (1976). “The Size of Coral Heads and the Community
Biology of Associated Decapod Crustaceans.” Journal of Biogeography 3: 35–47.
Abele, L. G., and K. Walters. (1979). “The Stability-Time Hypothesis: Reevaluation of the
Data.” American Naturalist 114: 559–568.
Adler, P. B., J. HilleRisLambers, and J. M. Levine. (2007). “A Niche for Neutrality.” Ecol-
ogy Letters 10: 95–104.
Adler, P. B., E. W. Seabloom, E. T. Borer, H. Hillebrand, Y. Hautier, A. Hector, W. S. Har-
pole, L. R. O’Halloran, J. B. Grace, T. M. Anderson, J. D. Bakker et al. (2011). “Pro-
ductivity Is a Poor Predictor of Plant Species Richness.” Science 333, 1750–1753.
Agrawal, A. F., and A. D. Wang. (2008). “Increased Transmission of Mutations by Low-
condition Females: Evidence for Condition-dependent DNA Repair.” PLoS Biology
6: e30.
Airoldi, L., and M. W. Beck. (2007). “Loss, Status and Trends for Coastal Marine Habitats
of Europe.” In Oceanography and Marine Biology, vol. 45, ed. R. N. Gibson, R.J.A.
Atkinson, and J.D.M. Gordon. Boca Raton, FL: CRC Press-Taylor, and Francis
Group, 345–405.
Algar, A. C., J. T. Kerr, and D. J. Currie. (2007). “A Test of Metabolic Theory as the
Mechanism Underlying Broad-scale Species-richness Gradients.” Global Ecology
and Biogeography 16: 170–178.
Allen, A. P., J. H. Brown, and J. F. Gillooly. (2002). “Global Biodiversity, Biochemical
Kinetics, and the Energetic-Equivalence Rule.” Science 297: 1545–1548.
Allen, A. P., and J. F. Gillooly. (2006). “Assessing Latitudinal Gradients in Speciation
Rates and Biodiversity at the Global Scale.” Ecology Letters 9: 947–954.
Allen, A. P., J. F. Gillooly, V. M. Savage, and J. H. Brown. (2006). “Kinetic Effects of
Temperature on Rates of Genetic Divergence and Speciation.” Proceedings of the
National Academy of Sciences USA 103: 9130–9135.
Allen, M. R., V. R. Barros, J. Broome, W. Cramer, R. Christ, J. A. Church, L. Clarke,
Q. Dahe, P. Dasgupta, and N. K. Dubash. (2014). IPCC Fifth Assessment Synthesis
Report—Climate Change 2014 Synthesis Report. Geneva: Intergovernmental Panel
on Climate Change.
Alroy, J., M. Aberhan, D. J. Bottjer, M. Foote, F. T. Fursich, P. J. Harries, A.J.W. Hendy,
S. M. Holland, L. C. Ivany, W. Kiessling, M. A. Kosnik et al. (2008). “Phanerozoic
Trends in the Global Diversity of Marine Invertebrates.” Science 321: 97–100.
Amend, A. S., T. A. Oliver, L. A. Amaral-Zettler, A. Boetius, J. A. Fuhrman, M. C. Horner-
Devine, S. M. Huse, D.B.M. Welch, A. C. Martiny, and A. Ramette. (2013). “Macro-
ecological Patterns of Marine Bacteria on a Global Scale.” Journal of Biogeography
40: 800–811.
Angel, M. V. (1993). “Biodiversity of the Pelagic Ocean.” Conservation Biology 7:
760–772.
186 REFERENCES
Chisholm, R. A., and S. W. Pacala. (2010). “Niche and Neutral Models Predict Asymp-
totically Equivalent Species Abundance Distributions in High-diversity Ecological
Communities.” Proceedings of the National Academy of Sciences 107: 15821–15825.
Chust, G., X. Irigoien, J. Chave, and R. P. Harris. (2012). “Latitudinal Phytoplankton Dis-
tribution and the Neutral Theory of Biodiversity.” Global Ecology and Biogeogra-
phy; 22:531–543.
Clarke, A., and K. J. Gaston. (2006). “Climate, Energy and Diversity.” Proceedings of the
Royal Society: Series B 273: 2257–2266.
Colwell, R. K., and D. C. Lees. (2000). “The Mid-domain Effect: Geometric Constraints on
the Geography of Species Richness.” Trends in Ecology and Evolution 15: 70–76.
Comes, H. P., and J. W. Kadereit. (1998). “The Effect of Quaternary Climatic Changes on
Plant Distribution and Evolution.” Trends in Plant Science 3: 432–438.
Connell, J. H. (1978). “Diversity in Tropical Rain Forests and Coral Reefs.” Science 199:
1302–1310.
Connell, J. H., and E. Orias. (1964). “The Ecological Regulation of Species Diversity.”
American Naturalist 98: 399–441.
Connor, E. F., A. C. Courtney, and J. M. Yoder. (2000). “Individuals-Area Relationships: The
Relationship between Animal Population Density and Area.” Ecology 81: 734–748.
Connor, E. F., and E. D. McCoy. (1979). “The Statistics and Biology of the Species-Area
Relationship.” American Naturalist 113: 791–833.
Corkrey, R., T. A. McMeekin, J. P. Bowman, D. A. Ratkowsky, J. Olley, and T. Ross.
(2014). “Protein Thermodynamics Can Be Predicted Directly from Biological
Growth Rates.” PLoS ONE 9: e96100.
Cotton, C. M. (1996). Ethnobotany: Principles and Applications. New York: John Wiley
& Sons.
Cowen, R. K, C. B. Paris, and A. Srinivasan. (2006). “Scaling of Connectivity in Marine
Populations.” Science 311: 522–527.
Cowling, R. M., R. L. Pressey, M. Rouget, and A. T. Lombard. (2003). “A Conservation
Plan for a Global Biodiversity Hotspot—the Cape Floristic Region, South Africa.”
Biological Conservation 112: 191–216.
Crame, J. A. (2000). “Evolution of Taxonomic Diversity Gradients in the Marine Realm:
Evidence from the Composition of Recent Bivalve Faunas.” Paleobiology 26:
188–214.
Crawley, M. J., and J. E. Harral. (2001). “Scale Dependence in Plant Biodiversity.” Science
291: 864–868.
Culver, S. J., and M. A. Buzas. (2000). “Global Latitudinal Species Diversity Gradient
in Deep-sea Benthic Foraminifera.” Deep Sea Research Part I: Oceanographic
Research Papers 47: 259–275.
Currie, D. J. (1991). “Energy and Large-scale Patterns of Animal and Plant Species Rich-
ness.” American Naturalist 137: 27–49.
Currie, D. J., and J. T. Kerr. (2008). “Tests of the Mid-domain Hypothesis: A Review of the
Evidence.” Ecological Monographs 78: 3–18.
Darwin, C. (1859). The Origin of Species. Modern Library ed. New York: Random House.
Davies, R. G., U. M. Irlich, S. L. Chown, and K. J. Gaston. (2010). “Ambient, Productive
and Wind Energy, and Ocean Extent Predict Global Species Richness of Procellari-
iform Seabirds.” Global Ecology and Biogeography 19: 98–110.
Davison, J., M. Moora, M. Öpik, A. Adholeya, L. Ainsaar, A. Bâ, S. Burla, A. G. Died-
hiou, I. Hiiesalu, T. Jairus, N. C. Johnson et al. (2015). “Global Assessment of
190 REFERENCES
Gillooly, J. F., A. P. Allen, G. B. West, and J. H. Brown. (2005). “The Rate of DNA Evolu-
tion: Effects of Body Size and Temperature on the Molecular Clock.” Proceedings
of the National Academy of Sciences USA 102: 140–145.
Gillooly, J. F., J. H. Brown, G. B. West, V. M. Savage, and E. L. Charnov. (2001). “Effects
of Size and Temperature on Metabolic Rate.” Science 293: 2248–2251.
Godfray, H.C.J., J. R. Beddington, I. R. Crute, L. Haddad, D. Lawrence, J. F. Muir, J. Pretty,
S. Robinson, S. M. Thomas, and C. Toulmin. (2010). “Food Security: The Challenge
of Feeding 9 Billion People.” Science 327: 812–818.
Gotelli, N. J., M. J. Anderson, H. T. Arita, A. Chao, R. K. Colwell, S. R. Connolly, D. J.
Currie, R. R. Dunn, G. R. Graves, and J. L. Green. (2009). “Patterns and Causes of
Species Richness: A General Simulation Model for Macroecology.” Ecology Letters
12: 873–886.
Grassle, J. F., and N. J. Maciolek. (1992). “Deep-sea Species Richness: Regional and Local
Diversity Estimates from Quantitative Bottom Samples.” American Naturalist 139:
313–341.
Grassle, J. F., and H. L. Sanders. (1973). “Life Histories and the Role of Disturbance.”
Deep Sea Research and Oceanographic Abstracts 20: 643–659.
Grassle, J. F., and K. I. Stocks. (1999). “A Global Ocean Biogeographic Information Sys-
tem (OBIS) for the Census of Marine Life.” Oceanography 12: 12–14.
Gravel, D., C. D. Canham, M. Beaudet, and C. Messier. (2006). “Reconciling Niche and
Neutrality: The Continuum Hypothesis.” Ecology Letters 9: 399–409.
Gravel, D., F. Guichard, and M. E. Hochberg. (2011). “Species Coexistence in a Variable
World.” Ecology Letters 14: 828–839.
Gregg, W. W., M. E. Conkright, Paul Ginoux, J. E. O’Reilly, and Nancy W. Casey. (2003).
“Ocean Primary Production and Climate: Global Decadal Changes.” Geophysical
Research Letters 30: 1909–1813.
Grenyer, R., C.D.L. Orme, S. F. Jackson, G. H. Thomas, R. G. Davies, T. J. Davies, K. E.
Jones, V. A. Olson, R. S. Ridgely, P. C. Rasmussen, T.-S. Ding et al. (2006). “Global
Distribution and Conservation of Rare and Threatened Vertebrates.” Nature 444:
93–96.
Grime, J. P. (1973). “Competitive Exclusion in Herbaceous Vegetation.” Nature 242:
344–347.
Grimm, V., and C. Wissel. (1997). “Babel, or the Ecological Stability Discussions: An
Inventory and Analysis of Terminology and a Guide for Avoiding Confusion.”
Oecologia 109: 323–334.
Haedrich, R. L., and G. T. Rowe. (1977). “Megafaunal Biomass in the Deep Sea.” Nature
269: 141–142.
Halpern, B., M. Frazier, J. Potapenko, K. Casey, K. Koenig, et al. 2015. Cumulative
Human Impacts: Supplementary Data. Knowledge Network for Biocomplexity.
doi:10.5063/F19Z92TW.
Halpern, B. S., S. Walbridge, K. A. Selkoe, C. V. Kappel, F. Micheli, C. D’Agrosa, J. F.
Bruno, K. S. Casey, C. Ebert, H. E. Fox, R. Fujita et al. (2008). “A Global Map of
Human Impact on Marine Ecosystems.” Science 319: 948–952.
Haney, J. C. (1986). “Seabird Aggregation at Gulf Stream Frontal Eddies.” Marine Ecol-
ogy Progress Series 28: 279–285.
Hansen, J., R. Ruedy, M. Sato, and K. Lo. (2010). “Global Surface Temperature Change.”
Reviews of Geophysics 48; doi:10.1029/2010RG000345.
REFERENCES 193
Hoegh-Guldberg, O. (1999). “Climate Change, Coral Bleaching and the Future of the
World’s Coral Reefs.” Marine and Freshwater Research 50: 839–866.
Hoffmann, A. A., and C. M. Sgrò. (2011). “Climate Change and Evolutionary Adaptation.”
Nature 470: 479–485.
Holbrook, S. J., R. J. Schmitt, and J. S. Stephens Jr.. (1997). “Changes in an Assemblage of
Temperate Reef Fishes Associated with a Climatic Shift.” Ecological Applications
7: 1299–1310.
Honjo, S., and H. Okada. (1974). “Community Structure of Coccolithophores in the Photic
Layer of the Mid-Pacific.” Micropaleontology 20: 209–230.
Hooper, D. U., E. C. Adair, B. J. Cardinale, J.E.K. Byrnes, B. A. Hungate, K. L. Matulich,
A. Gonzalez, J. E. Duffy, L. Gamfeldt, and M. I. O’Connor. (2012). “A Global Syn-
thesis Reveals Biodiversity Loss As a Major Driver of Ecosystem Change.” Nature
486: 105–108.
Hubbell, S. P. (2001). The Unified Neutral Theory of Biodiversity and Biogeography.
Princeton, NJ: Princeton University Press.
Hughes, T. P., J. T. Kerry, M. Álvarez-Noriega, J. G. Álvarez-Romero, K. D. Ander-
son, A. H. Baird, R. C. Babcock, M. Beger, D. R. Bellwood, and R. Berkelmans.
(2017). “Global Warming and Recurrent Mass Bleaching of Corals.” Nature 543:
373–377.
Hurlbert, A. H., and W. Jetz. ( 2010). “More Than ‘More Individuals’: The Nonequivalence
of Area and Energy in the Scaling of Species Richness.” American Naturalist 176:
E50–E65.
Huston, M. A. (1979). “A General Hypothesis of Species Diversity.” American Naturalist
113: 81–101.
———. (1994). Biological Diversity. Cambridge, UK: Cambridge University Press.
Hutchings, P., and P. Saenger. (1987). Ecology of Mangroves. St. Lucia, Australia:
Queensland University Press.
Hutchinson, G. E. (1957). “The Multivariate Niche.” Cold Spring Harbor Symposium
Quantitative Biology 22: 415–421.
———. (1959). “Homage to Santa Rosalia, or Why Are There So Many Kinds of Ani-
mals.” American Naturalist 93: 145–159.
Imhoff, M. L., L. Bounoua, T. Ricketts, C. Loucks, R. Harris, and W. T. Lawrence. (2004).
“Global Patterns in Human Consumption of Net Primary Production.” Nature 429:
870–873.
International Union for the Conservation of Nature (IUCN). (2016). Redlist Spatial Data.
Cambridge, UK: International Union for the Conservation of Nature Redlist Unit.
Available at http://www.iucnredlist.org/technical-documents/spatial-data.
Jablonski, D., C. L. Belanger, S. K. Berke, S. Huang, A. Z. Krug, K. Roy, A. Tomaso-
vych, and J. W. Valentine. (2013). “Out of the Tropics, but How? Fossils, Bridge
Species, and Thermal Ranges in the Dynamics of the Marine Latitudinal Diver-
sity Gradient.” Proceedings of the National Academy of Sciences 110: 10487–
10494.
Jablonski, D., K. Roy, and J. W. Valentine. (2006). “Out of the Tropics: Evolutionary
Dynamics of the Latitudinal Diversity Gradient.” Science 314: 102–106.
Jamieson, A. J., T. Malkocs, S. B. Piertney, T. Fujii, and Z. Zhang. (2017). “Bioaccumula-
tion of Persistent Organic Pollutants in the Deepest Ocean Fauna.” Nature Ecology
and Evolution 1: e0051.
REFERENCES 195
Micheli, F., and B. S. Halpern. (2005). “Low Functional Redundancy in Coastal Marine
Assemblages.” Ecology Letters 8: 391–400.
Miller, S. L., and L. E. Orgel. (1974). The Origins of Life on the Earth. Englewood Cliffs,
NJ: Prentice-Hall.
Milner-Gulland, E. J., E. L. Bennett, and SCB 2002 Annual Meeting Wild Meat Group.
(2003). “Wild Meat: The Bigger Picture.” Trends in Ecology and Evolution 18:
351–357.
Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross, L. H. Reynolds, R. B. Waide,
M. R. Willig, S. I. Dodson, and L. Gough. (2001). “What Is the Observed Relation-
ship between Species Richness and Productivity?” Ecology 82: 2381–2396.
Molinos, J. G., B. S. Halpern, D. S. Schoeman, C. J. Brown, W. Kiessling, P. J. Moore,
J. M. Pandolfi, E. S. Poloczanska, A. J. Richardson, and M. T. Burrows. (2016).
“Climate Velocity and the Future Global Redistribution of Marine Biodiversity.”
Nature Climate Change 6: 83–88.
Mongold, J. A., A. F. Bennett, and R. E. Lenski. (1996). “Evolutionary Adaptation to Tem-
perature. IV. Adaptation of Escherichia Coli at a Niche Boundary.” Evolution 35–43.
Mora, C., A. G. Frazier, R. J. Longman, R. S. Dacks, M. M. Walton, E. J. Tong, J. J. San-
chez, L. R. Kaiser, Y. O. Stender, J. M. Anderson, C. M. Ambrosino et al. (2013).
“The Projected Timing of Climate Departure from Recent Variability.” Nature 502:
183–187.
Mora, C., D. P. Tittensor, S. Adl, A.G.B. Simpson, and B. Worm. (2011). “How Many Spe-
cies Are There on Earth and in the Ocean?” PLoS Biology 9: e1001127.
Mora, C., D. P. Tittensor, and R. A. Myers. (2008). “The Completeness of Taxonomic
Inventories for Describing the Global Diversity and Distribution of Marine Fishes.”
Proceedings of the Royal Society B: Biological Sciences 275: 149–155.
Mulkidjanian, A. Y., A. Y. Bychkov, D. V. Dibrova, M. Y. Galperin, and E. V. Koonin.
(2012). “Origin of First Cells at Terrestrial, Anoxic Geothermal Fields.” Proceed-
ings of the National Academy of Sciences 109: E821–E830.
Muller, H. (1928). “The Measurement of Gene Mutation Rate in Drosophila, Its High Vari-
ability, and Its Dependence upon Temperature.” Genetics 13: 279–357.
Mumby, P. J., A. J. Edwards, J. E. Arias-González, K. C. Lindeman, P. G. Blackwell,
A. Gall, M. I. Gorczynska, A. R. Harborne, C. L. Pescod, H. Renken, C.C.C. Wab-
nitz, and G. Llewellyn. (2004). “Mangroves Enhance the Biomass of Coral Reef
Fish Communities in the Caribbean.” Nature 427: 533–536.
Myers, N., R. A. Mittermeier, C. G. Mittermeier, G.A.B. da Fonseca, and J. Kent. (2000).
“Biodiversity Hotspots for Conservation Priorities.” Nature 403: 853–858.
Nemani, R. R., C. D. Keeling, H. Hashimoto, W. M. Jolly, S. C. Piper, C. J. Tucker, R. B.
Myneni, and S. W. Running. (2003). “Climate-driven Increases in Global Terrestrial
Net Primary Production from 1982 to 1999.” Science 300: 1560–1563.
Newbold, T., L. N. Hudson, S. L. Hill, S. Contu, I. Lysenko, R. A. Senior, L. Börger, D. J.
Bennett, A. Choimes, and B. Collen. (2015). “Global Effects of Land Use on Local
Terrestrial Biodiversity.” Nature 520: 45–50.
Olson, D. B., G. L. Hitchcock, A. J. Mariano, C. J. Ashjan, G. Peng, R. W. Nero, and
G. P. Podesta. (1994). “Life on the Edge: Marine Life and Fronts.” Oceanography
7: 52–60.
Öpik, H., and S. A. Rolfe. (2005). The Physiology of Flowering Plants. Cambridge, UK:
Cambridge University Press.
REFERENCES 199
Rutherford, S., S. D’Hondt, and W. Prell. (1999). “Environmental Controls on the Geo-
graphic Distribution of Zooplankton Diversity.” Nature 400: 749–753.
Safi, K., M. V. Cianciaruso, R. D. Loyola, D. Brito, K. Armour-Marshall, and J.A.F.
Diniz-Filho. (2011). “Understanding Global Patterns of Mammalian Functional and
Phylogenetic Diversity.” Philosophical Transactions of the Royal Society B: Bio-
logical Sciences 366: 2536–2544.
Sala, O. E., F. S. Chapin III, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-
Sanwald, L. F. Huenneke, R. B. Jackson, A. Kinzig, R. Leemans et al. (2000).
“Global Biodiversity Scenarios for the Year 2100.” Science 287: 1770–1774.
Sanders, H. L. (1968). “Marine Benthic Diversity: A Comparative Study.” American Natu-
ralist 102: 243–282.
Sarmiento, J. L., R. Slater, R. Barber, L. Bopp, S. C. Doney, A. C. Hirst, J. Kleypas,
R. Matear, U. Mikolajewicz, P. Monfray, V. Soldatov et al. (2004). “Response
of Ocean Ecosystems to Climate Warming.” Global Biogeochemical Cycles 18;
doi:10.1029/2003GB002134.
Schipper, J., J. S. Chanson, F. Chiozza, N. A. Cox, M. Hoffmann, V. Katariya, J. Lamoreux,
A.S.L. Rodrigues, S. N. Stuart, H. J. Temple, J. Baillie et al. (2008). “The Status of
the World’s Land and Marine Mammals: Diversity, Threat, and Knowledge.” Sci-
ence 322: 225–230.
Schlüter, L., K. T. Lohbeck, M. A. Gutowska, J. P. Gröger, U. Riebesell, and T. B. Reusch.
(2014). “Adaptation of a Globally Important Coccolithophore to Ocean Warming
and Acidification.” Nature Climate Change 4: 1024–1030.
Selig, E. R., W. R. Turner, S. Troëng, B. P. Wallace, B. S. Halpern, K. Kaschner, B. G. Las-
celles, K. E. Carpenter, and R. A. Mittermeier. (2014). “Global Priorities for Marine
Biodiversity Conservation.” PLoS ONE 9: e82898.
Sepkoski, J. J. (1998). “Rates of Speciation in the Fossil Record.” Philosophical Trans-
actions of the Royal Society B: Biological Sciences 353: 315–326.
Short, F. T., B. Polidoro, S. R. Livingstone, K. E. Carpenter, S. Bandeira, J. S. Bujang, H. P.
Calumpong, T.J.B. Carruthers, R. G. Coles, W. C. Dennison, P.L.A. Erftemeijer et
al. (2011). “Extinction Risk Assessment of the World’s Seagrass Species.” Biologi-
cal Conservation 144, 1961–1971.
Silvertown, J., M. E. Dodd, K. McConway, J. Potts, and M. Crawley. (1994). “Rainfall,
Biomass Variation, and Community Composition in the Park Grass Experiment.”
Ecology 75: 2430–2437.
Sinervo, B., F. Méndez-de-la-Cruz, D. B. Miles, B. Heulin, E. Bastiaans, M. Villagrán-
Santa Cruz, R. Lara-Resendiz, N. Martínez-Méndez, M. L. Calderón-Espinosa,
R. N. Meza-Lázaro, H. Gadsden et al. (2010). “Erosion of Lizard Diversity by Cli-
mate Change and Altered Thermal Niches.” Science 328: 894–899.
Sommer, J. H., H. Kreft, G. Kier, W. Jetz, J. Mutke, and W. Barthlott. (2010). “Pro-
jected Impacts of Climate Change on Regional Capacities for Global Plant Spe-
cies Richness.” Proceedings of the Royal Society of London B: Biological Sciences;
doi:10.1098/rspb.2010.0120.
Sommer, U., and B. Worm. (2002). Competition and Coexistence. Berlin: Springer.
Sousa, W. P. (1979). “Disturbance in Marine Intertidal Boulder Fields: The Nonequilib-
rium Maintenance of Species Diversity.” Ecology 49: 1225–1239.
Stehli, F. G., R. G. Douglas, and N. D. Newell. (1969). “Generation and Maintenance of
Gradients in Taxonomic Diversity.” Science 164: 947–949.
REFERENCES 203
Worm, B., and H. S. Lenihan. (2013). “Threats to Marine Ecosystems: Overfishing and
Habitat Degradation.” In Marine Community Ecology and Conservation, ed. M. D.
Bertness. New York: Sinauer, 449–476.
Worm, B., H. K. Lotze, H. Hillebrand, and U. Sommer. (2002). “Consumer versus Resource
Control of Species Diversity and Ecosystem Functioning.” Nature 417: 848–851.
Worm, B., H. K. Lotze, I. Jonsen, and C. Muir. (2010). “The Future of Marine Animal
Populations.” In Life in the World’s Oceans: Diversity, Distribution and Abundance,
ed. A. McIntyre. Oxford, UK: Blackwell, 315–330.
Worm, B., H. K. Lotze, and R. A. Myers. (2003). “Predator Diversity Hotspots in the Blue
Ocean.” Proceedings of the National Academy of Sciences USA 100: 9884–9888.
Worm, B., and H. K. Lotze. (2015). “Marine Biodiversity and Climate Change.” In Climate
Change, ed. T. Letcher. Amsterdam: Elsevier, 195–212.
Worm, B., M. Sandow, A. Oschlies, H. K. Lotze, and R. A. Myers. (2005). “Global Patterns
of Predator Diversity in the Open Oceans.” Science 309: 1365–1369.
Worm, B., and D. P. Tittensor. (2011). “Range Contraction in Large Pelagic Predators.”
Proceedings of the National Academy of Sciences 108: 11942–11947.
Wright, D. H. (1983). “Species-Energy Theory: An Extension of Species-Area Theory.”
Oikos 41: 496–506.
Yasuhara, M., G. Hunt, H. J. Dowsett, M. M. Robinson, and D. K. Stoll. (2012). “Latitu-
dinal Species Diversity Gradient of Marine Zooplankton for the Last Three Million
Years.” Ecology Letters 15: 1174–1179.
Yasuhara, M., D. P. Tittensor, H. Hillebrand, and B. Worm. (2015). “Combining Marine
Macroecology and Palaeoecology in Understanding Biodiversity: Microfossils as a
Model.” Biological Reviews; 92: 199–215.
Zachos, J. C., G. R. Dickens, and R. E. Zeebe. (2008). “An Early Cenozoic Perspective on
Greenhouse Warming and Carbon-cycle Dynamics.” Nature 451: 279–283.
Zapata, F. A., K. J. Gaston, and S. L. Chown. (2005). “The Mid-domain Effect Revisited.”
American Naturalist 166: E144–E148.
Index
carnivores: sensitivity of biodiversity patterns 119; species richness of, 21, 22; synthesis of
to taxonomic resolution, 46–47 global biodiversity patterns, 52; univariate
cats, sensitivity of biodiversity patterns to relationship of species richness and sea sur-
taxonomic resolution, 45 face temperature, 83. See also pelagic sharks
Census of Marine Life, 7 coastal vertebrates: biodiversity of, 20–23;
cephalopods, pelagic, 28. See also coastal bony fishes, 20–21, 22; marine mammals,
cephalopods 23; pinnipeds, 22, 23; sharks, 21, 22; species
cetaceans, 182; environmental predictors of richness patterns, 22. See also coastal sharks
species richness, 78; global biodiversity coccolithophores, 27
data fit, 129; predicting biodiversity of community size, 94, 138; coastal biodiversity
endotherms, 140; species richness of, 29, 30; and, 127, 128; deep-sea biodiversity and,
synthesis of global biodiversity patterns, 53; 132–33; ecological theory, 174–75; frontal
univariate relationship of species richness habits and, 143; global species richness
and sea surface temperature, 83 patterns and, 176; habitat area and, 121, 125,
Challenger expedition, 5, 32 162, 169; local, 97–99, 109, 111, 115, 125,
clades, 13 146; metacommunity, 99, 103, 106, 128;
climate change, biodiversity changes by, 157–59 model implementation, 94, 156; optimal
coalescence approach, 95; implementation of, scaling of, 127; pelagic biodiversity and,
101–4; implementation of thermal niche in, 132, 133, 134; productivity and, 121, 125,
114; latitudinal biodiversity gradient, 108; 160, 167, 169; temperature and, 173–74;
transient biodiversity gradient comparison, terrestrial biodiversity and, 135–38
106; simulation, 96, 106, 107, 108, 127; community turnover, 62; in forward mode, 101,
temporal process variation, 179 102; hypothesized diversity mechanism, 58,
coastal biodiversity: hotspots of species rich- 59; in metabolic theory, 104–6; model imple-
ness across taxa, 154; model fit to global mentation, 156, 162; temperature and, 108,
biodiversity data, 129; predicting, 130; 110, 126, 157, 168, 173; thermal energy and,
predictions, 127–32; sensitivity of predic- 58, 94; transient biodiversity gradient, 106
tions for, 131 cone snails: environmental predictors of spe-
coastal cephalopods: environmental predictors cies richness, 91; global biodiversity data fit,
of species richness, 77, 91; global biodiver- 129; species richness patterns of, 16, 17, 18;
sity data fit, 129; species richness patterns, synthesis of global biodiversity patterns, 52;
16, 17, 18; synthesis of global biodiversity univariate relationship of species richness
patterns, 52; univariate relationship of spe- and sea surface temperature, 86
cies richness and sea surface temperature, 84 connectivity, global impact on biodiversity, 156
coastal ectotherms: changes in future richness conservation biology: applications of eco-
of, 168; predicting coastal biodiversity, 130; logical theory in, 149–50, 179–80; hotspot
sensitivity of predictions for, 131 approach to priorities, 150–55; ongoing
coastal habitats: biodiversity of invertebrates, biodiversity change and its drivers, 155–62;
13–18; biodiversity of plants, 18–20; biodi- projecting biodiversity change from theory,
versity of vertebrates, 20–23; definition of, 162–69; species-rich hotspots, 151–53
12; features of, 12–13; synthesis of global Continuous Plankton Recorder, 27
biodiversity patterns, 52; synthesis of latitu- Convention on Biological Diversity, 170
dinal biodiversity patterns, 54 copepods: species richness of, 27; synthesis of
coastal invertebrates: biodiversity, 13–18; spe- global biodiversity patterns, 52
cies richness of, 16–17, 87–88 corals, 43, 97; environmental predictors of
coastal plants: biodiversity, 18–20; spe- species richness, 77, 91; global biodiversity
cies richness of, 19. See also mangroves; data fit, 129; species richness of, 15, 16–17,
seagrasses 18; synthesis of global biodiversity patterns,
coastal sharks: environmental predictors of 52; temperature and climate change, 158;
species richness, 77, 91; global biodiversity univariate relationship of species richness
data fit, 129; ranges and thermal niches in, and sea surface temperature, 85
INDEX 209
coral triangle, 15, 17, 153 predictors of diversity, 58; limiting diversity,
crayfish: environmental predictors of species 58, 66–70; more individuals hypothesis,
richness, 89, 91; synthesis of global bio- 57, 59, 65, 93, 109; observed biodiversity
diversity patterns, 53 change and, 155–62; productivity as, 58,
64–65; promoting diversity, 58, 59–66;
Darwin, Charles, 3, 6 relating to habitat, 58, 71–73; seasonality as,
deep ocean: environmental predictors of spe- 58, 70; solar energy as, 59–60; spatial scale
cies richness in, 81 of biodiversity, 177–78; surface temperature
deep-sea biodiversity, 31–35; global patterns, as, 58, 60–64
33–34; model fit to global biodiversity data,
129; predicting, 135; predictions, 132–33; Earth: greenhouse periods of history, 43
regional studies, 34–35; sensitivity of predic- Earth System Models, 168
tions for, 136; synthesis of, 35 ecological equivalents, 97
deep-sea ectotherms: sensitivity of predictions ecological theory: neutral two-species com-
for, 136 munity model, 176; processes influencing
deep-sea habitat: synthesis of latitudinal bio- community ecology, 175–76; separate bodies
diversity patterns, 54 of, 173; synthesis in, 172
deep-sea species: ambient temperature as ecological timescale, 55, 59, 75, 76, 157, 177,
predictor, 85–87; benthic fauna, 34; benthic 178; evolutionary vs., 178–79
nematodes, 34 efficient theory, 94
diatoms, 24, 27, 43 Encyclopedia of Life, 7
dispersal, 76, 89; capacities for, 18, 20–21, environmental predictors: changes in, with
43; in coastal biodiversity, 131; in deep-sea depth, 79; of diversity, 76, 82–88; special
biodiversity, 136; ecological theory, 172, richness in pelagic realm, 78; species rich-
175; effect on global gradients in species ness in coastal realm, 77, 91; of species
richness, 118; effect of niches on, 142–43; richness in deep ocean, 81, 91; of species
influencing community ecology, 175–76; richness in surface ocean, 80, 91; of spe-
limitation, 31, 43–44, 51, 121, 123, 164; in cies richness on land, 81–82, 88–90, 91; of
metacommunity model, 94–104, 111, 112, vascular plant species richness on land, 79,
124; in model implementation, 156; neutral- 91; synthesis of 89–91; vertebrate species
metabolic-niche model, 114–17; in pelagic richness across scales, 74–75
biodiversity, 132, 134; period, 13; rates, 27, environmental stability: driver promoting
71, 104, 111, 112, 114, 117, 121, 127, 146, diversity, 58, 65–66
162; in terrestrial biodiversity, 139 environmental stress: definition of, 66; drivers
disturbance: biodiversity changes, 160–62; limiting diversity, 58, 67–69
drivers limiting diversity, 58, 69–70; inter- Eocene (late middle): biodiversity hotspots
mediate disturbance hypothesis, 70 during, 41, 42
diversity: hypothesized drivers of, 57–73; pat- Escherichia coli, 67
terns of and changes in, 4–5; sensitivity of eukaryotes, 1, 36, 51, 61
vertebrate patterns, to taxonomic resolution, euphausiids: environmental predictors of spe-
46–47; spatial and temporal scales, 73–76. cies richness, 78, 91; global biodiversity data
See also biodiversity fit, 129; species richness of, 25, 26, 27, 132;
domain, 73 synthesis of global biodiversity patterns, 52;
drivers, 56; disturbance as, 58, 69–70; univariate relationship of species richness
environmental, of diversity over deep and sea surface temperature, 86
time, 63; environmental stability as, 58, evolution: process of, 175; “Red Queen” effect
65–66; environmental stress as, 58, 67–69; of, 64
evolutionary time as, 58, 65–66; global evolutionary-speed hypothesis, 59, 60, 93
impacts on biodiversity, 156; habitat area evolutionary time: driver promoting diversity,
as, 58, 71–72; habitat complexity as, 58, 72; 58, 65–66; ecological vs., 178–79; time-
habitat domain as, 58, 73; hypothesized, and scales, 55, 59, 75, 157, 177, 178
210 INDEX
productivity: biodiversity changes, 159–60; selection, 172; ecological theory, 172; influenc-
driver promoting diversity, 58, 64–65, 90, ing community ecology, 175–76
92; ecological vs. evolutionary time, 178–79; sharks. See coastal sharks; pelagic sharks
global impact on biodiversity, 156; modeling snails: environmental predictors of species
effect on species richness, 109, 111, 121; richness, 77. See also cone snails
relationship of number of individuals to, 127 soil fungi: synthesis of global biodiversity
productivity hypothesis, 64–65 patterns, 53
pseudospecies, 164 solar energy: driver promoting diversity, 59–60
published works: species diversity in, 3 spatially explicit metacommunity model, 95–96
speciation, 172; ecological theory, 172, 175;
random drift, 172; influencing community ecol- influencing community ecology, 175–76
ogy, 175–76 species-area relationship (SAR), 71–72
Rapoport’s rule, 68, 118, 120–21, 182 species distribution models (SDMs): bio-
realism, 122, 174, 176, 181 diversity, 169; future projections of marine
red algae: genus richness, 20 species richness, 164; projecting biodiversity
“Red Queen” effect: of evolution, 64 change from, 163–65
regionalization: niches and diversity patterns, species diversity, 2; mentioning, in published
115 books, 3
reptiles: global biodiversity data fit, 129; species-energy hypothesis, 109
species richness of, 38, 39; synthesis of species-energy theory, 59
global biodiversity patterns, 53; terrestrial species richness, 2; coastal invertebrates,
biodiversity, 135, 138 16–17; coastal plants, 18, 19, 20; coastal
vertebrates, 22; conservation of hotspots,
salmon sharks (Lamna ditropis), 21 151–53; in coral triangle, 15, 153; environ-
sampling methods: robustness of biodiversity mental predictors of, in coastal realm, 77,
to, 45–48; sensitivity of fish diversity pat- 91; environmental predictors of, in deep-sea
terns, 48, 49 habitat, 85–87, 91; environmental predictors
Sanders, Howard, 6, 7 of, in freshwater taxa, 89–90, 91; environ-
seabirds: environmental predictors of species mental predictors of, in pelagic realm, 78,
richness, 78; predicting biodiversity of, 142; 91; environmental predictors of, on land,
species richness of, 29, 30; synthesis of 87, 88–89, 91; environmental predictors
global biodiversity patterns, 53; univari- of vertebrate. across scales, 74–75; final
ate relationship of species richness and sea outlook of, 182–83; future projections of
surface temperature, 86. See also pelagic coastal marine, 166; future projections of
seabirds land plant, 167; future projections of marine,
seagrasses: environmental predictors of species 164; hotspot approach to conservation
richness, 77, 91; global biodiversity data prioritization, 150–55; hotspots of, across
fit, 129; species richness of, 18, 19, 20; taxa, 152–53, 154; land plants, 37; land
synthesis of global biodiversity patterns, 52; vertebrates, 38–39; measure of biodiversity,
univariate relationship of species richness 11–12; patterns and changes in diversity,
and sea surface temperature, 85 4–5, 171–74; patterns of vertebrate and
sea lice: diversity of, 34 plants, 89; predictions of, 9; present-day
seasonality: drivers limiting diversity, 58, 70; environmental predictors of, 80–82, 171,
seasonality hypothesis, 70 173; stability-time hypothesis, 6; synthesis
sea surface temperature (SST): as environ- of global patterns, 50–55; temperature and
mental predictor, 80, 84; as environmental metabolic theory for, 62; thermal niche and,
predictor of species richness in coastal 176–77; thermal niches in predicting global
realm, 77; as environmental predictor of spe- biodiversity, 143, 144–45; total, and human
cies richness in pelagic realm, 78; observed impact, 152; univariate relationship of sea
global gradients in, 105; univariate relation- surface temperature and, 83–86. See also
ship of species richness and, 83, 83–86 environmental predictors
214 INDEX