See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/6016891

Species diversity can drive speciation:

Comment
Article in Ecology September 2007
DOI: 10.1890/05-2006.1 Source: PubMed

CITATIONS

READS

17

44

5 authors, including:
Moshe Kiflawi

Eran Brokovich

Ben-Gurion University of the Negev

Israel Society of Ecology and Environmental

34 PUBLICATIONS 763 CITATIONS

42 PUBLICATIONS 747 CITATIONS

SEE PROFILE

SEE PROFILE

Roi Holzman
Tel Aviv University
36 PUBLICATIONS 715 CITATIONS
SEE PROFILE

All content following this page was uploaded by Eran Brokovich on 04 January 2017.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.

Comments
Ecology, 00(0), 0000, pp. 000000
0000 by the Ecological Society of America

SPECIES DIVERSITY CAN DRIVE

SPECIATION: COMMENT
Moshe Kiawi,1,2,4 Jonathan Belmaker,1,2
Eran Brokovich,1,3 Shai Einbinder,1,3 and
Roi Holzman1,3
Introduction
The mechanisms that govern spatial patterns in
species richness, community composition and speciation
are fundamental issues in community and evolutionary
ecology. It has been argued recently (Emerson and
Kolm [2005a], Nature 434:10151017) that, in the
Hawaiian and Canary archipelagos, species diversity
itself has been driving the speciation rates of endemic
plants and arthropods, in accord with island biogeography theory. Here we show that the observed trends in
the distribution of endemics within these two archipelagos are in fact consistent with null predictions.
Moreover, we show that similar trends are followed by
non-endemic singletons (i.e., species not endemic to the
archipelago that occur on a single island), despite the
different mode by which they rst arose in the
archipelago. Together, these ndings undermine the
validity of the diversity-driven diversication hypothesis, at least in the systems considered here.
Is biodiversity simply the outcome of ecological and
evolutionary processes, or could it also be a causative
agent in driving speciation? The idea that species can
modify their environment, thereby making it available
for additional species, is not new (Jones et al. 1997,
Erwin 2005). Recently, Emerson and Kolm (2005a, b)
have gone further to suggest that species richness can
promote diversication, a prediction they ascribe to the
theory of island biogeography (MacArthur and Wilson
1967). Assuming that extinction and speciation rates
Manuscript received 19 December 2005; revised 10 October
2006; accepted 12 October 2006.Corresponding Editor: H.
Hillebrand.
1 The Interuniversity Institute for Marine Science at Eilat,
P.O. Box 469, Eilat 88103 Israel.
2 Department of Life Sciences, Ben-Gurion University,
POB 653, Beer Sheva, Israel.
3 Department of Evolution, Systematics and Ecology, The
Hebrew University, Givat Ram, Jerusalem, Israel.
4 E-mail: [email protected]

follow a similar curvilinear dependence on richness,

Emerson and Kolm predict that, under equilibrium
conditions, the proportional representation of endemic
singletons should increase with richness (but see Cadena
et al. 2005). They then proceed to demonstrate positive
relationships between relative endemism and total
species richness of the vascular ora and arthropod
fauna of the Canary and Hawaiian Islands. Assuming
that relative endemism proxies speciation rate, they
conclude that diversity drives diversication.
Implied by the diversity-driven diversication hypothesis is that the distribution of endemic singletons
across islands differ from the distribution expected by
chance, given the observed variation in species richness;
which itself varies due to differences in island properties
such as size, distance from the mainland, and so forth
(Table 1). In this note we show that, to a large extent,
the trends described by Emerson and Kolm coincide
with null expectations generated by random distribution
of singletons within archipelagos. Barring the effect of
sampling, species richness contributes little in explaining
the residual variance in relative endemism; contrary to
the diversity-driven diversication hypothesis. Consistent with this nding, we show that non-endemic
singletons (i.e., species occurring on a single island, but
which have been recorded outside the archipelago)
follow a similar trend to that of endemics; despite the
different mode by which they rst arose in the
archipelago.
The null model
Consider an archipelago of C islands, which acquire
species through immigration (either from within the
archipelago or from without) and speciation. Let pj be
the probability that a given species is currently present
on island j, such that variation in pj ensues differences in
island traits that affect species acquisition and persistence rates (e.g., size, distance to mainland/closest
island, and so on.). Consider next the probability that
a species is present on j and nowhere else within the
archipelago, hj. Assuming that the observed species
distribution was structured solely by variation in pj, hj
equals pj Pk6j (1  pk); which is a positive function of pj
(Fig. 1). Normalizing hj by division by RCj1 hj returns the
proportion of singletons on island j, which is expected if
species (or at least singletons) were randomly distributed
across islands. Given pj, we could thus test whether the
observed distribution of singletons is signicantly
different from random by evaluating the 95% condence
interval around the expected proportions. As it is the

COMMENTS

Ecology, Vol. 00, No. 0

TABLE 1. Results of multiple regression analyses, with total island richness as the dependent
variable.

Location, taxon
Canary Islands
Arthropods
Plants
Hawaiian Islands
Arthropods
Plants

Number
of islands
7
7

16
16

Variables in model

maximum elevation
distance to mainland
maximum elevation
distance to mainland

1.32
0.69
1.22
0.61

6.36
3.34
3.91
1.93

0.003
0.029
0.017
0.125

0.86
0.96

6.34
12.42

0.000
0.000

area
area

Adjusted r2
0.87**
0.71*

0.72**
0.91**

Notes: b is the standardized regression coefcient. Model signicance: *P  0.05; **P  0.001.

distribution of singletons that is of ultimate concern, this

constitutes a direct test of the diversity-driven hypothesis. Here we let pj equal island js relative richness: the
proportion of the total species pool found on island j.
This model implies that a positive association between
relative endemism and species richness is an inevitable
consequence of less ubiquitous species experiencing a
sharper increase in their probability of island occupancy,
which accompanies the increase in the species acquisition and persistence rate of islands (p). To see this we
need the probabilities of a species occurring on island j
and 1, 2, . . .C  1 other islands. The equations for these
probabilities are not as trivial as they are for singletons,
5

hhttp://hbs.bishopmuseum.org/ hbsdb.htmli

but the point is easily exemplied using simulation (Fig.

1).
Methods
Data set.Data for the plants and arthropods of the
Canary Islands were obtained directly from B. Emerson
and are based on Acebes-Ginoves et al. (2004) and
Oromi and Baez (2004). A list of Hawaiian plants and
arthropods was received from the manager of the Bishop
Museum (data available online).5 Hawaiian islands
included in the analyses are Kure, Midway, Laysan,
French Frigate Shoals, Lisianski, Necker, Pearl and
Hermes, Nihoa, Niihau, Kauai, Oahu, Molokai, Lanai,
Maui, Kahoolawe, and Hawaii. The data analyzed
excluded anthropogenic introductions, but included
uncertain occurrences.

FIG. 1. The proportion of singletons, doublets, triplets, and quadruplets (i.e., species occurring in i 1, 2, 3, and 4 islands,
respectively) expected to occur in an island with a probability p of being occupied. The proportions are based on a simulation
involving four cells (islands) that vary with respect to p (0.05, 0.15, 0.3, and 0.5). An array of four random numbers was iteratively
drawn from a uniform distribution (0  r  1). In each iteration, cell j was occupied if rj  pj, and the number of cells occupied was
tallied (i); iterations with i 0 were discarded. By the end of 10 000 iterations, we calculated the fraction of iterations with i 1, 2, 3,
and 4 contributed by each cell (qij; Rj qij 1). These fractions are depicted on the rst y-axis (proportion in cell j). Also shown is the
proportion of each cells total occurrences (richness) that is attributable to singletons: q1j/Ri qij (solid circles; second y-axis,
proportion of singletons).

Month 2007

COMMENTS

FIG. 2. The observed proportion of the total endemic (solid circle) and non-endemic (open circle) singletons found on each
island, along with the proportions expected under the null model (solid line). Relative richness, pj, is dened as the proportion of the
total species pool found on island j. The 95% condence interval used to test the null model is depicted by the dotted lines.

Island parameters of potential biogeographical importance were compiled from Price (2004; Hawaii: area,
maximum elevation, and age) and Fernandez-Palacios
and Mart n-Esquivel (2001; Canary Islands: area,
maximum elevation, and distance to mainland); distances to the nearest island were measured using ArcView
GIS 9 (ESRI, Redlands, California, USA) using a U.S.
general map (Hawaii) and The Times Concise Atlas of
the World (Canary Islands; Bartholomew et al. 1986).
Condence intervals and hypothesis testing.The 95%
condence intervals associated with the expected proportion of singletons were based on numerical estimation of hj; the probability that a species is present on
island j and nowhere else within the archipelago. Briey,
for each island j, a set of C random proportions were
iteratively (n 1000) drawn from a uniform distribution
(0  r  1). Comparing the observed and random
proportions, sets with rj  pj and rk  pk, for all k 6 j,
were scored as success. The proportion of successes
(out of 1000) constitutes a single estimate of hj. The
procedure was repeated 1000 times and the ensuing

sample used to estimate the condence interval (i.e.,

based on the 2.5th and 97.5th percentiles).
As a test of the diversity-driven-diversication hypothesis we (1) evaluated the extent by which observed
proportions fall within the expected condence intervals,
(2) evaluated the Spearman rank correlation between
island richness and the deviation of the observed
proportions from the expected, (3) compared (qualitatively) the trends followed by endemics with those of
non-endemics, expecting that presumed differences in
evolutionary and ecological histories should result in
different distribution patterns.
Results
To a very large extent, the observed proportion of
singleton endemics fell within the range expected by
chance (Fig. 2). Indication of a residual positive
relationship between richness and endemism, beyond
that which is expected by chance, is found only in the
case of plants in the Canary Islands (Table 2). Overall,
however, the trends followed by endemic and non-

COMMENTS
TABLE 2. Spearman rank correlation between island richness
and the deviation between the observed and expected
proportion of endemic and non-endemic singletons.
Endemic

Non-endemic

Location, taxon (n)

rs

rs

Canary arthropods (7)

Canary plants (7)
Hawaiian arthropods (16)
Hawaiian plants (16)

0.14
0.89
0.09
0.43

0.380
0.003
0.364
0.048

0.50
0.18
0.34
0.17

0.127
0.351
0.098
0.267

Notes: Positive correlations imply that richer islands contain

more singletons (and poorer islands contain fewer) than
expected by chance. One-tailed signicance levels are provided
for the correlation coefcients. Sample sizes in parentheses are
the number of taxa.

endemic singletons are remarkably similar (Fig. 2; Table

2).
Discussion
The use of null models in the study of community
ecology has been an accepted practice for over two
decades and has led to the rethinking of several
established hypotheses and paradigms (e.g., Gotelli
and Graves 1996, Colwell et al. 2004). Null models do
not suggest that evolutionary and ecological processes
are not operating, rather that these processes contribute
little structure to the variable in question, beyond that
which is expected by chance. With respect to the presentday distribution of the Hawaiian and Canary arthropod
and plant endemic singletons, we nd that it matches
null expectations with only minor exceptions (Fig. 2),
calling into question the hypothesis that diversity can
drive diversication. The similarity in the trends
followed by endemic and non-endemic singletons serves
only to underscore this conclusion.
It is important to remember that only with respect to
topographically and geographically driven variation in p
(an islands rate of species acquisition and persistence)
does species distribution appear random. In this sense,
our ndings would be consistent with any other
explanation of a positive relationship between relative
endemism and richness which invokes differences in
island attributes as the sole causative agent (e.g., Cadena
et al. 2005). To the extent that the proportion of
endemic singletons does indeed proxy local speciation
rates, we must conclude therefore that the overriding

View publication stats

Ecology, Vol. 00, No. 0

effect of island attributes affords species richness little

contribution to the spatial patterns of diversication
beyond the null sampling effect.
Acknowledgments
We are grateful to Tamar de la Concepcion Sanchez and
Richard L. Pyle for kindly assisting us in compiling the data set;
to B. Emerson and three anonymous reviewers for their
constructive comments; and to M. Spencer of his insightful
suggestions
Literature cited
Acebes Ginoves, J. R., et al. 2004. Spermatophyta. Pages 99
143 in Lista de Especies Silvestres de Canarias (Hongos,
Plantas y Animales Terrestres). Second edition. I. Izquierdo,
J. L. Martin, N. Zurita, and M. Archevaleta, editors.
Consejeria de Politica Territorial y Medio Ambienta
Gobierno de Canarias, La Laguna, Tenerife.
Bartholomew, J. C., R. Jones, H. A. G. Lewis, S. Lippard, D.
Rothery, and D. Whitehouse. 1986. The times concise atlas of
the world. Fifth edition. Times Books, London, UK.
Cadena, C. D., R. E. Ricklefs, I. Jimenez, and E. Bermingham.
2005. Is speciation driven by species diversity? Nature 438.
Online, [doi:10.1038/nature 04308].
Colwell, R. K., C. Rahbek, and N. J. Gotelli. 2004. The middomain effect and species richness patterns: what have we
learned so far. American Naturalist 163:E1E23.
Emerson, B. C., and N. Kolm. 2005a. Species diversity can
drive speciation. Nature 434:10151017.
Emerson, B. C., and N. Kolm. 2005b. Reply to: Is speciation
driven by species diversity? Nature 438. Online, [doi:10.1038/
nature 04308].
Erwin, D. H. 2005. Seeds of diversity. Science 308:17521753.
Fernandez-Palacios, J. M., and J. L. Mart n-Esquivel. 2001.
Las islas como experimento de laboratorio. In Naturaleza de
las Islas Canarias. Ecolog a y conservacion. J. M. FernandezPalacios and J. L. Mart n Esquivel, editors. Turquesa, Santa
Cruz de Tenerife.
Gotelli, N. J., and G. R. Graves. 1996. Null models in ecology.
Smithsonian Institution Press, Washington, D.C., USA.
Jones, C. G., J. H. Lawton, and M. Shachak. 1997. Positive and
negative effects of organisms as physical ecosystem engineers.
Ecology 78:19461957.
MacArthur, R. H., and E. O. Wilson. 1967. The theory of
island biogeography. Princeton University Press, Princeton,
New Jersey, USA.
Oromi, P., and M. Baez. 2004. Arthropoda. Pages 155300 in
Lista de Especies Silvestres de Canarias (Hongos, Plantas y
Animales Terrestres), 2nd edn. I. Izquierdo, J. L. Martin, N.
Zurita, and M. Archevaleta, editors. Consejeria de Politica
Territorial y Medio Ambienta Gobierno de Canarias, La
Laguna, Tenerife.
Price, J. P. 2004. Floristic biogeography of the Hawaiian
Islands: inuences of area, environment and paleogeography.
Journal of Biogeography 31:487500.