Conservation Genetics 1: 67–76, 2000.

© 2000 Kluwer Academic Publishers. Printed in the Netherlands.

67

Questioning species realities

Andrew P. Hendry1,2∗ , Steven M. Vamosi1 , Stephen J. Latham1 , Jana C. Heilbuth1 &

Troy Day1,3
1 Department of Zoology, University of British Columbia, 6270 University Blvd., Vancouver, BC, Canada V6T
1Z4; 2 Present address: Organismic and Evolutionary Biology Program, 319 Morrill Science Center, University of
Massachusetts Amherst, Amherst, MA 01003-5810, USA; 3 Present address: Department of Zoology, University of
Toronto, Toronto, Ontario, Canada M5S 3G5; ∗ author for correspondence (E-mail: [email protected])

Received 28 November 1999; accepted 25 April 2000

“In short, we shall have to treat species in the same manner as those naturalists treat genera, who admit that genera are merely artificial
combinations made for convenience. This may not be a cheering prospect; but we shall at least be free from the vain search for the
undiscovered and undiscoverable essence of the term species.”
(Darwin 1859, p. 485)

Key words: mitochondrial DNA, phylogeny, phylogeography, speciation, species concepts, taxonomy

Introduction designations (although not without their own caveats).

Many genetic studies have reexamined taxonomies
A frequent outcome when studying complex biolo- formerly based exclusively on morphology, and have
gical systems is that data collected to test a specific in some cases uncovered paraphyletic or polyphyletic
hypothesis can be interpreted several ways, and inter- groupings, thereby precipitating taxonomic rearrange-
pretations can be influenced by the paradigm through ment (e.g. de Jong 1998). Numerous genetic studies
which a given scientist views the world. A paradigm have also examined the validity of species designa-
to which most biologists subscribe is that biolo- tions in particular groups, in some cases confirm-
gical diversity can be meaningfully divided into least- ing and in others refuting previous interpretations
common evolutionary denominators, called ‘species’. (e.g. Riddle and Hafner 1999). An extension of this
Although delineating distinct species is often prob- approach, missing until recently, is the use of genetic
lematic, most biologists agree with Mayr (1957) that data to evaluate the overarching concept of assigning
“the living world is comprised of more or less distinct groups of organisms to specific bins, most notably
entities which we call species.” A contrasting view species.
(the one to which we subscribe) is a recognition of Perhaps the first, and certainly the most ambi-
more or less distinct clusters of organisms at varying tious attempt, to use genetic data for evaluating the
biological scales, without assuming some fundamental validity of species as a concept has its basis in a
and universal level of clustering that has evolution- series of reviews by Avise and colleagues (Avise and
ary significance out of proportion to all other levels Walker 1998, 1999; Avise et al. 1998; Johns and
of clustering. This view from outside the species Avise 1998). This series culminated in a paper titled
paradigm allows conclusions that depart significantly Species realities and numbers in sexual vertebrates:
from those commonly advanced. Perspectives from an asexually transmitted genome,
Taxonomic groups have historically been identified in which Avise and Walker (1999) used patterns of
using morphological criteria, leaving uncertainty in mitochondrial DNA (mtDNA) diversity to argue that
some cases as to the validity and meaning of groups “mtDNA data and traditional taxonomic assignments
thus delineated. Within the last few decades, molecu- tend to converge on what therefore may be real biotic
lar techniques have provided a powerful new tool units in nature”. Because this was the first major work
to independently evaluate the validity of taxonomic of its kind, with important implications for evolution,
68

ecology, and conservation, its methods and conclu- 1998). We reanalyzed the data in those reviews to see
sions deserve careful evaluation. Here, we provide a if we could bolster support for our interpretation or
reanalysis of the data sets and methods used by Avise for that of Avise and Walker (1999). First, we gener-
and colleagues, from which we conclude that mtDNA ated cumulative frequency distributions of mtDNA
discontinuities do not closely match recognized taxo- sequence divergence between taxonomic sister species
nomic species. and between intra-specific phylogroups. Sequence
This interpretation is ultimately independent of the divergence data used for this analysis was obtained
class of genetic markers used to identify biotic discon- from 277 sister species pairs and 183 phylogroup
tinuities, and instead reflects what we perceive as pairs of vertebrates (Table 1). These histograms were
fundamental flaws in the species paradigm. Recogni- used in the spirit of analyses performed by Avise
tion of these flaws would engender a broader consid- and colleagues to examine the degree of overlap in
eration of how delineating distinct species fails to sequence divergence between inter-specific and intra-
adequately capture the essence of biological diversity. specific pairs. Second, we matched data for diver-
We then discuss implications for two important gence among congeneric species [from Figures 1–4
goals of conservation biology – the identification of in Johns and Avise (1998)] with data for divergence
geographical regions and particular groups of organ- between conspecific phylogroups within each genus
isms that warrant protection. Finally, we propose an [from Table 2 in Avise and Walker (1998) and the
alternative solution to the species problem. To attain Appendix of Avise et al. (1998)]. This second analysis
this solution we suggest that biologists shift their focus was used to evaluate the degree to which divergence
from the elusive demarcation of species to more quant- among phylogroups within a species differs from
itative descriptions of variation within and among divergence among species in that genus.
groups (or clusters) of organisms. If taxonomic species designations converge on
mtDNA discontinuities, we would expect little overlap
in the cumulative frequency distributions of sequence
Reanalysis divergence within and between species. Thus, few
phylogroup pairs should exceed the levels of sequence
Avise and Walker (1999) used published data to evalu- divergence that separate species (i.e. the line for
ate patterns of mtDNA diversity within 252 taxonomic phylogroups in our Figure 1 should approach zero
species of vertebrates. Previous work by Johns and before the line for species begins to fall below unity).
Avise (1998) had indicated that mtDNA differences This was not the case (Figure 1). For example,
among species were large, with approximately 90% although 88% of bird sister species pairs exceeded
of sister species pairs showing at least 2% sequence 0.6% sequence divergence, 81% of bird phylogroup
divergence. Many taxonomic species are therefore pairs also exceeded that level. Exceeding 2% sequence
quite genetically distinct from each other, and have divergence were 66% of bird sister species pairs but
been evolving independently for a considerable length also 37% of bird phylogroup pairs. Considerable over-
of time (>1,000,000 years). Remarkably, however, lap in the amount of divergence within and between
56% of the species surveyed could be sub-divided into species was also evident for each of the other taxa
at least two “major intra-specific phylogroups”, and (Figure 1). We conclude that there is no clear separ-
5% had three or more such phylogroups (Avise and ation and no single threshold level for mtDNA diver-
Walker 1999). These phylogroups are envisioned as gence that distinguishes species from phylogroups.
independently-evolving, historical lineages equivalent Considerable overlap between the amount of genetic
to taxonomic species in all ways, except presumably divergence between phylogroups and species is also
the magnitude of divergence (Avise and Walker 1998). evident in other, less-extensive, reviews (Vogler and
The existence of distinct phylogroups within more DeSalle 1994; Klicka and Zink 1999; Riddle and
than half of the surveyed species seems at odds with Hafner 1999).
the conclusion that taxonomic species reliably unveil If taxonomic species designations converge on
distinct evolutionary lineages. mtDNA discontinuities, we would also expect that
Avise and Walker (1999) used data from three of divergence among phylogroups within species should
their own recent reviews of the primary literature on be considerably less than divergence between that
mtDNA diversity within and among species (Avise species’ congeners in paired comparisons. Although
and Walker 1998; Avise et al. 1998; Johns and Avise this was sometimes the case, intra-specific diver-
 69
Table 1. Characteristics of the data and sources used in our analysis, and by Avise and Walker (1999), to compare divergence between
sister species (‘Species’) and phylogroup pairs (‘Phylogroup’). Some species had more than two phylogroups, and so the total number
of species from which phylogroup pair divergence was estimated is indicated in parentheses. Information includes the publication from
which the data were obtained (‘Source’ – either Avise and Walker 1998, ‘A & W’, or Avise et al. 1998, ‘A et al.’), the relevant table or
figure in those publications (‘Location’), the number of species or phylogroup pairs (‘Pairs’), the type of mtDNA data used (‘Data’), the
number of phylogroup comparisons corrected for within phylogroup variation (‘Corrected’), the number of control region studies included
by Avise and colleagues (‘Control region’), the number of species excluded because geographically divergent samples did not demonstrate
major intra-specific phylogroups (‘Excluded’), and the median level of sequence divergence

Mammals Birds Herpetofauna Fish

Species Phylogroup Species Phylogroup Species Phylogroup Species Phylogroup

Source A et al.a A et al. A & Wb A&W A et al.a A et al. A et al.a A et al.

Location Figure 2a Figure 2b Figure 2a Table 1 Figure 3a Figure 3b Figure 4a Figure 4b
Pairs 92 72 (54) 35 37 (37) 42 47 (25) 108 26 (24)
Data cytbc >150 bpd cytbc >200 bpd cytbc >150 bpd cytbc >150 bpd
Corrected 0 ?e 0 14 0 ?e 0 ?e
Control regiong 0 9 0 2 0 2 0 1
Excluded 0 57f 0 26 0 37f 0 20f
Median divergence 6.4 2.6 5.6 3.1 6.1 3.1 4.1 2.6
a Data were originally compiled by Johns and Avise (1998). b Data were originally compiled by Klicka and Zink (1997). c A single sequence
of the mitochondrial cytochrome b gene (the longest >200 bp in GenBank, release 103.0) was used to represent each species (Johns and
Avise 1998). d Various regions of the mtDNA genome were used (Avise and Walker 1998; Avise et al. 1998). e Some phylogroup pairs were
corrected for within phylogroup variation but the number was not indicated by Avise and colleagues. f Excluded species were not reported
in the publications, so we estimated them using the reported percentage of excluded species for all vertebrates combined (44%, Avise
and Walker 1999). g The control region data are problematical because only a subset of control region studies were included by Avise and
colleagues. It was not indicated by Avise and Walker (1999) if control region data were corrected for its faster rate of evolution. Owing to
these cumulative ambiguities, we excluded control region sequence divergence data from our analysis.

gence often approached levels of inter-specific diver- reiterate and elaborate on those biases, and discuss
gence (Figure 2). On average, intra-specific diver- additional analytical concerns not addressed in their
gence was 44% as large as the average inter-specific manuscript. Some of the identified biases would tend
divergence within the same genus, and 87% as large to increase and others decrease apparent concordance
as the minimum inter-specific divergence. Diver- between traditional species and genetic discontinuities
gence among phylogroups within species actually (all biases will increase inferential uncertainty). It will
exceeded the minimum amount of divergence among be important for future studies to reduce these biases
congeners for 18 of 52 comparisons (35%). Thus, or to at least evaluate their strength. We will start with
although taxonomic species often recognize large the most general issues and work toward the more
mtDNA discontinuities, they fail to recognize many specific.
other such discontinuities. In fact, phylogroups iden- Avise and colleagues surveyed vertebrates, a group
tified in genetic surveys are often later interpreted as for which confusion as to species status is much
cryptic species [Klicka and Zink (1999) report some lower than for other taxa. It is therefore discour-
examples]. aging that Johns and Avise (1998) found “rather
poor equivalency of taxonomic rank across some
of the Vertebrates”. For example, they found that
Analytical hurdles in testing species realities “surveyed avian taxa on average show signific-
antly less genetic divergence than do same-rank taxa
Avise and Walker’s (1999) review sets an import- surveyed in other vertebrate groups . . . ” If poor corres-
ant precedent for the use of genetic data to pondence prevails among different vertebrate taxa,
evaluate species realities. Their methods therefore correspondence between vertebrates and other groups
deserve careful consideration to identify any potential is certainly lower. The greater ambiguity and incon-
concerns that should be addressed in future studies. sistency encountered in defining species within groups
Avise and Walker (1999) are careful to mention poten- such as eukaryotic algae, fungi, plants, marine inver-
tial sampling biases inherent in their data. Here we tebrates, nematodes, insect herbivores, and partheno-
70

Figure 1. Proportions of sister species pairs and phylogroup pairs exceeding different levels of sequence divergence. Note the large amount of
overlap in the distributions.

genic insects is apparent in the contributions to Clar- An obvious bias can arise from limited sampling
idge et al. (1997). The problem of non-equivalence of across a species’ range (Avise and Walker 1999). The
taxonomic rank among different types of organisms is resources available to any genetic survey are limited,
discussed by Avise and Johns (1999), along with their and sampling is usually restricted to a subset of the
proposed solution. locations in which any particular species is found.
Biases may arise owing to the subset of species Increased sampling from other locations would not
selected for analysis. Avise and Walker (1999) point decrease the amount of intra-specific genetic variation
out that species with large geographic ranges are often and might increase it substantially. This bias may have
chosen for phylogenetic analyses because they show been reduced (to an unknown degree) in Avise and
the greatest potential for multiple historical units. Walker (1999) through their aforementioned decision
Species with narrow ranges (and less expected intra- to only consider studies that sampled across much of
specific structure) may be underrepresented. This a species’ range. Of course their focus on studies over
effect was compounded in Avise and Walker (1999) large geographic ranges makes their conclusion that
because they only considered studies with “multiple 93% of the phylogroups “displayed a strong geograph-
samples from widely spaced localities across signific- ical orientation” less remarkable.
ant portions of a species’ range”. However, the extent The types of studies excluded from a review is
to which the geographical range of a group of organ- also an important consideration. For example, species
isms actually influences its potential for independent for which data were available but for which major
lineages has not been quantified. phylogroups were not detected (112 of 252 species)
 71

in number and composition to the biological units

currently recognized as taxonomic species” (Avise and
Walker 1999).
Another concern is that phylogroups must be fairly
distinctive to be recognized. In Avise and Walker
(1999), phylogroups were identified “by relatively
large genetic gaps between respective branches that
received strong bootstrap support in an estimated
mtDNA gene tree”. Phylogroups identified in this
manner typically “were distinguished consistently by
at least 0.6% sequence divergence”, which is equiv-
alent to about 300,000 years of separation (assuming
2% sequence divergence per million years). This level
of separation would exclude many groups that are
highly divergent in morphology, ecology, behavior,
and mate recognition. For example, many of the 300
or so endemic haplochromine cichlid fishes of Lake
Victoria would not be recognized even as intra-specific
phylogroups. Avise et al. (1998) acknowledged Lake
Victoria fishes, and the existence of other “examples
that depart radically from the vertebrate norm”, but
if taxonomists recognize species with very recent
origins, genetic surveys designed to test species real-
ities should also consider such groups. At a more
fundamental level, young species are important for
understanding the processes of diversification because
Figure 2. Amount of sequence divergence within conspecific they sometimes maintain their distinctiveness under
phylogroups relative to the average (top panel) and minimum the most difficult of circumstances – strict sympatry
(bottom panel) amount of sequence divergence between that species
and its congeners. Birds are represented by circles, mammals
(Schluter 1996; Taylor 1999).
by squares, amphibians and reptiles by triangles, and fishes by The type of genetic marker chosen for study
crosses. The diagonal line is the isoline, where divergence between may influence apparent concordance between taxo-
phylogroups is the same as that between species. Note that in nomic species and biotic discontinuities. Avise and
many instances divergence within species approaches or exceeds
divergence among congeners. Walker (1999) point out that mtDNA patterns will
fail to reflect the degree of interbreeding if gene flow
is male-biased. Another consideration is the poten-
were excluded from the analysis of Avise and Walker tial for ancient mtDNA lineages to be maintained
(1999), and therefore from our own. (We did contact for long periods of time within interbreeding popu-
Dr Avise to inquire about these excluded studies but he lations. These ancient lineages can then be sorted
indicated that records had not been kept of the specific between smaller, recently-diverged populations, arti-
studies that failed to detect phylogroups). Including ficially increasing their apparent age and distinctive-
these studies would perhaps have increased apparent ness. Alternatively, two groups with similar mtDNA
cohesiveness of taxonomic species. Working at cross- profiles can actually be independent if, for instance,
purposes is the exclusion of studies that monitored the mtDNA of one group has been “captured” in the other
rapidly-evolving control region, “except where ancient following an ancient hybridization event (Avise 1994).
and explicit divergence dates . . . were proposed in the Finally, the typical length of mtDNA surveyed in the
original publications” (Avise and Walker 1999). This bird studies (and presumably in the other taxa) is only
policy was adopted “to avoid a focus on unduly shal- 500 bp, and this may be too short to firmly establish
low mtDNA clades that are of little interest in the phylogenetic groups that originated less than 200,000
current context . . . ” However, we feel these shal- years ago (Avise and Walker 1998). We have illus-
low clades will be important in the context of testing trated the concerns associated with mtDNA because
“whether biotic discontinuities . . . bear resemblance this is the only class of markers yet employed for test-
72

ing species realities. Other genetic markers will have These problems have been the subject of unremit-
their own sets of constraints. ting and escalating debate for centuries (Darwin 1859;
Another bias can arise owing to adjustments for Mayr 1957; Otte and Endler 1989; Mallet 1995; Clar-
within-group variation – “When possible from the data idge et al. 1997; Wheeler and Meier 1997; Howard
provided, p (sequence divergence) values between and Berlocher 1998; Wilson 1999). Instead of revisit-
major phylogroups (e.g., ‘A’ and ‘B’) were correc- ing these intricate and often convoluted exchanges, we
ted for within-phylogroup variation according to the wish to highlight two critical problems, the considera-
following formula: pAB(net) = pAB − 0.5(pA + pB )” tion of which hints at what we feel may be the only
(Avise et al. 1998; see also Avise and Walker 1998; real solution. First, biologists are forced to decide,
Avise 1994, p. 96). In this manner, divergence between either explicitly or implicitly, what level of difference
phylogroups was sometimes discounted by the amount between two groups makes each worthy of its own
of divergence within those phylogroups, and within- species designation. Second, once species are identi-
group divergence was often quite large (Avise and fied, all are considered equal (except in phylogenetic
Walker 1999). The specific phylogroups subjected to weighting approaches, see below), despite the fact that
these corrections were indicated for birds (14 of 38 they may vary by orders-of-magnitude in the amount
phylogroups, Avise and Walker 1998) but not for the of difference from their nearest relative.
other taxa. The correction is not wrong in and of Regardless of the operational species concept
itself but the problem is that divergence was correc- chosen (reviewed by Mayden 1997), biologists are
ted in this manner for only some of the phylogroups, forced to grapple with the vexing choice of what
and divergence among species was not corrected level of difference (or amount of isolation) makes a
using a similar procedure (only a single mtDNA species. For example, the popular biological species
sequence was used for each species, Johns and Avise concept (BSC) states that species are “groups of actu-
1998). Avise and Walker (1998) did ponder a similar ally or potentially interbreeding natural populations
correction to between-species sequence divergence, which are reproductively isolated from other such
and concluded that its effects would be consider- groups” (Mayr 1940). If 100% reproductive isolation
able. In short, the absolute amount of divergence is used as a criterion for applying the BSC, identific-
between a considerable proportion of the phylogroups ation of distinct species is relatively straightforward
surveyed is higher than that reported by Avise and (although other problems remain). If this criterion
colleagues. was universally adopted, however, many current taxo-
Recognizing that relative strengths of the various nomic species would no longer be recognized, owing
biases are unknown, Avise and Walker (1999) suggest to rampant hybridization and introgression in the wild
that they “should partially cancel one another” and that (Arnold 1997). If some gene flow is permitted among
current taxonomic species agree with mtDNA discon- species when applying the BSC (the approach taken
tinuities “certainly within an order of magnitude”. The by most modern biologists), a threshold amount of
first view seems premature in the absence of suppor- isolation necessary to discriminate between species
tive analysis, and the second certainly leaves room must be assumed, and any such choice is largely
for improvement. Given the large number of uncer- arbitrary. Should two groups be considered separ-
tainties in the data and analyses performed thus far, ate species if they are 90% reproductively isolated,
we (and Avise and Walker 1999) encourage further or if they are 10% reproductively isolated? Mayr
efforts to extend tests of species realities to larger data (1996) suggested that species should be considered
sets, other taxa, and to the use of additional analytical distinct as long as “clandestine hybridization” does not
techniques. result in “the complete fusion of such species popula-
tions”. But how does one determine ‘complete fusion’
without retreat to thresholds? Compounding the prob-
Unsolvable problems with the concept of species lem, the threshold chosen will vary widely among
investigators.
Whether or not a group of organisms is recognized as Many taxonomists favor variants of the phylogen-
a distinct species is ultimately a dichotomous decision etic species concept (PSC) which defines species as
– yes or no. Notwithstanding the aesthetic simplicity the “smallest diagnosable cluster of individual organ-
of this approach, it remains burdened by unsolvable isms within which there is a parental pattern of ances-
philosophical, theoretical, and empirical problems. try and descent” (Cracraft 1983; Vogler and DeSalle
 73

1994). The PSC, however, provides no reprieve from Sigüenza 1999). These areas may contain extraor-
the arbitrary nature of species delineation because dinary levels of biological diversity or provide succor
some threshold level of difference must still be adop- for types of organisms not found elsewhere. A
ted (with enough traits all individuals are diagnosable second major goal is the winnowing of complexes of
from each other, Avise 1994). Either that or the closely related organisms into groups that would merit
analysis must be restricted to a subset of possible focused conservation efforts and those that would not
traits, and the resulting diagnostic characters will (e.g. Vogler and DeSalle 1994). Both of these goals
vary widely and often arbitrarily, making it difficult currently rely heavily on the identification or enumer-
to equate species counts and identities across even ation of distinct species or subspecific groups, and are
closely related taxa. Even if universal diagnosable therefore impacted by the operational and theoretical
criteria could be agreed upon, the resulting collec- problems discussed above.
tion of ‘species’ would have no biological meaning Scientific and popular pulpits used to appeal for the
other than their separation based on the chosen char- protection of particular geographical localities typic-
acters (Mallet 1995). Some purveyors of the PSC ally resort to superlatives based on counts of species,
profess to not assume that species are real entit- especially endemics (e.g. Wilson 1992). The conser-
ies but nevertheless use the method for delineating vation of areas with exceptional levels of irreplace-
distinct groups that merit conservation (Goldstein et able biological diversity should indeed be a prior-
al. 2000), an approach that does not obviate either ity but decisions made by prioritizing regions based
of the major problems discussed herein (it simply on species counts may not be the best approach
removes the term species from discussion). The arbit- (although it provides useful stop-gap information).
rary nature of diagnostic criteria such as specific Lurking beneath substantial operational difficulties,
characters or thresholds is a problem universal to such as non-equivalence of species counts obtained
operational species concepts. using different concepts (e.g. BSC vs. PSC, Peterson
The amount of difference between taxonomic and Navarro-Sigüenza 1999), are more fundamental
species (and at other taxonomic levels) in various problems. For example, some regions may contain
genetic or phenotypic traits is often very large. For many closely related species whereas others contain
example, 25 of the 109 sister species of birds we fewer but more diverse species. Counting genera or
considered showed less than 2% sequence diver- families is one attempt to circumvent this concern but
gence, whereas 11 of the pairs showed greater than becomes increasingly sensitive to the arbitrary nature
10% sequence divergence (Figure 1). The dichotom- of higher-level taxonomic categories, and ignores the
ous nature of species delineation ignores quantitative extent of diversity within those categories.
differences between species, and effectively considers A few inspired attempts have been made to
all equally distinctive. Thus, any study comparing quantify diversity at less inclusive levels of biological
species numbers among taxa, geographical regions, or organization, such as Hughes et al. (1997) for ‘popu-
time periods obscures the fact that biological diversity lations’. These attempts are unfortunately still limited
is poorly quantified simply by counting the number because they must likewise assume a threshold level
of taxonomic species. Comparative studies would of difference (Hughes et al. used statistical signific-
certainly benefit from the adoption of a standardized ance), and then the actual amount of difference among
temporal scheme for defining hierarchical taxonomic populations exceeding the threshold is ignored. It has
levels (e.g., Avise and Johns 1999) but such a scheme been suggested that ‘phylogeographic ESUs’ are an
remains compromised by the need to choose and apply improvement over species as measures of biological
thresholds. diversity (Riddle and Hafner 1999), but ESUs remain
hampered with the same two fundamental problems
we have discussed for species concepts. Mallet’s
Conservation implications (1995) ‘genotypic cluster’ view of biological diversity
is a step in the right direction but it does not go far
Conservation biologists often proceed by gathering enough. Mallet’s focus remains on identifying distinct
information designed to address several immediate clusters of organisms, which is important, but once
and long term goals. One goal is the identifica- clusters are identified the actual amount of difference
tion of geographical regions that warrant exceptional among them is forgotten. Biologists need to take the
efforts at protection (e.g. Peterson and Navarro- next step and divorce themselves from the idea of
74

discrete, equal bins into which organisms must be 1994; Crozier 1997). The last of these problems is the
forced, and begin to emphasize the level of variation most critical in the present context, owing to the afore-
within and among groups of organisms. mentioned ambiguities in species delineation, and the
The status that humans confer on a particular tendency to ignore within-group variation.
group of organisms under consideration for conser-
vation often hinges on whether or not that group has
somehow crossed an unseen threshold and become Quality in quantity – a promising direction
a species, evolutionarily significant unit (ESU),
management unit, or stock (Moritz 1994). The need Any system of biological classification that parti-
to identify discrete groups deserving conservation tions organisms into distinct species or ESUs (i.e.,
has it roots in natural history but has been promul- any species concept) fails to capture the essence of
gated beyond all biological reality in the legal arena biological diversity. In the time of Linnaeus, species
surrounding the US Endangered Species Act (Waples designations were a crucial contribution to natural
1991; Pennock and Dimmick 1997; Waples 1998). In history because they allowed biologists to commu-
this arena, the chosen criteria for delineating a species nicate using a common currency. Now, however,
or ESU becomes of critical importance to their future it should be recognized that strict adherence to the
welfare. Our illustration that operational thresholds for species paradigm may actually be impeding progress
species designation are arbitrary, inconsistent among in conservation and other areas of biology. We are
genetic (or phenotypic) markers, and not compar- joined in this view by a small chorus of philosoph-
able across taxa, suggests that important groups of ers and biologists (see Wilson 1999). We advocate
organisms will remain unacknowledged and therefore replacement of the current artificial view of life with
forever outside the umbrella of attempted protection. a system that describes groups of organisms based on
One possible solution to the difficulty of identi- the amount that they differ from other groups. Full
fying groups warranting conservation is to allocate description and justification of such a system must be
resources in quantitative proportion to the level of left for future analyses. Here, we merely wish to intro-
distinction of each group (as well as the nature and duce the germ of an idea that the species paradigm
intensity of the threat to their persistence). In this can be profitably replaced with a system based on the
fashion, all groups would be considered meritori- quantitative description of variation within and among
ous of protection, and the level of protection they groups (or clusters) of organisms.
received would be in proportion to their distinctive- Abandoning the concept of species and replacing
ness. This approach could be modified to consider it with a new system would ultimately constitute a
the degree of hybridization and introgression among radical change in the way biological organization is
groups, an issue that has been a traditional bane of conceptualized, but the implementation of such a
the ESA (Arnold 1997). For example, the distinctive- system would not be as painful as it might appear. The
ness of each group could be discounted by the amount change could initially entail specifying the amount
of introgression with other groups, weighted by the of difference in various traits (e.g. mtDNA, nuclear
distinctiveness of those other groups. DNA, morphology) among recognized groups of
The development of phylogenetic approaches to organisms at various hierarchical levels (e.g. genera,
conservation, which weight species in one way or species, subspecies, or populations in the current
another by their distinctiveness have been described system). Measuring biological diversity could involve
by several authors (e.g. Vane-Wright et al. 1991; quantifying total diversity for various genetic and
Crozier 1992; Faith 1992; reviewed by Faith 1994; phenotypic traits within and among geographical
Krajewski 1994; Crozier 1997). Modifications to regions or taxonomic groups. Studies of ‘speciation’
these approaches could be used to identify geograph- could be recast as studies of the evolution of repro-
ical regions warranting protection, as well as unique ductive isolation, and of genetic and phenotypic diver-
groups of organisms that merit focused conservation gence (e.g. how much isolation or divergence over
efforts. Phylogenetic methods still have some prob- how long a period). Comparative studies (e.g. sister-
lems, however, including non-equivalence of different species comparisons) could be replaced with compar-
weighting procedures, inconsistencies between char- isons of nearest-neighbor clusters in genetic space, and
acter sets, and a continued dependence on the reliable could incorporate the amount of difference between
and consistent identification of species (Krajewski the clusters. Many of our suggestions are consistent
 75

with an earlier recommendation for a standardized Avise JC, Walker D (1999) Species realities and numbers in sexual
temporal scheme of biological classification (Avise vertebrates: Perspectives from an asexually transmitted genome.
Proc. Natl. Acad. Sci. USA, 96, 992–995.
and Johns 1999), however that method differs concep- Avise JC, Walker D, Johns GC (1998) Pleistocene phylogeographic
tually from ours in its dependence on the identification effects on avian populations and the speciation process. Proc. R.
of discrete bins for organisms. Soc. London Ser. B, 265, 1707–1712.
The suggested resolution we have superficially Claridge MF, Dawah HA, Wilson MR (1997) Species: The Units of
Biodiversity. Chapman & Hall, New York.
sketched here will certainly raise numerous questions. Cracraft J (1983) Species concepts and speciation analysis. In:
For example, what traits should be used to identify and Current Ornithology (ed. Johnston R), pp. 159–187. Plenum
quantify continuous variation among groups? Many, Press, New York.
we hope, perhaps in a multivariate representation Crozier RH (1992) Genetic diversity and the agony of choice. Biol.
Cons., 61, 11–15.
of axes combining correlated traits. How would the Crozier RH (1997) Preserving the information content of species:
groups be identified, either before or during analysis? genetic diversity, phylogeny, and conservation worth. Annu. Rev.
Many groups should appear as emergent properties of Ecol. Syst., 28, 243–268.
the data, and in our approach these groups need not Darwin C (1859) On the Origin of Species by Means of Natural
Selection. John Murray, London.
be considered discrete – the degree of overlap can be de Jong WW (1998) Molecules remodel the mammalian tree. Trends
quantified. What should be done when different mark- Ecol. Evol., 13, 270–275.
ers or traits are in disagreement as to the strength of Faith DP (1992) Conservation evaluation and phylogenetic diversity.
Biol. Cons., 61, 1–10.
group identity? Perhaps they should all be retained
Faith DP (1994) Phylogenetic pattern and the quantification of
as useful descriptors of group distinctiveness. We do organismal biodiversity. Phil. Trans. R. Soc. Lond. B., 345,
not propose that these answers are definitive. Nor do 45–58.
we yet have answers to many other questions, some Goldstein PZ, DeSalle R, Amato G, Vogler AP (2000) Conservation
genetics at the species boundary. Cons. Biol., 14, 120–131.
of which have yet to be posed. Instead, we would Howard DJ, Berlocher SH (1998) Endless Forms: Species and
like to extend a pluralistic appeal to the community Speciation. Oxford University Press, Oxford.
of biologists to help us further explore what we feel is Hughes JB, Daily GC, Ehrlich PR (1997) Population diversity: its
a promising line of inquiry. extent and extinction. Science, 278, 689–692.
Johns GC, Avise JC (1998) A comparative summary of genetic
distances in the vertebrates from the mitochondrial cytochrome
b gene. Mol. Biol. Evol., 15, 1481–1490.
Acknowledgements Klicka J, Zink RM (1997) The importance of recent Ice Ages in
speciation: a failed paradigm. Science, 277, 1666–1669.
Our views were shaped during discussions with our Klicka J, Zink RM (1999) Pleistocene effects on North American
friends and colleagues at the University of British songbird evolution. Proc. R. Soc. London Ser. B, 266, 695–700.
Krajewski C (1994) Phylogenetic measures of biodiversity: a
Columbia, particularly Art Poon, Dolph Schluter, and comparison and critique. Biol. Cons., 69, 33–39.
Eric Taylor. Various manuscript iterations were read Mallet J (1995) A species definition for the modern synthesis.
by Jenny Boughman, Howard Rundle, Eric Taylor, Trends Ecol. Evol., 10, 294–299.
Mayden RL (1997) A heirarchy of species concepts: the denoue-
and an anonymous reviewer. Positive reinforcement ment in the saga of the species problem. In: Species: The Units
was provided by persistent entreaties from the lead of Biodiversity (eds. Claridge MF, Dawah HA, Wilson MR),
author’s word processing spell-checker that ‘speci- pp. 381–424. Chapman & Hall, New York.
ation’ be replaced with ‘speculation’. A. P. Hendry Mayr E (1940) Speciation phenomena in birds. Am. Nat., 74, 249–
278.
was supported by a Natural Sciences and Engineering Mayr E (1957) Species concepts and definitions. In: The Species
Council of Canada Postdoctoral Fellowship. Problem (ed. Mayr E), pp. 1–22. American Association for the
Advancement of Science, Washington.
Mayr E (1996) What is a species, and What is not? Philos. Sci., 63,
References 262–277.
Moritz C (1994) Defining ‘evolutionarily significant units’ for
Arnold ML (1997) Natural Hybridization and Evolution. Oxford conservation. Trends Ecol. Evol., 9, 373–375.
University Press, Oxford. Otte D, Endler JA (1989)Speciation and Its Consequences. Sinauer
Avise JC (1994) Molecular Markers, Natural History and Evolu- Associates Inc., Sunderland.
tion. Chapman & Hall, New York. Pennock DS, Dimmick WW (1997) Critique of the evolutionary
Avise JC, Johns GC (1999) Proposal for a standardized temporal significant unit as a definition for “distinct population segments”
scheme of biological classification for extant species. Proc. Natl. under the U.S. Endangered Species Act. Cons. Biol., 11, 611–
Acad. Sci., 96, 7358–7363. 619.
Avise JC, Walker D (1998) Species durations and Pleistocene effects Peterson AT, Navarro-Sigüenza AG (1999) Alternative species
on vertebrate phylogeography. Proc. R. Soc. London Ser. B, 265, concepts as bases for determining priority conservation areas.
457–463. Cons. Biol., 13, 427–431.
76

Riddle BR, Hafner DJ (1999) Species as units of analysis in ecology Waples RS (1991) Pacific salmon, Oncorhynchus spp., and the
and biogeography: time to take the blinders off. Global Ecol. definition of “species” under the Endangered Species Act. Mar.
Biogeogr., 8, 433–441. Fisheries Rev., 53, 11–22.
Schluter, D (1996) Ecological speciation in postglacial fishes. Phil. Waples RS (1998) Evolutionarily significant units, distinct popu-
Trans. R. Soc. Lond. Ser. B, 351, 807–814. lation segments, and the Endangered Species Act: reply to
Taylor EB (1999) Species pairs of north temperate freshwater fishes: Pennock and Dimmick. Cons. Biol., 12, 718–721.
evolution, taxonomy, and conservation. Rev. Fish Biol. Fisheries, Wilson EO (1992) The Diversity of Life. W. W. Norton and
9, 299–324. Company, New York.
Vane-Wright RI, Humphries CJ, Williams PH (1991) What to Wilson RA (1999) Species. The MIT Press, Cambridge.
protect?-Systematics and the agony of choice. Biol. Cons., 55, Wheeler QD, Meier R (1997) Species Concepts and Phylogenetic
235–254. Theory: A Debate. Columbia University Press, New York.
Vogler AP, DeSalle AP (1994) Diagnosing units of conservation
management. Cons. Biol., 8, 354–363.