Syst. Biol.

56(6):879–886, 2007
Copyright ! c Society of Systematic Biologists
ISSN: 1063-5157 print / 1076-836X online
DOI: 10.1080/10635150701701083

Species Concepts and Species Delimitation

K EVIN DE Q UEIROZ
Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution, Washington,
DC 20560-0162, USA; E-mail: [email protected]

Abstract.— The issue of species delimitation has long been confused with that of species conceptualization, leading to a
half century of controversy concerning both the definition of the species category and methods for inferring the boundaries
and numbers of species. Alternative species concepts agree in treating existence as a separately evolving metapopulation
lineage as the primary defining property of the species category, but they disagree in adopting different properties acquired
by lineages during the course of divergence (e.g., intrinsic reproductive isolation, diagnosability, monophyly) as secondary
defining properties (secondary species criteria). A unified species concept can be achieved by treating existence as a separately
evolving metapopulation lineage as the only necessary property of species and the former secondary species criteria as
different lines of evidence (operational criteria) relevant to assessing lineage separation. This unified concept of species
has several consequences for species delimitation, including the following: First, the issues of species conceptualization
and species delimitation are clearly separated; the former secondary species criteria are no longer considered relevant
to species conceptualization but only to species delimitation. Second, all of the properties formerly treated as secondary
species criteria are relevant to species delimitation to the extent that they provide evidence of lineage separation. Third, the

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

presence of any one of the properties (if appropriately interpreted) is evidence for the existence of a species, though more
properties and thus more lines of evidence are associated with a higher degree of corroboration. Fourth, and perhaps most
significantly, a unified species concept shifts emphasis away from the traditional species criteria, encouraging biologists to
develop new methods of species delimitation that are not tied to those properties. [Species concept; species criteria; species
delimitation.]

Readers of Systematic Biology hardly need to be Fortunately, this species concept problem is not as seri-
reminded of the importance of species in biology. Ac- ous as it appears. Despite the obvious differences among
cording to various authors, species are one of the fun- contemporary alternative species concepts and defini-
damental units of biology, making them comparable in tions, they exhibit an underlying conceptual unity, which
importance to genes, cells, and organisms, some of the provides the basis for a unified concept of species. As a
fundamental units at lower levels of biological orga- consequence, biologists are now in a position to free our-
nization (e.g., Mayr, 1982; see also de Queiroz, 2005a). selves from seemingly endless debates about the con-
However, because species exist at a higher level of orga- cept of species and thus also the definition of the species
nization than the humans observing them, species also category. One of the most significant benefits of a uni-
are generally much larger and longer lived than their hu- fied species concept is that it allows biologists to ap-
man observers. Moreover, the connections among their proach the problem of species delimitation in a more
parts (i.e., organisms) are ephemeral. This makes it more straightforward way. In this paper, I will review the
or less impossible for humans to perceive entire species species concept problem and a proposal about how di-
simply by looking at them, as they do for cells and organ- verse species concepts can be unified, which I have pub-
isms, which is why biologists have symposia devoted to lished previously (de Queiroz, 1998, 1999, 2005a, 2005b,
the topic of species delimitation. 2005c). I will then examine some of the consequences
To complicate matters, for roughly the past half cen- of a unified species concept for the problem of species
tury, the issue of species delimitation has been confused delimitation.
by a problem involving the concept of species itself. The
problem is that currently different subgroups of biol-
ogists advocate different and at least partially incom- ALTERNATIVE S PECIES CONCEPTS
patible species concepts (reviewed by Mayden, 1997; de Table 1 is a list of alternative species concepts. The list
Queiroz, 1998; Harrison, 1998). Mayden (1997) listed 24 consists of major categories of alternative species con-
different named species concepts, and there are even cepts advocated by contemporary biologists, with the
more alternative definitions (where a definition is a categories defined in terms of the properties upon which
concise description of a concept, so that any given they are based. Most readers of this journal are likely
species concept may be associated with definitions that knowledgeable about at least some of these proposed
differ in minor details of wording). Many of these concepts, which include the familiar biological, ecologi-
concepts and their associated definitions are incom- cal, evolutionary, and phylogenetic concepts, among oth-
patible in that they can lead to different conclusions ers. Importantly, all of these concepts have advocates
concerning the boundaries and numbers of species. Thus, among contemporary biologists. In addition, many of
the species concept problem—that is, current disagree- the concepts are at least partially incompatible. For ex-
ments about the theoretical concept of the species—is ample, several authors have called attention to situations
closely tied to the issue of species delimitation—that in which adoption of the biological species concept leads
is, how to determine the boundaries and numbers of to the recognition of fewer species taxa than adoption
species from emperical data. of one of the alternative species concepts, such as the
879
880 SYSTEMATIC BIOLOGY VOL. 56

diagnosable version of the phylogenetic species concept ogists who adopt a multidisciplinary approach, or those
(e.g., Bremer and Wanntorp, 1979; Cracraft, 1983; Zink, who can step back from their own personal investments
1996). and research interests, all of the concepts seem to have
The reason for these incompatibilities has to do with some merits. They are all based on important biological
the different biological properties upon which several of properties.
the alternative concepts are based; for example, intrinsic
reproductive isolation in the case of the isolation ver- R ECONCILIATION
sion of the biological species concept, occupation of a
distinct niche or adaptive zone in the case of the ecologi- The Common Element
cal species concept, and fixed character state differences As I have argued previously (e.g., de Queiroz, 1998,
in the case of the diagnosable version of the phyloge- 1999, 2005a, 2005b, 2005c), the key to reconciling the alter-
netic species concept (Table 1). Moreover, these differ- native species concepts is identifying a common element,
ences in emphasis are to be expected, because the various which implies a single, more general, concept of species.
properties are of greatest interest to different subgroups Previous attempts to solve the species concept problem
of biologists. For example, reproductive incompatibili- have tended instead to obscure the solution by empha-
ties are of central importance to biologists who study sizing the differences, rather than the similarities, among
hybrid zones, niche differences are paramount for ecolo- rival concepts. As it turns out, all contemporary species
gists, and diagnosability and monophyly are fundamen- concepts share a common element and, equally impor-

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

tal for systematists. Similarly, morphological differences tant, that shared element is fundamental to the way in
are central for paleontologists and museum taxonomists, which species are conceptualized. The general concept
whereas genetic ones are key for population geneticists to which I refer equates species with separately evolv-
and molecular systematists. On the other hand, for biol- ing metapopulation lineages, or more specifically, with

TABLE 1. Alternative contemporary species concepts (i.e., major classes of contemporary species definitions) and the properties upon which
they are based (modified from de Queiroz, 2005). Properties (or the converses of properties) that represent thresholds crossed by diverging
lineages and that are commonly viewed as necessary properties of species are marked with an asterisk (*). Note that under the proposal for
unification described in this paper, the various ideas summarized in this table would no longer be considered distinct species concepts (see de
Queiroz, 1998, for an alternative terminology). All of these ideas conform to a single general concept under which species are equated with
separately evolving metapopulation lineages, and many of the properties (*) are more appropriately interpreted as operational criteria (lines of
evidence) relevant to assessing lineage separation.

Species concept Property(ies) Advocates/references

Biological Interbreeding (natural reproduction resulting in Wright (1940); Mayr (1942); Dobzhansky (1950)
viable and fertile offspring)
Isolation *Intrinsic reproductive isolation (absence of Mayr (1942); Dobzhansky (1970)
interbreeding between heterospecific organisms
based on intrinsic properties, as opposed to
extrinsic [geographic] barriers)
Recognition *Shared specific mate recognition or fertilization Paterson (1985); Masters et al. (1987); Lambert and
system (mechanisms by which conspecific Spencer (1995)
organisms, or their gametes, recognize one
another for mating and fertilization)
Ecological *Same niche or adaptive zone (all components of Van Valen (1976); Andersson (1990)
the environment with which conspecific
organisms interact)
Evolutionary Unique evolutionary role, tendencies, and Simpson (1951); Wiley (1978); Mayden (1997)
historical fate
(some interpretations) *Diagnosability (qualitative, fixed difference) Grismer (1999, 2001)
Cohesion Phenotypic cohesion (genetic or demographic Templeton (1989, 1998a)
exchangeability)
Phylogenetic Heterogeneous (see next four entries) (see next four entries)
Hennigian Ancestor becomes extinct when lineage splits Hennig (1966); Ridley (1989); Meier and Willmann
(2000)
Monophyletic *Monophyly (consisting of an ancestor and all of Rosen (1979); Donoghue (1985); Mishler (1985)
its descendants; commonly inferred from
possession of shared derived character states)
Genealogical *Exclusive coalescence of alleles (all alleles of a Baum and Shaw (1995); see also Avise and Ball
given gene are descended from a common (1990)
ancestral allele not shared with those of other
species)
Diagnosable *Diagnosability (qualitative, fixed difference) Nelson and Platnick (1981); Cracraft (1983); Nixon
and Wheeler (1990)
Phenetic *Form a phenetic cluster (quantitative difference) Michener (1970); Sokal and Crovello (1970); Sneath
and Sokal (1973)
Genotypic cluster (definition) *Form a genotypic cluster (deficits of genetic Mallet (1995)
intermediates; e.g., heterozygotes)
2007 DE QUEIROZ—SPECIES CONCEPTS AND SPECIES DELIMITATION 881

segments of such lineages. To clarify, here the term lineage processes: mutation, natural selection, migration (or the
refers to an ancestor-descendant series (Simpson, 1961; lack thereof), and genetic drift. The characters affected by
Hull, 1980) in this case of metapopulations or simply those processes, however, are highly diverse. They may
a metapopulation extended through time (cf. Simpson, be genotypic or phenotypic; qualitative or quantitative;
1951). It is not to be confused with a clade or mono- selectively advantageous, disadvantageous, or neutral;
phyletic group, which is sometimes also called a lineage and they may involve many different aspects of organis-
but is generally made up of several lineages (separate mal biology, including genetics, development, morphol-
branches). The term metapopulation refers to an inclu- ogy, physiology, and behavior. With regard to the species
sive population made up of connected subpopulations concept problem, the important point is that changes in
(Levins, 1970; Hanski and Gaggiotti, 2004). It is used these characters lead to the acquisition of a number of
here to distinguish species, which are traditionally con- different properties by diverging lineages. Thus, as the
sidered to reside at the higher end of the population- lineages diverge, they (or their component organisms)
level continuum, from populations at the lower end of become phenetically distinguishable. They become diag-
the continuum, such as demes and family groups. Fi- nosable in terms of fixed character states. Their genitalia,
nally, a species is not an entire metapopulation lineage gametes, and developmental systems become incompat-
but only a segment of such a lineage. The point here ible. Their mate recognition systems diverge to the point
is that species give rise to other species, thereby form- where the organisms no longer recognize one another as
ing (species level) lineages. Any given species is but one potential mates. They evolve distinctive ecologies. And

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

of many segments that make up such a species level they pass through polyphyletic, paraphyletic, and mono-
lineage. phyletic stages in terms of their component genes. The
problem is that these changes do not all occur at the same
time, and they do not even necessarily occur in a regular
The Differences order (de Queiroz, 1998). The reason this is a problem is
Given that all contemporary species concepts share that each of several different species concepts adopts a
the common view that species are (segments of) sepa- different property from this set as a defining (necessary)
rately evolving metapopulation lineages (for evidence, property of the species category. This is the reason that
see de Queiroz, 1998), it is instructive to consider how the different species concepts, despite sharing a common
so much disagreement about species concepts can exist fundamental element, can nonetheless lead to different
in spite of this general conceptual agreement. Clarifica- conclusions concerning which lineages deserve to be rec-
tion emerges when we consider the differences among ognized as species.
alternative species concepts in the context of the com- Figure 1 is a highly simplified diagram representing
mon element (i.e., the idea that species are separately the process of lineage separation and divergence (i.e.,
evolving metapopulation lineages). If we consider the speciation). The shades of gray represent the daugh-
common element to be the primary defining property of ter lineages becoming more and more different from
the species category, then the diversity of species con- one another through time, and the numbered lines
cepts can be accounted for by positing that many of represent the times at which they acquire different
the properties that underlie alternative species concepts properties relative to each other—for example, when
(those marked with an asterisk in Table 1) have been they become phenetically distinguishable, diagnosable,
implicitly treated as secondary defining properties of reciprocally monophyletic, reproductively incompatible,
the species category. The point is that most of the al- ecologically distinct, and so forth. This set of properties
ternative species concepts adopt different properties of forms a large gray zone within which alternative species
lineages as secondary defining properties. Thus, under concepts come into conflict. On either side of the gray
all species concepts, a species is a separately evolving zone, there will be unanimous agreement about the num-
metapopulation lineage, but under the isolation version ber of species. Before the acquisition of the first prop-
of the biological species concept, the lineage also has to erty, everyone will agree that there is a single species,
be intrinsically reproductively isolated from other lin- and after the acquisition of the last property, everyone
eages; under the ecological species concept, the lineage will agree that there are two. In between, however, there
also has to occupy a different niche; under the phenetic will be disagreement. The reason is that each of sev-
species concept, it also has to be phenetically distinguish- eral different contemporary species concepts adopts a
able; under the phylogenetic species concept (mono- different property (represented by the horizontal lines)
phyly version), it also has be monophyletic in terms of its as its cutoff for considering a separately evolving lin-
component genes, organisms, or subpopulations, and so eage to have become a species. Thus, some people will
forth. draw the cutoff relatively early in the process of diver-
The reason that these different secondary properties gence, perhaps where differences in quantitative char-
(secondary species criteria) lead to incompatible species acters make the lineages phenetically distinguishable.
concepts is that they arise at different times during the Others will draw the cutoff somewhat later, perhaps
process of speciation (here used in a general sense to where the lineages develop an intrinsic reproductive bar-
encompass all of the phenomena that have been em- rier. And still others will draw the cutoff later yet, per-
phasized by contemporary biologists). Speciation can be haps where both lineages form exclusive (monophyletic)
conceptualized in terms of a few general evolutionary groups in terms of multiple gene trees. This is cause of the
882 SYSTEMATIC BIOLOGY VOL. 56

The solution has two components. First, it retains

the common element—the general concept of species
as separately evolving metapopulation lineages (or,
2 Species
more properly, segments thereof). Second, it treats this
property as the only necessary property of species. In
other words, all the other properties that have previously
been treated as necessary properties of species—the
SC9 properties that created the incompatibilities among alter-
SC8
native species concepts—are reinterpreted as no longer
being defining (necessary) properties of the species cate-
SC7 gory. Instead, they are considered contingent properties:
Gray Zone SC6 properties that species may or may not acquire during
(1 vs. 2 species)
SC5
the course of their existence. In other words, lineages
do not have to be phenetically distinguishable, diagnos-
SC4 able, monophyletic, intrinsically reproductively isolated,
SC3 ecologically divergent, or anything else to be considered
SC2
species. They only have to be evolving separately from
other lineages. If this proposal is accepted, then it is

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

SC1 no longer appropriate to refer to the ideas in question
(Table 1) as different species concepts, and a revised ter-
minology is needed (see de Queiroz, 1998).
Despite denying that certain properties are necessary
1 Species properties of species, an important part of the reason
that the species concept resulting from the aforemen-
tioned proposal can be considered unified is that it
continues to embrace the various properties that have
been considered important under the rival species con-
cepts. Those properties—the former secondary species
FIGURE 1. Lineage separation and divergence (speciation) and
species concepts (after de Queiroz, 1998, 1999, 2005a). This highly sim-
criteria—remain important in two ways. First, they serve
plified diagram represents a single lineage (species) splitting to form as important operational criteria or lines of evidence
two lineages (species). The gradations in shades of gray represent the relevant to assessing the separation of lineages. These
daughter lineages diverging through time, and the horizontal lines la- properties, attributes such as phenetic distinguishabil-
beled SC (species criterion) 1 to 9 represent the times at which they ity, reciprocal monophyly, pre- and postzygotic repro-
acquire different properties (i.e., when they become phenetically dis-
tinguishable, diagnosable, reciprocally monophyletic, reproductively ductive isolation, and so forth, are all properties that
incompatible, ecologically distinct, etc.). The entire set of properties lineages acquire as they separate and diverge from
forms a gray zone within which alternative species concepts come into one another and therefore provide evidence of lineage
conflict. On either side of the gray zone, there will be unanimous agree- separation and divergence. Because species are con-
ment about the number of species. Before the acquisition of the first
property, everyone will agree that there is a single species, and after
ceptualized as (segments of) separately evolving lin-
the acquisition of the last property, everyone will agree that there are eages, evidence of lineage separation is evidence for the
two. In between, however, there will be disagreement. The reason is existence of different species. Thus, the properties in
that different contemporary species concepts adopt different proper- question remain directly relevant to the issue of species
ties (represented by the horizontal lines) as their species criteria—that delimitation.
is, as their cutoffs for considering a separately evolving lineage to have
become a species. A second way in which these properties remain im-
portant is that they can be used to define subcategories
of the species category—that is, to recognize different
species concept problem. This is the reason for the exis- classes of species based on the properties that those
tence of so many incompatible definitions of the species species possess. However, in contrast to the way that
category despite widespread agreement about the gen- classes of species have been named under the alterna-
eral nature of species. tive species concepts—that is, using overly general and
therefore misleading adjectives (e.g., biological species,
A Unified Species Concept ecological species, phylogenetic species etc.)—a more
The situation I have just described suggests a simple precise and therefore more useful terminology can be
solution to the species concept problem. The solution in- developed under the unified species concept using ad-
volves a relatively minor yet still fundamental shift in jectives that describe the properties of interest (e.g., re-
the way that species are conceptualized. It retains the el- productively isolated species, ecologically differentiated
ement that is common to all contemporary species con- species, monophyletic species, etc.). Subcategories of the
cepts, and it eliminates the conflicts between those rival species category are important in that they are composed
concepts without denying the importance of the prop- of those species that are relevant to addressing partic-
erties that underlie their obvious differences. In short, it ular biological questions. For example, a study of re-
represents a unified species concept. inforcement (Butlin 1987) requires species that exhibit
2007 DE QUEIROZ—SPECIES CONCEPTS AND SPECIES DELIMITATION 883

postmating reproductive incompatibilities, whereas a of defining the species category (species conceptualiza-
study that uses a species-level phylogeny to make in- tion) from the methodological problem of inferring the
ferences about historical biogeography might be better boundaries and numbers of species (species delimita-
served using species that exhibit monophyly. In any case, tion). Previously these two issues were commonly con-
the point is that a unified species concept would continue fused in that the same properties that were used to infer
to embrace all of the properties that have been consid- species boundaries and numbers were also considered
ered important by previous authors; it just would not necessary for a lineage to be regarded as a species (i.e., for
treat any of those properties as necessary properties of deciding when a lineage had diverged enough to be con-
species. sidered a species). Moreover, because different authors
It is appropriate to point out here that the unified considered different properties to be necessary, they
species concept just described is not a new species con- commonly disagreed about the boundaries and numbers
cept but simply the clear separation of the theoretical of species. In other words, the issue of species delim-
concept of species (as separately evolving metapopula- itation was intimately intertwined with that of species
tion lineages) from operational criteria (lines of evidence) conceptualization and hopelessly confused by disagree-
that are used for its empirical application. As such, it is ments about the species concept.
not surprising that several previously proposed charac- In contrast, under a unified species concept, the prop-
terizations of the species category correspond closely to erties in question are no longer considered necessary
the unified species concept. Thus, the ideas that Mayr properties of species. This situation clarifies the issue

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

(1963) termed the “interbreeding-population concept” of species delimitation by revealing that those proper-
(e.g., Wright, 1940; Mayr, 1942; Dobzhansky, 1950) and ties have nothing to do with the conceptual problem
later (e.g., Mayr, 1969, 1970) the “biological species con- of defining the species category. Instead, they are more
cept” referred, at least initially, to a general theoretical appropriately viewed as lines of evidence relevant to
concept of species (though restricted to sexually repro- the fundamentally different methodological (rather than
ducing organisms) that should not be confused with conceptual) problem of inferring the boundaries and
Mayr’s popular species definition, which incorporates numbers of species—that is, species delimitation. Thus,
the operational criterion of intrinsic reproductive isola- under a unified concept of species, there should no longer
tion (de Queiroz, 2005a). Similarly, the species definitions be any disagreements about the boundaries and numbers
of Simpson (1951, 1961) and Wiley (1978) do not include of species that result purely from disagreements about
operational criteria and thus correspond closely to the the definition of the species category. Instead, disagree-
unified species concept (de Queiroz, 1998, 1999). More- ments about species delimitation should result from dis-
over, Mayden (1997, 1999) has recognized both that these agreements or differences concerning one or more of
characterizations represent a common general concept of the following issues: the reliability of particular meth-
species and that many of the alternative views are dis- ods (i.e., for inferring lineage separation), the relevance
tinguished primarily by operational criteria. Hennig’s of particular data, temporal scale (years versus decades
(1966) characterization of species is similar to the gen- versus centuries, etc.), prospective versus retrospective
eral biological species concept. Its distinctive property— perspectives, and cases of incomplete lineage separation.
the extinction of ancestral species when they give rise
to descendant species—was adopted in the interests of
strict conformity to a nested, hierarchical model (Hennig, Relevance of Diverse Properties
1966:64; Meier and Willmann, 2000) and is not an opera- Another consequence of a unified species concept is
tional criterion for deciding when a lineage is sufficiently that many different properties are relevant to the issue
divergent to be considered a species. And finally, the of species delimitation. Under most of the alternative
properties that Templeton (1989) identified as cohesion species concepts, in which various properties acquired
mechanisms relevant to his species definition, though by diverging lineages were viewed as necessary proper-
related to several operational species criteria, represent ties of species, a different one of these properties was con-
phenomena that are hypothesized to be responsible for sidered necessary under each alternative concept. This
the existence of metapopulation lineages (see Pigliucci, practice created the undesirable situation in which each
2003; de Queiroz, 2005c). alternative species concept unduly emphasized only one
of the various properties at the expense of the others
(Bush, 1995), with biologists engaged in an ongoing bat-
CONSEQUENCES FOR S PECIES D ELIMITATION tle over which property was to be considered the most
A unified species concept has consequences for the important.
issue of species delimitation, some of which I will briefly In contrast, under a unified species concept, most of
describe in the remainder of this paper. the properties emphasized under the alternative con-
cepts should be considered relevant to the issue of species
delimitation. In the context of a unified species concept,
Conceptualization versus Delimitation any property that provides evidence of lineage separa-
One of the most important consequences of a unified tion is relevant to inferring the boundaries and numbers
species concept is that it clarifies the issue of species de- of species. Considering the properties that have previ-
limitation by clearly separating the conceptual problem ously been adopted as secondary species criteria (those
884 SYSTEMATIC BIOLOGY VOL. 56

marked with an asterisk in Table 1), either the prop- Although presence of a single property provides evi-
erty itself (intrinsic reproductive isolation, monophyly, dence for lineage separation, a highly corroborated hy-
exclusive coalescence, diagnosability, deficits of genetic pothesis of lineage separation (i.e., of the existence of
intermediates), or its converse (incompatible fertiliza- separate species) requires multiple lines of evidence. In
tion systems, different niches, phenetic distinguishabil- general, the farther along lineages are in the process of
ity), provides evidence of lineage separation. Thus, all of divergence, the larger the number of differences they can
those properties are relevant (as lines of evidence) to the be expected to have acquired relative to one another,
problem of species delimitation. and therefore the easier it should be to find evidence
of separation. Conversely, the earlier lineages are in the
process of divergence, the more difficult it should be to
Quantity of Evidence find evidence of separation. In any case, multiple lines
Viewing the properties in question as evidence of lin- of evidence—that is, the possession of several proper-
eage separation has additional consequences. One is that ties that arise during lineage divergence—result in more
any evidence of lineage separation is sufficient to in- highly corroborated hypotheses of lineage separation,
fer the existence of separate species (compare Mayden, and thus of the existence of different species. This point
1999). To the extent that the possession (by a set of popu- may seem obvious, and some people have been using
lations) of even a single relevant property provides such multiple lines of evidence for years. Nonetheless, the ex-
evidence, it may be considered evidence for the existence istence of rival species concepts has worked against these

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

of a species. This is not to say that the properties are in- efforts by effectively asking people to choose a preferred
fallible; on the contrary, any line of evidence can be mis- (single) operational criterion.
leading if interpreted inappropriately. For example, the
existence of separate species is commonly inferred from
reciprocal monophyly of the alleles at a given locus in Alternatives to the Traditional Properties
allopartically or parapatrically distributed sets of popu- Among the most important consequences of adopting
lations (e.g., Moritz, 1994; Avise and Wollenberg, 1997). a unified species concept is that, by emphasizing sep-
However, if the locus is maternally inherited, as in the arately evolving lineages over contingent properties of
case of mitochondrial DNA, then a pattern of recipro- those lineages, it encourages biologists to shift their at-
cal monophyly can also result from low dispersal dis- tention away from the traditional species criteria and de-
tances of females even when autosomal and paternally velop new methods for species delimitation. Although
inherited genes are being exchanged regularly between properties such as intrinsic reproductive isolation, diag-
the same sets of populations (e.g., Irwin, 2002). In other nosability, (reciprocal) monophyly, and the like are cer-
words, two or more species might be inferred from such tainly relevant to the issue of lineage separation, many
data even though the populations in question form a sin- of them represent somewhat artificial cutoffs in the con-
gle metapopulation lineage. Thus, the point is not that tinuous process of divergence. Moreover, most of these
the presence of a single property guarantees that a set of properties are not very useful for detecting lineage sep-
populations possessing that property represents a sepa- aration in the early stages of divergence. In this context,
rate lineage (i.e., a species) but only that the presence of the development of new methods to test hypotheses of
a single property constitutes evidence (which is always lineage separation that are no longer based on the tra-
fallible) supporting that hypothesis. ditional species criteria represents significant progress.
On the other hand, the absence of any one or more of For example, consider new methods for species delimi-
the properties in question does not constitute evidence tation being developed in the context of coalescent the-
contradicting a hypothesis of lineage separation. In other ory (e.g., Knowles and Carstens, 2007). These methods
words, a lineage might lack one or more of those prop- use information from gene trees, which is the same sort
erties even if it is evolving separately from all other lin- of information that is commonly used to assess mono-
eages. The reason, of course, is that the lineage simply phyly under monophyletic and genealogical versions of
may not yet have evolved the properties, as might be the so-called phylogenetic species concept. However, in
expected if it is still in the early stages of divergence. the case of these new coalescent-based methods, mono-
Thus, an asymmetry exists concerning the evidence pro- phyly is not the focus. In fact, the methods in question
vided by the properties in question: the presence of any can provide evidence for lineage separation even when
one of those properties constitutes evidence for lineage none of the sampled loci exhibits monophyly within the
separation, but the absence of the same property does not sets of populations under consideration (Knowles and
constitute evidence against lineage separation—that is, Carstens, 2007).
against the hypothesis of separate species. When consid- Other new methods relevant to species delimitation
ering only the properties in question, only the absence make more direct use of geographic information than
of all of those properties should be considered evidence under traditional approaches. Geographic information
against the hypothesis that two (or more) sets of popula- is crucial because nearly all species exhibit geographic
tions represent different species, but even this is negative variation, and it is possible for larger differences to exist
evidence. On the other hand, it would seem to go with- between populations within the same old and geographi-
out saying that recognizing a species is inappropriate in cally widespread species than between populations from
the absence of any positive evidence for its existence. different but recently separated species (de Queiroz and
2007 DE QUEIROZ—SPECIES CONCEPTS AND SPECIES DELIMITATION 885

Good, 1997). This situation calls into question all meth- interminable debates about the definition of the species
ods that adopt as an operational criterion a particular category. Moreover, it provides a unified context for
level of divergence, whether derived from previously understanding the relevance of diverse methods to the
studied cases (e.g., Lefébure et al., 2006), theoretical mod- problem of species delimitation (i.e., as methods for eval-
els (e.g., Pons et al., 2006), or based on a more arbitrary uating whether sets of populations constitute separately
criterion, such as the threshold beyond which parsimony evolving lineages) and thus also for integrating the infor-
will no longer correctly estimate the number of muta- mation provided by different species delimitation meth-
tions with a probability greater than or equal to 0.95 (e.g., ods in empirical applications.
Cardoso and Vogler, 2005). (Such methods may still be
useful for obtaining first approximations when screen-
ing large numbers of samples from understudied taxa, ACKNOWLEDGMENTS
as in the cited papers.) Geographic information is neces- I thank J. Wiens for organizing the 2006 SSB symposium on species
sary to distinguish true discontinuities (i.e., lineage sep- delimitation and for inviting me to contribute this paper. L. Knowles
aration) from differentiation that occurs within species provided valuable information about methods based on coalescent
theory, and J. Sites, J. Wiens, and an anonymous reviewer provided
as the result of phenomena such as clines and isolation comments on an earlier version. I have previously acknowledged the
by distance. contributions of numerous colleagues to the development of my views
Although the direct use of geographic information on species concepts (see de Queiroz, 1998, 1999, 2005a, 2005b, 2005c).
in methods of (or related to) species delimitation is an

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

old idea (e.g., Gabriel and Sokal, 1969; Sokal and Oden,
1978a, 1978b; Sokal, 1979), with certain exceptions, in- R EFERENCES
cluding Mantel tests (Sokal, 1979) and nested clade anal-
Andersson, L. 1990. The driving force: Species concepts and ecology.
ysis (Templeton, 1998b), such methods are underused. In Taxon 39:375–382.
addition, most of the traditional species criteria (Table 1) Avise, J. C., and R. M. Ball, Jr. 1990. Principles of genealogical concor-
do not explicitly incorporate geographic information. dance in species concepts and biological taxonomy. Oxf. Surv. Evol.
This situation is likely to change with the recent and rapid Biol. 7:45–67.
development of geographic information system technol- Avise, J. C., and K. Wollenberg. 1997. Phylogenetics and the origin of
species. Proc. Natl. Acad. Sci. USA 94:7748–7755.
ogy and its application to problems involving species Baum, D. A., and K. L. Shaw. 1995. Genealogical perspectives on the
(see Raxworthy, 2007; Rissler and Apodaca, 2007). More- species problem. Pages 289–303 in Experimental and molecular ap-
over, methods that incorporate geographic information proaches to plant biosystematics (P. C. Hoch, and A. G. Stephenson,
have recently been proposed that are not based on tra- eds.). Missouri Botanical Garden, St. Louis.
Bremer, K., and H.-E. Wanntorp. 1979. Geographic populations or bio-
ditional operational criteria but instead are designed to logical species in phylogeny reconstruction. Syst. Zool. 28:220–224.
identify abrupt changes in surfaces defined by (genetic Bush, G. L. 1995. Reply from G. L. Bush. Trends Ecol. Evol. 10:38.
or phenotypic) characters that are indicative of at least Cardoso, A., and A. P. Vogler. 2005. DNA taxonomy, phylogeny and
partial lineage separation (e.g., Manel et al., 2003; Manni Pleistocene diversification of the Cicindela hybrida species group
et al., 2004; Miller, 2005). I will not say more about these (Coleoptera: Cicindelidae). Mol. Ecol. 14:3531–3546.
Cracraft, J. 1983. Species concepts and speciation analysis. Curr. Or-
or other new approaches to species delimitation, several nithol. 1:159–187.
of which will be described by other contributors to this de Queiroz, K. 1998. The general lineage concept of species, species cri-
issue (see also reviews by Sites and Marshall, 2003, 2004). teria, and the process of speciation: A conceptual unification and
My purpose is not to review new or existing methods but terminological recommendations. Pages 57–75 in Endless forms:
Species and speciation (D. J. Howard, and S. H. Berlocher, eds.).
only to point out that some of them represent movement Oxford University Press, New York.
away from the traditional species criteria toward a more de Queiroz, K. 1999. The general lineage concept of species and the
explicit treatment of the problem of species delimitation defining properties of the species category. Pages 49–89 in Species:
as the inference of separately evolving lineages. New interdisciplinary essays (R. A. Wilson, ed.). MIT Press, Cam-
bridge, Massachusetts.
de Queiroz, K. 2005a. Ernst Mayr and the modern concept of species.
CONCLUSION Proc. Natl. Acad. Sci. USA 102:6600–6607.
de Queiroz, K. 2005b. A unified concept of species and its consequences
My goal in this contribution has been to describe a for the future of taxonomy. Proc. Calif. Acad. Sci. 56:196–215.
unified concept of species and some of its consequences de Queiroz, K. 2005c. Different species problems and their resolution.
BioEssays 27:1263–1269.
for the problem of species delimitation. Several of these de Queiroz, K., and D. A. Good. 1997. Phenetic clustering in biology:
conclusions have been proposed by other authors; how- A critique. Q. Rev. Biol. 72:3–30.
ever, a long-standing confusion of the general concept Dobzhansky, T. 1950. Mendelian populations and their evolution. Am.
of species with the operational criteria by which species Nat. 84:401–418.
Dobzhansky, T. 1970. Genetics of the evolutionary process. Columbia
are recognized and the misleading terminology that has University Press, New York.
helped to perpetuate it (see Table 1) has interfered with Donoghue, M. J. 1985. A critique of the biological species concept and
their more thorough and widespread acceptance. The recommendations for a phylogenetic alternative. Bryologist 88:172–
realization that all modern species concepts are vari- 181.
ations on a common theme serves as the basis for a Gabriel, K. R., and R. R. Sokal. 1969. A new statistical approach to
geographic variation analysis. Syst. Zool. 18:259–278.
unified species concept under which conceptual and Grismer, L. L. 1999. An evolutionary classification of reptiles on
methodological issues are clearly separated. This uni- islands in the Gulf of California, Mexico. Herpetologica 55:446–
fied concept of species represents an end to seemingly 469.
886 SYSTEMATIC BIOLOGY VOL. 56

Grismer, L. L. 2001. An evolutionary classification and checklist of am- Nelson, G., and N. I. Platnick. 1981. Systematics and biogeography.
phibians and reptiles on the Pacific islands of Baja California, Mexico. Columbia University Press, New York.
Bull. South. Calif. Acad. Sci. 100:12–23. Nixon, K. C., and Q. D. Wheeler. 1990. An amplification of the phylo-
Hanski, I. and O. E. Gaggiotti. 2004. Metapopulation biology: Past, genetic species concept. Cladistics 6:211–223.
present, and future. Pages 3–22 in Ecology, genetics, and evolution Paterson, H. E. H. 1985. The recognition concept of species. Pages 21–
of metapopulations (I. Hanski and O. E. Gaggiotti, eds.). Elsevier, 29 in Species and speciation (E. S. Vrba, ed.). Transvaal Museum,
Amsterdam. Pretoria.
Harrison, R. G. 1998. Linking evolutionary pattern and process. Pages Pigliucci, M. 2003. Species as family resemblance concepts. BioEssays
19–31 in Endless forms: Species and speciation (D. J. Howard, and S. 25:596–602.
H. Berlocher, eds.). Oxford University Press, New York. Pons, J., T. G. Barraclough, J. Gomez-Zurita, A. Cardoso, D. P. Duran, S.
Hennig, W. 1966. Phylogenetic systematics. University of Illinois Press, Hazell, S. Kamoun, W. D. Sumlin, and A. P. Vogler. 2006. Sequence-
Urbana. based species delimitation for the DNA taxonomy of undescribed
Hull, D. L. 1980. Individuality and selection. Annu. Rev. Ecol. Syst. Insects. Syst. Biol. 55:595–609.
11:311–332. Raxworthy, C. J., C. Ingram, N. Rabibisoa, and R. Pearson. 2007. Ap-
Irwin, D. E. 2002. Phylogeographic breaks without geographic barriers plications of ecological niche modeling for species delimitation: A
to gene flow. Evolution 56:2383–2394. review and empirical evaluation using day geckos (Phelsuma) from
Knowles, L. L., and B. C. Carstens. 2007. Delimiting species without Madagascar. Syst. Biol. 56:This issue.
monophyletic gene trees. Syst. Biol. 56:This issue. Ridley, M. 1989. The cladistic solution to the species problem. Biol.
Lambert, D. M., and H. G. Spencer (eds.) 1995. Speciation and the recog- Philos. 4:1–16.
nition concept: Theory and application. Johns Hopkins University Rissler, L. J., and J. J. Apodaca. 2007. Adding more ecology into species
Press, Baltimore. delimitation: Ecological niche models and phylogeography help de-
Lefébure, T., C. J. Douady, M. Gouy, J. Gibert. 2006. Relationship be- fine cryptic species in the black salamander (Aneides flavipunctatus).

Downloaded from http://sysbio.oxfordjournals.org/ by guest on July 16, 2012

tween morphological taxonomy and molecular divergence within Syst. Biol. 56:This issue.
Crustacea: Proposal of a molecular threshold to help species delim- Rosen, D. E. 1979. Fishes from the uplands and intermontane basins of
itation. Mol. Phylogenet. Evol. 40:435–447. Guatemala: Revisionary studies and comparative geography. Bull.
Levins, R. 1970. Extinction. Pages 77–107 in Some mathematical prob- Am. Mus. Nat. Hist. 162:267–376.
lems in biology (M. Desternhaber, ed.). American Mathematical So- Simpson, G. G. 1951. The species concept. Evolution 5:285–298.
ciety, Providence, Rhode Island. Simpson, G. G. 1961. Principles of animal taxonomy. Columbia Uni-
Mallet, J. 1995. A species definition for the modern synthesis. Trends versity Press, New York.
Ecol. Evol. 10:294–299. Sites, J. W., and J. C. Marshall. 2003. Delimiting species: A renaissance
Manel, S., M. K. Schwartz, G. Luikart, and P. Taberlet. 2003. Land- issue in systematic biology. Trends Ecol. Evol. 18:462–470.
scape genetics: Combining landscape ecology and population ge- Sites, J. W., and J. C. Marshall. 2004. Operational criteria for delimiting
netics. Trends Ecol. Evol. 18:189–197. species. Annu. Rev. Ecol. Evol. Syst. 35:199–227.
Manni, F., E. Guérard, and E. Heyer. 2004. Geographic patterns of (ge- Sneath, P. H. A., and R. R. Sokal. 1973. Numerical taxonomy: The prin-
netic, morphologic, linguistic) variation: How barriers can be de- ciples and practice of numerical classification. W. H. Freeman, San
tected using Monmonier’s algorithm. Hum. Biol. 76:173–190. Francisco.
Masters, J. C., R. J. Rayner, I. J. McKay, A. D. Potts, D. Nails, J. W. Sokal, R. R. 1979. Testing statistical significance of geographic variation
Ferguson, B. K. Weissenbacher, M. Allsopp, and M. L. Anderson. patterns. Syst. Zool. 28:227–232.
1987. The concept of species: Recognition versus isolation. S. Afr. J. Sokal, R. R., and T. J. Crovello. 1970. The biological species concept: A
Sci. 83:534–537. critical evaluation. Am. Nat. 104:127–153.
Mayden, R. L. 1997. A hierarchy of species concepts: The denouement Sokal, R. R., and N. L. Oden. 1978a. Spatial autocorrelation in biology
in the saga of the species problem. Pages 381–424 in Species: The units 1. Methodology. Biol. J. Linn. Soc. 10:199–228.
of biodiversity (M. F. Claridge, H. A. Dawah, and M. R. Wilson, eds.). Sokal, R. R., and R. L. Oden. 1978b. Spatial autocorrelation in bi-
Chapman and Hall, London. ology 2. Some biological implications and four applications of
Mayden, R. L. 1999. Consilience and a hierarchy of species concepts: evolutionary and ecological interest. Biol. J. Linn. Soc. 10:229–
Advances toward closure on the species puzzle. J. Nematol. 31:95– 249.
116. Templeton, A. R. 1989. The meaning of species and speciation: A ge-
Mayr, E. 1942. Systematics and the origin of species. Columbia Univer- netic perspective. Pages 3–27 in Speciation and its consequences
sity Press, New York. (D. Otte and J. A. Endler, eds.). Sinauer Associates, Sunderland,
Mayr, E. 1963. Animal species and evolution. Belknap Press of Harvard Massachusetts.
University Press, Cambridge, Massachusetts. Templeton, A. R. 1998a. Species and speciation: Geography, popula-
Mayr, E. 1969. Principles of systematic zoology. McGraw-Hill, New tion structure, ecology, and gene trees. Pages 32–43 in Endless forms:
York. Species and speciation (D. J. Howard and S. H. Berlocher, eds.). Ox-
Mayr, E. 1970. Populations, species, and evolution. Belknap Press of ford University Press, New York.
Harvard University Press, Cambridge, Massachusetts. Templeton, A. R. 1998b. Nested clade analysis of phylogeographic data:
Mayr, E. 1982. The growth of biological thought: Diversity, evolution, Testing hypotheses about gene flow and population history. Mol.
and inheritance. Belknap Press of Harvard University Press, Cam- Ecol. 7:381–397.
bridge, Massachusetts. Van Valen, L. 1976. Ecological species, multispecies, and oaks. Taxon
Meier, R., and R. Willmann. 2000. The Hennigian species concept. Pages 25:233–239.
30–43 in Species concepts and phylogenetic theory (Q. D. Wheeler Wiley, E. O. 1978. The evolutionary species concept reconsidered. Syst.
and R. Meier, eds.). Columbia University Press, New York. Zool. 27:17–26.
Michener, C. D. 1970. Diverse approaches to systematics. Evol. Biol. Wright, S. 1940. The statistical consequences of Mendelian heredity
4:1–38. in relation to speciation. Pages 161–183 in The new systematics (J.
Miller, M. P. 2005. Alleles in space (AIS): Computer software for the Huxley, ed.). Oxford University Press, London.
joint analysis of interindividual spatial and genetic information. J. Zink, R. M. 1996. Bird species diversity. Nature 381:566.
Hered. 96:722–724.
Mishler, B. D. 1985. The morphological, developmental, and phylo-
genetic basis of species concepts in bryophytes. Bryologist 88:207–
214. First submitted 1 November 2006; reviews returned 16 January 2007;
Moritz, C. 1994. Defining ’evolutionarily significant units’ for conser- final acceptance 21 August 2007
vation. Trends Ecol. Evol. 9:373–375. Guest Associate Editor: John Wiens