Syst. Biol. 57(4):628–646, 2008
c Society of Systematic Biologists
Copyright ISSN: 1063-5157 print / 1076-836X online
DOI: 10.1080/10635150802302443
An Integrative Method for Delimiting Cohesion Species: Finding the Population-Species
Interface in a Group of Californian Trapdoor Spiders with Extreme Genetic Divergence
and Geographic Structuring
J ASON E. B OND AND AMY K. S TOCKMAN
Department of Biology, East Carolina University, Greenville, North Carolina 27858, USA; E-mail: [email protected] (J.E.B.)
Abstract.—Here we present an objective, repeatable approach to delineating species when populations are divergent and
highly structured geographically using the Californian trapdoor spider species complex Aptostichus atomarius Simon as a
model system. This system is particularly difficult because under strict criteria of geographical concordance coupled with
estimates of genetic divergence, an unrealistic number of population lineages would qualify as species (20 to 60). Our
novel phylogeographic approach, which is generally applicable but particularly relevant to highly structured systems, uses
genealogical exclusivity to establish a topological framework to examine lineages for genetic and ecological exchangeability
in an effort to delimit cohesion species. Both qualitative assessments of habitat and niche-based distribution modeling are
employed to evaluate selective regime and ecological interchangeability among genetic lineages; adaptive divergence among
populations is weighted more heavily than simple geographical concordance. Based on these analyses we conclude that
five cohesion species should be recognized, three of which are new to science. [Aptostichus stephencolberti; Cohesion species;
DNA barcoding; ecological interchangeability; Mygalomorphae; niche-based distribution modeling; species delineation.]
Although it is generally accepted that species comprise
lineages (De Queiroz, 1999, Sites and Marshall, 2004),
how lineages are diagnosed or recognized as species
remains a strong point of contention (Wake, 2006). It
is this point of contention that has captured the interest, thought, and imagination of many practicing systematists and evolutionary biologists because species
constructs, by definition, convey an almost immeasurable number of questions and hypotheses—Are populations isolated reproductively? Are they ecologically
interchangeable? Does migration occur and at what frequency? Are lineages genealogically exclusive (what
genes or subset of genes are exclusive)? Does introgression occur when populations come into contact? Despite the complexity of questions raised by virtually any
species construct, many of them lack the requisite rigor
they seemingly imply. For example, many species hypotheses are based on a single character system (e.g.,
genitalia) or single class of characteristics (somatic morphology) and are sometimes based on very few individual specimens (in some cases only one; see Huber, 2003,
for a summary of spider examples). Delimiting species
that represent real evolutionary lineages that summarize
a set of well-founded hypotheses requires an integrative
approach that accounts for multiple lines of evidence,
if informed decisions regarding conservation of habitat
and populations are to be made, and if we are to study
speciation pattern, diversification, and process in a rigorous manner.
Highly structured, genetically divergent, yet morphologically homogenous species (e.g., nonvagile cryptic
species), although often ignored or overlooked, provide one of the greatest challenges to delimiting species
(e.g., Bond et al., 2001; Hedin and Wood, 2002; Sinclair
et al., 2004; Boyer et al., 2007). Populations, or very small
groups of populations, constitute divergent genetic lineages but present somewhat of a contradiction because
they lack the “requisite” characteristics often used when
delimiting species. Morphological approaches to species
delimitation in many of these groups grossly oversimplify and underestimate diversity (Bond et al., 2001; Bickford et al., 2006); in short, these traditional applications
fail if our interests extend beyond what can simply be
diagnosed with a visual and/or anthropomorphic-based
assessment.
When genetically divergent, morphologically equivalent lineages exhibit microallopatric population structuring, lineage-based approaches to delineating species
are further confounded; virtually all population groups
are independent lineages and, thus, qualify as species
(Agapow et al., 2004; Hickerson et al., 2006), likely yielding specious results. For example, independent networks
(constructed using statistical parsimony, TCS; Templeton et al., 1987, 1992) are often considered a priori as
putative species available for further delimitation—for
example, “The status of these independent networks as
species was not questioned . . . ” (Pons et al., 2006). Under such a criterion, every population, and some haplotypes, composing a highly structured taxon potentially
fractionate into “putative species” (see Boyer et al., 2007).
Further confounding the issue of species delimitation in
nonvagile organisms are matters related to geographic
sampling (Hedin and Wood, 2002) and that deep phylogeographic breaks can arise stochastically (Irwin, 2002).
Although advocates of sequence-based species delimitation argue that DNA-based taxonomy should rely on
sequence information alone (e.g., Pons et al., 2006), organisms with highly structured systems, in particular,
dictate an integrative approach to investigating species
boundaries—approaches that, by design, must incorporate lines of evidence beyond simple molecular markers (Will et al., 2005) because population structuring and
divergence may only represent a maximum point of divergence; that is, populations may have diverged genetically long before they would be recognized as species.
Two major issues must be considered for this class of
organismal group if an integrative approach to species
delimitation is to be employed. First, how are lineages
628
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
delineated as “candidate” species; that is, how does one
identify “basal” lineages (“the oldest split or splits within
a species,” sensu Wiens and Penkrot, 2002) in a highly
structured, divergent system? And, second, once candidate lineages have been identified, what types of criteria
are used to delimit them as species such that adaptive diversity and evolutionary potential (Crandall et al., 2000)
are captured as part of the species recognition “process”? Herein we outline an operational evolutionary
lineage-based approach for delineating cohesion species
(Templeton, 1989) for organismal groups that exhibit
pronounced population genetic structure on a very fine
geographical scale. For a set of populations to qualify as
a cohesion species, they must be derived from a single evolutionary lineage (i.e., they share common ancestry) and
must be genetically exchangeable and/or ecologically interchangeable (Templeton, 2001). The cohesion species concept is ideal in its general applicability across all types
of taxa (Hull, 1997) and its ability to be couched as a
set of testable null hypotheses (Templeton, 2001). Our
approach, outlined below, employs an iterative methodology and set of rules to evaluate sister lineages for cohesion species criteria.
We apply our approach to the Aptostichus atomarius
trapdoor spider species complex, a group that is morphologically homogenous and nonvagile. We first evaluate
species crypsis among A. atomarius populations using
morphometric techniques to rigorously test the null hypothesis that these populations are morphologically homogenous for a set of traditionally used features (e.g.,
genitalia and secondary sexual characteristics). Second,
we evaluate the genetic structuring of populations using a set of mitochondrial and nuclear markers; a reconciled gene tree approach is used to identify basal lineages.
Third, we evaluate these lineages using cohesion species
criteria to test the null hypothesis that divergent lineages comprise a single species. Genetic exchangeability
(gene flow) is evaluated by assessing geographical concordance (agreement between geography and exclusive
haplotype groups) and evidence of allopatric fragmentation. Ecological interchangeability (habitat or niche of
a species) is evaluated using niche-based distribution
modeling (Stockman and Bond, 2007) and by assessing habitat differences through evaluation of adaptive
divergence within a phylogeographic framework. Genetically divergent lineages with parapatric geographic
distributions that are ecologically interchangeable are retained as a single species (sensu Crandall et al., 2000).
This approach seeks to evaluate populations separated
by many genetic changes such that species are not oversplit (Agapow et al., 2004), while balancing the reality
that these population groups represent independent evolutionary lineages.
The Aptostichus atomarius Complex
The Aptostichus atomarius (Araneae: Mygalomorphae,
Cyrtaucheniidae) species complex is an assemblage of
populations of trapdoor spider (Bond and Opell, 2002;
Bond and Hedin, 2006) distributed throughout southern
629
California and extending northward along the coastal
counties into Del Norte County (Fig. 1). These fossorial spiders are sit-and-wait predators that build silklined burrows covered by silk-soil trapdoors. The genus
Aptostichus is species rich, consisting of 30+ species
(most undescribed) found predominantly throughout
southern California, a recognized biodiversity hotspot
(Myers et al., 2000; Schoenherr, 1992). Species diagnosis is
based largely on male secondary sexual characteristics—
morphological modifications of the first walking legs
(see Fig. 1), typically referred to as the “mating clasper.”
Like many other mygalomorphs (tarantulas, trapdoor
spiders, and their kin), Aptostichus life-history characteristics manifestly differ from other spiders (Bond et al.,
2006): they are long lived (15 to 30 years), requiring 5
or more years to reach maturity (Main, 1978), and are
sedentary. Many mygalomorphs do not balloon—a common dispersal mode where spiders use silk draglines to
capture air currents. As a consequence, mygalomorphs
are prone to extreme population structuring (see Bond
et al., 2001, 2006; Starrett and Hedin, 2007; Arnedo and
Ferrández, 2007).
Based on a qualitative assessment of male mating
clasper morphology, all known populations of Aptostichus atomarius have been treated as a single species
(Bond, 1999); this diagnostic structure does not appear
to differ significantly for any of the populations surveyed (Fig. 1a). Variation among individuals collected
at the same location is comparable to that for individuals from widely separated sites (∼1500 specimens examined; Bond, 1999).
M ATERIALS AND M ETHODS
Specimen Sampling and Vouchering
We attempted to obtain ≥3 specimens (following
Wiens and Penkrot, 2002) per locality throughout the
known distribution of Aptostichus atomarius (see online appendix, available at www.systematicbiology.org).
However, due to the cryptic nature of these spiders’ burrows and overall rarity at some sites (in such cases purposely collecting fewer), some localities are represented
by <3 individuals. Each specimen was assigned a unique
voucher number and haplotype designation; all specimens collected as part of this study will be deposited
in various museum collections (see online appendix). To
establish the exclusivity of the focal taxon (Wiens and
Penkrot, 2002), 11 additional species were sampled.
Analyses of Mating Clasper and Spermathecae Shape
Digital images of male mating claspers (Leg I, left side
when possible) were recorded at the highest magnification possible for all available specimens using a digital
camera mounted on a stereomicroscope. Size calibrations were assessed by photographing a stage micrometer (accuracy to 0.02 mm) at magnification. Spermathecae
were removed from female specimens, cleared in clove
oil, temporarily mounted on slides, and photographed
using a compound microscope equipped with a digital
630
SYSTEMATIC BIOLOGY
VOL. 57
FIGURE 1. (a) Generalized distributional pattern of Aptostichus atomarius. Color scheme and geographic subdelineations correspond to clades
delimited by our phylogenetic analyses (legend corresponds to Fig. 3). Lines following coastal contours approximate distributions of clades
restricted to coastal dunes. Tibia and metatarsal segments are shown for the first left leg of male specimens from key areas sampled across
the species complex; spination patterns, particularly those at the distal-most aspect of the tibia, appear invariant. Aj, As, Am notation are to
denote references made in the taxonomic appendix. (b) PCA of mating clasper (left panel) and spermathecae (right panel) measurements; colors
correspond to populations above; point labels refer to collection accessions and county (museum material) or haplotype designations (this study).
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
camera (measurement from left side spermathecal bulb
when possible; size calibrations as noted above). Images were analyzed using the computer program ImageJ (Rasband, 2007). Quantitative measurements taken
from each mating clasper comprised tarsal, metatarsal,
tibia, proximal/ventral metatarsal depression length,
metatarsal apophysis height, and metatarsus width.
Meristic measurements comprised spine counts from the
following mating clasper articles/regions: distal prolateral tibia, retrolateral tibia, and prolateral patella, tibia,
and metatarsus. Spermathecal measurements comprised
the following: total width at base, base height, length
of bulb stalk, terminal bulb diameter, and separation of
bulb and base. Male mating clasper metatarsus and tarsus lengths were scaled as ratios of tibia size. Metatarsal
height was scaled to proximal metatarsal width and
the metatarsal depression length was scaled as a ratio
of total metatarsal length. Tibia length was retained in
subsequent analysis as a general metric of size. Spermathecal base height, stalk length, bulb diameter, and
separation were all scaled as a ratio of the total spermathecal width at the base. The total width at the base was
retained as a general metric of spermathecae size. Principal component analysis (PCA) of 10 mating clasper and 5
spermathecae measurements (analyzed separately) were
conducted using the computer program PC-ORD version 4 (McCune and Mefford, 1999).
DNA Preparation and Sequencing
Legs were removed from each specimen and preserved
in RNAlater (Qiagen, Valencia, CA) and stored at −80◦ C;
whole specimens were preserved in 80% ethanol. Genomic DNA was extracted using the Qiagen DNeasy
Tissue Kit. Standard PCR protocols were used to amplify an approximately 1500–base pair region of the mitochondrial genome spanning the region coding for the
12S rRNA, val-tRNA, and 16S rRNA genes (partial fragments of the 12S and 16S genes were amplified using the
primers LR-J-12887 CCGCTCTGAACTCAGATCACGT
and SR-N-14612spid AAGACAAGGATTAGATACCCT)
for all specimens included in the study. The internal transcribed spacer (ITS) units 1 and 2 were
amplified using the primers CAS18sF1 TACACACCGCCCGTCGCTACTA and CAS28sB1d TTCTTTTCCTCCSCTTAYTRATATGCTTAA and protocols as outlined in
Ji et al. (2003) for a subset of taxa representing each
of the major mitochondrial gene population groups (4–
5/mtDNA clade). Mitochondrial PCR products were purified and sequenced with both amplification primers
and an additional internal sequencing primer (LR-J13XXXa GGCAAATGATTATGCTACC) using an ABI automated DNA sequencer (Applied Biosystems, Foster
City, CA). ITS products were purified and sequenced
using the amplification primers. In an effort to detect
any possible intraindividual variation, PCR products
were cloned using the Invitrogen (Carlsbad, CA) TopoTA cloning kit; cloned products were amplified and sequenced using standard vector primers (M13 forward
and reverse, and T3 and T7, respectively). All sequences
631
were edited using the computer program Sequencher
(Genecodes, Madison, WI) and then aligned using the
software package ClustalX v1.8 (Thompson et al., 1997)
with the default gap opening and extension penalties.
Minor adjustments were made to the alignment to correct obvious mistakes. Gaps were treated as missing in all
likelihood analyses. Unambiguous insertion/deletions
(indels) were scored as binary characters (ITS data only;
Bayesian analyses).
Phylogenetic-Based Analyses
The computer program MrModelTest v2.1 (Nylander,
2004) was used to select an appropriate substitution
model, by Akaike information criterion (AIC), for each
data partition (12S, tRNA, 16S, and ITS). Using the model
of substitution indicated by AIC, analyses employing
Bayesian inference were conducted with MrBayes v3.1.2
(Ronquist and Huelsenbeck, 2003); the mtDNA and nuclear data sets were analyzed separately. Analyses consisted of two simultaneous runs each consisting of four
Markov chain Monte Carlo (MCMC) chains run initially
for 1,000,000 generations. Separate, simultaneous runs
were compared every 1000 to 5000 generations to check
for convergence (standard deviation of split frequencies
<0.01); runs were extended as necessary. Estimated parameters for each molecular partition were set to be independent, using unlink statefreq (all), revmat = (all),
shape = (all), pinvar = (all). MCMC runs were summarized and investigated for convergence of parameters, using the sump and sumt commands in MrBayes
and the computer program Tracer v3.1 (Rambaut and
Drummond, 2005). Trees prior to log likelihood stabilization (burn-in) and convergence were discarded; a majority rule consensus tree was produced using the contype =
allcompat command.
As a second measure of node support for major clade
groupings, we conducted a nonparametric bootstrap
analysis in PAUP* (Swofford, 2002) and Garli v0.951
(Zwickl, 2006) consisting of 100 replicates. Each run
comprised 10 random addition sequence replicates using TBR branch swapping; swapping iterations were
restricted to one million rearrangements. Gaps were
treated as missing; binary-scored indels were given a
weight of 2 (ITS data only).
The character trace facility in Mesquite v2.0 (Maddison and Maddison, 2007) was employed for ancestral character state reconstruction of spider habitat type
and to measure discord between gene trees and population subdivision (sof Slatkin and Maddison, 1989). Each
mitochondrial DNA haplotype was assigned either the
coastal dune or inland habitat character state designation. Character history on the Bayesian tree was then reconstructed using maximum-likelihood with a Markov
k-state 1-parameter model (Mk1, Lewis, 2001). Values for
swere computed for each major clade recovered from the
mtDNA Bayesian tree; we report s as the number of migration events required ≥ indicated by the tree structure
alone (s reported = s − (r −1); where r = the number of
sampling locations).
632
SYSTEMATIC BIOLOGY
A reconciled gene tree, based on the mitochondrial
and nuclear genealogies, was constructed in Mesquite.
The mitochondrial gene tree was pruned to include only
those taxa for which there was ITS data. Searches for the
reconciled gene tree, minimizing deep coalescences for
the two trees (Maddison, 1997), consisted of a heuristic
search using SPR branch swapping.
Niche-Based Distribution Modeling
Following Stockman and Bond (2007), ecological interchangeability is evaluated using two primary means
of assessment. First, locality coordinates for each specimen were imported into ArcMap (ESRI, Redlands, CA)
and converted into shape files. We used seven environmental layers thought to “likely influence the suitability
of the environment” (Phillips et al., 2006) for A. atomarius (see Stockman and Bond, 2007, for further justification of layer choice). Five climatic layers were obtained
from the WORLDCLIM data set (Hijmans et al., 2005):
annual precipitation, annual maximum temperature, annual minimum temperature, mean monthly temperature
range, and mean precipitation during the driest quarter.
A sixth layer, elevation, was constructed from a mosaic of
Digital Elevation Models (DEMs) derived from the National Elevation Dataset (USGS). DEMs were converted
to Raster format in ArcMap and resampled from 30-m
resolution to 1-km resolution using bilinear interpolation. The seventh layer, California level III bioregion data,
was obtained from the California Department of Forestry
and Fire Protection, Fire and Resource Assessment Program (CDF-FRAP) website and scaled to 1-km resolution using nearest-neighbor interpolation. All seven
layers were clipped to the same extent, cell size, and
projection.
Niche-based distribution models (DMs) were created
using the computer program Maxent (Phillips et al.,
2006). Maxent employs a maximum likelihood method
that estimates a species’ distribution that has maximum
entropy subject to the constraint that the environmental variables for the predicted distribution must match
the empirical average (Elith et al., 2006; Phillips et al.,
2006). Parameters for all Maxent analyses used the default values: convergence threshold = 10−5 , maximum
iterations = 500, regularization multiplier = 1, and auto
features selected. Additional larger values of the regularization multiplier were used to ensure that models
were not overfitting the data. Binary maps (predicted
presence or absence) were created from the Maxent generated DMs using the lowest presence threshold value
(LPT; see Pearson et al., 2006; Stockman and Bond, 2007).
The LPT is determined by selecting the lowest predicted
value for the set of presence points used to generate the
DM. We then assessed degree and significance of overlap
between DMs of closely related lineages using the procedure outlined in Stockman and Bond (2007) that employs
a Monte Carlo algorithm, implemented in the computer
program D-NOVL v1.3 (Stockman et al., 2007), to generate the probability distribution of overlap amounts expected for observed DMs.
VOL. 57
The second assessment of ecological interchangeability entails a principal components analysis of 19 climatic
variables (WorldClim data set; Hijmans et al., 2001) evaluated for each clade being compared, followed by a
MANOVA of the PC scores as outlined in Graham et al.
(2004b) and Stockman and Bond (2007). Clades were
treated as the fixed factor; PC scores were the dependent variables. A significant MANOVA score indicates
potential nonoverlap of ecological niche.
Methodological Framework to Delimiting Cohesion Species
As discussed by Wiens (2007), proportionally few papers address the development of explicit methodologies for delimiting species despite its importance as
a fundamental goal of systematics. We outline here a
methodological framework for delineating species. The
approach we employ is easily formulated as a set of
repeatable steps (outlined in Fig. 2). The initial criteria are modeled after Wiens and Penkrot (2002). A genealogy is used to evaluate geographical concordance
once exclusivity of the focal taxon has been established.
If haplotypes are not geographically structured, then
the possibilities of incomplete lineage sorting (if adaptive divergence is observed; sensu Masta and Maddison,
2002) or one widespread species with gene flow must be
considered. We establish a basal lineage starting point
for species evaluation through genealogical exclusivity
shared among lineages identified via a second nuclear
marker. This line can be established using alternative
procedures (e.g., Pons et al., 2006; Stockman and Bond,
2007; Wiens and Penkrot, 2002) in cases where divergence values do not preclude their application. This step
evaluates “hypothesis 1” of the cohesion species concept
(CSC)—Are the organisms sampled from a single evolutionary lineage?
The next step is to evaluate “hypothesis 2” of the
CSC—Are the lineages genetically exchangeable and/or
ecologically interchangeable? Exchangeability is assessed as a sister group comparison of the daughter
lineages (basal lineages) defined by the node directly
below the basal lineage designation. Each basal lineage sister pairing is first evaluated for the possibility
of gene flow between clades—genetic exchangeability.
Sampling must be sufficient to discriminate between
disjunct distributions due to nonoverlap versus insufficient sampling. When unsampled areas separate clades,
the strength of the phylogenetic break is reinforced by
genetic divergence that is uncorrelated with the geographic distance separating the clades. Second, basal
lineage pairings are evaluated for adaptive divergence
potential—ecological interchangeability. Adaptive divergence (AD) can be assessed by using standard approaches (e.g., morphological differentiation) or by other
less conventional means (e.g., overlap of DMs as already
discussed, see Graham et al., 2004a, 2004b, Wiens and
Graham, 2005; Swenson, 2005; Jakob et al., 2007; Stockman and Bond, 2007; Rissler and Apodaca, 2007). Our approach does not require that both criteria (nongenetic and
nonecological exchangeability) be met. However, when
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
633
FIGURE 2. General schematic for delimiting cohesion species. S1 to S3 refer to scenarios of geographic context. Reject = rejection of single
cohesion species (multiple species); accept = acceptance of lineages as a single cohesion species.
634
SYSTEMATIC BIOLOGY
populations are highly structured in the absence of major geographical or habitat breaks, considerable weight
is placed on evidence of AD. Alternatively, a deep phylogeographic break in the absence of AD requires concordance across other data sets to rule out sampling artifact.
Genetically structured population groups that lack significant geographical breaks and AD are treated as a single clade, or cohesion species. Alternatively, situations
involving a significant geographic/habitat break but no
AD are more ambiguous as such scenarios may be interpreted as speciation with niche conservation.
Basal lineages determined not to be genetically exchangeable or ecologically interchangeable are rejected
as cohesion species and are essentially designated as
emergent focal taxa (bottom of Fig. 2). The daughter lineages of the basal lineage nodes are evaluated in the
same manner as described above. Based on the degree of structuring and divergence in the nascent focal taxon, the newly established “basal lineage” may be
evaluated using finer scale approaches (e.g., coalescent
modeling/simulations if divergence values are shallow
enough) if applicable. Evaluation progresses through the
tree examining the daughters of each node towards the
tips until the null hypotheses of genetic exchangeability
and/or ecological interchangeability cannot be rejected
using the criteria outlined in Figure 2. Finally, it is important to note that character optimizations of putative
adaptive divergence may identify lineages towards the
tips of the tree that qualify as recent speciation events
via AD. Any such demarcation will render previous delineations in a focal taxon paraphyletic; otherwise, AD
would have been detected as a basal dichotomy in a previous step.
R ESULTS
Analyses of Mating Clasper and Spermathecae Shape
Figure 1b summarizes the results of the principal component analyses (color coded by geographic region). The
first three axes of the PCA of mating clasper dimensions describe 89.34% of the variation. However, there
appears to be no discrete groupings of populations based
on these features even when plotted along three axes.
The two individuals collected from populations along
the coast (in green) appear to be separated somewhat
from the remaining samples. It is, however, important
to note that these samples are each respectively separated along two different axes and thus do not technically fall out together as a distinct group. Results of the
PCA of spermathecae shape are similar. The first three
axes describe 95% of the variation, and no geographically distinct groups are observed along any of the three
axes. Based on qualitative and quantitative evaluations
of spider mating morphology we would conclude that
A. atomarius comprises a single, widely distributed morphological species.
Phylogenetic Analyses
Summary of genetic data.—Aligned matrices and trees
are accessioned TreeBase (S2114). We sampled 167 in-
VOL. 57
dividuals drawn from 75 sampling localities across
the known distribution of Aptostichus atomarius; 140
unique mtDNA haplotypes (GenBank accession numbers EU569898 to EU570037) were recovered from the
aligned data set comprising 1517 positions. Partial sequences (partial 16S) only were obtained for the Farallon
Island and Pinnacles sampling localities. Pairwise divergence values across the ingroup ranged from 0.007 to
0.239 (uncorrected P-values). With few exceptions (see
discussion below), haplotypes were restricted to single
localities, and multiple haplotypes sampled from the
same locality formed exclusive clades in all but one case.
Based on these results, we conclude that female gene flow
is severely limited, as has been documented for other
related mygalomorph groups (Bond et al., 2001, 2006,
Bond, 2004; Starrett and Hedin, 2007).
Internal transcribed spacer unit gene sequences (ITS1,
5.8S, and ITS2) for 22 individuals from 22 localities were
sampled (GenBank accession numbers EU569876 to
EU69897). A “stratigraphic” approach to subsampling
individuals for this data set was based on phylogenetic position within the mtDNA genealogy (see
haplotypes labeled in red, Fig. 3). We aimed to obtain
nuclear sequences for individuals corresponding to
basal, intermediate, and derived positions within each
mtDNA clade. All sequences recovered from the aligned
data set comprising 905 base pairs were unique; 12
unambiguous indels were scored as binary characters.
Pairwise divergence values ranged from 0.001 to 0.07
(uncorrected P-values).
Phylogenetic analyses.—The AIC analysis performed in
MrModelTest for the mitochondrial data indicated a general time-reversible model with a gamma distribution
and invariants model of rate heterogeneity (GTR++I)
for the 16S and tRNA partitions; GTR+ was indicated
for the 12S partition. Bayesian analysis of the data required 3 million generations to achieve convergence of
the two simultaneous runs. The results of the analysis, with outgroups removed, are summarized in Figure 3 (−ln = 31,256.83, harmonic mean evaluated from
both runs post-burn-in); branch lengths shown are averaged across the posterior distribution. Results of this
analysis have the following three key features relevant
to subsequent consideration for species delimitation:
(1) the focal taxon is genealogically exclusive with respect to the outgroup taxa (GenBank accession numbers
EU70038 to EU570050); (2) it exhibits extreme population structuring that is strongly geographically concordant across all phylogenetic levels (deep and shallow);
and (3) the five basal-most clades are strongly supported
(p > 0.95). Migration among lineages within each of the
major clades appears to be severely limited, with only a
few instances of geographically proximate populations
sharing haplotypes (clade 1, s = 1; clade 2, s = 1; clade
3, s = 0; clade 4, s = 1; clade 5, s = 0); no haplotypes
are shared among the five clades. The Bayesian analysis is largely congruent with likelihood and parsimony
results (not reported). Branch support for each of the
five major clades is also relatively high for these other
analyses.
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
635
FIGURE 3. Inferred mtDNA haplotype genealogy for all Aptostichus atomarius using Bayesian inference; to simplify, outgroups have been
removed. Posterior probabilities are listed at each internode; bootstrap values (likelihood followed by parsimony) are also listed for major
clades of interest. Haplotype designations reference localities listed in the online appendix according to county and location (alphanumeric
designations). Colored bars correspond to basal clades (see legend) and in Figure 1. Phylogenetic variation in abdominal color pattern is shown
by exemplars drawn from selected subclades; haplotypes corresponding to each abdominal pattern are listed on colored bars. Haplotypes in red
correspond to those sampled for the ITS data set. Blocked arrows indicate likelihood optimization of coastal dune habitat character state.
636
SYSTEMATIC BIOLOGY
VOL. 57
FIGURE 4. (a) Inferred ITS genealogy for a subsample of ingroup specimens. Tips are labeled by specimen number and corresponding mtDNA
haplotype designations. Thickened branches denote posterior probabilities >95%, values at nodes correspond to bootstrap values for parsimony
and likelihood analyses. (b) Reconciled mtDNA and ITS gene tree based on minimum coalescence optimality criterion.
For the nuclear ITS data, the AIC analysis indicated
a general time-reversible model with a gamma distribution and invariants model of rate heterogeneity
(SYM++I) for ITS1 and ITS2 and a Jukes-Cantor model
(JC; Jukes and Cantor, 1969) for the 5.8S partition. An
Mk+ model was used for the gap data partition. The
Bayesian analysis of these data comprised 10 million generations; burn-in was conservatively set at 7.5 million
generations. The results of the analysis are summarized
in Figure 4a (−ln = 2409.98, harmonic mean evaluated
from runs post-burn-in); branch lengths are averaged
across the posterior distribution. The ITS genealogy is
largely congruent with that based on the mtDNA data.
Five major groups are recovered in the unrooted tree
topology. With the exception of clade 2, all groups are
moderately to strongly supported in Bayesian, likelihood, and parsimony analyses. Clade 2 is paraphyletic
with respect to a couple of basal haplotypes; it is, however, monophyletic in a small percentage of the postburn-in trees. Figure 4b shows the reconciled species tree
minimizing deep coalescence for the mtDNA and ITS
gene trees. This tree is a strict consensus of 100 equally
optimal trees (score = 11,152,168 rearrangements examined) recovered from the analysis and show that all five
major clades are recovered but that clade 2 remains paraphyletic with respect to a single haplotype.
Maximum likelihood reconstruction of spider habitat
type (coastal dune versus inland) optimizes the coastal
character state as derived at two points on the tree (Fig. 3):
Clade 2 (proportional likelihood (pl) = 0.98) and a subset
of clade 5 (pl = 0.99). Clade 2 and the coastal subset of
clade 5 are considered nonecologically interchangeable
with their respective sister lineages as a consequence of
this habitat shift (see Discussion).
Niche-Based Distribution Modeling
Figure 5 summarizes the DMs constructed for clades 3
to 5. Only the inland (noncoastal) populations were used
to build predictive models for clade 5. As we will discuss
below, coastal dune populations are considered to be a
separate cohesion species. The clade 3 DM (Fig. 5a; LPT =
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
637
FIGURE 5. Predictive niche-based distribution models built using maximum entropy (implemented in the computer program Maxent). Grey
shaded areas denote areas of predicted occurrence. (a) Clade 3, dashed line roughly approximates the location of the Salinas Valley; (b) clade 4C;
(c) clade 4B; (d) clade 5.
40%) is based on nine presence observations and shows
the areas of predicted occurrence to be along the eastern slopes of the northern Santa Lucia ranges and mid
elevations of the Sierra Salinas, with some additional areas along the western slopes of the Coastal Range to the
south. The disjunct haplotype from Pinnacles (SBT01; see
Fig. 6c) was omitted from the analysis as an outlier. The
Salinas River Valley (Fig. 5a, dashed line) is consistently
an area of low probability in all of the predicted distributions and as such likely represents a historical barrier
to gene flow (see Discussion). Two separate DMs were
generated for the basal constituent lineages of clade 4,
as models produced from these clades combined were
prone to gross overprediction, likely due to the widely
varying habitat across this clade’s known distribution.
The predicted distribution of clade 4B (Fig. 5c; LPT =
50%) is restricted to the mostly xeric western edge of the
Central Valley; populations along this once more mesic
transect (see Harden, 2004) are rare and likely relictual
(Stockman and Bond, 2007). Conversely, predicted distribution of clade 4C (Fig. 5b; LPT = 34%) is found in
the more mesic coastal habitats, confined mainly to the
Gabilan and Diablo ranges just south of the San Francisco Bay. Areas of high probability for the inland occurrence of clade 5 (Fig. 5d, based on 14 data points; LPT =
40%) overlap almost the entire predicted range of clade
638
SYSTEMATIC BIOLOGY
VOL. 57
FIGURE 6. Detailed maps showing direct correspondence between haplotypes and geographic sampling localities. Phylogenetic results are
based on Bayesian inference and correspond directly to the overall phylogenetic picture shown in Figure 3. Posterior probabilities are given
at each node when >94%. (a) Clade 1; S1 and S2 refer to haplotypes sampled from more than one locality. (b) Clade 2; (c) clade 3; (d) clade 4;
(e) clade 5, inland sampling localities; (f) clade 5, coastal dune restricted localities.
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
4C (90.9%) and extend across into the Santa Cruz Mountains west of San Francisco Bay. Based on the analysis
in D-NOVL, the probability of observing this degree of
overlap among randomly distributed ranges is <0.001.
The MANOVA of the first three PC scores of clades 4
and 5 is nonsignificant (F (3, 28) = 2.71, P > 0.05), suggesting, like the overlapping DMs, that these clades are
ecologically interchangeable. The MANOVA comparing
the PC scores of clade 3 and clade 4+5 is significant (F (3,
37) = 4.26, P < 0.05), indicating noninterchangeability.
D ISCUSSION
Figure 6 shows the direct correspondence between the
mtDNA haplotype trees and geographic localities. Although the Aptostichus atomarius species complex lacks
morphological variation in male secondary sexual characteristics and in general female somatic morphology
(Fig. 1), it is highly structured on a fine geographic scale
(Fig. 6). Commonly employed, morphology-based approaches to species delimitation for trapdoor spiders
would fail to recognize the deep evolutionary history intrinsic to this group. Conversely, congruence among both
mitochondrial and nuclear loci recognize a number of divergent basal lineages. However, this profound withinlineage population structuring on a microallopatric scale
creates a certain degree of difficultly when attempting to
objectively determine a cutoff point for delineating lineages for speciation or conservation status using other
methods (e.g., Wiens and Penkrot, 2002). Haplotypes
within some populations fail to unite at the 95% confidence level using TCS; consequently, this methodology
clearly over splits the group into an unreasonable number of species-level lineages (∼60). If a criterion of geographical concordance is employed, more than 20 species
would be recognized within the A. atomarius complex.
Although this may be a rather simplified geographic
concordance delineation, most of these lineages are geographically disjunct (Fig. 6) and could be empirically
distinguished on the basis of unique nucleotide substitutions; that is, they are diagnosable at the molecular
level.
Topology-based evaluation of cohesion species criteria.—
Aptostichus atomarius is genealogically exclusive and
structured geographically; the dashed gray line in Figure 7 depicts the five basal mtDNA lineages designated
639
as prospective species. Table 1 summarizes species delineation inferences for each clade (see Figs. 3 and 6).
Evaluations of genetic exchangeability (GE ) and ecological interchangeability (EI ) are first considered one node
directly below the basal lineage designation (nodes A
to C; Fig. 7). At node A, clades 1 and 2 are geographically disjunct (Fig. 1) such that GE is unlikely given their
separation; that is, total nonoverlap spatially (clade 2
is restricted to the coast; S1, Fig. 2). Because clade 2 is
found only in coastal dune habitat, it has significant AD,
and thus these lineages are also not ecologically interchangeable. Trapdoor spiders inhabiting coastal dunes
construct deeper burrows with heavier silk linings (Bond
personal observation), cope with a continually shifting
sand substrate and salt-laden winds, encounter different
prey items, and experience relatively mild, regulated climatic conditions (Schoenherr, 1992; Ornduff et al., 2003).
Also, these psammophilic forms differ phenotypically:
their abdominal coloration and striping is much lighter
than that of the inland forms (Fig. 3).
The ecological distinction that can be made at node B,
which forms the dichotomy established by clade 3 and
clade 4+5, is not as straightforwardly resolved. These
clades form a parapatric grade of populations/lineages
extending from the coastal ranges inland and northward
(Figs. 1, 6c to f). The potential for GE would be difficult to dismiss in the absence of exhaustive geographic
sampling. However, the geographic range of clade 3 appears to be discordant from that of clade 4+5 as a consequence of the Salinas Valley, an ancient biogeographic
break (Jacobs et al., 2004) that has been documented for
several unrelated taxa (e.g., Shaffer et al., 2004; Matocq,
2002; Rissler et al. 2006; Starrett and Hedin, 2007). Although the DM for clade 5 overlaps minimally with that
of clade 3 (Fig. 5), the relatively low-lying area of the
Salinas Valley is likely unsuitable given the low values
of our predictive models for this region and failure to
obtain any specimen from this area. Whereas evidence
for non-EI is questionable, the location of this phylogenetic break along an established geographic landmark,
coupled with habitat unsuitability, supports delimitation
at this point in the genealogy (Fig 2, S2). Conversely,
as focus is shifted up to the sister pairing at node C,
we “collapse” clades 4 and 5 down to node C because
they overlap significantly in their predicted distributions
(Fig. 4b to d) and are not associated with any documented
FIGURE 7. Simplified tree topology redrawn from Figure 3, inferred directly from the Aptostichus atomarius mtDNA data set. Dashed gray
line is the basal lineage designation. Curved arrows correspond to tests for genetic exchangeability and/or adaptive divergence for the daughter
lineages at these nodes.
640
VOL. 57
SYSTEMATIC BIOLOGY
TABLE 1. “Species delimitation” results. The first two columns reference nodes and clades depicted in Figure 7; GE and E I refer to genetic
exchangeability and ecological interchangeability, respectively. Subclades are identified in FIGURE 6.
Node
A
B
C
E
F
G
I
Clades
GE
EI
Inference
1
2
3
4+5
4
5
1A
1B/C
2A
2B/C
3A
3B/C
Subclade 4+5
No
No
No
No
Noncohesive clade: geographical discordance with intervening samples; coastal clade is adaptively
diverged
Noncohesive clade: geographical discordance (Salinas Valley)
Yes
Yes
Cohesive clade: geographic concordance (parapatry); overlapping predicted distribution
Yes
Yes
Inconclusive outcome—sampling insufficient
Yes
No
Cohesive clade: geographic discordance; no adaptive divergence
No?
No?
No
No
Inconclusive outcome: geographic sampling inadequate but Pinnacles Haplotype, 3A, isolated and
outside area of prediction for clade 3B/C
Cohesive clade: geographical discordance; coastal clade, adaptively diverged—renders 4+5
paraphyletic
phylogeographic breaks (Fig. 2, S3). Although we cannot
entirely rule out the possibility that the phylogenetic discontinuity at this node is not a sampling artifact (sensu
Irwin, 2002; Kuo and Avise, 2005), this argument is weakened somewhat by the exclusivity of this clade in the ITS
genealogy (Fig. 4).
Thus far, of the five major clades, clades 1 to 3 were
retained above as nonexchangeable lineages (cohesion
species). As such, these lineages should now be evaluated for further delimitation (i.e., do these lineage contain
>1 species). This evaluation should proceed as before
such that the basal dichotomy within each clade is examined for GE and EI . Essentially nodes E, F, and G are reset as emergent focal taxa. Clade 1 exemplifies sampling
issues often encountered for highly structured populations. First, the two nodes at the basal-most dichotomy
(node E) lack strong support (p < 0.95; Figs. 3, 7). Second, geographical sampling is sparse along the southernmost extension of the Coast Ranges (Fig. 6a). Subclades
within the weakly supported basal lineage (Fig. 6a C1A)
are highly geographically discordant, with one subclade
in the Mojave Desert, the other near the coast, and both
separated by the next clade up in the genealogy. Also, individual haplotypes attributed to the Mojave Desert localities are phenotypically distinguishable from all other
individuals contained within the clade (i.e., they are
much lighter in color; see Fig. 3). In addition, clade C1B
contains the only examples of haplotypes shared among
localities (labeled as S1 and S2 in Fig. 6a). Although some
of the lineages embedded within clade 1 would likely fail
tests of GE and EI given additional sampling, and thus
represent distinct species, the existing data necessitate
their retention as a single species.
Due to complete sampling along the central California coast, clade 2 at node F (Fig. 7) is one of the few
instances in which GE can be evaluated using a more
fine-scale approach like NCA. As noted previously, clade
2 is restricted to the coastal dunes along the beaches
of Monterey, Santa Cruz, and San Francisco counties
(Fig. 6b). The basal-most haplotype MTY01 (Fig. 6b, subclade C2A) is isolated to the south and is genetically
separated from all other haplotypes by over 150 base
pairs. Although we have not been able to sample the
dunes at Point Sur, just 6.44 km to the north, it seems
highly unlikely that this branch would be significantly
broken up due to recovery of haplotypes from this locality. Although the NCA results are not shown here (for
brevity—see online supplementary material available at
www.systematicbiology.org), significant associations between haplotypes and geography as a consequence of
fragmentation occur at a number of hierarchical levels
within clade 2. For example, despite minimal geographic
separation with respect to total distance, subclades C2B
and C2C (Fig. 6b) appear to be fragmented across Año
Nuevo Point. Given the close geographical proximity of
these clades, the absence of both corroborative phylogeographic barriers and notable adaptive divergence (Fig. 2,
S3), we consider this to be an example of structuring due
to artifact and retain the species boundary at the clade 2
level (but see caveats regarding this decision below).
At node G (Fig. 7), clade 3, like clade 1, is an area
where additional sampling may be required. The
basal-most subclade (C3A; Fig. 6c) is represented by a
single locality, Pinnacles, situated at the southern end
of the Gabilan Range, across the Salinas Valley from the
remaining members of clade 3. As previously discussed,
the Salinas Valley represents a known geographic barrier. In consideration of this, plus the degree of genetic
divergence and geographic disjunction, we judge this
subclade to be non-GE with the others. It is also likely
to be non-EI , as indicated by the very low probability
of suitable habitat at Pinnacles according to the Maxent
model (Fig. 5a) despite the inclusion of that locality
in training earlier versions of the model (not shown).
However, due to the limited sampling, we consider an
interpretation of non-EI to be premature and retain all
subclades as a single unit. Alternatively, this locality
may represent a relictual population (see Stockman
and Bond, 2007). The clades defined by nodes H and D
(Fig. 7) are not evaluated at this point as they (clades 4
and 5) were collapsed due to apparent GE and EI at the
previous basal lineage demarcation.
The most derived subclades of clade 5 (Fig. 6f, node
I in Fig. 7) are unequivocally adaptively divergent and
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
form exclusive groupings in both the mtDNA and ITS
genealogies. Based on the character reconstruction of
spider habitat, this lineage is restricted to the northern
coastal dunes, and it exhibits the same distinctive characteristics as the clade 2 coastal specimens (e.g., lighter
coloration, deeper burrows, heavier silk). Therefore, we
consider this clade to be both non-GE and non-EI with
the other subclades and retain it for consideration for
species delineation despite the fact that it renders clade
4+5 paraphyletic (see discussion of this issue below).
Summary of species delimitation.—Based on the results
reported herein, we hypothesize that the Aptostichus
atomarius complex comprises minimally five cohesion
species (diagnosed and described in Appendix); that is,
groups of populations that meet the criteria of evolutionary lineage, genetic exchangeability, and ecological
interchangeability. Clades delimited as species are those
lineages defined by nodes E, F, G, C–I, and I (Fig. 7).
Clade 1 (as defined at node E) is restricted to southern
California’s Transverse and Peninsular Ranges, the Los
Angeles Basin, and the Mojave Desert (Fig. 6a). This clade
as presently constructed contains haplotypes sampled
from the putative type locality and thus would retain the
specific epithet, atomarius; however, given the extreme
habitat variability and phenotypic differences already
noted, bolstered sampling will likely result in the discovery of additional cohesion species, most notably a species
adaptively diverged for desert conditions. Clade 2 (Fig. 7,
node F), Aptostichus stephencolberti sp. nov., is restricted
to the coastal dunes that extend from the Big Sur area to
the San Francisco peninsula at Point Lobos and Golden
Gate (Fig. 6b). Members of this clade are considered to be
non-EI with populations of its sister clade (clade 1) due to
its specificity to coastal dunes and divergent phenotype
(Fig. 3). These populations are spatially separated from
all other “atomarius” complex members. Clade 3 (Fig. 7,
node G), Aptostichus angelinajolieae sp. nov., is geographically restricted to the Coastal Range south of the San
Francisco Bay (Fig. 6c). Our evaluation of EI based on
niche-based DM shows there to be little overlap in the
“selective regime” of clade 3 and other closely related
lineages. The Salinas Valley is a hypothesized barrier to
GE . Clade 4+5 (Fig. 7, defined at node C sans subclade
defined at node I), Aptostichus stanfordianus, has a distribution that is essentially east of the San Andreas Fault
(Fig. 6d, e). It includes San Francisco Peninsula populations and populations that extend northward throughout the Diablo, Gabilan, and northern coast ranges. The
subclade defined at node I, Aptostichus miwok sp. nov.,
comprises the northern coastal dune endemic populations whose distribution extends from the Golden Gate
at Point Bonita northward (Fig. 6f).
Assessment and Self-Critique
The approach that we have taken to delineating species
among these highly divergent lineages is not without issues and certain limitations and consequently requires
critical examination. We discuss below these limitations,
including use of mtDNA as a primary genetic marker, the
641
perils and pitfalls of evaluating ecological interchangeability, and the recognition of paraphyletic lineages as
species.
mtDNA as a population marker.—Although widely applied (Moore, 1995; Avise, 2000), there are a number of
reasons why mtDNA may be inadequate. Well documented are the potential problems of gene-tree/speciestree incongruence, introgression (McGuire et al., 2007),
intracellular symbionts, and selection (see Ballard and
Whitlock, 2004; Ballard and Rand, 2005; Hurst and
Jiggins, 2005). Moreover, exclusivity among mtDNA
haplotypes generally far outpaces time to exclusivity in
nuclear alleles (Hudson and Coyne, 2002). Despite these
criticisms, insights gained through mtDNA markers over
the past 25 years have been immense (Rubinoff and
Holland, 2005) and recent comparative phylogeographic
studies show obvious and compelling consistencies in
mtDNA genealogies among disparate taxa (see Lapointe
and Rissler, 2005; Rissler et al., 2006; Feldman and Spicer,
2006, for California phylogeography). The congruence
we observe between the nuclear (ITS) and mtDNA data
likewise supports the suitability of these data in highly
structured taxa and suggests that mtDNA alone may
be adequate for establishing a topological framework.
Still, we advocate an integrative approach to delineate
species, one that places strong consideration towards
adaptive divergence (sensu Crandall et al., 2000). For
example, although the genetics underlying adaptations
related to inhabiting a coastal dune environment remain
untested, the invariance of these features for coastal lineages suggests a potential nuclear genetic basis that is
consistent with mtDNA genealogical exclusivity. Furthermore, otherwise seemingly cryptic (i.e., morphologically) divergent mitochondrial lineages have been shown
to remain cohesive in zones of secondary contact (Bond
and Sierwald, 2002). All of these issues taken together,
coupled with the general successes realized through the
application of mtDNA-based phylogenetics (Rubinoff
and Holland, 2005) and demonstrated congruence with
nuclear markers here and in other spider studies (e.g.,
Hedin, 2001; Hendrixson and Bond, 2005; Starrett and
Hedin, 2007), warrant continued but cautious use of
these molecular data.
Evaluating ecological interchangeability.—The evaluation of ecological interchangeability is nontrivial. First,
our assessments of ecological interchangeability with
regard to the coastal lineages are qualitative judgments in need of further testing. The biotic and abiotic conditions for these psammophilic populations are
observably nonoverlapping with all inland lineages of
A. atomarius. Moreover, coastal populations are phenotypically distinct (see Fig. 3). Abdominal coloration is
likely an inherited trait that potentially enhances the survivability of mature wandering males by making them
more cryptic on a light, sandy surface (Hoekstra et al.,
2006). Females are likewise much lighter in color, but they
seldom leave their burrows; thus, selection on female
coloration is unlikely. However, dune environments are
relatively unstable and long-lived females may be displaced occasionally, providing instances whereby female
642
SYSTEMATIC BIOLOGY
crypsis would be advantageous. It appears that dune
endemic Aptostichus build deeper burrows perhaps in
response to the less stable dune environment. Future assessments of burrow architecture could provide key components to understanding adaptive divergence among
Aptostichus atomarius clades. One could argue that because these spiders are fossorial, they are buffered from
the environment and thus the impact of climate with respect to “selective regime” is minimized. However, Aptostichus burrows, in general, are very shallow (only a
few centimeters) and mature males leave their burrows
to wander. Consequently, male maturation times vary
tremendously across populations and species in apparent response to the onset of the rainy seasons and other
climatic factors (Bond, 1999).
Second, the evaluation of ecological interchangeability using niche-based distribution modeling requires
close scrutiny despite its obvious appeal and already
demonstrated utility in evaluating speciation pattern
and process (e.g., Graham et al., 2004a; Stockman and
Bond, 2007; Jakob et al., 2007; McGuire et al., 2007;
Rissler and Apodaca, 2007). Niche-based distribution
models are generally viewed as a prediction of the organism’s realized niche because occurrence localities
are drawn exclusively from the source habitat (Phillips
et al., 2006). The realized niche is thus a subset of
the fundamental niche, which defines the environmental conditions under which the organism can survive
in the absence of other biotic interactions or historical constraints. Because we can only model the realized niche, the possibility exists that lineages with
nonoverlapping predicted distributions could ostensibly
occupy the same fundamental niche; that is, nonexclusive subsets of the realized niche appear in the prediction. We therefore caution that nonoverlap of predicted
distribution warrants further testing (Rader et al., 2005,
see discussion below). Alternatively, although overlapping predicted distributions may support the null hypothesis of ecological interchangeability, they do not
necessarily preclude noninterchangeability. Overprediction indicates that a species’ environmental tolerances
are more extensive than observed and that other factors (e.g., biotic interactions) may be limiting geographic
range (Wiens and Graham, 2005). We envision a scenario in which the competitive interactions of two sister species whose distributions have been modeled are
precluding sympatry. Moreover, overlap of a predicted
distribution could suggest evidence of niche conservation among sister species or lineages—speciation occurring in the absence of ecological divergence (see Bond
et al., 2001). Sister species pairs have been demonstrated,
through niche modeling approaches (Peterson et al.,
1999), to occupy similar climatic niches. Such conservation would play a significant role in geographic speciation when climate across a barrier was unsuitable,
limiting gene flow between isolated populations (Wiens
and Graham, 2005). Niche-based distribution modeling
provides a robust evaluative tool for delimiting species
boundaries but one warranting careful consideration
and interpretation.
VOL. 57
What are the next steps regarding formalized tests of
ecological interchangeability? And, to what extent is further testing needed for seemingly obvious ecological disparities (e.g., coastal versus inland A. atomarius)? Rader
et al. (2005:239) define populations as ecologically interchangeable “when individuals can be moved between
different local populations and still occupy the same
ecological niche or selective regime”; that is, if distinctive populations are exposed to differing environmental conditions to which they become locally adapted,
they are no longer ecologically interchangeable. If genetically nonequivalent populations differ in “phenotypic”
expression of a trait(s), then they are considered to be
noninterchangeable. The ideal of reciprocal transplant
and common environment experimentation is without
question unfeasible for certain many organismal groups.
Thus, further work is needed to design tests of ecological
interchangeability that can be more easily implemented
and are feasible for the rapid biodiversity assessments required in some conservation situations. The life history
characteristics (e.g., long maturation times, >4 years)
make ecological experiments like those of Rader et al.
(2005) untenable for organisms like trapdoor spiders.
Moreover, reciprocal transplant experiments are potentially irresponsible and environmentally unethical under many circumstances. Alternatively, environmental
chamber-based experiments could be designed for A.
atomarius that involve experiments that test the ability of
non-dune-adapted spiders to build functional burrows
in sand and vice versa.
Is evaluation of ecological interchangeability necessary?—
Given the close correlation between ecological
divergence and reproductive isolation (Funk et al.,
2006—discussed by Rissler and Apodaca, 2007) and
consequently the putative role of ecological differentiation in speciation process, an evaluation of ecological
interchangeability should play some significant role in
species delimitation. At issue is whether genetic divergence in the absence of ecological change or adaptive
divergence of some other type is sufficient for species
recognition. We would argue that minimally some
variation or disparity in present day selective regime
must be demonstrated among morphologically identical
lineages, particularly when they are parapatric (i.e.,
future gene flow is plausible). This perspective must be
balanced with the potential for what might be characterized as non-adaptive radiation or niche conservatism
among sister lineages. However, highly structured
systems by definition are predisposed to allopatric
fragmentation and have the relatively unique characteristic of achieving exclusivity prior to the evolution of
isolation or cohesion mechanisms (see Harrison, 1998).
The requirement of adaptive divergence, or capacity for
adaptive divergence through the occupation of different
selective regimes, echoes Crandall et al. (2000), who
argued that recognition of evolutionary significant units
(ESUs) for the purposes of conservation required more
than measures of “genetic isolation.” Adaptive divergence in the form of ecological or morphological change,
in concert with genetic isolation, better addresses
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
the goal of conserving the evolutionary potential of
populations. Indeed, a review of the current speciation
literature reveals that many studies assessing species
boundaries are multidisciplinary (Johnson et al., 2004;
Daniels et al., 2005; Yoder et al., 2005; Sanders et al., 2006;
Schlick-Steiner et al., 2006; Roe and Sperling, 2007) and
likely meet the adaptive divergence criterion set forth
here and by Crandall et al. (2000). Far fewer studies seem
to advocate the recognition of lineages solely on genetic
divergence, exclusivity, or geographic concordance (e.g.,
Bond et al., 2001; Kronauer et al., 2005).
Nonexclusivity of species.—Species-level paraphyly (reviewed by Funk and Omland, 2003), is more prevalent among species groups than previously thought. Our
recognition of lineage I (Figs. 6f, 7), the northern coastal
dune endemic, as a species, renders clade 4+5 paraphyletic. We prefer nonexclusivity in favor of ignoring
the adaptive divergence shared by all members of the
coastal lineage. Consequently, we also elected not to elevate the remaining lineages within clade 5 to species to
correct for nonexclusivity (what Templeton, 1998, characterized as “speciation by remote control”). Such hierarchical recalibration would mask the budding nature of
this incipient speciation event (i.e., peripatric) and would
be inconsistent with the criterion of adaptive divergence
we enforced on other parts of the tree. The northern
coastal populations occur at a much shallower level of divergence and are in fact the most derived lineage on the
tree, yet this lineage is phenotypically distinguishable
from the remaining members of clade 5. This suggests
that adaptive divergence has been rapid and recent and
thus nonexclusivity is a likely consequence of incomplete
lineage sorting rather than issues related to phylogenetic
signal or imperfect taxonomy (Hendrixson and Bond,
2005). The recency of this speciation event is consistent
with north/south breaks observed in other Pacific coast
endemic taxa (Edmands, 2001) and is the likely consequence of post-Pleistocene range expansion to the north
(Burns and Barhoum, 2006).
S UMMARY AND FUTURE D IRECTIONS
We present in this paper an approach to species delineation for highly geographically structured species that
seeks to abrogate vexing problems encountered when
testing species boundaries in such groups. Many nonvagile organisms have populations that are genetically
unique, deeply divergent, and often morphologically homogenous. As such, commonly employed methodologies and conceptual approaches to species delineation
are either rendered inapplicable by the data or run
the risk of fractionating lineages into numerous species
when applied haphazardly. Once a set of basal lineages
are identified, we then use an iterative process to examine them, starting at a predesignated basal point moving upward through the phylogeny, examining lineages
for genetic exchangeability and ecological interchangeability. A final examination towards the tips of the phylogeny ensures that instances of incipient speciation are
not overlooked due to gene-tree/species-tree incongru-
643
ence. We evaluate genetic exchangeability through assessments of geographical concordance and evidence of
allopatric fragmentation. Ecological interchangeability is
evaluated using niche-based distribution modeling and
general assessment of habitat differences using character
state reconstructions. For divergent lineages with parapatric geographic distributions, but appearing to remain
ecologically interchangeable, we elect to retain these lineages as cohesion species (sensu Crandall et al., 2000).
The approach we outline is applied to the Aptostichus
atomarius species complex. This group is widely distributed (Figs. 1, 6) across diverse habitats, a situation
that predisposes populations, lineages, to become locally adapted (Calsbeek et al., 2003). And, parallel origins
of psammophilic phenotypes underscore the adaptive
nature of the species radiation within this group of spiders at both deep and shallow phylogenetic levels. Phylogenetic evaluation of the Aptostichus atomarius complex
identifies five lineages that should be considered cohesion species.
Although our work on the Aptostichus atomarius
species complex remains unfinished, the approach taken
here provides a framework that will direct future questions and areas of research emphasis in this group of
spiders. Increased sampling across the Mojave and the
Transverse Ranges will likely resolve two additional
species. Second, the coastal dune endemic clades provide an opportunity to study the parallel evolution of
coastal dune adaptation and the evolution of ecological
interchangeability within a sister group context (as discussed earlier with regards to the ideas of Rader et al.,
2005). Finally, the complex geology, climatic, and biotic
environment of the California Floristic Province provides
a remarkable backdrop for which to consider the historical biogeography of this widespread species complex
within a comparative phylogeographic framework (e.g.,
Calsbeek et al., 2004; Lapointe and Rissler, 2005; Rissler
et al., 2006, Feldman and Spicer, 2006).
ACKNOWLEDGMENTS
This research was supported by National Science Foundation grant
DEB 0315160 (REVSYS). The manuscript was greatly improved through
the comments provided by Trip Lamb, Marshal Hedin, and Jack Sites.
Stacey Smith, Jennifer Roberts, Michael Brewer, and Brent Hendrixson
assisted in collection of DNA sequence data; Paul Marek and Marshal
Hedin assisted with collection of specimens. Collecting permits were
provided to us by the California State Parks System (Arthur Fong).
R EFERENCES
Agapow, P. M., O. R. P. Bininda-Edmonds, K. A. Crandall, J. L. Gittleman, G. M. Mace, J. C. Marshall, and A. Purvis. 2004. The impact of
species concept on biodiversity studies. Q. Rev. Biol. 79:161–179.
Arnedo, M. A., and M. A. Ferrandez. 2007. Mitochondrial markers
reveal deep population suibdivision in the European protectred spider Macrothele calpeiana (Walkernaer, 1805) (Araneae, Hexathelidae).
Con. Gen. 8:1147–1162.
Avise, J. C. 2000. Phylogeography. The history and formation of species.
Harvard University Press, Cambridge, Massachusetts.
Ballard, J. W. O., and D. M. Rand. 2005. The population biology of
mitochondrial DNA and its phylogenetic implications. Annu. Rev.
Ecol. Evol. Syst. 36:621–642.
Ballard, J. W. O., and M. C. Whitlock. 2004. The incomplete natural
history of mitochondria. Mol. Ecol. 13:729–744.
644
SYSTEMATIC BIOLOGY
Bickford, D., D. J. Lohman, N. S. Sodhi, P. K. L. Ng, R. Meier, K. Winker,
K. K. Ingran, and I. Das. 2006. Cryptic species as a window on diversity and conservation. TREE 22:148–155.
Bond, J. E. 1999. Systematics and evolution of the Californian trapdoor
spider genus Aptostichus Simon (Araneae: Mygalomorphae, Euctenizine). PhD thesis, Virginia Polytechnic Institute and State University, Blacksburg, Virginia.
Bond, J. E. 2004. Systematics of the Californian euctenizine spider genus
Apomastus (Araneae: Mygalomorphae: Cyrtaucheniidae): The relationship between molecular and morphological taxonomy. Invertebr.
Syst. 18:361–376.
Bond, J. E., D. A. Beamer, T. Lamb, and M. Hedin. 2006. Combining genetic and geospatial analyses to infer population extinction in Mygalomorph spiders endemic to the Los Angeles region. Anim. Conserv.
9:145–157.
Bond, J. E., and M. Hedin. 2006. A total evidence assessment of the phylogeny of North American euctenizine trapdoor spiders (Araneae,
Mygalomorphae, Cyrtaucheniidae) using Bayesian inference. Mol.
Phylogenet. Evol. 41:70–85.
Bond, J. E., M. C. Hedin, M. G. Ramirez, and B. D. Opell. 2001. Deep
molecular divergence in the absence of morphological and ecological change in the Californian coastal dune endemic trapdoor spider
Aptostichus simus. Mol. Ecol. 10:899–910.
Bond, J. E., and B. D. Opell. 2002. Phylogeny and taxonomy of the genera of south-western North American Euctenizinae trapdoor spiders
and their relatives (Araneae, Mygalomorphae, Cyrtaucheniidae).
Zool. J. Linn. Soc. 136:487–534.
Bond, J. E., and P. Sierwald. 2002. Cryptic speciation in the Anadenobolus
excisus species complex on the island of Jamaica. Evolution 56:1123–
1135.
Boyer, S. L., J. M. Baker, and G. Giribet. 2007. Deep genetic divergences in Aoraki denticulate (Arachnida, Opiliones, Cyphophthalmi):
A widespread “mite harvestman” defies DNA taxonomy. Mol. Ecol.
16:4999–5016.
Burns, K. J., and D. N. Barhoum. 2006. Population-level history of the
wrentit (Chamaea fasciata): Implications for comparative phylogeography in the California Floristic Province. Mol. Phylogenet. Evol.
38:117–129.
Calsbeek R. C., J. N. Thompson, and J. E. Richardson. 2003. Patterns
of molecular evolution and diversification in a biodiversity hotspot:
The California Floristic Province. Mol. Ecol. 12:1021–1029.
Crandall, K. A., O. R. P. Bininda-Emonds, G. M. Mace, and R. K. Wayne.
2000. Considering evolutionary processes in conversation biology.
TREE 15:290–295.
Daniels, S. R., N. J. L. Heideman, M. G. J. Hendricks, M. E. Mokone, and
K. A. Crandall. 2005. Unraveling evolutionary lineages in the limbless fossorial skink genus Acontias (Sauria: Scincidae): Are subspecies
equivalent systematic units? Mol. Phylogenet. Evol. 34:645–654.
De Queiroz, K. 1999. The general lineage concept of species and the
defining properties of the species category. Pages 49–90 in Species,
new interdisciplinary essays (R. A. Wilson, ed.). MIT Press, Cambridge, Massachusetts.
Edmands, S. 2001. Phylogeography of the intertidal copepod Tigriopus
californicus reveals substantially reduced population differentiation
at northern latitudes. Mol. Ecol. 10:1743–1750.
Elith, J., C. H. Graham, R. P. Anderson, M. Dudik, S. Ferrier, A. Guisan,
R. J. Hijmans, F. Huettmann, J. R. Leathwick, A. Lehmann, J. Li, L.
G. Lohmann, B. A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y.
Nakazawa, J. M. Overton, A. T. Peterson, S. J. Phillips, K. Richardson,
R. Scachetti-Pereira, R. E. Schapire, J. Soberon, S. Williams, M. S. Wisz,
and N. E. Zimmermann. 2006. Novel methods improve prediction of
species’ distributions from occurrence data. Ecography 29:129–151.
Feldman, C. R, and G. S. Spicer. 2006. Comparative phylogeography
in a community of reptiles: Repeated patterns of cladogenesis and
population expansion in California. Mol. Ecol. 15:2201–2222.
Funk, D. J., P. Nosil, and W. J. Etges. 2006. Ecological divergence exhibits
consistently positive associations with reproductive isolation across
disparate taxa. Proc. Nat. Acad. Sci. USA 103:3209–3213.
Funk, D. J., and K. E. Omland. 2003. Species-level paraphyly and polyphyly: Frequency, causes, and consequences, with insights from animal mitochondrial DNA. Ann. Rev. Ecol. Evol. Syst. 34:397–423.
Graham, C. H., S. Ferrier, F. Huettman, C. Moritz, and A. T. Peterson. 2004a. New developments in museum-based informatics
VOL.
57
and applications in biodiversity analysis. Trends Ecol. Evol. 19:497–
503.
Graham, C. H., S. R. Ron, J. C. Santos, C. J. Schneider, and C. Moritz.
2004b. Integrating phylogenetics and environmental niche models
to explore speciation mechanisms in dendrobatid frogs. Evolution
58:1781–1793.
Harden, D. R. 2004. California geology. Pearson Prentice Hall, Upper
Saddle River, New Jersey.
Harrison, R. G. 1998. Linking evolutionary pattern and process: The
relevance of species concepts for the study of speciation. Pages 19–
31 in Endless forms: Species and speciation (D. J. Howard and S. H.
Berlocher, eds.). Oxford University Press, New York, New York.
Hedin, M. 2001. Molecular insights into species phylogeny, biogeography, and morphological stasis in the ancient spider genus Hypochilus
(Araneae: Hypochilidae). Mol. Phylogenet. Evol. 18:238–251.
Hedin, M., and D. A. Wood. 2002. Genealogical exclusivity in geographically proximate populations of Hypochilus thorelli Marx (Araneae,
Hypochilidae) on the Cumberland Plateau of North America. Mol.
Ecol. 11:1975–1988.
Hendrixson, B. E., and J. E. Bond. 2005. Testing species boundaries in
the Antrodiaetus unicolor complex (Araneae: Mygalomorphae: Antrodiaetidae): “Paraphyly” and cryptic diversity. Mol. Phylogenet. Evol.
36:405–416.
Hickerson, M. J., C.P. Meyer, and C. Moritz. 2006. DNA barcoding will
often fail to discover new animal species over broad parameter space.
Syst. Biol. 55:729–739.
Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis.
2005. Very high resolution interpolated climate surfaces for global
land areas. Int. J. Climatol. 25:1965–1978.
Hoekstra, H. E., R. J. Hirschmann, R. J. Bundey, P. Insel, and J. P. Crossland. 2006. A single amino acid mutation contributes to adaptive
color pattern in beach mice. Science 313:101–104.
Huber, B. A. 2003. Rapid evolution and species-specificity of arthropod
genitalia: Fact or artifact? Org. Divers. Evol. 3:63–71.
Hudson, R. R., and J. A. Coyne. 2002. Mathematical consequences of
the genealogical species concept. Evolution 56:1557–1565.
Hull, D. L. 1997. The ideal species concept—And why we can’t get it.
Pages 357–380 in Species: The units of biodiversity (M. F. Claridge,
H. A. Dawah, and M. R. Wilson, eds.). Chapman & Hall, London.
Hurst, G. D. D., and F. M. Jiggins. 2005. Problems with mitochondrial
DNA as a marker in population, phylogeographic and phylogenetic
studies: The effects of inherited symbionts. Proc. R. Soc. Lond. B Biol.
272:1525–1534.
Irwin, D. E. 2002. Phylogeographic breaks without geographic barriers
to gene flow. Evolution 56:2383–2394.
Jacobs, D. K., T. A. Haney, and K. D. Louie. 2004.Genes, diversity, and
geologic process on the Pacific Coast. Annu. Rev. Earth Planet. Sci.
32:601–652.
Jakob, S. S., A. Ihlow, and F. R. Blattner. 2007. Combined ecological niche
modeling an molecular phylogeography revealed the evolutionary
history of Hordeum marinum (Poaceae)—Niche differentiation, loss
of genetic diversity, and speciation in Mediterranean Quaternary
refugia. Mol. Ecol. 16:1713–1727.
Ji, Y. J., D. X. Zhang, and L. J. He. 2003. Evolutionary conservation
and versatility of a new set of primers for amplifying the ribosomal
internal transcribed spacer regions in insects and other invertebrates.
Mol. Ecol. Notes 3:581–585.
Johnson, J. B., T. E. Dowling, and M. C. Belk. 2004. Neglected taxonomy
of rare desert fishes: Congruent evidence for two species of leatherside chub. Syst. Biol. 53:841–855.
Jukes, T. H., and C. R. Cantor. 1969. Evolutions of protein molecules.
Pages 21–132 in Mammalian protein metabolism (H. N. Munro, ed.).
Academic Press, New York.
Kronauer, D. J. C., P. J. Bergmann, J. M. Mercer, and A. P. Russell.
2005. A phylogeo graphically distinct and deep divergence in the
widespread Neotropical turnip-tailed gecko, Thecadactylus rapicauda.
Mol. Phylogenet. Evol. 34:431–437.
Kuo, C. H., and J. Avise. 2005. Phylogeographic breaks in low-dispersal
species: The emergence of concordance across gene trees. Genetica
124:179–186.
Lapointe F. J., and L. J. Rissler. 2005. Congruence, consensus, and the
comparative phylogeography of codistributed species in California.
Am. Nat. 166:290–299.
2008
BOND AND STOCKMAN—SPECIES DELIMITATION WITH EXTREME POPULATION STRUCTURING
Lewis, P. O. 2001. Phylogenetic systematics turns over a new leaf.
Trends Ecol. Evol. 16:30–37.
Maddison, W. P. 1997. Gene trees in species trees. Syst. Biol. 46:523–536.
Maddison, W. P., and D. R. Maddison. 2007. Mesquite: A
modular system for evolutionary analysis. Version 2.0.
http://mesquiteproject.org.
Main, B. Y. 1978. Biology of the arid-adapted Australian trapdoor spider
Anidiops villosus (rainbow). Bull. Br. Arachnol. Soc. 4:161–175.
Masta, S. E., and W. P. Maddison. 2002. Sexual selection driving diversification in jumping spiders. Proc. Natl. Acad. Sci. USA 99:4442–4447.
Matocq, M. D. 2002. Morphological and molecular analysis of a contact
zone in the neotoma fuscipes species complex. J. Mammal. 83:866–883.
McCune, B., and M. J. Mefford. 1999. PC-ORD: Multivariate analysis of
ecological data. Version 4.0. MjM Software Design, Gleneden Beach,
Oregon.
McGuire, J., C. W. Linkem, M. S. Koo, D. W. Hutchison, A. K. Lappin, D. I. Orange, J. Lemos-Espinal, B. R. Riddle, and J. R. Jaeger.
2007. Mitochondrial introgression and incomplete lineage sorting
through space and time: Phylogenetics of crotaphytid lizards. Evolution 61:2879–2897.
Moore, W. S. 1995. Inferring phylogenies from mtDNA variation—
mitochondrial-gene trees versus nuclear-gene trees. Evolution
49:718–726.
Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. da Fonseca,
and J. Kent. 2000. Biodiversity hotspots for conservation priorities.
Nature 403:853–858.
Nylander, J. A. A., F. Ronquist, J. P. Huelsenbeck, and J. L. NievesAldrey. 2004. Bayesian phylogenetic analysis of combined data. Syst.
Biol. 53:47–67.
Ornduff, R., P. M. Faber, and T. Keeler-Wolf. 2003. Introduction to California plant life. University of California Press, Berkeley, California.
Pearson, R. G., C. J. Raxworthy, M. Nakamura, and A. T. Peterson. 2006.
Predicting species distributions from small numbers of occurrence
records: A test case using cryptic geckos in Madagascar. J. Biogeogr.
34:102–117.
Peterson, A. T., J. Soberon, and V. Sanchez-Cordero. 1999. Conservatism
of ecological niches in evolutionary time. Science 285:1265–1267.
Phillips, S. J., R. P. Anderson, and R. E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model.
190:231–259.
Pons, J., T. G. Barraclough, J. Gomez-Zurita, A. Cardoso, D. P. Duran, S.
Hazell, S. Kamoun, W. D. Sumlin, A. P. Vogler. 2006. Sequence-based
species delimitation for the DNA taxonomy of undescribed insects.
Syst. Biol. 55:595–609.
Rader, R. B., M. C. Belk, D. K. Shiozawa, and K. A. Crandall. 2005.
Empirical tests for ecological exchangeability. Anim. Conserv. 8:239–
247.
Rambaut, A., and A. Drummond. 2005. Tracer, A program for
analyzing results from Bayesian MCMC programs such as
BEAST & MrBayes. Version 1.3. Oxford, UK. Available at
http://evolve.zoo.ox.ac.uk/software.html?id=tracer (last accessed
12 February 2006).
Rasband, W. S. ImageJ. 2007. U. S. National Institutes of Health,
Bethesda, Maryland. http://rsb.info.nih.gov/ij/, 1997–2007.
Rissler, L. J., and J. J. Apodaca. 2007. Adding more ecology into species
delimitation: Ecological niche models and phylogeography help define cryptic species in the Black Salamander (Aneides flavipunctatus).
Syst. Biol. 56:924–942.
Rissler, L. J., R. J. Hijmans, C. H. Graham, C. Moritz, and D. B. Wake.
2006. Phylogeographic lineages and species comparisons in conservation analyses: A case study of California herpetofauna. Am. Nat.
167:655–666.
Roe, A. D., and F. A. H. Sperling. 2007. Population structure and species
boundary delimitation of cryptic Dioryctria moths: An integrative
approach. Mol. Ecol. 16:3617–3633.
Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574.
Rubinoff, D., and B. S. Holland. 2005. Between two extremes: Mitochondrial DNA is neither the panacea nor the nemesis of phylogenetic and
taxonomic inference. Syst. Biol. 54:952–961.
Sanders, K. L., A. Malhotra, and R. S. Thorpe. 2006. Combining molecular, morphological and ecological data to infer species boundaries
in a cryptic tropical pitviper. Biol. J. Linn. Soc. 87:343–364.
645
Schlick-Steiner, B. C., F. M. Steiner, K. Moder, B. Seifert, M. Sanetra,
E. Dyreson, C. Stauffer, and E. Christian. 2006. A multidisciplinary
approach reveals cryptic diversity in Western Palearctic Tetramorium
ants (Hymenoptera: Formicidae). Mol. Phylogenet. Evol. 40:259–273.
Schoenherr, A. A. 1992. A natural history of California. University of
California Press, Berkeley, California.
Shaffer, H. B., G. B. Pauly, J. C. Oliver, and P. C. Trenham. 2004. The
molecular phylogenetics of endangerment: Cryptic variation and
historical phylogeography of the California tiger salamander, Ambystoma californiense. Mol. Ecol. 13:3033–3049.
Sinclair, E. A., R. L. Bezy, K. Bolles, J. L. Camarillo, K. A. Crandall, and J.
W. Sites. 2004. Testing species boundaries in an ancient species complex with deep phylogeographic history: Genus Xantusia (Squamata:
Xantusiidae). Am. Nat. 164:396–414.
Sites, J. W., and J. C. Marshall. 2004. Empirical criteria for delimiting
species. Ann. Rev. Ecol. Evol. Syst. 35:199–227.
Slatkin, M., and W. P. Maddison. 1989. A cladistic measure of gene flow
inferred from the phylogenies of alleles. Genetics 123:603–613.
Starrett, J., and M. Hedin. 2007. Multilocus genealogies reveal multiple cryptic species and biogeographic complexity in the California
turret spider Antrodiaetus riversi (Myglomorphae, Antrodiaetidae).
Mol. Ecol. 16:583–604.
Stockman, A. K, and J. E. Bond. 2007. Delimiting cohesion species: Extreme population structuring and the role of ecological interchangeability. Mol. Ecol. 16:3374–3392.
Stockman, A. K., R. M. Danell, and J. E. Bond. 2008. D-NOVL: A
program to simulate overlap between two niche-based distribution
models. Mol. Ecol. Resources 8:290–294.
Swenson, N. G. 2005. GIS-based niche models reveal unifying climatic
mechanisms that maintain the location of avian hybrid zones in a
North American suture zone. J. Evol. Biol. 19:717–725.
Swofford, D. L. 2002. PAUP*: Phylogenetic analysis using parsimony
(*and other methods). Version 4.0b8a. Sinauer Associates, Sunderland, Massachusetts.
Templeton, A. R. 1989. The meaning of species and speciation: A genetic perspective. Pages 3–27 in Speciation and its consequences
(D. Otte and J. A. Endler, eds.). Sinauer Associates, Sunderland,
Massachusetts.
Templeton, A. R. 1998. Species and speciation: Geography, population structure, ecology and gene trees. Pages 32–41 in Endless forms:
Species and speciation (D. J. Howard and S. H. Berlocher, eds.). Oxford University Press, New York, New York.
Templeton, A. R. 2001. Using phylogeographic analyses of gene trees
to test species status and processes. Mol. Ecol. 10:779–791.
Templeton, A. R., E. Boerwinkle, and C. F. Sing. 1987. A cladistic
analysis of phenotypic associations with haplotypes inferred from
restriction endonuclease mapping. I. Basic theory and an analysis
of alcohol dehydrogenase activity in Drosophila. Genetics 117:343–
351.
Templeton, A. R., K. A. Crandall, and C. F. Sing. 1992. A cladistic
analysis of phenotypic associations with haplotypes inferred from
endonuclease mapping and DNA sequence data. 3. Cladogram estimation. Genetics 132:619–633.
Thompson J., T. J. Gibson, F. Plewniak, and F. Jeanmougin, D. G. Higgins. 1997. The ClustalX windows interface: Flexible strategies for
multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25:4876–4882.
Wake, D. B. 2006. Problems with species: Patterns and processes of
species formation in salamanders. Ann. Missouri Bot. Gard. 93:8–23.
Wiens, J. J. 2007. Species delimitation: New approaches for discovering
diversity. Syst. Biol. 56:875–878.
Wiens, J. J., and C. H. Graham. 2005. Niche convervatism: Integrating
evolution, ecology, and conservation biology. Annu. Rev. Ecol. Evol.
Syst. 36:519–39.
Wiens, J. J., and T. A. Penkrot. 2002. Delimiting species using DNA
and morphological variation and discordant species limits in spiny
lizards (Sceloporus). Syst. Biol. 51:69–91.
Will, K. W., B. D. Mishler, and Q. D. Wheeler. 2005. The perils of DNA
barcoding and the need for integrative taxonomy. Syst. Biol. 54:844–
851.
Yoder, A. D., L. E. Olson, C. Hanley, K. L. Heckman, R. Rasoloarison,
A. L. Russell, J. Ranivo, V. Soarimalala, K. P. Karanth, A. P. Raselimanana, and S. M. Goodman. 2005. A multidimensional approach
646
SYSTEMATIC BIOLOGY
for detecting species patterns in Malagasy vertebrates. Proc. Natl.
Acad. Sci. USA 102:6587–6594.
Zwickl, D. J. 2006. Genetic algorithm approaches for the phylogenetic
analysis of large biological sequence datasets under the maximumlikelihood criterion. PhD dissertation, the University of Texas at
Austin.
First submitted 7 December 2007; reviews returned 13 February 2008;
final acceptance 24 April 2008
Associate Editor: Jack Sullivan
APPENDIX
Species descriptions and diagnoses are given in a rapid format to
expedite formal documentation of the biodiversity assessed as part
of this study. Detailed descriptions and illustrations of these taxa will
follow in a monograph treating the diversity of the entire genus. We
rely on unique combinations of nucleotide substitutions for diagnosis (sensu Bond and Sierwald, 2002; Bond, 2004). Nucleotide changes
given in bold type are changes that occur in the clade in question and
are not positions that show reversals to a more common state found
immediately outside the species being diagnosed. Alignment positions
used for diagnosis can be ascertained from the reference alignment deposited in TreeBASE (accession numbers S2114). Species authorship is
to be attributed to the first author (J. E. Bond).
Species of the Atomarius complex
(Cyrtaucheniidae: Euctenizinae)
Aptostichus atomarius Simon 1891, clade 1, Southern California.
Aptostichus stanfordianus Smith 1908, clade 4/5 Central California.
Aptostichus stephencolberti Bond sp. nov., clade 2, California: Coastal
Beaches of Monterey, Santa Cruz, San Mateo, and San Francisco
counties.
Aptotichus angelinajoleae Bond sp. nov., clade 3, California: Monterey
and San Benito Counties.
Aptostichus miwok Bond sp. nov., defined at node I, California: Coastal
Beaches of Marin, Sonoma, Mendocino, Humboldt, and Del Norte
counties.
Aptostichus stephencolberti Bond sp. nov. (Figs. 1, 6b)
Type specimens.—Male holotype (MY3515; EU570029) and female paratype (MY3513, EU570028) deposited in the CAS. USA, California, San Mateo County, Pescadero State Beach. 122.41219, −37.26598,
elev. 4 m. Colls. A. Stockman, P. Marek, 29 Jan 2006.
Diagnosis.—Aptostichus stephencolberti can be distinguished from
the other closely related species on the basis of the following unique
mtDNA nucleotide substitutions at the following reference alignment
positions: A (154), A (161), G (307), G (596), T (683), A (684), G (691), G
(706), T (709), G (742), G (747), C (771), A (867), G (909) A (948), T (963), G
(1023), G (1062), A (1068), A (1111), T (1378), T (1383), T (1385), T (1389),
T (1475). This species is lighter in coloration than A. angelinajolieae and
is restricted to beach habitats, whereas A. angelinajolieae is darker in
color and found inland.
Etymology.—The specific epithet is a patronym, named in honor
of Mr. Stephen Colbert. Mr. Colbert is a fellow citizen who truly has
the courage of his convictions and is willing to undertake the very
difficult and sometimes unpopular work of speaking out against those
who have done irreparable harm to our country and the world through
both action and inaction. He will be especially remembered by many
of Jason Bond’s generation for his speech at the 2006 White House
Correspondents Dinner.
Description.—Male holotype. Carapace, chelicerae, legs brownish
yellow (10YR 6/6). Abdomen uniform, lighter with dusky stripes. Carapace 6.38 long, 5.63 wide, hirsute, margin with fringe of black setae.
Eye group 1.00 wide. Outer margin of cheliceral furrow with 6 teeth.
VOL. 57
PTl 3.25, PTw 1.00, Bl 1.40. Cymbium with 6 spines. Leg I: 5.63, 3.00,
3.72, 3.68, 2.40; tibia and metatarsus of leg I; spination as illustrated in
Fig. 1 (Aj); TSp 6, TSr 6, TSrd 8. Female paratype. Carapace, chelicerae
coloration like that of male. Abdomen with light dusk stripes. Carapace 6.00 long, 5.19 wide, lightly hirsute. Eye group 1.05 wide. Labium
with 2 cuspules. Rastellum distinct, small cluster of short, stout spines
(5). Outer margin of cheliceral furrow with 5 teeth. Walking legs: Leg
I: 4.12, 2.69, 2.31, 1.87, 1.25; tarsal scopula heavy to moderate on legs I
and II.
Aptostichus angelinajolieae Bond sp. nov. (Figs. 1 and 6c)
Type specimens.—Female holotype (MY3310; EU569958) and
male paratype (AP167) deposited in the California Academy of Sciences. Type locality: USA, California, Monterey County, Carmel Valley
Rd/G16, 3.7 miles North of Arroyo Seco Rd. 36.29045, −121.46594, elev.
337 m, oak scrub habitat. Coll. A. Stockman, 9 Jun 2005.
Etymology.—The specific epithet is a patronym in honor of Ms.
Angelina Jolie in recognition of her work on the United Nations High
Commission for Refugees.
Diagnosis.—Aptostichus angelinajolieae differs from the other
closely related species on the basis of the following unique mtDNA nucleotide substitutions at the following reference alignment positions:
G (70), C (99), G (112), G (123), T (199), G (208), A (231), G/T (236), A
(272), G (317), G (342), C (360).
Description.—Female holotype. Carapace, chelicerae, legs yellowish red (5YR 4/6). Abdomen gray in hue with distinct chevron pattern
(Fig. 4). Carapace 8.40 long, 7.00 wide, lightly hirsute. Eye group raised,
1.25 wide. Labium with 3 cuspules. Rastellum distinct, small cluster of
short, stout spines (6). Outer margin of cheliceral furrow armed with
6 teeth. Walking legs: Leg I: 6.00, 3.90, 3.69, 2.81, 1.87. Male paratype.
Like female specimen. Carapace 6.30 long, 5.50 wide, hirsute, margin with fringe of heavy black setae. Eye group 1.05 wide. AER and
PER straight. Labium lacks cuspules. Outer margin of cheliceral furrow
armed with 6 teeth, inner margin lacks denticles. Palp articles relatively
slender, PTl 2.63, PTw 0.85, Bl 1.30. Cymbium with 4 spines. Leg I: 5.37,
3.00, 3.76, 3.76, 2.40; Tarsae scopulae light on legs I and II. Tibia and
metatarsus of legs I; spination as illustrated in Fig. 1 (As); TSp 8, TSr 5,
TSrd 4.
Aptostichus miwok Bond sp. nov. (Figs. 1, 6f)
Type specimens.—Female holotype (MY301; EU69907) and male
paratype (AP149) deposited in the CAS. USA, California, Humboldt
County, Clam Beach County Park. 41.01333 −124.10923, elev. 1 m. Colls.
J. Bond and M. Hedin, 13 Jan 2002. Male paratype from USA, California, San Francisco County, Farallon Island. Coll. M. G. Ramiriz,
2 Sep 1986.
Etymology.—The specific epithet is in honor of the Coast Miwok
Indian tribe known to have inhabited the coastal areas of California
from the Golden Gate northward prior to European settlement.
Diagnosis.—Aptostichus miwok can be distinguished on the basis of
the following unique mtDNA nucleotide substitutions at the following
reference alignment positions: T (54), T (192), G (303), A (310), A (342),
C (459), A (513), G (714), T (726), T (1352).
Description.—Female holotype. Carapace, chelicerae, legs brownish yellow (10YR 6/6). Abdomen uniform much lighter with light dusk
stripes (Fig. 4). Carapace 6.25 long, 5.81 wide, lightly hirsute. Eye group
1.23 wide. Labium with 8 cuspules. Rastellum distinct cluster of short,
stout spines (6). Outer margin of cheliceral furrow armed with 8 teeth,
inner margin with small proximal patch of denticles. Walking legs: Leg
I: 4.56, 2.80, 2.88, 2.20, 1.36. Male paratype. Specimen badly damaged.
Only leg I measurements and spination patterns are reported here as a
reference to paratype specimen. Leg I: 4.80, 2.40, 3.20, 3.20, 1.80. Tibia
and metatarsus of legs I; spination as illustrated in Figure 1 (Am); TSp
5, TSr 3, TSrd 4.