Journal of Heredity, 2023, 114, 131–142

https://doi.org/10.1093/jhered/esad001
Advance access publication 13 January 2023
Original Article

Original Article
Predictors of genomic diversity within North American
squamates
Ivy E. Larkin1, , Edward A. Myers2, , Bryan C. Carstens1, , Lisa N. Barrow1,3,
1
Department of Evolution, Ecology and Organismal Biology, The Ohio State University, Columbus, OH, United States,
2
Department of Biological Sciences, Clemson University, Clemson, SC, United States
3
Present address: Museum of Southwestern Biology and Department of Biology, University of New Mexico, Albuquerque, NM 87131, United
States.
Address correspondence to L.N. Barrow at the address above, or e-mail: [email protected].
Corresponding Editor: Sara Ruane

Abstract
Comparisons of intraspecific genetic diversity across species can reveal the roles of geography, ecology, and life history in shaping biodiversity.
The wide availability of mitochondrial DNA (mtDNA) sequences in open-access databases makes this marker practical for conducting analyses
across several species in a common framework, but patterns may not be representative of overall species diversity. Here, we gather new and ex-
isting mtDNA sequences and genome-wide nuclear data (genotyping-by-sequencing; GBS) for 30 North American squamate species sampled in
the Southeastern and Southwestern United States. We estimated mtDNA nucleotide diversity for 2 mtDNA genes, COI (22 species alignments;
average 16 sequences) and cytb (22 species; average 58 sequences), as well as nuclear heterozygosity and nucleotide diversity from GBS data
for 118 individuals (30 species; 4 individuals and 6,820 to 44,309 loci per species). We showed that nuclear genomic diversity estimates were
highly consistent across individuals for some species, while other species showed large differences depending on the locality sampled. Range
size was positively correlated with both cytb diversity (phylogenetically independent contrasts: R2 = 0.31, P = 0.007) and GBS diversity (R2 =
0.21; P = 0.006), while other predictors differed across the top models for each dataset. Mitochondrial and nuclear diversity estimates were not
correlated within species, although sampling differences in the data available made these datasets difficult to compare. Further study of mtDNA
and nuclear diversity sampled across species’ ranges is needed to evaluate the roles of geography and life history in structuring diversity across
a variety of taxonomic groups.
Key words: American Southwest, Florida, genotyping-by-sequencing, mitochondrial DNA, nucleotide diversity, Serpentes

Introduction The type of genetic data chosen for comparative studies

Genetic information is key for investigating the evolutionary is important to consider. Most work to date has used widely
histories and demographic trajectories of populations and available markers from organellar genomes, but the extent
species. Since the advent of DNA sequencing in the late to which these data are representative of broader genomic
1970s (Sanger et al. 1977), analysis of sequence variation has patterns remains unclear. Mitochondrial DNA (mtDNA)
enabled biologists to discover species, determine relationships is easily sequenced and has been used for decades, making
among lineages, reveal demographic histories, and provide it readily accessible from public databases for comparisons
assessments of population structure and diversity (Avise across large numbers of species (Zink and Barrowclough
2000; Hebert et al. 2003; Brito and Edwards 2009). Genetic 2008; Miraldo et al. 2016). While certain characteristics of
diversity within species is especially critical for persistence in mtDNA, such as a high mutation rate and lack of recom-
changing environments and is clearly of interest for global bination, make it desirable for population-level studies and
conservation (Laikre 2010; Laikre et al. 2020), but several the inference of phylogenetic relationships at shallow scales
challenges remain for gathering and comparing genetic in- (Avise et al. 1987), mtDNA may be poorly suited as a rep-
formation from large numbers of species. The extent to which resentation of genomic diversity in a given species (Galtier
genetic diversity is predictable given species characteristics et al. 2009). As a nonrecombinant unit that is matrilineally
such as range size or life history traits is an area of active in- inherited, mtDNA represents a single realization of many
terest (e.g. Leffler et al. 2012; Romiguier et al. 2014; Singhal possible coalescent histories of sampled taxa, and it is well
et al. 2017b; Pelletier and Carstens 2018; Barrow et al. 2021). appreciated that any single genealogy may not accurately re-
These studies demonstrate that predictors often differ across flect population or species history (Hudson and Turelli 2003)
different taxonomic groups and scales, prompting the need particularly in cases where sex-biased dispersal or multiple
for additional investigations into understudied clades. divergences in a short timespan have occurred (Morin et al.

Received July 19, 2022; Accepted January 13, 2023

© The American Genetic Association. 2023.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/
licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For
commercial re-use, please contact [email protected]
132 Journal of Heredity, 2023, Vol. 114, No. 2

2004; Galtier et al. 2009). Notably, several investigations Life history and ecological traits also correspond with ge-
comparing genetic diversity from mitochondrial and nuclear netic diversity across broad taxonomic groups (Romiguier et
genomes within species have found conflicting results across al. 2014; Chen et al. 2017). Body size is negatively correlated
different taxonomic groups (Bazin et al. 2006; Mulligan et with genetic diversity within and between populations for
al. 2006; Singhal et al. 2017b). Bazin et al. (2006) showed some groups (e.g. mammals: Brüniche-Olsen et al. 2018; bees:
that mtDNA did not correspond with expectations of popula- López-Uribe et al. 2019; butterflies: Mackintosh et al. 2019),
tion abundance when comparing across broad animal groups, which could relate to limits on population size (leading to
e.g. invertebrate groups did not have higher mtDNA diversity lower within-population diversity) or to higher dispersal
than vertebrate groups, while nuclear sequence and allozyme (leading to lower genetic structure) in larger-bodied species
datasets did meet this expectation. Comparisons within ver- (White et al. 2007; Paz et al. 2015). Fecundity, or clutch size,
tebrate groups such as mammals (Mulligan et al. 2006) and is also expected to relate to abundance, where species that
lizards (Singhal et al. 2017b) have demonstrated a positive have more offspring will have larger and more stable popu-
correlation between estimates of mtDNA and nuclear diver- lation sizes through time and therefore higher neutral diver-
sity, however, suggesting that mtDNA may be a useful marker sity (Kimura 1979). Indeed, clutch size is positively correlated
for understanding patterns of genetic diversity in animals with genetic diversity across animals (Romiguier et al. 2014),
with smaller population sizes. Taken together, the contrasting where species that have many, small offspring have higher
results of these studies also illustrate how the taxonomic scale within-population genetic diversity than long-lived species
of a given study can lead to different findings. with few, large offspring. Aspects of mating system or repro-
Sampling many nuclear loci enables more robust estimates ductive mode have also been associated with genetic varia-
of species relationships, population history, diversity tion in different groups. In plants, within-population genomic
estimates, and demographic parameters of interest (Edwards diversity is higher in outcrossing species compared with
and Beerli 2000; Carling and Brumfield 2007). The growing selfing species (Chen et al. 2017), which is expected because
availability of genome-scale nuclear datasets is promising for inbreeding reduces effective population size.
comparative studies (Garrick et al. 2015), but sampling many Different reproductive strategies may also lead to differences
individuals and populations for multiple species in a single in dispersal distance and resulting genetic structure. For ex-
study remains cost prohibitive. One solution implemented ample, in Panamanian frogs, species with direct development
in previous comparative studies is to obtain genome-wide exhibit greater genetic structure compared with larval devel-
estimates of diversity from a single or few individuals as a rep- oping species (Paz et al. 2015); and in marine invertebrates,
resentative of each species (e.g. Romiguier et al. 2014; Singhal benthic direct-developing species have higher genetic struc-
et al. 2017b; Grundler et al. 2019). It is not yet clear how con- ture than those with a pelagic larval stage (Collin 2001; Lee
sistent genomic diversity estimates are across populations of and Boulding 2009). Another aspect of reproduction that
wide-ranging species that have experienced different histories has not been thoroughly addressed is parity mode, whether
across their ranges (but see Nazareno et al. 2017 for an ex- species lay eggs (oviparous) or have live young (viviparous).
ample in plant populations separated by 20 km). We begin to Oviparity could be associated with higher dispersal, and less
address this question in squamates by evaluating genomic di- genetic structure, if females must travel long distances to lo-
versity from multiple individuals and localities within species cate suitable nest sites, while viviparous species have reduced
in the present study. movements because of increased energetic costs (Shine 2015).
In addition to considering appropriate sampling strategies, In the common lizard (Zootoca vivipara), greater genetic
identifying potential predictors of genomic diversity within structure was found in an oviparous lineage compared with
species is challenging because many species-level character- a viviparous lineage, but in this case, the pattern could be
istics may correspond with diversity within populations, explained by differences in demographic history (Cornetti
diversity between populations (i.e. genetic structure), or et al. 2015). Comparisons of species-wide genetic diversity
both. Species with larger census population sizes are ex- across multiple oviparous and viviparous species are lacking
pected to have larger effective population sizes and therefore thus far.
higher levels of neutral genetic diversity within populations
(Kimura 1979). This assumed relationship may also extend
to total range size, which has been considered as a proxy North American squamates in regions of
for census population size (e.g. Leffler et al. 2012; Singhal contrasting topographic complexity
et al. 2017b) because species that occupy large areas are This study focuses on the genetic diversity of squamates
presumed to be more locally abundant. The abundance– from North America in 2 regions that were climatically suit-
range size relationship may be explained by a variety of able during the Last Glacial Maximum (LGM; Waltari et al.
mechanisms (Gaston et al. 1997). In analyses of some taxo- 2007), but differ in their topographic complexity (Badgley et
nomic groups, population density appears to be unrelated to al. 2017). We investigated species from the North American
range size (Novosolov et al. 2017), while empirical examples Coastal Plain and the Desert Southwest (Fig. 1), both of
of island and mainland bird species have demonstrated the which harbor high reptile diversity (Jenkins et al. 2015)
expected positive relationship between genomic diversity and are currently under threat from habitat fragmenta-
and range size (Brüniche-Olsen et al. 2019; Leroy et al. tion, climate change, and change in wildfire regimes (Archer
2021). Furthermore, the size, shape, and characteristics of a and Predick 2008; Noss et al. 2015; Briggs et al. 2020).
species’ range can impact how individuals move across the The North American Coastal Plain includes the Gulf and
landscape, thus influencing rates of dispersal, gene flow, and Atlantic coasts in the Southeastern United States and has
the maintenance of genetic structure and overall diversity been described as a biodiversity hotspot (Noss et al. 2015).
within species (Wright 1943; Sexton et al. 2014; Myers et Although the region was never glaciated, when climatic
al. 2019). and habitat shifts during Pleistocene glacial cycles occurred
Journal of Heredity, 2023, Vol. 114, No. 2 133

Fig. 1. Individual sample localities and nucleotide diversity estimates for species included in the study. a) Map of all samples for which GBS data were
analyzed. Desert Southwest samples are shown in turquoise and Coastal Plain species in brown. Dotted black boxes indicate the inset maps to the
right, which depict localities sampled within 2 species as examples. Shapes indicate individuals were sampled from the same locality within a species,
but do not necessarily indicate the same locality across different species. b) Phylogenetic relationships for the 30 species included in the study based
on Tonini et al. (2016). Within-individual nucleotide estimates are shown to the right with colors and shapes corresponding to the maps in (a).

(Williams et al. 2004), the region likely provided stable refugial populations during the Pleistocene. Furthermore,
refugial areas for many of the species in this study (Soltis the high squamate diversity in these regions provides the op-
et al. 2006; Weinell and Austin 2017). Within the Desert portunity to compare species with a variety of range sizes,
Southwest, located in the southwestern United States and body sizes, and life history traits.
north-central Mexico, the Chihuahuan Desert is the most In this study, we combined new and existing mtDNA and
biologically diverse desert and is considered among the 200 genome-scale nuclear sequence data for 30 North American
most biologically valuable ecoregions globally (Olson and squamate species to address 3 main questions. First, how con-
Dinerstein 1998; Briggs et al. 2020). This region is also sistent are nuclear genomic estimates across individuals and
predicted to have provided refugial areas during the LGM localities within species? Second, are measures of mtDNA
(Waltari et al. 2007), but is more topographically complex and nuclear diversity within species correlated, suggesting
with high environmental heterogeneity that plays an impor- that mtDNA is a useful proxy for within-species diversity?
tant role in population genetic structure and lineage diver- Third, are species geographic range size, body size, or life
gence (Badgley et al. 2017; Myers et al. 2019). Both regions history traits (number of offspring and reproductive mode)
should retain much of the ancestral genetic diversity of the associated with either mtDNA or nuclear diversity within
sampled species since each is suspected to have harbored squamates?
134 Journal of Heredity, 2023, Vol. 114, No. 2

Methods S3). Additional mtDNA sequences for 2 genes, COI and cy-
tochrome b (cytb), were obtained from NCBI GenBank for
Sample collection
species with data available. Sequences for each species were
We gathered and analyzed sequence data from 30 spe- aligned in Geneious 2020.1.1 (https://www.geneious.com)
cies representing 4 snake families (Viperidae, Colubridae, using MAFFT 7.450 (Katoh and Standley 2013).
Natricidae, and Dipsadidae) and 2 lizard families (Anolidae
and Anguidae) (Fig. 1). We followed the most recent taxo-
nomic revisions in the literature, ensuring mtDNA sequences
Nuclear data collection
could be assigned to a single species based on their specific Genotyping-by-sequencing (GBS) libraries were prepared
localities, and excluding sequences without locality informa- following a modified version of the protocol described in
tion (Supplementary Table S1). Tissue samples for 12 Coastal Elshire et al. (2011). For 4 samples per Coastal Plain species
Plain species (4 to 7 individuals each) were collected in (48 individuals), we digested 100 ng of input DNA with the
northern Florida from 2009 to 2018, preserved in either 95% enzyme Pst1, ligated 8 µL of adapter mix including unique
ethanol or DMSO tissue buffer, frozen at or below −20 °C, barcodes, pooled libraries, and performed a bead cleanup
and subsequently archived at the Museum of Southwestern with Sera-Mag Speedbeads (Rohland and Reich 2012). Final
Biology (Fig. 1; Supplementary Table S2). We also included PCR amplification was conducted with 16 cycles in 8 repli-
previously published data for 18 snake species distrib- cate reactions, followed by pooling, bead cleanup, and quanti-
uted across the Desert Southwest (Myers et al. 2019, 2020; fication via a Qubit fluorometer (Life Technologies). Libraries
Myers 2021). For the nuclear dataset of these 18 species, we were size selected to 200 to 500 bp using a Blue Pippen
chose 4 individuals with a similar sampling strategy as the (Sage Science Inc.), quantified with a Bioanalyzer (Agilent
Coastal Plain species. Briefly, we chose 2 individuals each Technologies), and sequenced at the OSU Comprehensive
from 2 localities per species when possible, focusing prima- Cancer Center Genomics Shared Resource on an Illumina
rily on the Chihuahuan Desert and a similar distance (~200 HiSeq 4000 with 150 bp paired-end sequencing. For 4
km) between localities. This standardized distance enabled samples per Desert Southwest species (72 individuals), we
us to make comparisons among species without the poten- downloaded GBS reads from the NCBI Sequence Read
tially confounding effects of geographic distance on genetic Archive (Supplementary Table S2).
structure. Sequenced GBS reads from all 30 species were assembled
through the ipyrad 0.7.28 pipeline (Eaton and Overcast
Range maps and trait data 2020) with the following settings. After demultiplexing (no
Species range maps were downloaded as shapefiles from the mismatches allowed in barcodes), the R1 reads for the 4
IUCN Red List of Threatened Species Version 6.1 (www. individuals per species were assembled to generate within-
iucnredlist.org/). Maps were edited in QGIS 3.12 as needed to species datasets. We trimmed reads for the Southeastern US
exclude nonnative parts of the range where species have been species to 125 bp prior to assembly using the built-in ipyrad
introduced and to edit species ranges according to recent tax- option which uses the software tool “cutadapt” (trim_reads
onomic changes (Supplementary Table S1). The R packages = 0, −25). We used the de novo assembly method with a
“rgdal” (Bivand et al. 2021) and “geosphere” (Karney 2013; clustering similarity threshold of 85% (clust_threshold =
Hijmans 2021) were used to visualize ranges and calcu- 0.85), maximum clustering depth of 10,000 reads (maxdepth
late total range area in kilometers squared. Species trait in- = 10,000), maximum of 5 low quality base calls per read
formation was obtained primarily from Burbrink and Myers (max_low_qual_bases = 5; with Q < 20), minimum depth
(2015), including maximum body size (log-transformed), av- for base calling of 6 reads (mindepth_statistical = 6 and
erage clutch size (log-transformed), and parity (viviparous or mindepth_majrule = 6), strict filtering for adapters (filter_
oviparous). Trait data for the 2 lizard species were obtained adapters = 2), and we retained reads longer than 35 bp after
from field guides and online natural history accounts (Powell trimming adapters (filter_min_trim_len = 35). We allowed
et al. 2016; https://www.virginiaherpetologicalsociety.com; a maximum of 8 heterozygous sites in consensus sequences
https://animaldiversity.org). (max_Hs_consens = 5), a maximum of 20 single nucleotide
polymorphisms (SNPs) per locus (max_SNPs_locus = 20),
Mitochondrial sequences and required all 4 individuals (no missing data) to be included
For the 12 Coastal Plain species, genomic DNA was extracted in a locus to retain that locus in output files (min_samples_
from tail or liver tissue with the E.Z.N.A. DNA Tissue Kit locus = 4).
(Omega Bio-Tek) according to the manufacturer’s instructions.
We amplified a 658-base pair portion of the mtDNA cyto- Genetic diversity metrics
chrome c oxidase 1 gene (COI) using primers ReptBCF_COI Measures of mtDNA diversity were calculated for each spe-
and ReptBCR_COI (Castañeda and de Queiroz 2011). The cies using the R package “pegas” (Paradis 2010). We calcu-
thermal profile included an initial denaturation step at 94 °C lated nucleotide diversity (pi or π, Nei 1987) for COI and
for 3 min, 35 cycles of denaturing at 94 °C for 45 s, annealing cytb alignments of each species with the nuc.div function.
at 53 °C for 30 s, elongating at 72 °C for 60 s, and a final Nuclear diversity within species was estimated from each
extension step at 72 °C for 10 min. PCR products were GBS dataset using the R packages “adegenet” (Jombart
visualized on agarose gels to ensure successful amplification, 2008; Jombart and Ahmed 2011) and “PopGenome” (Pfeifer
cleaned with EXOSAP-IT (Affymetrix Inc.), and sequenced et al. 2014). Expected heterozygosity (Hs) was computed
at The Ohio State University (OSU) Comprehensive from the unique SNPs outfile from ipyrad for each species
Cancer Center Genomics Shared Resource. Species-specific using “adegenet.” To calculate π, “PopGenome” reads in a
sequencing primers were designed when the original PCR directory of FASTA files including an alignment for each
primers did not perform adequately (Supplementary Table locus; we used a custom Python script (available on Dryad)
Journal of Heredity, 2023, Vol. 114, No. 2 135

to generate this directory for each species using the alleles Results
outfiles (full sequences) from ipyrad. We calculated π within
Data summary
each sample (2 alleles per individual) and then determined
the mean and SD across individuals, hereafter referred to as We generated 62 new COI sequences (GenBank accession
“mean within-individual π.” For comparison, we also calcu- numbers ON911378–ON911439) and combined these with
lated π within each species from the alignments including 4 available sequences on GenBank. Of the 30 squamates in
individuals (8 alleles) per species, hereafter “within-species our study, we were able to analyze COI alignments for 22
π.” Species-level and within-population nuclear diversity species (4 to 69 sequences per alignment; average 16) and
can be estimated from a single individual when sufficient cytb alignments for 22 species (8 to 143 sequences; average
loci are sampled (e.g. Chen et al. 2017; Grundler et al. 58). Three species (Crotalus oreganus, Hypsiglena jani, and
2019). Where possible, we included 2 individuals per species Trimorphodon vilkinsonii) did not have sufficient mtDNA
from each sampled locality to investigate how consistent data available for either COI or cytb and 17 species had
genomic diversity estimates are across different individuals alignments for both genes (Table 1). We sequenced new GBS
and localities within species. data for 47 individuals from 12 species (NCBI SRA accession
numbers SAMN31800422–SAMN31800468) and calculated
genomic diversity metrics for 118 individuals from 30 species.
One Thamnophis sirtalis sample had poor sequencing success
Statistical analyses and was removed from subsequent analyses. The assembled
We compared mtDNA (COI π, cytb π) and nuclear (mean GBS datasets included 2,923 to 25,086 loci for the unique
within-individual π, within-species π) diversity estimates SNP datasets and 6,820 to 44,309 loci for the full sequence
within species using pairwise correlations in R. We then datasets depending on the species (Table 1).
constructed phylogenetic generalized least square (PGLS)
models to assess the importance of different predictors on Mitochondrial and nuclear diversity
genetic diversity, while accounting for the possibility that Mitochondrial diversity within species ranged from 0.0015
closely related species may share similar genomic diversity to 0.0812 for COI π and from 0.0005 to 0.0717 for cytb π.
and species traits. We downloaded 100 phylogeny subsets Diversity estimates for the 2 mtDNA genes were correlated
for the 30 species in our study from VertLife.org, which within species (Supplementary Fig. S1; Spearman’s rho (ρ) =
includes the squamate relationships from Tonini et al. (2016), 0.897, P < 2.2e−16). For the nuclear GBS data, within-species
and generated a least-squares consensus tree using phytools heterozygosity ranged from 0.3 to 0.346 and within-species
(Revell 2012) for our analyses. We constructed a model set π ranged from 0.0005 to 0.0083. Mean within-individual π
for each of 3 responses: mtDNA COI π, mtDNA cytb π, ranged from 0.0004 to 0.0056 and there was substantial varia-
and nuclear within-species π. For each response, predictors tion among individuals and localities for some species (Fig. 1).
of interest included geographic range size, maximum body For species such as Thamnophis marcianus and Masticophis
size, average clutch size, and parity. To assess effects of flagellum, the within-individual π estimates were consistent
sample size on diversity metrics, we included the number of between the 2 individuals within a locality, but values for 1
individuals sampled as a covariate in mtDNA models, and locality were nearly double the values for the second locality.
the number of GBS loci as a covariate for nuclear models. For other species such as Crotalus adamanteus and Sonora
We included the region (Coastal Plain or Desert Southwest) episcopa, the within-individual π estimates were highly con-
in our initial PGLS models for nuclear within-species π and sistent for all 4 individuals. For still other species such as
used Welch’s 2-sample t-tests to check for differences between Thamnophis sauritus and Lampropeltis splendida, there was
the GBS datasets generated in this study and those generated large variation in π estimates even for individuals collected
previously. from the same locality.
Analyses were conducted using the R packages “ape” The within-species π estimates were strongly correlated
(Paradis and Schliep 2019), “geiger” (Harmon et al. 2008), with the mean within-individual π estimates (Supplementary
and “nlme” (Pinheiro et al. 2022). We constructed PGLS Fig. S1; Spearman’s ρ = 0.986, P < 2.2e−16), and we used
models using 2 models of evolution, Brownian Motion (BM) the within-species estimate to represent overall species nu-
and Ornstein–Uhlenbeck (OU). The BM model assumes traits clear diversity for subsequent analyses. Metrics of diversity
evolve according to a random walk, while the OU model for mtDNA and nuclear GBS data were not correlated for the
incorporates a trait optimum toward which traits are pulled species represented in both datasets (Fig. 2). Nuclear within-
(Felsenstein 1985; Martins and Hansen 1997). For each species π was not correlated with mtDNA COI π (Spearman’s
model set, all combinations of the predictors of interest were ρ = 0.16, P = 0.476) or mtDNA cytb π (Spearman’s ρ = 0.293,
included and models were ranked based on the corrected P = 0.185). We did not detect any clear differences in nuclear
Akaike Information Criterion (AICc). We ran model selec- diversity between datasets sampled from the Desert Southwest
tion analyses using the R package “MuMIn” (Bartón 2022) compared with the Coastal Plain (Supplementary Fig. S2).
to determine which model of evolution had a better fit. We
then constructed models incorporating all combinations of Predictors of squamate genetic diversity
the predictors of interest and ran model selection analyses The OU model of evolution was the better fit for all 3 re-
to assess which sets of predictors were included in the best- sponse variables (model weights >0.986; Table 2). Different
fit models for each diversity measure. We further assessed sets of predictors were included in the top models for the 3
relationships between continuous variables using Spearman’s response variables (Table 3). For mtDNA COI, parity was
rank correlation tests, phylogenetically independent contrasts the only variable consistently included in the top models.
(PIC) implemented in “ape” (Felsenstein 1985), and linear Diversity between viviparous and oviparous species was not
models. significantly different according to the top PGLS model (Table
136 Journal of Heredity, 2023, Vol. 114, No. 2

Table 1. Sample sizes for mtDNA and GBS datasets for each species.

Species Family # COI seqs # cytb seqs # GBS samples # GBS loci (unique SNPs) # GBS loci (seqs)

Ophisaurus ventralis Anguidae 6 NA 4 22,951 42,875

Anolis carolinensis Anolidae 16 NA 4 15,708 28,909
Agkistrodon conanti Viperidae 5 46 4 5,961 15,540
Sistrurus miliarius Viperidae 10 31 4 14,356 23,527
Crotalus ornatus Viperidae NA 27 4 9,047 29,251
Crotalus atrox Viperidae 5 66 4 6,926 19,307
Crotalus scutulatus Viperidae 4 74 4 14,696 29,502
Crotalus oreganus Viperidae NA NA 4 11,032 27,615
Crotalus viridis Viperidae NA 8 4 15,565 29,627
Crotalus tigris Viperidae 5 NA 4 9,003 37,903
Crotalus pyrrhus Viperidae 6 NA 4 3,274 6,820
Crotalus adamanteus Viperidae 10 126 4 8,424 27,387
Nerodia fasciata Natricidae 30 143 4 15,130 25,148
Thamnophis marcianus Natricidae NA 22 4 7,467 17,648
Thamnophis sauritus Natricidae 7 NA 4 14,540 17,945
Thamnophis sirtalis Natricidae 7 37 3 15,664 25,409
Hypsiglena jani Dipsadidae NA NA 4 19,304 29,950
Sonora episcopa Colubridae 7 20 4 2,923 12,181
Trimorphodon vilkinsonii Colubridae NA NA 3 5,566 26,644
Salvadora hexalepis Colubridae 8 31 4 3,513 20,475
Masticophis flagellum Colubridae 55 136 4 7,512 18,350
Coluber constrictor Colubridae 51 22 4 18,737 22,267
Lampropeltis splendida Colubridae NA 25 4 6,485 19,095
Cemophora coccinea Colubridae 69 9 4 15,437 25,907
Rhinocheilus lecontei Colubridae 9 105 4 4,713 15,022
Arizona elegans Colubridae 8 92 4 16,879 37,596
Pantherophis quadrivittatus Colubridae 12 20 4 14,352 26,136
Pituophis catenifer Colubridae 9 98 4 18,205 32,161
Pantherophis emoryi Colubridae NA 129 4 25,086 44,309
Pantherophis guttatus Colubridae 11 11 4 14,377 25,517

4), although oviparous species did have higher COI diversity Discussion
than viviparous species without phylogenetic correction (Fig.
We compared genetic diversity estimates from mitochondrial
3a; t-test: t = 2.34, P = 0.03). For mtDNA cytb, range size
and nuclear genomes for 30 North American squamate spe-
and parity were always included in the top models. Within-
cies. Our analyses demonstrated that range size is positively
species cytb diversity had a positive relationship with range
correlated with mtDNA and nuclear diversity in squamates
size (Fig. 3), suggesting that species with larger ranges tend
and that life history traits are inconsistent predictors of diver-
to have higher mtDNA diversity. This relationship was con-
sity for these datasets. We found that mtDNA and nuclear di-
sistent for both uncorrected (Spearman’s ρ = 0.64; P = 0.002;
versity estimates were not correlated, but we note that several
Fig. 3c) and phylogenetically corrected datasets (phylogenet-
aspects of sampling made these datasets difficult to compare.
ically independent contrasts: R2 = 0.31; P = 0.007; Fig. 3d).
Here, we discuss sampling considerations for this and future
Oviparous species had higher cytb diversity compared with
studies, including the number of individuals, localities, spe-
viviparous species (Fig. 3b; t-test: t = 3.11, P = 0.008).
cies, and loci available or feasible to sample. Despite these
Range size and clutch size were consistently included in the
challenges, we provide empirical evidence that geography is a
top models for nuclear within-species π (Table 2). Nuclear
key element for understanding patterns of genomic diversity
within-species π had a positive relationship with both range
within and across species.
size and average clutch size (Table 4; Fig. 3). The relationship
between nuclear π and range size was consistent for both un-
corrected (Spearman’s ρ = 0.56; P = 0.001; Fig. 3e) and phylo- Sampling considerations across geography,
genetically corrected datasets (PIC: R2 = 0.21; P = 0.006; Fig. taxonomy, and genomes
3f). The relationship between nuclear π and average clutch Comparative population genomics is a developing field that
size was significant for the uncorrected data (Spearman’s ρ = aims to uncover the forces underlying genomic diversity by
0.53; P = 0.003; Fig. 3g), but not based on phylogenetically comparing across multiple species (Ellegren and Galtier 2016;
independent contrasts (R2 = 0.04; P = 0.288; Fig. 3h). Edwards et al. 2022). Studies typically focus on the roles of
Journal of Heredity, 2023, Vol. 114, No. 2 137

selection, neutrality, and recombination in structuring ge- al. 2017). In contrast, comparative phylogeography takes a
netic variation in a primarily nongeographic context, thus geography-centric view to understanding historical processes
sampling usually consists of a single or few individuals to underlying current genetic variation within species and
represent each species (e.g. Romiguier et al. 2014; Chen et lineages. Ideally, species have been sampled across their entire
range and as densely as possible, often revealing the presence
of multiple lineages within nominal species (e.g. Schield et al.
Nuclear Within−Species Pi

2018). A third emerging field, macrogenetics, focuses on an

rho = 0.16, p = 0.476 even broader geographic and taxonomic scale, typically using
0.008

repurposed data from relatively few genetic markers to inves-

tigate large-scale patterns and predictors of genetic variation
within species (Leigh et al. 2021). Macrogenetic studies rely
0.004

on sampling choices and data from previous studies, but until

relatively recently, genome-wide datasets were not common
in phylogeographic studies (Garrick et al. 2015). The gradual
0.000

accumulation of these data over time and the introduction of

0.00 0.02 0.04 0.06 0.08 new data aggregators (e.g. Pelletier et al. 2022) potentially
increases the scale of comparative studies that use opportun-
mtDNA COI pi
istic sampling, but the novel data collected here offer a direct
Nuclear Within−Species Pi

comparison between 2 ecoregions.

rho = 0.293, p = 0.185 Thoroughly sampling many individuals from many spe-
0.008

cies across geographic space remains costly, thus a pressing

question relevant to the design of comparative studies is:
How representative are genomic diversity estimates from a
0.004

few individuals or from a narrow part of a species’ range?

Our GBS data indicate that this answer depends on several
factors—certain species (e.g. C. adamanteus) had highly con-
0.000

sistent genomic diversity estimates from localities sampled

0.00 0.02 0.04 0.06 0.08 ~200 km apart, while others (e.g. T. marcianus) had 2-fold
differences in genomic diversity between localities (Fig. 1).
mtDNA cytb pi This result indicates that using a single individual or lo-
Fig. 2. Lack of correlation between mtDNA and nuclear diversity cality to represent diversity for an entire species does not al-
estimates within species. Spearman’s rank correlation sample estimates ways provide a complete picture of diversity within species.
and P values are shown. We expected to find similar levels of genetic diversity across

Table 2. Model comparison of PGLS models under the BM or OU models of evolution.

Response Model log Likelihood AIC ΔAIC Weight

mtDNA COI π OU 57.452 −98.9 0.00 0.998

BM 50.320 −86.6 12.26 0.002
mtDNA cytb π OU 66.965 −117.9 0.00 0.997
BM 60.257 −106.5 11.42 0.003
Within-species π OU 157.485 −297.0 0.00 0.994
BM 151.435 −286.9 10.1 0.006

Table 3. Model comparison of PGLS models with different sets of predictors.

Response df log Likelihood AIC ΔAIC Weight Range size Body size Clutch size Parity

mtDNA COI π (OU models) 6 55.745 −99.5 0.00 0.129 + +

5 54.731 −99.5 0.03 0.127 +
6 55.729 −99.5 0.03 0.127 + +
mtDNA cytb π (OU models) 7 66.402 −118.8 0.00 0.379 + + +
8 66.964 −117.9 0.88 0.245 + + + +
6 64.615 −117.2 1.58 0.172 + +
Within-species π (OU models) 7 156.994 −300.0 0.00 0.342 + + +
6 155.531 −299.1 0.92 0.215 + +
8 157.310 −298.6 1.37 0.172 + + + +

The top 3 models for each response are shown, which all had a delta AIC value of <2. Note that none of the models for mtDNA COI π were substantially
improved over others (delta AIC all <4.4). The plus (“+”) sign in the predictor columns indicates that predictor was included in that model.
138 Journal of Heredity, 2023, Vol. 114, No. 2

Table 4. Model coefficients, standard errors (SE), t, and P values for the top model listed in Table 3 for each response.

Response Coefficient Value SE t P value

mtDNA COI π (OU model) Intercept −0.0565 0.0662 −0.8540 0.4043

log(Range size) 0.0062 0.0047 1.3184 0.2039
Parity_vivip −0.0165 0.0100 −1.6530 0.1157
N sequences −0.0001 0.0003 −0.2777 0.7844
mtDNA cytb π (OU model) Intercept −0.1522 0.0539 −2.8252 0.0117
log(Range size) 0.0078 0.0033 2.3328 0.0322*
log(Body size) 0.0315 0.0182 1.7320 0.1014
Parity_vivip −0.0179 0.0061 −2.9503 0.0090*
N sequences 0.0001 0.0001 0.8222 0.4223
Within-species π (OU Intercept −0.0099 0.0041 −2.4112 0.0236
model) log(Range size) 0.0010 0.0003 3.8158 0.0008*
log(Body size) −0.0022 0.0014 −1.6001 0.1221
log(Clutch size) 0.0043 0.0017 2.5383 0.0177*
N loci 0.0000 0.0000 0.0665 0.9475

*indicates P < 0.05.

individuals because we sampled them from historically stable (Marques et al. 2013). On the other hand, the GBS data consist
regions. Thus, it may seem surprising that such differences are of several thousand loci and provide a better representation of
already emerging. We predict even greater differences when genome-wide diversity within species compared with a single
populations are sampled from topographically complex re- mtDNA gene. These differences in power and geographic rep-
gions or in areas that have been influenced by historical cli- resentation, along with relatively few data points for compar-
mate change (e.g. Howes and Lougheed 2008; Nali et al. ison (22 species), may explain the lack of correlation between
2020), a consideration that should continue to be addressed mtDNA and nuclear diversity metrics in our study. Previous
in future studies. studies have found a relationship between mtDNA and nu-
Our study provides an initial comparison of squamate clear diversity in some taxonomic groups including mammals
genomic diversity from 6 families sampled in 2 species-rich (47 species; Mulligan et al. 2006) and Australian lizards (60
regions, the North American Desert Southwest and North species; Singhal et al. 2017b). Further study of this topic is
American Coastal Plain. We found considerable variation in important for comparative studies because mtDNA is readily
diversity among species within some families (e.g. Colubridae available for many more species, but reduced representation
and Natricidae; Fig. 1), but further taxon sampling is needed nuclear genome datasets are increasing and have the potential
for robust comparisons of genomic diversity across families. to provide new insights on overall species diversity.
In general, the Desert Southwest has more heterogenous
environments and landscapes compared with the Coastal Additional challenges for repurposed data
Plain, and Florida in particular, where populations were One challenge for comparative studies such as ours is the diffi-
sampled (Noss et al. 2015; Badgley et al. 2017). Thus far, we culty of gathering different data types from species complexes
did not detect any clear differences in average genomic di- that have undergone recent taxonomic revisions. When pos-
versity levels between the 2 regions (Supplementary Fig. S2), sible, we manually edited mtDNA alignments and IUCN
although several Desert Southwest species had high variation range maps to reflect data from a single described lineage
in genomic diversity among individuals and localities (Fig. 1). (Supplementary Table S1). In some cases, however, recent tax-
This result is expected given the heterogeneous nature of the onomic revisions and limited geographic information about
region, where large differences in climate, elevation, and hab- existing sequences and traits made it impossible to assign
itat on relatively small spatial scales can conceivably lead to data to newly described species. For those taxa, we excluded
differences in population sizes and standing genomic diversity mtDNA data and assumed trait information from former
among nearby localities (Myers et al. 2019). species designations would be representative for the whole
We found that mtDNA and GBS diversity were not complex. We provide the alignments and GenBank accession
correlated, suggesting that mtDNA may not be a reliable proxy numbers for the data we analyzed (https://doi.org/10.5061/
for within-species diversity in North American squamates. dryad.qfttdz0ks) such that they can be reassigned and
Several differences in our sampling for mtDNA and nuclear reanalyzed in follow-up studies. Future efforts to link geo-
data made these datasets difficult to compare, however. The graphic coordinates with genetic sequences in open-access
number of mtDNA sequences available for some species was databases will continue to improve the prospects of compar-
limited, producing COI and cytb alignments for only 22 spe- ative studies (Pelletier et al. 2022). Frequent updates to geo-
cies each. These mtDNA sequences most likely represent a graphic distribution and trait databases will also be needed to
much broader geographic area within each species compared facilitate large-scale comparative studies.
with our GBS data, though it is not possible to quantify the The approaches used to obtain and analyze nuclear
influence of geographic distance because sample localities genome-wide datasets are important to consider for compar-
were not associated with the majority of GenBank sequences ative studies of genetic diversity. One appealing characteristic
Journal of Heredity, 2023, Vol. 114, No. 2 139

Fig. 3. Predictors of mtDNA (a–d) and nuclear diversity (e–h) within species. Scatter plots and dotted lines depict simple linear relationships without
phylogenetic correction. a, b) Welch’s 2-sample t-test statistics and P values are shown. c, e, g) Spearman’s rank correlation sample estimates and P
values are shown. d, f, h) Phylogenetically corrected R2 values and P values are shown.

of mtDNA data for data repurposing and comparing species of the pipeline used for analysis (Harvey et al. 2016; Shafer
sequenced across different studies is the ease of aligning known et al. 2017). Our study used existing squamate datasets from
genes. Restriction site-associated methods such as GBS pro- the Desert Southwest that we reassembled to ensure the same
vide many more loci for analysis, but without an annotated ipyrad version and parameters were used for comparison with
reference genome for most species of interest, these loci are new GBS data we collected for the Coastal Plain species. We
anonymous. Furthermore, the locus assemblies and resulting found that heterozygosity values were highly sensitive to the
parameters are highly dependent on the parameter settings parameter settings used. Prior to reassembly, heterozygosity
140 Journal of Heredity, 2023, Vol. 114, No. 2

estimates for the Desert Southwest species were an order of most of the oviparous species included in this study belong to
magnitude smaller than those of the Coastal Plain species the family Colubridae, but viviparity in squamates has evolved
(Supplementary Table S4). These contrasting results demon- from oviparity at least 34 and perhaps more than 100 times
strate the additional challenges of comparing GBS and similar (Pyron and Burbrink 2014; Blackburn 2015). Squamates there-
datasets collected from different studies. Targeted sequence fore present an excellent opportunity to continue investigating
capture approaches may be preferable to increase compara- whether life history traits such as clutch size or parity are as-
bility across studies, though they remain more expensive and sociated with intraspecific genomic diversity while taking phy-
require more initial effort to design probe kits (Harvey et al. logeny into account.
2016; Singhal et al. 2017a; Hutter et al. 2022). Overall, we provide new insights into the predictors of diver-
sity within squamates considering both mitochondrial and nu-
clear genome patterns. We showed that geographic range size
Predictors of squamate genomic diversity corresponds with intraspecific genomic diversity even when a
Range size was positively correlated with nucleotide diver- relatively small part of the range is sampled. We also demonstrate
sity within squamate species for both mtDNA (cytb) and nu- that genomic diversity can vary widely between populations on
clear data, despite differences in geographic and locus sampling. small spatial scales, suggesting that sampling a few individuals
Previous studies have found mixed results in a variety of tax- may not adequately represent diversity levels within species.
onomic groups and with different sampling strategies. For ex- As genome-wide datasets continue being generated, compara-
ample, a global study including single-gene sequences from more tive studies of intraspecific diversity across species ranges will
than 8,000 eukaryotic species found that total range size was increase understanding of the roles of geography and life history
one of the most important predictors of genetic structure within in structuring diversity across a variety of taxonomic groups.
species (Pelletier and Carstens 2018). In contrast, (Romiguier et
al. 2014) analyzed transcriptome data from 2 to 10 individuals
of 76 animal species and found no correlation between genomic Supplementary material
diversity and range size. Detailed focus within taxonomic groups Supplementary material is available at Journal of Heredity
demonstrated correlations between range size and genetic diver- online.
sity in Drosophila (Leffler et al. 2012), Australian lizards (Singhal
et al. 2017b), and cetaceans (Vachon et al. 2018), but not in
butterflies (Mackintosh et al. 2019). To our knowledge, we pro- Funding
vide the first evidence that total range size may be a useful pre- This work was supported by The Ohio State University
dictor of squamate genomic diversity. Future studies including Undergraduate Research Scholarship awarded to IEL and by
additional species, biogeographic regions, and nuclear sampling The Ohio State University President’s Postdoctoral Scholars
across species’ ranges would provide important insights about Program through an award to LNB. Computational resources
the generality of these findings and the potential influence of were provided by the Ohio Supercomputer Center via a grant
fluctuations in population size on genomic variation. to BCC (PAA0202). LNB and BCC were supported via grants
The life history and ecological traits we included were less from the National Science Foundation (LNB: DEB-2112946
consistent predictors of mitochondrial and nuclear diversity and BCC: DBI-1910623).
in squamates. Body size has negative associations with genetic
diversity in various taxonomic groups including mammals
(Brüniche-Olsen et al. 2018), frogs (Paz et al. 2015), and Acknowledgments
insects (López-Uribe et al. 2019; Mackintosh et al. 2019). We thank Moses Michelsohn for providing tissues from
One possible explanation is that larger-bodied animals can Florida and Megan Smith for laboratory and analytical sup-
disperse over longer distances, leading to less genetic struc- port. Samples were obtained in accordance with collecting
ture and overall diversity across a species range. Our results permits from the Florida Fish and Wildlife Conservation
do not provide compelling evidence for squamates, perhaps Commission and animal care protocols approved by the
indicating that body size is not a good indicator of dispersal Florida State University Animal Care and Use Committee.
for this group. Our results for clutch size are consistent with
the prediction that species with larger clutch sizes harbor
higher genomic diversity (e.g. Romiguier et al. 2014), though Data availability
this relationship did not remain after phylogenetic correction We have deposited the primary data underlying these analyses
(Fig. 3h). These patterns will require further scrutiny with as follows:
increased species sampling in future studies.
The potential association of parity with mitochondrial diver- • Sampling locations, field numbers, museum catalogue
sity, where oviparous (egg-laying) species had higher diversity numbers, R/Python scripts, and datasets used for analyses
on average than viviparous (live-bearing) species, is intriguing and visualization: Dryad link—https://doi.org/10.5061/
and should also be investigated further with more compre- dryad.qfttdz0ks.
hensive sampling. Since the mitochondrial sequences were • DNA sequences: GenBank accession numbers
sampled across species ranges, our results may reflect greater ON911378–ON911439; NCBI SRA accession numbers
genetic structure for oviparous species, in contrast with the pre- SAMN31800422–SAMN31800468.
diction that oviparous species must disperse to find suitable
nest sites (Shine 2015). The evolution of viviparity is associ-
ated with colder climates and faster development times in cool References
temperatures (Ma et al. 2018), but our study lacked sampling Archer SR, Predick KI. Climate change and ecosystems of the south-
across broad latitudinal or climatic gradients. Furthermore, western United States. Rangelands. 2008;30:23–28.
Journal of Heredity, 2023, Vol. 114, No. 2 141

Avise JC. Phylogeography: the history and formation of species. Cam- Felsenstein J. Phylogenies and the comparative method. Am Nat.
bridge (MA): Harvard University Press; 2000. 1985;125:1–15.
Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, Neigel JE, Reeb Galtier N, Nabholz B, Glémin S, Hurst GDD. Mitochondrial DNA
CA, Saunders NC. Intraspecific phylogeography: the mitochondrial as a marker of molecular diversity: a reappraisal. Mol Ecol.
DNA bridge between population genetics and systematics. Annu 2009;18:4541–4550.
Rev Ecol Syst. 1987;18:489–522. Garrick RC, Bonatelli IAS, Hyseni C, Morales A, Pelletier TA, Perez MF,
Badgley C, Smiley TM, Terry R, Davis EB, DeSantis LRG, Fox DL, Rice E, Satler JD, Symula RE, Thomé MTC, et al. The evolution of
Hopkins SSB, Jezkova T, Matocq MD, Matzke N, et al. Biodi- phylogeographic datasets. Mol Ecol. 2015;24:1164–1171.
versity and topographic complexity: modern and geohistorical Gaston KJ, Blackburn TM, Lawton JH. Interspecific abundance-range
perspectives. Trends Ecol Evol. 2017;32:211–226. size relationships: an appraisal of mechanisms. J Anim Ecol.
Barrow LN, Masiero da Fonseca E, Thompson CEP, Carstens BC. 1997;66:579–601.
Predicting amphibian intraspecific diversity with machine learning: Grundler MR, Singhal S, Cowan MA, Rabosky DL. Is genomic diver-
challenges and prospects for integrating traits, geography, and ge- sity a useful proxy for census population size? Evidence from a
netic data. Mol Ecol Resour. 2021;21:2818–2831. species-rich community of desert lizards. Mol Ecol. 2019;1664–
Bartón K. MuMIn: Multi-Model Inference. R package version 1.46.0. 1674.
2022. [accessed 2022 November 15]. https://CRAN.R-project.org/ Harmon LJ, Weir JT, Brock CD, Glor RE, Challenger W. GEI-
package=MuMIn GER: investigating evolutionary radiations. Bioinformatics.
Bazin E, Glémin S, Galtier N. Population size does not influence mi- 2008;24:129–131.
tochondrial genetic diversity in animals. Science. 2006;312:570– Harvey MG, Smith BT, Glenn TC, Faircloth BC, Brumfield RT. Se-
572. quence capture versus restriction site associated DNA sequencing
Bivand R, Keitt T, Rowlingson B. rgdal: bindings for the “Geospa- for shallow systematics. Syst Biol. 2016;65:910–924.
tial” Data Abstraction Library. R package version 1.5-23. 2021. Hebert PDN, Cywinska A, Ball SL, deWaard JR. Biological identifications
[accessed 2022 November 15]. https://CRAN.R-project.org/ through DNA barcodes. Proc Biol Sci. 2003;270:313–321.
package=rgdal Hijmans RJ geosphere: Spherical Trigonometry. R package version 1.5-
Blackburn DG. Evolution of viviparity in squamate reptiles: reversibil- 14. 2021. [accessed 2022 November 15]. https://CRAN.R-project.
ity reconsidered. J Exp Zool B Mol Dev Evol. 2015;324:473–486. org/package=geosphere
Briggs MK, Lozano-Cavazos EA, Mills Poulos H, Ochoa-Espinoza J, Howes BJ, Lougheed SC. Genetic diversity across the range of a temper-
Rodriguez-Pineda JA. The Chihuahuan Desert: a binational con- ate lizard. J Biogeogr. 2008;35:1269–1278.
servation response to protect a global treasure. In: Goldstein MI, Hudson RR, Turelli M. Stochasticity overrules the “three-times rule”:
DellaSala DA, editors. Encyclopedia of the world’s biomes. Elsevier genetic drift, genetic draft, and coalescence times for nuclear loci
Inc; 2020. p. 126–138. versus mitochondrial DNA. Evolution. 2003;57:182–190.
Brito PH, Edwards SV. Multilocus phylogeography and phylogenetics Hutter CR, Cobb KA, Portik DM, Travers SL, Wood PL Jr, Brown RM.
using sequence-based markers. Genetica. 2009;135:439–455. FrogCap: a modular sequence capture probe-set for phylogenomics
Brüniche-Olsen A, Kellner KF, Anderson CJ, DeWoody JA. Runs of ho- and population genetics for all frogs, assessed across multiple phy-
mozygosity have utility in mammalian conservation and evolution- logenetic scales. Mol Ecol Resour. 2022;22:1100–1119.
ary studies. Conserv Genet. 2018;19:1295–1307. Jenkins CN, Houtan KSV, Pimm SL, Sexton JO. US protected lands mis-
Brüniche-Olsen A, Kellner KF, DeWoody JA. Island area, body size and match biodiversity priorities. PNAS. 2015;112:5081–5086.
demographic history shape genomic diversity in Darwin’s finches Jombart T. adegenet: a R package for the multivariate analysis of ge-
and related tanagers. Mol Ecol. 2019;28:4914–4925. netic markers. Bioinformatics. 2008;24:1403–1405.
Burbrink FT, Myers EA. Both traits and phylogenetic history influence Jombart T, Ahmed I. adegenet 1.3-1: new tools for the analysis of
community structure in snakes over steep environmental gradients. genome-wide SNP data. Bioinformatics. 2011;27:3070–3071.
Ecography. 2015;38:1036–1048. Karney CFF. Algorithms for geodesics. J Geod. 2013;87:43–55.
Carling MD, Brumfield RT. Gene sampling strategies for multi-locus Katoh K, Standley DM. MAFFT multiple sequence alignment software
population estimates of genetic diversity. PLoS One. 2007;2:e160. version 7: improvements in performance and usability. Mol Biol
Castañeda MR, de Queiroz K. Phylogenetic relationships of the Evol. 2013;30:772–780.
Dactyloa clade of Anolis lizards based on nuclear and mitochon- Kimura M. The neutral theory of molecular evolution. Sci Am.
drial DNA sequence data. Mol Phylogenet Evol. 2011;61:784–800. 1979;241:98–100, 102, 108 passim.
Chen J, Glémin S, Lascoux M. Genetic diversity and the efficacy of Laikre L. Genetic diversity is overlooked in international conservation
purifying selection across plant and animal species. Mol Biol Evol. policy implementation. Conserv Genet. 2010;11:349–354.
2017;34:1417–1428. Laikre L, Hoban S, Bruford MW, Segelbacher G, Allendorf FW,
Collin R. The effects of mode of development on phylogeography and Gajardo G, Rodríguez AG, Hedrick PW, Heuertz M, Hohenlohe
population structure of North Atlantic Crepidula (Gastropoda: PA, et al. Post-2020 goals overlook genetic diversity. Science.
Calyptraeidae). Mol Ecol. 2001;10:2249–2262. 2020;367:1083–1085.
Cornetti L, Ficetola GF, Hoban S, Vernesi C. Genetic and ecological Lee HJ, Boulding EG. Spatial and temporal population genetic struc-
data reveal species boundaries between viviparous and oviparous ture of four northeastern Pacific littorinid gastropods: the effect of
lizard lineages. Heredity. 2015;115:517–526. mode of larval development on variation at one mitochondrial and
Eaton DAR, Overcast I. ipyrad: interactive assembly and analysis of two nuclear DNA markers. Mol Ecol. 2009;18:2165–2184.
RADseq datasets. Bioinformatics. 2020;36:2592–2594. Leffler EM, Bullaughey K, Matute DR, Meyer WK, Ségurel L, Venkat
Edwards SV, Beerli P. Perspective: gene divergence, population diver- A, Andolfatto P, Przeworski M. Revisiting an old riddle: what
gence, and the variance in coalescence time in phylogeographic determines genetic diversity levels within species? PLoS Biol.
studies. Evolution. 2000;54:1839–1854. 2012;10:e1001388.
Edwards SV, Robin VV, Ferrand N, Moritz C. The evolution of compar- Leigh DM, van Rees CB, Millette KL, Breed MF, Schmidt C,
ative phylogeography: putting the geography (and more) into com- Bertola LD, Hand BK, Hunter ME, Jensen EL, Kershaw F, et al.
parative population genomics. Genome Biol Evol. 2022;14:evab176. Opportunities and challenges of macrogenetic studies. Nat Rev
Ellegren H, Galtier N. Determinants of genetic diversity. Nat Rev Genet. 2021;22:791–807.
Genet. 2016;17:422–433. Leroy T, Rousselle M, Tilak M-K, Caizergues AE, Scornavacca C,
Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, Recuerda M, Fuchs J, Illera JC, De Swardt DH, Blanco G, et al.
Mitchell SE. A robust, simple genotyping-by-sequencing (GBS) ap- Island songbirds as windows into evolution in small populations.
proach for high diversity species. PLoS One. 2011;6:e19379. Curr Biol. 2021;31:1303–1310.e4.
142 Journal of Heredity, 2023, Vol. 114, No. 2

López-Uribe MM, Jha S, Soro A. A trait-based approach to predict pop- Pinheiro J, Bates D, DebRoy S, Sarkar D; R Core Team. nlme: Linear
ulation genetic structure in bees. Mol Ecol. 2019;28:1919–1929. and Nonlinear Mixed Effects Models. R package version 3.1-155.
Ma L, Buckley LB, Huey RB, Du W-G, Pincheira-Donoso D. A global 2022. [accessed 2022 November 15]. https://CRAN.R-project.org/
test of the cold-climate hypothesis for the evolution of viviparity of package=nlme
squamate reptiles. Glob Ecol Biogeogr. 2018;27:679–689. Powell R, Conant R, Collins JT. Peterson field guide to reptiles and
Mackintosh A, Laetsch DR, Hayward A, Charlesworth B, Waterfall M, amphibians of eastern and central North America. 4th ed. Boston
Vila R, Lohse K. The determinants of genetic diversity in butterflies. (MA): Houghton Mifflin Harcourt; 2016.
Nat Commun. 2019;10:1–9. Pyron RA, Burbrink FT. Early origin of viviparity and multiple
Marques AC, Maronna MM, Collins AG. Putting GenBank data on the reversions to oviparity in squamate reptiles. Ecol Lett. 2014;17:13–
map. Science. 2013;341:1341. 21.
Martins EP, Hansen TF. Phylogenies and the comparative method: a Revell LJ. phytools: an R package for phylogenetic comparative biol-
general approach to incorporating phylogenetic information into ogy (and other things). Methods Ecol Evol. 2012;3:217–223.
the analysis of interspecific data. Am Nat. 1997;149:646–667. Rohland N, Reich D. Cost-effective, high-throughput DNA sequencing
Miraldo A, Li S, Borregaard MK, Flórez-Rodríguez A, Gopalakrishnan libraries for multiplexed target capture. Genome Res. 2012;22:939–
S, Rizvanovic M, Wang Z, Rahbek C, Marske KA, Nogués- 946.
Bravo D. An Anthropocene map of genetic diversity. Science. Romiguier J, Gayral P, Ballenghien M, Bernard A, Cahais V, Chenuil
2016;353:1532–1535. A, Chiari Y, Dernat R, Duret L, Faivre N, et al. Comparative pop-
Morin PA, Luikart G, Wayne RK; the SNP workshop group. SNPs in ecol- ulation genomics in animals uncovers the determinants of genetic
ogy, evolution and conservation. Trends Ecol Evol. 2004;19:208–216. diversity. Nature. 2014;515:261–263.
Mulligan CJ, Kitchen A, Miyamoto MM. Comment on “Population Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-
size does not influence mitochondrial genetic diversity in animals”. terminating inhibitors. Proc Natl Acad Sci USA. 1977;74:5463–
Science. 2006;314:1390. 5467.
Myers EA. Genome-wide data reveal extensive gene flow during the Schield DR, Adams RH, Card DC, Corbin AB, Jezkova T, Hales NR,
diversification of the western rattlesnakes (Viperidae: Crotalinae: Meik JM, Perry BW, Spencer CL, Smith LL, et al. Cryptic genetic
Crotalus). Mol Phylogenet Evol. 2021;165:107313. diversity, population structure, and gene flow in the Mojave rattle-
Myers EA, McKelvy AD, Burbrink FT. Biogeographic barriers, Pleisto- snake (Crotalus scutulatus). Mol Phylogenet Evol. 2018;127:669–
cene refugia, and climatic gradients in the southeastern Nearctic 681.
drive diversification in cornsnakes (Pantherophis guttatus com- Sexton JP, Hangartner SB, Hoffmann AA. Genetic isolation by envi-
plex). Mol Ecol. 2020;29:797–811. ronment or distance: which pattern of gene flow is most common?
Myers EA, Xue AT, Gehara M, Cox CL, Davis Rabosky AR, Lemos- Evolution. 2014;68:1–15.
Espinal J, Martínez-Gómez JE, Burbrink FT. Environmental het- Shafer ABA, Peart CR, Tusso S, Maayan I, Brelsford A, Wheat CW,
erogeneity and not vicariant biogeographic barriers generate Wolf JBW. Bioinformatic processing of RAD-seq data dramatically
community-wide population structure in desert-adapted snakes. impacts downstream population genetic inference. Methods Ecol
Mol Ecol. 2019;28:4535–4548. Evol. 2017;8:907–917.
Nali RC, Becker CG, Zamudio KR, Prado CPA. Topography, more than Shine R. The evolution of oviparity in squamate reptiles: an adapta-
land cover, explains genetic diversity in a Neotropical savanna tree tionist perspective. J Exp Zool B Mol Dev Evol. 2015;324:487–
frog. Divers Distrib. 2020;26:1798–1812. 492.
Nazareno AG, Bemmels JB, Dick CW, Lohmann LG. Minimum sample Singhal S, Grundler M, Colli G, Rabosky DL. Squamate Conserved
sizes for population genomics: an empirical study from an Amazo- Loci (SqCL): a unified set of conserved loci for phylogenomics
nian plant species. Mol Ecol Resour. 2017;17:1136–1147. and population genetics of squamate reptiles. Mol Ecol Resour.
Nei M. Molecular evolutionary genetics. New York: Columbia Univer- 2017a:17:e12–e24.
sity Press; 1987. Singhal S, Huang H, Title PO, Donnellan SC, Holmes I, Rabosky DL.
Noss RF, Platt WJ, Sorrie BA, Weakley AS, Means DB, Costanza J, Genetic diversity is largely unpredictable but scales with museum
Peet RK. How global biodiversity hotspots may go unrecognized: occurrences in a species-rich clade of Australian lizards. Proc R Soc
lessons from the North American Coastal Plain. Divers Distrib. B Biol Sci. 2017b:284:20162588.
2015;21:236–244. Soltis DE, Morris AB, McLachlan JS, Manos PS, Soltis PS. Comparative
Novosolov M, Rodda GH, North AC, Butchart SHM, Tallowin OJS, phylogeography of unglaciated eastern North America. Mol Ecol.
Gainsbury AM, Meiri S. Population density–range size relationship 2006;15:4261–4293.
revisited. Glob Ecol Biogeogr. 2017;26:1088–1097. Tonini JFR, Beard KH, Ferreira RB, Jetz W, Pyron RA. Fully-sampled
Olson DM, Dinerstein E. The Global 200: a representation approach phylogenies of squamates reveal evolutionary patterns in threat
to conserving the earth’s most biologically valuable ecoregions. status. Biol Conserv. 2016;204:23–31.
Conserv Biol. 1998;12:502–515. Vachon F, Whitehead H, Frasier TR. What factors shape genetic diver-
Paradis E. pegas: an R package for population genetics with an sity in cetaceans? Ecol Evol. 2018;8:1554–1572.
integrated-modular approach. Bioinformatics. 2010;26:419–420. Waltari E, Hijmans RJ, Peterson a T, Nyári AS, Perkins SL, Guralnick
Paradis E, Schliep K. ape 5.0: an environment for modern phylogenetics RP. Locating pleistocene refugia: comparing phylogeographic and
and evolutionary analyses in R. Bioinformatics. 2019;35:526–528. ecological niche model predictions. PLoS One. 2007;2:e563.
Paz A, Ibáñez R, Lips KR, Crawford AJ. Testing the role of ecology and Weinell JL, Austin CC. Refugia and speciation in North American scar-
life history in structuring genetic variation across a landscape: a trait- let snakes (Cemophora). J Herpetol. 2017;51:161–171.
based phylogeographic approach. Mol Ecol. 2015;24:3723–3737. White EP, Ernest SKM, Kerkhoff AJ, Enquist BJ. Relationships be-
Pelletier TA, Carstens BC. Geographical range size and latitude pre- tween body size and abundance in ecology. Trends Ecol Evol.
dict population genetic structure in a global survey. Biol Lett. 2007;22:323–330.
2018;14:20170566. Williams JW, Shuman BN, Webb T, Bartlein PJ, Leduc PL. Late-
Pelletier TA, Parsons DJ, Decker SK, Crouch S, Franz E, Ohrstrom J, Quaternary vegetation dynamics in north America: scaling from
Carstens BC. phylogatR: phylogeographic data aggregation and taxa to biomes. Ecol Monogr. 2004;74:309–334.
repurposing. Mol Ecol Resour. 2022;22:2830–2842. Wright S. Isolation by distance. Genetics. 1943;28:114–138.
Pfeifer B, Wittelsbuerger U, Ramos-Onsins SE, Lercher MJ. PopGenome: Zink RM, Barrowclough GF. Mitochondrial DNA under siege in a-
an efficient Swiss army knife for population genomic analyses in R. vian phylogeography: nuDNA vs. mtDNA in phylogeography. Mol
Mol Biol Evol. 2014;31:1929–1936. Ecol. 2008;17:2107–2121.