bs_bs_banner Zoological Journal of the Linnean Society, 2015. With 4 figures Using geometric morphometrics for integrative taxonomy: an examination of head shapes of milksnakes (genus Lampropeltis) SARA RUANE* Department of Biology, College of Staten Island/CUNY Graduate Center, 2800 Victory Blvd., Staten Island, NY 10314 Received 6 November 2014; revised 13 January 2015; accepted for publication 14 January 2015 Species discovery and identification has long relied on traditional morphometric analyses, although molecular methods for species delimitation are becoming increasing popular and important. Despite an increase in studies that rely solely on molecular data to differentiate between species, additional evidence that supports genealogically-based species delimitation is desirable at least for field and museum identification of species and is part of an integrative approach to taxonomy. The present study uses geometric morphometric (GM) analyses to examine six species of milksnake (genus Lampropeltis) that have recently been delimited based on multilocus data in a coalescent frame- work. Landmarks are plotted onto the dorsal view of 487 specimens and canonical variate analysis (CVA) is used to determine whether the differences in head shape of these six species can be used to correctly classify specimens. For five of the six species, CVA accurately classifies individuals >70% of the time. The present study illustrates that, although GM-based analyses may not correctly differentiate between species 100% of the time, GM methods can be useful for detecting shape differences between species and help to corroborate species delimitation. © 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015 doi: 10.1111/zoj.12245 ADDITIONAL KEYWORDS: morphology – species delimitation – triangulum. Morphological analyses have long been the stand- of informative or discretely different characters or have ard for the discovery and description of species. For elevated intraspecific variation, traditional morpho- many species, morphological characters provide infor- logical methods may not be able to detect differences mation pertaining to taxonomic identity and evolu- and alternative approaches are necessary. One such tionary relationships among taxa. However, for some approach is geometric morphometrics (GM), which com- species, the number of morphological characters may prises a collection of shape-analysis techniques that be limited or may not be useful with respect to assess the relative spatial distribution of a set of pre- phylogenetic inference; relying solely on morphology determined landmarks, such as points where the sutures to define species has long been recognized as prob- of a skull come into contact with one another; the re- lematic (Mayr, 1942). In particular, within species com- sulting set of coordinates represent a shape that is scaled plexes or among cryptic and pseudocryptic taxa (Saez to be independent of size and is typically analyzed using & Lozano, 2005), molecular methods are often used multivariate statistics (Zelditch, Swiderski & Sheets, to disentangle phylogeny (Bickford et al., 2007). That 2012). These GM methods have been used to de- does not mean morphology is unable to provide addi- scribe variation in many taxonomic groups, including tional insights but, for taxa that have a limited number turtles (Claude et al., 2003), lizards (Stayton, 2005; Kaliontzopoulou, Carretero & Llorente, 2007; Leaché et al., 2009), and mammals (Cardini et al., 2009). Geo- metric morphometric-based analyses have recently in- *Current address: Department of Herpetology, American Museum of Natural History, 200 Central Park West, creased in popularity for herpetological studies New York, NY 10024, USA. E-mail: [email protected] specifically, with studies using GM methods to examine © 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015 1 2 S. RUANE differences related to sexual dimoprhism, allometry, and Gentilli et al., 2009), although GM techniques show taxonomy (Kaliontzopoulou, 2011). A GM approach has promise for adding support to molecular taxonomic also been found to be more powerful compared to tra- hypotheses. A study on European vipers found that ditional morphological analyses based on linear mensural subspecies that were distinct clades based on a mo- data for discriminating between species and popula- lecular phylogeny were also distinct morphologically tions of ‘morphologically ambiguous’ taxa (e.g. moths: (Gentilli et al., 2009). Mutanen & Pretorius, 2007; bats: Evin et al., 2008; Geometric morphometrics may be particularly useful cichlids: Maderbacher et al., 2008). for groups or complexes where morphology has been Snakes have a limited number of categorical mor- misleading. One such group comprises the milksnakes phological traits as a result of their lack of append- (genus Lampropeltis).Until recently, milksnakes have ages and generalized elongate body form. Accordingly, been been considered a single taxon (formerly GM could be useful for capturing shape variation with L. triangulum; Lacépède, 1788) with 25 subspecies respect to head shape, which, for snakes, may be closely ranging from south-eastern Canada to Ecuador. These tied with feeding ecology (Lillywhite & Henderson, subspecies designations were based almost entirely on 1993; Shine et al., 2002; Vincent, Herrel & Irschik, colour pattern (Williams, 1988). Colour patterns within 2004). Additionally, there are already a few GM- the former milksnake subspecies are highly variable based studies that have identified significant differ- and not always diagnostic if locality information is un- ences in head shape between sexes and subspecies of available for an individual (Williams, 1988). Subse- snake (Vincent et al., 2004; Gentilli et al., 2009), al- quent molecular studies using multiple loci and a variety though studies on snakes using GM are rare of methods have shown that milksnakes from differ- (Kaliontzopoulou, 2011). Morphological features that ent geographical locations are not monophyletic within are more commonly used in snake systematics include Lampropeltis (Bryson et al., 2007; Pyron & Burbrink, colour pattern and scale count. Both of these charac- 2009a; Ruane et al., 2014). The most recent molecu- ters may be misleading with respect to taxonomy and lar study on milksnakes (Ruane et al., 2014), which evolutionary history. Colour pattern is often found to defined species using the general lineage species concept be variable within species and thus is not a reliable (De Queiroz, 2007) and is followed in the present study, indicator of evolutionary relationships (Burbrink, Lawson determined that there are seven species that form three & Slowinski, 2000), whereas scale counts can be in- distinct clades within Lampropeltis: these are fluenced by both biotic (e.g. diet; Fabien et al., 2004) Lampropeltis abnorma (Bocourt, 1886), Lampropeltis and abiotic (e.g. temperature; Fox, 1948) factors through- annulata Kennicot 1861, Lampropeltis elapsoides out the range of a species. This lack of suitable char- (Holbrook, 1838), Lampropeltis gentilis (Baird & Girard, acters is also confounded by the many species of snakes 1853), Lampropeltis micropholis Cope 1861, Lampropeltis containing cryptic diversity, as revealed by the number polyzona Cope 1861, and Lampropeltis triangulum of phylogeographical studies demonstrating the exist- (Lacepédè, 1788). The present study examines six of ence of numerous independently evolving lineages within these taxa that have specimens readily available within wide-ranging taxa (e.g. Agkistrodon contortrix: Guiher a GM framework to determine whether there are dis- & Burbrink, 2008; Lampropeltis pyromelana: Burbrink tinct head shape differences between the species. Con- et al., 2011; Lampropeltis triangulum: Ruane et al., sidering the disparity in diet and habitat use among 2014). Not only have phylogeographical studies un- milksnakes (Williams, 1988; Werler & Dixon, 2000; covered this previously unknown diversity, but also Ernst & Ernst, 2003), head shape may be a good start- new multilocus coalescent-based methods are able to ing point for determining whether morphological dif- further delineate between these lineages as distinct ferences exist among these species, as well as potentially species (e.g. L. pyromelana: Burbrink et al., 2011; providing integrative taxonomic support to the mo- L. triangulum: Ruane et al., 2014; Lampropeltis zonata: lecular species delimitation for these snakes Myers et al., 2013), although there has been some con- tention regarding classification based solely on DNA (Bauer et al., 2011). Molecular data alone are suffi- MATERIAL AND METHODS cient evidence of speciation under the general species concept (De Queiroz, 2007). Nevertheless, determin- SAMPLES ing whether there are identifiable morphological fea- In the present study, the dorsal view of the head of tures inherent to a species is desirable and may provide 487 individuals representing six species of milksnake additional information with respect to ecological factors, was photographed (Fig. 1; see also Appendix, Table A1). such as diet (Lillywhite & Henderson, 1993; Shine Although seven species have been recently elevated et al., 2002; Vincent et al., 2004). There is a paucity (Ruane et al., 2014), it was not possible to obtain samples of studies that have used GM to examine shape vari- from Mexico that were definitively L. annulata and so ation in snakes (Manier, 2004; Vincent et al., 2004; this species