Lamella and Scansor Numbers in Thecadactylus Rapicauda (Gekkonidae): Patterns Revealed Through Correlational Analysis and Implic

Short Notes

Lamellaand scansor numbers in Thecadactylus rapicauda (Gekkonidae): Patternsrevealed through correlational analysis and implicationsfor systematicand functional studies

PhilipJ. Bergmann,Anthony P .Russell

Vertebrate MorphologyResearch Group,Department ofBiologicalSciences, Universityof Calgary,2500 UniversityDr. N.W .,Calgary,Alberta, Canada T2N1N4 e-mail: [email protected],[email protected]

Manytaxa of lizardsbear transversely enlarged subdigital scales that are often a source ofsystematic characters. Among gekkonid and polychrotid lizards, elaborations of such scales,called lamellae and scansors, are employed in adhesivelocomotion (Colette, 1961; Hiller,1968; Russell, 1975). Subdigital lamellae and scansors are derivatives of scales, the form, shapeand number of whichare thought to re ect function (Russell, 1981) as well asindicatepatterns of relationship(Couper et al. ; 1993 Raxworthyand Nussbaum, 1993; I Grismer etal. ; 1994/.Whilelamellae in the loose sense encompass all subdigital scales thatare transversely enlarged, they can be divided into two subsets ( g. 1): scansors, thedistal lamellae most closely involved in adhesion and therefore occurring on the hyperextensibleportion of adigit(Russell, 1981, 1986); and basal lamellae, which differ morphologicallyfrom scansors(Russell, 1981), and are situated more proximally. Scale patternsin squamatesare generally regarded as being genetically determined and discrete (Hecht,1952). Counts of total lamellae feature in both systematic (Grismer etal. ; 1994, etc.)and functional (Hecht, 1952; Peterson, 1983a, b; Mayerin Losos, 1990; Glossip and Losos,1997) studies. Numbersof totallamellae vary both within and between digits and, as a consequence, mostauthors choose to countthese structures on asmallerset of digits,on theassumption

Figure 1. Ventralview of digitIV ,left pes of Thecadactylusrapicauda ,showingthe location of thetotal lamella numberand its subsets: scansors andbasal lamellae, theformer beingpositioned beneath the hyperextensible regionof the digit.

thatthese digits will be representativeor maximallyinformative (e.g. Digit IV :Broadley, 1977;King, 1962. Digits I, III, andIV :Otaet al., 1995). Although such selectivity is pragmaticand potentially justi able, we areunaware of any studies that have formally addressedthe questionsof whichdigits are maximally informative, and which are the most appropriatefor aparticulartaxon. As boththe manus and pes of lizardsgenerally exhibit asymmetryof digitallength, digits three and four are often chosen for totallamella counts (Hecht,1952; Colette, 1961). This has been justi ed because, being the largest, these digitsare assumed to be most load-bearing (Hecht, 1952), and the easiest to distinguish andcount (Colette, 1961). However, asymmetry of the length of digits varies markedly acrosstaxa (Russell et al., 1997), even when phalangeal counts are identical. Given that discrepancyexists between the differential length of digitsin different taxa, the question ofhowrepresentative or informativeselected total lamella counts are of a particulartaxon remains,and should be assessed before a selectionis made of which data to record and employ. Short Notes 381

Hereinwe investigatethe variation within and between total lamella, scansor, and basal lamella( g. 1)countson manualand pedal digits of thegekkonid Thecadactylusrapicauda asan exemplar .Stemmingfrom this,we outlinea methodby whichmaximally informative datafor systematicand functional studies might be selected. W eexaminethe question ofwhether there are high levels of correlation between total lamella, scansor, and basal lamellacounts on all digits of the manus and pes and explore ways to determine, on asmallersubsample, which of these counts are of most interest in the context of the question(systematic, functional, or both)at hand.W ealsoaddress the question of whether correlationalpatterns exhibited by scansor and basal lamella counts mirror thoseof total lamellacounts, or ifthesetwo component counts exhibit independent patterns. Byundertakingsuch a correlationalstudy of theseiterative homologues, we investigate whetherdifferent digits provide new or merely repeated information, and which, if any, digitsare most representative (being most highly correlated) of the other digits on the limbin question. Both highly representative and poorly representative counts form valid charactersfor systematicstudies and focal points for functionalconsiderations. The inclusionof theformer isjusti ed asaproxyfor theother digits, while the inclusion of the latteris justi ed as anindicator of theuniqueness of aparticulardigit, which may be of particularimportance in functional studies. Totallamella number( g. 1) was recordedfor all digits of the right manus and pes of40 specimens of Thecadactylusrapicauda fromacross itsbroad range (see Appendix1 fora listof specimens andtheir localities). Asample size of40was selected arbitrarilyas atrade-offbetween time involvedin data collectionand a sufcient sample. We suggestthat twenty would be a minimumsample size forthis methodology. Additionally, the scansor numberwas recordedby countingthose enlarged plates thatoccur only beneath the hyperextensible portion of each digit(Russell, 1981). This was achievedby re ecting the distal portion of each digitwith an entomological pinto determine thepoint of in ection at whichhyperextension occurs (Russell and Bels, 2001).Due tothe intimateinvolvement of scansors inadhesion, they may bemore constrainedby selection than the more extensive totallamellar series (as countedby Hecht, 1952). Finally, basal lamella numberwas calculated bysubtracting scansor numberfrom the total lamella count( g. 1).These twogroups of ‘subset’counts are includedhere as potentiallyuseful, albeit neglected (but see Russell,1981; Russell and Bels, 2001),characters forsystematic and functionalstudies. Theresulting data (table1 forsummary) were tested fornormality using the Kolmogorov-Smirnov test and were foundto be highlynon-normal for total lamella, scansor,and basal lamella counts.For this reason, and due totheirdiscontinuous nature, Spearman correlations(Sokal and Rohlf, 1995) were usedin the analysis. Six sets ofcorrelationswere conducted:one for each combinationof total lamella, scansor,and basal lamella countsfor thefore and hindlimb. For each digitin each set, Spearman correlationcoef cients withthe remaining four digits were averaged.This resulted in a singleAverage Spearman CorrelationCoef cient for each digit.Although this doesnot represent arigorousstatistic, we believethat it does give a goodidea ofhowrepresentative a countis of othersuch counts, and is easier tointerpret than a correlationof a particulardigital count with the total respective countfor the entire limb (which would eliminate theuse ofaverages). These average Spearman’s correlation coefcients were thenused to determine howrepresentative a particularcount was ofall othercounts in a set (withoutaccounting for differential digital lengths). AverageSpearman correlation coef cients for totallamella, scansor, and basal lamella countsof fore andhindlimbs follow several patterns ( g. 2).Similar general patterns are evidentfor bothtotal lamella and scansor counts for theforelimb, with digit I beingleast representativeof theothers (see Russell and Bauer, 1990: 464-466 for possiblereasons), 382 Short Notes

Table 1. Summaryof statistics forcounts of totallamellae, scansors,and basal lamellae onall digitsof themanus andpes for40 specimens of Thecadactylusrapicauda .Allvalues are mean standarddeviation. § Digit I II III IV V

Totallamellae 15 :7 2:0 18:6 1:6 20:1 1:7 20:6 1:6 18:8 2:2 § § § § § ManusScansors 8:3 1:3 11:0 1:1 11:5 1:0 11:8 0:9 10:9 0:9 § § § § § Basal lamellae 7 :4 1:5 7:7 1:2 8:6 1:3 8:7 1:4 7:9 1:7 § § § § § Totallamellae 17 :6 2:0 19:7 1:9 21:1 1:9 21:6 1:8 20:3 1:4 § § § § § Pes Scansors 9:5 1:3 11:6 0:8 12:1 1:1 12:3 0:9 11:5 0:9 § § § § § Basal lamellae 8 :3 2:2 8:1 1:5 9:1 1:4 9:3 1:5 8:9 1:0 § § § § §

Figure 2. Average Spearman correlationcoef cients foreach digitwith all otherdigits for total lamella (a, d), scansor (b,e), andbasal lamella (c, f) countsfor fore (a, b,c) andhindlimbs (d, e, f) for Thecadactylusrapicauda (n 40). D digitsII, III andIV beingincreasingly representative of theother digits, and digit V being lessrepresentative ( g. 2a,b). A similarpattern is revealed for thetotal lamella counts for thehindlimb ( g. 2d),except that digit V istheleast representative. The pattern changes, however,when the scansor counts for thehindlimb and basal lamella counts for bothlimbs areconsidered ( g. 2e,c, f, respectively).For hindlimbscansor counts, digit II isthe least representativewhile digit V isthe most representative. When basal lamella counts are analyzed,digit III ismostrepresentative on both limbs ( g. 2c,f). Theleast representative digitsfor frontand hindlimbs are I andV (g. 2c, f), respectively,corresponding to the leastrepresentative digits for totallamella counts ( g. 2a,d). Short Notes 383

Theseresults suggest that for Thecadactylusrapicauda ,atotallamella count for the forelimbwould best be representedby digitsI andIV ,whilefor thehindlimb it would be bestrepresented by digitsIV andV .Representativescansor counts for theforelimb would mirror thesituation for thetotal lamella counts of that limb, but the hindlimb would be mosteffectively represented by countsof scansorson digitsII andV .Basallamella counts wouldbe bestrepresented by digits I andIII for theforelimb and by digits III andV for thehindlimb. These ndingsalso demonstrate that scansor and basal lamella counts contain different informationfrom totallamella counts, validating their systematic utility. Since scansors are mostclosely involved with adhesion and their counts follow different patterns than those ofbasal lamellae, from afunctionalperspective they may be ofgreaterinterest than basal lamellae.Furthermore, since phalangeal length differs withindigits (Russell and Bauer, 1990;Russell et al.,1997), it isnotsurprising that scansor and basal lamella counts also containindependent information. Themorphology and relative degree of development (large versus reduced) of digit I ishighly variable amongst geckos in general (Russell and Bauer, 1990), which suggests thattotal lamella counts in general will be highlyvariable both within and between taxa. Incertain taxa there may be major differences in patterns of expression of this digit betweenthe fore andhindlimbs (even in the same species). Thus, any attempt to include totallamella, scansor or basallamella counts of digitI insystematicanalyses should take accountof thehigh degree of variabilityof this digit across taxa. Hecht(1952) and Colette(1961) both pointed out thegenerally held contention that digit IV isthelongest and may therefore be representativeof totallamella counts and patterns for theentire manus and pes. While this prediction holds true for totallamella counts and scansorcounts of themanus of Thecadactylusrapicauda ,andtotal lamella counts of the pes,it isnotre ected in scansorcounts of thepes orbasallamella counts of boththe manus andpes. Pedal (vs. manual) geometry in geckos(Russell et al.,1997) might contribute to thisdiscrepancy, with digit IV ofthe pes being generally shorter than digit III (Russell etal., 1997: g.2a,b). Proportions of thebasal (non-scansor-bearing) to distal (scansor- bearing)portions of digit IV amonggeckos vary considerably (Russell et al., 1997: 784 — Rhoptropus)andmay result in unexpected patterns of totallamella, scansor and basal lamellacounts for thatdigit. Thedata presented for Thecadactylusrapicauda andthe comparative data available for pedalproportions and structure in geckos (Russell and Bauer, 1990; Russell et al., 1997)illustrate that the most appropriate characters for aparticulartaxon to beincluded ina systematic,functional or ecologicalstudy may not necessarily be themost intuitive. Suchdata further indicate that total lamella counts (which include all transversely widened subdigitalscales), and scansor and basal lamella counts (which represent subsets of the totallamella count) do not necessarily exhibit similar patterns. Thus, the distinction betweenscansors and basal lamellae (which functionally are restricted to the basal part 384 Short Notes ofthe digit in many taxa) is real, and each type of modi ed scale may be subjected to differentselective pressures. For bothfore andhindlimbs, average correlation coef cients are greater overall for total lamellacounts than for scansorcounts or basallamella counts. The range for alllamella countsis comparable between limbs. Autocorrelation does not appear to be a prevalent issue,given that the highest average correlation in the analysis is 0.727 ( g. 2d, digit IV;thelowest is 0.257 — g.2f, digit V). Thereis considerable inherent variation, and anappropriate choice of characters to investigate may well yield more discriminatory informationthan a defaultchoice of atotallamella count for digitIV onmanus and pes (for example). Determiningwhich morphological characters are most appropriate to includein a study canbe timeconsuming, and data collection can be arduousand lengthy. Our correlational approachprovides a potentialavenue for exploringhow to maximize the utility of the datacollected by asking some initial questions about a smallersubset of the entire setof organisms available for examination.The initial effort involved in running the correlationalanalysis may well circumvent time spent in gathering data that provide no furtherdiscriminatory or explanatory power. This approach is applicable to an array of comparablevariables where serially repeated counts are involved. Care should be taken insuchcorrelational analysis to avoidor accountfor thepotential confounding effects of ontogenyand selection. For example,Hecht (1952) noted that total lamella number was subjectto selection in the gekkonid Aristelliger ,withthe count range differing between subadultand adult individuals. Unfortunately, Hecht (1952) did not distinguish between basallamellae and scansors, so the subtleties of such differences cannot be dissected further.

Acknowledgements. We thankLarry Powell and Larry Linton for discussions of statistical matters. Two anonymousreviewers improvedthe presentation of themanuscript. We are alsograteful to the following people andinstitutions for making specimens availablefor examination: Jens Vindum,California Academy ofSciences (CAS);Alan Resetar, FieldMuseum of Natural History(FMNH); Linda Trueb, Kansas University(KU); José Rosado,Museum of Comparative Zoology (MCZ); Laurie Vitt and Amy Estep,Oklahoma Museum of Natural History;and Greg Schneider,University of Michigan Museum of Zoology(UMMZ).

References

Broadley,D.G. (1977): A review ofthe northeastern forms ofthe Pachydactyluscapensis complex(Sauria: Gekkonidae).Arnoldia 8: 1-19. Colette,B.B. (1961): Correlations between ecologyand morphology in anoline lizards fromHavana, Cuba and southernFlorida. Bulletin of the Museum of ComparativeZoology 125: 137-161. Couper,P .J.,Covacevich, J.A., Moritz, C. (1993):A review ofthe leaf-tailed geckosendemic toeastern Australia: Anew genus,four new species, andother new data.Memoirs of theQueensland Museum 34: 95-124. Glossip,D., Losos,J.B. (1997): Ecological correlates ofnumberof subdigital lamellae inanoles. Herpetologica 53: 192-199. Short Notes 385

Grismer, L.L.,Ota, H.,Tanaka,S. (1994):Phylogeny, classi cation, and biogeography of Goniurosaurus kuroiwae (Squamata:Eublepharidae) from the Ryukyu Archipelago, Japan, with description of a new subspecies.Zoological Science 11: 319-335. Hecht,M.K. (1952): Natural selectionin the lizard genus Aristelliger . Evolution 6: 112-124. Hiller,U. (1968):Untersuchungen zum Feinbauund zur Funktion der Haftborsten von Reptilien. Z. Morph.Tiere 62: 307-362. King,W. (1962): Systematics ofLesser Antilleanlizards ofthe genus Sphaerodactylus .Bulletinof the Florida State Museum 7: 1-52. Leviton,A.E., Gibbs Jr., R.H., Heal, E.,Dawson,C.E. (1985): Standards in herpetology and ichthyology: Part I. Standardsymbolic codes for institutional resource collectionsin herpetology and ichthyology. Copeia 1985: 802-832. Losos,J.B. (1990): Ecomorphology, performance capability,and scaling of West Indian Anolis lizards: An evolutionaryanalysis. Ecological Monographs 60: 369-388. Ota, H.,Lau,M.W., Weidenhofer, T., Y asukawa, Y.,Bogadek,A. (1995):Taxonomic review ofthe geckos allied to Gekko chinensis Gray 1842(Gekkonidae Reptilia) from China and Vietnam. TropicalZoology 8: 181-196. Peterson,J.A. (1983a): The evolution of the subdigital pad in Anolis.1.Comparisonsamong anoline genera. In: Advancesin Herpetology and Evolutionary Biology, p. 245-283. Rhodin, A.G.J., Miyata, K., Eds, Cambridge, Massachusetts, Museumof ComparativeZoology. Peterson,J.A. (1983b): The evolution of the subdigital pad in Anolis.2.Comparisons among the iguanid genera related tothe anolines and a viewfrom outside the radiation. J. Herpetol. 17: 371-397. Raxworthy,C.J., Nussbaum, R.A. (1993): A new Madagascan Phelsuma,witha review of Phelsumatrilineata andcomments on Phelsumacepediana inMadagascar (Squamata:Gekkonidae). Herpetologica 49: 342-349. Russell,A.P .(1975):A contributionto the functional analysis of thefoot of the Tokay, Gekko gecko (Reptilia: Gekkonidae).J. Zool.,London 176: 437-476. Russell,A.P .(1981):Descriptive and functional anatomy of thedigital vascular system oftheTokay, Gekko gecko. J. Morphol. 169: 293-323. Russell,A.P .(1986):The morphological basis ofweight-bearing in the scansors ofthe tokay gecko (Reptilia: Sauria).Can. J. Zool. 64: 948-955. Russell,A.P .,Bauer,A.M. (1990): Digit I inpad-bearing gekkonine geckos: Alternate designsand the potential constraintsof phalangealnumber. Memoirs of the Queensland Museum 29: 453-472. Russell,A.P .,Bauer,A.M., Laroiya, R. (1997):Morphological correlates ofthe secondarily symmetrical pes of gekkotanlizards. J. Zool.,London 241: 767-790. Russell,A.P .,Bels, V.(2001):Digital hyperextension in Anolissagrei .Herpetologica 57: 58-65. Sokal,R.R., Rohlf, F.J. (1995): Biometry (3 rd ed.).New York,W.H. Freeman andCo.

Appendix 1.

Museums,catalogue numbers, and collection localities for specimens used.Abbreviations follow Leviton et al. (1985),except Oklahoma Museum of Natural History. CAS: 8767— St.Croix; 13263 — Ecuador,Pichincha; 15816, 15817 — Ecuador,Napo-Pastaza. FMNH: 45449,109825 — Peru,Loreto; 49109, 49112, 49114 — Mexico,Y ucatan;168128, 228257 — Peru, Madrede Dios. KU: 96489,96490 — Panama, Bocas delToro; 130197 — Brasil,Para; 194933,204963, 204964, 204965, 207765,215009, 220185, 220186 — Peru,Madre de Dios; 220485, 222360 — Peru,Loreto; 229881 — Dominica. MCZ: 4744— Brasil,Amazonas; 60816,60818, 60819 — Dominica;81220, 171650 — Guyana,Mazaruni- Potaro. OklahomaMuseum of NaturalHistory: 36427,36434 — Ecuador,Sucumbios; 36753 — Brazil, Pará; 37333, 37334— Brazil, Rondônia. UMMZ: 80812,80813, 83284 — Mexico,Y ucatan;83325 — Dominica.

Received: August27, 2002. Accepted: December 2,2002.