1995. The Journal of Arachnology 23:60–64

A TEST OF THE CENTRAL-MARGINAL MODEL USING SAND POPULATION S (PAR MESAENSIS, VAEJOVIDAE)

Tsunemi Yamashita' and Gary A. Polis: Department of Biology, Box 1812 Station B, Vanderbilt University, Nashville, Tennessee 37235 USA

ABSTRACT. The central-marginal model proposes marginal populations contain a lower density of individ- uals, lower levels of genetic variation, and are more isolated than populations in the center of a species range . Previous tests of the model used Drosophila, organisms capable of extended dispersal . We test the central - marginal model with , organisms with restricted dispersal abilities . We measured genetic variation through allozyme analysis of eight loci (five polymorphic, three rnonomorphic) to obtain estimates of hetero- zygosity. We compared differences between the two types of populations with a split-plot ANOVA . We also compared central and marginal populations using standard parametric tests. We found marginal populations contain lower genetic variation than central populations . These populations may be important as models in conservation to study the effects of fragmentation .

Peripheral or marginal populations are thos e tion that allozyrne heterozygosity is reduced at on the boundaries of a species' geographic range. range margins is not valid (Brossard 1984). How- They exhibit unique properties not evident in ever, Drosophila probably disperse relatively great populations in the center (Brossard 1984) . Gen- distances compared to most non-flying taxa. Non- erally, as one moves outward from the center o f flying taxa (vertebrates) show a decline in alleli c a species' range, populations are hypothesized to diversity in marginal populations (Soule' 1973) . become less dense, more isolated, and less vari- We propose scorpions, unable to disperse great able genetically within populations (da Cunha et distances, also may illustrate the predictions of al. 1959; da Cunha & Dobzhansky 1954; Carson Lewontin's (1974) central-marginal model . 1959; Soule 1973 ; Brussard 1984) . These trends are embodied in the central-marginal model (Le- wontin 1974). Several explanations exist . One METHODS hypothesis (da Cunha & Dobzhansky 1954) states Relevant scorpion biology.—The scorpion Pa- genetic polymorphism is positively correlate d ruroctonus mesaensis Stahnke is restricted to sand with the number of niches an organism occupie s dunes and sandy substrates scattered throughout — more niches are available at the center of a the southwestern United States and northern species range. Yet Brussard's (1984) free recom- Mexico. It occurs in dense populations (range = bination hypothesis proposes linkage disequilib- 1600—5000/ha) (Polis & Yamashita 1991) and it rium is favored in central populations because is an ecologically important species, as a gener- extreme phenotypes are selected against, i. e., alist predator in desert food webs (Polis 1979; stabilizing selection occurs. Here, favored gene s Polis & McCormick 1986, 1987) . This scorpio n are linked together to create a stable phenotype is dispersal limited because specialized morpho- with maximal fitness. In theory, at the range mar- logical features adapt and restrict it to sand. The gins, linkage equilibrium is selected in stressfu l species possess numerous modified setae on their environments to create novel phenotypes better tarsi (sand shoes) to facilitate sand movemen t able to survive sub-optimal conditions (Brossard and burrow construction (Polis et al . 1986). Fur- 1984). ther, their ability to detect substrate vibration s Research with Drosophila suggests the predic- to localize prey only functions well on sand (Brownell & Farley 1979). These specializations reduce the likelihood of extended migration. 'Present address: Department of Biology, Northeas t However, reproductive males move extensively Louisiana University, Monroe, Louisiana 7120 9 when searching for mates (Polis & Farley 1979 , USA 1980). 60

YAMASHITA POLIS—SAND SCORPION CENTRAL-MARGINAL POPULATIONS 6 1

Collection sites : 0 4 0 Central Populations: Marginal Populations: A. Glamis G. Windy Point Kilometers B.Yuma Desert H. Kelso C. Yuma Army I. Needles D. Ogilby J. Bouse E.N Fontana K.Phoenix F.Glamis East L.Agua Caliente Figure 1 .—A partial map of the desert areas of California, Arizona, and Baja California Norte where population s of mesaensis were collected . A proposed range of P. mesaensis is indicated by the dashed line .

Collection and electrophoresis procedures.— and other pertinent methodology are described We collected the scorpions over three years, 19 89- in Yamashita (1993). 1991 . Paruroctonus mesaensis is easy to collect Central and marginal population determina- as it (and all scorpions) fluoresces when illumi- tion. —We determined the geographical center nated with a ultraviolet light (Sylvania F8T5 / of P. mesaensis populations from collection data BLB). We transported the scorpions alive to the (Haradon 1983; D. Gaffin pers . comm.; Yama- laboratory, froze them with liquid nitrogen, an d shita 1993) (Fig. 1). We determined the following stored them at — 70 °C until electrophoretic anal - range extremes: in the north (Death Valley, Cal- ysis. ifornia); in the south, Cabo Lobos (Sonora, Mex - We electrophoresed five polymorphic and three ico); in the east (Phoenix, Arizona); and in the monomorphic loci from 28 populations . This re- west (Windy Point in the Coachella Valley, Cal- search was the first allozyme analysis of any scor - ifornia). We determined the range center to be pion populations, and these eight loci were th e located 65 km north of Yuma, Arizona by lo- only resolvable loci from a screening of 25 en- cating the midpoint between the extreme north- zyme loci on 11 different buffer systems . We as- ern and southern and the western and eastern sayed a mean of 29 individuals (± 5 .7) from each populations . We designated six population s population for each locus. The specific protocols within a 32 km radius of the range center as

62 THE JOURNAL OF ARACHNOLOGY

Central populations: Glamis, Yuma desert, Yum a Table 1 .-Split plot ANOVA analysis of two pop- Army, Ogilby, N. Fontana, and Glamis East. We ulation categories (Central and Marginal). This design delineated peripheral populations as those near- is taken after Weir 1990 . est to edges of the range. These include Windy Point, Kelso, Needles, Bouse, Phoenix, and Agua df MS F-value P-value Caliente. Category 1 3.54 21 .60 5_0.00 1 Statistical analyses.—We performed two type s Individuals within of analyses to determine if differences betwee n categories 222 0.166 1.01 >_0 .437 central and marginal populations exist . The first, Loci 4 4.29 26.17 5_0.001 5_0 a method outlined by Weir (1990), tested if het- Category x loci 4 1 .29 7.86 .00 1 Error 888 0.164 erozygosity differences exist between populatio n types. This design, similar to a split-plot ANO- VA, considered variation from five sources : pop- ulations, individuals within populations, loci, loci from the central and marginal populations wer e by populations, and loci by individuals within significantly different (Tables 1, 2). The results populations. Heterozygotes are entered as 1's, of the split-plot ANOVA (Table 1) show value s homozygotes, as 0's and the data from each cen- from all levels of analysis (populations ; individ- tral or marginal populations were pooled . We uals within populations ; loci; population x loci) used electrophoretic data from 20 individual s were significant (P 0 .001) except individuals from each of six central and six marginal pop- within populations (P 0.437). These results ulations (120 individuals in each population type) establish that central populations are signifi- and five loci . We used Weir's analysis because cantly different from marginal ones in hetero- estimates of heterozygosity often exhibit large zygosity. Furthermore, the significant among loci interlocus variances and non-normal distribu- effects suggests that each locus expressed a dif- tions; therefore, many standard parametric tests ferent pattern of heterozygosity from other loci. may be inappropriate (Archie 1985) . Loci within marginal populations were signifi- Second, we performed standard parametric cantly different from loci within central popu- statistics to determine differences between cen- lations, which suggests within each population tral and marginal populations. The variables ex- type (central or marginal), the same locus ex- amined were observed average heterozygosity , pressed significantly different heterozygosities. mean allele number, and percent polymorphis m The mean genetic variability (observed het- of each population. We calculated these variable s erozygosity) in marginal populations (0 .106 ± using BIOSYS-1 (Swofford & Selander 1989) . 0.025, n = 6) was significantly less than central Observed heterozygosity per locus is the fraction populations (0 .164 ± 0.018, n = 6; t = 3 .80, 0.05 of heterozygous individuals from a given sample • P > 0.01) supporting the central-marginal for a particular locus (Ferguson 1980; Weir 1990). model. The mean allele number for the central Observed average heterozygosity is the mean populations (1 .73 ± 0.144) was marginally great - value from all loci. Although heterozygosity val- er (t = 2.21, 0.1 >_ P > 0.05) than that of mar- ues commonly undergo an arcsine transforma- ginal populations (1 .52 ± 0.095). The mean per- tion, our data did not require such a procedure cent polymorphism for central populations (45 .83 because most heterozygosity values fell betwee n ± 6 .45) was also marginally greater (35 .42 ± 0.30 and 0 .70, a range that does not require trans- 9.41, t = 2.71, 0.05 >_ P> 0.01). formation (Sokal & Rohlf 1981) . We used two other indices of genetic vari- DISCUSSION ability. Percent polymorphism is the mean num - The significant differences between central and ber of polymorphic loci in a population. Here, a marginal populations for allozyme heterozygos- locus is polymorphic if the frequency of the mos t ity, mean allele number, and percent polymor- common allele is 0 .95 or less. The mean number phism are consistent with the central-margina l of alleles per locus is the number of alleles at model (Brussard 1984). Paruroctonus mesaensis each locus averaged across all loci. is one of the species that fits the predictions o f this model ; tests of the model using Drosophila RESULTS allozymes failed to exhibit similar pattern s The split-plot ANOVA and standard para- (Brussard 1984). In our study, the decrease in metric tests showed the mean heterozygosities genetic variability in marginal populations prob-

YAMASHITA & POLIS—SAND SCORPION CENTRAL-MARGINAL POPULATIONS 63

Table 2.-A comparison of genetic variability between central and marginal populations . See text for discus- sion.

Mean Mean allele heterozygosity number % Polymorphism Central populations Glamis 0.153 1 .75 37.5 Yuma Desert 0.176 2.00 50.0 Yuma Army 0.136 1 .63 50.0 Ogilby 0.167 1 .63 37 .5 N. Fontana 0.166 1 .63 50.0 Glamis East 0.187 1 .75 50.0 Mean 0.164 1 .73 45.8 SD 0.01 8 0.144 6. 5 Marginal populations Windy Point 0.103 1 .50 25.0 Phoenix 0.078 1 .38 25.0 Bouse 0.13 1 1 .63 50.0 Agua Caliente 0.140 1.50 37.5 Kelso 0.084 1.63 37.5 Needles 0 .097 1 .50 37.5 Mean 0.106 1 .52 35.4 SD 0.025 0.095 9 .4 Central vs margina l t-statistic 3 .80 2.21 2.7 1 P values 0 .05 P > 0 .01 0.1 >_P> 0 .05 0.05 P> 0.01

ably stems from reduced gene flow or smalle r al. No P . mesaensis were observed in > 30 hours overall population size. Because scorpion dis- of searching on rocky habitats adjacent to sandy persal is local, populations at the range margin areas (Polis, unpubl. data). are less likely to receive migrants from other pop- The Phoenix population is the most eastern ulations compared to more central populations . and one of the most genetically depauperate pop- Central populations exhibit the highest allel e ulations . Emigration into this area probably oc- number and percent polymorphism . These pop- curred along the dry river beds of the Salt and ulations may maintain higher genetic variability Gila rivers. This population relies on unidirec- because exchange with other nearby populations tional gene flow since the substrate outside the is more frequent and population size is generally river bed is a dispersal barrier to the psammo- larger in the center of the range. However, mod- philic scorpion and no populations exist to th e els suggest that a very small effective population east. The low values of genetic variability ma y size (ne < 10 individuals) is required to reduce be a result of two primary factors: it is a periph - significantly the number of alleles per locus with- eral population with a low population size an d in a population (Nei et al . 1975 ; Rice & Mack receives little gene flow from other populations . 1991). Peripheral isolate formation may have bee n Some marginal populations (Needles, Bouse, enhanced by large scale floods in the Holocene and Phoenix) are geographically isolated fro m (Ely et al. 1993). The Salt and Gila rivers ex- other populations . Needles, north of a mountain perienced large-scale floods in the last 5600 years range present on either side of the Colorado Riv- (Ely et al . 1993). Periods of minor and majo r er, is effectively isolated . The Bouse population floods were interspersed . These extreme floods exists on the eastern edge of the Cactus Plain, a may have created expansion corridors for scor- large sandy region in western Arizona . It is sur- pion movement and isolated populations b y rounded by rocky habitat and isolated from th e fragmenting previous habitat . nearest population by 40 km . Although Bouse is Although several studies have compared cen- not separated by a large distance, the interme- tral and marginal populations, a clear trend is diate rocky substrate effectively curtails dispers - not evident (Hoffman & Parsons 1991). Some

64 THE JOURNAL OF ARACHNOLOGY

report no decrease in genetic variability within Lewontin, R . C. 1974. The genetic basis of evolu- marginal populations, e. g., Drosophila allo- tionary change. Columbia Univ. Press. New York , zymes (Brussard 1984) and an annual grass (Ric e New York . & Mack 1991). Analysis of dispersal limited an- Nei, M, T. Maruyama, & R. Chakraborty . 1975. The bottleneck effect and genetic variability in popula- imals (frogs) report a decrease in genetic vari- tions. Evolution, 29:1-10. ability among marginal populations (Sjogren Polis, G. A. 1979. Prey and feeding phenology of th e 1991). Further research into the properties of desert sand scorpion Paruroctonus mesaensis (Scor- marginal populations is warranted because mar- pionidae, Vaejovidae). J. Zool., London, 188 :333- ginal populations, with their insular or penin- 346. sular properties, are similar to populations frag- Polis, G. A. & R. D. Farley. 1979. Characteristics mented through man's encroachment upon th e and environmental determinants of natality, growth , environment . and maturity in a natural population of the desert scorpion, Paruroctonus = mesaensis (Scorpionidae : ACKNOWLEDGMENTS Vaejovidae). J. Zool., London, 187 :517-542. Polis, G. A. & R . D. Farley. 1980. Population biology We are grateful to M r . Hedin and M . Rosen fo of a desert sand scorpion: Survivorship, microhab- field collection . We are also grateful to C . Bas- itat, and the evolution of life history strategy . Ecol- kauf, D. McCauley, and W . Eickmeier for dis- ogy, 61:620-629 . cussion and comments. Financial support was Polis, G. A. & S. J. McCormick. 1986. Patterns of provided by Vanderbilt University graduat e resource use and age structure among a guild of travel and research awards . desert scorpions. J. Ecol ., 55:59-73. Polis, G . A. & S. J. McCormick. 1987. Competitio n LITERATURE CITED and predation among species of desert scorpions. Archie, J.W. 1985. Statistical analysis of heterozy- Ecology, 68:332-343 . gosity data: Independent sample comparisons. Evo- Polis, G. A., C. A. Myers & M. A. Quinlan . 1986. lution, 39 :623-637. Burrowing biology and spatial distribution of desert Brownell, P . H. & R. D. Farley. 1979. Prey-localizing sand scorpions. J. Arid Environ ., 10:137-146 . behaviour of the nocturnal desert sand scorpion, Polis, G. A. & T. Yamashita. 1991 . The ecology and Paruroctonus mesaensis: Orientation to substrate importance of predaceous in desert com - vibrations . Anim. Behay., 27:185-193 . munities. Pp. 180-222. In The Ecology of Desert Brussard, P. F. 1984. Geographic patterns and en- Communities . (G.A. Polis, ed.) Univ. of Arizona vironmental gradients: The central-marginal mode l Press . Tucson, Arizona . in Drosophila revisited. Ann. Rev. Ecol. Syst., 15: Rice, K. J. & R . N. Mack . 1991 . Ecological genetics 25-64. of Bromus tectorum . I. A hierarchical analysis o f Carson, H. L. 1959. Genetic conditions that promote phenotypic variation . Oecologia, 88 :77-83. or retard the formation of species . Cold Spring Har- Sjogren, P. 1991 . Genetic variation in relation to bor Symp. Quant . Biol., 24:87-103 . demography of peripheral pool frog population s da Cunha, A. B., & T. Dobzhansky. 1954. A further (Rana lessonae) . Evol. Ecol., 5:248-271 . study of chromosomal polymorphism in Drosophila Sokal, R. R. & F. J. Rohlf. 1981 . Biometry, 2nd ed. willistoni in relation to environment. Evolution , W. H. Freeman & Co., San Francisco . 8:119-134 . Soule', M. 1973. The epistasis cycle: A theory o f da Cunha, A. B., T. Dobzhansky, O . Pavlovsky, & B . marginal populations. Ann. Rev. Ecol. Syst., 4:165- Spassky. 1959. Genetics of natural populations . 187. XXVIII. Supplementary data on the chromosomal Swofford, D . L. & R . B. Selander . 1989. BIOSYS-1 : polymorphism in Drosophila willistoni in relation A computer program for the analysis of allelic vari - to the environment . Evolution, 13:389-404 . ation in population genetics and biochemical sys- Ely, L. L, Y. Enzel, V. R. Baker, & D. R. Cayan. 1993. tematics. User's Manual . Illinois Nat. Hist. Survey. A 5000-year record of extreme floods and climati c Weir, B. S. 1990. Genetic Data Analysis . Sinauer, change in the southwestern United States. Science , Sunderland, Massachusetts. 262:410-412 . Yamashita, T . 1993. Genetic and Morphological Ferguson, A . 1980. Biochemical Systematics an d Variation in Scorpion Populations on Habitat Is - Evolution . Wiley & Sons, New York, New York . lands. Ph. D. dissertation . Vanderbilt University , Haradon, R. M. 1983 . Smeringurus, a new species Nashville, Tennessee . of Paruroctonus Werner (Scorpionies, Vaejovidae). J. Arachnol ., 11 :251-270 . Manuscript received 5 January 1995, revised 21 March Hoffman, A. A. & P. A. Parsons. 1991. Evolutionary 1995. ' Genetics and Environmental Stress . Oxford Univ . Press, Oxford .