Spatial variation in colour morph, spotting and ... - Springer Link

Genetica (2006) 128:51–62 DOI 10.1007/s10709-005-5367-7


Springer 2006

Spatial variation in colour morph, spotting and allozyme frequencies in the candy-stripe spider, Enoplognatha ovata (Theridiidae) on two Swedish archipelagos G.S. Oxford1 & B. Gunnarsson2 1

Department of Biology (Area 18), University of York, P. O. Box 373, YO10 5YW, York, UK (Phone: +441904-328640; Fax: +44-1904-328505; E-mail: [email protected]); 2Department of Applied Environmental Science, University of Go¨teborg, Box 464, S-405 30, Go¨teborg, Sweden

Received 17 February 2005 Accepted 22 November 2005

Key words: local selection, morph identification, null model, population differentiation, thermal selection

Abstract The selective significance, if any, of many invertebrate visible polymorphisms is still not fully understood. Here we examine colour- and black spotting-morph frequencies in the spider Enoplognatha ovata in populations on two Swedish archipelagos with respect to different spatial scales and, in one archipelago, against the background of variation at four putative neutral allozyme marker loci. Every population studied was polymorphic for colour and 28 out of 30 contained all three colour morphs – lineata, redimita and ovata. We found no evidence for a breakdown in the traditional colour morph designation previously suggested for other northern European populations of this species. For colour there is no significant heterogeneity at spatial scales greater than between local populations within islands. Black spotting frequencies show a similar lack of pattern over larger spatial scales except that there are significant differences between the Stockholm and Go¨teborg archipelagos. Measures of population differentiation (h) within the Stockholm islands for the two visible systems show them to be significantly more differentiated than the neutral markers, suggesting local selection acting on them in a population-specific manner. On the basis of previous observations and the distribution of spotting phenotypes on a European scale, it is argued that thermal selection might operate on black spotting during the juvenile stages favouring more spots in continental climates. It is not clear what selective forces act on colour.

Introduction Although the colours and patterns exhibited by some invertebrates may have relatively obvious adaptive functions, for instance, crypsis, camouflage, mimicry and aposematism, in other cases the selective advantage of genetically-determined, visible variation is far from clear. This is true for many polymorphic spiders (Holl, 1987; Oxford & Gillespie, 1998). For example, the adaptive significance of the exuberant visible polymorphism in the endemic Hawaiian happy-face spider, Theridion grallator (Theridiidae) (Oxford & Gillespie,

1996a, b, c) has not been fully established although tantalising indications of the presence, and mode of action, of selective forces are beginning to emerge (Gillespie & Oxford, 1998). Another case of extensive polymorphic variation awaiting an explanation involves the candy-stripe spider, Enoplognatha ovata (Clerck) (Theridiidae). The widespread distribution and abundance of this species throughout the western Palaearctic and in North America, and its rather simple visible variation, has resulted in a plethora of studies of its population genetics dating back to the 1930s (e.g., Bristowe, 1931; Hippa & Oksala, 1979; Reillo &

52 Wise, 1988a; Oxford & Reillo, 1993; Oxford, 2005 and references therein). Three principal colour morphs are recognised: in lineata the opisthosoma is plain yellow; redimita has, in addition, a pair of dorso-lateral carmine stripes; while ovata has the entire dorsal surface of the opisthosoma carmine. These morphs are controlled by three alleles at an autosomal locus (Oxford, 1983) and in mature females form a dominance hierarchy, with ovata dominant to both redimita and lineata, and redimita dominant to lineata. Despite the effort spent in studying this species, the factor(s) responsible for the maintenance of the polymorphism and for determining morph frequencies in local populations are still poorly understood (but see Oxford, 2005). There is circumstantial evidence, for example from the ubiquity of the polymorphism (including its presence in a sister species, E. latimana Hippa & Oksala) and a very consistent rank-order of colour morphs within populations, that the polymorphism is ultimately maintained in some way (Oxford & Reillo, 1993; Oxford, 2005). Here we examine phenotypic variation in samples of E. ovata from two archipelagos off the Swedish coast. Our primary aim was to apply an indirect method to determine whether selection acts on the visible variation. To do this we compare the degree of population differentiation at the colour locus with that deduced from variation at other, putatively neutral, loci. Island systems are ideal for this purpose in that they hold discrete populations with limited gene exchange between them. Our general method has been used in the past for both polygenic characters and polymorphic loci in order to identify selection acting between subpopulations (e.g., Prout & Barker, 1993; Mithen, Raybould & Giamoustaris, 1995), and to reveal how selection operates differently on different character sets (e.g., Long & Singh, 1995; Taylor, Shen & Kreitman, 1995). Recently the approach has been applied, for the first time, to an investigation of a colour polymorphism in a spider (Gillespie & Oxford, 1998). Population differentiation for neutral genetic markers is determined by the balance between genetic drift within populations and migration between them (Hartl & Clark, 1997). As both processes should affect all neutral loci equally, the degree of genetic differentiation among populations should, on average, be the same at all of these marker loci. If the extent of

population differentiation at a colour locus is not significantly different from that at the marker loci, within sampling error, then the colour alleles can also be regarded as being neutral. However, if the degree of differentiation is significantly different from that indicated by the neutral markers, selection can be inferred to be acting over and above the background processes of migration and drift. The direction of differences between the two genetic systems provides information about the sort of selection that may be operating. If interpopulational differentiation is greater for colour than for the neutral markers, this would imply that local selection is acting on the colour locus in different directions in different populations. Conversely, if differentiation for colour is less than that for neutral markers this would suggest global selection for rather similar colour-allele frequencies among populations. We applied this general method to material from one of the island archipelagos. Allozyme electrophoresis was used to generate allele frequencies from which estimates of the degree of genetic differentiation among populations were derived. These estimates were then compared with those obtained from variation in two visibly polymorphic systems, controlling the colour polymorphism and the presence or absence of dorso-lateral black spots (Oxford, 1989). A secondary aim was to examine visible polymorphisms in populations close to the species’ probable northern limit in Europe (Oxford & Reillo, 1994). The distributions of colour and black spotting phenotypes from Swedish populations are compared with those in Britain and from further south in continental Europe. The populations studied are also of considerable interest with respect to the nature of the colour variation. Earlier work, primarily in Finland but including material collected at Gustavsberg, Stockholm, Sweden (Hippa & Oksala, 1979), had suggested that the traditional division of the colour variation into three distinct morphs is artificial and that in mature females there is ‘a gradating series of forms from narrow redimita type to distinct ovata type’ (Hippa & Oksala, 1979: 205). This conclusion was important in shaping the model they developed to explain the genetics of colour variation in E. ovata and its regulation during development (Hippa & Oksala, 1979). Observations of populations elsewhere in Europe (e.g., Oxford, 1985; Oxford &

53 Reillo, 1993) and in North America (e.g., Reillo & Wise, 1988a; Reillo, 1989) have not supported this view and we wished to investigate whether there was indeed a breakdown in the polymorphism in northern European populations.

Materials and methods Spiders were collected from islands in the Stockholm (samples 1–17) and Go¨teborg (samples 18–30) archipelagos (Figures 1 and 2) between 5 and 15 August, 1997. They were sought in roadside patches of broad-leaved vegetation and samples were obtained within as short a linear distance as possible (up to c. 30 m), depending on density. In August mature females are established within rolled leaves; searching for leaves rather than spiders directly ensured that sampling was random with respect to visible morphs. Sampled leaves were sought in adjacent stretches of vegetation by the two authors and collected into separate bags. After a sufficient number had been gathered the first c.50 animals drawn equally from both bags

Figure 2. Map of the Go¨teborg archipelago showing sampling sites: for island names see Table 1. The approximate boundary of Go¨teborg is shown in darker shading. For analyses of geographical trends (see text), islands were grouped as follows: (18, 19); (20, 21, 22); (23, 24); (25, 26, 27); (28, 29, 30). See inset to Figure 1 for the location of the study area (G).

Figure 1. Map of the Stockholm archipelago showing sampling sites: for island names see Table 1. The approximate boundary of Stockholm is shown in darker shading. For analyses of geographical trends (see text), islands were grouped as follows: (1, 2, 3); (5, 6, 7, 8); (10, 11); (12, 13); (14, 15, 16, 17). The inset shown the locations of the more detailed Stockholm (S) and Go¨teborg (G) maps (Figures 1 and 2, respectively).

were placed live into individual Eppendorf tubes for subsequent electrophoresis. The lids of the tubes were perforated for ventilation and the spiders placed in a cool box. The remaining individuals were preserved in 70% ethanol and sorted into separate tubes according to colour morph within 24 h (Oxford & Reillo, 1993). Live spiders were stored under cool conditions through the fieldwork period, transported to the University of York in a cool box and preserved at )80C until required.

54 Alcohol-preserved spiders from each site were subsequently scored for colour and the number of black spots. Allozyme frequencies were determined using cellulose acetate electrophoresis (Richardson, Baverstock & Adams, 1986). Individual animals were homogenised in two volumes of 50 mM Tris buffer (pH 8.0), centrifuged, and the supernatant applied to Titan III cellulose acetate sheets (Helena Laboratories). Appropriate running buffers were taken from Richardson, Baverstock and Adams (1986) and stain recipes and procedure adapted from those of Hebert and Beaton (1993). Initially 30 enzyme systems were tested on E. ovata material from Yorkshire, England, of which 12 were found to give strong staining on gels. Of these, nine (probably representing 13 loci) produced consistently resolvable bands: aspartate amino transferase [AAT – 2 loci], a-amylase [AMY], glucose-6-phosphate isomerase [GPI], glyceraldehyde-3-phosphate dehydrogenase [G3PDH – 2 loci], a-glycerol-3-phosphate dehydrogenase [aGPDH], isocitrate dehydrogenase [IDH, 2 loci], lactate dehydrogenase [LDH], malic enzyme [ME], and phosphoglucomutase [PGM, 2 loci]. Three systems (AAT, ME and PGM-2), however, were not easily interpretable and five loci proved to be monomorphic across the Swedish populations (G3PDH-1, G3PDH-2, aGPDH, IDH-1 and, IDH-2). Thus, only four loci (AMY, GPI, LDH and PGM-1) were polymorphic and informative with respect to determining genetic structure. For all enzyme systems, extracts from between 24 and 36 individuals from each of 14 sites were run at 200 V in Tris–Glycine buffer (pH 8.5) for 30–40 min. The genetic structure of populations, as revealed by the allozyme data, was investigated using the FSTAT v.2.9.3 (Goudet, 1995) and TFPGA v.1.3 (Miller, 1997) programs. Estimates of FST (h) for the visible morph loci were calculated according to Weir (1996: 177–179), assuming Hardy–Weinberg equilibria. Expected numbers in some of the R  C contingency tables comparing numbers of visible morphs between populations were too small to allow conventional chi-squared analyses. In these cases unbiased estimates of the exact probabilities were calculated for the tables using the Metropolis algorithm, as implemented in the R  C program of Miller (1997).

Results Visible variation Colour morph was scored for all individuals (alcohol-preserved and frozen) whereas spotting was only recorded in alcohol-preserved material. The scoring of individuals for the presence and number of black spots requires microscopic examination and as the frequency of spotting was extremely high in alcohol-preserved specimens from the Stockholm archipelago, it was decided not to risk enzyme denaturation by scoring this character in frozen material prior to homogenisation. Data on these two polymorphic systems are summarised in Table 1. Contingency v2 tests were used to determine whether colour-morph frequencies were homogeneous between alcohol-preserved and frozen samples from the Stockholm archipelago. This was so in all 14 cases after appropriate Bonferroni correction. Table 1 shows that all populations were polymorphic for at least the lineata and redimita colour morphs and that in 28 out of 30 samples the ovata morph was also present. One of the samples in which the ovata morph was absent was relatively small (site 14, n = 20), but the other was large (site 10, n = 74). In no instance was there any difficulty in determining whether a spider was of the redimita or ovata morph i.e., no intermediates were encountered. Frequencies of the three colour morphs varied markedly from sample to sample (for n ‡ 20, Stockholm archipelago, v2(28) = 163.6, p