Contributions to Zoology, 71 (4) (2002)
Genetic and ecological differentiation between the butterfly sisterspecies Colias alfacariensis and Colias hyale
Daniel F.R. Cleary , Henri Descimon , Steph B.J. Menken
Keywords: Lepidoptera, Pieridae, allozymes, population structure, gene flow
Clear habitat separation between the sister species Colias alfacariensis and C. hyale is shown when occurring sympatrically. Colias hyale is found more often in moist cultivated pastures while Colias alfacariensis is more abundant in dry uncultivated habitat. Out of a total of 16 loci, no diagnostic loci were found between C. alfacariensis and C. hyale, and both species shared most major polymorphisms. Exceptions were the marked differences in allele frequencies at the HK locus and only C. hyale, but not C. alfacariensis was further invariable at the GOT2 locus, which is usually highly polymorphic in the Pieridae. Colias hyale has a significantly lower level of heterozygosity than its sister species C. alfacariensis. In Colias alfacariensis heterozygosity is highest in the Alps and lowest in the low-lying region of Northern France. Both species show high levels of gene flow over a large geographic area. Within C. alfacariensis, but not in C. hyale, the FST value of the PGI locus is significantly different from zero effectively separating the species into populations with high levels of the ’b’ allele to the west and North, and low levels of the allele in the Alps and Italy. This could point to selection within the PGI locus in line with the well established pattern of selection at the PGI locus in other species of Colias. Glaciations have been an important force in shaping the evolutionary history of European biota, leading to extinction, but also allowing new species to evolve into the newly available land as the ice sheets retreated. The genetic and distributional pattern found between both Colias species suggests that habitat shifts and subsequent adaptation during glaciations could have played an important role in their speciation.
Coexistence of species requires the existence of evolutionarily continual interspecific trade-offs in the abilities of species to deal with factors that limit their fitness and abundance. These include trade-offs between competitive abilities and dispersal abilities, susceptibility to disease, herbivory or predation, between abilities to live off average conditions or abilities to exploit resource pulses, and between abilities to compete for alternative resources in a heterogeneous landscape (Tilman, 2000). Feeding ecology, reproductive behaviour, and physiology are characteristics of a species niche. Niches tend to be conservative over time scales associated with sister taxon pairs in, for example, birds, mammals, and butterflies but this conservatism breaks down at the level of families (Peterson et al., 1999). By analysing the present day difference in genetic makeup and the often subtle differences in ecology between species within a pair of sister species we can get an idea of the evolutionary forces that led to their differentiation into separate species.
The present study concerns two species of the genus Colias [Lepidoptera: Pieridae] namely C. alfacariensis and C. hyale (Brunton, 1998). Colias is one of the most widespread butterfly genera in the world. It is present on all continents except Australia, and ranges from the Antarctic (e.g., Colias antarctica) to the Arctic (e.g., Colias boothi), across Asia, Africa, North and South America, from the lowlands to the highest mountains.
Colias alfacariensis and C. hyale are sister species (Brunton, 1998), and are difficult to tell apart due to the large variation in wing patterns and other morphological characters of the butterflies. In general, C. hyale has a more northerly and C. alfacariensis a more southerly distribution with a large region of overlap in central Europe. In recent years C. hyale has become increasingly rare in northern Europe due to abandonment in agricultural areas of its larval food plant Medicago sativa (Descimon, (1994(5)).
Colias alfacariensis flies in May-June and August-September in two, and in southern Europe sometimes in three generations. Food plants are Hippocrepis comosa and Coronilla varia. Colias hyale has similar flight periods and the food plants are Medicago sativa, Vicia cracca, Coronilla spp., Lotus spp., Cytisus spp., Trifolium spp., and other Fabaceae (Higgins and Riley, 1970). Both species are thus oligophagous on Fabaceae although C. hyale has a larger range of potential food plants.
The goal of this study is to analyse the ecology and genetical population structure of both species and to try to gain insight into the differentiation process that produced both present-day taxa. Lepidoptera are an important source for understanding the genetics of speciation (Menken et al., 1992; Ritchie and Phillips, 1998). The analysis of spatial pattern (e.g., relative densities and habitat preferences), direct measures of migration (e.g., Mark-release-recapture), and indirect measures (e.g., genetic differentiation) are all important for understanding how populations are structured. Lewis et al. (1997; and refs therein) suggest that at least two of these methods should be applied in order to provide a good indication of population structure.
The specific questions we address are the following. 1) Do both species differ in their habitat preference? 2) What is the spatial, genetical population structure of both species?
Materials and methods
Individuals of both C. alfacariensis and C. hyale were sampled for an allozyme study in France and Italy in the summer of 1992. Species determination was based on the studies of Berger and Fontaine (1947), Jarvis (1955), and Dutreix (1980). Habitat preference between C. alfacariensis and C. hyale was studied by means of field observations in the French Alps where both species occur sympatrically.
A total of 920 adult C. alfacariensis and 1155 adult C. hyale were observed from the 14th of August until the 8th of September, 1974 by HD. The location and geographical position of all sites is presented in Table 1. During the field observations the following environmental variables were noted at each locality: 1. Aspect, 2. Vegetation (cultivated – uncultivated), 3. Humidity (dry – irrigated), 4. Altitude 5. Date, 6. Weather (warm – cold), in addition to the frequency of occurrence of adults of C. alfacariensis and C. hyale.
The relative abundances of both species were regressed onto the first axis of a principal components analysis (PCA based on the environmental variables and using Statistica for Windows, version 5, StatSoft Inc.) to see if both species varied significantly (positively or negatively) in their response to this synthetic environmental gradient. The relative abundance here is the absolute abundance of species A at a given site divided by the absolute abundances of both species at the same site. We use relative abundance here because of the high variance in the absolute abundance among sites.
For the allozyme electrophoresis analyses, populations of both Colias species were sampled in the French Alps and other localities, including the Provence, Northern France and Italy (Table 1). After capture the butterflies were stored in liquid nitrogen. The head and thorax were subsequently homogenised in the laboratory in 180 ml of a 0.1M Tris-HCl buffer (pH 8.0) and the extracts were centrifuged for 6 minutes at 9000 x g. The supernatant from the extract was then frozen until used for electrophoresis.
Two gel matrices were used, namely cellulose acetate and starch gels. The starch gels were prepared according to Harris and Hopkinson (1977) and the exact preparation of buffers and gels follows Raijmann et al. (1997). Two buffer solutions were used in the starch gels, namely a Tris-Citric acid buffer (TC 8.0) and a Tris-Malic acid buffer (TM 7.4). Agar overlays were applied to all gel slices.
The following loci (locus abbreviations in parentheses) gave consistently good results and were used for the analyses: Aldolase (ald), Creatine Kinase 1 (ck1) and 2 (ck2), Glutamate-oxaloacetate transaminase 1 (got1) and 2 (got2), Glycerol-3-phosphate dehydrogenase (gpd), Hexokinase (hk), Isocitrate dehydrogenase 1 (idh1) and 2 (idh2), Malate dehydrogenase 1 (mdh1) and 2 (mdh2), Malic enzyme (me), Mannose phosphate isomerase (mpi), Phosphogluconate dehydrogenase (6pgdh), Glucose phosphate isomerase (pgi), and Phosphoglucomutase (pgm). The loci ck1, ck2, gpd, got2, and pgm gave the best resolution with TM; got1, hk, mdh1, mdh2, me, mpi, idh1, idh2, and pgi gave the best resolution with TC; 6pgdh gave the best resolution with a Tris-Glycine buffer (pH 8.5) on cellulose acetate. Locus number 1 refers to the least anodally migrating locus. Allelic designations use ‘a’ as the least anodally migrating allele, then ‘b’ and so forth.
Individual genotypes were used as input data for BIOSYS-1 (Swofford and Selander, 1981). The allelic frequencies and mean heterozygosity were derived directly from the BIOSYS output. Variation in heterozygosity between C. alfacariensis and C. hyale was tested for significance with a one-way Anova (using Statistica for Windows, version 5, StatSoft Inc.).
The inbreeding coefficients (FST FIT , and FIS ; Wright, 1951) were calculated according to Weir and Cockerham’s (1984) F, f, and Theta respectively. FST values were tested for departure from zero by the method of Workman and Niswander (1970), and inbreeding coefficients (FIS ) were tested by the method of Li and Horvitz (1953).
The amount of gene flow (Ne M) was estimated by the F statistic (FST ) method of Weir and Cockerham (1984), which uses Theta as an estimate of FST . Theta is assumed to be a better estimate of FST than Wright (1978) estimate and does not make assumptions about number of populations, sample sizes, or heterozygote frequencies so that it is more appropriate for small data sets (Weir and Cockerham, 1984). FST values were calculated using the computer programme THETA (Ellis, 1994), which uses the individual genotypes as input. FST is a useful estimator of gene flow according to the equation Ne M ≈ _([1/ FST ]-1) (Wright, 1951). Since there is an inverse relationship between FST and Ne M we consider FST values that do not differ significantly from zero to indicate unlimited gene flow, i.e., a panmictic populations structure.
Along the synthetic environmental axis (PCA-1) C. australis has a highly significant (R2 = 0.599, F1,33 = 49.20, P < 0.001) positive relation to PCA-1 whereas C. hyale has a highly significant (R2 = 0.599, F1,33 = 49.24, P < 0.001) negative relation (Fig. 1). Of the six environmental variables, vegetation and altitude were positively correlated with PCA-1, aspect and humidity negatively correlated, and date and weather uncorrelated. Humidity is thus probably the most important environmental determinant of both species since there is an obvious relation (e.g., exposed ridges tend to be drier than valley bottoms) between humidity and the other three variables that had high loadings for PCA-1 (vegetation, altitude, and aspect).
Fig. 1. Regression of C. alfacariensis and C. hyale onto the first Principal component axis (PCA-1). The slope of both lines is highly significant (P < 0.001).
Four of the loci (ald, mdh1, ck2, and got1) are monomorphically fixed for the same allele in the populations of both species. The remaining set of 12 allozyme loci did not contain loci that were diagnostic between C. alfacariensis and C. hyale (Appendix I). Four loci (got2, mpi, pgi, and pgm) show a very high degree of genetic polymorphism, but got2 is monomorphic in all populations of C. hyale.An intermediate level of genetic polymorphism is found among the loci hk, idh1, and me. Low levels of genetic variability are present in the remaining loci (Appendix 1).
In general, C. alfacariensis tends to have both a higher mean number of alleles and percentage of polymorphic loci (Table 2). Consequently, Colias alfacariensis also has a significantly higher heterozygosity (range: 0.148-0.194) than C. hyale (range: 0.084-0.094) (F1,11 = 72.70, P < 0.001). Heterozygosity is highest for C. alfacariensis populations in the geographically complex Alps and drops off towards the lower-lying regions to the South (Provence and Italy) and the North (Metz and Seine), with the lowest values occurring in Northern Europe. Within C. alfacariensis, the Alpine/Italian on the one hand and the Provence/Northern populations on the other hand consistently differ at the PGI locus (Fig. 2).
Fig. 2. Map of the localities where the populations of C. alfacariensis were sampled. Five main localities are shown, representing a single population with the exception of the Alps and Provence where three and four populations, respectively, were sampled. Within each circle average heterozygosity of 12 loci and the frequency of the pgi-b allele are given. For the Provence and Alps the values are means of the three populations. H = heterozygosity of a single populations, Hm = mean heterozygosity from more than one population.
In C. alfacariensis, estimated single locus FIS significantly differs from zero at me (Table 3). In C. hyale the PGM locus significantly differs from zero. FIS values were generally slightly positive in both species indicating a small deficiency of heterozygotes. In both species total genetic diversity (FIT ) was mainly the result of inbreeding (FIS ), among population variation (FST ) being less important. Within populations of both species the most common allele at a locus is usually the same (Appendix 1). Within C. hyale none of the loci have a FST value significantly different from zero indicating that, at least until recently, there was a panmictic population structure with high levels of gene flow. Over the same geographic range, C. alfacariensis has an estimated Ne M of less than 10. The difference here though is solely due to the PGI locus, which effectively divides the species into a Northern-Provence and Alpen-Italian component. Excluding this locus gives a panmictic population structure, since none of the other loci differ significantly from zero, quite similar to its sister species.
Colias alfacariensis and C. hyale both have a pronounced habitat preference, with C. hyale found on moist cultivated areas and C. alfacariensis found on dry uncultivated areas. Colias hyale has a consistently lower level of heterozygosity than C. alfacariensis, has fewer alleles per locus, and most populations are monomorphic at loci that Geiger (1982) has determined as being usually highly polymorphic in the Pieridae, such as got2. The fact that no diagnostic loci were found between C. alfacariensis and C. hyale is not uncommon in Lepidoptera. For instance, Aubert et al. (1997) also found no diagnostic loci between the sister species Papilio machaon and P. hospiton.
Both Colias species, but especially C. hyale are known to be migrants (Higgins and Riley, 1970). This explains why both species maintain a panmictic population structure over a large geographical area with little genetic differentiation between the populations. This is also the case with the Monarch butterfly (Danaus plexippus) (Eanes and Koehn, 1978) famed for its migratory ability. Other species of North American Colias also show a high level of movements between patches of their habitat (Watt et al., 1977).
The loss of genetic variation in C. hyale and its preference for high humidity conditions suggest that it may represent one of the many examples of northern split-off species that evolved from species that were able to survive the periods of glaciation in ice-free areas of the Mediterranean (Hewitt, 2000). Such populations were subsequently able to colonise the area left behind after the retreating ice sheet.
Split-offs usually result in reduced genetic variation coupled with independent evolution to sub-specific or specific status (Marchi et al., 1996). Populations at the northern edge of their range would be able to expand, usually via long-distance dispersal and rapidly colonize the newly available habitat allowing the best-adapted hybrid genomes to spread. These repeated colonising and founding events would eventually lead to a loss of alleles and thus to homozygosity. Many European and North American species show a lower genetic diversity in their northern populations (Hewitt, 2000). The same holds for other northern species, for example, the Dall sheep, Ovis dalli, only has a heterozygosity of 0.015 compared to an average of 0.041 for 184 mammalian taxa Hewitt (1989). Exactly such a pattern in levels of genetic variability can be found between C. alfacariensis and C. hyale. Not only does C. hyale have a lower genetic diversity than C. alfacariensis, but also Northern populations of C. alfacariensis have lower levels of heterozygosity than southern populations. The loss of heterozygosity in this and other Northern species is argued to be due to repeated fluctuation of ice fronts, which will cause a series of shifts in the periglacial range of the species involved. Range changes and consequently changes in population size provide evidence for a general loss of heterozygosity in northern species compared to southern species. This can further be predicted for subdivided species with extensive North-South ranges, depending on the life history and population dynamics of the species involved.
For the southern populations, the temporal predictability of a suitable habitat patch should decrease the rate of extinction, making the colonisation of other suitable habitat patches unnecessary for the continued survival of the species in a particular area. Such decreases in extinction probably would be associated with lower gene flow (Liebherr, 1988).
Selection at pgi has been demonstrated in a number of Colias species and PGI allozyme polymorphisms occur in parallel among many Colias species (Watt et al., 1996). In different American Colias complexes lowland and alpine groups have genotypes that display major functional differences, related to energy, allowing predictions about flight performance and fitness consequences which have been tested and confirmed (Watt et al., 1996; Watt, 1992). PGI polymorphs of Colias taxa may have diverged from an ancestral polymorphism and have been maintained till today (Watt et al., 1996).
Males of kinetically favoured pgi genotypes show greater flight capacity, survivorship, and overall mating success. There is also active female choice in selecting these favoured genotypes, where favoured males have an even greater advantage in mating older, more discriminating females than they do in mating previously unmated less discriminating females (Watt et al., 1986). A similar selection might be occurring in C. alfacariensis, but interestingly enough not in its sister species C. hyale. As in the American Colias species the main difference in the PGI locus among C. alfacariensis populations is between the alpine populations of France and the high elevation population in Italy and the low-lying populations in the Provence and Northern France.
Aubert J, Barascud B, Descimon H, Michel F. 1997. Ecology and genetics of interspecific hybridization in the swallowtails, Papilio hospiton Géné and P. machon L., in Corsica (Lepi-doptera: Papilionidae). Biological Journal of the Linnean Society. 60: 467-492
Berger LA, Fontaine M. 1947. Une espece meconue du genre Colias F. Lambillionea XLVII 11-12: 91-111
Brunton CFA. 1998. The evolution of ultraviolet patterns in European Colias butterflies (Lepidoptera, Pieridae): a phylogeny using mitochondrial DNA. Heredity 80: 611-616
Descimon H. 1994(5). Colias hyale (L.): une espèce en voie de disparition? Alexanor 18: 440-442
Dutreix C. 1980. Étude des deux espèces affines Colias hyale Linné et Colias alfacariensis Verity [Lepidoptera, Pieridae]. Alexanor 11: 297-316
Eanes WF, Koehn RK. 1978. An analysis of genetic structure in the monarch butterfly, Danaus plexippus L. Evolution 32: 784-797
Ellis WN 1994. Theta version 2.2. A program for the Macintosh computer for the analysis of population structure and estimation of gene flow levels based on the methods of Weir and Cockerham (1984). Available at no charge from the author.
Geiger H. 1982. Biochemisch-genetische Untersuchungen zur Systematik und Evolution von Weisslingen des Europäischen Faunengebietes (Lepidoptera, Pieridae) Thesis, Univ. von Bern: 1-80
Harris H, Hopkinson DA. 1977. Handbook of enzyme electrophoresis in human genetics. North Holland publishing company
Hewitt GM. 1989. The subdivision of species by hybrid zones. In: Speciation and its Consequences [Ed.Daniel Otte and John A. Endler]: 85-110 Sinauer Ass. Inc. Publishers
Hewitt G. 2000. The genetic legacy of the quaternary ice ages. Nature 405: 907-913
Higgins LG, Riley ND. 1970. Field guide to the butterflies of Britain and Europe. Dutch edition: Elsevier, Amsterdam.
Jarvis FLV. 1955. Etude biologique comparee de Colias alfacariensis (Verity) et de Colias hyale (L.) Lambillionea LV, 3-4: 28-34 and 5-6: 37-50
Lewis OT, Thomas CD, Hill JK, Brookes MI, Crane TPR, Graneau YA, Mallet JLB, Rose OC. 1997. Three ways of assessing metapopulation structure in the butterfly Plebejus argus. Ecological Entomology 22: 283-293
Li, C.C. and D.G. Horvitz. 1953. Some method of Estimating the inbreeding coefficient. Am. J. Hum. Gen. 5: 107~117.
Liebherr JK. 1988. Gene flow in ground beetles (Coleoptera: Carabidae) of differing habitat preference and flight-wing development. Evolution 42: 129-137
Marchi A, Addis Exposíto Hermosa V, and Crnjar R. 1996. Genetic divergence and evolution of Polyommatus corridon gennargenti (Lepidoptera: Lycaenidae) in Sardinia. Heredity 77: 16-22
Menken SBJ, Herrebout WM, Wiebes JT. 1992. Small ermine moths (Yponomeuta): their host relations and evolution. Annu. Rev. Entomol. 37: 41-66.
Peterson AT, Soberón J, Sánchez-Cordero V. 1999. Conservation of ecological niches in evolutionary time. Science 285: 1265-1267
Raijmann LEL, VanGinkel WE, Heckel DG, Menken SBJ. (1997). Inheritance and linkage of isozymes in Yponomeuta padellus (Lepidoptera, yponomeutidae). Heredity 78: 645-654
Ritchie MG, Phillips SDF. 1998. The genetics of sexuals isolation. In: Endless Forms: species and speciation. Oxford University Press. P 470: 291-308
Statistica for Windows (version 5, StatSoft Inc.).
Swofford, D. L., and R. B. Selander. 1981. BIOSYS-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72:281-283.
Tilman D. 2000 Causes, consequences and ethics of biodiversity. Nature 405: 208-211
Watt WB, Chew FS, Snyder LRG, Watt AG, Rothschild DE. 1977. Population structure of pierid butterflies. I. Numbers and movements of some montane Colias species. Oecologia 27:1-27.
Watt WB, Carter PA, Donahue K. 1986. Females’ choice of “good genotypes” as mates is promoted by an insect mating system. Science 233: 1187-1190
Watt WB. 1992. Eggs, enzymes, and evolution: natural genetic variants change insect fecundity. PNAS 89: 10608-10612
Watt WB, Donohue K, and Carter PA. 1996. Adaptation at specifc loci. VI. Divergence vs. parellelism of polymorphic allozymes in molecular function and fitness-component effects among Colias species (Lepidoptera, Pieridae). Mol. Biol. Evol. 13: 699-709
Weir BS, and Cockerham CC. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370.
Workman, P. L. and J. D. Niswander. 1970. Population genetic studies on southwestern Indian tribes. II. Local genetic differentiation in the Papago. Am. J. Hum. Gen. 22: 24-49.
Wright S. 1951. The genetical structure of populations. Annals of Eugenetics 15: 323-354
Wright S. 1978. Evolution and the genetics of populations. Vol. 4. Variability within and among natural populations. University of Chicago Press, Chicago.
We wish to thank Leon Raijmann and Wil van Ginkel for help in the lab. Barbra Schütt and Filipo Fabiani provided help with field collections.