Selection in sympatric populations of Cepaea

On a preleve 95 echantillons de populations allopatriques et sympatriques de Cepaea nemoralis et de Cepaea hortensis a proximite de l'extremite meridionale de l'aire d'extension de Cepaea hortensis. On a observe des variations de frequences phenotypiques, en passant de l'allopatrie a la sympatrie, dans les deux especes

Changes in phenotype frequencies were observed passing from allopatry to sympatry in both species. In the North, a decrease in the frequencies of yellow, yellow-banded and effectively banded yellow snails was detected in sympatric populations of Cepaea nemoralis and, at the same time, the frequencies of the banded, five-banded, and effectively banded phenotypes in Cepaea hortensis increased. These changes in phenotype frequencies between allopatric and sympatric populations in one species may depend upon the phenotype frequency of the other. Such frequency dependent selection may interact with other selective forces and with competitive selection.
We believe that, in southern populations, these changes are due to climatic selection in both species.
The range of these two species overlaps. In Northern Europe, Cepaea hortensis reaches higher latitudes, while Cepaea nemoralis reaches more southerly latitudes. This may indicate that Cepaea hortensis is more resistant to cold climates than Cepaea nemoralis, which in turn, is better adapted to milder climates (Cameron, 1970a, b, Jones et. al., 1977. In mixed populations, different morph frequencies are often found for each species, and some local correlations have also been found between morphs of the two. This could be due to several factors: to visual selection on shell pattern or size (Clarke, 1960(Clarke, , 1962aCarter, 1967;Bantock & Bayley, 1973;Bantock et al., 1976); to interspecific competition (Arthur, 1978(Arthur, , 1980; to climatic selection (Arthur, 1982a) or to interspecific competition and intra and interspecific effects of population density (Cameron & Carter, 1979).
Various modes of evolutionary change resulting from interspecific competition have been proposed: character displacement (unilateral or bilateral), convergence of characters and alteration of the variance of morphological characters, evolution of competitive capacity and &dquo;genetic feedback&dquo; (Arthur, 1982b).
Here, we study populations of Cepaea nemoralis and Cepaea hortensis in allopatry and sympatry, to analyse whether climatic selection or interspecific competition cause any differences between allopatric and sympatric populations. We study them in a zone where Cepaea hortensis is at the edge of its range. Marginal populations are interesting since they may be found in limiting environmental or competitive conditions (Bantock & Price, 1975).

MATERIAL AND METHODS
Sampling was carried out in the western part of the Iberian Mountains (Fig. 1). This zone has high peaks (San Lorenzo 2,262 meters, Picos de Urbion 2,228 meters, etc.) and deep valleys. It is very cold in winter and hot in summer. Each sample was taken in an area of approximately 400 m 2 .
A total of 7 235 snails was collected, of which 5 220 were Cepaea nemoralis and 2 015 were Cepaea hortensis. Fifty allopatric and 32 sympatric populations were found for Cepaea nemoralis (Table I), while for Cepaea hortensis, there were 13 allopatric and 32 sympatric. For the purpose of analysis, only 23 sympatric populations of nemoralis and 26 of hortensis were considered. The difference in the number of sympatric populations of the two species is due to the fact that some populations had a very small number of individuals. Samples were scored according to the criteria of Lamotte (1951), Cain and Sheppard (1954) and Arnold (1968).
For each sample, data on vegetation, pH, soil characteristics, climate, distance from the sea and altitude were collected.

RESULTS
Cepaea nemoralis is more common and more widespread, than is Cepaea hortensis. Each species is polymorphic in the region. For Cepaea nemoralis, the area sampled was divided into two zones, the north and south-facing slopes of the Iberian Mountains (Table II). Phenotype differences appeared, principally in the patterns of bands (Fig. 2), whereas, for coloured, no differences were apparent.
We examined allopatric and sympatric populations in each zone. In the southern zone there were no significant differences in either species. In the north-facing slopes, allopatric populations of Cepaea nemoralis exhibit, a significant increase in yellows, banded yellows and effectively banded yellows (Table II), while in the sympatric zone the pinks are more frequent.
In allopatric populations of Cepaea hortensis, unbanded yellow, effectively unbanded and unbanded phenotypes were significantly more frequent than in sympatric populations, while in the latter, effectively banded, banded and yellow 12 345 were more frequent in the northern zone (Table II).
When frequencies are compared between species in sympatry and in allopatry in each zone, the number of significant differences are much the same for allopatric and sympatric comparisons in the south. In the north, however, allopatric comparisons produced many fewer significant differences than sympatric ones. There were no significant differences of phenotypic frequencies related to soil characteristics, pH or vegetation. However, there were significant differences in the altitude of populations between northern and southern slopes (Table III).
There was a series of significant correlations (P < 0.05) between morph frequencies and altitude in Cepaea hortensis, but not with very high correlation coefficients.
In the northern zone, the unbanded yellow and effectively unbanded yellow phenotypes were positively correlated with altitude. In the southern zone, yellow 12 345 and 12 345 were negatively correlated, whereas, fusion-yellow and fusion-banded were positively correlated. Cepaea nemoralis phenotypes, positively correlated with altitude, include pink 00300 and yellow 00300 in the North. Trush predation was found in only five populations, of which, three were allopatric and the other two, sympatric (Table IV). In two populations, the predation is not selective, and in the remaining three, the most heavily predated are the fusion banded or effectively banded for Cepaea nemoralis, (San Andr6s, banded shells, X3 = 8.12; Vizmanos, colour shell, xi = 4.98, effectively banded shell, ki = 4.39; Diustes, banded shell, x3 = 8.06), while for Cepaea hortensis, there is no selective predation. In our study, each population is composed of 16 different phenotypes. Up to now, we have made comparison phenotype by phenotype. This gives an idea of what is happening with just one phenotype at a time in each species, but causes a loss of perspective with respect to what is occurring overall. Therefore, we looked for an analysis where the 16 phenotypes and all the populations could be introduced, and whose results would be easy to interpret; an analysis which fulfills all these demands is Factorial Correspondence Analysis (FCA). Two large groups of Cepaea nemoradis and Cepaea hortensis were separated (Fig. 3), so that the populations of two species would be sufficiently different to be grouped into two different areas in the figure. Within Cepaea nemoralis, there was a subgroup including the sympatric populations, although in this case, the separation was not as clear as between species. FCA represents phenotypes and populations and also keeps the proximity relationship (i.e. the Cepaea hortensis populations are close to phenotypes yellow 12 345, fusion banded yellow and hyalozonate), which indicates that, in Cepaea hortensis, these are the most frequent phenotypes. This separation may also be observed in the cluster (Fig. 4), made by nearest neighbour/paired sample analysis. The two species are separated by a level of affinity of 0.423. Within Cepaea hortensis, there is a separation of populations at 0.165 with a high proportion of unbanded yellow. For Cepaea nemoralis, there is a group at 0.102, where the majority of populations are allopatric.

DISCUSSION AND CONCLUSIONS
In our area, Cepaea nemoralis populations seem to be divided into two zones corresponding to the north and south-facing slopes of the Iberian Mountains.
Marked climatic differences characterize the two zones. The southern zone is drier and hotter with more severe climatic conditions than in the northern zone. The observed differences in the phenotype frequencies could be produced by climatic selection or by geographic isolation, due to the Iberian Mountains.
When analysing the passage from allopatry to sympatry within each species, changes may be observed in the frequencies of colour patterns and shell banding. On north-facing slopes of the Iberian Mountains, in sympatric populations of Cepaea nemoralis, the effectively banded yellows and other yellow morphs decreased. With regards to colour, pink shells increased in sympatry, while yellow phenotypes decreased. For Cepaea hortensis, the banded phenotypes also increased in sympatric populations, while the unbanded diminished. This is similar to Arthur's (1978Arthur's ( , 1980Arthur's ( , 1982a observations. The gene for shell colour of Cepaea hortensis does not seem to be affected. Why are there such frequency differences between allopatric and sympatric populations? Some authors have proposed visual selection as the main cause (Clarke, 1962a, b, Carter, 1967Bantock et al., 1976). In our case, this does not seem to be so, since trush predation is very slight, and not significant. Others authors have suggested interspecific (Arthur, 1978(Arthur, , 1980 or climatic selection (Cameron, 1970a, b, c, Arthur, 1980, although it is very difficult to separate the two (Arthur, 1982a). Climatic selection may be important in our southern populations, as there are no differences between the morph-frequencies of the allopatric or the sympatric populations within a species, while in the north there are differences. Another argument in favour of this kind of selection, is that the Cepaea hortensis populations sampled are marginal, and may be in environmentally limiting conditions. Moreover, the sympatric populations of Cepaea hortensis in the southern zone are significantly higher in altitude than those in the north, as happens with allopatric populations of Cepaea nemoralis.
For sympatry, there are more significant differences between the two species in the north, than in the south. However, for allopatry, they are higher in the south. Allopatric populations present more differences in morph-frequencies probably due to differences in niches or adaptative strategies. It is known that Cepaea nemoralis and Cepaea hortensis have different strategies to reach similar darkening of the shell (Jones et al., 1977).
In the northern zone, the allopatric populations in both species have few significant differences in their morph-frequencies. That implies that there is little effect of climatic selection on the northen slopes, with a greater importance of interspecific competition, since the differences between species increase in sympatry. The phenotypes of Cepaea nemoralis or Cepaea hortensis, which present significant differences between allopatry and sympatry, would suffer interspecific competition. A possible strategy would be displacement of morph-frequencies towards phenotypes infrequent in the other sympatric species, in an attempt to make the two species appear very different as far as their phenotype frequencies are concerned (i.e., within the phenotypes which suffer displacement, if one species has high frequencies of a phenotype, the other species tends to decrease its frequency when it passes from allopatry to sympatry, thus increasing the differences between the morph-frequencies of both species). This is reflected in the FCA (Fig. 3), where sympatric populations of Cepaea nemoralis appear on average farther from Cepaea hortensis than allopatric populations. Also sympatric populations of C. hortensis appear, on average, farther from C. nemoralis than the allopatric populations, although genetic flow between same species populations may exist. However, Cowie and Jones (1987) showed that there is competition and habitat separation in Cepaea, but found no evidence of an interspecific interaction that might be a precursor of character displacement in experimentally mixed population of C. nemoradis and C. hortensis. They consider that the competition may reflect a balance between invasion and extinction. On the other hand, Arthur (1982a) observed differences between allopatry and sympatry, possibly due to climatic factors, since allopatric populations of Cepaea nemoralis had less vegetation density and more sunlight. The data of Cameron and Dillon (1984), also show differences in wooded habitats. In our study, we did not observe such a difference.
This leads us to believe that, in the south, the limiting factor is climatic, while in the north, there may be a mixture of climatic factors and competitive displacement.
Taking into account these results, climatic selection may be acting upon one or both species and competitive selection can be detected in the species or phenotypes not subject to climatic selection, although it is not clear whether there is an interaction between the two.
We agree with Arthur (1982a), that there is no reason why unbandeds should be weaker interspecific competitors, or why climatic selection should be the cause of the frequency changes when passing from allopatry to sympatry. We think rather, that a phenotype will be a stronger or weaker competitor depending on the frequency it has in the other species. This is an extension of the concept of selection depending on frequencies to a two species situation (see Arthur, 1982b). Frequency dependent selection had already been noted for mixed colonies, but from the point of view of selection (Clarke, 1962b, Bantock et al., 1976. There are cases where differences have not been observed in morph-frequencies between allopatric and sympatric populations; this may be due to interaction between various selective forces which neutralise character displacement, or to an equilibrium in the mixed populations.