Printer Friendly

Morphological differentiation and genetic cohesiveness over a microenvironmental gradient in the marine snail Littorina saxatilis.

The study of conspicuous small-scale morphological variation within populations with cross-morph gene flow is both taxonomically and evolutionarily interesting. It reminds us that superficial investigations may indicate that conspecifics are different species. It can also contribute to discussions about how species should be defined, whether gradual sympatric speciation is possible, and how natural selection and gene flow may interact to maintain clines over small distances.

Several species concepts are discussed in the literature, and they serve different purposes (Endler 1989). But we should at least include within the same species groups of individuals with considerable gene flow among them. If we consider different morphs of the same gene pool members of different species, our interpretations of evolution will be misleading or completely wrong.

Differential selection can counteract the equalizing effect of gene flow (Endler 1977), and in species of relatively poor dispersal, abrupt phenotypic clines may appear over tens of meters (e.g., Jain and Bradshaw 1966; Murray and Clarke 1984; Hazel and Johnson 1990). However, reports of very steep clines over less than 10 m, or so, are rare (reviewed by Endler 1977, see also Mitchell-Olds 1992).

The gastropod Littorina saxatilis is a marine intertidal taxon that has an extremely variable shell morphology. Different forms show pronounced differences in, for example, size at maturity, shell thickness, spire height, aperture area and shell color. Thus, in earlier studies, groups of morphologically different forms were often considered distinct species (Heller 1975; Smith 1981; Fretter and Graham 1980). Later analyses of protein variation have indicated that many of the different morphs belong to the same gene pool (Janson and Ward 1984; Janson and Ward 1985; Johannesson and Johannesson 1990b). However, groups of reproductively separated individuals have sometimes been overlooked (Hannaford Ellis 1979; Ward and Warwick 1980; Ward and Janson 1985; Sundberg et al. 1990). Now the consensus is that the old species L. saxatilis includes three taxa that deserve species status, namely Littorina arcana Hannaford Ellis, Littorina nigrolineata Gray, and L. saxatilis (Olivi). The three true species are closely related. Genetic distances (Nei's D) range from 0.035 to 0.083 (22 loci, see Knight and Ward 1991), which is in the lower range of genetic distances between congeneric animal species (Thorpe 1982).

Fossil records together with a recent phylogenetic analysis based on morphological variation of the approximately 20 existing species of the genus Littorina (Reid 1990a) suggest that in the Upper Pliocene at least two independent lineages immigrated from the Pacific Ocean to the northern Atlantic, and led to the six recent species of Littorina in this area (Reid 1990b). One clade includes L. saxatilis, L. arcana, L. nigrolineata, Littorina mariae, and Littorina obtusata, which all have direct development, whereas the other clade contains only one species [Littorina littorea (L.)], which has planktotrophic development. This may reflect a greater speciation rate for the clade with direct development.

All species of the evolutionary lineage with direct development are polymorphic in, for example, shell characters (Reimchen 1981; Janson 1982a; Naylor and Begon 1982; Janson and Sundberg 1983; Grahame and Mill 1986; Warmoes 1986). The only species of the other clade (L. littorea) has a high potential of dispersal through a 4- to 6-wk pelagic stage, and this species is much less variable in shell morphology (Janson 1987).

Is there any causal connection between low dispersal rates and high levels of polymorphism? Is it possible that isolation mechanisms can evolve between sympatric or adjacent populations with different morphs, or does allopatric speciation dominate within the group of direct developing snails? If we knew more about gene flow, natural selection, population structure, and development of isolation mechanisms among morphs, we would, perhaps, be able to suggest answers.

If the phenotypic variation is inherited and the gene flow is not so heavy that it will counterbalance differentiation, the most likely explanation of within-population variation in phenotypic characters is that the environment of a population is heterogeneous with different parts of the population being under different regimes of selection. Unfortunately, we often do not know if the observed variation is inherited. Furthermore, both changed environments and changed gene frequencies cause altered heritabilities. Therefore, it can be misleading to use estimates of heritabilities from studies of other populations in other environments (Lewontin 1974; Sultan 1987). However, studies of phenotypic variation in L. saxatilis suggest that important genetic components may be found at least in some shell characters (three parameters of shell shape, Newkirk and Doyle 1975; growth rate, Janson 1982b; shell color, Atkinson and Warwick 1983 and Ward et al. pers. comm.). Furthermore, Boulding and Hay (1993) found significant additive genetic variance for shell shape and shell thickness in one species of Littorina (L. sp.). Other studies, however, show that the development of shell characters can be influenced by, for example, growth rate (L. littorea, Kemp and Bertness 1984) and crab predation (Nucella lamellosa, Appleton and Palmer 1988; Nucella lapillus, Palmer 1990).

Intraspecific variation in shell form in L. saxatilis is most typically recognized as variation between groups of individuals living in shores of different types (reviewed by Raffaelli 1982 and Faller-Fritsch and Emson 1985). However, Johannesson and Johannesson (1990a) found that much intraspecific variation may be present between upper and lower levels of the same shore, especially in wave-exposed rocky sites. In some parts of Europe, for example, around the British Isles, a large component of within-shore variation has earlier been attributed to the presence of two separate species: Littorina neglecta in the lower shore and L. saxatilis in the upper shore (e.g., Heller 1975; Fish and Sharp 1985). But L. neglecta is not genetically different from L. saxatilis and the two forms are therefore ecotypes of the same species (Johannesson and Johannesson 1990b).

The Galician populations of L. saxatilis that we have investigated in the present study are dimorphic in shell characters but the differences in morphologies of these two forms are not those typical of L. saxatilis or L. neglecta populations in Great Britain. However, the Galician populations reflect a similar situation of dimorphism. The upper-shore form of Galician L. saxatilis is large, has a pronounced ornamentation and a black-banded shell and is easily distinguished from the lower-shore form that is small, smooth, and unbanded. The two forms occupy different microenvironments, an upper-shore barnacle belt and a lower-shore mussel belt. Their distributions overlap only in a narrow border zone.

The objectives of this study were to describe the pattern of allozyme differentiation within and between shores, and to find out if this differentiation is related to the morphological differentiation observed. Furthermore we wanted to know if differences in shell sculpture and banding patterns are inherited, and if differences in other phenotypic characters are associated with them.

Although we found genetic differences in shell ornamentation and color between the two morphs, we found no correlation between morphological distances and allozyme genetic distances when we compared populations kilometers apart. Therefore, we concluded that the two morphs belong to the same species (L. saxatilis), and that strong divergent selection keeps them separate in spite of a continuous distribution and a gene flow over the contact zone. On a local scale, gene flow between morphs is, however, somewhat reduced compared with gene flow within morphs, and preliminary data suggest this is caused by nonrandom mating between the two morphs.


The Organism

Individuels of Littorina saxatilis are unisexual and have only sexual reproduction; females give birth to small crawl-away juveniles. The scale of active dispersal of adults, as estimated for Swedish populations, is in the range of only a few meters per month (Janson 1983). It seems, however, that sometimes passive dispersal of adults and juveniles over long distances (in the range of kilometers or more) may occur (Johannesson and Warmoes 1990). The species occupies a wide range of environments (e.g., brackish water lagoons, salt marshes, sheltered boulder shores, and exposed rocky shores). Shell morphology generally converges within the same type of environments. This suggests that selection is indeed important. Furthermore, one study has shown that the survival rates of different morphological forms are substantially different between different habitats (Janson 1983).

The Galician lower-shore form of L. saxatilis is different from the British neglecta form because it is about twice as large, and it lacks a dark band running into the aperture. Furthermore, the typical habitat of British neglecta is inside empty barnacle shells, whereas the Galician lower-shore form lives among blue mussels. (See Johannesson and Johannesson [1990a] for a discussion of the characters of Littorina neglecta.) The Galician upper-shore form is similar to some populations of L. saxatilis from, for example, the Dale area of southwest Wales (Grahame and Mill pers. comm. 1990; pers. obs.).

Sampling Design

We wanted to analyze gene flow over large distances (tens of kilometers), and both vertically and horizontally over short distances (tens of meters and less). We therefore sampled three pairs of transects from different geographical positions (Corrubedo Cape, 42 [degrees] 33'N, 9 [degrees] 6'W; Silleiro Cape, 42 [degrees] 6'N, 8 [degrees] 53'W; La Cetarea, 41 [degrees] 55'N, 8 [degrees] 53'W; fig. 2). Each transect (A and B) included three samples, one upper-shore (US), one midshore (MS) and one lower-shore (LS) sample. Note that we labeled the transects A and B for convenience of reference only. One cannot a priori suppose that samples with the same label have more in common with each other than with samples designated with the other label. This also applies to samples within localities.

The upper-shore samples were from the top of the barnacle (Chthamalus stellatus) belt, or somewhat above. Ridged and banded forms of L. saxatilis dominated here. The lower-shore samples were from within the blue mussel (Mytilus galloprovincialis) belt. Smooth and unbanded forms of L. saxatilis dominated here. The midshore samples were from the border areas between the mussel and barnacle zones. Ridged and banded and smooth and unbanded forms occurred in about equal proportions here.

We added three additional samples to one of the six transects (Corrubedo Cape B). Two were sampled from between the upper-shore and the midshore samples, and one was taken from between the midshore and the lower-shore samples. We took each sample from within 1-2 [m.sup.2] of shore.

Sampling Sites

All three geographic areas (Corrubedo Cape, Silleiro Cape, and La Cetarea) represented very wave-exposed shores with smooth granite rocks interrupted by cracks and crevices, in particular at Corrubedo Cape. The slopes of the shores were 25 [degrees]-30 [degrees] at Corrubedo Cape, 5 [degrees]-10 [degrees] at Silleiro Cape, and 15 [degrees]-20 [degrees] at La Cetarea. The tidal range is about 1 m at neap tides and about 4 m at spring tides.

Determination of Statistical Significance

We used randomization tests instead of parametric tests to evaluate most of our null hypotheses. The advantage of randomization tests is that they are distribution-free, and therefore not restricted by the same assumptions as parametric tests. It is only the way of determining the significance that is different in a randomization test compared to its standard equivalent (Edgington 1987; Manly 1991).

Qualitative Characters

We examined all individuals of all samples (range of N, 34-102) to see if they had dark bands parallel to the growth spiral at least on the last (the biggest) whorl, and if they had ridges. We classified all individuals that had both shell ridges and dark bands into ridged and banded forms, and those that lacked both these characters into smooth and unbanded forms. We designated individuals that were banded and smooth, or unbanded and ridged, as "mixed forms."

Shell Size and Shape

We performed a principal-component analysis (PCA) on the upper-shore, midshore, and lower-shore samples from one transect (Corrubedo Cape, transect B) using the same individuals that were analyzed genetically. We used the following shell characters in the analysis: aperture height, aperture width, columella width, lip height, shell height, shell thickness, shell width, whorl breadth (the diameter at the middle of whorls), whorl diameter (at the suture of the whorls), and whorl height. The three last characters were measured on the last and second last whorls, and numbered 1 and 2.

We measured the shell dimensions, except shell thickness, on the magnified picture of each shell displayed on a video monitor. The precision of these measurements ranges from 2.6 [[Micro]meter] to 44 [[Micro]meter] depending on the dimension. We measured shell thickness with a digital dial indicator about 1 mm from the edge of the aperture. The precision was 1 [[Micro]meter]. See Johannesson and Johannesson (1990a) for more details.

PCA is useful for summarizing and presenting results from measurements of several characters (e.g., Manly 1986). It can separate size and shape differences (Jolicoeur 1963) if they are not confounded because of allometry (e.g., Airoldi and Flury 1988; Tissot 1988). In addition, ordinary PCA is intended for single-group analysis, and applying this technique to analysis of several populations may also create problems (e.g., Airoldi and Flury 1988; Tissot 1988). The significant results we got would probably, however, have persisted if we had used a PCA that is intended for the multigroup case, because the differences among means are so large. Also direct examination of the size and shape variation among the groups verities the results of the PCA.

We could break up the 162 individuals analyzed into seven groups according to shore level (upper, mid, and lower) and shell category (ridged and banded, smooth and unbanded, and mixed form). No smooth and unbanded or mixed morphs appeared in the upper-shore sample. The number of snails in each group range from 32 to 42. To look for differences among the seven groups we performed a one-factor analysis of variance (ANOVA) for each of the three first principal components. We estimated the probabilities of F-ratios as high as or higher than the observed if no real differences exist. This was done by comparing the number of F-ratios as high as, or higher than the F-ratios for the observed data, when the computer randomly allocated the shell height data over the seven groups 4999 times and added the observed value. For the two first principal components we used Fisher's modified least significant difference procedure to see which of the means differed significantly. We permuted the scores 19,999 times and added the observed ordering for each of the 21 comparisons to determine the P-values.

Shell Height

The PCA revealed that in transect B from Corrubedo Cape the largest differences between groups were in mean shell size. Shell height is a widely used measure of shell size and was highly correlated with PC 1 (N = 162, [R.sup.2] = 0.916, P [is less than] 0.01), which we then regard as a measure of multivariate allometric size. We could therefore use shell height to study size variation among snails from all localities and shore levels, using all the 18 main samples.

To define shell height, we used a line from the apex through the center of the columella and another line at a right angle to the first, which touched the edge of the lip. Shell height was then the distance between the shell apex and the intercept between the two lines.

We analyzed the variation in mean shell height among the 18 samples in a three-factor nested ANOVA. Locality (random, three levels) and shore level (fixed, three levels) were orthogonal factors. Transect (random, two levels) was nested within both the orthogonal factors, as each pair of samples (labeled A and B) constituted two replicates within each of the nine combinations of localities and shore levels.

We randomly deleted 0 to 68 individuals in each sample to make the ANOVA balanced with N = 34. Several approximations or F-values when the ANOVA is unbalanced have been proposed. But they can be biased, and in our analyses the deletion of individuals affects only the degrees of freedom for the residual, which has an unbiased value of 594. The deletions have very little effect on power.

We estimated the probabilities of F-ratios as high as or higher than the observed if no real differences exist. This was done by comparing the number of F-ratios as high as or higher than the F-ratios for the observed data when the computer randomly allocated the shell height data over the seven groups 1300 times and added the observed value.

We could not use ANOVA to compare shell heights of the different morphs from different shore levels, because we did not find all morphs in all shore levels. Instead we present mean shell heights and standard errors of the means for each morph. We do this for the 18 main samples from each shore level and site.

Genetic Variation

We analyzed on average 48 individuals (range 4-104) for five polymorphic enzyme loci: phosphoglucose isomerase (Pgi, EC, mannose phosphate isomerase (Mpi, EC, aspartate aminotransferase (Aat- 1, EC, phosphoglucomutase (Pgm-1, EC, and nucleoside phosphorylase (Np, EC These loci represent the most polymorphic loci found in L. saxatilis (see for example, Sundberg et al. 1990, and see Ward et al. 1986; Knight and Ward 1986 for breeding experiments that confirm Mendelian inheritance of these loci).

We prepared samples and analyzed enzymes essentially as in Ward and Warwick (1980) and Janson and Ward (1984). Allele designations follow that of Johannesson and Johannesson (1989), although three new alleles appeared in the Spanish populations, which have not been reported before [i.e., [Pgi.sup.75], [Pgi.sup.50], and [Pgi.sup.35]--the first one may be equivalent to the rare allele [Pgi.sup.80] described by Johannesson and Johannesson (1989) and others].

We analyzed two sets of samples. In the first set we included all 12 upper-shore and lower-shore samples. In the second set, we analyzed the six midshore samples, which we first divided into groups of ridged and banded and smooth and unbanded form (the groups containing individuals of mixed morphology were too small to be included in further analysis).

We used the BIOSYS-1 package, release 1.7 (Swofford and Selander 1981) to calculate allele frequency data and to analyze the pattern of differentiation between groups. In this, we applied two different clustering techniques to two different estimates of genetic relationships. We used the Unweighted Pair-Group Method with Arithmetic Averages (UPGMA) to cluster estimates of Nei's genetic identity between pairs of groups, and the distance Wagner procedure to analyze Rogers' genetic distance matrix. However, the results of the two different types of analyses were essentially the same, and we have chosen to present only the UPGMA phenograms of Nei's I (Nei 1972).

A gene diversity analysis (Nei 1973; Chakraborty 1980) was performed to trace the sources of the genetic variation over samples. In this, we included all upper-shore and lower-shore samples of the six transects to compare differentiation within and among the three geographic areas and differentiation within and between habitats (i.e., upper and lower shores).

We furthermore used Mantel's test (e.g., Manly 1991) to see if there was any positive correlation between morphological and genetic distances between samples. We constructed a morphological distance matrix in which the distance between any two populations was

d = 0.5 [Sigma][absolute of][p.sub.i] - [q.sub.i]

(Manly 1985, p. 180), where [p.sub.1], [p.sub.2], and [p.sub.3] were the proportions of the ridged and banded, mixed, and smooth and unbanded morphs in one of the populations, and [q.sub.1], [q.sub.2], and [q.sub.3] were the corresponding proportions in the other population. We used Nei's (1972) genetic distance in the genetic matrix. As the purpose of this test was to find out if, on a larger geographic scale, samples of snails that were morphologically similar also were genetically related, we did not use replicate samples from the same shore level of a site (e.g., Corrubedo Cape A, upper shore and Corrubedo Cape B, upper shore) as these were likely to be genetically as well as morphologically correlated because of the small distances between them. Instead we randomly chose one upper-shore, one midshore, and one lower-shore sample from each site (Corrubedo Cape, Silleiro Cape, and La Cetarea) in the Mantel's test. In our case, the test estimated the probability of finding a correlation as high as, or higher than the observed, if no real correlation exists (one sided test). This was done by comparing the number of correlations as high as or higher than the correlation for the observed data when the computer randomly allocated the genetic distances 4999 times and added the observed value.

Heredity of Morphological Characters

We isolated 11 ridged and banded females from upper-shore sites and six smooth and unbanded females from lower-shore sites in transparent plastic containers (280 mL) with top and bottom replaced by plankton nets (400 [[micro]meter]). The containers were placed in outdoor gutters with a continuous flow of sea water. It seems most probable that the females had mated in the wild with males of similar morphological types. Each female produced offspring continuously, and we left them together with the female until the largest offsprings had reached a shell height of about 8.5 mm. Then we stopped the experiment and noted the shell color pattern and shell sculpture in small ([is less than] 3 mm) and large ([is greater than] 3 mm) offspring.

In addition, we measured sculpture height and shell height in sibling groups of 9 to 15 individuals. These came from five of the ridged and banded mothers. We performed an analysis of covariance to see if the relationships between sculpture height (dependent variable) and shell height (covariable) differed between sibling groups. We determined the P-values by randomizing the sculpture height data over the five groups 990 times and adding the observed value.


Qualitative Characters

The ridged and banded form made up 92% to 100% of the individuals in the upper-shore samples, whereas in the lower-shore samples between 66% and 97% were smooth and unbanded forms. We found no smooth and unbanded forms in the upper-shore samples, but we identified between 0% and 7% of the individuals in the lower-shore samples as ridged and banded. Either the ridged and banded or the smooth and unbanded form dominated in the midshore samples, but between 11% and 29% of the individuals had mixed characters.

Transect B from Corrubedo Cape included three extra samples in between the upper-shore and lower-shore samples. The distribution of forms within these extra samples revealed that the switch from complete dominance of ridged and banded forms to dominance of smooth and unbanded forms was abrupt. (Note that the total length of the Corrubedo Cape B transect was only 13 m and that the distances between successive samples were only 2-3 m.)

Shell Size and Shape

The principal-component analysis (PCA) was successful in reducing the original values into a few transformed variables that explain much of the variation, because all the measured shell dimensions were highly correlated with each other. The first component describes 90.7% of the total variation, and when the first three components have been considered, less than 5% of the variance remains.
TABLE 1. Eigenvalues and component loadings for the first three principal
components analyzed in Littorina saxatilis from Corrubedo Cape, transect B.
Seven groups are included in the analysis: ridged and banded snails from the
upper shore (N = 42), ridged and banded from the midshore (N = 40), mixed from
the midshore (N = 23), smooth and unbanded from the midshore (N = 18), ridged
and banded from the lower shore (N = 3), mixed from the lower shore (N = 11),
and smooth and unbanded from the lower shore (N = 25). The principal-component
analysis (PCA) is based on a covariance matrix among 13 quantitative shell
characters. The eigenvalues indicate the variances of each component, and the
component loadings show (for each component) which characters contribute most
to the total variation.

Component:              1         2         3
Eigenvalue:           0.348     0.010     0.008
Cumulative %:        90.7      93.3      95.6

                          Component loadings

Shell height          0.207     0.108     0.188
Shell width           0.189     0.132     0.236
Aperture height       0.171     0.052     0.246
Aperture width        0.155     0.087     0.278
Lip height            0.330    -0.580    -0.062
Columella width       0.284    -0.286     0.153
Whorl diameter 1      0.246     0.059     0.169
Whorl breadth 1       0.247     0.058     0.119
Whorl height 1        0.273     0.462     0.394
Whorl diameter 2      0.331     0.057    -0.080
Whorl breadth 2       0.392     0.104    -0.286
Whorl height 2        0.351     0.350    -0.677
Shell thickness       0.311    -0.431     0.020

F-values(*):          53.84      7.22      4.05
Probabilities         0.0002    0.0002     0.001

  * From ANOVA of the seven shore-level-morph group means of principal
component scores. We randomized the data 4999 times and added the observed
ordering to calculate the probabilities.

The elements of the first vector were all positive and of similar magnitude (table 1). In addition, correlations between original values for all characters and the scores of the first principal component were significant (all P [much less than] 0.01). We therefore conclude that the scores of the first component mainly express variation caused by shell size.

The mean principal component (PC) scores of shore-level-morph groups differ significantly for PC 1, PC 2, and PC 3. The multiple-comparison test revealed that the ridged and banded morphs from upper and midshores were on average larger than snails from the other groups.

We could not detect any clear pattern of differences in PC 2 although the smooth and unhanded morphs had less pronounced aperture lips, thinner shells, and higher whorls than the other morphs.

Shell Height

Shell size, represented by its correlate shell height, was a major difference between lower- and upper-shore snails in all localities. The term for transect nested within locality and shore level is also significant. This means that shell height varied on a smaller spatial scale than our samples within localities represented.

The ridged and banded midshore snails were, on the average, larger than snails of the two other morphs when we broke up the midshore samples into their three different morphological groups. Nonetheless, the average size of ridged and banded midshore snails was smaller than that of upper-shore snails, whereas smooth and unbanded midshore snails were similar in size to the lower-shore snails. Many of the mixed snails from the midshore were only about 3 mm large and had probably not yet developed their final rib and banding character states.

Thus, the analyses of metric shell characters TABULAR DATA OMITTED indicate that there are large differences in size and minor differences in form between ridged and banded and smooth and unhanded morphs of the upper and lower shores. Some of these differences (in particular size differences) remain in the midshore zone where the two forms overlap in distribution.

Genetic Variation

Details of genetic variation of all samples can be found in Appendix 1. We first considered all upper- and lower-shore samples from the six transects. A gene diversity analysis of these samples revealed that between 3.7% and 15% of the variation was caused by differentiation between samples. Averaged over the five loci the differentiation among samples within geographic areas ([G.sub.SA]) was about as large as the differentiation among geographic areas ([G.sub.AT]). Furthermore, differentiation among samples within habitats (i.e., differentiation among upper-shore samples and among lower-shore samples, [G.sub.SH]) was even more important than the differentiation caused by the samples being from different microhabitats (upper and lower shores, [G.sub.HT]). This indicates the near lack of consistent differences between samples from different habitats.

Interestingly though, a cluster analysis of the same set of samples revealed that the most genetically related snails always came from samples from the same shore level and from the same geographic site. As the length of each vertical transect was approximately the same as the distance between two replicate transects, we may conclude that on a microgeographic scale the vertical (between microhabitat) genetic differentiation was somewhat greater than the horizontal (within microhabitat) genetic differentiation. However, the upper- and lower-shore samples did not form two separate groups at the TABULAR DATA OMITTED next level of branching. This means that the pattern of genetic differentiation on a large geographic scale was unrelated to shore level, and thus to the morphology of the snails. (See also the results of Mantel's test below.)

When we analyzed the midshore samples we first divided each sample into groups of each morph. But here we found no consistent trend in the result of the clustering analysis. Sometimes the groups of snails with the same morphology from replicate samples joined, and sometimes groups of snails with different morphologies but from the same original sample were most genetically similar. The lack of genetic differentiation between morphs on a macrogeographic scale was evident also in this analysis.

The results of Mantel's comparison of the morphological and genetic distance matrices indicated that there was no positive correlation between morphological and genetic differentiation on a geographic scale (r = 0.088, P = 0.29). This supports the view of one species, extremely polymorphic in some shell characters.

Genetic Variation in the Aat-1 Locus

The allele frequency distribution in Aat-1 suggested selection as a possible source of variation in this locus, as the fast allele Aat-[I.sup.120] increased downshore in at least three transects (Corrubedo Cape A: from 0.17 to 0.44 over 10 m, Corrubedo Cape B: from 0.11 to 0.47 over 13 m, and Silleiro Cape A: from 0.47 to 0.87 over 20 m, see Appendix 1, P [is less than] 0.001 for all in Workman and Niswander's (1970) gene contingency [[Chi].sup.2] test). The difference between upper- and lower-shore samples in Willeiro Cape B was not significant ([[Chi].sup.2] = 3.04, df = 1), but this was probably a Type II error caused by the low sample size of the upper-shore sample (N = 8).

The frequencies of the 100 and 120 alleles of the midshore samples were in between the frequencies of the upper- and lower-shore samples. In Corrubedo and Silleiro Capes the Aat-[I.sup.120] allele was, however, always more common among smooth and unbanded snails than among ridged and banded snails from the same midshore sample (P [is less than] 0.05 in gene contingency [[Chi].sup.2] tests of the four cases, Appendix 2).

Heredity of Morphological Characters

We reared 187 offspring from 11 ridged and banded females kept in cages at the laboratory. The largest ([is greater than] 3 mm), and thus presumably oldest offspring (105 individuals), had all developed ridges and bands. The 82 small ([is less than] 3 mm) offspring were, unlike their mothers, all smooth. However, all of them had developed bands except one that was unbanded. The nine large ([is greater than] 3 mm) offspring, which we were able to rear from six smooth and unbanded females, were all smooth and unbanded. However, also this time all 45 small ([is less than] 3 mm) offspring were smooth and unbanded. Thus, the differences in morphological characters used to distinguish between upper-shore and lower-shore forms of Galician Littorina saxatilis were probably inherited, but, at least in the laboratory, they did not develop until after a shell height of approximately 3 mm.

The regressions of sculpture depth on shell size differed significantly in slope among groups of siblings from different ridged and banded mothers (fig. 8, ANCOVA; [F.sub.4,56] = 4.23, P = 0.007). This suggests a genetic component of sculpture depth variation as well.


We found two distinct morphological forms of Littorina saxatilis in our samples from exposed rocky shores in northwestern Spain. They dominated on different

levels of the littoral zone, but occurred in about equal proportions in a narrow central zone where we also found low frequencies of intermediates. In a common environment, differences in shell banding pattern and spiral ridging remained in the [F.sub.1] generation, and they had an association with differences in shell size and shape.

Differences in shell ornamentation and color between the two Galician forms developed at a shell height of about 3 mm in the laboratory. The ontogenetic changes in color pattern that we found among the smooth and unbanded snails (i.e., bands that break up to a tessellated pattern) have been described earlier in British Littorina neglecta populations (Hannaford Ellis 1984). However, Struhsaker (1968) found that the degree of shell ornamentation already differed in the pelagic larvae of Littorina picta, another species polymorphic for this character.

In our experiments in a common laboratory environment, all surviving offspring of ridged mothers became ridged at an approximate shell height of 3 mm, whereas offspring of smooth mothers remained smooth. This suggested inherited differences in shell ornamentation between the two morphs. Boulding et al. (1993) report shell ornamentation to be a highly plastic character in Littorina sitkana Philippi, as laboratory-reared offspring of deeply ridged parents developed ornamentation only if cultured at a slow growth rate. This reminds us of the problem of extrapolating laboratory results to natural populations, which is indeed impossible if we have not studied norms of reactions for the sets of environments or genotype frequencies that correspond to those in the populations that we are interested in (see, e.g., Lewontin 1974; Sultan 1987).

Although the presence of two distinct forms and few intermediates in the border zones is compatible with a model of two hybridizing species, an alternative interpretation is that the morphologyical transitions are intraspecific and caused by sharp selection gradients. Because we found no positive correlation between morphological and genetic distances among geographically separated populations, we reject the alternative that the two forms are different species. We conclude instead that they are ecotypes (sensu Turesson 1922) within the same species (L. saxatilis).

Studies made in the last decade have revealed sharp phenotypic gradients in populations of L. saxatilis from other parts of its distribution. Janson (1982a) describes horizontal clines between alternating rocky and boulder microhabitats over 1 km of rocky shore in Sweden. Also in this case, there was no correlation between genetic and morphological differences among subpopulations although significant genetic heterogeneity was present among subpopulations only meters apart (Janson and Ward 1984).

The presence of microclines in this species is most probably caused by a poor dispersal capacity with no pelagic larvae and monthly migration distances within a range of only a few meters (Janson 1983). Poor dispersal is, however, obviously not enough to cause phenotypic clines of the magnitude reported here, but in combination with differential selection over sharp microenvironmental gradients we might expect strong phenotypic clines (e.g., Felsenstein 1976, p. 260; Endler 1977, pp. 58-61). Selection and not drift must be the operational force because the morphological characters converge in similar environments.

Today we know nothing about how selection acts to maintain the dimorphism in the Galician populations of L. saxatilis, but the Swedish clines have been somewhat investigated in this respect. In Sweden, the switch from one microenvironment to another occurs over 5 to 10 m (Janson and Sundberg 1983; pers. obs.), and in the transition zones snails with intermediate characters are most numerous (Janson and Sundberg 1983). Transplant experiments suggest not only that the two extreme forms (exposed rocky form and sheltered boulder form) are selected against, if transplanted to the wrong side of the cline, but also that the intermediate forms have survival rates roughly twice those of the extreme forms in the middle of the cline (Janson 1983).

In the Galician shores, there is an abrupt transition between the two distinct microhabitats, and in contrast to the situation in Sweden, nowhere did we find that snails of the mixed form dominated in numbers. (Ten percent to 30% of the midshore individuals had mixed form, and these figures may be overestimates as the intermediate samples included many small snails that may not have developed ornamentation or tessellation.) Different occurrences of intermediate forms in Swedish and Galician transitional zones may indicate crucial differences in the evolution of the populations. By a computer iterated model Johannesson and Sundberg (1992) conclude that the polymorphism found in Sweden, with intermediate snails being superior in the middle of the cline, will not evolve to speciation. Is Galicia a more likely place for sympatric speciation events in L. saxatilis?

An analysis of mating structure of Galician midshore populations of L. saxatilis indicates nonrandom mating between ridged and banded and smooth and unbanded forms with fewer interform mating pairs found than expected from chance (Johannesson et al. 1994). This suggests that a partial barrier to gene flow was already present.

The genetic variation found in the Aat-1 locus gives further support for an interrupted gene flow among upper- and lower-shore subpopulations. The association between allozyme frequencies and shore levels present at two sites (Corrubedo and Silleiro Capes) seems likely to be caused by selection because even stronger cross-shore clines have been found in other populations of L. saxatilis (Sweden, Iceland, Norway, and Isle of Man; Johannesson and Johannesson 1989). (Noticeably, however, the directions of the Galician clines are opposite to all the others!) However, the variation in Aat-1 persisted in the midshore samples, that is, the frequencies of the alleles differed significantly between smooth and unbanded and ridged and banded forms in four of six samples. This supports the existence of a partial reproductive barrier between the extreme forms, or less likely, a genetic coupling between morphological characters and genotype of this locus.

The nature of this partial mating barrier is not yet clear, but size differences among ridged and banded and smooth and unbanded snails may be important (Johannesson et al. 1994). Although the size differences are both obvious and consistent among sites, we do not know the factors behind them or their temporal variation (McCormack 1982). However, the cause for the size differences may in part be caused by non-genetic growth factors and in part by genetic differences in growth rates, as in Swedish L. saxatilis populations (Janson 1982b).

Independent of the ultimate causes of size differences the presence of such between potential mates may cause assortative mating if, for any reason, large males mate with large females more often than with small females, and vice versa. Such positive relationships between the sizes of mates have been found in a Swedish population of L. saxatilis (Saur 1990) and in three of six Galician subpopulations of this species (Johannesson et al. 1994).

Obviously, a reduced gene flow may be the first step towards complete isolation between the ridged and banded and the smooth and unbanded forms of L. saxatilis in this area. However, the barrier is incomplete now and strong reinforcement is needed to reach levels of genetic isolation that are comparable with the effect of isolation by distance between geographically separated populations of the species (e.g., Knight et al. 1987). Although we are fairly convinced that the limited dispersal capacity of a direct developing species like L. saxatilis does indeed support the development of extensive polymorphisms, it is difficult to guess how often these polymorphisms may evolve into sympatric speciation (Maynard Smith 1966). Possibly sympatric speciation, besides allopatric speciation, may contribute substantially to raise the speciation rate among clades with nonpelagic development.


We thank M. Lindegarth for writing two of the computer programs for the randomization tests, U. Stenfelt for laboratory aid, A. Falck-Wahlstrom for drawing figure 1 and P. R. Jonsson, P. Nilsson, and M. Lindegarth for statistical discussions. We received many useful comments from R. D. Ward, E. G. Boulding, and one anonymous reviewer. We received financial support from grants from the Swedish Natural Science Research Council (to K.J.), the Xunta de Galicia, Spain (to E.R.A.), and the foundation of Collianders (to K.J. and B.J.).


Airoldi, J., and B. K. Flury. 1988. An application of common principal component analysis to cranial morphometry of Microtus californicus and M. ochrogaster (Mammalia, Rodentia). Journal of Zoology (London) 216:21-36.

Appleton, R. D., and A. R. Palmer. 1988. Waterborne stimuli released by predatory crabs and damaged prey induce more predator-resistant shells in a marine gastropod. Proceedings of the National Academy of Sciences, USA 85:4387-4391.

Atkinson, W. D., and T. Warwick. 1983. The role of selection in the colour polymorphism of Littorina rudis Maton and Littorina arcana Hannaford-Ellis (Prosobranchia: Littorinidae). Biological Journal of the Linnean Society 20:137-151.

Boulding, E. G., J. Buckland-Nicks, and K. L. Van Alstyne. 1993. Morphological and allozyme variation in Littorina sitkana Philippi and related Littorina species from Northeastern Pacific. Veliger 36:43-68.

Boulding, E. G., and T. K. Hay. 1993. Quantitative genetics of shell form of an intertidal snail: constraints on short-term response to selection. Evolution 47:576-592.

Chakraborty, R. 1980. Gene diversity analysis in nested subdivided populations. Appendix 1 in R. Beckwitt. Genetic structure of Pileolaria pseudomilitaris (Polychaeta: Spirorbidae). Genetics 96:711-726.

Edgington, E. S. 1987. Randomization tests. Marcel Dekker, New York.

Endler, J. A. 1977. Geographic variation, speciation, and clines. Princeton University Press, Princeton, N.J.

-----. 1989. Conceptual and other problems in speciation. Pp. 625-648 in D. Otte and J. A. Endler, eds. Speciation and its consequences. Sinauer, Sunderland, Mass.

Faller-Fritsch, R. J., and R. H. Emson. 1985. Causes and patterns of mortality in Littorina rudis (Maton) in relation to intraspecific variation: a review. Pp. 157-177 in P. G. Moore, and R. Seed, eds. The ecology of rocky coasts. St. Edmundsbury Press, Bury St. Edmunds, UK.

Felsenstein, J. 1976. The theoretical population genetics of variable selection and migration. Annual Review of Genetics 10:253-280.

Fish, J. D., and L. Sharp. 1985. The ecology of the periwinkle, Littorina neglecta Bean. Pp. 143-156 in P. G. Moore and R. Seed, eds. The ecology of rocky coasts. St. Edmundsbury Press, Bury St. Edmunds, UK.

Fretter, V., and A. Graham. 1980. The prosobranch molluscs of Britain and Denmark. Part 5. Marine Littorinacea. Journal of Molluscan Studies, Suppl. 7:243-284.

Grahame, J., and P. J. Mill. 1986. Relative size of the foot of two species of Littorina on a rocky shore in Wales. Journal of Zoology (London) 208:229-236.

Hannaford Ellis, C. J. 1979. Morphology of the oviparous rough winkle, Littorina arcana Hannaford Ellis 1978, with notes on the taxonomy of the L. saxatilis species-complex (Prosobranchia: Littorinidae). Journal of Conchology 30:43-56.

-----. 1984. Ontogenetic change of shell colour patterns in Littorina neglecta Bean (1844). Journal of Conchology 31:343-347.

Hazel, W. N., and M. S. Johnson. 1990. Microhabitat choice and polymorphism in the land snail Theba pisana (Muller). Heredity 65:449-454.

Heller, J. 1975. The taxonomy of some British Littorina species, with notes on their reproduction (Mollusca: Prosobranchia). Zoological Journal of the Linnean Society 56:131-151.

Jain, S. K., and A. D. Bradshaw. 1966. Evolutionary divergence among adjacent plant populations. I. The evidence and its theoretical analysis. Heredity 21:407-441.

Janson, K. 1982a. Phenotypic differentiation in Littorina saxatilis Olivi (Mollusca, Prosobranchia) in a small area on the Swedish west coast. Journal of Molluscan Studies 48:167-173.

-----. 1982b. Genetic and environmental effects on the growth rate of Littorina saxatilis. Marine Biology 69:73-78.

-----. 1983. Selection and migration in two distinct phenotypes of Littorina saxatilis in Sweden. Oecologia 59:58-61.

-----. 1987. Allozyme and shell variation in two marine snails (Littorina, Prosobranchia) with different dispersal abilities. Biological Journal of the Linnean Society 30:245-256.

Janson, K., and P. Sundberg. 1983. Multivariate morphometric analysis of two varieties of Littorina saxatilis from the Swedish west coast. Marine Biology 74:49-53.

Janson, K., and R. D. Ward. 1984. Microgeographic variation in allozyme and shell characters in Littorina saxatilis Olivi (Prosobranchia: Littorinidae). Biological Journal of the Linnean Society 22:289-307.

-----. 1985. The taxonomic status of Littorina tenebrosa Montagu as assessed by morphological and genetic analyses. Journal of Conchology 32:9-15.

Johannesson, B., and K. Johannesson. 1990a. Littorina neglecta Bean, a morphological form within the variable species Littorina saxatilis (Olivi)? Hydrobiologia 193:71-87.

Johannesson, K., and B. Johannesson. 1989. Differences in allele frequencies of Aat between high- and mid-rocky shore populations of Littorina saxatilis (Olivi) suggest selection in this enzyme locus. Genetical Research 57:7-11.

-----. 1990. Genetic variation within Littorina saxatilis (Olivi) and Littorina neglecta Bean: Is L. neglects a good species? Hydrobiologia 193:89-97.

Johannesson, K., and P. Sundberg. 1992. Speciation in Littorina saxatilis (Olivi)?--a one-dimensional selection-migration model. Pp. 1-8 in J. Grahame, P. J. Mill, and D. G. Reid, eds. Proceedings of the Third International Symposium on Littorinid Biology. Malacological Society of London, London.

Johannesson, K., and T. Warmoes. 1990. Rapid colonization of Belgian breakwaters by the direct developer, Littorina saxatilis (Olivi) (Prosobranchia, Mollusca). Hydrobiologia 193:99-108.

Johannesson, K., E. Rolan-Alvarez, and A. Ekendahl. 1994. Incipient reproductive isolation between two sympatric morphs of the intertidal snail Littorina saxatilis. Evolution. In press.

Jolicoeur, P. 1963. The multivariate generalization of the allometry equation. Biometrics 19:497-499.

Kemp, P., and M. D. Bertness. 1984. Snail shape and growth rates: evidence for plastic shell allometry in Littorina littorea. Proceedings of the National Academy of Sciences USA 81:811-813.

Knight, A. J., R. N. Hughes, and R. D. Ward. 1987. A striking example of the founder effect in the mollusc Littorina saxatilis. Biological Journal of the Linnean Society 32:417-426.

Knight, A. J., and R. D. Ward. 1986. Purine nucleoside phosphorylase polymorphism in the genus Littorina (Prosobranchia: Mollusca). Biochemical Genetics 24:405-413.

-----. 1991. The genetic relationships of three taxa in the Littorina saxatilis species complex (Prosobranchia: Littorinidae). Journal of Molluscan Studies 57:81-91.

Lewontin, R. C. 1974. The analysis of variance and the analysis of causes. American Journal of Human Genetics 26:400-411.

Manly, B. F. J. 1985. The statistics of natural selection. Chapman and Hall, London.

-----. 1986. Multivariate statistical methods: a primer. Chapman and Hall, London.

-----. 1991. Randomization and Monte Carlo methods in biology. Chapman and Hall, London.

Maynard Smith, J. 1966. Sympatric speciation. American Naturalist 100:637-650.

McCormack, S. M. D. 1982. The maintenance of shore-level size gradients in an intertidal snail (Littorina sitkana). Oecologia 54:177-183.

Mitchell-Olds, T. 1992. Does environmental variation maintain genetic variation? A question of scale. Trends in Ecology and Evolution 7:397-398.

Murray, J., and B. Clarke. 1984. Movement and gene flow in Partula taeniata. Malacologia 25:343-348.

Naylor, R., and M. Begon. 1982. Variation within and between populations of Littorina nigrolineata Gray on Holy Island, Anglesey. Journal of Conchology 31:17-30.

Nei, M. 1972. Genetic distance between populations. American Naturalist 106:283-292.

-----. 1973. Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences, USA 70:3321-3323.

Newkirk, G. F., and R. W. Doyle. 1975. Genetic analysis of shell-shape variation in Littorina saxatilis on an environmental cline. Marine Biology 30:227-237.

Palmer, A. R. 1990. Effect of crab effluent and scent of damaged conspecifics on feeding, growth, and shell morphology of the Atlantic Dogwhelk Nucella lapillus. Hydrobiologia 193:155-182.

Raffaelli, D. G. 1982. Recent ecological research on some European species of Littorina. Journal of Molluscan Studies 48:342-354.

Reid, D. G. 1990a. A cladistic phylogeny of the genus Littorina (Gastropoda): implications for evolution of reproductive strategies and for classification. Hydrobiologia 193:1-19.

-----. 1990b. Trans-Arctic migration and speciation induced by climatic change: the biogeography of Littorina (Mollusca, Gastropoda). Bulletin Marine Science 47:35-49.

Reimchen, T. E. 1981. Microgeographical variation in Littorina mariae Sacchi & Rastelli and a taxonomic consideration. Journal of Conchology 30: 341-350.

Saur, M. 1990. Mate discrimination in Littorina littorea (L.) and L. saxatilis (Olivi) (Mollusca: Prosobranchia). Hydrobiologia 193:261-270.

Smith, J. E. 1981. The natural history and taxonomy of shell variation in the periwinkles Littorina saxatilis and Littorina rudis. Journal of the Marine Biological Association, UK 61:215-242.

Struhsaker, J. W. 1968. Selection mechanisms associated with intraspecific shell variation in Littorina picta. Evolution 22:459-480.

Sultan, S. E. 1987. Evolutionary implications of phenotypic plasticity in plants. Pp. 127-178 in M. K. Hecht et al., eds. Evolutionary biology, Vol. 21. Plenum Press, N.Y.

Sundberg, P., A. J. Knight, R. D. Ward, and K. Johannesson. 1990. Estimating the phylogeny in mollusc Littorina saxatilis (Olivi) from enzyme data: methodological considerations. Hydrobiologia 193:29-40.

Swofford, D. L., and R. K. Selander. 1981. BIOSYS-I: a Fortran program for the comprehensive analysis of electrophoretic data in population genetics and systematics. Journal of Heredity 72:281-283.

Thorpe, J. P. 1982. The molecular clock hypothesis: biochemical evolution, genetic differentiation, and systematics. Annual Review of Ecology and Systematics 13:139-168.

Tissot, B. N. 1988. Multivariate analysis. Pp. 35-51 in M. L. McKinney, ed. Heterochrony in evolution: a Multidisiplinary Approach. Plenum Press, NY.

Turesson, G. 1922. The genotypical response of the plant species to the habitat. Hereditas 3:211-350.

Ward, R. D. and K. Janson. 1985. A genetic analysis of sympatric subpopulations of the sibling species Littorina saxatilis Olivi and Littorina arcana Hannaford Ellis. Journal of Molluscan Studies 51:86-94.

Ward, R. D., and T. Warwick. 1980. Genetic differentiation in the molluscan species Littorina rudis and Littorina arcana (Prosobranchia: Littorinidae). Biological Journal of the Linnean Society 14:417-428.

Ward, R. D., T. Warwick, and A. J. Knight. 1986. Genetic analysis of ten polymorphic enzyme loci in Littorina saxatilis (Prosobranchia: Mollusca). Heredity 57:233-241.

Warmoes, T. 1986. A preliminary systematic and taxonomic study on the genus Littorina (Gastropoda, Prosobranchia). Licentiaatsthesis, Universitaire Instelling Antwerpen, Antwerp.

Workman, P. L., and J. D. Niswander. 1970. Populations studies on southwestern Indian tribes. II. Local genetic differentiation in the Papago. American Journal of Human Genetics 22:24-49.
COPYRIGHT 1993 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Johannesson, Kerstin; Johannesson, Bo; Rolan-Alvarez, Emilio
Date:Dec 1, 1993
Previous Article:Gene interaction affects the additive genetic variance in subdivided populations with migration and extinction.
Next Article:Genetic correlation between a female mating preference and the preferred male character in seaweed flies (Coelopa frigida).

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters