Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Shrews on small islands: epigenetic variation elucidates population stability

Shrews on small islands: epigenetic variation elucidates population stability Hanski. I. and Kuitunen, J. 1986. Shrews on small islands: epigenetic variation elucidates population stability. - Holarct. Ecol, 9: 193-204, Three large (4 to 8 ha) and 14 small islands (0.3 to 2 ha) in a lake in eastern Finland, all situated less than 0,5 km from the mainland, were surveyed for small mammals. Three species of shrew and two species of vole were resident in July 1982: Sorex araneus on 10. S, caecuiiens on 2. S- mimiius on 5, Microius agrestis on 12 and Clelhriotiomys glareolus on 4 islands. Immigrants were trapped from tiny islets, ;uid the data indicate that S. caecutiens and M. agre.sris are better dispersers than S. minulus and C. glareotus. respectively. Microuis agrestis. S. araneus and C. ^iareolus occurred nonrandomly. on subsets of the larger islands, while the two small Sorex species occurred more erratically, possibly because of competition with S. araneus. Juvenile sex ratio was miile-biased on the mainland but female-biased on large islands, possibly because juvenile males move more and are more likely to emigrate from an island than juvenile females. Metrical and non-metrical (epigenetic) cranial traits gave similar patterns of population differentiation in S. araneus. Two of the three large-island populations have differentiated from the mainland populations and from each other, suggesting that the populations are relatively slable. Small-island populations, which are often less than 10 individuals in size, showed liltlc differentiation but had more epigenetic traits fixed than large-island and mainland populations (founder effect). This suggests that the small-island populations are unstable, have a high extinction rate. /. Hanski and J. Kuitunen, DepI of Zoology, Univ. of Helsinki, P. Rautatiekaiu 13. SF'OOWO Helsinki. Finland. I. Introduction PopuUitions of small mammals on real or habitat islands provide an opportunity to study the classical biogcographic relationships (MacArthur and Wilson 1967) of insular cotnmunities (Beer et al. 1954. Webb 1965, Lomolino 1982. l9S4a. Adier and Wilsoti 1985), but such studies may also elucidate the spatial dynamics of populations, e.g. extinction of small populations (Smith 1980. Pokki 1981). dispersal (Smith 1974. Pokki 1981, Lomolino 1984b). establishment of populations (Crowcll 1973), ititra- and interspecific competition (Kostian 1970) and predation (Lomolino i984b). as well as genetic differentiation of local populations (Berry and Warwick 1974. Berry et aL 1978. Sikorski 1982). These aspects of population dynamics pose particular diffiAcecpted 12 November 1985 © HOIARCTtC ECOLOGY culties for anyone studying Soricinae. because of the small size and covert habits of shrews, Radiotelemetry seems to be impractical for so small mammals (M. Genoud. pers. comm.). Island population studies cannot answer many questions about individual behaviour, but at the population level they can be useful. Here we report on trapping results of small mammals on the 17 small islands of a lake in eastern Finland, in a region with six species of shrew {Sorex and Neomys) and two speeies of vole (Microtus agrestis and Clethrionomys glareolus). Our main interest lies in the shrews, for which little has been published about occurrence on smiill islands (the ecologically interesting scale), though some information exists for large, mostly oceanic islands (the evolutionary interesting scale: McTaggart Cowan 1941. Corbet 1966, Olsen 1969, Kock 1974, 01- HOl.ARCriC ECOLOGY 9:3 (1986) ert 1975. Okhotina 1976. 1977. Spitzenberger 1977, Grainger and Fairley 1978. Pieper 1978, Ellenbroek 1980. Jensen et al. 1980. Roberts 1980. Thomas et al. 1980, Michielson et al, 1980. O'Keeffe and Fairley 1981. Yalden 1981, Malmqvist 1985). The purpose of this paper is to describe the occurrence of the species on the islands in 1982. prior to experimental perturbations (work started in 1983), and to try to tie together ecological and morphometrical data on the island populations. Metrical and non-metrical (epigenetic) morphological variation may help to understand the dynamics of populations on small islands, where genetic drift may be assumed to be the main mechanism of differentiation. Fig. 2. The 6 metrical measures of the lower jaw. 2. Material and methods 2.1. Field work Lake Sysma in eastern Finland (Ilomantsi, 63°N. 32°E) has three "large" islands (3.8 to 8.0 ha). 14 "small" islands (0.3 to 2.2 ha), and 25 tiny islets, less than 01 ha in size. All islands have a maximum elevation of less than 10 m and most less than 2 m. Isolation varies from 10 to 500 m (Fig. 1). All islands and most islets have a more than 100 years old. mixed coniferous forest cover: pine Pinus sylvesiris and spruce Picea abies with some birch Betula spp. The dwarf sbrubs Ledum and Vaccinium are abundant in the field layer, and the humus layer is very thick. Inevitably, there are some minor habitat differences, but generally the islands have a uniform forest cover. All islands, most islets, and 5 mainland study sites (Fig. 1) representing the same forest vegetation were trapped with metal cones {Pankakoski 1979) for usually three nights in July 1982. Mainland sites were trapped with 3(l-cone traplines, traps spaced 10 m apart. Large islands were trapped with 50 to 1(K) cones, small islands with 15 to 30 cones, and islets with 3 to 10 cones, depending on island size (Tab. 1), Traps were spread evenly across the islands. Two small islands (S4 and S12) and one islet (15) were trapped more intensively to remove all shrews (details in Section 3,2.). Shrews were preserved in 70'^ alcohol. 2.2. Metrical traits Fig. I. A map of the study area, lake Sysma in eastern Finland, Ml-5 iire 5 mainland study sites, Ll-3 are "large" islands (4 to K ha. light shading). Sl-14 arc -small" islands (0,3 to 2 ha. strong shading), and the small numbers from 1 to 25 indicate the locations of tiny islets, less than 0.1 ha (11-25 in the text). Body weight was measured to the nearest 0.1 g with a spring balance after the shrew had been dried for 5 min on blotting paper. A weight meastirement taken after preservation in alcohol is not comparable to live weight but allows comparisons within this set of data. Hind foot length was measured to the nearest 0.1 mm with a ruler. Following these measurements tbe animals were sexed and aged (young of the year vs. overwintered) and the breeding state of females was reeorded. Skinned carcasses were boiled for 5 min. and the skeletons were prepared enzymatically. in papain (2-4 d at 38°C. as recommended by Searie 1954). The following cranial measurements were taken under a dissecting microscope with an ocular micrometer (accuracy 1-2%); Condylobasal length. Six bilateral measures of the lower Jaw (Fig. 2). The average value of the two sides was used in the analysis. HOLARCTIC ECOLOGY 2.3. Non^metrical traits Ihirty-tour traits were scored, mostly following Peter King's (pers. comm.) work on the common shrew in the United Kingdom. See also Berry and Searle (1963). The traits are described in Appendix A. 2.4. Statistical methods Non-metrical divergence between populations was measured with C. A. B. Smith's Mean Measure of Divergence. MMD. Sjovold (1977) gives a thorough statistical treatment of the MMD, and we have followed elosely his recommendations. Side-by-side eorrelation in the manifestation of bilateral traits was measured with the rp cocfficicnl of Pearson. Correlation between the trait manifestation and size was tested with the point-biserial correlation coefficient of Pearson, r^,, with the condylobasal length of the skull used as a measure of size. We did not ealculate correlations between the traits, because our material is relatively small and previous studies have shown that such correlations, if they exist, are generally small (Truslove 1961. Berry and Berry 1967. Kellock and Parsons 1970, Berry 1976, Sj0vold 1977. Sikorski 1982). Sj0vold (1977) concludes that "with regard to sample sizes available in general, eorrelation does not affect the results to a greater extent than would random fluctuations from independent variables." Metrical traits were first analysed with univariate ANOVA, then with stepwise discriminant function analysis (HYLPS programs of the Computing Centre of the University of Helsinki). i. Results 3.1. Species composition on islands Trapping results are summarized in Tab. I. Sorex isodon was caught from the mainland sites M2 (3 individuals) and M4 (I), and S. minutissimu.s from M4 (2), but Neomus fodicns was not found in the study area until 1983 (Hanski unpubl.). These three species were not re- Tab. I. Trapping results for mainland study sites (MI-5), large (Ll-3) and small islands (S1-I4), and islets (11-24) in July 1982 (for their iocalions see Fig. 1). This table gives relative densities (individuals per KX) trap-nights) or actual numbers caught (figures in brackets) for Sara - Sorex araneus, Scae - .V. caeculiens. Smin - .V. minufus, Magr - Microtus agrestis. and Cgia = Clelhrionomys glareolus. Islets 115, 120, and 125 were not trapped, other islets not included were empty. Population Ml M2 M3 M4 M3 LI L2 L3 SI S2 S3 S4 S.^ S6 S7 S8 S9 Sin SI I S12 S13 SI4 Area ha Traps M) 30 30 30 30 46 10() I(X) 30 20 20 160 20 20 15 25 20 20 15 90 25 20 Nights 4-5 3 4 3 3 3 3 3 3 3 3 6-7 3 3 3 3 3 3 3 7 3 3 Sara 24 Seae Smin Magr Cgla Trapping period 3.8 8.1) 4.4 2.2 0.7 0.8 0.8 0.7 1.2 0.4 1.0 0.8 0.4 0.3 0.3 1.0 0.7 8 13 10 - 9 (1) 4 (1) (1) (21) .. . 2 4 - 1 9 - 22 (1) 12 5 (1) 6 7 (I) 8 3 6 + 2 + (21) 8 42 (39) 12 3 (4) (1) (14) 4 _ 5 (1) 18 - 6 1 1 - (t) (I) (2) (3) 2 2 _ + — — — - II 12 13 15 16-8 110 III 113 122 124 (1) (3) (1) (4) (1) (2) 8-13.7 8-13.7 8-13.7 20-23.7 19-22.7 2()-23.7 18-21,7 19-22.7 6- 9.7 6- 9.7 6- 9.7 14-20.7 7-10,7 7-10.7 7-10.7 7-10.7 7-1(1.7 20-23.7 20-23.7 1.5-22.7 20-23.7 19-22.7 7-10.7 7-10.7 7-10.7 8-14.7 7-10.7 20-23.7 2()-23.7 2()-23.7 19-22.7 19-22.7 + The catches were lost but the species was present. tCOLOGY '>;3 (iy«i) • LOCAL POPULATION 3.2. Population densities O IMMIGRANTS Tab. 1. gives relative population densities, individuals per 100 trap-nights, for the mainland study sites and the islands. Two small islands, S4 (0.8 ha) and S12 (0.3 ha), and MINUTUS the islet 15, were intensively kill-trapped with cones in 0 0 0 CAECUTIENS 0 July 1982 to remove all shrews (Tab. 1). Equally inten0 0 ARANEUS sive trapping in May to June 1983 proved that the islands had been emptied (Hanski unpubl.; an adult male •O 00 CD O of Neomys fodiens was caught on S4 in May 1983. but it od AREA ho O O O r-' o o o o o o o had obviously moved there from the mainland). These ':^ ISLAND results suggest that trapping with the intensity of 150 Fig. 3. Presence of breeding populations and catches of probpitfalls per ha for 7 days can remove all shrews from an able immigrants on the islands, arranged in an order of inisland (even a lower intensity might be sufficient; cf. creasing island size. See Fig. 1 for the island locations. Aulak 1967, Pudek 1969, Pankakoski 1979). Densities may go up to at least 30-40 individuals per ha on small islands in late summer. The following are corded from the islands, but this is not surprising beminimum estimates (July 1982): S. caeculiens on S4. 26 cause of their rarity on the mainland. per ha, and S. minulus on S11,38 per ha. The densest isSorex araneus, S. caecutiens and S. minutus were land population was however S. araneus on LI (3.8 ha), present at all five mainland sites. Based on the numbers where 76 individuals were trapped with 46 cones in 3 of shrews trapped, and their age and sex (below), we nights; population size was probably 200-300 individuconclude that there were 10 breeding populations of S. als, density >50 individuals per ha. Apart from this araneus, 2 populations of S. caecutiens, and 5 popu- population, densities were not different between the islations of 5, minulus on the 17 islands (Tab. 1). The field lands and the mainland (Tab. I), though admittedly our vole Microtus agresiis and the bank vole Cleihrionomys mainland and island results are not strictly comparable. glareolus had populations on 12 and 4 islands, respeetively (Tab. 1). The field vole was uncommon on the mainland, though it was the dominant vole on small is- 3.3. Dispersal lands. Webb (1965) found the same for the ecological Overwintering and breeding are practically impossible equivalent Microtus pennsylvanicus in North America on the tiny islets, and any small mammal caught there is (see also Crowell 1973). an immigrant, except perhaps the field vole, which Species number increased with island size from I to 4 might manage to breed on some islets (cf. Pokki 1981). (Fig. 3). as expected (MacArthur and Wilson 1967, In view of the trapping intensity on islands, a single indiConnor and McCoy 1979; see Dueser and Brown 1981). vidual caught on an island in July is also likely to be an Lomolino 1984a for mammals), but this standard result immigrant, though we cannot exclude the possibility masks interesting biology. The field vole was present on that it belonged to an extremely small population, permost islands regardless of island size. The eommon haps of size 1. shrew was resident on most islands larger than 0.7 ha, Looking first at the numbers of shrews trapped on iswhile the bank vole was found on only the three largest lets, there are too many S. caeculiens and too few S. araislands and a single small one (Fig. 3). Island size is a neus and S. minutus in comparison with the mainland good predictor of the presence of the latter two species densities, the source of most immigrants, but the num(one-tailed Mann-Whitney U test, P <().O1 and 0.(125, bers are small and the deviations from the expeeted frerespeetively; cf. Schoener and Schoener 1983). In Min- quencies are nonsignificant (Tab. 2). Including the nesota (Beer et ai. 1954), M. pennsylvanicus occurred probable immigrants trapped on islands, there is a sigon islands of all sizes, down to (at least) 0.2 ha, while nificant difference between S. caeculiens and S. minutus Cleihrionomys gapperi and Peromyscus maniculalus (observed immigrants 10:1, expeeted 6.8: 4.2, P <0.05). were present on islands greater than 1 ha. Sorex araneus cannot be fairly compared because it had Compared with the above-mentioned speeies, the a local population on most islands and hence many postwo small Sorex show a different pattern. Excluding the sible immigrants cannot be identified as such. 5. caeculiens population on the largest island (L2), the The field vole is clearly a superior colonizer of small small shrews were absent as breeding populations from islands to the bank vole (Tab. 2: Ppkki 1981), Carter the 4 largest islands, though 6 populations were found and Merritt (1981) have experimentally confirmed the on smaller islands (Fig. 3). good swimming ability of Microtus pennsylvanicus. in conclusion, these data suggest that S. caecutiens is a better disperser than S. minutus, which is expected, as the smaller S. minutus has a shorter starvation time and GLAREOLUS AGRESTiS HOi.ARCTlC ECOLOGY 9.3 rial. The 9 metrical traits were tested for equality with respect to sex in S. araneus. No significant difference was found in any of the traits and data for the two sexes were pooled. Pi)pulations Sara Scae Smin Magr Cgia Univariate ANOVAs were run for 13 populations of .V. araneus (Ml-5. Ll-3, SL S3. S5. S8 and S14). 5 Mainland 110 31 19 9 13 populations of S. caecutiens (MI, M3, M5, L2 and S4) Islands 196 33 21 146* 24* and 4 populations of 5. minutus (M4. M5, S8 and SIO). Island Pooled sample sizes are 246, 41 and 15. respectively. immigrants 3+ 5 1 ? ? Untransformed and log-transformed analyses gave simiIslets 6 5 27 lar results. 'Slight underestimates because eatches from SI, S3, and S5 Variation among S. araneus populations was signifiwere lost (see Tab. I). cant (P <0.01) in all 9 metrical traits. In S. minutus, the F values were nonsignificant, except two jaw measures (2 and 3, P <().()5). while in S. caecutiens the following traits varied significantly (P <0.01) between the popuis more likely to be drowned while swimming than S. lations: hind foot length, and jaw measures 1, 2 and 6. caecutiens. However, with these data we cannot exclude the alternative explanation that S. caecutiens simply Sorex caecutiens had two island populations, on L2 moves more than S. minutus. with .V. araneus and on S4 alone. A (naive) competition hypothesis would predict that in the absence of S. minutus but in the presence of 5. araneus the L2 population 3.4. Demography has small individuals, while the S4 population, in the abThe trapping results give some information on breeding sence of S. araneus, has shifted towards tbe seemingly successful S. araneus niche. However, the opposite was and the sex ratio. We restrict this section to shrews. The frequency of adult female .V. araneus was 4 to 5% observed: in most metrical traits the L2 population conon the mainland and large islands in July 1982. Adult sisted of large while the S4 population of small indivimales comprised 20 to 25% on the mainland and on L2 duals. The 13 S. araneus populations were subjected to a ;md L3. but only 3% on LI (2 individuals). The difference between the islands is highly significant (x" — 14.5. stepwise discriminant function analysis. The first 2 funcdf - 2, P <(l.!)()l). It may be significant that LI had the tions accounted for 56 and 17% of discrimination, and densest S. araneus population of all islands. Immature were statistically significant. Fig. 4 shows the population shrews establish feeding territories (Michielson 1%6. means on the first 2 functions, and the correlations of Hawes 1977), and if these are densely packed, adult these functions with the original metrical traits. males, which have the greatest home ranges (MichielOverlap between the populations is extensive, less son 1%6, Hawes 1977), may suffer from frequent fights than half (467o) of the individuals being classified corand may have difficulty in obtaining enough food. rectly. Although the a posteriori classification is of limIn young S. araneus, the sex ratio was male-biased ited interest when overlap is so extensive, it is worth (58% males) at the five mainland sites hut female-biased (39% males) on targe islands. Differences between the mainland sites (x' = 3.41, df ^ 4. NS) and large islands (x" = 0.19, df = 2. NS) are nonsignificant, but the difference between the pooled mainland sample and the pooled large-island sample is significant (x' = 6.26, df = 1, P<().02). The simplest explanation is that juvenile males move M3 more than juvenile females, as in most mammals (see reviews by Greenwood 1980, Dobson 1982). This hyS3 pothesis explains male-biased trap-catches on ihe mainLI land, and it explains the female-biased sex ratio on isS'A S5 lands, assuming (1) that males, because they move more, are more likely than females to leave an island, ss and (2) tbat emigration from an island exceeds immigration to the island (almost certainly true). Tab. 2. Pooled numbers at the mainland sites, on islands and on islets (species abbreviations as in Tab. i). "Island immigranls" are individuals trapped on islands and considered to be immigrants (see the text). .V5. DifTcrentiation between populations: metrieal traits Metrical analyses were restricted to juveniles because of I he small number of overwintered shrews in the mateIIOI.ARCTIC ECOl OGY 9:3 (1986) Fig. 4. Discriminant function analysis of the island populations: Ihf first 2 functions The insert shows correlations between the functions and the metrical traits. See Fig, 1 for the island locations. have especially low scores on the first function (small individuals). To some extent these data conform to the Krebs' or island effect (Krebs et al. 1969): island populations are often dense and have large individuals (sec also Jewell 1966, Delany 1970). Our small-island populations are, however, so unstable that individual selection has little time to work towards larger individual size, and in fact we found small individuals on small islands. Small size offers tbe (perhaps slight) advantage to a small-island population that less resources are used per individual, hence extinction probability may be smaller (individual and/or group selection; see also discussion in Lomolino 1985). Thirteen young common shrews were trapped at sites not included in the discriminant function analysis. The result of this analysis can be used to classify these extra individuals into the 13 populations, to indicate dispersal sources. Sucb an exercise has heuristic value, though the result must be interpreted with caution. Most of the 13 movements implied (or relatednesses of small populations: islands S6 and S9), are reasonable (Fig. 5). Of particular interest is that the 3 individuals caught from 15 all have the strongest connection (the largest a posteriori probability) to a different surrounding population. This strongly suggests that these 3 individuals have in fact dispersed from other populations to the islet 15, and, thus have not been born there. Fig. 5. Discriminant function analysis suggests the movements or relatednesses between populations shown here for 13 individuals of S. araneus trapped at sites not included in the analysis. The thick line indicates the a posteriori probability (of an individual "belonging" to the population) greater than 0.7, the thin line indicates probabilities from 0.4 to 0.69. and the broken line shows probabilities less than 0.4. 3.6. DiFTerentiation between populations: non-metrieal traits Details of the calculations are given in Appendix B. This section is restricted to S. araneus, as sample sizes are small for the other species. The 3 large-island populations have all diverged more from the mainland populations than the remaining 10 populations (P <0.01), and the 5 small-islatid popu- noting that all except one of the mainland populations had a smaller proportion of correctly classified individuals than the island populations {Mann-Whitney U test, P <0.05). Tbe exceptional mainland population is M3, distinct also in the analysis of epigenetic traits (below). Of the 3 large-island populations, LI and L3 are more distinct (have more individuals classified correctly) than L2, and of the small islands SI and S3 have the most distinct populations. (The largest population LI has many individuals in the a posteriori classification, which may be partly an artefact, but the Mann-Whitney test above is conservative, because the mainland samples were generally the largest.) Mainland populations, excepting M3, delimit a relatively small area in the discriminant space, and most islands are situated outside this subspace. Of the 3 largeisland populations, L2 has not diverged from the mainland populations, while the others have, but to opposite directions on the 2nd function. The first discriminant function is strongly correlated with hind foot length and weight (Fig. 4). All 3 large-island populations have higher scores on this function (larger individuals) than the 5 small-island populations (P <0.05). SI and S3 198 • I MEDIAN MU FROM MAINLAND I MEDIAN MU FROM ISLANDS LARGE 12 L3 LI SMALL SI 56 53 55 Sli, ISLANDS Ml M2 M3 M4 Mb MAINLAND Fig. 6. Median measures of uniqueness (MLJ) of 13 .V, anmeus populations with respeet to the remaining island and mainland populations (non-metrical analysis). The MMU is the average of the MMDs with the populations involved. See Fig. 1 for the locations of the populations. HOLARCTIC ECOLOGY 9:3 (IV«6) Tab. 3. The number of constant traits in a random sample of 8 accurate answer to the question: how many traits would individtials from 13 populations of 5. araneus. For the locations show no variation in a random sample of 8 shrews? of the study sites see Fig. 1. Means and 95% confidence interTab. 3. summarizes the results. The 95% confidence vals were calculated with Monte Carlo simulation (see text). Population Mean Ml M2 3.48 2.44 1.82 3.09 3.37 3.98 95% CI 2-6 1-4 1-3 1-6 2-6 1-5 3-8 2-6 9-9 1-6 Mainland m. m Large Island Small Island LI L2 L3 SI S3 S14 intervals mostly overlap, but mainland populations have less traits fixed on average (2.84) than island populations (5.13; P <0.U5), Small-island populations tend to have more traits fixed than large-island populations but the difference is not significant. Trait B9 is a prime example of founder effect and/or genetic drift: it had a frequency between 0.07 and 0.19 on the mainland, it was absent (in the samples) from 4 islands, including one large one, but it had a high frequency on 2 small islands (0.25 and 0.36, Tab. Bl). 4. Discussion lations tend to be more differentiated than the 5 mainland one (P <().O5; Fig. 6). Median divergences of the island populations from other island populations are greater than their divergenees from mainland populations (Fig. 6), suggesting that island populations have diverged in different directions. There is no elear differenee between large- and small-island populations in this respect. One mainland population (M3) is clearly more differentiated than the remaining ones (Fig. 6). A careful look at the frequencies in Tab. Bl reveals 3 traits in which island and mainland populations appear to fall into distinct groups. All small-island frequencies are higher than or equal to the large-island and mainland frequencies in B14and D6 (for each trait P <0.01). Both traits were excluded from the calculation of the MMD because of a difference between age classes (B14) or correlation with size (D6). A surprising result, however, is that D6 was positively correlated with size, while small-island individuals tended to be small (Fig. 4) but had a high frequency of D6. All 5 mainland populations had higher or equal frequency of D4 than the 8 island populations (P <0.01). Additionally, all 5 smallisland populations had a lower frequency of B13 than the 3 large-island populations (P <().O5). In summary, there is an indication of a significant differenee between small-island and large-island or mainland populations in 20% of the 21 traits studied. The difference is between newly-established, small populations, and older, larger populations. Some populations in Tab. Bl show more variation than others, but a direct comparison may be misleading, because sample sizes are different: less variation is observed in a smaller sample, other things being equal. To correct for sample size, we used Monte Carlo simulation to reduce all samples to 8 (randomly selected) individuals, the observed minimum. Calculations were based on the number of sides, because most of the traits in Tab. Bl are bilateral, and the frequencies reported are averages of the two sides. This method gives a sufficiently 13- HOLARCTIC ECOLOGY ^;3( 1986) Metrical and epigenetic analyses of microdifferentiation gave much the same results, which suggests, as the two kinds of variation are independent of each other and hardly affected by the same non-genetic factors, that differences between the populations are primarily genetic. Reduced epigenetic variation in populations on small islands points to the same direction. Good agreement between metrical and epigenetic results has not always been reported (Berry 1968, Berry and Jakobson 1975, Berry et al. 1978). Why different studies should differ in this respect we do not know, but note that our spatial and temporal scales are smaller than in Berry's and his assoeiates' works. Amongst the mainland S. araneus populations, M3 showed the greatest differentiation in both metrical and epigenetic traits. The explanation is obvious when one looks at the map of the study area (Fig. 1): M3 is surrounded by dispersal barriers to the west (the lake) and to the north (a dry ridge and another lake). This result suggests that in other, less obvious cases a study of morphometrical variation may reveal unexpected dispersal barriers between populations. What are the ecological implications of the observed morphological differentiation of the island populations? First of all, the main mechanism of differentiation is likely to be genetic drift, though admittedly some traits showed significant differences between island and mainland populations (Section 3.6). If the populations would not go extinct, we would expect more differentiation on small than large islands, because of increasing rate of drift with decreasing effective population size. The epigenetic divergence of S. araneus was indeed smaller on the largest island (8 ha) than on the two other large islands (3.8 and 4.4 ha), even though isolation was the same (Fig. 1). Genetic drift in relatively stable populations explains this difference. Data for 17 5. araneus populations on islands greater than 2 ha in another lake gave the same result (Hanski 1986). The results for the small-island S. araneus populations are, however, opposite to the prediction: populations on small islands (0.7 to 2.2 ha) are less differentiated than populations on the two larger islands (3.8 and 4.4 ha). Isolation cannot explain this result (Fig. 1), but population stability can: if small-island populations have a high extinction probability, they have less time to diverge than large-island populations. The large number of constant epigenetic traits in small-island populations is likely to stem from the initial founder effect and is not inconsistent with this hypothesis. Although more work on both the ecology and morphometrics of the island populations is needed, and is in progress, these results suggest that morphometrical studies can help to understand the dynamics of semi-isolated populations. It would be nice to know the time periods needed to achieve the observed levels of differentiation in the island populations, but this of course is not known. Apart from the problems of relating differentiation in epigenetic traits to differentiation in genetic traits, for largely unknown reasons some populations diverge fast, e.g. Apodemus agrarius in parks in Warsaw has shown significant divergence in 30 years (Sikorski 1982), while others diverge slowly or not at all, e.g. a population of the same species in Edinburgh has not diverged in 100 years (Berry and Warwick 1974). Another complication is that the island populations are not entirely isolated (Skaren 1980 and Hanski 1986 present more data and discuss overwater dispersal in shrews). Something more positive can be said about the sizes of the populations that are relatively stable and show morphological divergence versus populations which appear unstable and show little divergence. Our data suggest a critical island size between 2 and 4 ha separating the relatively stable and the relatively unstable populations of S. araneus. Numbers are likely to fluctuate mostly between 20 and 200 on the large islands (4 to 8 ha), which means that demographic stochasticity is relatively unimportant, and the low of 20 individuals may be large enough to buffer the population against much of environmental stochasticity. In contrast, small islands, less than 2 ha in size, have populations that may reach the size of 50-100 individuals in late summer, but in early spring, before breeding has started, these populations are often very small. To give an example, S3 (0.8 ha) and S5 (0.7 ha), which were removal-trapped in May to June 1983, had S. araneus populations of 3 and 8 individuals, respectively (Hanski, unpubl.). Both populations had both sexes in spring 1983, and the females were pregnant and/or iactating, but clearly demographic stochasticity must be a major factor In the dynamics of such small populations. The above comments were limited to 5. araneus, the best competitor, and probably little affected by the other species. The data on the island occupancy (Fig. 3) clearly showed that the dynamics of the two small species, S. caecutiens and 5. minutus, was affected by something else apart from island size. Isolation cannot explain the near absence of the small species from the 4 largest islands for a number of reasons: (1) these islands are not more isolated than the smaller islands where S. 200 caecutiens and S. minutus were present; (2) the bank vole, a poor disperser (Crowell 1973), was present only on the three largest islands and one small one (Fig. 3); (3) two 5. caecutiens, most probably immigrants, were caught from L2 and L3, again suggesting that dispersal to these islands is not infrequent; and (4) in later years (1983-85), both S. caecutiens and S. minutus have been recorded as residents or immigrants on some of the large islands. Hanski (1986) describes the colonisation of LI by S. minutus in 1984. The most plausible hypothesis we can suggest is that interspecific competition has affected the distributions of the small species on the islands. It seems reasonable to assume that competition could cause the extinction of an inferior competitor on these islands, less than 8 ha in area, though we expect that spatial heterogeneity prevents exclusion taking place on much larger islands, as on the mainland. Competition is not less severe on small islands, but chance becomes important and leads to a proportion of empty islands, available for all species to colonize. We recognize that the competition hypothesis cannot be properly tested without experiments, which are now in progress. Acknowledgements - We thank T. and J.-L. Tast for conducting the field work. We are especially grateful to P. King for sending us information on the non-metrical traits that he has scored in the eommon shrew. O. Jarvinen, P. King, E. Pankakoski and R. Vaisancn made useful comments on the manuscript. The study was supported by the Academy of Finland. Appendix A: Non-metrical traits This appendix gives the non-metrical (epigenetic) traits studied. If not mentioned otherwise, the trait is bilateral, and presence is scored. See Fig. Al. XI. Degree of fusion of processus spinosus of coalesced sacral vertebrae (Dolgov 1961). Several variants can be distinguished (Fig. Al). The trait was dichotomized by combining variants 3-4 and 5-8; variants 1-2 were not recorded. Unilateral. Medial palatine foramen. Posterior palatine foramen (occasionally accessory posterior palatine foramen present). Accessory terminal palatine foramen, on rear end of palate. The main foramen was always present. Foramen ovale divided into two. Accessory foramen posterior to the basisphenoid foramen. Basisphenoid process, projection sharper than a right angle was scored. Median basioccipital foramen. Unilateral. Basioccipital foramen. Accessory hypoglossal foramina (1 or 2). Prcmaxillary foramina, including the ones posterior to a line drawn up from the mid-point of HOLARCTIC ECOLOGY a. Al. A2. A3. A4. A5. A6. A7. A8. A9. Bl. B5 B4 B3 B2 D3 Dl B1 V B14 CI B11 B12B14 B15 A1 A2 A4 A6 A9 B8 B10 "" B16 Fig. AI. The non-metrical traits. The different variants of the traits B14 and XI are shown separately. See text for full names. the 2nd unicuspid and anterior to the junction of the 3rd and 4th unicuspids. From none to 5 foramina. The trait was dichotomized by combining variants 0-1 and 2-5 foramina. B2. Anterior maxillary foramen (rarely 2 but scored merely as present or absent). B3. Accessory foramen outside infraorbital foramen. B4. Accessory foramen above the rim of infraorbital foramen. B5. Accessory foramen posterior to B4. B6. Accessory foramen posterior to lacrimai foramen. B7. Accessory foramen below lacrimai foramen. B8. Accessory posterior alisphenoid foramen (rarely 2 but scored as present or absent; rarely the main foramen absent but pooled with the variant without accessory foramen). B9. Pncutnatic foramen, between the two articular condyles. BIO. Anterolateral parietal foramen, near the edge of the bone. Bl I. Anterior lateral sinus foramen. Occasionally absent, or with I or 2 accessory foramina dorsal to the main one (not scored). HOLARCTIC ECOLOGY 9:3 (1986) B12. Mesial lateral sinus foramina. From none to 5 foramina. The trait was dichotomized by combining variants 0-1 and 2-5 foramina. B13. Squamosal-parietal suture. Occasionally a branch or branches extend upwards in the parietal, which variant we scored. B14. Posterior lateral sinus foramen. Several variants (Fig. A I). The trait was dichotomized by combining variants O-I and 2-5. B15. Accessary posterior lateral sinus foramen. B16. Pro-otic foramen. CI. Parieto-frontal foramen reduced to a slit. C2. Median supraoccipital foramen. Unilateral. Dl. Anterior accessory foramen to the major mental foramen (rarely two or three accessory foramina but pooled with the major variant). D2. Lower mental foramen. D3. Dentary canal. Several variants, we scored absence of the major canal. D4. Lower coranoid foramen. D5. Upper coranoid foramen. D6. Coranoid fossa cross-ridge. Appendix B: Non-metricaE analy.sis The pooled frequencies of the variants in the 34 nonmetrical traits are available by request from the senior author (IH) for juvenile and adult .S'. araneus, and for juvenile S, caecutiens and S. minutus. The two sides of the 31 bilateral traits bave been treated separately. All three species pairs differ highly significantly in 6 to 9 traits each, though not always in the same traits. Adult and juvenile 5. araneus bave similar frequencies, excepting B5 and B14. Marginally significant differences on one side only in the bilateral traits B4, B6, BIO, D4 and D5 cause less coticem and do not prevent pooling of the age classes. The frequencies of B4 and B5, wbich are situated close to each other (Fig. Al), arc higher in juvenile than adult 5. araneus. The significant difference in B14 is caused by an increase in the frequency of variant (I and a decrease in the frequency of variant 1 from juveniles to adults (see Fig. Al). Homogeneity of incidence between the two sexes in juvenile S. araneus was tested in the largest sample, from Island LI (71 individuals), and in the pooled mainland sample (78 individuals). Only one significant dif- Tab. BL Frequencies of 21 non-metrical traits in 13 populations of S. araneus (multiplied by UXl). ln bilateral traits the average of the two sides is given. For the locations of the populations see Fig. 1, and for the traits see Appendix A. n is ihe number of individuals studied (for some traits n is slightly smaller because the trait could not be scored on some skulls). Trait XI A2 A9 Bl B2 Ml M2 M3 M4 M5 LI Population L2 L3 SI S5 S8 60 S14 •Ul •Jl B4 B5 B6 ^ 5 23 5 32 55 3t 4t) B7 B8 By BI2 BL3 BI4 3K f5 > 0 'Ul ( 1 I ) •Jl BI5 C2 Dl D3 D4 D5 D6 n 36 5 HH) Tab. B2. The Mean Measure of Divergence (upper triangle) and its standard deviation (lower triangle) between 13 populations of 5, araneus. TTie MMDs that are at least twice as great as the standard deviation can be considered significant (bold face). L2 L3 LI S I S8 25 . S3 Populations S5 Si4 Ml M2 M3 M4 12 , 19 . 62 43 . . 38 . 55 . 0 3 -4,5 . -2.8 04 , -0.9 05 . -2.1 12 , 0 8 -4,6 . -L4 -0,2 -0,4 -2,5 0 4 -0,7 . -2,4 22 . L2 L3 LI SI S8 S3 7.3 8.3 3,7 14.9 60 , S5 SI4 Ml M2 M3 M4 M5 3,4 3.4 1.8 2,3 3,4 1.7 2.0 60 , 70 . 27 . 59 . 36 . 50 . 20 32 . . 5 8 5 0 -2.1 , 02 , . -2.4 07 , 0 5 -1.3 , 47 , -4,5 2 5 2.0 . 50 44 , . 7 4 -0,5 . 5 0 4 4 47 . -0.2 . 34 . 29 . 32 32 . . . 4 2 3.6 3 9 3.9 2.3 , 5 2 4,7 5.0 5.0 3.4 , . . . 35 3 0 3 3 3 3 17 , . 39 , 33 . 36 . 36 , 20 . 30 , -LI n.2 12 71 . . -1,9 72 . 2 4 12.1 . -0,9 08 . 29 . 03 . 3-1 -2.0 46 . 53 . -0.2 49 . 00 . 39 . -1.4 42 . 25 . 35 , 28 . 39 , HOLARrnC ECOLOGY <i:i (i'tNd) ference was found (pooled mainland sample, trait XI, P<(),(15), which is not more than expected by chance. There is consequently no reason to analyse the two sexes separately. Of the 34 traits, those were excluded tbat did not show significant between-population variation. Our criterion was conservative: if a variant had a significant (P <O,(I5) difference in frequency between one or more pairs of populations, the trait was included in the analysis. The number of traits remaining was 21. of whicb 19 were bilateral. More than two variants were scored for 6 traits but these traits were subsequently dichotomized as indicated in Appendix A. Seventeen of the 19 bilateral traits show significant correlation between the two sides; A2 and A9 do not. Excluding the latter two, tbe average value of tp is 0.41. close to the average in Sjflvold's (1977) comprehensive study of the red fox in Sweden {cp = 0.46). Homogeneity of lateral incidence was tested in the bilateral traits. Seventeen of the 19 traits show no significant deviation from the null hypothesis, but A2 and B4 do. the incidence being too high on the left side in both traits. Correlation with size was tested using the point-biserial correlation coefficient. Full results for all three species are available from the senior author by request. Bilateral traits with only one side showing a marginally significant correlation are considered uncorrelated witb size. In .S". araneus, the incidence of 4 traits was correlated with size: B7, B8, B12, and D6. Based on the above results the MMD was calculated ft)r 13 populations and 15 traits of S. araneus. These traits are the 21 traits with significant variation between populations minus the ones that are heterogeneous with respect to age classes (B5 and B14) or are correlated with size (B7. B8, B12 and D6), Tab. Bl gives the frequencies in the 13 populations and Tab, B2 gives the MMDs and their standard deviations. the Faroe Islands: a study in microdifferentiation. - J. Zool., Lond. 185:73-92. - and Searie, A. G. 1963. Epigenetic polymorphism of the rodent skeleton. - Proc. Zool, Soc. Lond, 140: 577-615, - and Warwick, T. 1974. Field mice (Apodemus sytvaticus) on the Castle Rock. Edinburgh: an isolated population, - J, Zoo!.. Lond. 174: 325-331, Carter. J. L, and Merritt, J, F. 1981, Evaluation of swimming ability as a means of island invasion !iy sma!! mamma!s in coastal Virginia. - Annl, Carnegie Mus, 50: 31-46. Connor, E, F. and McCoy. E. D. 1979. The statistics and biology of the species-area relationship. - Am. Nat, 113: 791833. Corbet. G. B. 1966. Records of mammals and their ectoparasites from four Scottish islands. - Glasgow Nat, 18: 426434. Crowel!. K. L. 1973. Experimental zoogeography: introductions of mice to small islands, - Am. Nat. 107: 535-558. Delany, M. J. 1970, Variation and ecology of island populations of the long-tailed field mouse {Apodetnus sylvaticus (L.)). - Symp. Zool. Soc. Lond. 26: 283-295. Dobson. F. S, 1982, Competition for mates and predominant juvenile male dispersal in mammals. - Anim. Behav. 30: 1183-1192. Dolgov, V. A. 1961. Variation in some bones of postcranial skeleton of the shrews (Mammalia, Sorieidae). - Acta therio!. 5: 203-228, Dueser, R. D, and Brown. W. C. 1980. Ecological correlates of insu!ar rodent diversity, - Eco!ogy 61: 50-56. Ellenbroek, E. J. M, 1980, Interspecific competition in the shrews Sorex araneus and Sorex minulus (Sorieidae, Insectivora): a population study of the Irish pygmy shrew. - J. Zoo!.. Lond, !92: 119-136. Grainger. J. P. and Fairley. J. S. 1978. Studies on the biology of the pygmy shrew Sorex minutus in ihe west of Ireland. J, Zoo!,. Lond, !H6: 109-!42. Greenwood. P, J. !9t<0. Mating systems. phi!opatry and dispersal in birds and mamma!s. - Anim. Behav. 28: 114{>-1162. Hanski, !. 19S6, Dynamics of shrews on small islands accord with the equi!ibrium mode!.-Bio!, J. Linn. Soc. in press. Hawes, M, L, !977, Home range, territoriality and ecological separation in sympatric shrews Sorex vagrans and Sorex obscurus. - J. Mamma!. 58: 354-367. Jensen. B., Lau!und. B, and Engelstoft. C, 1980. The common shrew Sorex araneus and small rodents in the Hanherred Island on northwest coast of Denmark, - Flora Fauna 86: 7376. Jewell. P. A. 1966. Breeding season and recruitment in some British mamma!s confined to small islands. - In: Rowland, L W. (ed). Comparative biology of reproduction in mamReferences mals. Zoo!. Soc. Symp. No. 15. Academic Press. Adlcr, G. H- and Wilson, M, L. 1985, Small mammals on Mas- Keltock, W, L. and Parsons, P. A. 1970. Variation of minor non-metrical cranial variants in Australian Aborigines. sachusetts islands: the case of probability functions in clariAm, J. Phys. Anthrop. 32: 409-422. fying biogeographic relationships. - Oecologia (Berl.) 66: Kock. D. 1974. A Suncus mertensi, new speeies from Flores, [7S-IS6, Lesser Sunda Islands (Mammalia. Insectivora), - SenAulak. W. 1967. Estimation of small mammal density in three ekenb. Biol, 55: 197-204. forest biotopes, - Ekol. Pol. Scr. A. 15. 39: 755-778, Kostian. E, 1970. Habitat requirements and breeding biology Beer. J. R.. Lukcns. P. R. and Olson, D, 1954. Small mammal of the Root Vo!e, Microtus oeconomux (Pallas), on shore populations on islands of Basswood Lake. Minnesota. meadows in the Gu!f of Bothnia. Finland. - Ann. Zool. Ecology 35: 437-W5, Fenn. 7: 329-340. Berry, A, C, 1976. The anthropological value of minor variants Krebs. C. J.. Ke!!er, B, L. and -Bimarin. R. H. 1969. Microtus of Ihc dental crown.-Am. J, Phys. Anthrop. 45: 257-268, population bio!ogy: demographic changes in f!uctuating - and Berry. R. J. 1967, Epigenetic variation in the human populations of M. ochrogaster and M. pennsytvanicus in cranium,"-J- Anat. 101: 361-.379. southern Indiana. - Ecology 50: 557-607, Berry. R. J. 1968. The biology of non-metrical varianis in mice and men. - In: Brothwcll. D. R. (ed.). The skeletal biology Lomolino. M, V, 1982, Species-area and species-distance relationships of terrestrial mammals in t!ie Thousand Island of earlier human populations. Pergamon. London. Region, - Oecologia (Ber!.) 54: 72-75. - and Jakobson, M, E. 1975. Ecological genetics of an island - !984a, Mamma!ian island biogeography: effects of area, population of the House mouse (Musmusculus).-i. Zoo!., isolation and vagility. - Oecologia (Berl.) 61: 376-382. Lond. 175; 523-540. - , Jakobson, M. E. and Peters. J, 1978. The House mice of - 1984b. Immigrant selection, predation. and the distribu- nOLARCnC ECOLOGY Q:3 {XWh) tions of Microtus pennsylvanicus and Blarina brevicauda on Pucek. Z, 1969. Trap response and estimation of numbers of is!ands. - Am, Nat. 123: 468-483, shrews in removal catches. - Acta iheriol. 2S: 403-426. - 1985. Body size of mammals on islands: the island rule reRoberts. P, J. 1980. Diet of a kestre! {Falco tinnunculus) on examined. -Am. Nat. 125: 310-316. Bardsey Island UK. - Bird Study 27: 116. MacArthur. R, H- and Wilson. E, O. 1967, The theory of is- Schoener. T, W. and Schoener. A. 1983. Distribution of verteland biogeography, - Princeton Univ. Press, Princeton, brates on some very sma!! islands. I. Occurrence sequences Malmqvist, M, G. 1985. Character displacement and biogeogof individual species, - J, Anim. Ecoi. 52: 209-235. raphy of the pygmy shrew in Northern Europe. - Ecology Searie, A, G. 1954, Geneticai studies on the skeleton of the 66: 372-377. mouse. IX. Causes of skeletal variation within pure lines. McTaggart Cowan. I. 1941, Insularity in the genus 5orpjt on the J, Genet. 52: 68-102. north coast of British Columbia. - Proc. Biol. Soc. Wash- Skaren. U. 1980. The water as a barrier in spreading of shrews. ington 54: 95-108. (In Finnish with English summary). - Savon Luonto 12: 44Michielson. N, C. 1966. intraspecific and interspecific eompe47, tition in the shrews Sorex araneus L.. and Sorex minutus L. Sikorski. M. D. 1982. Non-metrical divergence of isolated - Arch. Neer!, de Zool. 17: 73-174. populations of Apodemus agrarius in urban areas. - Acta - . Gerink, A. and Moerel. L. 1980, De dwergspitsmuis, de theriol. 27: 169-180, enige Sorex-soort op Ameland, - Lutra 23: 33-41. Sj0vo!d, T. 1977. Non-metrica! divergence between ske!eta! O'Keeffe.D, A. and Fairley. J. S, 1981, 2 population studies of populations. - Ossa 4. supp!. 1. 133 pp. Irish pygmy shrews Sorex minutus. - Ir. Nat. J, 20: 269-275. Smith, A. T. 1974. The distribution and dispersal of pikas: conOkhotina, M, V. 1976. A new form of shrew (Insectivora, Sorsequences of insular pt>pulation structure. - Ecology 55: ieidae) from Moncron Islands. Russian SFSR, USSR, 1112-!119, Zool. Zh. 55: 59(K595. - 1980. Temporal changes in insu!ar populations of the pika - 1977. Shrews (Insectivora, Sorieidae) of Sakhalin I.s!and, {Ochotona princeps). - Ecology 6!: S-13, Zoo!. Zh, 56: 243-249. Spitzenberger. F. !977, The mammalian fauna of Cyprus. Part O!ert. J, 1975. A contribution to the knowledge of the smal! I: Insectivora and Rodentia. - Ann, Naturhist. Mus. Wien mamma! fauna of the Lofoteti Islands. - Zool. Beitr. 21: 81:401^142, 135-142, Thomas. H, H,, Jones. G. S, and Dibblee. R. L. 1980. Sorex Olsen, O. W, 1969. Hymenolepis prihilofensis new species of palustris, new record on Prince-Edward Island, Canada, Cestoda (Hymenolepidipae) from the Pribilof shrew Sorex Can, Fie!d-Nat. 94: .329-33L prihilofensis from the Pribilof Is!ands, A!aska. - Can. J. Trus!ove. G. M, !%1, Genetica! studies on the skeleton of the Zoo!. 47: 449-454. mouse. XXX, A search for corre!ations between some miPankakoski. E, 1979. The cone trap - a useful tool for index nor variants, - Genet. Res,. Camb. 2: 431^38. trapping of sma!! mammals. - Ann, Zool, Fenn. 16: !44- Webb. W. L, 1965. Sma!! mamma! popu!ations on islands. 150. Ecology 46: 479^88, Pieper. H, 1978. A new species of Crocidura (Mammalia, Sor- Yalden. D. W, 1981. The occurrence of the Pigmy shrew Sorex ieidae) from the Island of Crete, Greece. - Bonn Zool. minutus on moorland and the implications for its presence Beitr. 29: 281-286. in Ireland, - J. Zool., Lond. 195: 147-156. Pokki. J. 1981, Distribution, demography and dispersal of the field vole {Microtus agrestis) in Ihe Tvarminne archipelago, Finland. - Acta Zooi. Fenn. 164. 48 pp. HOIARCTtC ECOLOGY 9:.l (mSf.; http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Ecography Wiley

Shrews on small islands: epigenetic variation elucidates population stability

Hanski, Ilkka; Kuitunen, Juhani
Ecography , Volume 9 (3) – Oct 1, 1986
Download PDF
13 pages

Loading next page...
 
/lp/wiley/shrews-on-small-islands-epigenetic-variation-elucidates-population-Vp8r1huZLH

References (61)

Publisher
Wiley
Copyright
Copyright © 1986 Wiley Subscription Services, Inc., A Wiley Company
ISSN
0906-7590
eISSN
1600-0587
DOI
10.1111/j.1600-0587.1986.tb01209.x
Publisher site
See Article on Publisher Site

Abstract

Hanski. I. and Kuitunen, J. 1986. Shrews on small islands: epigenetic variation elucidates population stability. - Holarct. Ecol, 9: 193-204, Three large (4 to 8 ha) and 14 small islands (0.3 to 2 ha) in a lake in eastern Finland, all situated less than 0,5 km from the mainland, were surveyed for small mammals. Three species of shrew and two species of vole were resident in July 1982: Sorex araneus on 10. S, caecuiiens on 2. S- mimiius on 5, Microius agrestis on 12 and Clelhriotiomys glareolus on 4 islands. Immigrants were trapped from tiny islets, ;uid the data indicate that S. caecutiens and M. agre.sris are better dispersers than S. minulus and C. glareotus. respectively. Microuis agrestis. S. araneus and C. ^iareolus occurred nonrandomly. on subsets of the larger islands, while the two small Sorex species occurred more erratically, possibly because of competition with S. araneus. Juvenile sex ratio was miile-biased on the mainland but female-biased on large islands, possibly because juvenile males move more and are more likely to emigrate from an island than juvenile females. Metrical and non-metrical (epigenetic) cranial traits gave similar patterns of population differentiation in S. araneus. Two of the three large-island populations have differentiated from the mainland populations and from each other, suggesting that the populations are relatively slable. Small-island populations, which are often less than 10 individuals in size, showed liltlc differentiation but had more epigenetic traits fixed than large-island and mainland populations (founder effect). This suggests that the small-island populations are unstable, have a high extinction rate. /. Hanski and J. Kuitunen, DepI of Zoology, Univ. of Helsinki, P. Rautatiekaiu 13. SF'OOWO Helsinki. Finland. I. Introduction PopuUitions of small mammals on real or habitat islands provide an opportunity to study the classical biogcographic relationships (MacArthur and Wilson 1967) of insular cotnmunities (Beer et al. 1954. Webb 1965, Lomolino 1982. l9S4a. Adier and Wilsoti 1985), but such studies may also elucidate the spatial dynamics of populations, e.g. extinction of small populations (Smith 1980. Pokki 1981). dispersal (Smith 1974. Pokki 1981, Lomolino 1984b). establishment of populations (Crowcll 1973), ititra- and interspecific competition (Kostian 1970) and predation (Lomolino i984b). as well as genetic differentiation of local populations (Berry and Warwick 1974. Berry et aL 1978. Sikorski 1982). These aspects of population dynamics pose particular diffiAcecpted 12 November 1985 © HOIARCTtC ECOLOGY culties for anyone studying Soricinae. because of the small size and covert habits of shrews, Radiotelemetry seems to be impractical for so small mammals (M. Genoud. pers. comm.). Island population studies cannot answer many questions about individual behaviour, but at the population level they can be useful. Here we report on trapping results of small mammals on the 17 small islands of a lake in eastern Finland, in a region with six species of shrew {Sorex and Neomys) and two speeies of vole (Microtus agrestis and Clethrionomys glareolus). Our main interest lies in the shrews, for which little has been published about occurrence on smiill islands (the ecologically interesting scale), though some information exists for large, mostly oceanic islands (the evolutionary interesting scale: McTaggart Cowan 1941. Corbet 1966, Olsen 1969, Kock 1974, 01- HOl.ARCriC ECOLOGY 9:3 (1986) ert 1975. Okhotina 1976. 1977. Spitzenberger 1977, Grainger and Fairley 1978. Pieper 1978, Ellenbroek 1980. Jensen et al. 1980. Roberts 1980. Thomas et al. 1980, Michielson et al, 1980. O'Keeffe and Fairley 1981. Yalden 1981, Malmqvist 1985). The purpose of this paper is to describe the occurrence of the species on the islands in 1982. prior to experimental perturbations (work started in 1983), and to try to tie together ecological and morphometrical data on the island populations. Metrical and non-metrical (epigenetic) morphological variation may help to understand the dynamics of populations on small islands, where genetic drift may be assumed to be the main mechanism of differentiation. Fig. 2. The 6 metrical measures of the lower jaw. 2. Material and methods 2.1. Field work Lake Sysma in eastern Finland (Ilomantsi, 63°N. 32°E) has three "large" islands (3.8 to 8.0 ha). 14 "small" islands (0.3 to 2.2 ha), and 25 tiny islets, less than 01 ha in size. All islands have a maximum elevation of less than 10 m and most less than 2 m. Isolation varies from 10 to 500 m (Fig. 1). All islands and most islets have a more than 100 years old. mixed coniferous forest cover: pine Pinus sylvesiris and spruce Picea abies with some birch Betula spp. The dwarf sbrubs Ledum and Vaccinium are abundant in the field layer, and the humus layer is very thick. Inevitably, there are some minor habitat differences, but generally the islands have a uniform forest cover. All islands, most islets, and 5 mainland study sites (Fig. 1) representing the same forest vegetation were trapped with metal cones {Pankakoski 1979) for usually three nights in July 1982. Mainland sites were trapped with 3(l-cone traplines, traps spaced 10 m apart. Large islands were trapped with 50 to 1(K) cones, small islands with 15 to 30 cones, and islets with 3 to 10 cones, depending on island size (Tab. 1), Traps were spread evenly across the islands. Two small islands (S4 and S12) and one islet (15) were trapped more intensively to remove all shrews (details in Section 3,2.). Shrews were preserved in 70'^ alcohol. 2.2. Metrical traits Fig. I. A map of the study area, lake Sysma in eastern Finland, Ml-5 iire 5 mainland study sites, Ll-3 are "large" islands (4 to K ha. light shading). Sl-14 arc -small" islands (0,3 to 2 ha. strong shading), and the small numbers from 1 to 25 indicate the locations of tiny islets, less than 0.1 ha (11-25 in the text). Body weight was measured to the nearest 0.1 g with a spring balance after the shrew had been dried for 5 min on blotting paper. A weight meastirement taken after preservation in alcohol is not comparable to live weight but allows comparisons within this set of data. Hind foot length was measured to the nearest 0.1 mm with a ruler. Following these measurements tbe animals were sexed and aged (young of the year vs. overwintered) and the breeding state of females was reeorded. Skinned carcasses were boiled for 5 min. and the skeletons were prepared enzymatically. in papain (2-4 d at 38°C. as recommended by Searie 1954). The following cranial measurements were taken under a dissecting microscope with an ocular micrometer (accuracy 1-2%); Condylobasal length. Six bilateral measures of the lower Jaw (Fig. 2). The average value of the two sides was used in the analysis. HOLARCTIC ECOLOGY 2.3. Non^metrical traits Ihirty-tour traits were scored, mostly following Peter King's (pers. comm.) work on the common shrew in the United Kingdom. See also Berry and Searle (1963). The traits are described in Appendix A. 2.4. Statistical methods Non-metrical divergence between populations was measured with C. A. B. Smith's Mean Measure of Divergence. MMD. Sjovold (1977) gives a thorough statistical treatment of the MMD, and we have followed elosely his recommendations. Side-by-side eorrelation in the manifestation of bilateral traits was measured with the rp cocfficicnl of Pearson. Correlation between the trait manifestation and size was tested with the point-biserial correlation coefficient of Pearson, r^,, with the condylobasal length of the skull used as a measure of size. We did not ealculate correlations between the traits, because our material is relatively small and previous studies have shown that such correlations, if they exist, are generally small (Truslove 1961. Berry and Berry 1967. Kellock and Parsons 1970, Berry 1976, Sj0vold 1977. Sikorski 1982). Sj0vold (1977) concludes that "with regard to sample sizes available in general, eorrelation does not affect the results to a greater extent than would random fluctuations from independent variables." Metrical traits were first analysed with univariate ANOVA, then with stepwise discriminant function analysis (HYLPS programs of the Computing Centre of the University of Helsinki). i. Results 3.1. Species composition on islands Trapping results are summarized in Tab. I. Sorex isodon was caught from the mainland sites M2 (3 individuals) and M4 (I), and S. minutissimu.s from M4 (2), but Neomus fodicns was not found in the study area until 1983 (Hanski unpubl.). These three species were not re- Tab. I. Trapping results for mainland study sites (MI-5), large (Ll-3) and small islands (S1-I4), and islets (11-24) in July 1982 (for their iocalions see Fig. 1). This table gives relative densities (individuals per KX) trap-nights) or actual numbers caught (figures in brackets) for Sara - Sorex araneus, Scae - .V. caeculiens. Smin - .V. minufus, Magr - Microtus agrestis. and Cgia = Clelhrionomys glareolus. Islets 115, 120, and 125 were not trapped, other islets not included were empty. Population Ml M2 M3 M4 M3 LI L2 L3 SI S2 S3 S4 S.^ S6 S7 S8 S9 Sin SI I S12 S13 SI4 Area ha Traps M) 30 30 30 30 46 10() I(X) 30 20 20 160 20 20 15 25 20 20 15 90 25 20 Nights 4-5 3 4 3 3 3 3 3 3 3 3 6-7 3 3 3 3 3 3 3 7 3 3 Sara 24 Seae Smin Magr Cgla Trapping period 3.8 8.1) 4.4 2.2 0.7 0.8 0.8 0.7 1.2 0.4 1.0 0.8 0.4 0.3 0.3 1.0 0.7 8 13 10 - 9 (1) 4 (1) (1) (21) .. . 2 4 - 1 9 - 22 (1) 12 5 (1) 6 7 (I) 8 3 6 + 2 + (21) 8 42 (39) 12 3 (4) (1) (14) 4 _ 5 (1) 18 - 6 1 1 - (t) (I) (2) (3) 2 2 _ + — — — - II 12 13 15 16-8 110 III 113 122 124 (1) (3) (1) (4) (1) (2) 8-13.7 8-13.7 8-13.7 20-23.7 19-22.7 2()-23.7 18-21,7 19-22.7 6- 9.7 6- 9.7 6- 9.7 14-20.7 7-10,7 7-10.7 7-10.7 7-10.7 7-1(1.7 20-23.7 20-23.7 1.5-22.7 20-23.7 19-22.7 7-10.7 7-10.7 7-10.7 8-14.7 7-10.7 20-23.7 2()-23.7 2()-23.7 19-22.7 19-22.7 + The catches were lost but the species was present. tCOLOGY '>;3 (iy«i) • LOCAL POPULATION 3.2. Population densities O IMMIGRANTS Tab. 1. gives relative population densities, individuals per 100 trap-nights, for the mainland study sites and the islands. Two small islands, S4 (0.8 ha) and S12 (0.3 ha), and MINUTUS the islet 15, were intensively kill-trapped with cones in 0 0 0 CAECUTIENS 0 July 1982 to remove all shrews (Tab. 1). Equally inten0 0 ARANEUS sive trapping in May to June 1983 proved that the islands had been emptied (Hanski unpubl.; an adult male •O 00 CD O of Neomys fodiens was caught on S4 in May 1983. but it od AREA ho O O O r-' o o o o o o o had obviously moved there from the mainland). These ':^ ISLAND results suggest that trapping with the intensity of 150 Fig. 3. Presence of breeding populations and catches of probpitfalls per ha for 7 days can remove all shrews from an able immigrants on the islands, arranged in an order of inisland (even a lower intensity might be sufficient; cf. creasing island size. See Fig. 1 for the island locations. Aulak 1967, Pudek 1969, Pankakoski 1979). Densities may go up to at least 30-40 individuals per ha on small islands in late summer. The following are corded from the islands, but this is not surprising beminimum estimates (July 1982): S. caeculiens on S4. 26 cause of their rarity on the mainland. per ha, and S. minulus on S11,38 per ha. The densest isSorex araneus, S. caecutiens and S. minutus were land population was however S. araneus on LI (3.8 ha), present at all five mainland sites. Based on the numbers where 76 individuals were trapped with 46 cones in 3 of shrews trapped, and their age and sex (below), we nights; population size was probably 200-300 individuconclude that there were 10 breeding populations of S. als, density >50 individuals per ha. Apart from this araneus, 2 populations of S. caecutiens, and 5 popu- population, densities were not different between the islations of 5, minulus on the 17 islands (Tab. 1). The field lands and the mainland (Tab. I), though admittedly our vole Microtus agresiis and the bank vole Cleihrionomys mainland and island results are not strictly comparable. glareolus had populations on 12 and 4 islands, respeetively (Tab. 1). The field vole was uncommon on the mainland, though it was the dominant vole on small is- 3.3. Dispersal lands. Webb (1965) found the same for the ecological Overwintering and breeding are practically impossible equivalent Microtus pennsylvanicus in North America on the tiny islets, and any small mammal caught there is (see also Crowell 1973). an immigrant, except perhaps the field vole, which Species number increased with island size from I to 4 might manage to breed on some islets (cf. Pokki 1981). (Fig. 3). as expected (MacArthur and Wilson 1967, In view of the trapping intensity on islands, a single indiConnor and McCoy 1979; see Dueser and Brown 1981). vidual caught on an island in July is also likely to be an Lomolino 1984a for mammals), but this standard result immigrant, though we cannot exclude the possibility masks interesting biology. The field vole was present on that it belonged to an extremely small population, permost islands regardless of island size. The eommon haps of size 1. shrew was resident on most islands larger than 0.7 ha, Looking first at the numbers of shrews trapped on iswhile the bank vole was found on only the three largest lets, there are too many S. caeculiens and too few S. araislands and a single small one (Fig. 3). Island size is a neus and S. minutus in comparison with the mainland good predictor of the presence of the latter two species densities, the source of most immigrants, but the num(one-tailed Mann-Whitney U test, P <().O1 and 0.(125, bers are small and the deviations from the expeeted frerespeetively; cf. Schoener and Schoener 1983). In Min- quencies are nonsignificant (Tab. 2). Including the nesota (Beer et ai. 1954), M. pennsylvanicus occurred probable immigrants trapped on islands, there is a sigon islands of all sizes, down to (at least) 0.2 ha, while nificant difference between S. caeculiens and S. minutus Cleihrionomys gapperi and Peromyscus maniculalus (observed immigrants 10:1, expeeted 6.8: 4.2, P <0.05). were present on islands greater than 1 ha. Sorex araneus cannot be fairly compared because it had Compared with the above-mentioned speeies, the a local population on most islands and hence many postwo small Sorex show a different pattern. Excluding the sible immigrants cannot be identified as such. 5. caeculiens population on the largest island (L2), the The field vole is clearly a superior colonizer of small small shrews were absent as breeding populations from islands to the bank vole (Tab. 2: Ppkki 1981), Carter the 4 largest islands, though 6 populations were found and Merritt (1981) have experimentally confirmed the on smaller islands (Fig. 3). good swimming ability of Microtus pennsylvanicus. in conclusion, these data suggest that S. caecutiens is a better disperser than S. minutus, which is expected, as the smaller S. minutus has a shorter starvation time and GLAREOLUS AGRESTiS HOi.ARCTlC ECOLOGY 9.3 rial. The 9 metrical traits were tested for equality with respect to sex in S. araneus. No significant difference was found in any of the traits and data for the two sexes were pooled. Pi)pulations Sara Scae Smin Magr Cgia Univariate ANOVAs were run for 13 populations of .V. araneus (Ml-5. Ll-3, SL S3. S5. S8 and S14). 5 Mainland 110 31 19 9 13 populations of S. caecutiens (MI, M3, M5, L2 and S4) Islands 196 33 21 146* 24* and 4 populations of 5. minutus (M4. M5, S8 and SIO). Island Pooled sample sizes are 246, 41 and 15. respectively. immigrants 3+ 5 1 ? ? Untransformed and log-transformed analyses gave simiIslets 6 5 27 lar results. 'Slight underestimates because eatches from SI, S3, and S5 Variation among S. araneus populations was signifiwere lost (see Tab. I). cant (P <0.01) in all 9 metrical traits. In S. minutus, the F values were nonsignificant, except two jaw measures (2 and 3, P <().()5). while in S. caecutiens the following traits varied significantly (P <0.01) between the popuis more likely to be drowned while swimming than S. lations: hind foot length, and jaw measures 1, 2 and 6. caecutiens. However, with these data we cannot exclude the alternative explanation that S. caecutiens simply Sorex caecutiens had two island populations, on L2 moves more than S. minutus. with .V. araneus and on S4 alone. A (naive) competition hypothesis would predict that in the absence of S. minutus but in the presence of 5. araneus the L2 population 3.4. Demography has small individuals, while the S4 population, in the abThe trapping results give some information on breeding sence of S. araneus, has shifted towards tbe seemingly successful S. araneus niche. However, the opposite was and the sex ratio. We restrict this section to shrews. The frequency of adult female .V. araneus was 4 to 5% observed: in most metrical traits the L2 population conon the mainland and large islands in July 1982. Adult sisted of large while the S4 population of small indivimales comprised 20 to 25% on the mainland and on L2 duals. The 13 S. araneus populations were subjected to a ;md L3. but only 3% on LI (2 individuals). The difference between the islands is highly significant (x" — 14.5. stepwise discriminant function analysis. The first 2 funcdf - 2, P <(l.!)()l). It may be significant that LI had the tions accounted for 56 and 17% of discrimination, and densest S. araneus population of all islands. Immature were statistically significant. Fig. 4 shows the population shrews establish feeding territories (Michielson 1%6. means on the first 2 functions, and the correlations of Hawes 1977), and if these are densely packed, adult these functions with the original metrical traits. males, which have the greatest home ranges (MichielOverlap between the populations is extensive, less son 1%6, Hawes 1977), may suffer from frequent fights than half (467o) of the individuals being classified corand may have difficulty in obtaining enough food. rectly. Although the a posteriori classification is of limIn young S. araneus, the sex ratio was male-biased ited interest when overlap is so extensive, it is worth (58% males) at the five mainland sites hut female-biased (39% males) on targe islands. Differences between the mainland sites (x' = 3.41, df ^ 4. NS) and large islands (x" = 0.19, df = 2. NS) are nonsignificant, but the difference between the pooled mainland sample and the pooled large-island sample is significant (x' = 6.26, df = 1, P<().02). The simplest explanation is that juvenile males move M3 more than juvenile females, as in most mammals (see reviews by Greenwood 1980, Dobson 1982). This hyS3 pothesis explains male-biased trap-catches on ihe mainLI land, and it explains the female-biased sex ratio on isS'A S5 lands, assuming (1) that males, because they move more, are more likely than females to leave an island, ss and (2) tbat emigration from an island exceeds immigration to the island (almost certainly true). Tab. 2. Pooled numbers at the mainland sites, on islands and on islets (species abbreviations as in Tab. i). "Island immigranls" are individuals trapped on islands and considered to be immigrants (see the text). .V5. DifTcrentiation between populations: metrieal traits Metrical analyses were restricted to juveniles because of I he small number of overwintered shrews in the mateIIOI.ARCTIC ECOl OGY 9:3 (1986) Fig. 4. Discriminant function analysis of the island populations: Ihf first 2 functions The insert shows correlations between the functions and the metrical traits. See Fig, 1 for the island locations. have especially low scores on the first function (small individuals). To some extent these data conform to the Krebs' or island effect (Krebs et al. 1969): island populations are often dense and have large individuals (sec also Jewell 1966, Delany 1970). Our small-island populations are, however, so unstable that individual selection has little time to work towards larger individual size, and in fact we found small individuals on small islands. Small size offers tbe (perhaps slight) advantage to a small-island population that less resources are used per individual, hence extinction probability may be smaller (individual and/or group selection; see also discussion in Lomolino 1985). Thirteen young common shrews were trapped at sites not included in the discriminant function analysis. The result of this analysis can be used to classify these extra individuals into the 13 populations, to indicate dispersal sources. Sucb an exercise has heuristic value, though the result must be interpreted with caution. Most of the 13 movements implied (or relatednesses of small populations: islands S6 and S9), are reasonable (Fig. 5). Of particular interest is that the 3 individuals caught from 15 all have the strongest connection (the largest a posteriori probability) to a different surrounding population. This strongly suggests that these 3 individuals have in fact dispersed from other populations to the islet 15, and, thus have not been born there. Fig. 5. Discriminant function analysis suggests the movements or relatednesses between populations shown here for 13 individuals of S. araneus trapped at sites not included in the analysis. The thick line indicates the a posteriori probability (of an individual "belonging" to the population) greater than 0.7, the thin line indicates probabilities from 0.4 to 0.69. and the broken line shows probabilities less than 0.4. 3.6. DiFTerentiation between populations: non-metrieal traits Details of the calculations are given in Appendix B. This section is restricted to S. araneus, as sample sizes are small for the other species. The 3 large-island populations have all diverged more from the mainland populations than the remaining 10 populations (P <0.01), and the 5 small-islatid popu- noting that all except one of the mainland populations had a smaller proportion of correctly classified individuals than the island populations {Mann-Whitney U test, P <0.05). Tbe exceptional mainland population is M3, distinct also in the analysis of epigenetic traits (below). Of the 3 large-island populations, LI and L3 are more distinct (have more individuals classified correctly) than L2, and of the small islands SI and S3 have the most distinct populations. (The largest population LI has many individuals in the a posteriori classification, which may be partly an artefact, but the Mann-Whitney test above is conservative, because the mainland samples were generally the largest.) Mainland populations, excepting M3, delimit a relatively small area in the discriminant space, and most islands are situated outside this subspace. Of the 3 largeisland populations, L2 has not diverged from the mainland populations, while the others have, but to opposite directions on the 2nd function. The first discriminant function is strongly correlated with hind foot length and weight (Fig. 4). All 3 large-island populations have higher scores on this function (larger individuals) than the 5 small-island populations (P <0.05). SI and S3 198 • I MEDIAN MU FROM MAINLAND I MEDIAN MU FROM ISLANDS LARGE 12 L3 LI SMALL SI 56 53 55 Sli, ISLANDS Ml M2 M3 M4 Mb MAINLAND Fig. 6. Median measures of uniqueness (MLJ) of 13 .V, anmeus populations with respeet to the remaining island and mainland populations (non-metrical analysis). The MMU is the average of the MMDs with the populations involved. See Fig. 1 for the locations of the populations. HOLARCTIC ECOLOGY 9:3 (IV«6) Tab. 3. The number of constant traits in a random sample of 8 accurate answer to the question: how many traits would individtials from 13 populations of 5. araneus. For the locations show no variation in a random sample of 8 shrews? of the study sites see Fig. 1. Means and 95% confidence interTab. 3. summarizes the results. The 95% confidence vals were calculated with Monte Carlo simulation (see text). Population Mean Ml M2 3.48 2.44 1.82 3.09 3.37 3.98 95% CI 2-6 1-4 1-3 1-6 2-6 1-5 3-8 2-6 9-9 1-6 Mainland m. m Large Island Small Island LI L2 L3 SI S3 S14 intervals mostly overlap, but mainland populations have less traits fixed on average (2.84) than island populations (5.13; P <0.U5), Small-island populations tend to have more traits fixed than large-island populations but the difference is not significant. Trait B9 is a prime example of founder effect and/or genetic drift: it had a frequency between 0.07 and 0.19 on the mainland, it was absent (in the samples) from 4 islands, including one large one, but it had a high frequency on 2 small islands (0.25 and 0.36, Tab. Bl). 4. Discussion lations tend to be more differentiated than the 5 mainland one (P <().O5; Fig. 6). Median divergences of the island populations from other island populations are greater than their divergenees from mainland populations (Fig. 6), suggesting that island populations have diverged in different directions. There is no elear differenee between large- and small-island populations in this respect. One mainland population (M3) is clearly more differentiated than the remaining ones (Fig. 6). A careful look at the frequencies in Tab. Bl reveals 3 traits in which island and mainland populations appear to fall into distinct groups. All small-island frequencies are higher than or equal to the large-island and mainland frequencies in B14and D6 (for each trait P <0.01). Both traits were excluded from the calculation of the MMD because of a difference between age classes (B14) or correlation with size (D6). A surprising result, however, is that D6 was positively correlated with size, while small-island individuals tended to be small (Fig. 4) but had a high frequency of D6. All 5 mainland populations had higher or equal frequency of D4 than the 8 island populations (P <0.01). Additionally, all 5 smallisland populations had a lower frequency of B13 than the 3 large-island populations (P <().O5). In summary, there is an indication of a significant differenee between small-island and large-island or mainland populations in 20% of the 21 traits studied. The difference is between newly-established, small populations, and older, larger populations. Some populations in Tab. Bl show more variation than others, but a direct comparison may be misleading, because sample sizes are different: less variation is observed in a smaller sample, other things being equal. To correct for sample size, we used Monte Carlo simulation to reduce all samples to 8 (randomly selected) individuals, the observed minimum. Calculations were based on the number of sides, because most of the traits in Tab. Bl are bilateral, and the frequencies reported are averages of the two sides. This method gives a sufficiently 13- HOLARCTIC ECOLOGY ^;3( 1986) Metrical and epigenetic analyses of microdifferentiation gave much the same results, which suggests, as the two kinds of variation are independent of each other and hardly affected by the same non-genetic factors, that differences between the populations are primarily genetic. Reduced epigenetic variation in populations on small islands points to the same direction. Good agreement between metrical and epigenetic results has not always been reported (Berry 1968, Berry and Jakobson 1975, Berry et al. 1978). Why different studies should differ in this respect we do not know, but note that our spatial and temporal scales are smaller than in Berry's and his assoeiates' works. Amongst the mainland S. araneus populations, M3 showed the greatest differentiation in both metrical and epigenetic traits. The explanation is obvious when one looks at the map of the study area (Fig. 1): M3 is surrounded by dispersal barriers to the west (the lake) and to the north (a dry ridge and another lake). This result suggests that in other, less obvious cases a study of morphometrical variation may reveal unexpected dispersal barriers between populations. What are the ecological implications of the observed morphological differentiation of the island populations? First of all, the main mechanism of differentiation is likely to be genetic drift, though admittedly some traits showed significant differences between island and mainland populations (Section 3.6). If the populations would not go extinct, we would expect more differentiation on small than large islands, because of increasing rate of drift with decreasing effective population size. The epigenetic divergence of S. araneus was indeed smaller on the largest island (8 ha) than on the two other large islands (3.8 and 4.4 ha), even though isolation was the same (Fig. 1). Genetic drift in relatively stable populations explains this difference. Data for 17 5. araneus populations on islands greater than 2 ha in another lake gave the same result (Hanski 1986). The results for the small-island S. araneus populations are, however, opposite to the prediction: populations on small islands (0.7 to 2.2 ha) are less differentiated than populations on the two larger islands (3.8 and 4.4 ha). Isolation cannot explain this result (Fig. 1), but population stability can: if small-island populations have a high extinction probability, they have less time to diverge than large-island populations. The large number of constant epigenetic traits in small-island populations is likely to stem from the initial founder effect and is not inconsistent with this hypothesis. Although more work on both the ecology and morphometrics of the island populations is needed, and is in progress, these results suggest that morphometrical studies can help to understand the dynamics of semi-isolated populations. It would be nice to know the time periods needed to achieve the observed levels of differentiation in the island populations, but this of course is not known. Apart from the problems of relating differentiation in epigenetic traits to differentiation in genetic traits, for largely unknown reasons some populations diverge fast, e.g. Apodemus agrarius in parks in Warsaw has shown significant divergence in 30 years (Sikorski 1982), while others diverge slowly or not at all, e.g. a population of the same species in Edinburgh has not diverged in 100 years (Berry and Warwick 1974). Another complication is that the island populations are not entirely isolated (Skaren 1980 and Hanski 1986 present more data and discuss overwater dispersal in shrews). Something more positive can be said about the sizes of the populations that are relatively stable and show morphological divergence versus populations which appear unstable and show little divergence. Our data suggest a critical island size between 2 and 4 ha separating the relatively stable and the relatively unstable populations of S. araneus. Numbers are likely to fluctuate mostly between 20 and 200 on the large islands (4 to 8 ha), which means that demographic stochasticity is relatively unimportant, and the low of 20 individuals may be large enough to buffer the population against much of environmental stochasticity. In contrast, small islands, less than 2 ha in size, have populations that may reach the size of 50-100 individuals in late summer, but in early spring, before breeding has started, these populations are often very small. To give an example, S3 (0.8 ha) and S5 (0.7 ha), which were removal-trapped in May to June 1983, had S. araneus populations of 3 and 8 individuals, respectively (Hanski, unpubl.). Both populations had both sexes in spring 1983, and the females were pregnant and/or iactating, but clearly demographic stochasticity must be a major factor In the dynamics of such small populations. The above comments were limited to 5. araneus, the best competitor, and probably little affected by the other species. The data on the island occupancy (Fig. 3) clearly showed that the dynamics of the two small species, S. caecutiens and 5. minutus, was affected by something else apart from island size. Isolation cannot explain the near absence of the small species from the 4 largest islands for a number of reasons: (1) these islands are not more isolated than the smaller islands where S. 200 caecutiens and S. minutus were present; (2) the bank vole, a poor disperser (Crowell 1973), was present only on the three largest islands and one small one (Fig. 3); (3) two 5. caecutiens, most probably immigrants, were caught from L2 and L3, again suggesting that dispersal to these islands is not infrequent; and (4) in later years (1983-85), both S. caecutiens and S. minutus have been recorded as residents or immigrants on some of the large islands. Hanski (1986) describes the colonisation of LI by S. minutus in 1984. The most plausible hypothesis we can suggest is that interspecific competition has affected the distributions of the small species on the islands. It seems reasonable to assume that competition could cause the extinction of an inferior competitor on these islands, less than 8 ha in area, though we expect that spatial heterogeneity prevents exclusion taking place on much larger islands, as on the mainland. Competition is not less severe on small islands, but chance becomes important and leads to a proportion of empty islands, available for all species to colonize. We recognize that the competition hypothesis cannot be properly tested without experiments, which are now in progress. Acknowledgements - We thank T. and J.-L. Tast for conducting the field work. We are especially grateful to P. King for sending us information on the non-metrical traits that he has scored in the eommon shrew. O. Jarvinen, P. King, E. Pankakoski and R. Vaisancn made useful comments on the manuscript. The study was supported by the Academy of Finland. Appendix A: Non-metrical traits This appendix gives the non-metrical (epigenetic) traits studied. If not mentioned otherwise, the trait is bilateral, and presence is scored. See Fig. Al. XI. Degree of fusion of processus spinosus of coalesced sacral vertebrae (Dolgov 1961). Several variants can be distinguished (Fig. Al). The trait was dichotomized by combining variants 3-4 and 5-8; variants 1-2 were not recorded. Unilateral. Medial palatine foramen. Posterior palatine foramen (occasionally accessory posterior palatine foramen present). Accessory terminal palatine foramen, on rear end of palate. The main foramen was always present. Foramen ovale divided into two. Accessory foramen posterior to the basisphenoid foramen. Basisphenoid process, projection sharper than a right angle was scored. Median basioccipital foramen. Unilateral. Basioccipital foramen. Accessory hypoglossal foramina (1 or 2). Prcmaxillary foramina, including the ones posterior to a line drawn up from the mid-point of HOLARCTIC ECOLOGY a. Al. A2. A3. A4. A5. A6. A7. A8. A9. Bl. B5 B4 B3 B2 D3 Dl B1 V B14 CI B11 B12B14 B15 A1 A2 A4 A6 A9 B8 B10 "" B16 Fig. AI. The non-metrical traits. The different variants of the traits B14 and XI are shown separately. See text for full names. the 2nd unicuspid and anterior to the junction of the 3rd and 4th unicuspids. From none to 5 foramina. The trait was dichotomized by combining variants 0-1 and 2-5 foramina. B2. Anterior maxillary foramen (rarely 2 but scored merely as present or absent). B3. Accessory foramen outside infraorbital foramen. B4. Accessory foramen above the rim of infraorbital foramen. B5. Accessory foramen posterior to B4. B6. Accessory foramen posterior to lacrimai foramen. B7. Accessory foramen below lacrimai foramen. B8. Accessory posterior alisphenoid foramen (rarely 2 but scored as present or absent; rarely the main foramen absent but pooled with the variant without accessory foramen). B9. Pncutnatic foramen, between the two articular condyles. BIO. Anterolateral parietal foramen, near the edge of the bone. Bl I. Anterior lateral sinus foramen. Occasionally absent, or with I or 2 accessory foramina dorsal to the main one (not scored). HOLARCTIC ECOLOGY 9:3 (1986) B12. Mesial lateral sinus foramina. From none to 5 foramina. The trait was dichotomized by combining variants 0-1 and 2-5 foramina. B13. Squamosal-parietal suture. Occasionally a branch or branches extend upwards in the parietal, which variant we scored. B14. Posterior lateral sinus foramen. Several variants (Fig. A I). The trait was dichotomized by combining variants O-I and 2-5. B15. Accessary posterior lateral sinus foramen. B16. Pro-otic foramen. CI. Parieto-frontal foramen reduced to a slit. C2. Median supraoccipital foramen. Unilateral. Dl. Anterior accessory foramen to the major mental foramen (rarely two or three accessory foramina but pooled with the major variant). D2. Lower mental foramen. D3. Dentary canal. Several variants, we scored absence of the major canal. D4. Lower coranoid foramen. D5. Upper coranoid foramen. D6. Coranoid fossa cross-ridge. Appendix B: Non-metricaE analy.sis The pooled frequencies of the variants in the 34 nonmetrical traits are available by request from the senior author (IH) for juvenile and adult .S'. araneus, and for juvenile S, caecutiens and S. minutus. The two sides of the 31 bilateral traits bave been treated separately. All three species pairs differ highly significantly in 6 to 9 traits each, though not always in the same traits. Adult and juvenile 5. araneus bave similar frequencies, excepting B5 and B14. Marginally significant differences on one side only in the bilateral traits B4, B6, BIO, D4 and D5 cause less coticem and do not prevent pooling of the age classes. The frequencies of B4 and B5, wbich are situated close to each other (Fig. Al), arc higher in juvenile than adult 5. araneus. The significant difference in B14 is caused by an increase in the frequency of variant (I and a decrease in the frequency of variant 1 from juveniles to adults (see Fig. Al). Homogeneity of incidence between the two sexes in juvenile S. araneus was tested in the largest sample, from Island LI (71 individuals), and in the pooled mainland sample (78 individuals). Only one significant dif- Tab. BL Frequencies of 21 non-metrical traits in 13 populations of S. araneus (multiplied by UXl). ln bilateral traits the average of the two sides is given. For the locations of the populations see Fig. 1, and for the traits see Appendix A. n is ihe number of individuals studied (for some traits n is slightly smaller because the trait could not be scored on some skulls). Trait XI A2 A9 Bl B2 Ml M2 M3 M4 M5 LI Population L2 L3 SI S5 S8 60 S14 •Ul •Jl B4 B5 B6 ^ 5 23 5 32 55 3t 4t) B7 B8 By BI2 BL3 BI4 3K f5 > 0 'Ul ( 1 I ) •Jl BI5 C2 Dl D3 D4 D5 D6 n 36 5 HH) Tab. B2. The Mean Measure of Divergence (upper triangle) and its standard deviation (lower triangle) between 13 populations of 5, araneus. TTie MMDs that are at least twice as great as the standard deviation can be considered significant (bold face). L2 L3 LI S I S8 25 . S3 Populations S5 Si4 Ml M2 M3 M4 12 , 19 . 62 43 . . 38 . 55 . 0 3 -4,5 . -2.8 04 , -0.9 05 . -2.1 12 , 0 8 -4,6 . -L4 -0,2 -0,4 -2,5 0 4 -0,7 . -2,4 22 . L2 L3 LI SI S8 S3 7.3 8.3 3,7 14.9 60 , S5 SI4 Ml M2 M3 M4 M5 3,4 3.4 1.8 2,3 3,4 1.7 2.0 60 , 70 . 27 . 59 . 36 . 50 . 20 32 . . 5 8 5 0 -2.1 , 02 , . -2.4 07 , 0 5 -1.3 , 47 , -4,5 2 5 2.0 . 50 44 , . 7 4 -0,5 . 5 0 4 4 47 . -0.2 . 34 . 29 . 32 32 . . . 4 2 3.6 3 9 3.9 2.3 , 5 2 4,7 5.0 5.0 3.4 , . . . 35 3 0 3 3 3 3 17 , . 39 , 33 . 36 . 36 , 20 . 30 , -LI n.2 12 71 . . -1,9 72 . 2 4 12.1 . -0,9 08 . 29 . 03 . 3-1 -2.0 46 . 53 . -0.2 49 . 00 . 39 . -1.4 42 . 25 . 35 , 28 . 39 , HOLARrnC ECOLOGY <i:i (i'tNd) ference was found (pooled mainland sample, trait XI, P<(),(15), which is not more than expected by chance. There is consequently no reason to analyse the two sexes separately. Of the 34 traits, those were excluded tbat did not show significant between-population variation. Our criterion was conservative: if a variant had a significant (P <O,(I5) difference in frequency between one or more pairs of populations, the trait was included in the analysis. The number of traits remaining was 21. of whicb 19 were bilateral. More than two variants were scored for 6 traits but these traits were subsequently dichotomized as indicated in Appendix A. Seventeen of the 19 bilateral traits show significant correlation between the two sides; A2 and A9 do not. Excluding the latter two, tbe average value of tp is 0.41. close to the average in Sjflvold's (1977) comprehensive study of the red fox in Sweden {cp = 0.46). Homogeneity of lateral incidence was tested in the bilateral traits. Seventeen of the 19 traits show no significant deviation from the null hypothesis, but A2 and B4 do. the incidence being too high on the left side in both traits. Correlation with size was tested using the point-biserial correlation coefficient. Full results for all three species are available from the senior author by request. Bilateral traits with only one side showing a marginally significant correlation are considered uncorrelated witb size. In .S". araneus, the incidence of 4 traits was correlated with size: B7, B8, B12, and D6. Based on the above results the MMD was calculated ft)r 13 populations and 15 traits of S. araneus. These traits are the 21 traits with significant variation between populations minus the ones that are heterogeneous with respect to age classes (B5 and B14) or are correlated with size (B7. B8, B12 and D6), Tab. Bl gives the frequencies in the 13 populations and Tab, B2 gives the MMDs and their standard deviations. the Faroe Islands: a study in microdifferentiation. - J. Zool., Lond. 185:73-92. - and Searie, A. G. 1963. Epigenetic polymorphism of the rodent skeleton. - Proc. Zool, Soc. Lond, 140: 577-615, - and Warwick, T. 1974. Field mice (Apodemus sytvaticus) on the Castle Rock. Edinburgh: an isolated population, - J, Zoo!.. Lond. 174: 325-331, Carter. J. L, and Merritt, J, F. 1981, Evaluation of swimming ability as a means of island invasion !iy sma!! mamma!s in coastal Virginia. - Annl, Carnegie Mus, 50: 31-46. Connor, E, F. and McCoy. E. D. 1979. The statistics and biology of the species-area relationship. - Am. Nat, 113: 791833. Corbet. G. B. 1966. Records of mammals and their ectoparasites from four Scottish islands. - Glasgow Nat, 18: 426434. Crowel!. K. L. 1973. Experimental zoogeography: introductions of mice to small islands, - Am. Nat. 107: 535-558. Delany, M. J. 1970, Variation and ecology of island populations of the long-tailed field mouse {Apodetnus sylvaticus (L.)). - Symp. Zool. Soc. Lond. 26: 283-295. Dobson. F. S, 1982, Competition for mates and predominant juvenile male dispersal in mammals. - Anim. Behav. 30: 1183-1192. Dolgov, V. A. 1961. Variation in some bones of postcranial skeleton of the shrews (Mammalia, Sorieidae). - Acta therio!. 5: 203-228, Dueser, R. D, and Brown. W. C. 1980. Ecological correlates of insu!ar rodent diversity, - Eco!ogy 61: 50-56. Ellenbroek, E. J. M, 1980, Interspecific competition in the shrews Sorex araneus and Sorex minulus (Sorieidae, Insectivora): a population study of the Irish pygmy shrew. - J. Zoo!.. Lond, !92: 119-136. Grainger. J. P. and Fairley. J. S. 1978. Studies on the biology of the pygmy shrew Sorex minutus in ihe west of Ireland. J, Zoo!,. Lond, !H6: 109-!42. Greenwood. P, J. !9t<0. Mating systems. phi!opatry and dispersal in birds and mamma!s. - Anim. Behav. 28: 114{>-1162. Hanski, !. 19S6, Dynamics of shrews on small islands accord with the equi!ibrium mode!.-Bio!, J. Linn. Soc. in press. Hawes, M, L, !977, Home range, territoriality and ecological separation in sympatric shrews Sorex vagrans and Sorex obscurus. - J. Mamma!. 58: 354-367. Jensen. B., Lau!und. B, and Engelstoft. C, 1980. The common shrew Sorex araneus and small rodents in the Hanherred Island on northwest coast of Denmark, - Flora Fauna 86: 7376. Jewell. P. A. 1966. Breeding season and recruitment in some British mamma!s confined to small islands. - In: Rowland, L W. (ed). Comparative biology of reproduction in mamReferences mals. Zoo!. Soc. Symp. No. 15. Academic Press. Adlcr, G. H- and Wilson, M, L. 1985, Small mammals on Mas- Keltock, W, L. and Parsons, P. A. 1970. Variation of minor non-metrical cranial variants in Australian Aborigines. sachusetts islands: the case of probability functions in clariAm, J. Phys. Anthrop. 32: 409-422. fying biogeographic relationships. - Oecologia (Berl.) 66: Kock. D. 1974. A Suncus mertensi, new speeies from Flores, [7S-IS6, Lesser Sunda Islands (Mammalia. Insectivora), - SenAulak. W. 1967. Estimation of small mammal density in three ekenb. Biol, 55: 197-204. forest biotopes, - Ekol. Pol. Scr. A. 15. 39: 755-778, Kostian. E, 1970. Habitat requirements and breeding biology Beer. J. R.. Lukcns. P. R. and Olson, D, 1954. Small mammal of the Root Vo!e, Microtus oeconomux (Pallas), on shore populations on islands of Basswood Lake. Minnesota. meadows in the Gu!f of Bothnia. Finland. - Ann. Zool. Ecology 35: 437-W5, Fenn. 7: 329-340. Berry, A, C, 1976. The anthropological value of minor variants Krebs. C. J.. Ke!!er, B, L. and -Bimarin. R. H. 1969. Microtus of Ihc dental crown.-Am. J, Phys. Anthrop. 45: 257-268, population bio!ogy: demographic changes in f!uctuating - and Berry. R. J. 1967, Epigenetic variation in the human populations of M. ochrogaster and M. pennsytvanicus in cranium,"-J- Anat. 101: 361-.379. southern Indiana. - Ecology 50: 557-607, Berry. R. J. 1968. The biology of non-metrical varianis in mice and men. - In: Brothwcll. D. R. (ed.). The skeletal biology Lomolino. M, V, 1982, Species-area and species-distance relationships of terrestrial mammals in t!ie Thousand Island of earlier human populations. Pergamon. London. Region, - Oecologia (Ber!.) 54: 72-75. - and Jakobson, M, E. 1975. Ecological genetics of an island - !984a, Mamma!ian island biogeography: effects of area, population of the House mouse (Musmusculus).-i. Zoo!., isolation and vagility. - Oecologia (Berl.) 61: 376-382. Lond. 175; 523-540. - , Jakobson, M. E. and Peters. J, 1978. The House mice of - 1984b. Immigrant selection, predation. and the distribu- nOLARCnC ECOLOGY Q:3 {XWh) tions of Microtus pennsylvanicus and Blarina brevicauda on Pucek. Z, 1969. Trap response and estimation of numbers of is!ands. - Am, Nat. 123: 468-483, shrews in removal catches. - Acta iheriol. 2S: 403-426. - 1985. Body size of mammals on islands: the island rule reRoberts. P, J. 1980. Diet of a kestre! {Falco tinnunculus) on examined. -Am. Nat. 125: 310-316. Bardsey Island UK. - Bird Study 27: 116. MacArthur. R, H- and Wilson. E, O. 1967, The theory of is- Schoener. T, W. and Schoener. A. 1983. Distribution of verteland biogeography, - Princeton Univ. Press, Princeton, brates on some very sma!! islands. I. Occurrence sequences Malmqvist, M, G. 1985. Character displacement and biogeogof individual species, - J, Anim. Ecoi. 52: 209-235. raphy of the pygmy shrew in Northern Europe. - Ecology Searie, A, G. 1954, Geneticai studies on the skeleton of the 66: 372-377. mouse. IX. Causes of skeletal variation within pure lines. McTaggart Cowan. I. 1941, Insularity in the genus 5orpjt on the J, Genet. 52: 68-102. north coast of British Columbia. - Proc. Biol. Soc. Wash- Skaren. U. 1980. The water as a barrier in spreading of shrews. ington 54: 95-108. (In Finnish with English summary). - Savon Luonto 12: 44Michielson. N, C. 1966. intraspecific and interspecific eompe47, tition in the shrews Sorex araneus L.. and Sorex minutus L. Sikorski. M. D. 1982. Non-metrical divergence of isolated - Arch. Neer!, de Zool. 17: 73-174. populations of Apodemus agrarius in urban areas. - Acta - . Gerink, A. and Moerel. L. 1980, De dwergspitsmuis, de theriol. 27: 169-180, enige Sorex-soort op Ameland, - Lutra 23: 33-41. Sj0vo!d, T. 1977. Non-metrica! divergence between ske!eta! O'Keeffe.D, A. and Fairley. J. S, 1981, 2 population studies of populations. - Ossa 4. supp!. 1. 133 pp. Irish pygmy shrews Sorex minutus. - Ir. Nat. J, 20: 269-275. Smith, A. T. 1974. The distribution and dispersal of pikas: conOkhotina, M, V. 1976. A new form of shrew (Insectivora, Sorsequences of insular pt>pulation structure. - Ecology 55: ieidae) from Moncron Islands. Russian SFSR, USSR, 1112-!119, Zool. Zh. 55: 59(K595. - 1980. Temporal changes in insu!ar populations of the pika - 1977. Shrews (Insectivora, Sorieidae) of Sakhalin I.s!and, {Ochotona princeps). - Ecology 6!: S-13, Zoo!. Zh, 56: 243-249. Spitzenberger. F. !977, The mammalian fauna of Cyprus. Part O!ert. J, 1975. A contribution to the knowledge of the smal! I: Insectivora and Rodentia. - Ann, Naturhist. Mus. Wien mamma! fauna of the Lofoteti Islands. - Zool. Beitr. 21: 81:401^142, 135-142, Thomas. H, H,, Jones. G. S, and Dibblee. R. L. 1980. Sorex Olsen, O. W, 1969. Hymenolepis prihilofensis new species of palustris, new record on Prince-Edward Island, Canada, Cestoda (Hymenolepidipae) from the Pribilof shrew Sorex Can, Fie!d-Nat. 94: .329-33L prihilofensis from the Pribilof Is!ands, A!aska. - Can. J. Trus!ove. G. M, !%1, Genetica! studies on the skeleton of the Zoo!. 47: 449-454. mouse. XXX, A search for corre!ations between some miPankakoski. E, 1979. The cone trap - a useful tool for index nor variants, - Genet. Res,. Camb. 2: 431^38. trapping of sma!! mammals. - Ann, Zool, Fenn. 16: !44- Webb. W. L, 1965. Sma!! mamma! popu!ations on islands. 150. Ecology 46: 479^88, Pieper. H, 1978. A new species of Crocidura (Mammalia, Sor- Yalden. D. W, 1981. The occurrence of the Pigmy shrew Sorex ieidae) from the Island of Crete, Greece. - Bonn Zool. minutus on moorland and the implications for its presence Beitr. 29: 281-286. in Ireland, - J. Zool., Lond. 195: 147-156. Pokki. J. 1981, Distribution, demography and dispersal of the field vole {Microtus agrestis) in Ihe Tvarminne archipelago, Finland. - Acta Zooi. Fenn. 164. 48 pp. HOIARCTtC ECOLOGY 9:.l (mSf.;

Journal

Ecography – Wiley

Published: Oct 1, 1986

There are no references for this article.