THE VELIGER © CMS, Inc., 2001 The Veliger44(1):8-19 (January 2. 2001) Influence of Vegetation Type on the Constitution of Terrestrial Gastropod Communities in Northwest Spain PAZ ONDINA Dpto. Bioloxia Animal, Facultade de Bioloxi'a, Universidade de Santiago de Compostela, A Coruna, Espaiia; e-mail: [email protected] AND SALUSTIANO MATO Dpto. Ecoloxi'a e Bioloxi'a Animal, Facultade de Ciencias, Universidade de Vigo, Pontevedra, Espaiia Abstract. We investigated the influence ofthree different vegetation types on the establishment ofterrestrial gastropod communities in the northwestIberianPeninsula, usingboth an ordination technique (DetrendedCorrespondenceAnalysis) and a classification technique (Two-Way Indicator Species Analysis) applied to a 498 X 47 site-by-species abundance matrix (total number of individuals 17,902). The results of these analyses indicate that meadow sites are characterized by a group of species comprising Cionella lubrica (Miiller, 1174), Nesovitrea hammonis (Strom, 1765), and Zonitoides excavatus (Alder, 1830). Woodland sites are characterized by a group comprising Columella aspera Walden, 1966, Oxychilus alliarius (Miller, 1822), Acanthinula aculeata (Miiller, 1774), and Punctum pygmaeum (Drapamaud, 1801). Vitrea contracta (Westerlund, 1871), Discus rotundatus, and Aegopinella nitidula (Drapamaud, 1805) form a group of companion species to this latter community. INTRODUCTION (Detrended Correspondence Analysis) and a hierarchical classification technique (TWINSPAN). The view that vegetation type affects the distribution of terrestrial gastropods has been advanced from various an- DESCRIPTION OF the STUDY AREA gles and conclusions by authors including Baba, 1974; Beyer & Saari, 1977; Andre, 1982; Stamol, 1991, 1993; The study area (12,400 km-) comprises the Provinces of Cowie et al., 1995, and Hermida et al., 1995. Some au- A Coruiia and Pontevedra in western Galicia (northwest thors have suggested that distribution is not primarily re- Spain) (Figure 1). This area is bounded to the east by a lated to vegetation but rather to litter characteristics mountain system running north-south and reaching ele- (Bishop, 1977; Locasciulli & Boag, 1987). By contrast, vations ofup to 1100 m. Soils are generally poor, in view other authors have gone so far as to define gastropod ofthe predominance ofgranites and schists, togetherwith communities corresponding to specified phytosociologi- the high rainfall and hilly topography: cambisols, lepto- cal communities (Frank, 1981; Kornig, 1989; Stamol, sols, and regosols are the most frequent. Climate is oce- 1992; Baba & Bagi, 1997). It has also been suggested anic, characterized by high rainfall and mild tempera- that microhabitat characteristics are the most important tures. The potential vegetation overmostofthestudy area determinants of gastropod distribution (Drozdowski, (which falls within the Eurosiberian Region) is Quercus 1968; Boag & Wishart, 1982; Young & Evans, 1991). robur L. woodland. Depending on series, the codominant Most studies in this field are based on simple descrip- tree species may be Betula pubescens Ehm. or Castanea tions of the habitats in which different gastropod species sativa Miller, with Laurus nobilis L., Ilex aquifolium L., are found, and relatively few studies have applied quan- Crataegus monogyna Jacq., or Frangus alnus Miller in titative methods. In the present study, with the aim of the shrub layer. However, much of the study region is furthering understanding of the influence of vegetation currently occupied by the introduced species Pinus pi- type on the distribution of terrestrial gastropods, we col- naster Alton, P. radiata D., and Eucalyptus globulus La- lected498 samples fromthree biotopes (woodland, mead- bifl. ow, riverbank) in the northwest Iberian Peninsula (Figure Riverbank vegetation is strongly influencedby ground- 1). These three biotopes have distinct characteristics, and water level, with two different associations being recog- couldbe expectedtohave distinctmalacofaunas. Thedata nized, both dominated by Alnus glutinosa (L.), namely were analyzed with the aid of an ordination technique Valeriana pyrenaicae-Alnetum glutinosae, richer in ferns p. Ondina & S. Mato, 2001 Page 9 The analysis was performed with CANOCO 3.1 (Ter Braak, 1990) using the rare-species downweighting op- tion, by which all species with frequency below 20% of the maximumfrequency ofany species are downweighted in proportion to their frequency (Hill, 1979). For this analysis, the data matrix was first log-transformed (log[;2 -I- 1], where n is number of individuals). CANOCO can supply centroids (weighted averages) of environmental variables in the ordination diagram. To find out the centroids, a matrix was built in which the variable biotope was converted into a nominal variable, so that it was possible to represent it by points in the ordenation diagram (Ter Braak, 1988). To verify the results obtained by ordination, we used a divisive hierarchical classification technique (Two-Way Indicator Species Analysis, TWINSPAN; Hill, 1979). Figure 1 One of the basic ideas in TWINSPAN stems is that Location of the study area showing the 100 km squares of the each group of sites can be characterized by a group of UTM (Universal Transversal Mercator) Grid. differential species, species that appear to prevail in one side of a dichotomy. The idea of a differential species is essentially qualitative, and to be effective with quantita- and nemoral herbs, and the more disturbed Senecio bay- tive data must be replaced by a quantitative equivalent. onesis-Alnetum glutinosae. Meadow vegetation in the study area falls into the phy- This equivalent is the "pseudospecies." Each species abundance is replaced by the presence of one or more tosociological class Molinio-Arrhenatheretea elatioris. The most frequent species within this association include pseudospecies. The more abundant a species is, the more pseudospecies are defined. The levels of abundance that Agrostis capillaris L., Linum bienne Miller, Lolium per- are used in TWINSPAN to define the crude scale are here enne L., Trifolium dubium Sibth., Bellis perennis L., termed "pseudospecies cut levels." Plantago lanceolata L., and Holcus lanatus L. RESULTS MATERIALS and METHODS Ordination Samples were collected, by stratified random sampling, from a total of 498 woodland, riverbank, and meadow The results ofcorrespondence analysis ofthe log-trans- sites (166 sites in each biotope). At each site we obtained formed 498 X 47 site-by-species matrix are summarized a soil and litter sample of 100 X 50 X 5 cm (length X in Figure 2 and Tables 1 and 2. The first four axes ex- width X depth). In the laboratory, the samples were wet- plained the majority of total variance. Significance was sieved through a 7 mm mesh over a 0.5 mm mesh. Ma- calculated using a x~ procedure. terial retained by the second sieve was carefully exam- Absolute contribution values for each species were cal- ined under a magnifying glass, with the aim of finding culated according to ludez (1989). Assuming that the 47 all gastropods. Gastropods foundwere immersedin water, taxa have the same absolute contribution (defined as uni- then fixed in 70° alcohol. Only live specimens were in- form absolute contribution), every species would contrib- cluded, since otherwise the abundance oftestaceous spe- ute with 2.12% to the total variance ofa single axis (100/ cies is likely tobe overestimated (Andre, 1981; Phillipson 47). Species with absolute contribution value higher than & Abel, 1983). The 47 species found, together with the 2.12% would be the ones which better explain the axes species codes used in the tables and figures, are listed in (Judez, 1989). Such species are listed for axes I and II in Appendix I. Table 3. The resulting 498 X 47 site-by-species abundance ma- The interpretation of the results of this analysis in trix was analyzed by Detrended Correspondence Analy- terms of potential cause-effect hypotheses is made diffi- sis. DCA is a modification of Correspondence Analysis cult by the large number of samples and species, and by (CA) developed to overcome some conspicuous faults. the low inertias of the first three axes (despite their sta- The main modification, from which the technique obtains tistical significance). With the aim ofreducing data noise its name, is detrending, which attempts to remove the and betterrevealing the data structure, we thus performed "arch efect" in which the second and subsequent axes a second analysis (Figure 3) using those species that were appear as polynomial functions of the first axis and thus not downweighted, i.e., species that had been assigned a obscure the underlying gradient structure (Ter Braak, downweighting factor of 1, namely, Acanthinula aculea- 1986, 1988). ta, Arion intermedius, Aegopinella nitidula, Coumella as- Page 10 The Veliger, Vol. 44, No. 1 II Table 1 Downweighting values for each of the 47 species, as calculated by CANOCO. Aa 1.00 De 0.966 Os 0,355 Ma 0.095 •Mb Ai 1.00 Do 0.836 Pu 0.281 Eb 0.059 An 1.00 Ef 0.787 Zs 0.268 Ap 0.053 Ca 1.00 Ob 0.775 Ci 0.265 Ag 0.029 CI 1.00 Ps 0.751 Oc 0.237 Ce 0.029 Ct 1.00 Dl 0.659 Ve 0.234 Da 0.029 Dr 1.00 Cb 0.656 Mg 0.203 Dp 0.029 Nh 1.00 Lc 0.650 Og 0.174 Pe 0.029 Oa 1.00 H.S 0.459 Vi 0.161 Rd 0.029 Pp 1.00 Vp 0.477 Ah 0.128 Va 0.029 Vc 1.00 Cr 0.468 Bp 0.112 Zn 0.029 Ze 1.00 Oe 0.394 La 0.096 Pe The absolute contribution values for each species on Cr axes I and II were calculated (Judez, 1989) and listed in Os Table 5. Species with absolute contribution value higher Vi than 8.3% (100/12) are indicated in bold. Foraxis I, these Cb Mg species are (in decreasing order of contribution) N. ham- p(j, Do Zs .Ma monis, C. lubrica, C. aspera, A. aculeata, Z. excavatus, AP Og and O. alliarius, which together explain 77% of the var- p, ^" Ci iance on this axis. For axis II, these species are A. inter- Ob medius, O. alliarius, D. rotundatus, and C. tridentatum, •.Aa .Rd •^CI'VpP.s-'VVa®-Di ,1 which together explain 84% of variance on this axis. We base our interpretation on axis I, since axis II was not n '-^ BD •Dp -Ah significant at the 5% level (Judez, 1989). Ze •Zn According to Figure 3, it can be seen that there is a Ag •Ai group of three species (TV. hammonis. Z. excavatus, and •Nh C. lubrica) which can be clearly differentiated from the rest in relation to their position along the axis. Toward •Ef •Da the far right side of this axis, C. aspera, P. pygmaeum, and A. aculeata show a high correlation with this axis, •Ce although A. nitidula and V. contractu also show certain proximity, but with lower contributions to axis I. The two species with lowest absolute contributions to axis I were •Oa C. tridentatum andA. intermedius, which located close to •Oc the origin of this axis. O. alliarius shows a higher con- tribution to axis II than to axis I. In the plot of sites (Figure 4), most of the woodland sites are located on the right side of axis I, whereas most Figure 2 Table 2 Ordination of the 47 species on the first two axes extracted by Detrended Correspondence Analysis. Eigenvalues, percentage inertias, significance levels, and degrees of freedom for the first four axes extracted by DCA of the 498 X 47 site-by-species abundance matrix. pera, Cionella lubrica, Carychium tridentatum, Discus rotundatus. Nesovitrea hammonis, Oxychilus alliarjus, Axis Eigenvalue % Inertia df Piinctum pygmaeum. Vitrea contractu and Zonitoides ex- ctcheaenvtasatpgueesci(ieTnsearbtdliieasst,1r)i.buEtiivgoaenlnsuveasa,lluoeansgnd(tmhdeeeagosrruedreiesnaootffiosnferpaeaxreiadst)oi,monpaerore-f IIIIIIVI 0000....2422308621063823 4558....8864 ><<< 0000....00001155 444498881579 listed in Table 4. & p. Ondina S. Mato, 2001 Page 11 Table 3 Coordinates, weights, and absolute contributions to the first two axes extracted by DCA of the 498 X 47 matrix (see Table 2), for the 18 species whose contribution to at least one of these axes was greater than the average. Species Coord, axis I Coord, axis II Weight Contrib. axis I (%) Contrib, axis A. aculeata -1.7307 0.2755 0.0270 8.0873 0,2049 A. intermedius 0.1809 -0.7525 0.1129 0.3694 6.3930 A. nitidula -0.6896 0.6063 0.0909 4.3227 3.3414 C. aspera -1.8361 -0.5628 0,0280 9.4395 0,8868 C. lubrica 0.9731 0.2486 0.1255 11.8838 0,7756 C. barbara 1.3789 2.3641 0.0068 1.2929 3.8004 C. tridentatum -0.2055 0.7473 0.1030 0,4349 5.7521 D. reticulatum 1.5687 0.7886 0.0150 3.6912 0,9328 D. lombricoides 2.0974 0.1239 0.0052 2.2875 0,0079 D. rotundatus -0.5257 1.1147 0.0879 2.4292 10.9220 L. cylindracea -1.3697 -0.1054 0.0126 2.3638 0,0139 N. hammonis 1.0101 -0.9244 0.1497 15.2739 12.7920 O. alliarius -1.4451 -2.7767 0.0426 8.8962 32.8448 O. elegans 2.6413 5.3712 0.0030 2,0923 8.6549 P. pygmaeum -1.6918 -0.3795 0.0158 4.5245 0.2275 P. subvirescens 1.4076 0.3170 0.0114 2.2587 0.1145 v. contracta -1.1475 0.7172 0.0496 6.5311 2.5513 Z. excavatus 0.9669 -0.3249 0.0551 5.1512 0.5816 of the meadows are located on the left side. In view of this plot, the woodland gastropod community can be con- sideredto comprise C. aspera, A. aculeata, P. pygmaeum, and O. alliarius, accompanied by V. contracta, A. niti- •Oa dula, and D. rotundatus, while the community of open areas, meadows, comprises A^. hammonis, C. lubrica, and Z. excavatus. To confirm these conclusions, we calculated the cen- troids. To find out the centroids, a matrix was built in which the variable biotope was converted into a nominal Ai variable. So we assigned to each sample the value 1 or 0, according to their presence or absence into the consid- ered variable. As expected, the woodland centroid lies to the right of the plot, whereas the open-site centroids are plotted to the left (Figure 5). The riverbank centroid lies close to the origin, which is attributable to the fact that such sites represent various biotopes with highly variable •Nh •Ca characteristics intermediate between woodland and mead- ow. This heterogeneity of the riverbank sites makes it difficult to discriminate a clearly defined species corn- Pp Ze •An Table 4 •Aa •CI Eigenvalues, percentage inertias, significance levels and •Vc Ct degrees of freedom for the first four axes extracted by DCA •Dr of the 12 species. % Axis Eigenvalue Inertia X- df I 0.388 17,7 < 0.01 440 Figure 3 II 0.287 13.1 > 0.05 438 III 0.220 10.6 > 0.05 436 Ordination ofthe 12 species that were not downweighted on the IV 0.167 7.6 > 0.05 434 first two axes extracted by Detrended Correspondence Analysis. Page 12 The Veliger, Vol. 44, No. 1 On »• 0» o o ** S O ,° * •W cOrO•* ooO oo * *it o** . Figure 5 Plotofsamples (•, meadowF;ig*u,rreiv4erbank; O, woodland)onthe WCen—trwooioddslaonfd)theonthtrheeefibristottowpoesax(eMs—exmteraacdtoewd;byR—DCriAverobfantkh;e first two axes extracted by DCA ofthe 12 species. 12 species. munity. The characteristic that these riverbank samples dentata, and a second group (group B) including C. lu- have in common is constant moisture supply, which brica, N. hammonis, Z. excavatus, Deroceras reticulatum, would discriminate aquatic species (not considered in the Ponentina subvirescens, Deroceras lombricoides, Dero- present study). ceras laeve, and Vertigopygmaee. The results ofclassification of sites considering the 12 species used in DCA are summarized in Figure 8. Con- Classification sidering the groups resulting from the first split, the in- The species-abundance cut-offs selected for definition dicator pseudospecies for group A (with abundance class of pseudospecies for TWINSPAN were 0, 3, 6, 10, 20, in brackets) are A^. hammonis (1) and C. lubrica (1), while 40, and 100, giving up to seven pseudospecies per spe- those for group B are V. contracta (1), A. nitidula (1), cies. The results of the subsequent classification of sites andA. aculeata (1). Group A contains 79% ofwoodland by pseudospecies are summarized in Figure 6. The first sites, 57% of riverbank sites, and 34% of meadow sites, split separated a group (group A) containing most (73%) whereas group B contains most meadow sites, somewhat of the woodland sites, about half (48%) of the riverbank less than half of the riverbank sites, and a small propor- sites, and some (26%) ofthe meadow sites, from a group tion ofwoodland sites. The first split within group B sep- (group B) containing most (74%) of the meadow sites, arates group BB, with indicator pseudospecies C. lubrica the other half (52%) of the riverbank sites, and some (2), C. tridentatum (1), and Z. excavatus (2), containing (27%) of the woodland sites. most of the meadow sites and only a single woodland The results of the classification of species by sites are site, from group BA, containing all other group-B wood- summarized in Figure 7. The two species groups defined land sites. The first split within group A separates group by the first split are very similar to those obtained by AA, with indicator pseudospecies C. tridentatum (1), A. DCA: a first group (group A) including A. aculeata, C. nitidula (1), D. rotundatus (1), V. contracta (1), C. lu- aspera, O. alliarius, P. pygmaeum, A. nitidula, C. triden- brica (1), and A. aculeata (1), from group AB, with in- tatum, V. contracta, Euconulus fulvus, and Clausilia bi- dicator species O. alliarius (1). & p. Ondina S. Mato, 2001 Page 13 N=214 N=233 V. contracta 1 (95, 23) N. hammonis 1 (58, 177) A. nitidula 1 (120, 55) C. lubrica 1 (53, 126) A. aculeata 1 (62, 8) D. rotundatus 1 (96, 11) N=182 C. aspera 1 (63, 11) N. hammonis2 (114,4) A. intermedius 1 (72, 39) N=57 N=157 C. lubrica2 (92, 3) C. tridentatum 1 (62, 0) O. alliarius 1(35, 30) D. rotundatus 1 (7, 89) Z. excavatus 1 (56, 2) C. aspera2 (16,9) C. tiidentatum 1 (4, 89) A. nitidula 1 (13, 107) C. lubrica 1 (0, 53) AA AB BA BB 9 meadow 32 meadow 89 meadow 27meadow 5 riverbank 67 riverbank 69 riverbank 15 riverbank 43woodland 58woodland 29woodland 9woodland 41 meadow 116meadow 72 riverbank 79 riverbank 101 woodland 38woodland Figure 6 TWINSPAN classification ofthe samples considering the 47 species, showing the indicator pseudospecies foreach split. The classification of these 12 species by sites (Figure and buffer variation in moisture levels and temperature. 9) is again very similar to that obtained by DCA. The The resulting habitat characteristics are critical for the first split separates a group (group B) comprisingN. ham- establishment of certain species, and variations in these monis, C. lubrica, and Z. excavatus from the rest (group factors are the cause of the observed differences among A). The first species to split from group A are C. triden- communities. tatum and A. intermedius, in accordance with the more Taken together, the results ofthe different analyses in- variable behavior of these taxa (as was indicated by dicate that the C. lubrica, N. hammonis, and Z. excavatus DCA). The remaining species split into two groups: one characterize the snail communities present in open sites comprising V. contracta, D. rotundatus, and A. nitidula, (meadow). The other group comprises C. aspera, O. al- the other P. pygmaeum, O. alliarius, C. aspera, and A. liarius, A. aculeata, and P. pygmaeum, with preference aculeata. for wooded areas with more vegetation cover V. contracta, D. rotundatus, and A. nitidula form a DISCUSSION group of companion species to the woodland gastropod community, but are also important in riverbank commu- Gastropod populations exist in complex environments re- nities, where they are in fact more abundant. flecting the interaction of numerous factors, including A. intermedius and C. tridentatum show more irregular characteristics of the soil-humus-litter-vegetation system. behavior, their presence being more homogeneously dis- This system is clearly dependent on the herb and woody tributed, though both appearto have a strongerpreference layers, which contribute to litter formation, filter light. for open areas than for woodland sites. Page 14 The Veliger, Vol. 44, No. 1 A B Aip,'uMb ALnc,, MCgb,, COtb,,DVr,cHs ACae,, AEfg,,LAap,,OCaa Cr, Do BNp,h,CiP,s,CI,ZeDe OAeh,, VDaa,,VDle,,DVpp,,MZan Vi, Zs Oc, Og, Os, Pe Pp, Rd Figure 7 TWINSPAN classification of the 47 species. Although these are the only species for which statisti- more abundant in meadow. E. fulvus, by contrast, shows cally significant conclusions may be drawn, in view of a preference for woodland sites. their abundances and contribution to the ordination, valid Some authors, including Boycott (1934), whose study conclusions may also be inferred for a number of other provided the starting pointforthe majority ofmore recent species. Notably, some species that are relatively infre- studies, have concluded that gastropods show no specific quent in the sample as a whole may in fact be important association with vegetation, and that apparent relation- components of particular habitat types that form subcat- ships between gastropods and vegetation are due more to egories of the major categories (woodland, riverbank, environmental conditions than to the fact that the plants meadow). Species of this type may include the agrioli- in question are food sources (since the variety ofthe gas- macids D. reticulatum, D. lombricoides, and D. laeve tropoddiet means that this wouldbe alimitingfactoronly (which appear to show a preference for open areas), and under extreme conditions). Bishop (1977) considered that the group comprising P. subvirescens. Cochlicella bar- vegetation has no important effect on the composition of bara, and V. pygmaea (particularly the latter), which is the malacofauna, but that the litter layer (which provides Table 5 Coordinates, weights, and absolute contributions to the first two axes extracted by DCA for all 12 species included in the analysis. Species Coord, axis I Coord, axis II Weight Contrib. axis I (%) Contrib. axis A. aculeata 1.8491 -0.5263 0.0304 10.39 0.84 A. intennedius -0.1065 1.6338 0.1269 0.14 33.87 A. nitidula 0.7637 -0.4383 0.1022 5.96 1.96 C. aspera 1.9244 0.4041 0.0315 11.66 0.51 C. lubhca -1.0413 -0.5492 0.1411 15.29 4.25 c. tridentatum 0.1860 -0.8620 0.1163 0.40 8.64 D. rotundatus 0.5363 -1.0580 0.0989 2.84 11.07 N. hammonis -1.1624 0.4360 0.1683 22.74 3.20 O. alliarius 1.3454 2.5484 0.0479 8.67 31.10 P. pygrnaeum 1.7727 -0.1755 0.0177 5.56 0.05 V. contracta 1.1568 -0.7654 0.0558 7.46 3.26 z. excavatus -1.1854 -0.4309 0.0620 8.71 1.15 p. Ondina & S. Mato, 2001 Page 15 N=191 N=240 B N. hammonis 1 (168, 67) V. contracts 1 (17, 101) C. lubhca 1 (123, 56) A. nitidula 1 (43, 132) A. aculeata 1 (6, 64) N=77. ^N=114 N=169 C. tridentatum 1 (92, 2) O. alliarius 1 (33, 36) C. lubrica2 (2, 90) A. nitidula 1 (117, 15) C. tridentatum 1 (3, 58) D. rotundatus 1(89, 6) Z excavatus2 (0, 42) V. contracta 1 (92, 9) C. /uftnca 1 (56, 0) A. aculeata 1 (58, 6) BA BB AA AB 26 meadow 75 meadow 33 meadow 19 meadow 23 riverbank 38 riverbank 67 riverbank 15 riverbank 28woodland 1 woodland 69woodland 37woodland 101 meadow 52 meadow 61 riverbank 82 riverbank 29woodland 106woodland Figure 8 TWINSPAN classification ofthe samples, considering the 12 species considered in the second DCA. food and shelter) has a marked effect, important factors tions created by that biotope that determine the establish- being mineral content, surface bacterial and fungal pop- ment of the malacofauna. ulations, and the amount of usable litter. Locasciulli & Woodland vegetation may itself provide the gastropod Boag (1987) pointed out that litter layer characteristics fauna with the necessary conditions for survival: it buff- depend on the overlying vegetation, and took as their ers variations in temperature and humidity, and provides starting point the assumption that gastropods do not use shelter and varied microhabitats such as tree roots, fallen the litter layer directly, but rather nutrients derived from trunks, hollows, together with a usable humus and litter it; nevertheless, they stressed the importance of the litter layer. The tree cover performs a basic microclimatic role, layer for providing a stable microclimate. which is probably more important than the composition Inthe presentstudy, we have takenas ourstartingpoint of the litter layer. It is indeed the presence or absence of the view thateventhemostcommon andmosteuryecious climate-buffering tree cover that largely explains the dif- species do not occur in all vegetation types; all species ference in malacofauna between woodland and meadow prefer some biotopes over others, so that different bio- biotopes: the latter are exposed to marked daily and sea- topes can be considered rich or poor depending on the sonal variations in temperature, humidity, and light inten- conditions that they offer to the specific demands of the sity (Dillon, 1980; Boag & Wishart, 1982; Curry, 1994). species in question. Thus each biotope may contain char- Species that occur preferentially in meadow are probably acteristic associations, and it is thus probably the condi- more resistant to such variation, and escape its effects by Page 16 The Veliger, Vol. 44, No. 1 B A Z. excavatus C. tridentatum V. contracta P. pygmaeum A. intermedius N. hammonis D. rotundatus o. alliarius C. lubhca A. nitidula C. aspera A. aculeate Figure 9 TWINSPAN classification ofthe 12 species considered in the second DCA. burrowing into the soil (Stephenson, 1966 in Peake, 1978; considered were woodland biotopes, characterized by Outeiro et al., 1989; Outeiro et al., 1993), or are more various moisture and temperature indices; there was thus influenced by other non-microclimatic factors first, such no comparison with other biotopes. Similar conclusions as biotope factors, or soil factors, or factors not consid- were reached by Warebom (1982), who considered A^. ered in the present study, such as the quality of the litter hammonis and C. lubrica to be characteristic ofwoodland or the herb layer (again the only biotope considered), though it should be Riverbanks constitute such a heterogeneous category stressed that the characterization of biotopes in terms of that fauna ofboth types are present, especially species at malacofauna was not one of the goals of this study, so the limits ofthe requirements ofthe otherbiotopes. These that there was no comparison with other biotopes. This habitats thus probably behave as ecotones, as well as be- author also found that C. lubrica may occur at high abun- ing favorable because of water availability. dance in meadow. Walden (1955) reported that C. lubrica In order to facilitate comparison of our results with is a species typical ofwoodlands, but this conclusion was those of previous studies, we have summarized previous not based on a uniform sampling strategy (16 samples results in Table 6. Like us, most authors have reported C. were obtained from woodland sites, versus onlyfourfrom lubrica and A^. hammonis to be species of meadow, open sites). This author likewise reported that C. lubrica though only Riballo (1990) also included Z excavatus as may be abundant in swamps and marshes. Note, however, the third characteristic species of this community. Previ- that Walden's data analysis (using percentage abundanc- ous reports of species characteristic of woodland com- es) is not directly comparable with that used in the pre- munities, however, show some discrepancies. For exam- sent study. ple, Alonso (1977) concluded that C. lubrica is charac- The few previous studies to considerZ. excavatus have teristic of poplar groves. This is probably attributable to concluded that it is a woodland species (Boycott, 1934; the particular characteristics of poplar groves, in which Bishop, 1977; Anderson, 1983); this contrasts with the trees are planted in existing meadow, where C. lubrica present study, and with Riballo (1990), who found that was probably already present. Z. excavatus appears to be characteristic ofmeadow. Ri- Baba (1981) likewise considered C. lubrica to be a ballo (1990) states that the distribution ofZ. excavatus is constant species within the malacofauna of woodland wider than has been suggested by some previous authors; sites. This author's aim was to identify relationships be- indeed, Anderson (1983) also found this species in mead- tween the successional series of gastropod communities ow, and remarked that in acid soils (as in the present and plant communities along a river The only habitats study) it typically occurs in association with N. hammon- p. Ondina & S. Mato, 2001 Page 17 Table 6 Summarized findings of previous studies for the 12 species considered showing whether each species is judged to be characteristic of meadow (^), characteristic of woodland (•), or ubiquitous (). Authors CI Nh Ze Ca Oa Aa Pp Vc An Dr Ct Ai Alonso, 1977 • • • Anderson, 1983 • • • • Andre, 1982 * • • • • Baba, 1981 • • • • Badie & Rondelaud, 1979 * Bishop, 1977 • • • • • • • • • Boycott, 1934 • • • • • • Cameron, 1973, 1978 • • • • • • Cameron et al. 1980 • • • • Frank, 1979 * * • • • • Hermida et al., 1994 • • • • Jourdin et al., 1985 * * * Kornig, 1989 Mason, 1974 • • • • • • • * Matzke, 1976 * Meier, 1987 Mordan, 1977 * Ojea & Anadon, 1983 * • Ojea et al., 1987 * * • • • • Outeiro, 1988 * • • • • • Outeiro et al., 1989 * Paul 1975a h • • • • Paul, 1978a,b • • • Phillipson & Abel, 1983 • • • • Radea & Mylonas, 1992 • Riballo, 1990 * * * • • • • Riballo et al., 1995 • • • • Solhoy, 1981 • South, 1992 Stamol, 1993 • • Tattersfield, 1990 Valovirta, 1967, 1979 • • • • Walden, 1955 • • • • Wareborn, 1982 • • • • • • is. Boycott (1934) did not find this species in his wood- The present results indicate that C. tridentatum and A. land sites and, although he referred to it as a woodland intermedins should be considered as ubiquitous species, taxon, he stated that its special characteristics (notably whereas some previous studies have classified these spe- calciphobia and sensitivity to competition) lead it to live cies as characteristic of woodland. This difference is "where it can" and not "where it would like to." In any probably attributable to the fact that these previous stud- case, there have been few ecological studies of this spe- ies considered only woodland sites, therefore the presence cies, and there is a shortage ofdata to facilitate statistical in this biotope ofubiquitous species seems logical. More- investigation of putative relationships with particular over, some of the species classified in the present study types of biotope. as characteristic ofmeadow or riverbank have previously Our results for both P. pygmaeum and A. aculeata been classified as ubiquitous by previous authors. agree closely with previous reports; both have consis- It should be borne in mind that there have been few tently been identified as woodland species. Similarly, our studies in which the sampling method or the statistical results for C. aspera and O. alliarius, classified in the treatment ofthe data have been directly comparable with present study as characteristic of woodland, likewise those used in the present study. Furthermore, most pre- show close agreement with previous studies. The species vious studies included consideration of species that are classified in the present study as woodland-community absent from our region, or that did not show significant companion species have often been described as wood- habitat preferences in the present study. Andre (1982) land species, supporting our findings. studied terrestrial mollusk populations in Quercuspubes-