Acta Herpetologica 4(2): 143-151, 2009 Habitat features and distribution of Salamandra salamandra in underground springs Raoul Manenti1, Gentile F. Ficetola1,2, Barbara Bianchi3, Fiorenza De Bernardi1 1 Dipartimento di Biologia, Università degli Studi di Milano, Via Celoria, 26, I-20133 Milano, Italy. Corresponding author. E-mail: raoul.manenti@unimi.it. 2 Dipartimento di Scienze dell’Ambiente e Del Territorio, Università degli Studi di Milano Bicocca, Piazza della Scienza, 1, I-20126 Milano, Italy. 3 Comitato per la difesa delle Bevere, Via Garibaldi, 10, I-20040 Capriano di Briosco, Briosco (Milano), Italy. Submitted on: 2009, 22nd February; revised on 2009, 15th August; accepted on 2009, 31st August . Abstract. Subterranean habitats are among the less known terrestrial habitats, but can reveal an unexpected biodiversity, and can play an underestimated role for amphibians. The fire salamander Salamandra salamandra is sometimes found in underground envi- ronments, but the factors affecting its distribution in subterranean spaces remain sub- stantially unexplored. We repeatedly surveyed some hypogeous springs, such as drain- ing galleries and “bottini” in NW Italy, in order to evaluate the relationship between environmental features and distribution of S. salamandra in these underground springs. We performed visual encounter surveys to assess the occurrence of larvae, juveniles or adults in springs. We also recorded four habitat variables: easy of access, isolation, mac- robenthos richness and forest cover of the surrounding landscape. We used generalized linear models to evaluate the relationships between habitat features and occurrence of larvae. We observed larvae of S. salamandra in 13 out of 22 springs; their presence was associated to springs with high easy of access and with relatively rich macrobenthos communities. In underground springs, larval development apparently required longer time than in nearby epigeous streams. Nevertheless, S. salamandra can attain metamor- phosis in this environment. The occurrence of S. salamandra in underground environ- ments was not accidental, but repeated in the time and interesting from an ecological point of view, confirming the high plasticity of the species. Keywords. Amphibians, cannibalism, cave, draining galleries, ecology, fire salaman- der, habitat, larval development. INtRoDuctIoN In the last years, biospeleology has had a remarkable development. Biospeleology has extended its attention not only to cave habitats, but also to interstitials ones, espe- cially ground waters (Danielopol and Griebler, 2008). An increasing number of studies 144 R. Manenti et alii has shown that the subterranean domain, formerly considered as a species-poor environ- ment, often reveals an unexpected biodiversity (Botosaneanu and Stock, 1997; Bottazzi et al., 2008). Subterranean communities are composed by both hypogean and epigean taxa (Danielopol and Griebler, 2008). The occurrence of amphibians in underground habitats is reported in both the herpetological and biospeleological literature (Gimenez-Lopez and Guarner Deu, 1982). The life cycle of some taxa is completely (e.g., Proteus) or partially (e.g., Speleomantes) connected with caves and interstitial habitats. Furthermore, several amphibians can use underground habitats even if they do not complete their life cycle there, i.e., they are occasional trogloxenes (Romero, 2001; Sket, 2008). A number of Euro- pean amphibians have been more or less regularly found inside caves. Multiple mecha- nisms can explain the underground presence of amphibians: underground spaces can be winter shelters, hiding places during the active season, and feeding habitats (Baumgart, 1981; uhrin and Lesinsky, 1997). Furthermore, caves sometime act as natural traps where amphibians fall or are transported by the water flow (uhrin and Lesinsky, 1997). The fire salamander Salamandra salamandra has been repeatedly found in under- ground environments (Baumgart, 1981; Gimenez-Lopez and Guarner Deu, 1982; uhrin and Lesinsky, 1997; Razzetti et al., 2001). Adults and metamorphs are often observed dur- ing latency periods, with fidelity of the individuals to shelter places (Feldman, 1967; Baun- gart, 1981). Furthermore, some studies reported the presence of larvae in subterranean damp biotopes. Gimenez-Lopez and Guarner Deu (1982) point out the finding of a larva in natural cave of catalonia; uhrin and Lesinsky (1997) suppose that salamanders reproduce in Slovakian underground spaces on the basis of records of larvae in caves and galleries; and Veith (1986) supposes that reproduction in underground environments is common in the Rhine valley. In Italy, several authors have observed larvae of S. salamandra in both natural and artificial underground environments (Bressi, 1995; Bressi and Dolce, 1999; Razzetti et al., 2001). Nevertheless, data on the occurrence and development of S. salaman- dra in hypogeous habitats remains anecdotic. Furthermore, the factors affecting the distri- bution of salamanders in subterranean environments are substantially unexplored. Recent observations showed that S. salamandra sometimes lays larvae inside the brooks or in the pools of some artificial hypogeous biotopes, such as draining galleries and other subterranean springs in Northern Italy (Manenti, 2007, 2008). The aim of this study was evaluating the distribution of S. salamandra in subterranean springs, to under- stand the ecological determinants that affect their occurrence, and whether these environ- ments can be a suitable reproductive habitat. MAtERIALS AND MEthoDS Underground springs considered We considered two typologies of subterranean artificial springs: draining galleries and the so called ‘bottini’. Draining galleries (Fig. 1A) are characterized by an almost horizontal tunnel that penetrates the side of a slope, to catch the subterranean water of a spring and bring it out- ward. Draining galleries are associated to traditional agriculture, and are found in Europe, Asia and Northern Africa (Balland, 1992). Even if they are less known than other spring typologies, draining 145Salamandra salamandra in underground springs galleries are widespread in Italy. Some of them are very ancient, such as the Etruscan spring flow tunnels (caponetti, 2005) or the galleries in the towns of Matera and Siena (Kucher, 2005). They house small brooks and/or water reservoirs. ‘Bottini’ (Fig. 1B) are small buildings with limited sub- terranean development and with basins for water collecting. Both typologies are frequently obsolete and unused. In most cases, they are hidden and difficult to reach (Manenti, 2008). In the study area, draining galleries have a limited width (max 2 m); and a length varying from 5 up of 100 meters. The brooks and the pools within galleries usually have limited depth (average 20 cm). The length of Bottini is usually 3-4 m; the average water depth is 41.7 cm. Study area and surveys We surveyed 22 subterranean springs in Lombardy (Northern Italy) between the districts of Lecco, como, and Milan (Fig. 2). The study area is comprised in the catchment basins of the Lam- bro, Adda and Seveso rivers. It is characterized by hilly and mountainous reliefs with a good cover of broadleaved woodlands. In order to identify and reach the springs, we used information available in local studies on troglophyle molluscs and springs features (Nardo and Guglielmin, 1996; Pezzoli, 2007), we collected information from local people and local environmental organisation, and we directly explored the countryside. All springs were permanent or nearly permanent. From September 2007 till January 2009, we surveyed each springs 2-8 (median: 3) times. Most of springs were surveyed 2-5 times; two springs with larvae were surveyed 7-8 times, at inter- vals of one-two months, to better monitor the larval development. The occurrence of larvae in springs was not related to the number of surveys (logistic regression, c21 = 1.05, P = 0.31), indicat- ing that variation in the number of surveys did not bias the results of our analyses. We performed visual encounter surveys to assess the presence and the abundance of S. salamandra larvae, juve- niles and adults. The surveyed waterbodies are usually small, always have very clear water, therefore visual census allow a reliable estimation of presence/absence of larvae in the water. When larvae of S. salamandra are present, their per-visit detection probability is > 90% (Manenti et al., 2009), there- fore with two-three surveys the cumulative probability of undetecting present larvae is < 1%. Larvae Fig. 1. Example of the access tunnel of (A) a draining gallery and (B) a “bottino”. 146 R. Manenti et alii were assigned to three development stages on the basis of total length and morphology (Jusczcyk and Zakrzewski, 1981): 27-45 mm; 46-65 mm; metamorphs with body size > 60 mm. We recorded four environmental variables to describe these subterranean habitats and to evaluate the relationship between habitat features and S. salamandra (Manenti et al., 2009): a) easy of access for salamanders, measured using a rank scale (1 = completely closed by doors or other obstacles and apparently inaccessible; 2 = difficult access because of doors or other obstacles; 3 = open and accessible); b) isolation, measured as the distance between each spring and the nearest known laying site of S. salamandra, obtained from previous studies (Manenti et al., 2008; Ficetola et al., 2009; Manenti et al., 2009) or by direct observations during surveys; c) richness of the com- munity of benthonic macro-invertebrates, measured as the number of taxonomic units (see Ghetti, 1997; in every spring, we sampled macrobenthos by moving the substrate for 5-10 minutes, and we used a thin-mesh dip net to collect the invertebrates); d) forest cover measured as forest cover per- centage within 400 m from each sampling point on the basis of the 1:10000 Vector Map of Lom- bardy, using the ESRI ArcView 3.2 GIS. Forests are the major habitat of adults, and forest cover at this scale is strongly associated with populations of S. salamandra (Ficetola et al., 2009). Statistical analyses We used generalized linear models (GLMs), assuming binomial error, to evaluate the relation- ships between habitat features and occurrence of larvae. First, we analyzed all univariate relation- ships. We also tried to build GLMs describing the relationships between multiple habitat features and Fig. 2. Study area. White circles show the location of the draining galleries, while triangles of the “botti- ni”; grey: hidrographic network. Some circles and triangles are partially superimposed, due to geographic proximity. 147Salamandra salamandra in underground springs occurrence of S. salamandra, subjected to the constraint that all variables in a given GLM must have variance inflation factor ≤ 5 (Bowerman and o’connell, 1990). We built GLMs with all combina- tions of two or three variables, testing also for interactions. however, none of these GLMs with mul- tiple variables showed Akaike Information criterion lower than the best univariate model, probably because of the limited sample size (see Burnham and Anderson, 2002). Therefore, in the results we report the univariate models only. We also calculated Nagelkerke’s R2 (R2N) as a measure of the vari- ance explained by GLMs. If necessary, we transformed variables to improve normality (see table 1). Table 1. habitat features of springs with and without larvae, univariate relationships between habitat fea- tures and presence / absence of larvae of Salamandra salamandra, and comparison with epigeous streams with larvae from the same study area (sites in Ficetola et al., 2009). For easy of access, we report the medi- an and the range of variation; for isolation, macrobenthos richness and forest cover, we report mean ± SE. R2N: Nagelkerke’s R2. Presence of larvae c21 P R2N Epigeous streams Yes (n = 13) No (n = 9) (n = 45) Easy of access 3 (2-3) 2 (1-3) 10.57 0.001 0.51 - Isolation (m)1 475 ± 250 471 ± 292 0.177 0.674 0.01 - Macrobenthos taxa2 2.38 ± 0.39 1.33 ± 0.62 5.48 0.019 0.30 14.9 ± 1.5 Forest cover3 0.66 ± 0.07 0.64 ± 0.09 0.06 0.810 0.00 0.65 ± 0.03 1: log transformed prior to analyze; 2: square-root transformed prior to analyze; 3: square-root arcsine transformed prior to analyze RESuLtS We observed larvae of S. salamandra in 13 out of 22 springs. Furthermore, we found a recently metamorphosed individual (65 mm) in one spring where we did not record lar- vae. We found adults in five tunnels. We observed recently laid larvae in November 2007, May-April 2008 and December 2008. In nearly all the cases, these larvae overlapped with older larvae previously laid. All the springs with larvae in autumn 2007 received at least a new deposition in spring or autumn 2008. In all springs, we observed multiple develop- ment stages during the same visit, indicating the presence of larvae from different clutch- es. The richness of macrobenthos was usually low. The average number of taxonomic units observed was 1.95 per spring (SE = 0.35), and was much lower than observed in epigeous laying streams (Manenti et al., 2009; table 1). univariate tests showed that the presence of larvae was associated to the springs with high easy of access and with rich macrobenthos. The relationships with isolation and for- est cover were non significant (table 1). Easy of access was the variable with the largest explanatory power, and explained 51% of variation of the distribution of S. salamandra larvae (table 1). We observed metamorphs in four springs. Furthermore, in all springs with larvae we recorded the presence of late development stages. however, in four sites, the exit of meta- morphs from springs seemed to be very difficult. We directly observed episodes of can- nibalism in two galleries, in which larvae at late development stages fed on small larvae. 148 R. Manenti et alii DIScuSSIoN The occurrence of S. salamandra in the hypogeous springs was not accidental, but repeated in time and interesting from an ecological point of view. All the study springs are underground headwaters. Therefore, the presence of larvae can not be explained by drift- ing or trapping from epigeous environments. Instead, the presence of larvae suggests that these headwaters are used by females for laying. Furthermore, in all springs with larvae we observed repeated laying during multiple seasons, indicating that laying in these sites was not occasional. Easy of access was the variable most strongly related to the presence of larvae. In prac- tice, we observed larvae in most of springs accessible to adults. This suggests that under- ground springs are selected by females for laying. underground springs have permanent hydroperiod, and can be particularly important in the areas where epigeous streams are temporary. Furthermore, we observed a significant association with macrobenthos richness. Benthonic invertebrates constitute the major prey items of larvae, and therefore this rela- tionship is not surprising (see Manenti et al., 2009). The presence of benthos can be partic- ularly important in these habitats, where invertebrates never reach high densities. It should also be noted that easy of access and macrobenthos richness were positively related (Spear- man’s correlation, rS = 0.55, P = 0.008), probably because easy of access increases also colo- nization by insects. Therefore, it is possible that these two variables have a synergic effect. Nevertheless, the richness of invertebrates remained low in all springs (table 1). This can be a critical factor for the survival and development of larvae in this environment. Indeed, the scarcity of shelter and of invertebrates might favours cannibalism. During our surveys, we observed some episodes of cannibalism; furthermore, in several cases we observed larvae with wounds and bite marks in areas without predator insects (see below), suggesting that cannibalism can be frequent. S. salamandra often performs cannibalism; in suboptimal habitats, cannibalism can be the only strategy allowing some larvae (usually the oldest ones) to reach metamorphosis (Eitam et al., 2005). Previous studies showed that, in epigeous streams, larvae are associated with high forest cover, probably because forests are the major habitat of adults (Ficetola et al., 2009; Manenti et al., 2009). The lack of a significant relationship (table 1) does not mean that forest pres- ence is unimportant. Most of the studied streams are in highly forested landscapes (table 1); the average forest cover across all sites (65%) was similar to the cover in epigeous streams used for laying by salamanders (table 1: Ficetola et al., 2009; Manenti et al., 2009). There- fore, forest presence was not a limiting factor in the landscapes studied. We did not observe a relationship with isolation, and we observed reproductions also in springs very far from the nearest occupied epigeous stream (table 1). Amphibians often live in network of metap- opulations, and isolation has strong negative effects on the occupancy of wetlands (Ficetola and De Bernardi, 2004; Zanini et al., 2009). however, our knowledge of breeding streams is probably incomplete, and we might have missed nearby sites. Amphibians living in cave environments usually have delayed development (clergue- Gazeau, 1975). We found a similar pattern, with the coexistence of multiple larval stages. our observations suggest that larval development in these caves can require more than eight months. This is more than twice the age at metamorphosis in epigeous streams (usu- ally 3-4 months: Nöllert and Nöllert, 1992). The slow development rate can be caused by 149Salamandra salamandra in underground springs cold temperature, lack of light, and scarcity of food items (e.g., macrobenthos). on the other hand, underground habitats can have advantages, for example because of the con- stant thermal environment, which allows development also during winter, or the scar- city of predators. For instance, during surveys and macrobenthos samplings, we never observed fish or predatory insects, even those that typically inhabits other S. salamandra breeding sites (e.g., Nepa cinerea; larvae of Cordulegaster boltoni or other odonata). The observation of metamorphs in several springs confirm that these areas can be suitable habitat for larval development. Nevertheless, some of the springs can act as traps. In three sites, the vertical banks did not allow the adults or metamorphs to leave the springs, and we found drown animals. S. salamandra shows high plasticity, has surprising local adaptations, and can perform larval development in habitats ranging from fast running streams to ponds (Weitere et al., 2004). The observation of larvae in underground springs is a further confirmation of this plasticity. Indeed, the occurrence of S. salamandra in underground, artificial springs is also known for other countries (e.g., Switzerland: K. Grossenbacher pers. comm.) and similar breeding sites are probably present also in other areas of its distribution range (Baumgart, 1981). underground environments are among the less known terrestrial habitats, and can play an underestimated role for amphibians. Draining galleries and other underground springs could be very useful to study plasticity and adaptations in S. salamandra. AcKNoWLEDGEMENtS We are grateful to: Suor Donata, E. Pezzoli, L. Kalcich, P. Pozzoli, F. Prada, M. cappelli and D. Magni for helping in locating springs and logistic support during surveys. The comments of one anonymous reviewer improved a previous draft of the manuscript. REFERENcES Balland, D. (1992): Les eaux cachées. Études géographiques sur les galleries drainantes souterraines. Dèpartement de Géographie, université Sorbonne, Paris. Baumgart, G. (1981): observations sur l’hibernation de quelques amphibiens dans les anciennes mines vosgiennes: Salamandre tachetée, Grenouille rousse et crapaud commun. Aquarama 58: 42-72. Botosaneanu, L., Stock, J.h. (1997): Stygofauna of oman .1. A new freshwater stygobiont cyathura (Isopoda, Anthuridae), from interstitia of coarse wadi sediments in oman. Ann. Limnol.-Int. J. Limnol. 33: 79-84. Bottazzi, E., Bruno, M.c., Mazzini, M., Pieri, V., Rosetti, G. (2008): First report on copep- oda and ostracoda (crustacea) from northern Apenninic springs (N. Italy): a faunal and biogeographical account. J. Limnol. 67: 56-63. Bowerman, B.L., o’connell, R.t. (1990): Linear statistical models. PWS-Kent, Boston. Bressi, N. (1995): catalogo della collezione Erpetologica del Museo civico di Storia Natu- rale di trieste. I-Amphibia. Museo civico di Storia Naturale di trieste, trieste. 150 R. Manenti et alii Bressi, N., Dolce, S. (1999): osservazioni di Anfibi e Rettili in grotta. Riv. Idrobiol. 38: 475-481. Burnham, K.P., Anderson, D.R. (2002): Model selection and multimodel inference: a prac- tical information-theoretic approach. Springer Verlag, New York. caponetti, L. (2005): Gallerie drenanti e sistemi idraulici Etruschi: il caso di tuscania. Ecomusei Regione Piemonte, trino (Vc). clergue-Gazeau, M. (1975): Effets de la vie cavernicole sur la reproduction des Amphi- biens. Bull. Soc. Zool. Fr. 100: 665-666. Danielopol, D.L., Griebler, c. (2008): changing Paradigms in Groundwater Ecology - from the ‘Living Fossils’ tradition to the ‘New Groundwater Ecology’. Int. Rev. hyd- robiol. 93: 565-577. Eitam, A., Blaustein, L., Mangel, M. (2005): Density and intercohort priority effects on lar- val Salamandra salamandra in temporary pools. oecologia 146: 36-42. Feldman, R. (1967): Nachweis der ortstreue des Feuer-salamanders, Salamandra salaman- dra terrestris Lacépède, 1788, gegenuber seinem Winterquartier. Zool. Anz. Leipzig 178: 42-48. Ficetola, G.F., De Bernardi, F. (2004): Amphibians in an human-dominated landscape: the community structure is related to habitat features and isolation. Biol. conserv. 119: 219-230. Ficetola, G.F., Padoa-Schioppa, E., De Bernardi, F. (2009): Influence of landscape elements in riparian buffers on the conservation of semiaquatic amphibians. conserv. Biol. 23: 114-123. Ghetti, P.F. (1997): Indice Biotico Esteso (I.B.E.): Manuale di applicazione. Provincia Autonoma di trento, trento. Gimenez-Lopez, S., Guarner Deu, N. (1982): Distribucion hipogea de Salamandra sala- mandra. Laurenti (Amphibia Salamandridae) en San Lorenç del Munt i Serra de l’obac (terrasa, provincia de Barcelona, Espana). P. cent. pir. Biol. Exp. 13: 43-45. Jusczcyk, W., Zakrzewski, M. (1981): External morphology of larval stages of the spotted salamander Salamandra salamandra (L.). Acta Biol. cracoviensa 23: 127-135. Kucher, M. (2005): The water supply system of Siena, Italy. Routledge Ed., Florence. Manenti, R. (2007): considerazioni sulla cenosi erpetologica delle gallerie drenanti. Riv. Idrobiol. 43: 114-118. Manenti, R. (2008): Amphibiens des sources et galeries drainantes en territoire préalpin, l’exemple du Mont Barro et du Mont de Brianza (Lombardie, Italie). Bull. Soc. herp. Fr. 128: 25-40. Manenti, R., Ficetola, G.F., De Bernardi, F. (2009): Water, stream morphology and lands- cape: complex habitat determinants for the fire salamander Salamandra salamandra. Amphibia-Reptilia 30: 7-15. Manenti, R., Ficetola, G.F., Padoa-Schioppa, E., De Bernardi, F. (2008): Spatial autocorre- lation and distribution of Salamandra salamandra – a preliminary analysis. In: her- petologia Sardiniae, pp. 341-344. c. corti, Ed, Edizioni Belvedere, Latina. Nardo, A., Guglielmin, M. (1996): Le Sorgenti del Barro. Quaderni del Parco Monte Barro, 3: 1-48. Nöllert, A., Nöllert, c. (1992): Die Amphibien Europas. Kosmos, Stuttgart. Pezzoli, E. (2007): I molluschi e i crostacei delle sorgenti e delle acque sotterranee della Lom- bardia: censimento delle stazioni. Parco Regionale del Monte Barro, Galbiate (Lc). 151Salamandra salamandra in underground springs Razzetti, E., Bonini, L., Barbieri, F. (2001): Riproduzione in grotta di Salamandra sala- mandra e Salamandrina terdigitata negli Appennini settentrionali. Atti 3° congresso nazionale ShI, (Pavia, 2000). Pianura 13: 181-184. Romero, A. (2001): The biology of hypogean fishes. Kluwer, Dordrecht. Sket, B. (2008): can we agree on an ecological classification of subterranean animals? J. Nat. hist. 42: 1549-1563. uhrin, M., Lesinsky, G. (1997): Mechanism of occurrence of amphibians in an under- ground spaces in Slovakia: preliminary data evaluation. Proceedings of the 12th International congress of Speleology, La chaux de fonds Switzerland: 3: 325-327. Veith, M. (1986): Feuersalamander - Salamandra salamandra. In: Die Amphibien und Reptilien in Rheinland-Pfalz A. Bitz, Ed, Ges. fur Naturschutz und ornithologie Rheinland-Pfalz. Weitere, M., tautz, D., Neumann, D., Steinfartz, S. (2004): Adaptive divergence vs. envi- ronmental plasticity: tracing local genetic adaptation of metamorphosis traits in sal- amanders. Mol. Ecol. 13: 1665-1677. Zanini, F., Pellet, J., Schmidt, B. (2009): The transferability of distribution models across regions: an amphibian case study. Divers. Distrib. 15: 469-480.