SALAMANDRA 48(3) 157–165 30M Oetcatmoboerrp 2h0o1si2s of SIaSlSaNm a0n0d3r6a– i3n3f7ra5immaculata Age and size at metamorphosis of Salamandra infraimmaculata larvae born in the laboratory and raised under different density regimes with food ad libitum Michael R. Warburg Department of Biology, Technion – Israel Institute of Technology, Haifa 32000, Israel e-mail: [email protected] Manuscript received: 29 November 2011 Abstract. Age and size at metamorphosis were studied in two half-sibling Salamandra infraimmaculata larval cohorts born and raised in the laboratory under three different density regimes, and fed ad libitum until they metamorphosed. There was no significant effect of density on the number of larvae metamorphosing. In the two cohorts studied, a signifi- cant positive relationship was observed with age at metamorphosis that increased with density, and a significant negative relationship between mass and density. In both cohorts, density did not appear to have an effect on either minimal or maximal age, mass or length at metamorphosis, nor on the range between maximum and minimum. There was no signifi- cant difference between the two cohorts in either age or length at metamorphosis. The difference in mass increased with density. The evolutionary significance of density effects on size (mass, length) and age at metamorphosis under unlimited resource (food) conditions, is by spreading out emergence of post-metamorphs onto land and their subsequent dispersal, and by affecting their size as adults and thereby their eventual maturity. Key words. Caudata, larval cohorts, density effects, metamorphosis. Introduction the season, lies in the advantage these larvae have when rainfall is continuous, since they are able to grow more Urodeles exhibit plasticity in the timing of, and in achiev- rapidly by feeding on newborn larvae (Cohen et al. 2005, ing an optimal size at metamorphosis (Wilbur & Collins Warburg 1992, 2009a, 2010). The larvae require at least 1973, Warburg 2011). The duration of the larval period five weeks to undergo metamorphosis (Warburg 2011). If (i.e., their age or the time to metamorphosis), as well as breeding is delayed because of delayed rainfall, there is a their size at metamorphosis depend on their larval growth risk that time could be insufficient to allow metamorphosis history (discussed in Cohen et al. 2005, 2006). Both the (Warburg 2010, 2011). timing of, and the size at, metamorphosis are critical for Aquatic fauna is rather poor when the ponds first form survival of the young post-metamorphs and as a conse- early in the season. Larvae then depend almost entirely on quence, for the recruitment of juveniles into the terrestrial their cannibalistic (sibling predation) trait, preying on late- population (Warburg 2011). born larval cohorts in order to obtain enough food to facil- This is especially true when conditions are suboptimal itate their own growth (Cohen et al. 2005). The survival of as are those prevailing on Mt. Carmel, Israel, during the larvae in spite of all these risks will enable their successful breeding season of Salamandra infraimmaculata Mar- survival as post-metamorphs. tens, 1885. Winter rains may start in October, eliciting In addition, the emergence on land of the young post- breeding in the salamanders. In some years, rainfalls may metamorphs will have to be timed (i.e., actual date) so that be sufficient to fill the rock pools, which are their main the soil will still be moist and temperatures sufficiently low breeding sites. However, in about 50% of the time, the pools to make successful dispersal possible (Warburg 2011). dry out, resulting in the loss of entire larval cohorts (War- This depends on the amount of rains falling until then, and burg 2009a). Nevertheless, when rains eventually resume, on the season. Emergence in winter is preferable to spring, some larvae may be saved, since they can survive for about which is characterized by hot spells (sciroccos). Moreover, a week on wet mud (Warburg 1986a,b). both the larvae’s age at metamorphosis (measured from The question remains: why start breeding early in the their birth in the laboratory because it can rarely be re- rainy season when about half of the broods might be lost? corded in nature), as well as the duration of the larval pe- The advantage in starting larval life as early as possible in riod (i.e., the time it takes to metamorphosis), are of great © 2012 Deutsche Gesellschaft für Herpetologie und Terrarienkunde e.V. (DGHT), Mannheim, Germany All articles available online at http://www.salamandra-journal.com 157 Michael R. Warburg significance. As a rule, larval period is positively related to two to five years after metamorphosis in order to release age at metamorphosis. Finally, the size attained by the lar- them into the wild as juveniles (Warburg unpubl. data). vae when metamorphosing is of great consequence for the Larvae were placed in glass ‘finger bowls’ (13.5 cm in di- survival of the post-metamorphs and later of the adults. ameter) or enamel troughs (15 × 25 cm) filled with stale However, metamorphosis can be delayed due to slow lar- (aged) tap water, 2 cm in depth. They were fed ad libitum val growth rates, resulting in small size at metamorphosis. (i.e., the larvae had available unlimited amounts of food To conclude: successful survival to adulthood depends on until they completed metamorphosis) with live Tubifex starting life as a larva as early in the season as made pos- worms or minced beef liver (no significant differences were sible, by rainfall (Warburg 1992, 2010), growing rapidly found in the effects of these two types of food on either age (Cohen et al. 2006), avoiding cannibalistic predation by or size at metamorphosis). The Tubifex were added after half-siblings (Cohen et al. 2005), metamorphosing as ear- cleaning the troughs and were always available to larvae. ly in the season as possible and at an early age (Warburg The water was changed every day. 2011), attaining a greater size, and, finally, dispersing as ear- Larval salamanders usually stay on the bottom most of ly in the season as possible in order to escape dehydration the time, hiding among stones. Only towards the end of of the soil. the metamorphic cycle will they swim to the surface for Density (i.e., larval crowding) was shown to have an ef- air. Consequently, density is calculated per surface area fect on urodelan larvae with regard to both growth rate rather than volume. Larvae were placed either individually and development to metamorphosis (Brunkow & Col- into finger bowls (42.4 cm2 /larva), or in enamel troughs lins 1996). Ohdachi (1994) has shown that density was in groups of either five (‘low density’, 28.7 cm2 /larva) or 10 negatively related to both time and size at metamorphosis. larvae (‘high density’, 14.3 cm2 /larva). Three replicates of High density can cause failure in metamorphosis (Wilbur each constellation were used. Towards the end of their lar- 1976). Scott (1990) in Ambystoma opacum (Graven- val period, they were transferred into spacious plastic con- horst, 1807), and Brodman (1996) have shown in both tainers with both water and soil that enabled them to crawl Ambystoma maculatum (Shaw, 1802) and A. jeffersonia­ onto land and metamorphose. num (Green, 1827) that both size and larval period are As the sample size was rather small, I used both t-tests negatively affected by density. and regression analysis for statistical analyses with Sigma- The objective of the present study is to establish to which Plot 9.0. The relationships between age, mass, and length, extent three different density regimes affect both age and and the three density conditions were tested using regres- size (mass, length) at metamorphosis of half-sibling sala- sion analysis. The difference between the two cohorts was mander larval cohorts born in the laboratory to the same examined by t-tests. female (i.e., a cohort), and fed ad libitum. Results Materials and methods The data for means and ranges in age and dimensions The data discussed here are based on a long-term study (mass, length) of the two half-sibling larval cohorts are giv- (1974–1998) of a single breeding population of Salaman­ en in Tables 1 and 2. Cohort I consisted of 96 larvae, and co- dra infraimmaculata (Warburg 2006, 2007a). This species hort II of 106 larvae. There was no significant difference be- is a ‘rare’ and endangered species found in the northern, tween the two cohorts (Appendix 1). No significant effect of mountainous part of Israel (IUCN Red List status “Near densities on the numbers of emerging metamorphs could Threatened”; www.iucnredlist.org). The metapopulation be seen in either cohort (Figs. 1, 2) (Appendix 2). A signifi- inhabiting Mt. Carmel survives in the most southeastern cant positive relationship between increasing age and den- portion of the genus’ distribution. sity was noted in cohort I (Figs. 3A–C) (Appendix 3). Like- Gravid females collected in the field delivered their lar- wise in cohort II, age at metamorphosis related significantly vae in the laboratory and were thereafter released into the to densities as did mass and length (Fig. 4A–C). In both wild (Warburg 2007b, 2008, 2009a). This is an ovovivipa- cohorts, no significant effect of density could be demon- rous species, laying eggs that hatch upon contact with water strated as affecting either mass or length with age (Figs. 5A– (Warburg et al. 1978/79). During the study period, a to- C, 6 A–C) (Appendix 4). Density did not appear to have tal of 74 half-sibling larval cohorts were born in the labora- an effect on either minimal or maximal age except at high tory (Warburg 2009c, 2010), two of which were used for densities in cohort II. In both cohorts, minimal and maxi- the study described in this paper. Thus, the identity of their mal mass and length did not show any significant difference mothers, their dates of birth, and their sizes at birth were (Fig. 7 A–C). The range between minimal and maximal age all known. These cohorts contained 4,085 larvae, all born to at metamorphosis showed an insignificant positive rela- these freshly collected females. Most of the larvae were re- tionship with increasing density in both cohorts (Fig. 8 A), leased into the ponds where their mothers had been collect- whereas both mass and length at metamorphosis decreased ed. Of the remaining larvae, 396 were raised in the laborato- with density. Finally, there was no significant difference in ry (at room temperature) until they metamorphosed (Co- either age, mass or length between metamorphosing larvae hen et al. 2005, 2006). Others were raised for an additional belonging to the two cohorts (Fig. 9 A–C) (Appendix 1). 158 Metamorphosis of Salamandra infraimmaculata Table 1. Age, mass, and length at metamorphosis of specimens in cohort I, fed ad libitum and raised under different density regimes (single: larvae kept individually in glass jars (42.4 cm2/larva); low = larvae kept at a density of 28.7 cm2/larva; high = larvae kept at a density of 14.3 cm2/larva). Means are followed by standard deviation, the range in parentheses and the sample size (N). single low high age (days) 44.1 ± 2.04 (41–47); N = 6 45.45 ± 2.56 (40–51); N = 6 46 ± 3.28 (41–76); N = 6 mass (g) 1.63 ± 0.14 (1.46–1.6); N = 16 1.36 ± 0.22 (1.03–1.79); N = 16 1.29 ± 0.23 (1.07–1.55); N = 16 length (mm) 66.3 ± 3.76 (64–70); N = 6 60.1 ± 1.87 (57–64); N = 16 60.2 ± 3.16 (56–70); N = 10 Table 2. Age, mass, and length at metamorphosis of specimens in cohort II, fed ad libitum and raised under different density regimes. For abbreviations see Table 1. single low high age (days) 44.4 ± 2.07 (61–70); N = 10 44.6 ± 4.28 (40–51); N = 9 47.9 ± 7.91 (41–76); N = 16 mass (g) 1.69 ± 0.09 (1.52–1.83); N = 10 1.51 ± 0.23 (1.24–1.92); N = 9 1.44 ± 0.21 (1.08–1.97); N = 16 length (mm) 65.6 ± 2.95 (61–70); N = 11 61.8 ± 3.68 (57–68); N = 9 59.2 ± 3.54 (54–65); N = 16 A A B B C C Figure 1. Density effects on numbers of metamorphs in cohort I. Figure 2. Density effects on numbers of metamorphs in cohort (A = single, B = low, C = high). See Appendix 2. II. (A = single, B = low, C = high). See Appendix 2. 159 Michael R. Warburg A A B B C C Figure 3. Average (A) age, (B) mass, and (C) length at metamor- Figure 4. Average (A) age, (B) mass, and (C) length at metamor- phosis in cohort I when raised under three different densities. phosis in cohort II when raised under three different densities. See Appendix 3. See Appendix 3. Discussion both age and larval period were known since individu- al clutches of half-siblings were either collected (Doody The main problem with studies on aquatic amphibian lar- 1996), or adult females laid eggs that hatched in the labora- vae concerns their origin and age, both of which are gener- tory (Ziemba & Collins 1999, Cohen et al. 2005, 2006). ally unknown. Thus, in some studies on density effects on Some of the variability found in the period to metamor- metamorphosis in urodeles, neither age nor larval period phosis and size of post-metamorphs can be attributed to could be known with accuracy since the larvae studied had these facts. Thus, it is not surprising that Wilbur (1976) been collected in the field (Chazal et al. 1994). In others, noticed a high variance in both time and size at metamor- individual egg clutches in nests attended by females were phosis within the same pond. Variation in duration of lar- collected in the field, separated, and hatched in the labora- val period and body size at metamorphosis was attribut- tory. Consequently, although the larval period was known ed to time of oviposition and density of conspecific larvae. (since the eggs hatched in the laboratory) in such cases, Such individual variation can affect population dynamics it was nevertheless not identical with age (from the actual (Brunkow & Collins 1996). point of time when the eggs had been laid) (Kusano 1981, Most studies attempted to show the effects of density Collins & Cheek 1983, Fauth, Resetarits & Wilbur (i.e., crowding or sharing the space with one or more lar- 1990, Doody 1996, Nishihara 1996, Walls 1998). More- vae), and resource (i.e., food resources) on time required over, in neither scenario can it be ascertained whether the until metamorphosis and size of post-metamorphs. Few larvae were indeed half-siblings, because it cannot be ruled studies tried to isolate and analyse these factors separate- out that either the larvae or the batches of eggs indeed orig- ly: the effect of density when food is available at all times, inated from more than one female. In some other studies, or the effect of food when larvae are raised singly without 160 Metamorphosis of Salamandra infraimmaculata A A B B C Figure 5. Age, mass, and length at metamorphosis in cohort I (A: single, B: low density, C: high density). See Appendix 4. any potential density effect (Brodman 1999, Alcoben- das et al. 2004, Warburg 2009b, 2010). In the last men- C tioned study it was concluded that metamorphosis appears to be affected by food resources when density could not have been a factor since these larvae were raised individu- Figure 6. Age, mass, and length at metamorphosis in cohort II (A = single, B = low density, C = high density). See Appendix 4. ally. Table 3 summarizes density conditions in six different studies of larvae belonging to four different urodelan spe- cies that were raised on an unlimited food supply. The con- cult to compare them. Thus, the high density of 42.7 cm2 ditions vary greatly between different experiments. Thus, in A. texanum larvae (Petranka 1984), is close to a mod- low-density conditions ranged from 16.5 cm2 per larva of erate density of 43.7 cm2 in S. salamandra larvae (Dega- Hynobius retardatus (Dunn, 1923) (see Wakahara 1995) ni 1993). Raising larvae of A. maculatum at high densities to 502.7 cm2 per larva of Ambystoma tigrinum nebulosum significantly affected the number of metamorphs (Walls Hallowell, 1853 (Collins & Cheek 1983): a 30.5-fold dif- 1998). Generally, urodelan larvae raised under high-den- ference. Likewise, high-density conditions ranged from sity conditions took longer to metamorphose and did so 7.5 cm2/larva of Salamandra salamandra (Linnaeus, 1758) at a smaller size. It was shown that when A. opacum larvae (Degani 1993) to 42.7 cm2/larva of Ambystoma texa num were raised under high–density conditions, which results (Matthes, 1855) (petranka 1984): a 5.7-fold difference. in smaller post-metamorphs, growth and first reproduc- In the present study, S. infraimmaculata larvae were raised tion were delayed by three years (Taylor & Scott 1997). at a low density of 28.7 cm2/larva or a high density of 14.3 In the present study, it was shown that although neither cm2/larva. Since experimental conditions vary so much competition nor sibling-predation (i.e., cannibalism) were between different studies on the significance of density possible (as potential cannibal larvae were removed once in urodelan larval growth and metamorphosis, it is diffi- identified as such), and food was constantly available, den- 161 Michael R. Warburg Table 3. Density (cm2 per larva) data from the literature for larvae of different salamander species raised on an unlimited food supply (ad libitum). Total numbers of larvae are given in parentheses. In order of appearance data were taken from: Degani (1993), Reques & Tejedo (1996), Cohen et al. (2006), Collins & Cheek (1983), Petranka (1984), Wakahara (1995). Species low density medium density high density food type Salamandra salamandra 75 (20) 43.7–75 (20) 7.5 (20) aquatic invertebrates Salamandra salamandra 300 (2) pond zooplankton Salamandra salamandra 28.7 (4–5) 14.3 (9–10) Tubifex and minced liver Ambystoma tigrinum nebulosum 502.7 (3) 215.4 (7) Artemia nauplii Ambystoma texanum 342 (1) 85.5 (4) 42.7 (8) aquatic invertebrates Hynobius retardatus 16.5–33 (5) 13.2–16.5 (20) 8.25–11 (40) urodelan larvae sity affected both age and size at metamorphosis. This sup- situation existed in the experiment conducted by Eitam et ports hypothesis 2 of Petranka & Sih (1986), which states al. (2005). The true meaning of this term needs yet to be that larval growth was density-dependent because of ‘in- clarified. It is the behavioural aspects of density effects that terference competition’, a hypothesis that did not find sup- are largely unknown. What could be the exact nature of port in their own study on A. texanum larvae. A similar such interference? This term could comprise a number of A A B B C C Figure 7. Minimum and maximum (A) age, (B) mass, and Figure 8. Range between minimal and maximal (A) age, (B) mass, (C) length at metamorphosis in the two cohorts. and (C) length at metamorphosis in the two cohorts. 162 Metamorphosis of Salamandra infraimmaculata Table 4. Differences between low and high density (Δ%) in number, age, and mass at metamorphosis when larvae of different sala- mander species at low and high densities were raised and fed ad libitum. species number of larvae (Δ%) age (Δ%) mass (Δ%) source Ambystoma opacum 981 (81.7 %) 2.6 % 47.3 % Scott (1990) Ambystoma opacum 38.8 % 49.2 % Scott (1994) Salamandra infraimmaculata 5 (50 %) 4.4 % 4.9 % this study factors: (1) Since food (especially live food) is generally not In S. infraimmaculata, the differences between both age evenly distributed, it is possible that some larvae are quick- and mass when raised at low and high densities were 4.4 er than others in locating it. These will have an advantage and 4.9% respectively (Table 4). as they may keep others from feeding. (2) It is possible that Differential growth rates of larvae have previously been a sensory response (optical, olfactory or tactile) is involved demonstrated in this species (Cohen et al. 2006). These that is sufficiently deterrent to keep some larvae from feed- were due to different growth modes identified within a ing regularly. Thus, it is theoretically possible that in spite single half-sibling larval cohort and resulted in variable of the fact that food was constantly available, some larvae growth rates. Consequently, the variability in the age of fed less or less frequently than others. larvae at metamorphosis could range from two to four In A. opacum, density affected both larval period (i.e., months. Such wide variability could have catastrophic ef- age at metamorphosis), which varied from 2.6 to 38.8%, fects on the timing of metamorphosis and subsequent ju- and mass at metamorphosis, which differed by about 48% venile dispersal. In this harsh Mediterranean climate, it when raised at low and high densities (Scott 1990, 1994). makes all the difference whether larvae metamorphose during winter rather than spring, which can well be too hot and dry for the post-metamorphs to survive. What could be the cause of the definite effect of density (larvae raised in the company of other larvae) under seem- ingly food resource-independent conditions that was dem- onstrated here? On the face of it, food could not possibly be the cause of this effect since it was always freely avail- able, but perhaps the presence of other larvae already de- ters some from feeding ad libitum? This point is of great in- terest and definitely needs further investigations. The evo- lutionary significance of these findings could be in show- A ing the way how larvae are capable of spreading out the timing of metamorphosis (i.e., age at metamorphosis), and thereby spreading out post-metamorphic growth, which would result in spreading out the age at maturation. This would mean that the aquatic larval population were ca- pable of regulating the size of the adult terrestrial popula- tion. Keeping the larval population low in small ponds un- der unpredictable hydroperiod conditions would improve chances of survival in species living under unpredictable ambient conditions (Warburg 2007b). B References Alcobendas, M., D. Buckley & M. Tejedom (2004): Varibility in survival, growth and metamorphosis in the larval fire sala- mander (Salamandra salamandra): effects of larval birth size, sibship and environment. – Herpetologica, 60: 232–245. Brodman, R. (1996): Effects of intraguild interactions on fitness and microhabitat use of larval Ambystoma salamanders. – Co- C peia, 1996: 372–378. Brodman, R. (1999). Food and space dependent effects during Figure 9. Average of (A) age, (B) mass, and (C) length at meta- the interactions of two species of larval salamanders. – Journal morphosis in cohorts I and II. See Appendix 1. of Freshwater Ecology, 14: 431–437. 163 Michael R. Warburg Brunkow, P. E. & J. P. Collins (1996): Effects of individual; varia- Wakahara, M. (1995): Cannibalism and the resulting dimor- tion in size on growth and development of larval salamanders. phism in larvae of a salamander Hynobius retardatus, inhab- – Ecology, 77: 1483–1492. ited in Hokkaido, Japan. – Zoological Science, 12: 467–473. Chazal, A. C., J. D. Krenz & D. E. Scott (1994): Relationship Walls, S. C. (1998): Density dependence in a larval salamander. of larval density and heterozygosity to growth and survival of The effect of interference and food limitation. – Copeia, 1998: juvenile marbled salamanders (Ambystoma opacum). – Cana- 926–935. dian Journal of Zoology, 74: 1122–1129. Warburg, M. R. (1986a): A relic population of Salamandra sala­ Cohen, M., R. Flam, R. Sharon, H. Ifrach, D. Yeheskely- mandra on Mt. Carmel: A ten-year study. – Herpetologica, 42: Hayon & M. R. Warburg, M.R. (2005): The evolutionary sig- 174–179. nificance of intra-cohort cannibalism in larvae of a xeric-in- Warburg, M. R. (1986b): Observations on a relic population of habiting salamander: an inter-cohort comparison. – Current Salamandra salamandra on Mt. Carmel during eleven years. Herpetology, 24: 55–66. – pp. 389–394 in Roček, Z. (ed.): Studies in Herpetology. – Cohen, M., D. Yeheskeli-Hayon, M. R. Warburg, D. Da vid- Prague: Charles University Press. son, G. Halevi & R. Sharon (2006): Differential growth Warburg, M. R. (1992): Breeding patterns in a fringe population identified in salamander larvae half-sib cohorts: a survival of fire salamanders, Salamandra salamandra. – Herpetological strategy? – Development, Growth and Differentiation, 48: Journal, 2: 54–58. 537–548. Warburg, M. R. (2006): Breeding site tenacity during a quar- Collins, J. P. & J. E. Cheek (1983): Effect of food and density on ter of a century, in a rare xeric-inhabiting Salamandra within development of typical and cannibalistic salamander larvae in an isolated metapopulation. – Bulletin Societe Herpetologique Ambystoma tigrinum nebulosum. – American Zoologist, 23: France, 118: 1–18. 77–84. Warburg, M. R. (2007a): Longevity in Salamandra infraimmacu­ Degani, G. (1993): Cannibalism among Salamandra salamandra lata from Israel with a partial review on other salamanders. – (L.) larvae. – Israel Journal of Zoology, 39: 125–129. Salamandra, 43: 21–34. Doody, S. (1996): Larval growth rate of known age Ambystoma Warburg, M. R. (2007b): The phenology of a rare salamander opacum in Louisiana under natural conditions. – Journal of (Salamandra infraimmaculata) in a population breeding un- Herpetology, 30: 294–297. der unpredictable conditions: a 25 year study. – Acta Herpe- Eitam, A., L. Blaustein & M. Mangel (2005): Density and in- tologica, 2: 247–257. tercohort priority effects on larval Salamandra salamandra in Warburg, M. R. (2008): Changes in recapture rate of a rare sala- temporary pools. Oecologia, 146: 36–42. mander in an isolated metapopulation studied. – Russian Fauth, J. E., W. J. Resetarits & H. M. Wilbur (1990): Interac- Journal of Herpetology, 15: 11–18. tions between larval salamanders: a case of competitive equal- Warburg, M. R. (2009a): Long-term study on the reproductive ity. – Oikos, 58: 91–99. strategy of Salamandra infraimmaculata (Amphibia: Urodela: Kusano, T. (1981): Growth and survival rate of the larvae of Salamandridae) females in a single population breeding under Hynobius nebulosus tokyoensis Tago (Amphibia, Hynobiidae). precarious ambient conditions. – Herpetozoa, 22: 27–42. – Research in Population Ecology, 23: 360–378. Warburg, M. R. (2009b): Age and size at metamorphosis of half- Nishihara, A. (1996): Effects of density on growth of head size in sib Salamandra infraimmaculata larvae born in the laboratory larvae of the salamander Hynobius retardatus. – Copeia, 1996: and raised singly under three different food regimes. – Belgian 478–483. Journal of Zoology, 139: 156–165. Ohdachi, S. (1994): Growth, metamorphosis, and gape-limited Warburg, M. R. (2010): Studies on 74 half-sib larval cohorts cannibalism and predation on tadpoles of salamanders Hyno­ born to individually-identifiable Salamandra infraimmaculata bius retardatus. – Zoological Science., 11: 127–131. females belonging to a single population; a long-term study. – Petranka, J. W. (1984): Success of interpopulational variation in Journal of Zoology London, 280: 332–342. growth responses of larval salamanders. – Ecology, 65: 1857– Warburg, M. R. (2011): Long-term study on the variability in du- 1865. ration of larval period and timing of metamorphosis in a sala- Petranka, J. W. & A. Sih (1986): Environmental instability, com- mander: a way to regulate dispersal. – Revue suisse Zoologie, petition, and density-dependent growth and survivorship of a 118: 231–249. stream-dwelling salamander. – Ecology, 67: 729–736. Warburg, M. R., G. Degani & I. Warburg (1978/79): Ovo vivi- Reques, R. & M. Tejedo (1996): Intraspecific aggressive behav- parity in Salamandra salamandra (L.) (Amphibia, Urodela) iour in fire salamander larvae (Salamandra salamandra): The from northern Israel. – Vie Milieu, 28/29C: 247–257. effects of density and body size. – Herpetological Journal, 6: Wilbur, H. M. (1976): Density-dependent aspects of metamor- 15–19. phosis in Ambystoma and Rana sylvatica. – Ecology, 57: 1289– Scott, D. E. (1990): Effects of larval density in Ambystoma opa­ 1296. cum: an experiment in large-scale field enclosures. – Ecology, Wilbur, H. M. & J. P. Collins (1973): Ecological aspects of am- 71: 296–306. phibian metamorphosis. – Science, 182: 1305–1314. Scott, D. E. (1994): The effect of larval density on adult demo- Ziemba, R. E. & J. P. Collins (1999): Development size structure graphic traits in Ambystoma opacum. – Ecology, 75: 1383–1396. in tiger salamanders: the role of intraspecific interference. – Taylor, B. E. & D. E. Scott (1997): Effects of larval density de- Oecologia, 120: 524–529. pendence on population dynamics of Ambystoma opacum. – Herpetologica, 53: 132–145. 164 Metamorphosis of Salamandra infraimmaculata Appendix 1 Cohort I: low density Results of t-tests of data on cohort I versus those on cohort II. age mass length Age Age Mass Length R2 0.9779 0.9989 0.9996 single low density high density P 0.5879 0.0001 < 0.0001 T value 0.330005 0.231763 0 degree of freedom 20 20 20 P value 0.745007 0.819489 1 degree of freedom 19 17 30 Cohort I: high density Mass age mass length single low density high density Age Mass Length R2 0.9779 0.9982 0.9989 T value 0.981452 1.650592 1.889448 P 0.6382 0.0261 < 0.0001 P value 0.343027 0.111849 0.068216 degree of freedom 17 17 17 degree of freedom 14 24 31 Cohort II: single Length age mass length single low density high density R2 0.9963 0.095 2 T value -0.25577 1.543785 -0.87974 P 0.9993 < 0.0001 2 P value 0.801845 0.135725 0.385772 degree of freedom 0.9986 0.0198 2 degree of freedom 14 24 31 Cohort II: low density age mass length Appendix 2 R2 0.983 0.9984 1 Results of regression analysis of age under different density P < 0.0001 < 0.0001 0 regimes in relation to Figures 1 and 2. degree of freedom 16 16 16 Cohort I Cohort II: high density R2 P degree of freedom age mass length single versus age 0.9313 0.531 3 R2 0.9974 0.9984 0.9982 low density versus age 0.7767 0.2889 4 P 0.1882 0.0703 0.0085 high density versus age 0.757 0.806 3 degree of freedom 10 10 10 Cohort II R2 P degree of freedom Appendix 4 single versus age 0.8401 0.0598 9 Results of regression analysis of age under different density low density versus age 0.7281 0.7432 13 regimes in relation to Figures 5 and 6. high density versus age 0.4973 0.8015 14 Cohort I R2 P degree of freedom age versus mass 0.9941 0.5193 6 Appendix 3 age versus length 0.9981 0.1829 6 Results of regression analysis of age under different density mass versus length 1 0.06 6 regimes in relation to Figures 3 and 4. Cohort I: single Cohort II age mass length R2 P degree of freedom R2 0.9941 0.9981 0.9981 age versus mass 0.9974 0.1882 10 P 0.5193 0.1829 0.1829 age versus length 0.9984 0.0703 10 degree of freedom 6 6 6 mass versus length 1 0 10 165