Oecologia (2001) 128:360–367 DOI 10.1007/s004420100672
Louise Bardsley · Trevor J.C. Beebee
Non-behavioural interference competition between anuran larvae under semi-natural conditions Received: 27 June 2000 / Accepted: 12 February 2001 / Published online: 4 April 2001 © Springer-Verlag 2001
Abstract Two major types of indirect competition have been recognised in natural communities, notably scramble for resources and interference to gain advantage by various alternative mechanisms. Interference effects are however often difficult to demonstrate in the field and their significance in nature requires more extensive study. Larvae of the anurans Bufo bufo and Bufo calamita compete strongly both in the field and in the laboratory, with the former species generally superior to the latter. Under laboratory conditions an interference component to this competition is readily demonstrable so we carried out an experiment to determine whether interference effects between the larvae of these species are also detectable in natural ponds. Larvae were reared at natural densities in cages immersed in a sand dune pool. Survival and growth rates were measured under conditions of no interaction, partial interaction (permitting interference but not resource competition) and full interaction between the species. For partial interaction, larvae were separated by mesh such that water and faeces could pass between compartments but the animals could not. Numbers of Anurofeca [=Prototheca] richardsi, a mediator of interference competition between these anurans under laboratory conditions, food acquisition and food availability were also determined during the course of the experiment. Asymmetric interspecific competition, manifested as reduced survival and growth rates of B. calamita, occurred under conditions of both full and partial interactions between the species though overall competition strength was much greater under fully interacting compared with partially interacting conditions. Interference competition, probably mediated by A. richardsi, was implicated under the partially interacting conditions in which resource competition was prevented.
L. Bardsley · T.J.C. Beebee (✉) School of Biological Sciences, University of Sussex, Falmer, Brighton BN1 9QG, UK e-mail: [email protected] Tel.: +44-1273-606755, Fax: +44-1273-678433
Keywords Anuran larvae · Anurofeca richardsi · Interference competition
Introduction It is now widely recognised that the relative strengths of predation, competition and environmental stochasticity vary widely in different situations. Competition, in particular, is a complex process with often poorly understood mechanisms of operation. Resource and interference-based competition were recognised as distinct processes many years ago (Brian 1956) but even this straightforward distinction is frequently difficult to demonstrate. Experiments with many taxa including rotifers, crustaceans, insects and fish have successfully demonstrated varying levels of resource and interference competition (e.g. Anholt 1990; MacIsaac and Gilbert 1991; Rodriguez 1991; Holway 1999) but there remains a need for more empirical work in this field because underlying mechanisms of interference competition and their significance under natural conditions have rarely been elucidated. The study of competition between anuran larvae is well documented (e.g. Morin 1983; Wilbur 1987). Larval growth rate, timing of metamorphosis and size at metamorphosis are well-characterised and important fitness attributes (Wilbur and Collins 1973; Wilbur 1980). Multiple studies have shown that predation, competition and environmental stochasticity (pond desiccation time) can all be important and can interact to produce a variety of outcomes. Predation, for example, can relieve competition and improve the prospects of competitively inferior species (Morin 1983; Wilbur 1987). Similarly, pond chemistry in the form of pH differences can affect the outcome of competition between hylid frogs (Warner et al. 1993). However, there have been relatively few attempts to dissect the mechanisms of competition in any of these situations. Interference competition may result from behavioural interactions, such as aggression between tadpoles restricting access to food (e.g. Faragher
361
and Jaeger 1998) or from indirect effects such as the production of pathogens or growth inhibitors. Indirect interference competition between tadpoles is of particular interest because it has been detected in several studies over the past 40 years (e.g. Richards 1958, 1962; Licht 1967; Steinwascher 1979; Banks and Beebee 1987). In the laboratory a unicellular eukaryotic organism, Anurofeca (= Prototheca) richardsi (Baker et al. 1999), mediates interference competition between numerous species of anuran larvae (Beebee 1991). A. richardsi proliferates in larval guts, is passaged in faeces and apparently diverts small (competitively inferior) larvae into non-productive coprophagy. Anurofeca cells are indigestible and constitute a high proportion of faecal mass under these circumstances (Beebee and Wong 1992). Although all anuran larvae passage Anurofeca, their effects on growth are inversely proportional to tadpole size (Beebee 1991; Beebee and Wong 1992). Anurofeca-mediated interference competition has also been implicated in experimental replicated pond systems (e.g. Griffiths et al. 1993) and we were therefore interested to know whether this mechanism might also occur in wild anuran populations. Interference effects of this kind seem relatively rare in natural ponds (e.g. Petranka 1989; Werner 1992; Biesterfeldt et al. 1993), perhaps because biotic factors such as competitor microorganisms or predators usually limit the abundance of Anurofeca to low levels (Wong et al. 1994; Baker and Beebee 1997). The two toads Bufo bufo and Bufo calamita are normally allotopic in Britain but in some areas B. bufo is replacing B. calamita as scrub develops and favours syntopy (Denton et al. 1997). This is usually a temporary situation, followed by local extinction of B. calamita if B. bufo numbers become large. Larvae of B. bufo are competitively superior to those of B. calamita because B. bufo breeds earliest in spring, and their larvae are generally abundant by the time B. calamita spawns. Priority effects based on different breeding season times are not uncommon in anurans (e.g. Alford and Wilbur 1985; Wilbur and Alford 1985; Morin et al. 1990; Lawler and Morin 1993). We have already shown that competition between B. bufo and B. calamita larvae can be intense in natural pools (Bardsley and Beebee 1998a), with B. calamita experiencing reduced growth rates and survival in the presence of B. bufo. We therefore set out to test the hypothesis that this competition includes an interference component. In a sand dune pond, a habitat used by both species, we measured the effects on B. calamita larvae of both full and partial interaction with larvae of B. bufo. Partial interaction permitted only interference competition mediated by A. richardsi and excluded resource competition as well as behavioural interference.
Materials and methods Study site The pond initially chosen for this study was a natural dune “slack” on the Ainsdale Sand Dunes National Nature Reserve in northwest England. It had a perimeter of about 700 m at the start of the
experiment, with a shallow shelving profile and a maximum depth of 30 cm. However, rapid desiccation part-way through the experiment (day 40) necessitated moving the trials to a nearby, slightly deeper dune slack with a perimeter of 500 m shelving to 60 cm. This was accomplished within a few hours and without any loss of animals. Variance between dune slacks with respect to environmental conditions is generally small and all are shallow, open to the sun, poorly vegetated and with sandy substrates. Previous studies using multiple ponds in the same dune area demonstrated that patterns of anuran larval development were consistent to the extent that pond effects were negligible (Bardsley and Beebee 1998a, b). Amphibian larvae from natural spawnings were almost completely removed by netting from both ponds (leaving less than 0.1 per litre) prior to running the experiment. Animal collection Eggs were obtained from multiple spawn strings of B. bufo and B. calamita (>5 of each) laid in other ponds within the same sand dune system and hatched in holding tanks in the field. The eggs of each species were laid within a few hours of each other but there was a gap of about 4 weeks between the spawnings of the two species, and B. bufo spawn for the experiment was therefore collected 4 weeks earlier than that of B. calamita. Batches of larvae were assigned randomly to cages soon after they became free-swimming [Gosner (1960) stage 27 for B. bufo and stage 25 for B. calamita], at which point B. calamita larvae are safe from direct predation by B. bufo. Average sizes of larvae of each species were initially equal in all treatments at around 9 mm for B. bufo and 8 mm for B. calamita. Larval development periods in natural ponds are typically 70–80 days for B. bufo and 30–60 days for B. calamita, depending on weather conditions. Treatment cages and experimental design A set of collapsible 80×30 cm rectangular cages was constructed using wooden frames covered around all four sides with green nylon mesh (Netlon, UK) of 2.0 mm pore diameter. The cage bottoms were open and 10-cm spikes at each corner together with 3 cm of mesh all around the cage bases were driven into the sediment. Cage height was 50 cm, with the top well above water level, and each contained approximately 50 l of pond water at the start of the experiment (i.e. with an average water depth of just over 20 cm, but with a gradient from 10 to 30 cm). Larvae therefore had access to sediment and some choice of water depth but could not escape or enter the cages. Cages also had tight-fitting lids made of wooden frames with white nylon mesh of 0.25 mm pore size. All the cages were netted exhaustively to remove predators before adding anuran larvae and any predators subsequently found in them were removed immediately. Cages were arranged randomly, but at similar depths, and were installed about 4 weeks before the start of experiment. Water levels within the cages dropped naturally during the study. At day 40 the cages were moved because of imminent pond desiccation, when mean water levels were less than 7 cm. In the new pond, less than 1 km distant, they were installed with a mean depth of 20 cm as at the start of the experiment. We had four treatments, each in triplicate (Table 1), to compare intraspecific competition at two different densities of B. calamita with interspecific competition between B. bufo and B. calamita under fully interacting or partially interacting conditions. There were also cages containing B. bufo for the partial interactions (see below) and others without any anuran larvae as controls to monitor food availability in the absence of larval grazing. Treatments were allocated randomly to the cages. We added 50 B. bufo larvae to each of the appropriate cages (Table 1) on 8 April (day 1), and 50 or 100 B. calamita to each appropriate cage (Table 1) on May 13 (day 35). By this time the mean size of B. bufo larvae was 18.2 mm, with no significant differences between cages. We decided the initial larval densities on the basis of field studies (Bardsley and Beebee 1998a) and they fell within the range of
362 Table 1 Experimental treatments
Treatment
No. Bufo calamita added to cage
No. Bufo bufo added to cage
Control Intraspecific competition Interspecific competition (full interaction) Interference competition (partial interaction) Control and source of B. bufo faeces Resource availability control
50 100 50 50 – –
– – 50 Faeces only 50 –
those found most frequently in the dune ponds. No food was added to the cages. For partially interacting conditions we placed all the B. bufo larvae from the faecal source treatment cages in 75×25 cm mesh trays (0.5 mm nylon), 10 cm deep, and suspended them inside the main cages of the partial interaction treatment, above the B. calamita larvae but physically separated from them. No food was added. Larvae from one faecal source cage were suspended over any one partial interaction treatment cage, and the trays temporarily replaced the lids normally used to cover the cage tops. Faeces from the B. bufo larvae passed through the tray floors for the period of suspension (4–6 h, repeated twice each week starting on day 7), after which the B. bufo larvae were returned to the faecal source treatment cages. This period (about 6% of the total time) was chosen following laboratory studies indicating that it should be sufficient to elicit a growth inhibitory response in the B. calamita larvae if Anurofeca production was comparable in the pond situation (Beebee 1991). The process of placing larvae in suspended cages triggers defecation (Beebee 1991) and thus generates an effect greater than would be expected simply from the proportion of time spent there. Interspecific resource competition for suspended algae was minimised by the short partial interaction period but in any case should be low because B. calamita is primarily a benthic feeder (Bardsley and Beebee 1998a). Day 42 was used as the 100% reference point to avoid any effects from moving the cages to the new pond on day 40, although mortality between day 35 and day 42 was in all cases less than 10% of the starting numbers. We terminated the experiment on 1 July (day 84) when the first B. calamita were at Gosner stage 42, immediately prior to metamorphosis but before detectable tail shortening occurred. Metamorphosis of B. bufo started around 6 June 6 (day 63) and almost all had emerged by day 84. Toadlets of B. calamita drown very rapidly if prevented from leaving the water, as might happen in cages, so no attempt was made to collect data on metamorphs. Measured responses Every cage was inspected once per week from experimental day 1 onwards and numbers of surviving tadpoles of each species were determined by exhaustive netting. The species could be identified directly due to the large differences in size and the development of white chin-patches in B. calamita larvae (Beebee 1983). Every larva was measured (snout-tail tip) to the nearest 0.5 mm on every weekly visit. Anurofeca production was measured as a response to treatment conditions. Up to 20 larvae of each species, depending on how many still survived, were removed from each treatment cage and placed in mesh cages (separate for each species) immersed in water within plastic boxes. Very small amounts of food (boiled lettuce) were added to stimulate defecation, the larvae were left for 2 h and then returned to their respective treatment cages. Faeces were collected after pouring off excess water from the plastic box, preserved in 50 ml of 10 µg/ml CuSO4 and later centrifuged and resuspended in 10 ml sterile distilled water. A. richardsi cells were then counted as described elsewhere (Bardsley and Beebee 1998a). Average numbers of cells from three 50 µl aliquots per sample were used as data points. Mean numbers of A. richardsi excreted per tadpole per hour in each treatment were calculated
and total numbers produced per hour by all the larvae in each treatment cage were estimated by extrapolation. Sample collection started on 8 April and occurred every 2 weeks until 24 June (day 77). Algal cells in sediment and periphyton constitute the main food of these bufonid larvae in dune ponds (Bardsley and Beebee 1998a, b). Arbitrary 1 cm2 of bottom sediment (to a depth of ≤1 mm, by pipette) and 4 cm2 of a scrape made along the side of the cage, areas continually browsed by larvae, were carefully removed and pooled into 50 ml water taken from within the cage. Samples of this kind were taken from every cage on 8 April and then once every 2 weeks until 17 June (day 70), always at comparable locations within each cage. All algal cells present were preserved by addition of CuSO4 to a final concentration of 10 µg/ml. Each suspension was later centrifuged at 600×g for 2 min; most of the supernatant was discarded and the pellet of cells resuspended in 10 ml of retained water. Duplicate 50 µl aliquots of each sample were subsequently examined on a haemocytometer slide. Total numbers of all cells combined were recorded as an indicator of food availability. For two sampling times (day 28 and day 70) all microorganisms present were identified to the most practicable taxonomic level. More than 95% fell within five distinctive groups of eukaryotic algae: three categories of Chlorophyta were scored (Desmidales, Zygnematales and other Chlorophyta), the Bacillariophyta (diatoms), and pooled “other algae”. For all analyses an average of the two 50 µl samples was used as the datum point for each cage. Analysis of larval gut contents was carried out on a single occasion (27 May, day 49). This date was chosen because it corresponds with the growth period in which competitive effects are likely to be most severe (Beebee 1991). Three B. calamita and/or three B. bufo were removed from each cage for this purpose and subsequent survivorship calculations took account of these abstractions. Tadpoles were preserved in ethanol and gut contents were subsequently identified and counted as described elsewhere (Bardsley and Beebee 1998a) as indicators of food acquisition. Briefly, gut contents were suspended in 2 ml distilled water and duplicate 50 µl aliquots examined on a haemocytometer slide. Data were corrected for mean larval size in each treatment (i.e. represent cell numbers per centimetre of tadpole) after establishing that there was a strong positive correlation between cell count and tadpole size in all treatments. Data analysis All data were tested for normality of residuals and transformed where necessary before analysis by STATISTIX or SPSS statistical packages. In all cases, means of samples within cages were used as single datum points. Survival, larval sizes, numbers of cells in the gut and food availability (cell counts in cage samples) were compared between treatments using ANOVAs or, in one case, repeated-measures ANOVAs together with Tukey tests to determine which groups were significantly different (Fowler and Cohen 1990). All proportion data were arcsine transformed and compared using MANOVAs. Rank sum and Wilcoxon sign rank tests were employed to compare algal floras between ponds. Pearson correlations were carried out to ascertain the strengths of relationships between survival, growth rate, food availability, food acquisition and A. richardsi abundance. Multiple linear re-
363 gression was then applied to investigate relationships between these variables. Feeding niche overlap was estimated using index C (Hurlbert 1978). Unfortunately the complete batch of preserved samples (Anurofeca, gut contents and food availability) from treatment 1 was lost and therefore not available for analysis.
Results Survival and growth of anuran larvae Survival and growth rates of B. calamita larvae during the course of the experiment varied between treatments (Fig. 1). Survival on day 84 ranged between 28% and 67% and varied significantly as a function of treatment (ANOVA F=24.89, df=3,8, P=0.0002). Final survival of B. calamita larvae at high density (100) was on average about 85% of that at low density (50) but differences between these conditions were not significant by Tukey test at P≤0.05. However, final survival in the presence of 50 B. bufo (fully interacting conditions) was
Louise Bardsley · Trevor J.C. Beebee
Non-behavioural interference competition between anuran larvae under semi-natural conditions Received: 27 June 2000 / Accepted: 12 February 2001 / Published online: 4 April 2001 © Springer-Verlag 2001
Abstract Two major types of indirect competition have been recognised in natural communities, notably scramble for resources and interference to gain advantage by various alternative mechanisms. Interference effects are however often difficult to demonstrate in the field and their significance in nature requires more extensive study. Larvae of the anurans Bufo bufo and Bufo calamita compete strongly both in the field and in the laboratory, with the former species generally superior to the latter. Under laboratory conditions an interference component to this competition is readily demonstrable so we carried out an experiment to determine whether interference effects between the larvae of these species are also detectable in natural ponds. Larvae were reared at natural densities in cages immersed in a sand dune pool. Survival and growth rates were measured under conditions of no interaction, partial interaction (permitting interference but not resource competition) and full interaction between the species. For partial interaction, larvae were separated by mesh such that water and faeces could pass between compartments but the animals could not. Numbers of Anurofeca [=Prototheca] richardsi, a mediator of interference competition between these anurans under laboratory conditions, food acquisition and food availability were also determined during the course of the experiment. Asymmetric interspecific competition, manifested as reduced survival and growth rates of B. calamita, occurred under conditions of both full and partial interactions between the species though overall competition strength was much greater under fully interacting compared with partially interacting conditions. Interference competition, probably mediated by A. richardsi, was implicated under the partially interacting conditions in which resource competition was prevented.
L. Bardsley · T.J.C. Beebee (✉) School of Biological Sciences, University of Sussex, Falmer, Brighton BN1 9QG, UK e-mail: [email protected] Tel.: +44-1273-606755, Fax: +44-1273-678433
Keywords Anuran larvae · Anurofeca richardsi · Interference competition
Introduction It is now widely recognised that the relative strengths of predation, competition and environmental stochasticity vary widely in different situations. Competition, in particular, is a complex process with often poorly understood mechanisms of operation. Resource and interference-based competition were recognised as distinct processes many years ago (Brian 1956) but even this straightforward distinction is frequently difficult to demonstrate. Experiments with many taxa including rotifers, crustaceans, insects and fish have successfully demonstrated varying levels of resource and interference competition (e.g. Anholt 1990; MacIsaac and Gilbert 1991; Rodriguez 1991; Holway 1999) but there remains a need for more empirical work in this field because underlying mechanisms of interference competition and their significance under natural conditions have rarely been elucidated. The study of competition between anuran larvae is well documented (e.g. Morin 1983; Wilbur 1987). Larval growth rate, timing of metamorphosis and size at metamorphosis are well-characterised and important fitness attributes (Wilbur and Collins 1973; Wilbur 1980). Multiple studies have shown that predation, competition and environmental stochasticity (pond desiccation time) can all be important and can interact to produce a variety of outcomes. Predation, for example, can relieve competition and improve the prospects of competitively inferior species (Morin 1983; Wilbur 1987). Similarly, pond chemistry in the form of pH differences can affect the outcome of competition between hylid frogs (Warner et al. 1993). However, there have been relatively few attempts to dissect the mechanisms of competition in any of these situations. Interference competition may result from behavioural interactions, such as aggression between tadpoles restricting access to food (e.g. Faragher
361
and Jaeger 1998) or from indirect effects such as the production of pathogens or growth inhibitors. Indirect interference competition between tadpoles is of particular interest because it has been detected in several studies over the past 40 years (e.g. Richards 1958, 1962; Licht 1967; Steinwascher 1979; Banks and Beebee 1987). In the laboratory a unicellular eukaryotic organism, Anurofeca (= Prototheca) richardsi (Baker et al. 1999), mediates interference competition between numerous species of anuran larvae (Beebee 1991). A. richardsi proliferates in larval guts, is passaged in faeces and apparently diverts small (competitively inferior) larvae into non-productive coprophagy. Anurofeca cells are indigestible and constitute a high proportion of faecal mass under these circumstances (Beebee and Wong 1992). Although all anuran larvae passage Anurofeca, their effects on growth are inversely proportional to tadpole size (Beebee 1991; Beebee and Wong 1992). Anurofeca-mediated interference competition has also been implicated in experimental replicated pond systems (e.g. Griffiths et al. 1993) and we were therefore interested to know whether this mechanism might also occur in wild anuran populations. Interference effects of this kind seem relatively rare in natural ponds (e.g. Petranka 1989; Werner 1992; Biesterfeldt et al. 1993), perhaps because biotic factors such as competitor microorganisms or predators usually limit the abundance of Anurofeca to low levels (Wong et al. 1994; Baker and Beebee 1997). The two toads Bufo bufo and Bufo calamita are normally allotopic in Britain but in some areas B. bufo is replacing B. calamita as scrub develops and favours syntopy (Denton et al. 1997). This is usually a temporary situation, followed by local extinction of B. calamita if B. bufo numbers become large. Larvae of B. bufo are competitively superior to those of B. calamita because B. bufo breeds earliest in spring, and their larvae are generally abundant by the time B. calamita spawns. Priority effects based on different breeding season times are not uncommon in anurans (e.g. Alford and Wilbur 1985; Wilbur and Alford 1985; Morin et al. 1990; Lawler and Morin 1993). We have already shown that competition between B. bufo and B. calamita larvae can be intense in natural pools (Bardsley and Beebee 1998a), with B. calamita experiencing reduced growth rates and survival in the presence of B. bufo. We therefore set out to test the hypothesis that this competition includes an interference component. In a sand dune pond, a habitat used by both species, we measured the effects on B. calamita larvae of both full and partial interaction with larvae of B. bufo. Partial interaction permitted only interference competition mediated by A. richardsi and excluded resource competition as well as behavioural interference.
Materials and methods Study site The pond initially chosen for this study was a natural dune “slack” on the Ainsdale Sand Dunes National Nature Reserve in northwest England. It had a perimeter of about 700 m at the start of the
experiment, with a shallow shelving profile and a maximum depth of 30 cm. However, rapid desiccation part-way through the experiment (day 40) necessitated moving the trials to a nearby, slightly deeper dune slack with a perimeter of 500 m shelving to 60 cm. This was accomplished within a few hours and without any loss of animals. Variance between dune slacks with respect to environmental conditions is generally small and all are shallow, open to the sun, poorly vegetated and with sandy substrates. Previous studies using multiple ponds in the same dune area demonstrated that patterns of anuran larval development were consistent to the extent that pond effects were negligible (Bardsley and Beebee 1998a, b). Amphibian larvae from natural spawnings were almost completely removed by netting from both ponds (leaving less than 0.1 per litre) prior to running the experiment. Animal collection Eggs were obtained from multiple spawn strings of B. bufo and B. calamita (>5 of each) laid in other ponds within the same sand dune system and hatched in holding tanks in the field. The eggs of each species were laid within a few hours of each other but there was a gap of about 4 weeks between the spawnings of the two species, and B. bufo spawn for the experiment was therefore collected 4 weeks earlier than that of B. calamita. Batches of larvae were assigned randomly to cages soon after they became free-swimming [Gosner (1960) stage 27 for B. bufo and stage 25 for B. calamita], at which point B. calamita larvae are safe from direct predation by B. bufo. Average sizes of larvae of each species were initially equal in all treatments at around 9 mm for B. bufo and 8 mm for B. calamita. Larval development periods in natural ponds are typically 70–80 days for B. bufo and 30–60 days for B. calamita, depending on weather conditions. Treatment cages and experimental design A set of collapsible 80×30 cm rectangular cages was constructed using wooden frames covered around all four sides with green nylon mesh (Netlon, UK) of 2.0 mm pore diameter. The cage bottoms were open and 10-cm spikes at each corner together with 3 cm of mesh all around the cage bases were driven into the sediment. Cage height was 50 cm, with the top well above water level, and each contained approximately 50 l of pond water at the start of the experiment (i.e. with an average water depth of just over 20 cm, but with a gradient from 10 to 30 cm). Larvae therefore had access to sediment and some choice of water depth but could not escape or enter the cages. Cages also had tight-fitting lids made of wooden frames with white nylon mesh of 0.25 mm pore size. All the cages were netted exhaustively to remove predators before adding anuran larvae and any predators subsequently found in them were removed immediately. Cages were arranged randomly, but at similar depths, and were installed about 4 weeks before the start of experiment. Water levels within the cages dropped naturally during the study. At day 40 the cages were moved because of imminent pond desiccation, when mean water levels were less than 7 cm. In the new pond, less than 1 km distant, they were installed with a mean depth of 20 cm as at the start of the experiment. We had four treatments, each in triplicate (Table 1), to compare intraspecific competition at two different densities of B. calamita with interspecific competition between B. bufo and B. calamita under fully interacting or partially interacting conditions. There were also cages containing B. bufo for the partial interactions (see below) and others without any anuran larvae as controls to monitor food availability in the absence of larval grazing. Treatments were allocated randomly to the cages. We added 50 B. bufo larvae to each of the appropriate cages (Table 1) on 8 April (day 1), and 50 or 100 B. calamita to each appropriate cage (Table 1) on May 13 (day 35). By this time the mean size of B. bufo larvae was 18.2 mm, with no significant differences between cages. We decided the initial larval densities on the basis of field studies (Bardsley and Beebee 1998a) and they fell within the range of
362 Table 1 Experimental treatments
Treatment
No. Bufo calamita added to cage
No. Bufo bufo added to cage
Control Intraspecific competition Interspecific competition (full interaction) Interference competition (partial interaction) Control and source of B. bufo faeces Resource availability control
50 100 50 50 – –
– – 50 Faeces only 50 –
those found most frequently in the dune ponds. No food was added to the cages. For partially interacting conditions we placed all the B. bufo larvae from the faecal source treatment cages in 75×25 cm mesh trays (0.5 mm nylon), 10 cm deep, and suspended them inside the main cages of the partial interaction treatment, above the B. calamita larvae but physically separated from them. No food was added. Larvae from one faecal source cage were suspended over any one partial interaction treatment cage, and the trays temporarily replaced the lids normally used to cover the cage tops. Faeces from the B. bufo larvae passed through the tray floors for the period of suspension (4–6 h, repeated twice each week starting on day 7), after which the B. bufo larvae were returned to the faecal source treatment cages. This period (about 6% of the total time) was chosen following laboratory studies indicating that it should be sufficient to elicit a growth inhibitory response in the B. calamita larvae if Anurofeca production was comparable in the pond situation (Beebee 1991). The process of placing larvae in suspended cages triggers defecation (Beebee 1991) and thus generates an effect greater than would be expected simply from the proportion of time spent there. Interspecific resource competition for suspended algae was minimised by the short partial interaction period but in any case should be low because B. calamita is primarily a benthic feeder (Bardsley and Beebee 1998a). Day 42 was used as the 100% reference point to avoid any effects from moving the cages to the new pond on day 40, although mortality between day 35 and day 42 was in all cases less than 10% of the starting numbers. We terminated the experiment on 1 July (day 84) when the first B. calamita were at Gosner stage 42, immediately prior to metamorphosis but before detectable tail shortening occurred. Metamorphosis of B. bufo started around 6 June 6 (day 63) and almost all had emerged by day 84. Toadlets of B. calamita drown very rapidly if prevented from leaving the water, as might happen in cages, so no attempt was made to collect data on metamorphs. Measured responses Every cage was inspected once per week from experimental day 1 onwards and numbers of surviving tadpoles of each species were determined by exhaustive netting. The species could be identified directly due to the large differences in size and the development of white chin-patches in B. calamita larvae (Beebee 1983). Every larva was measured (snout-tail tip) to the nearest 0.5 mm on every weekly visit. Anurofeca production was measured as a response to treatment conditions. Up to 20 larvae of each species, depending on how many still survived, were removed from each treatment cage and placed in mesh cages (separate for each species) immersed in water within plastic boxes. Very small amounts of food (boiled lettuce) were added to stimulate defecation, the larvae were left for 2 h and then returned to their respective treatment cages. Faeces were collected after pouring off excess water from the plastic box, preserved in 50 ml of 10 µg/ml CuSO4 and later centrifuged and resuspended in 10 ml sterile distilled water. A. richardsi cells were then counted as described elsewhere (Bardsley and Beebee 1998a). Average numbers of cells from three 50 µl aliquots per sample were used as data points. Mean numbers of A. richardsi excreted per tadpole per hour in each treatment were calculated
and total numbers produced per hour by all the larvae in each treatment cage were estimated by extrapolation. Sample collection started on 8 April and occurred every 2 weeks until 24 June (day 77). Algal cells in sediment and periphyton constitute the main food of these bufonid larvae in dune ponds (Bardsley and Beebee 1998a, b). Arbitrary 1 cm2 of bottom sediment (to a depth of ≤1 mm, by pipette) and 4 cm2 of a scrape made along the side of the cage, areas continually browsed by larvae, were carefully removed and pooled into 50 ml water taken from within the cage. Samples of this kind were taken from every cage on 8 April and then once every 2 weeks until 17 June (day 70), always at comparable locations within each cage. All algal cells present were preserved by addition of CuSO4 to a final concentration of 10 µg/ml. Each suspension was later centrifuged at 600×g for 2 min; most of the supernatant was discarded and the pellet of cells resuspended in 10 ml of retained water. Duplicate 50 µl aliquots of each sample were subsequently examined on a haemocytometer slide. Total numbers of all cells combined were recorded as an indicator of food availability. For two sampling times (day 28 and day 70) all microorganisms present were identified to the most practicable taxonomic level. More than 95% fell within five distinctive groups of eukaryotic algae: three categories of Chlorophyta were scored (Desmidales, Zygnematales and other Chlorophyta), the Bacillariophyta (diatoms), and pooled “other algae”. For all analyses an average of the two 50 µl samples was used as the datum point for each cage. Analysis of larval gut contents was carried out on a single occasion (27 May, day 49). This date was chosen because it corresponds with the growth period in which competitive effects are likely to be most severe (Beebee 1991). Three B. calamita and/or three B. bufo were removed from each cage for this purpose and subsequent survivorship calculations took account of these abstractions. Tadpoles were preserved in ethanol and gut contents were subsequently identified and counted as described elsewhere (Bardsley and Beebee 1998a) as indicators of food acquisition. Briefly, gut contents were suspended in 2 ml distilled water and duplicate 50 µl aliquots examined on a haemocytometer slide. Data were corrected for mean larval size in each treatment (i.e. represent cell numbers per centimetre of tadpole) after establishing that there was a strong positive correlation between cell count and tadpole size in all treatments. Data analysis All data were tested for normality of residuals and transformed where necessary before analysis by STATISTIX or SPSS statistical packages. In all cases, means of samples within cages were used as single datum points. Survival, larval sizes, numbers of cells in the gut and food availability (cell counts in cage samples) were compared between treatments using ANOVAs or, in one case, repeated-measures ANOVAs together with Tukey tests to determine which groups were significantly different (Fowler and Cohen 1990). All proportion data were arcsine transformed and compared using MANOVAs. Rank sum and Wilcoxon sign rank tests were employed to compare algal floras between ponds. Pearson correlations were carried out to ascertain the strengths of relationships between survival, growth rate, food availability, food acquisition and A. richardsi abundance. Multiple linear re-
363 gression was then applied to investigate relationships between these variables. Feeding niche overlap was estimated using index C (Hurlbert 1978). Unfortunately the complete batch of preserved samples (Anurofeca, gut contents and food availability) from treatment 1 was lost and therefore not available for analysis.
Results Survival and growth of anuran larvae Survival and growth rates of B. calamita larvae during the course of the experiment varied between treatments (Fig. 1). Survival on day 84 ranged between 28% and 67% and varied significantly as a function of treatment (ANOVA F=24.89, df=3,8, P=0.0002). Final survival of B. calamita larvae at high density (100) was on average about 85% of that at low density (50) but differences between these conditions were not significant by Tukey test at P≤0.05. However, final survival in the presence of 50 B. bufo (fully interacting conditions) was
