Anurans contain a diversity of defensive chemicals in their skin secretions, which function in protection against pathogens, parasites and/or predators (Conlon Reference CONLON2011, Mina et al. Reference MINA, PONTI, WOODCRAFT, JOHNSON and SAPORITO2015). These defensive chemicals include amines, peptides, proteins and steroids, which are manufactured by anurans, as well as lipophilic alkaloids, which are obtained from dietary arthropods (reviewed in Saporito et al. Reference SAPORITO, SPANDE, GARRAFFO and DONNELLY2009). Although several studies have examined the defensive function of particular chemicals among anuran species (Formanowicz & Brodie Reference FORMANOWICZ and BRODIE1982, Fritz et al. Reference FRITZ, STANLEY and DEPAMPHILIS1981, Szelistowski Reference SZELISTOWSKI1985, Weldon et al. Reference WELDON, KRAMER, GORDON, SPANDE and DALY2006), very little research has compared the effectiveness of different defensive chemicals between sympatric species exposed to the same predators. Given the diversity of defensive chemicals present in anuran skin, it is possible that some frog species are more or less protected from certain types of predators than others. Understanding how different defensive chemicals function against similar predators will provide important insight into the ecology and evolution of chemical defences in frogs. The aim of our study was to gain an understanding of how bufadienolide and alkaloid-based defences protect anurans from similar predators in their natural habitats.

Bufonid toads (Bufonidae) are typically characterized by their cryptic colouration and behaviour, which provides camouflage in their leaf-litter habitats (Heinen Reference HEINEN1985). Most bufonids produce a variety of steroids (e.g. bufadienolides; Erspamer Reference ERSPAMER, Heatwole and Barthalmus1994) that are stored in a few localized parotoid glands on their dorsum (Hostetler & Cannon Reference HOSTETLER and CANNON1974), which defend these toads and their tadpoles from predation by both invertebrates and vertebrates (Brodie et al. Reference BRODIE, FORMANOWICZ and BRODIE1978, Formanowicz & Brodie Reference FORMANOWICZ and BRODIE1982, Shine Reference SHINE2010, Toledo et al. Reference TOLEDO, RIBEIRO and HADDAD2007). Unlike bufonids, certain dendrobatid frogs (Dendrobatidae) possess alkaloid-based chemical defences, which are acquired entirely from a diet mainly consisting of mites and ants (reviewed in Saporito et al. Reference SAPORITO, SPANDE, GARRAFFO and DONNELLY2009, Reference SAPORITO, DONNELLY, SPANDE and GARRAFFO2012). Alkaloids are stored in thousands of poison (granular) glands that are distributed throughout the dorsum of these frogs (Saporito et al. Reference SAPORITO, ISOLA, MACCACHERO, CONDON and DONNELLY2010). Alkaloid defences are unpalatable, and in some cases toxic, to a variety of invertebrate and vertebrate predators (Brodie & Tumbarello Reference BRODIE and TUMBARELLO1978, Fritz et al. Reference FRITZ, STANLEY and DEPAMPHILIS1981, Murray et al. Reference MURRAY, BOLTON, BERG and SAPORITO2016, Szelistowski Reference SZELISTOWSKI1985). Chemically defended dendrobatids have conspicuous colours, which serve as warning (aposematic) signals to colour-visioned predators such as birds (Maan & Cummings Reference MAAN and CUMMINGS2012, Paluh et al. Reference PALUH, HANTAK and SAPORITO2014, Summers & Clough Reference SUMMERS and CLOUGH2001; however, see Alvarado et al. Reference ALVARADO, ALVAREZ and SAPORITO2013).

Bufonids and dendrobatids are largely sympatric in the neotropics, and many species share microhabitats, and very likely, the same predators; however, it is not currently known how bufadienolide and alkaloid defences compare against the same predators. Many arthropods utilize chemoreception as a dominant sense by which to detect their prey, and therefore represent an important group of predators to study differences in the effectiveness of chemical defences. Experimental studies with the dendrobatid frog Oophaga pumilio in Costa Rica have demonstrated that adult frogs are protected from predation by the predatory bullet ant, Paraponera clavata, and the ctenid spider, Cupiennius coccineus (Fritz et al. Reference FRITZ, STANLEY and DEPAMPHILIS1981, Szelistowski Reference SZELISTOWSKI1985). Murray et al. (Reference MURRAY, BOLTON, BERG and SAPORITO2016) recently demonstrated that lower quantities of alkaloid defences in juvenile O. pumilio lead to increased predation by the bullet ant P. clavata, suggesting that these predators can detect differences in chemical defences. A study by Gray et al. (Reference GRAY, KAISER and GREEN2010) demonstrated that alkaloid defences in adult Dendrobates auratus in Panama were effective against the red rump tarantula, Sericopelma rubronitens. Many studies have demonstrated that the bufonid toad, Rhinella marina, is protected from predators (reviewed in Shine Reference SHINE2010); however, little is known about the predator defence in other toad species. The common litter toad, Rhaebo haematiticus (Bufonidae), is sympatric with both O. pumilio and D. auratus in Central and South America (Guyer & Donnelly Reference GUYER and DONNELLY2005) and is likely exposed to the same arthropod predators. As a bufonid, R. haematiticus contains bufadienolide-based defences (Ferreira et al. Reference FERREIRA, LIMA, DEBIASI, SOARES, MACHADO, NORONHA, RODRIGUES, SINHORIN, PESSOA and JÚNIOR2013), which probably provides chemical protection from predators.

In the present study, the effectiveness of bufadienolide-based defences in R. haematiticus and alkaloid-based defences in D. auratus were compared against two potential predators – the bullet ant (P. clavata) and ctenid spider (C. coccineus). In addition, predation between adult and juvenile D. auratus was compared using bullet ant predators.

Predation trials took place from 9 June to 30 July 2012 at the Organization for Tropical Studies (OTS), La Selva Biological Research Station, Costa Rica (10°26′N, 83°59′W). Dendrobates auratus is a diurnal, conspicuously coloured, alkaloid-containing poison frog that is present in leaf-litter, and Rhaebo haematiticus is a diurnal, cryptically coloured, steroid-containing bufonid frog that also occurs in leaf-litter at La Selva (Guyer & Donnelly Reference GUYER and DONNELLY2005). Craugastor fitzingeri is a common leaf-litter frog that occurs in the same microhabitat as D. auratus and R. haematiticus, and is similar in size to both of the experimental frogs; however, it is not chemically defended, and therefore served as a negative control in all of the predation experiments. The predatory invertebrates used in this study were the bullet ant, P. clavata, and the ctenid spider, C. coccineus, both of which are abundant at the study site. All individuals of C. fitzingeri, R. haematiticus, 16 adult D. auratus and six juvenile D. auratus were collected from La Selva. The number of D. auratus at La Selva was fewer than expected, and therefore a second location (Chilamate, Costa Rica, 10°26′N, 84°05′W) was used to acquire an additional 15 D. auratus. The D. auratus obtained from this population were only presented to ctenid spiders, and therefore no D. auratus from La Selva were presented to ctenid spiders. All frogs were held in individual containers with leaf-litter for no more than 3 d preceding an experiment.

Bullet ant predation trials consisted of 16 adult D. auratus (mean SVL = 32.8 mm), 15 adult R. haematiticus (mean SVL = 33.6 mm), and 16 adult C. fitzingeri (mean SVL = 32.9 mm). In addition six trials were conducted with juvenile D. auratus (mean SVL = 22.6 mm) and juvenile C. fitzingeri (mean SVL = 22.6 mm). Paraponera clavata (mean length = 23.3 mm) actively nests between buttresses of trees, and forages during the day (Young & Hermann Reference YOUNG and HERMANN1980). Paraponera clavata is a generalist predator and has been found to prey on plant parts, arthropods and small vertebrates (Fritz et al. Reference FRITZ, STANLEY and DEPAMPHILIS1981, Young & Hermann Reference YOUNG and HERMANN1980). Predation experiments involving P. clavata were conducted at nests between 16h00 and 17h00. Trials were conducted along foraging trails on tree trunks. Following Fritz et al. (Reference FRITZ, STANLEY and DEPAMPHILIS1981), individual frogs were presented to ants in the foraging line by holding each frog by its right hind limb with a pair of 30.5-cm forceps. Individual frogs of each species were randomly presented to a single nest of ants and individual frogs were used only for a single trial. Ant nests were not used more than once a day, and the same nest was used an average of six times during the study. Each bullet ant trial ran for 5 min, and began when an ant made contact with the frog. Modifying the methods of Fritz et al. (Reference FRITZ, STANLEY and DEPAMPHILIS1981), frogs were scored as: (1) preyed upon (not released after attack) or (2) not preyed upon (touched with antennae and avoided). Following trials, all surviving frogs were held for 24 h to allow recovery from any attacks, and then returned to their original site of capture.

Ctenid spider predation trials consisted of 15 adult D. auratus (mean SVL = 35.0 mm), 15 adult R. haematiticus (mean SVL = 34.0 mm) and 15 adult C. fitzingeri (mean SVL = 35.0 mm). Cupiennius coccineus (mean length = 24.4 mm) is typically found on vegetation at night and is a sit-and-wait predator (Barth et al. Reference BARTH, SEYFARTH, BLECKMANN and SCHÜCH1988). Cupiennius coccineus has been found to prey on large invertebrates and small frogs (Szelistowski Reference SZELISTOWSKI1985, MMH and RAS, pers. obs.). Predation experiments with C. coccineus were conducted at night between 20h30 and 24h00. Trials were conducted on the vegetation upon which the spiders were found. Following Fritz et al. (Reference FRITZ, STANLEY and DEPAMPHILIS1981), individual frogs were randomly presented to an individual spider using 30.5-cm forceps in the same manner as described above. Individual frogs were only used for a single trial, and individual spiders were not utilized more than once in the entire experiment. Ctenid spider trials were conducted for 2 min, beginning when the frog was presented to the spider. Modified from the methods of Szelistowski (Reference SZELISTOWSKI1985), frogs were scored as: (1) preyed upon (not released after attack) or (2) not preyed upon (not attacked or released after an attack). Following trials, all surviving frogs were held for 24 h, and then returned to their original site of capture.

Binary logistic regression was used to determine if frog species and frog size (adult vs. juvenile) were significant predictors of bullet ant and ctenid spider predation. All statistical analyses were conducted in SPSS version 18.0.

Bullet ants preyed upon zero (0/16) adult D. auratus, one (1/15) adult R. haematiticus and 12 (12/15) adult C. fitzingeri. Frog species was a significant predictor of bullet ant predation, and adult C. fitzingeri were 45 times more likely to be preyed upon when compared with D. auratus (P = 0.001; odds ratio = 45.0; CI_0.95 = 4.4–457.5) and 20 times more likely when compared with R. haematiticus (P = 0.002; odds ratio = 19.5; CI_0.95 = 3.0–126.5). There was no difference in bullet ant predation between adult D. auratus and R. haematiticus (P = 0.571). Bullet ants preyed upon zero (0/6) juvenile D. auratus and five (5/6) juvenile C. fitzingeri. Frog species was a significant predictor of bullet ant predation, and juvenile C. fitzingeri were 25 times more likely to be preyed upon when compared with juvenile D. auratus (P = 0.038; odds ratio = 25.0; CI_0.95 = 1.2–521.0). Frog age/size class was not a significant predictor of predation between juvenile and adult D. auratus (P = 0.466).

Ctenid spiders preyed upon one (1/15) adult D. auratus, one (1/15) adult R. haematiticus and seven (7/15) adult C. fitzingeri. Frog species was a significant predictor of ctenid spider predation, and adult C. fitzingeri were 12 times more likely to be preyed upon when compared with D. auratus (P = 0.030; odds ratio = 12.3; CI_0.95 = 1.3–118) and R. haematiticus (P = 0.030; odds ratio = 12.3; CI_0.95 = 1.3–118). There was no difference in ctenid spider predation between adult D. auratus and R. haematiticus (P = 0.999).

Bufonid and dendrobatid frogs are both considered chemically defended anurans, yet they utilize different types of chemicals in their defence against predators – bufonids possess synthesized bufadienolides, whereas dendrobatids contain sequestered alkaloids. The present field-based study demonstrates that both R. haematiticus (bufonid) and D. auratus (dendrobatid) are similarly protected from predation by the predatory ant P. clavata and ctenid spider C. coccineus. These results provide evidence that bufadienolide-based and alkaloid-based chemical defences are equally effective at deterring predation by the same arthropod predators. In both cases, predator avoidance was likely associated with the presence of chemical defences, whereas the control frogs were palatable (Brodie et al. Reference BRODIE, FORMANOWICZ and BRODIE1978, Formanowicz & Brodie Reference FORMANOWICZ and BRODIE1982, Fritz et al. Reference FRITZ, STANLEY and DEPAMPHILIS1981, Szelistowski Reference SZELISTOWSKI1985).

Bullet ants always made contact with frogs before deciding to prey upon them. In several experimental trials, bullet ants would wipe their mandibles on a substrate after coming into contact with R. haematiticus or D. auratus, suggesting that the defensive chemicals were distasteful (i.e. unpalatable). Interestingly, this behaviour never occurred after a bullet ant came into contact with C. fitzingeri (control frogs). Similar results were reported by Fritz et al. (Reference FRITZ, STANLEY and DEPAMPHILIS1981) and Murray et al. (Reference MURRAY, BOLTON, BERG and SAPORITO2016), both of which demonstrated that O. pumilio are chemically defended from P. clavata. In several instances throughout the experiment, after coming into contact with D. auratus, C. coccineus wiped their pedipalps with their anterior appendages, suggesting that alkaloids were considered unpalatable, as has been observed in other studies (Murray et al. Reference MURRAY, BOLTON, BERG and SAPORITO2016, Szelistowski Reference SZELISTOWSKI1985). Gray et al. (Reference GRAY, KAISER and GREEN2010) observed a similar behaviour (fang wiping) in the tarantula Sericopelma rubronitens after it was presented to adult Dendrobates auratus. These behaviours never occurred after C. coccineus came in contact with C. fitzingeri or R. haematiticus; however, the spiders always retreated similarly after coming in contact with D. auratus or R. haematiticus. The cleaning behaviours exhibited by each arthropod predator after making contact with R. haematiticus (bullet ants) and D. auratus (bullet ants and ctenid spiders) provide evidence that both bufadienolides and alkaloids are considered unpalatable and are equally avoided.

Juvenile poison frogs are smaller in size, and have smaller poison glands and reduced quantities of alkaloids (Saporito et al. Reference SAPORITO, ISOLA, MACCACHERO, CONDON and DONNELLY2010, Stynoski et al. Reference STYNOSKI, TORRES-MENDOZA, SASA-MARIN and SAPORITO2014). Therefore, it may be possible for predators to detect this reduced amount of alkaloids and more successfully prey on juvenile frogs. Murray et al. (Reference MURRAY, BOLTON, BERG and SAPORITO2016) reported that juvenile O. pumilio were preyed on significantly more often than adults by P. clavata, which was attributed to the smaller quantities of alkaloids in juveniles. The results of the present study, however, demonstrate that juvenile D. auratus appear to be as equally protected against predation by P. clavata when compared with adults. The differences observed between the two studies are likely due to differences in alkaloid quantities between juvenile D. auratus and O. pumilio. Although juvenile D. auratus and O. pumilio are both smaller than adults of their respective species, the juvenile D. auratus used in the present study were approximately the same size as adult O. pumilio. Therefore, it is likely that juvenile D. auratus contain relatively similar quantities of alkaloids when compared with adult O. pumilio, which appears to provide them equal protection from bullet ant predators.

Overall, the findings of the present study provide evidence that although the defensive chemicals present in D. auratus and R. haematiticus are different, they are both equally effective at deterring predation by the same invertebrate predators.