Dietary specialization predicts toxicity in recently diverged lineages of poison frogs

Texto completo

(1)1. Dietary specialization predicts toxicity in recently diverged lineages of poison frogs. 2. Lina María Arenas, Emilio Realpe, Adolfo Amézquita. 3. ABSTRACT. 4. Bright colourations allow dart-poison frogs to advertise bad taste and reduce predation. Several. 5. studies have proposed that skin alkaloids of poison frogs come from dietary sources, namely. 6. arthropods such as ants and mites. Evidence suggests concomitant variations in toxicity and. 7. diet specialization among poison frog genera. However, no studies have evaluated whether such. 8. patterns hold at the intraspecific level. We tested whether diet and toxicity were correlated. 9. among five Oophaga lineages, through toxicity bioassays and stomach contents analyses.. 10. Despite all species have similar diets, and behave as specialist predators, we found that small. 11. variations in diet composition imply noticeable variations in diet breadth and toxicity within. 12. lineages. We provide the first line of evidence confirming concomitant variations between dietary. 13. specialization and toxicity and bringing further support for the diet–toxicity hypothesis at the. 14. finest phylogenetic scale.. 15 16. INTRODUCTION:. 17. Species that use only a subset of the available resources or prey in lower proportions. 18. than the average field availability are usually called diet specialists (Kelley and Farrell. 19. 1998; Nosil 2002; Toft 1980). Most of these species are considered active foragers as. 20. they spend more time and energy during prey search, persecution and capture, being. 21. subject to higher risks of predation (Abrahams and Dill 1989; Kohler and Mcpeek 1989).. 22. In turn, predation pressures have promoted the evolution of several defense. 23. mechanisms in active foragers, such as bright colourations that advertise bad taste or. 24. toxicity to potential predators. These signals, that are learned and avoided by predators,. 25. are called aposematic. Aposematism is a widespread phenomenon in nature (Cott. 26. 1940; Inbar and Lev-Yadun 2005) and has been studied in several organisms including.

(2) 27. marine invertebrates (Lindquist and Hay 1996), butterflies (Mallet and Joron 1999),. 28. plants (Rubino and Mccarthy 2004) and amphibians (Cott 1940), such as the species. 29. belonging to the dart-poison frog family Dendrobatidae (Silverstone 1975, 1976;. 30. Summers and Clough 2001).. 31. Most dendrobatids are diurnal and inhabit the leaf litter of tropical rain forests from. 32. Costa Rica to Bolivia. Dart-poison frogs are thought to be distributed along a continuum. 33. where specialized feeding habits have been consistently related to the accumulation of. 34. toxic lipophilic skin alkaloids sequestered from dietary sources (Daly and Myers 1967;. 35. Saporito et al. 2004; Toft 1995). Eating preferences seem to be biased towards prey. 36. such as ants, mites and termites (Toft 1980), which have been proposed as the major. 37. sources of alkaloids in dendrobatids (Daly et al. 1994; Dumbacher et al. 2004; Saporito. 38. et al. 2007b). As a consequence, most research that addresses the origins of. 39. aposematism in dendrobatids has focused on the differences among poison frog. 40. genera, comparing cryptic generalists with aposematic specialists (Darst et al. 2005;. 41. Toft 1995). Recent evidence suggests that intraspecific diet preferences (Valderrama-. 42. Vernaza et al. 2009) and alkaloid composition (Saporito et al. 2007a) have seasonal. 43. variation, suggesting that coupled variations between diet and toxicity should be weak. 44. or absent at the population or individual levels.. 45. No studies have tested whether diet and toxicity vary concomitantly at phylogenetic. 46. resolutions finer than the genera.. 47. Oophaga histrionica is a dendrobatid species widely distributed along the Chocoan. 48. lowlands of western Colombia. Evidence suggests this species behaves as a specialist. 49. predator, having at least 15 different colour variants which also differ in their alkaloid. 50. compositions (Myers and Daly 1976;Darst et al. 2005). In fact, one of these variants is. 51. considered to be a different species (Oophaga lehmanni) based on its particular skin. 52. alkaloids, although it is separated from other populations of O. histrionica by just a few.

(3) 53. kilometers. We propose the histrionica species complex as a model to test whether. 54. intra and interspecific variations in toxicity are coupled to concomitant differences in. 55. dietary specialization. We estimated and described diet and toxicity and because both. 56. invertebrate availability and frog ecological traits may be affected by the geographic. 57. distance among the study localities, we further tested for correlations between. 58. geographic / genetic distances and the studied traits.. 59 60. METHODS. 61. STUDY SPECIES AND TISSUE COLLECTION. 62. The genus Oophaga includes at least eight species (Grant et al. 2006) within which both. 63. O. histrionica and O. lehmanni are endemic to Colombia and O. sylvatica occurs also in. 64. Ecuador. All species inhabit the tropical rain forests, occurring from 100 to 1200 m.a.s.l.. 65. (Myers and Daly 1976; Silverstone 1975). While O. lehmanni has a narrow distribution. 66. and little colour variation with only three known morphs (i.e., black with blue, red or. 67. yellow stripes), O. histrionica is distributed across the biogeographic Chocó region in. 68. Colombia (Myers 1976; Medina & Amézquita in prep.).. 69. All skin and stomach samples were collected during April and May 2008 to avoid. 70. seasonal changes in the relative field abundances of arthropods (and possibly toxins) as. 71. suggested by previous studies (Berazategui et al. 2007; Valderrama-Vernaza et al.. 72. 2009). Frogs were captured using a small plastic cup in order to reduce stress and. 73. avoid the release of toxins that could hamper test results. Each individual’s identity,. 74. locality, georeference and sex (when possible) were recorded. We photographed each. 75. animal over a flat surface and next to a centimeter scale using a Sony Cyber-shot®. 76. DSC-H5 camera. Photographs were used to analyze each individual’s snout – vent. 77. length (SVL) using the software Image J (available online at http://rsbweb.nih.gov/ij/).. 78. Individuals were euthanized using a 0.1 mL, 100% ethanol shot to the heart, and.

(4) 79. skinned right after to avoid any contamination or toxin loss. Each skin sample was. 80. divided and stored in three plastic vials containing 1.5 mL of 100% methanol and taken. 81. to the lab for analyses. We collected most of the stomach contents by inducing the. 82. eversion of the stomach, a procedure that has proven to be as effective as the. 83. examination of the stomach contents of preserved animals (Amézquita in prep.),. 84. although we also included the stomach contents of the sacrificed individuals in our. 85. analyses. Contents were extracted and stored in separate vials with 1.5 mL of 90%. 86. ethanol each. To avoid any post- mortem digestion, we made a small cut on each. 87. stomach to allow ethanol inside the tissue. Table 1 summarizes the number of skin and. 88. stomach samples collected for each species.. 89 91. Table 1: Species location and number of samples collected. 93 95 97 99 101 102 103. DIET ANALYSES. 104. We separated and determined each prey to the lowest taxonomic category possible. 105. (usually family), using a Zeiss® StemiSV6 stereoscope and following the dichotomy. 106. keys by Wolff (2003) & Harvey (1992). Each item was photographed using a Canon®. 107. PowerShot G5 camera adapted to the stereoscope. After determination, each prey was. 108. stored in a 0.2mL tube and each stomach was stored in a plastic bag. Prey. 109. abundances for each species were used to describe each species’ diet and to calculate. 110. dietary niche breadths using the approximation proposed by Colwell and Futuyma.

(5) 111. (1971). As the Oophaga species we analyzed in this study do not coexist, we did not. 112. use any of the niche similarity indexes that have been proposed (Hurlbert 1978; Pianka. 113. 1974; Van Horne and Ford 1982). Instead, we calculated dietary niche dissimilarities. 114. using a canonical discriminant analysis combining overall prey abundances in all. 115. species. For this analysis we used the software JMP statistics version 8.0.. 116 117. TOXICITY ANALYSES. 118. To prepare skin samples for the toxicity tests, we cut each skin into small pieces and. 119. liquefied them with a PRO® 200 homogenizer (Pro Scientific Inc.) using the same. 120. methanol in which they were first stored. After homogenization, extracts were filtered. 121. using a fine screen mesh in order to sieve out toxin remnants from large skin pieces in. 122. the sample. Methanol excesses were evaporated using a JOUAN® RC1010 vacuum. 123. centrifuge calibrated at 38 ̊C for an average period of three hours. Because we had. 124. separated each skin sample in three different tubes, dry pellets were merged into one. 125. tube and then resuspended in 0.2 mL of 0.9% sterile saline solution.. 126. We used a total of 20 common laboratory mice (Mus musculus, strain CFW) to conduct. 127. the toxicity tests. Only one individual was used per skin extract and their body weights. 128. ranged from 20g to 25 g; a random number of males and females were used, as our. 129. goal was not to measure differential toxin effects according to gender. We believed that. 130. a possible way of quantifying each population’s toxic effect on the nervous system was. 131. to estimate the differences in mice open field activities before and after intoxication. 132. (Katz et al. 1981; Fride and Mechoulam 1993). Preliminary tests, using 0.1 mL, 0.9%. 133. sterile saline solution shot, indicated that the injection per-se did not have any effects on. 134. mice displacement when properly administered.. 135. The differences in open field activities were measured using as experimental arena a 29. 136. cm tall x 38 cm wide x 20 cm depth glass aquarium printed out with a 5 x 5 cm grid on.

(6) 137. its base. Each experiment consisted on two main test periods: The control period. 138. began by placing each mouse inside the aquarium and registering the cumulative. 139. number of squares wandered every 20 seconds during three minutes. Subsequently,. 140. the experimental period consisted of injecting each mouse intraperitoneally with 0.1 mL. 141. out of the 0.2 mL saline-toxin extract suspension described above. Immediately after. 142. the injection, individuals were placed back in the aquarium where we registered the. 143. number of squares wandered after the injection. If the test was fatal, we registered the. 144. time until total paralysis, as well as the time until the last breath or muscle contraction.. 145. On the other hand, if the mouse did not die, the experiment would continue for 18. 146. minutes, which was considered the mean standardized time taken for recovery (i.e.. 147. regain of all motor functions) during preliminary tests. After each test, we tagged,. 148. weighted, sexed and monitored each mouse for possible side or late toxic effects. We. 149. measured toxicity values for each Oophaga lineage as the difference in the mean field. 150. activity between test periods (after injection - before injection).. 151. We used separate analyses of variance (ANOVA) to test wether the lineages we. 152. analyzed differed in their snout-vent lengths, to evaluate the change of overall activities. 153. between the two periods of the toxicity tests and to probe whether the different. 154. Oophaga morphs had different prey preferences. All the ANOVAs were followed by. 155. post-hoc Honestly Significant Difference (HDS) tests when relevant. In order to evaluate. 156. whether diet breadths among species were correlated to their correspondent toxicities. 157. we run a linear regression using the logarithmic values of niches and toxicity. All. 158. statistical tests were made using the software SPSS version 16.0 (SPSS Inc.). 159 160. GENETIC AND GEOGRAPHIC RELATIONS WITH DIET AND TOXICITY. 161. Oophaga lineages have recent divergence times (Medina & Amézquita in prep.). Thus,. 162. we used a patristic approach rather than a phylogenetic one to estimate correlations.

(7) 163. across the different lineages and their corresponding variations in diet breadth and. 164. toxicity. Patristic genetic distances between each pair of populations were calculated. 165. using the software PATRISTIC (Fourment and Gibbs 2006) using genomic DNA. 166. sequences from Medina & Amézquita (in prep.). Geographic distances were calculated. 167. using the Geographic Distance Matrix Generator (version 1.2.3) available online at:. 168. http://biodiversityinformatics.amnh.org/open_source/gdmg. We used linear distances,. 169. as there seem to be no true geographic barriers separating the different populations.. 170. The mean differences in toxicity and diet breadth made up two different data matrices,. 171. which were tested for correlation with geographic and genetic data using Le Progiciel R. 172. Mantel test (Casgrain and Legendre 2001).. 173 174. RESULTS:. 175. Despite all lineages’ diets included at least 18 different arthropod orders, 60 to 85% of. 176. the stomach contents consisted only of ants (Fig. 1a). Among ants, myrmecines were. 177. the most abundant prey, constituting almost 95% of all the ants found. About 3 to 5 %. 178. of all ants corresponded to the two other ant subfamilies we found: Formicinae and. 179. Ponerinae (Fig. 1b). Hymenopterans were both, the most abundant and the most. 180. prevalent prey item in all five populations (Fig. 2). Acari, larvae and coleopterans were. 181. also abundant items, although these orders combined only made up about 20% of all. 182. prey items analyzed. The remaining prey orders were found sporadically (i.e. only one to. 183. 10 items). Oophaga histrionica from La Delfina was found to be the largest species and. 184. O. histrionica from El Naranjo, the smallest (ANOVA, df= 4, F= 84.58, p < 0.001; Tukey. 185. HSD p<0.001). The later also consumed more Acari than all other lineages (ANOVA, df=. 186. 4, F= 0.847, p= 0.507; Tukey HSD p= 0.085). Furthermore, this population ate more. 187. prey categories, having the highest values of niche breadth, followed by the two other. 188. O. histrionica morphs, O. sylvatica and O. lehmanni (ANCOVA, df= 4, F= 6.201, p=.

(8) 189. 0.001, Tukey HSD p=0.054), (Fig. 3). Niche overlap analysis showed two canonical. 190. functions together explaining 79.4% of the variance in the relative abundance of prey. 191. items. CA1 separated lineages with higher abundances of prey such as ants, larvae,. 192. spiders and opiliones, from species having mites, hemipterans, pseudroscorpions and. 193. other hymenopterans on their stomachs.. 194. Toxicity tests revealed that overall activity decreased after the injection (ANOVA, n=2,. 195. df= 1 F= 22.525, p<0.001). We found that the Oophaga lineages we analyzed differed. 196. in their toxicity (ANOVA, df= 4, F= 6.969, p<0.001) The Oophaga histrionica lineage. 197. was less toxic than O. sylvatica and O. lehmanni (Tukey HSD p= 0.09) (Fig. 3). Aside. 198. from having lower activity, 35% of the mice injected with the toxin extracts from O.. 199. histrionica died and 14% had visible difficulties breathing and moving. In all cases,. 200. toxicity increased with narrower dietary niche breadths (linear regression, R2= 0.295,. 201. ANOVA df=1, F=13.826, p=0.001); (Fig. 4). Neither geographic nor patristic genetic. 202. distances were correlated to toxicity or diet breadth (Mantel r= 0.4782, p=0.1529). 203 204. DISCUSSION. 205. Our data suggest that even small differences in dietary specialization are correlated with. 206. concomitant variation in toxicity. Accordingly, our results revealed just small variations in. 207. the diet compositions of the Oophaga lineages we analyzed. Ants, and more precisely. 208. myrmecines, were consistently the most abundant prey item, while other prey were. 209. relatively rare. We found that the most toxic lineages ate slightly larger proportions of. 210. hymenopterans thus, having narrower dietary niches.. 211. The overall predominance of ants in the diets of the different Oophaga lineages is not. 212. surprising. Several studies have proposed that toxic dendrobatid genera behave as. 213. specialists, eating larger amounts of ants than non-toxic genera (Toft 1995; Daly et al..

(9) 214. 1997; Santos et al. 2003). Nevertheless, ants are one of the most abundant arthropod. 215. orders in the wild, thus, a few authors have proposed that aposematic poison frogs. 216. should be considered generalists (Simon and Toft 1991; Lima and Moreira 1993;. 217. Woodhead et al. 2007). Even if we did not measure prey abundances in the wild, we. 218. could still suggest that the Oophaga lineages we analyzed are in fact diet specialists,. 219. since the specialist continuum remains the same with or without this analysis (Toft. 220. 1980; Darst et al. 2005). Ants have also been suggested as the primary source of. 221. alkaloids (Daly et al. 2000; Saporito et al. 2004) and evidence suggests that they. 222. secrete a wide variety of alkaloids through their poison and mandibular glands (Numata. 223. & Ibuka, 1987). Thereby, it is not strange that all frog lineages prefer hymenopterans. 224. above other items.. 225. An interesting result was that the small frog lineages had overall high mite abundances. 226. in their stomachs. However, size differences may only implicate that these species have. 227. smaller gapes, and thus, are obliged to consume larger amounts of small prey, due to. 228. morphological constraints (Huey and Pianka 1981; Schmitt and Holbrook 1984). Given. 229. that mites are generally considered highly chitinous prey they should have poor. 230. digestibility, meaning that larger numbers should be consumed in order to have the. 231. appropriate energy rewards (Toft 1995). Nonetheless, mites have been proposed as a. 232. major source of alkaloids in Oophaga pumilio (Saporito et al. 2007b; Saporito et al.. 233. 2004), a species that has also been reported to have overall small body sizes, ranging. 234. from 19 to 22 cm (Stynoski 2009). Consequently, if small Oophaga lineages consume. 235. more mites, they may be sequestering a different subset of alkaloids than that expected. 236. if they ate more ants, and this could explain the differences in toxicity we have found.. 237. Our results indicate that small variations in prey consumption resulted in striking. 238. differences in niche breadths and toxicities of the Oophaga lineages we analyzed.. 239. Furthermore, this study takes a step forward supporting the diet-toxicity hypothesis.

(10) 240. even at the species level, which supposes the finest evolutionary scale to test. 241. concomitant relationships between these two characters. Our findings suggest that the. 242. evolution of diet and toxicity could be related to foraging efficiency, to different abilities. 243. within lineages to sequester specific alkaloids from dietary sources or to the a. 244. combination between these patterns and the influence of prevailing ecological. 245. conditions within each locality. Although Formicidae continued to be the most abundant. 246. prey in stomach contents as most studies have found ( Toft 1995; Santos et al. 2003;. 247. Darst et al. 2005), our study indicates that higher proportions of myrmecine ants are. 248. correlated with toxicity, a novel result worth studying in detail.. 249. Analyzing specialization at finer, intraspecific scales has led our study to conclude that. 250. diet and toxicity within the histrionica species complex do vary concomitantly. Our. 251. results verify the diet-toxicity hypothesis indicating that narrower diet breadths predict. 252. higher toxicities even at fine intraspecific levels. We propose a new possible source of. 253. alkaloids for dart poison frogs, suggesting that the evolution of diet preferences in this. 254. genus may have arisen as a consequence of selection pressures over the different. 255. Oophaga lineages.. 256 257 258 259 260 261 262 263.

(11) 264. FIGURE LEGENDS:. 265. Figure1a: Diets of the five Oophaga lineages analyzed in this study. Although these. 266. frogs include at least 18 different prey categories, ants remain the most abundant prey.. 267. Figure 1b: Proportion of ant subfamilies consumed by each of the Oophaga lineages.. 268. Myrmecines (green bars) were far more abundant than the other two subfamilies found,. 269. Ponerina (red bars) and Formicinae (yellow bars).. 270. Figure 2: Abundance and prevalence of the different prey categories found in the. 271. stomach contents of the five Oophaga lineages. Alkaloid contributing prey such as ants,. 272. mites, larvae (Diptera and Coleoptera) and adult Coleoptera were present in most. 273. stomachs and had greater amounts than other prey types.. 274. Figure 3: Relationship between the Oophaga lineages size (SVL) and the four main. 275. variables analyzed in this study. Dietary niche breadth, as well as ant and mite. 276. proportions eaten had no relation with size, but were correlated to frog morph. Black. 277. bars on each dot represent the standard error for the mean body size (horizontal bars). 278. and each variable (vertical bars).. 279. Figure 4: Differences in niche breadth and toxicity between the five Oophaga lineages.. 280. Black bars on each dot represent the standard error for the mean niche breadth. 281. (horizontal bars) and toxicity (vertical bars).. 282 283 284 285 286 287.

(12) 288. REFERENCES:. 289. Abraham s, M . & Dill, L. 1989. A determination of the energetic equivalence of the. 290. risk of predation. Ecology, 70, 999-1007.. 291. Berazategui, M., Camargo, A. & Maneyro, R. l. 2007. Environmental and. 292. seasonal variation in the diet of elachistocleis bicolor (guérin-méneville 1838) (anura:. 293. Microhylidae) from northern uruguay. Zoological Science, 24, 225-231.. 294. Casgrain, P. & Legendre, P. 2001. The r package for multivariate and spatial. 295. analysis. Département de sciences biologiques, Université de Montréal.. 296. Colwell, R. K. & Futuyma, D. J. 1971. On the measurement of niche breadth and. 297. overlap. Ecology, 52, 567-576.. 298. Cott, H. B. 1940. Adaptive coloration in animals. London: Methuen.. 299. Daly, J. W . & M yers, C. W . 1967. Toxicity of panamanian poison frogs. 300. (dendrobates): Some biological and chemical aspects. Science, 156, 970-973.. 301. Daly, J. W ., M artin Garraffo, H., Hall, G. S. E. & Cover, J. F. 1997. Absence. 302. of skin alkaloids in captive-raised madagascan mantelline frogs (mantella) and. 303. sequestration of dietary alkaloids. Toxicon, 35, 1131-1135.. 304. Daly, J. W ., Secunda, S. I., Garraffo, H. M ., Spande, T. F., W isnieski, A.. 305. & Cover, J. F., Jr. 1994. An uptake system for dietary alkaloids in poison frogs. 306. (dendrobatidae). Toxicon, 32, 657-663.. 307. Daly, J. W ., Garraffo, H. M ., Jain, P., Spande, T. F., Snelling, R. R.,. 308. Jaramillo, C. & Rand, A. S. 2000. Arthropod–frog connection: Decahydroquinoline. 309. and pyrrolizidine alkaloids common to microsympatric myrmicine ants and dendrobatid. 310. frogs. Journal of Chemical Ecology, 26, 73-85.. 311. Darst, C. R., M enndez-Guerrero, P. A., Colom a, L. A. & Cannatella, D. C.. 312. 2005. Evolution of dietary specialization and chemical defense in poison frogs. 313. (dendrobatidae): A comparative analysis. The American Naturalist, 165, 56-69..

(13) 314. Dum bacher, J. P., W ako, A., Derrickson, S. R., Sam uelson, A., Spande,. 315. T. F. & Daly, J. W . 2004. Melyrid beetles (choresine): A putative source for the. 316. batrachotoxin alkaloids found in poison-dart frogs and toxic passerine birds.. 317. Proceedings of the National Academy of Sciences of the United States of America, 101,. 318. 15857-15860.. 319. Fourm ent, M . & Gibbs, M . 2006. Patristic: A program for calculating patristic. 320. distances and graphically comparing the components of genetic change. BMC. 321. Evolutionary Biology, 6, 1.. 322. Fride, E. & M echoulam , R. 1993. Pharmacological activity of the cannabinoid. 323. receptor agonist, anandamide, a brain constituent. European journal of pharmacology,. 324. 231, 313-314.. 325. Grant, T., Frost, D. R., Caldwell, J. P., Gagliardo, R. O. N., Haddad, C. F.. 326. B., Kok, P. J. R., Means, D. B., Noonan, B. P., Schargel, W . E. &. 327. W heeler, W . C. 2006. Phylogenetic systematics of dart-poison frogs and their. 328. relatives (amphibia: Athesphatanura: Dendrobatidae). Bulletin of the American Museum. 329. of Natural History, 1-262.. 330. Harvey, M. 1992. The phylogeny and classification of the pseudoscorpionida. 331. (chelicerata : Arachnida). Invertebrate Systematics, 6, 1373-1435.. 332. Huey, R. & Pianka, E. 1981. Ecological consequences of foraging mode. Ecology,. 333. 62, 991-999.. 334. Hurlbert, S. H. 1978. The measurement of niche overlap and some relatives.. 335. Ecology, 59, 67-77.. 336. Inbar, M. & Lev-Yadun, S. 2005. Conspicuous and aposematic spines in the. 337. animal kingdom. Naturwissenschaften, 92, 170-172..

(14) 338. Katz, R. J., Roth, K. A. & Carroll, B. J. 1981. Acute and chronic stress effects. 339. on open field activity in the rat: Implications for a model of depression. Neuroscience &. 340. Biobehavioral Reviews, 5, 247-251.. 341. Kelley, S. T. & Farrell, B. D. 1998. Is specialization a dead end? The phylogeny of. 342. host use in dendronoctonus bark beetles (scolytidae). 52, 13.. 343. Kohler, S. & M cPeek, M . 1989. Predation risk and the foraging behavior of. 344. competing stream insects. Ecology, 70, 1811-1825.. 345. Lim a, A. P. & M oreira, G. 1993. Effects of prey size and foraging mode on the. 346. ontogenetic change in feeding niche of colostethus stepheni (anura: Dendrobatidae).. 347. Oecologia, 95, 93-102.. 348. Lindquist, N. & Hay, M . E. 1996. Palatability and chemical defense of marine. 349. invertebrate larvae. 66. 350. M allet, J. & Joron, M. 1999. Evolution of diversity in warning color and mimicry:. 351. Polymorphisms, shifting balance, and speciation. Annu. Rev. Ecol. Syst., 30, 201-233.. 352. M yers, C. W . & Daly, J. W . 1976. Preliminary evaluation of skin toxins and. 353. vocalization in taxonomic and evolutionary studies of poison-dart frogs (dendrobatidae).. 354. . Bulletin of the American Museum of Natural History, 175, 175 – 262.. 355. Nosil, P. 2002. Transition rates between specialization and generalization in. 356. phytophagous insects. 56, 6.. 357. Pianka, E. R. 1974. Niche overlap and diffuse competition. Proceedings of the. 358. National Academy of Sciences of the United States of America, 71, 2141-2145.. 359. Rubino, D. L. & McCarthy, B. C. 2004. Presence of aposematic (warning). 360. coloration in vascular plants of southeastern ohio. Journal of the Torrey Botanical. 361. Society, 131, 252–256..

(15) 362. Santos, J. C., Colom a, L. A. & Cannatella, D. C. 2003. Multiple, recurring. 363. origins of aposematism and diet specialization in poison frogs. PNAS, 100, 12792-. 364. 12797.. 365. Saporito, R. A., Garraffo, H. M., Donnelly, M. A., Edwards, A. L.,. 366. Longino, J. T. & Daly, J. W . 2004. Formicine ants: An arthropod source for the. 367. pumiliotoxin alkaloids of dendrobatid poison frogs. Proceedings of the National. 368. Academy of Sciences of the United States of America, 101, 8045-8050.. 369. Saporito, R. A., Donnelly, M . A., Jain, P., M artin Garraffo, H., Spande, T.. 370. F. & Daly, J. W . 2007a. Spatial and temporal patterns of alkaloid variation in the. 371. poison frog oophaga pumilio in costa rica and panama over 30 years. Toxicon, 50, 757-. 372. 778.. 373. Saporito, R. A., Donnelly, M . A., Norton, R. A., Garraffo, H. M ., Spande,. 374. T. F. & Daly, J. W . 2007b. Oribatid mites as a major dietary source for alkaloids in. 375. poison frogs. Proceedings of the National Academy of Sciences, 104, 8885-8890.. 376. Schm itt, R. & Holbrook, S. 1984. Eny of prey selection by black surfperch. 377. embiotoca jacksoni (pisces: Embiotocidae): The roles of fish morphology, foraging. 378. behavior, and patch selection. Mar. Ecol. Prog. Ser, 18, 225-239.. 379. Silverstone, P. A. 1975. A revision of the poison-arrow frogs of the genus. 380. dendrobates wagler. Natur. Hist. Mus. Los Angeles Co., Sci. Bull., 21, 1-55.. 381. Silverstone, P. A. 1976. A revision of the poison-arrow frogs of the genus. 382. phyllobates bibron in sagra (family dendrobatidae). Natur. Hist. Mus. Los Angeles Co.,. 383. Sci. Bull., 27, 1-53.. 384. Sim on, M . P. & Toft, C. A. 1991. Diet specialization in small vertebrates: Mite-. 385. eating in frogs. Oikos, 61, 263-278.. 386. Stynoski, J. 2009. Discrimination of offspring by indirect recognition in an egg-feeding. 387. dendrobatid frog, oophaga pumilio. Animal Behaviour, 78, 1351-1356..

(16) 388. Sum m ers, K. & Clough, M . E. 2001. The evolution of coloration and toxicity in the. 389. poison frog family (dendrobatidae). PNAS (Proc. Nat. Acad. Sci.), 98, 6227-6232.. 390. Toft, C. A. 1980. Feeding ecology of thirteen syntopic species of anurans in a. 391. seasonal tropical environment. Oecologia, 45, 131-141.. 392. Toft, C. A. 1995. Evolution of diet specialization in poison-dart frogs (dendrobatidae).. 393. Herpetologica, 51, 202-216.. 394. Valderram a-Vernaza, M ., Ram írez-Pinilla, M . P. & Serrano-Cardozo, V. H.. 395. 2009. Diet of the andean frog ranitomeya virolinensis (athesphatanura: Dendrobatidae).. 396. Journal of Herpetology, 43, 114-123.. 397. Van Horne, B. & Ford, R. G. 1982. Niche breadth calculation based on. 398. discriminant analysis. Ecology, 63, 1172-1174.. 399. W olff, M . 2003. Insectos de colombia: Guía básica de familias Medellín.. 400. W oodhead, C., Vences, M ., Vieites, D. R., Gam boni, I., Fisher, B. L. &. 401. Griffiths, R. A. 2007. Specialist or generalist? Feeding ecology of the malagasy. 402. poison frog mantella aurantiaca. 17, 12.. 403 404 405 406 407 408 409 410 411.

(17) 412. Figures:. 413. Figure 1:. (15). Myrnecinae Formicinae Ponerinae. 414 415 416 417 418 419 420 421. (10). (7). (12). (10).

(18) Figure 2:. Percentage of stomachs containing each prey. 422. 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438. O. histrionica El Naranjo (15) Ants. O. histrionica El Amargal (10). O. histrionica La Delfina (7). O. sylvatica La Codicia (12). Acari Larvae Coleoptera. Average prey per stomach contents. O. lehmanni Anchicayá (10).

(19) 439. 440. Figure 3:.

(20) 441. 442. Figure 4:.

(21)