SALAMANDRA 54(1) 30–36 15 FebruaryJ 2o0s1e8phineI SAS.N N 0o0t3h6a–c3k37e5r et al. Homing behavior in the Neotropical poison frog Ameerega trivittata Josephine A. Nothacker, Clara P. Neu, Michael Mayer, Norman Wagner & Stefan Lötters Trier University, Department of Biogeography, 54286 Trier, Germany Corresponding author: Stefan Lötters, e-mail: [email protected] Manuscript received: 3 October 2016 Accepted: 8 June 2017 by Jörn Köhler Abstract. Neotropical poison frogs (Aromobatidae, Dendrobatidae) are known for their complex behavior including site fidelity and home range maintenance. It has been shown in a few poison frog species that these amphibians are able to re- turn to their home ranges after experimental translocation. In this study we asked if Ameerega trivittata can be allocated to the species performing homing behavior. In this taxon, males and females show home range behavior, while sexes were not distinguished in our study. Fieldwork was carried out in a wild population at Panguana (Peru), using replacement dis- tances of 150 m, 600 m and 900 m. In total, 79 frogs were translocated. Most rapidly returned to their home ranges from all translocation distances, with a decrease of the homing success with longer distance. Among the poison frogs studied so far, it is remarkable that A. trivittata is the only one known to be able to return from 900 m (which perhaps is a remark- able homing distance for anurans in general), while maximum return distances in other species are less than 50% of this. Ameerega trivittata is one of the largest poison frogs (maximum snout–vent length 55 mm). However, long distance hom- ing was not explained by the species’ body size. We rather expect that ‘good knowledge’ of the general area (in terms of integration of learned landmarks) that frogs live in is the reason for the ability of long distance homing in our focal taxon. Key words. Amphibia, Anura, Dendrobatidae, experimental translocation, Peru, site fidelity. Introduction Holenweg Peter et al. 2001). In recent years, there has been some research progress with regard to diurnal terres- Homing behavior in animals is defined as the ability of an trial anurans from the Neotropics, mainly in the intriguing individual to return over little or unknown terrain to the group of poison frogs of the families Aromobatidae and place where it ‘lives’, its home range (Burt 1943). This phe- Dendrobatidae (Nowakowski et al. 2013, Pašukonis et al. nomenon was reviewed by Papi (1992) and Lavar & Kelly 2013, Pichler et al. 2017). (2008) and is found in many arthropods and vertebrates, Poison frogs are a diverse group of small (ca. 2–6 cm such as decapod crustaceans (Pittman & McAlpine adult size), almost exclusively diurnal amphibians, as- 2003), salmons (Quinn & Dittman 1990) or pigeons sociated with humid habitats from rain forest to páramo (Walcott 1996). Various advantages of such site fidelity (Grant et al. 2006). They are well known for having bright have been discussed, in particular the maximization of re- vivid colorations and skin toxins, although about two thirds source exploitation, which often is related to reproduction. of the ca. 300 taxa lack these traits (cf. Kahn et al. 2016). Among the Amphibia, many species are known to Poison frogs show complex reproductive and social be- show site fidelity (Sinsch 2014). This is frequently accom- haviors, including larval transport by parent frogs to small panied by seasonal migratory return, such as to a ‘tradi- water bodies and other parental brood care strategies, in- tional’ breeding pond (Duellman & Trueb 1986, Sinsch traspecific aggressiveness and complex interspecific mim- 2006, 2014, Wells 2007). As a result of this, numerous am- icry (Lötters et al. 2007). Many species show strong site phibians are able to perform homing behavior; what can fidelity and defend territories in one or both sexes (Pröhl be studied through experimental translocation (reviewed 2005). This behavior is suggested to be associated with the by Sinsch 1990). As summarized by Wells (2007), hom- utilization of beneficial resources, such as food or shelter ing behavior in salamanders and newts is relatively well availability, or to be profitable in mate attraction. Resource understood, compared to other amphibians. In frogs and use also explains homing behavior observed in some poi- toads, homing behavior so far has mostly focused on noc- son frog species (Nowakowski et al. 2013, Pašukonis et al. turnal taxa from the temperate region (e.g., Bogert & Sta- 2013). An additional reason for pronounced homing abili- tion 1947, Dole 1968, Jameson 1957, Twitty et al. 1964, ties is that water bodies utilized for larval deposition are of- © 2018 Deutsche Gesellschaft für Herpetologie und Terrarienkunde e.V. (DGHT), Mannheim, Germany A3v0ailable at http://www.salamandra-journal.com Homing behavior in Ameerega trivittata Table 1. Summary of homing behavior in poison frog species after experimental translocation. Species, sex, country Translocation distance Successful homing rate Source Aromobatidae Allobates femoralis, 50 m 80% (N = 10) Pašukonis et al. (2013) males, French Guiana 100 m 100% (N = 10) (SVL ≤ 33.5 mm) 200 m 80% (N = 10) 400 m 30% (N = 10) 800 m 0% (N = 10) Allobates talamancae, 20 m 83.32% (N = 12) Pichler et al. (2017) males, Costa Rica 200 m 70% (N = 10) (SVL ≤ 24.0 mm) Dendrobatidae Oophaga pumilio, 3 m (females only) 80% (N = 5) McVey et al. (1981) males and females, Costa Rica 6 m 90% (N = 10) (SVL ≤ 24.0 mm) 12 m 90.91% (N = 11) 20 m (females only) 83.33% (N = 6) O. pumilio, 20 m 67% (N = 30) Nowakowski et al. (2013) sexes not distinguished, Costa Rica 30 m 57% (N = 30) ten limited and specimens have to carry tadpoles over sev- their home ranges over longer distances than the previous- eral hundred meters away from their home ranges (Pröhl ly studied, smaller poison frog species. In concert with this, & Hödl 1999, Ringler et al. 2009). As summarized in we hypothesized that (3) successful homing increases with Table 1, of the 18 genera in the two poison frog families, larger body size among conspecifics of the study species. homing after experimental translocation has been exam- ined in three species of two genera only (McVey et al. 1981, Nowakowski et al. 2013. Pašukonis et al. 2013, Pichler et Material and methods al. 2017). All of them displayed a clear tendency to return, Translocation experiments but return rates decreased with distance. In this paper, we study homing behavior in a species of the dendrobatid ge- Fieldwork was carried out between 25 February and 8 April nus Ameerega. 2015 at Panguana Biological Field Station, Region Huánu- Ameerega trivittata (Spix, 1824) is a day-active, wide- co, Peru. It is situated in the Amazon lowland rain forest spread Amazonian lowland taxon (Silverstone 1976). on the lower Río Llullapichis (9°37’ S, 74°56’ W, ca. 260 m Both sexes in this comparatively large poison frog (maxi- above sea level), a tributary to the Río Pachitea (Schlüter mum size 55 mm) perform home range behavior and males 2005). The population of A. trivittata in the surroundings are suggested to defend part of their home ranges as ter- of this field station was already investigated in previous life ritories (Roithmair 1994, Luiz et al. 2015). Home ranges history studies (e.g., Roithmair 1994, Neu et al. 2016). are highly variable in size and can be remarkably large. Ac- A total of 91 individuals were collected for experimen- cording to Neu et al. (2016), the average home range size tal displacement. All were caught after sunset (latest on 21 in males and females is 420.08 ± 148.02 m² and 524.48 ± March) when sleeping on vegetation. Males and females of 87.56 m², respectively. Site fidelity in males is suggested to this species, if at all, only slightly differ in external mor- be related to the presence of resources that are advanta- phology so that when collecting them the sexes are hard geous to reproduction, such as exposed habitat structures to tell apart. In this study we did not determine the sex- (e.g., fallen trees) for emitting vocalizations with the goal es. For clarification, in another study on A. trivittata (Neu to attract females (Roithmair 1994, Aciolo & Neckel- et al. 2016), we distinguished males and females, but this Oliveira 2014). The motivation for female site fidelity re- was done on day-long observations of calling and mating. mains unknown (cf. Neu et al. 2016). Specimens had mean snout-to-vent length (SVL) 40.18 ± Because of the observed site fidelity in A. trivittata we 0.04 mm, ranging from 35.0 to 48.0 mm. Given these data, expected this species to exhibit homing behavior (like oth- we considered all test individuals as adults (cf. Acioli & er poison frogs). The goal of our study was to explore this Neckel-Oliveira 2014, Lötters et al. 2007). Digital pho- in a wild population via experimental translocations. We tographs of the dorsal pattern of each frog were taken us- hypothesize that (1) homing behavior of A. trivittata de- ing a Fuji FinePix S2500HD digital camera. Photographic crease with an increase of translocation distance. Given the data were used for individual recognition from a distance species’ large body size and home range size, we hypoth- of up to 3 m (cf. Silverstone 1976). Captured individuals esize that (2) A. trivittata individuals are able to return to were singly placed in plastic containers with humid sub- 31 Josephine A. Nothacker et al. Table 2. Distance classes used in translocation experiments and df = 15, p = 0.02174). For further analyses, the actual dis- the number of specimens of Ameerega trivittata randomly as- tances (m) were used. signed to each cardinal direction. We calculated a Generalized Linear Model (GLM) with ‘logit link’ for data with binomial distribution, i.e., a lo- Distance class East South West North Total gistic regression model. Homing of individuals was the binary response variable and translocation distance (m) 150 m 6 8 9 8 31 and cardinal direction (ordinal data) of individuals, SVL 600 m 9 13 3 7 32 (mm) and altitudinal difference between capture and re- 900 m 4 5 3 4 16 lease place (m) were considered as potential predictor vari- Total 19 26 15 19 79 ables. The global model was simplified using stepwise AIC (Akaike Information Criterion) selection and model fit was obtained by calculating the area under the receiver op- strate until they were released on the next morning before erating characteristic curve (AUC) of the best fitting mod- 09:00 h. Collection sites were marked with a plastic tag; the el. AUC values range from 0.5 to 1.0 whereas values > 0.7 coordinates and elevation above sea level were taken with a describe “usable”, > 0.8 “good” and ≥ 0.9 “very good” mod- Garmin eTrex 30 GPS receiver. els (Swets 1988). In total, 91 specimens were collected. By flipping a coin, With the goal to identify effects of the three different we randomly decided if a specimen was translocated or re- group size of distance classes and the control group, a leased the other day at the exact collection sites. The latter Fisher’s exact test was conducted. We tested for significant was performed to examine the recapture potential without differences between the numbers of days after which frogs translocation (control group). In accordance with other were recaptured among the three distance classes via a chi- field-based studies on poison frog homing behavior (e.g., square test. Pašukonis et al. 2013, Pichler et al. 2017), for 79 frogs to All statistical analyses were performed in R (R Develop- be translocated, we used (each chosen using random num- ment Core Team 2008: “R: a language and environment for bers) three distance classes and all four cardinal directions statistical computing”; http://www.rproject.org), using the for translocations (Table 2). The GPS receiver was used to packages mass and verification. determine distances and directions in the field; in addition, coordinates and information on elevation above sea level of capture and release points were GPS-recorded. Results For indication of successful homing, subsequent to translocations or release at collection site (i.e., control Homing behavior in A. trivittata was found for all three group), always the same fieldworker searched for the in- distance classes. Thirty-four of the 79 translocated individ- dividuals within the area of their collection sites at least uals (i.e. 43%) were recaptured within the area of their orig- every third day. We allowed a ca. 30 m radius around the inal collection sites, i.e., their assumed home ranges (Ta- original collection site, due the relatively large home ranges ble 3). Seven of the 12 control group frogs (i.e. 58%) were occupied by this species (cf. Roithmair 1994, Neu et al. retraced at their capture sites. There was neither a statisti- 2016). Due to the diurnal activity of A. trivittata, we main- cally significant difference in homing success between the ly conducted searches for frogs during day time. Twen- distance classes, nor between each of them and the control ty-minute-searches were performed. For individuals that group (Fisher’s exact test: p > 0.05). However, an excep- were retraced, this was 1–16 searches (4.34 ± 3.40, N = 41); tion from this was noted between the 150 m and 600 m dis- for those not found, this was 5–14 searches (9.70 ± 2.42, tance classes (Fisher’s exact test: p = 0.014). The time pe- N = 50). riod after which frogs had returned varied markedly (Ta- ble 3) and there was no significant difference among these time periods (chi-square test: p > 0.05). Data analysis The best fitting model (with AUC = 0.7) only included the translocation distance of the individuals as predictor In the field, it was difficult to translocate frogs at the ex- variable. Translocation distance negatively affected A. tri­ act distance. Thus, differences between intended distanc- vittata homing behavior (Z = -3.72, P < 0.001), while the es (i.e., 150, 600 and 900 m) and the actual distances (m) other considered variables had no effect on it. Already at between capture sites and release points (i.e., GPS marks) about 400 m translocation distance, the predicted homing were determined with ArcGIS 10.3 (Esri®). Student’s t-tests of individuals was 50% only (Fig. 1). for paired samples (normally distributed data with homo- geneity of variance was previously confirmed) was con- ducted to identify differences between the intended (i.e. Discussion exact) and the actual translocation distance in the field. Successful homing behavior Results revealed statistically significant differences for the 600 m and 900 m distance (150 m: t = 1.563, df = 30, p = Like other poison frog species (Table 1), A. trivittata dis- 0.1286; 600 m: t = 2.466, df = 30, p = 0.0194; 900 m: 2.561, plays site fidelity (cf. Roithmair 1994, Neu et al. 2016). 32 Homing behavior in Ameerega trivittata Table 3. Information on successful homing behavior Ameerega trivittata for three distance classes studied plus a control group (cf. Table 2). Number of specimens Mean period after which specimens were recaptured Distance class (with percentage of each all translocated specimens) (with range in parentheses) 150 m 23 (74.19%) 9.09 ± 1.7 (2–30) days 600 m 7 (21.88%) 14.14 ± 2.38 (5–21) days 900 m 4 (25.00%) 13.63 ± 1.44 (11–17) days total 34 (43,04%) 12.29 ± 1.61 (2–30) days 0 m (control group) 7 (58.33%) 6.57 ± 1.49 (2–14) days As hypothesized by us, our results clearly indicate that more than the 41 A. trivittata individuals had returned to this species performs homing behavior after experimen- their home ranges. tal translocation, decreasing with translocation distance. Remarkable is the relatively low number of recaptured In fact, homing success was explained in our GLM by specimens that were not displaced (i.e., the control group), translocation distance. Comparing observed return rates especially when compared to other poison frog homing be- to those in other aromobatid and dendrobatid species (Ta- havior studies (e.g., Pašukonis et al. 2013, Pichler et al. ble 1; Fig. 1), A. trivittata behaved largely similar to them. 2017). It appears unlikely to us that all untraced individu- However, as already pointed out by Pichler et al. (2017), als had disappeared from their home ranges in response comparisons are generally hampered due to the absence of to collecting them. On other occasions, we often collected standardized methods. and handled A. trivittata individuals and they seemed un- Actually, we expect that true returns in our study were affected by this (cf. Neu et al. 2016). We argue that rath- even higher than determined and that we had overlooked er we might have overlooked these frogs for other reasons some frogs. Reasons for this include: (i) the probability of such as again a low detection probability and the species’ detection/redetection of individuals remains an imperfect large home size already mentioned. issue (cf. Holenweg Peter et al. 2001, Schmidt 2003; Regarding the duration after which individuals could be Wagner et al. 2011); (ii) home ranges in A. trivittata can retraced near their collection sites, i.e., after homing, varied be considerably large (Roithmair 1994, Luiz et al. 2015, greatly and did not depend on distance. This illustrates that Neu et al. 2016) so that a priori retracing specimens re- even after displacement over long distances, A. trivittata is mains a hard undertaking; (iii) the searches for specimens able to return rapidly. This is an observation shared by oth- that we considered as ‘not retraced’ was (for logistical rea- er aromobatid and dendrobatid species (McVey et al. 1981, sons) not standardized, leaving the possibility that actually Nowakowski et al. 2013, Pašukonis et al. 2013, Pichler Figure 1. Predicted homing success of Ameerega trivittata individuals translocated at different distance classes (0, 150, 600, 900 m). 33 Josephine A. Nothacker et al. et al. 2017). While this certifies a highly developed sense large distance has not been examined yet in any other poi- of orientation, the navigation mechanisms in poison frogs son frog taxon – and its confirmation might generally be remain unknown, as in general amphibians can make use considered remarkable in small- to mid-size anurans (cf. of various modes (compass orientation, beaconing, pilot- Holenweg Peter et al. 2001). Return rates in Allobates ta­ age, path integration, true navigation; Sinsch 1990, 2006). lamancae and Oophaga pumilio (McVey et al. 1981, Nowa- Allobates femoralis, after experimental displacement and kowski et al. 2013, Pichler et al. 2017) suggest that cer- radio-tracking, moved straight paths to their home rang- tainly these frogs are unable to find back to their home es (Pašukonis et al. 2014 a, b). The authors suggested that ranges when displaced for such a large distance (Fig. 2). spatial learning and orientation on landmarks could be the Only in Allobates femoralis, frogs were experimentally dis- orientation mechanism in this poison frog species. Such placed for 800 m – but none returned (Pašukonis et al. landmarks remain unknown but could be represented by 2013). Thus, our hypothesis is corroborated. fallen trees, water bodies etc. Ameerega trivittata homing One rationale for our hypothesis was that Ameerega tri­ behavior in our study was neither influenced by altitudinal vittata is considerably larger than the other studied species differences between collection site versus release point nor (cf. Table 1). For instance, Allobates femoralis, the largest of by cardinal direction, perhaps suggesting that frogs in gen- them, has only about two thirds of the SVL of A. trivitta­ eral know the terrain and use landmarks for orientation. ta (Silverstone 1976, Lötters et al. 2007). As body size As with regard to poison frogs in general, beside aro- might be related to interspecific differences, we hypothe- mobatids, not only dendrobatids in the genus Oophaga but sized in addition that SVL among the replaced A. trivittata also in the distantly related (cf. Grant et al. 2006) genus specimens had an effect on homing success, i.e., successful Ameerega are able to conduct homing. homing increased with larger body size. However, SVL was negated in our GLM as a potential predictor variable with regard to translocation distance class, so that our hypoth- Homing over long distance and the role of body size esis has to be rejected. Also, Pichler et al. (2017) proposed that differences in body size between Allobates femora lis One of our hypotheses was that A. trivittata is able to re- and A. talamancae did not have an effect on homing per- turn over longer distances than the four other poison frog formance. species that have been studied so far. Our data demonstrate As another reason, we assumed that the large home that A. trivittata specimens are able to find back to their ranges recorded in our target species could explain long home ranges after displacement for up to 900 m. Such a distance homing behavior. Although, we have not tested Figure 2. Successful homing in four poison frog species after experimental translocation (based on data in Tables 1 and 3). Size of iconized frogs is smaller than natural size but proportions to each other resembles natural conditions. Note than one species was translocated for 800 m with return rate zero. Photographs used are available via the worldwide web and were taken by B. Wilson, B. Gratwicke, J. P. Lawrence, Rana Verde. 34 Homing behavior in Ameerega trivittata this further, there might be some rationale behind this. be interesting to repeat this study with more standardized Poison frogs are suggested to have ‘good knowledge’ about numbers of individuals per translocation class and perhaps the general area they live in by integration of learned land- even longer translocation distances. In this way, it can also marks (see above). In many species, this spatial learning is be assessed if homing success decreases gradually, which especially essential, as in many species, tadpoles are car- cannot be stated with certainty from our results (Fig. 2). ried to water bodies which can be fairly distant from the place of terrestrial egg deposition (Pröhl 2005, Ringler et al. 2013, Pašukonis et al. 2014a). This was confirmed for Acknowledgments A. trivittata in which home ranges often do not contain We are most grateful to people from the Panguana Biological water bodies suitable for larval development (Roithmair Field Station, especially to Juliane Diller and Moro and Nery 1994, Luiz et al. 2015). This might promote frogs’ orienta- Modena. Permissions were kindly made available by the Peru- tion in geographic space. In this context, Pašukonis et al. vian Ministry of Agriculture (No. 0196-2014-MINAGRI-DGFFS/ (2013) suggested that in A. femoralis, the decrease in hom- DGEFFS) and the Servicio Nacional Forestal y de Fauna Silves- ing performance was correlated with the maximum tad- tre (No. 0050-2015-SERFOR-DGGSPFFS). Our research in Peru pole transport distance. In the majority of poison frogs, also benefited from a cooperation agreement between Trier Uni- larvae are carried by males only – including A. trivittata versity, Department of Biogeography, and Centro de Ornitología (Schlüter 2005, Lötters et al. 2007). In this species, it is y Biodiversidad (CORBIDI) in Lima. Grants were kindly made available to J.A.N. and C.P.N. by the German Academic Exchange suggested further that males do not carry all tadpoles from Service (DAAD) and to M.M. by the Konrad-Adenauer-Stif- one clutch at once (Roithmair 1994, Aciolo & Neckel- tung (KAS). We are also thankful to an anonymous reviewer of Oliveira 2014), which poses another strong argument the original manuscript of this paper who made valuable com- that males are reliable homers. Altogether these are good ments. arguments for male homing over long distances in our fo- cal species. In our study on A. trivittata, sexes could not be distin- References guished. In concert with the concept of ‘good knowledge’ due to repeated larval transport, we will not rule out that Aciolo, E. C. S. & S. Neckel-Oliveira (2014): Reproductive biology of Ameerega trivittata (Anura: Dendrobatidae) in an A. trivittata specimens performing long-distance homing area of terra firme forest in eastern Amazonia. – Acta Ama- in our study exclusively were males. On the other hand, zonica, 44: 473–480. females are less well studied and even the motivation for Bogert, C. M. (1947): A field study of homing in the Carolina them to maintain home ranges remains unknown. This toad. – American Museum Novitates, 1355: 1–24. illustrates a noteworthy research deficiency, especially Burt, W. H. (1943): Territoriality and home range concepts as ap- since females possess home ranges equal in size to those plied to mammals. – Journal of Mammalogy, 24: 346–352. of males (Neu et al. 2016). In the case of O. pumilio it has been demonstrated that both sexes show solid homing Dole, J. W. (1968): Homing in leopard frogs Rana pipiens. – Ecol- ogy, 49: 386–399. behavior (McVey et al. 1981). In this dendrobatid, the fe- male is strongly involved in brood care (i.e., the contrary to Duellman, W. E. & L. Trueb (1986). Biology of amphibians. – McGraw-Hill, New York, NY. A. tri vittata; Aciolo & Neckel-Oliveira 2014). Oophaga pumilio males might perform homing as calling sites for Grant, T., D. R. Frost, J. P. Caldwell, R. Gagliardo, C. F. mate attraction are beneficial, which is likewise not ap- B. Haddad, P. J. Kok, D. B. Means, B. P. Noonan, W. E. Schargel & W. C. Wheeler (2006): Phylogenetic systemat- plicable to A. trivittata females. In addition, Staudt et al. ics of dart-poison frogs and their relatives (Anura: Athesphat- (2010) indicated that in O. pumilio food density (formicine anura: Dendrobatidae). – Bulletin of the American Museum ants) was higher in home range core areas than outside. It of Natural History, 299: 1–262. remains to be studied if this is a motivation for homing in Holenweg Peter, A.-K., H.-U. Reyer & G. Abt Tietje (2001): poison frogs. Homing behavior of Rana lessonae, R. ridibunda and their hybridogenetic associate R. esculenta after experimental dis- placement. – Amphibia-Reptilia, 22: 475–480. Conclusions and outlook Jameson, D. L. (1957): Population structure and homing respons- es in the Pacific tree frog. – Copeia, 1957: 221–228. We have shown that A. trivittata is an additional poison Kahn, T. R., E. La Marca, S. Lötters, J. L. Brown, E. Twomey frog that performs homing behavior and that this species & A. Amézquita (2016): Aposematic poison frogs (Dendro- is able to return from 900 m after experimental transloca- batidae) of the Andean countries: Bolivia, Colombia, Ecua- tion. This is a remarkably long distance, not only for the dor, Perú and Venezuela. Conservation International Tropical group of Neotropical poison frogs but small- to mid-sized Field Guide Series. – Conservation International, Arlington, anurans in general. While the relatively large SVL of the VA. study species may play a role for long distance homing, Laver, P. N. & M. J. Kelly (2008): A critical review of home range within-species variation of SVL did not explain homing studies. – The Journal of Wildlife Management, 72: 290–298. success. It remains to be studied if home range size or sex Luiz, L. F., F. A. Leon Contrera & S. Neckel-Oliveira (2015): play a role in A. trivittata homing behavior. Also, it might Diet and tadpole transportation in the poison dart frog 35 Josephine A. Nothacker et al. Ameerega trivittata (Anura, Dendrobatidae). – Herpetological Schlüter, A. (2005): Amphibien an einem Stillgewässer in Journal, 25: 187–190. Peru. Mit einer illustrierten Checkliste der Amphibien und Reptilien des unteren Río Llullapichis. – Edition Chimaira, Lötters, S., K.-H. Jungfer, F.-W. Henkel & W. Schmidt Frankfurt/M. (2007): Pfeilgiftfrösche. Biologie, Haltung, Arten. – Chimaira, Frankfurt/M.. Silverstone, P. A. (1976): A revision of the poison-arrow frogs of the genus Phyllobates Bibron in Sagra. – Natural History McVey, M. E., R. G. Zahary, D. Perry & J. MacDougal (1981): Museum of Los Angeles County Science Bulletin, 27: 1–53. Territoriality and homing behavior in the poison dart frog (Dendrobates pumilio). – Copeia, 1981: 1-8. Sinsch, U. (1990). Migration and orientation in anuran amphi- bians. – Ethology, Ecology and Evolution, 2: 65–79. Neu, C. P., S. S. Bisanz, J. A. Nothacker, M. Mayer & S. Löt- ters (2016): Male and female home range behavior in the Ne- Sinsch, U. (2006): Orientation and navigation in Amphibia. – otropical poison frog Ameerega trivittata (Anura, Dendrobati- Marine and Freshwater Behaviour and Physiology, 39: 65–71. dae) over two consecutive years. – South American Journal of Sinsch, U. (2014): Movement ecology of amphibians: from in- Herpetology, 11: 149–156. dividual migratory behaviour to spatially structured popu- Nowakowski, A. J., B. Otero Jiménez, M. Allen, M. Diaz- lations in heterogeneous landscapes. – Canadian Journal of Escobar & M. A. Donnelly (2013): Landscape resistance to Zoo logy, 92: 491–502. movement of the poison frog, Oophaga pumilio, in the low- Staudt, K., S. Meneses Ospina, D. Mebs & H. Pröhl (2010): lands of northeastern Costa Rica. – Animal Conservation, 16: Foraging behaviour and territoriality of the strawberry poison 188–197. frog (Oophaga pumilio) in dependence of the presence of ants. Papi, F. (1992): Animal homing. – Chapman & Hall, London. – Amphibia-Reptilia, 31: 217–227. Pašukonis, A., M. Ringler, H. B. Brandl, R. Mangione, E. Swets, K. (1988): Measuring the accuracy of diagnostic systems. Ringler & W. Hödl (2013): The homing frog: High homing – Science, 240: 1285–1293. performance in a territorial dendrobatid frog Allobates femo­ Twitty, V., D. Grant & O. Anderson (1964): Long distance ralis (Dendrobatidae). – Ethology, 119: 762–768. homing in the newt Taricha rivularis. – Proceedings of the Na- Pašukonis, A., I. Warrington, M. Ringler & W. Hödl (2014a): tional Academy of Sciences of the United States of America, Poison frogs rely on experience to find the way home in the 51: 51–58. rainforest. – Biology Letters, 10: 20140642. Twomey, E., V. Morales & K. Summers (2008): Evaluating con- Pašukonis, A., M. C. Loretto, L. Landler, M. Ringler & W. dition-specific and asymmetric competition in a species-dis- Hödl (2014b): Homing trajectories and initial orientation in tribution context. – Oikos, 117: 1175–1184. a Neotropical territorial frog, Allobates femoralis (Dendrobati- Wagner, N., J. Pellet, S. Lötters, B. R. Schmidt & T. Schmitt dae). – Frontiers in Zoology, 11: 29. (2011): The superpopulation approach for estimating the pop- ulation size of ‘prolonged’ breeding amphibians: Examples Pichler, C., S. Weinlein, L. Kopeinig & A. Pašukonis (2017): from Europe. – Amphibia-Reptilia, 32: 323–332. Homing performance in a territorial dendrobatid frog, Allo­ bates talamancae. – Salamandra, 53: 309–313. Walcott, C. (1996): Pigeon homing: observations, experiments and confusions. – The Journal of Experimental Biology, 199: Pittman, S. J. & C. A. McAlpine (2003): Movements of marine 21–27. fish and decapod crustaceans: process, theory and application. – Advances in Marine Biology, 44: 205–294. Wells, K. D. (2007): The ecology and behavior of amphibians. – University of Chicago Press, Chicago, IL. Pröhl, H. (2005): Territorial behavior in dendrobatid frogs. – Journal of Herpetology, 39: 354–365. Pröhl, H. & W. Hödl (1999): Parental investment, potential re- productive rates, and mating system in the Strawberry dart- poison frog, Dendrobates pumilio. – Behavioral Ecology and Sociobiology, 46: 215–220. Quinn, T. P. & A. H. Dittman (1990). Pacific salmon migra- tions and homing – mechanisms and adaptive significance. – Trends in Ecology & Evolution, 5: 174–177. Ringler, M., E. Ursprung & W. Hödl (2009): Site fidelity and patterns of short- and long-term movement in the brilliant- thighed poison frog Allobates femoralis (Aromobatidae). – Be- havioral Ecology and Sociobiology, 63: 1281–1293. Ringler, E., A. Pašukonis, W. Hödl & M. Ringler (2013): Tad- pole transport logistics in a Neotropical poison frog: indica- tions for strategic planning and adaptive plasticity in anuran parental care. – Frontiers in Zoology, 10: 67. Roithmair, M. E. (1994): Male territoriality and female mate selection in the dart-poison frog Epipedobates trivittatus (Dendro batidae, Anura). – Copeia, 1994: 107–115. Schmidt, B. R. (2003): Count data, detection probabilities, and the demography, dynamics, distribution, and decline of am- phibians. – Comptes Rendus Biologies, 326: 119–124. 36