1 Female-female aggression is linked to food defence in a poison frog
2
3
4
5
6
7 Heike Pröhl1*§, Martin G. Scherm2§, Santiago Meneses3,4, Corinna E. Dreher1, Ivonne
8 Meuche1, Ariel Rodríguez1
9
10
11 12 1 Institute of Zoology, University of Veterinary Medicine, Bünteweg 17, 30559 Hannover, Germany,
13 [email protected], [email protected], [email protected],
15 2 Institute of Diabetes Research, Helmholtz Zentrum München, German Research Center for
16 Environmental Health (GmbH), Heidemannstraße 1, 80939 München, Germany,
18 3 Smithsonian Tropical Research Institute, Luis Clement Ave., Bldg. 401 Tupper, Balboa Ancon,
19 Panama, Republic of Panama, [email protected]
20 4Department of Biological Sciences, George Washington University, Washington D.C., 20052, USA. 21
22 *Correspondence: Heike Pröhl
23 § These authors contributed equally to this work.
24
25 Total number of words (including references): 6728
1
26 Abstract
27 Habitat occupancy by territorial animals is expected to depend on the distribution of critical
28 resources. Knowledge on female territoriality is scarce but it has been suggested as a
29 mechanism to defend limited resources for reproduction. A previous study showed female
30 intrasexual aggression to be connected to territorial behaviour in the strawberry poison frog
31 Oophaga pumilio, a diurnal aposematic species with complex maternal care. Here we
32 investigate the link between spatial distribution of resources important for reproduction and
33 female distribution and behaviour. We observed focal females in their natural habitat in
34 Costa Rica, and recorded the distribution of ecological predictor variables in a grid system.
35 The data were used for calculating home range and territory sizes and for connecting female
36 habitat use to the distribution of potential resources by computing spatial habitat occupancy
37 models. Even though we found females to occupy large home ranges, they were highly
38 aggressive towards other females only inside a small part of their home range, here termed
39 core area. Among the ecological factors the sustained abundance of ants (as the main food
40 item of the frogs), the presence of leaf litter, and suitable rearing sites for tadpoles predicted
41 female site occupancy patterns. The number of ants was twice as high in the core areas
42 compared to the rest of the female home ranges. Our results suggest that female spacing
43 behaviour is principally driven by the spatial distribution of its main food resource, but that
44 hiding places (leaf litter) and tadpole rearing sites also play a role. The defence of places with
45 high sustained abundance of ants could be relevant for this prolonged-breeding species as
46 egg production and maternal care are energetically highly demanding. Regarding the link
47 between resource defence and maternal care, the reproductive strategy of female
48 strawberry poison frogs resembles that of the females of small mammals comprising same-
49 sex-competition for food and high investment in producing and rearing young.
2
50
51
52 KEYWORDS
53 feeding behaviour, frogs, habitat occupancy, resource distribution, spatial autocorrelation,
54 territoriality
55 1 INTRODUCTION
56 In natural populations individuals compete for limited resources to maximize their fitness.
57 Competitive strategies differ between the sexes, because males and females depend on
58 distinct resources for survival and reproduction, and include a broad repertoire of
59 behavioural components, e.g. aggressive encounters and territoriality (Clutton-Brock &
60 Huchard, 2013). Territoriality, i.e. the defence of a limited area, ensures the territory holder
61 prior access to critical resources like food, water, mates, hiding or nesting places (Wells,
62 1977; Pröhl, 2005; Alcock 2006).Territoriality is associated with time and energetic costs as
63 well as increased conspicuousness to predators and risk of injury (Carpenter & Mcmillen
64 1976; Marler, Walsberg, White & Moore, 1995; Alcock 2006). However, the benefits of
65 territoriality must outweigh the costs (Brown, 1964; Schoener, 1983; Adams, 2001) and the
66 resources must be defensible and limited in time or space (Wilson, 1975).
67 In numerous vertebrate species territoriality is well studied, especially for males (review
68 Wells 1977; Davies, 1991; Alcock, 2005). Males often compete for females and therefore
69 defend resources that females need for reproduction such as breeding sites (Emlen & Oring,
70 1977). When some males are stronger than others they can gain priority access to clumped
3
71 resources and thereby increase their probability for mating with multiple females (mammals:
72 Clutton-Brock, 1989; snakes: Webb, Scott, Whiting & Shine, 2015).
73 In general, female intrasexual competition and territoriality has attracted less scientific
74 attention than male competition (Clutton-Brock, 2007) but is particularly well understood in
75 small mammals (Ostfeld, 1990). Some discussion about the purpose, i.e. the resources
76 defended by females, has stimulated this research area (Wolff & Peterson, 1998). Female
77 territorial behaviour has typically been suggested as a mechanisms to defend food resources
78 (Ostfeld, 1990; Rosvall, 2011), however later studies indicated that females directly defend
79 their young in nest sites with the ultimate purpose to reduce the risk of infanticide by
80 conspecifics (Bonnatto, 2017; Steinmann, Priotto & Polop, 2009).
81 Because of the high importance of resources for reproductive success the distribution of
82 resources should influence the distribution of the sexes, while the resource distribution itself
83 is affected by other ecological factors. The link between ecological factors and distribution of
84 a species can be analysed via habitat occupancy models. These models normally analyse the
85 distribution of one or several species over larger landscapes (e.g. Joseph, Preston & Johnson,
86 2016). The questions addressed often involve which ecological factors influence the
87 presence/habitat selection and reproduction of species in a patchy landscape. Hereby spatial
88 autocorrelation often plays a role, because the proximity to other (sub-) populations or
89 occupied habitat produces emigrants and has an effect on habitat choice and therefore
90 occupancy of this habitat (e.g. Miró, O’Brien, Hall & Jehle, 2017). Here we use a habitat
91 occupancy model to relate resource distribution to the distribution and behaviour of females
92 in a frog species.
93 In frogs, defence of resources in territories occurs mainly in species with longer breeding
94 periods. Territoriality was mainly observed in males of aquatic, arboreal as well as terrestrial
4
95 egg layers (review: Wells, 2007). The resources the males defend are mostly those that
96 females need for successful reproduction such as oviposition sites. These are aggressively
97 protected against other males in ponds or streams, or other sheltered places. For the poison
98 frog family Dendrobatidae complex resource-based mating systems have been described
99 (review: Wells, 2007; Pröhl, 2005). Males often defend all-purpose territories for longer time
100 periods containing feeding, shelter, oviposition, calling and breeding sites (e.g. Roithmair,
101 1992; Poelmann & Dicke, 2008). Within anurans, female territoriality was only reported for
102 this family. Female territoriality seems to be mainly connected to defence of food and retreat
103 sites (e.g. Mannophryne trinitatis (Test, 1954; Wells, 1980a); Colostethus panamansis (Wells,
104 1980b)).
105 The strawberry poison frog (Oophaga pumilio) is a diurnal species which inhabits
106 Caribbean lowland rainforests in Nicaragua, Costa Rica and Western Panama (Myers & Daly
107 1983; Savage, 2002). Both sexes are polygamous (Bunnell, 1973; Pröhl & Hödl 1999; Meuche,
108 Linsenmair & Pröhl, 2011) and their ecology and complex reproductive behaviour are well
109 studied (e.g. Rudh, Rogell, Håstad & Qvarnström, 2013). Since females provide eggs for
110 mating and unfertilized eggs as a food source for their tadpoles, female reproduction and
111 especially maternal parental care are associated with higher energetic and temporal costs
112 than reproduction in males which only provide sperm and moisten the eggs (Limerick, 1980;
113 Weygoldt 1980; Pröhl & Hödl 1999). The diet of the strawberry poison frog consists mainly of
114 ants and mites (Donnelly, 1991), from which the frogs accumulate toxic alkaloids into their
115 skin (Saporito, Donnelly, Spande & Garraffo, 2012). Together with their aposematic
116 coloration, toxicity provides an effective protection from potential predators (Daly, Myers &
117 Whittaker, 1987; Saporito, Zuercher, Roberts, Gerow & Donnelly, 2007).
5
118 The main purposes for aggressive male territorial behaviour in O. pumilio seem to be the
119 availability of calling sites, access to females, which are the limiting sex in this species, and
120 place for courtship and oviposition (Pröhl, 2003; Pröhl & Berke 2001; Meuche, Linsenmair &
121 Pröhl 2012). Males also spend more time feeding in the core areas of their territories, but no
122 more insect prey has been found in these areas than outside of their territories (Staudt,
123 Meneses Ospina & Pröhl, 2010). Recent studies showed that also the females of O. pumilio
124 exhibit aggressive behaviour (Haase & Pröhl, 2002) particularly in the intensely used core
125 areas of their home ranges (Meuche, Linsenmair & Pröhl, 2011).There was no evidence that
126 females compete for males or that oviposition or tadpole-rearing sites play a role for female
127 territoriality (Pröhl & Berke, 2001; Meuche, Linsenmair & Pröhl 2011). However, female
128 home ranges are spatially closely associated with tadpole rearing sites and overlap in the
129 vicinity of these (Pröhl & Berke, 2001). Because of the much higher energetic costs of
130 maternal care compared to male care and the importance of access to high quality food,
131 necessary for toxin sequestration (Staudt, Meneses Ospina, Pröhl, 2010), we assume female
132 habitat occupancy and aggressive behaviour to be linked to the exploitation of food
133 resources.
134 Here, we use the strawberry poison frog as a model species for a detailed analysis of the
135 connection between the distribution of resources in the habitat, and how this in turn affects
136 the occupancy and spacing pattern as well as the territorial and feeding behaviour of the
137 females. Because our main hypothesis is that females defend areas rich in food resources, we
138 test whether females aggressively defend areas with high prey abundance and whether they
139 mainly feed in these areas. Since water-filled leaf axils are used as tadpole rearing sites and
140 therefore are important reproductive resources we also test whether their presence
141 influences female distribution. To our knowledge this is one of the first studies investigating
6
142 female territorial behaviour in association with spatial abundance of highly important but
143 potentially limited resources in frogs. Providing new data on female territoriality enables us
144 concentrating a part of our discussion on the difference between male and female
145 territoriality
146
147
148
149 2 METHODS
150 2.1 Study area
151 This field study was conducted from April to June 2012 in a Caribbean lowland rainforest in
152 the Biological Reserve Hitoy Cerere in Costa Rica (N: 09°40‘, E: 83°05’). The study area was
153 located at approximately 100m above sea level in an abandoned banana plantation with
154 transition to a young secondary forest and measured 400 m2 (20 x 20 m). It was divided into
155 grids of 1 m2 using nylon strings at a height of 2 m above the ground to facilitate behavioural
156 observations and measurement of ecological variables. All behavioural observations are in
157 compliance with the Guidelines for the use of animals (2012) and have been approved by the
158 Costa Rican authorities (SINAC: ACLAC-PI-005-2012).
159 2.2 Behavioural observations
160 All behavioural observations took place during the breeding season of O. pumilio. At the
161 beginning of the study period as many adult frogs as possible (N = 19 females, N = 15 males)
162 were caught within the study area during three consecutive days. The sex of the animals was
163 determined by the presence of a dark vocal sac at the throat of the males. Furthermore the
7
164 length (SVL in mm) and weight (in g) was measured using a manual calliper and a precision
165 balance (WEDO Precision 500). To facilitate the recognition of individual frogs in the study
166 area photos were taken from the front, lateral, ventral and dorsal body parts. Greyish or
167 black spots and pattern on the red background allowed recognition of individuals during
168 observations.
169 Eleven females were observed by the same person (M. Scherm) for five days each using
170 focal animal sampling. The focal female was observed continuously from 6:00 until 12:00 h
171 while the following parameters were recorded every minute on a check sheet: 1) Position of
172 the frog in the grid system of the study area at the beginning of each one minute interval. 2)
173 Feeding: during feeding the frog turns towards the prey and captures it with a stroke of the
174 tongue (M. Scherm, personal observation). Because it is often not possible to determine if
175 the feeding attempt was successful, every feeding attempt was recorded as a feeding event.
176 3) Agonistic female-female interactions: A female was considered aggressive when it directly
177 approached or followed the other female or initiated any kind of physical contact. A
178 submissive female tried to escape the conflict by moving away from the aggressive female,
179 hiding in the substrate or remaining motionless. We also intended to monitor agonistic
180 interactions between females and males, but there were none.
181 To minimize the impact of the observer on the animals, fast movements were avoided
182 and a distance of at least 2 m to the focal animal was maintained. Since no female fled or
183 showed any change in behaviour during the observation, we consider the influence of the
184 observer as negligible.
185 2.3 Habitat and vegetation
186 Several ecological factors were recorded for every 1 m2 grid: 1) Number of ants captured in a
187 trap (details see below). 2) Leaf litter: we determined if more or less than 50% of the area of 8
188 the grid were covered with dead plant material. 3) Debris cluster: the presence of clusters of
189 leaves or wood, exceeding the normal ground coverage (more than 20 cm deep) which could
190 offer shelter for the frogs (M. Scherm & I. Meuche, personal observation). 4) Plants
191 harbouring water filled leaf axils for tadpole rearing. In the study area these were mainly
192 Musa and Heliconia species.
193 2.4 Ant collection
194 For prey collection we used simple pitfall traps, consisting of buried plastic cups, filled up to a
195 height of approximately 3 cm with a mixture of water and a small amount of dishwashing
196 liquid (B. Hölldobler, personal communication). In each grid one trap covered with a lid was
197 placed into the ground with the edge directly at the substrate in the centre of the grid,
198 except in grids where the centre was blocked by trees or other vegetation. The traps were
199 activated at 6:00 am by removing the lid and deactivated after six hours at 12:00 noon. The
200 traps were covered in the same order as they were activated to ensure a high accuracy for
201 the activation time. In the following the ant specimens in the traps were counted, recorded
202 and transferred to vials filled with 70% ethanol. Altogether the traps were activated four
203 times in April and May 2012 separated by 7-8 days intervals. . Ants with a body length larger
204 than 1 cm (mainly Pachycondyla spec., Ectatomma spec. and Atta cephalotes) were excluded
205 from the analysis because they were not eaten by any of the focal females but provoked
206 escape reactions (M. Scherm, personal observation). In an earlier study ants were identified
207 as main food item in O. pumilio especially for the females (Donnelly, 1991), and therefore we
208 expected to mainly find ants in the traps. Our expectation was fulfilled, yet we only found
209 ants, but no other insects in the traps (see results).
210 2.5 Classification of ants
9
211 All ant specimens were placed in a petri dish filled with ethanol (75%). One specimen from
212 each genus and morphospecies that had the legs spread and directed ventrally was selected.
213 The selected specimens were dried in absorbent paper. After mounting, the specimens were
214 placed in a dryer for at least 14 hours and stored away from light, extreme temperatures,
215 humidity and insect pests to avoid deterioration.
216 For the identification of ant genera the keys from Bolton (1994) and Hölldobler & Wilson
217 (1990) were used. The genera Brachymyrmex and Paratrechina were identified to species or
218 morphospecies using John T. Longino’s web tools at http://academic.evergreen.edu/-
219 projects/ants/AntsofCostaRica.html.
220 We calculated the mean ant abundance for each cell and estimated a sustained
221 abundance index (SAI) as the product of mean abundance and the frequency of non-zero ant
222 counts. This index down weights the mean abundance values of those cells where no ants
223 were observed in one or more survey days and should capture the relative importance of
224 cells as food sources for frogs. This index behaves better than other alternatives using the
225 standard deviation, or other central tendency metrics, which end up giving equal values to
226 cells with similar stability regardless of their low or high mean counts. The density of ants for
227 the core areas and the rest of the home range were determined by averaging the number of
228 ants of all grids falling into the core area or the home range. The average number of feeding
229 attempts per grid for home ranges and core areas were calculated in the same manner.
230
231 2.6 Data analysis
232 The position data of the focal females were transformed into x- and y-coordinates where
233 the individual coordinates correspond to the centre of the grids. To eliminate
10
234 pseudoreplication, position data of every fifth observation minute were used (Meuche,
235 Linsenmair & Pröhl, 2011). For calculating home ranges and core areas we applied the
236 program ArcView GIS 3.3 (Environmental Systems Research Institute, ESRI), as well as the
237 plugins Spatial Analyst 2.0a (ESRI), Cad2Shape 3.0 and Home Range Extension (Rodgers &
238 Carr, 1998). We used the adaptive-kernel method (Worton, 1989) for estimating the home
239 range sizes by the density function of 95% of the recorded position data. The density
240 function of 50% of the position data was calculated and defined as the core area of the home
241 range (Pröhl & Berke, 2001; Meuche, Linsenmair & Pröhl, 2011). Based on SVL and weight
242 the condition of the frogs was assessed following Murphy (1994) and Jakob et al. (1996). We
243 calculated Product -Moment correlations between the SVL, weight and condition and home
244 range and core area size.
245 We used a spatial generalized mixed effect model to estimate the relative influence of
246 habitat predictors (ants SAI, debris clusters, leaf litter and rearing sites) on the habitat
247 occupancy by females while accounting for the random effect of females and the spatial
248 relationship of the data. Spatial regression analyses were performed in R (R Core Team, 2016)
249 with the R-package spaMM, which fits the models described in Rousset and Ferdy (2014). We
250 implemented a binomial generalized mixed effect model in which the proportion of time a
251 female spent on a given grid cell (minutes in cell / total minutes - minutes in cell) was
252 regressed against the four habitat predictors (fixed factors), while accounting for spatial
253 relationships and individual females as a random factors. We evaluate the statistical
254 significance of the fixed predictors by calculating their 95% confidence intervals and by
255 comparing alternative models with and without each predictor with a likelihood ratio test (
256 alpha = 0.05).
11
257 We present boxplots for ecological and behavioural differences between core areas and
258 the rest of the home range. We hypothesized the abundance of prey (ants), presence of leaf
259 litter, presence of debris clusters and tadpole rearing sites to be higher in the core areas than
260 in surrounding home ranges. Because we expected a higher abundance of resources in the
261 core areas of the females we also expected them to feed more and being more aggressive
262 here than in the rest of the home range.
263 3 RESULTS
264 3.1 Size of home ranges and core areas
265 The sizes of the home ranges of the focal females (n = 11) ranged from 4.8m2 to 71.1m2
266 (median, interquartile range: 54.8m2, 31.2m2). The size of the core areas ranged from 0.41m²
267 to 9.9m² (median, interquartile range: 6.7m2, 8.5m2). On average the core areas accounted
268 for 12.0% of the total area of the home ranges. There was no significant correlation between
269 the size of the home ranges or the core areas and SVL (snout-vent length: mean = 2.2cm, SD
270 = 0.12), weight (mean = 1.9g, SD = 0.18) or condition (mean = 0.001, SD = 0.08) of the focal
271 females (all P > 0.05). However there was a trend of heavier females using larger home
272 ranges (r = 0.55, P = 0.08).
273
274 3.2 Habitat occupancy
275 The analyses of home range/core area calculation and the habitat occupancy model are
276 different methods for analyzing female spacing behaviour. However, the location of the
277 female core areas corresponds to the grids which were most occupied by the females (Figure
278 1, 2). In Figure 1 and 2 only those grids are shown with at least one observation of a present
279 female. Grids without female presence were excluded from further analyses because we
12
280 cannot exclude that non-focal females occupied these areas. Note that female core areas are
281 exclusive ranges (our data; Meuche, Linsenmair & Pröhl, 2011) making it unlikely that other
282 females not observed by us occupied these areas.
283 Results of the spatial generalized linear model indicated that the sustained abundance of
284 ants and the presence of thick leaf litter and rearing sites are statistically significant
285 predictors of habitat occupancy by females (Table 1). Sustained ant abundance and thick leaf
286 litter showed a statistically highly significant relationship with female occupancy whereas the
287 relationship with the presence of rearing sites was barely significant. Females spent more
288 time in those grid cells with higher values of these three variables while the presence or
289 absence of debris clusters was not statistically associated with female occupancy.
290 3.3 Agonistic interactions
291 Eight of the eleven focal females were involved in a total of 18 agonistic female-female
292 interactions. Three females were involved in one aggressive interaction, three females in
293 two, one female in four and one female in five aggressive interactions. All except one of the
294 interactions took place in the core area of a female home range. In all of these aggressive
295 interactions the resident female initiated and dominated the conflict (Figure 2b).
296 3.4 Ecological and behavioural differences between core areas and home ranges
297 Only ants but no other insects were captured in the traps. In the core areas (median,
298 interquartile range: 5.33, 1.62) we trapped significantly more ants per grid m² than in the
299 remaining home range (median, interquartile range: 2.52, 1.91; Figure 3a). Accordingly most
300 feeding attempts (77%) occurred in the core areas (Figure 3b). In the home ranges, the
301 average number of feeding attempts per m² varied from 0.06 to 17.8 (median, interquartile
302 range: 2.24, 3.48). In the core area, the number of feeding attempts per grid ranged from
13
303 19.5 to 110 and was much higher than in the remaining home range (median, interquartile
304 range: 31.7, 20.7). The percentage of grids with more than 50% ground covered with leaf
305 litter was higher in the core areas (median, interquartile range: 72.7, 13.3) than in the home
306 ranges (median, interquartile range: 51.4, 18.7; Figure 3c). The percentage of grids with
307 debris cluster overlapped largely between both core areas (median, interquartile range 12.5,
308 33.3) and home ranges (median, interquartile range 8.6, 11.2; Figure 3d). The percentage of
309 grids with plants that offer tadpole rearing sites was variable but twice as high in core areas
310 (median, interquartile range: 38.1, 33.3) than in the remaining home range (median,
311 interquartile range: 17.6, 8.7; Figure 3e).
312 4 DISCUSSION
313 Our study combined behavioural observations, capture of prey items in pitfall traps, records
314 of ecological predictor variables with the calculation of the home range sizes and habitat
315 occupancy patterns to investigate the spatial relationship between female territorial and
316 feeding behaviour to the distribution of potential resources. The habitat occupancy of
317 territorial female frogs was mainly determined by the distribution of their main food source:
318 ants. Other important factors for female distribution were the presence of leaf litter and
319 potential tadpole rearing sites. While ants provide the female with important energy, leaf
320 litter might offer shelter and humidity to the frogs and maybe ants. Tadpole rearing sites
321 were more loosely connected to female distribution but obviously females visit them to feed
322 their tadpoles. Females defended only a small part of the larger home range which
323 corresponds to the core area or 50% kernel density function. Behavioural observations
324 revealed that most feeding events and almost all aggressive interactions with other females
325 took place here. The defence of food might be particularly important in females of species
14
326 displaying high mating activities and energy demanding brood care. Besides reproduction,
327 the consumption of ants and other small insects is critical to these aposematic frogs since
328 they obtain toxic alkaloids from their insect prey for subsequent storage in their skin glands.
329 In conclusion our data confirm the hypothesis that female frogs defend rewarding feeding
330 sites.
331 4.1 Territoriality, resource use and occupancy patterns in females
332 Our results also support the ecologically more general hypothesis, initially developed by
333 Emlen & Oring (1977), that the evolution of territoriality is driven by the spatial distribution
334 of resources and that these resources must be economically defensible. Female strawberry
335 poison frogs defend food resources against other females and are able to successfully repel
336 conspecific females which try to invade their territories. While data on female territoriality
337 are scarce for amphibian species, territoriality is better studied in mammals. In this group
338 some studies demonstrated that females defend access to food (Ostfeld, 1990) since
339 maternal investment is linked to high demands for energy during pregnancy and lactation.
340 The behaviour of female strawberry poison frogs mostly coincides with the strategy of
341 female mammals defending food. Like in mammals our females heavily invest in maternal
342 care. Maternal care mainly consists of tadpole provisioning with nutritive eggs which
343 requires high amounts of energy. There is another aspect in which female frogs behave like
344 mammal females: as mammal females during pregnancy and lactation (Rosvall, 2011)
345 strawberry poison frog females are more aggressive towards other females during periods of
346 tadpole feeding (Haase & Pröhl, 2001).
347 Besides insect prey, tadpole rearing sites, i.e. small water filled leaf axils, are critical
348 resources to strawberry poison frog females (Weygoldt, 1980; Donnelly, 1989). After
15
349 hatching of the terrestrial clutch the females transport their tadpoles to these small water
350 bodies; since food for tadpoles is scarce in these tiny pools, female return to their strictly
351 oophagous tadpoles every few days to feed them with nutritive eggs. Since females use
352 several tadpole rearing sites at different places in the forest (Brust, 1990; Pröhl & Hödl,
353 1999) they are not able to stay in close proximity and defend their tadpoles or tadpole
354 rearing sites. Also females are continuously assessing potential rearing sites, probably
355 because these are not always filled with enough water and therefore present no stable and
356 reliable resources for reuse (Brust, 1990). Nevertheless we found more tadpole rearing sites
357 inside the core areas of the females than in the rest of the home ranges. The proximity of
358 rearing sites might help females to save time and energy when there is nedd to feed the
359 tadpoles.
360 While the diversity of parental care strategies is impressive in amphibians, maternal
361 provisioning of the tadpoles with food (mostly unfertilised eggs) is restricted to the genus
362 Oophaga, some other dendrobatid species, and a number of tropical tree frogs (Wells, 2007).
363 To our knowledge provisioning of the young with maternal produced material (e.g. eggs,
364 secretions, nutrients via uterus) does not occur in fish or reptiles, but is the rule in mammals
365 and birds (Alcock, 2005; Kvarnemo, 2010). In frogs and mammals only the females provide
366 food. However in many bird species, parents of both sexes work hard to find food for their
367 brood, therefore live in pairs (social monogamy) and defend a common territory with nest
368 and feeding sites. This difference between mammals and frogs on one hand and birds on the
369 other might be explained by the the high energy needs of endothermous bird young and the
370 need to grow quickly for minimizing time in the nest and predation risk (Alcock 2005).
371 The distribution of female strawberry poison frogs is governed by two independent
372 resources – prey and tadpole rearing sites – and therefore shows that it is not sufficient to
16
373 restrict data collection to single resources. Together with male territoriality (Pröhl & Hödl,
374 1999), our frogs provide a beautiful example for the notion that habitat structure and
375 resource distribution in a species are fundamental for the evolution of its, in this case
376 polygamous, mating system.
377 4.2 Aposematism and diet specialisation
378 Oophaga pumilio is an aposematic species, i.e. the frogs combine their conspicuous
379 coloration with toxic alkaloids in skin glands for avoiding to be eaten by predators. The
380 alkaloids are sequestered from their insect prey, mainly ants and mites (Saporito et al., 2004,
381 2007). In our study area we mainly found myrmicine ants (Table 2). These ants are
382 considered as potential dietary sources for histrionicotoxins (HTX) (Saporito et al. 2007)
383 which are the most frequent toxins in male frogs in Hitoy Cerere (Staudt, Meneses Ospina &
384 Pröhl, 2010). Our study showed ants to be more abundant in the territories of the females,
385 thus ant distribution influences the distribution and territorial behaviour of the females.
386 Even though ants seem to present the most important food items to female frogs (Donnelly,
387 1991) mites also provision the frogs with toxic alkaloids (Takada et al., 2005; Saporito et al.,
388 2007). For unknown reasons we were unable to capture mites with our pitfall traps. Future
389 studies should apply appropriate ecological methods for mite collection (Saporito, Zuercher,
390 Roberts, Gerow & Donelly, 2007) and relate mite distribution to frog distribution will be a
391 nice complement to this study and will help to improve our understanding of how different
392 prey types shape the distribution of their predators.
393 Aposematic animals are supposed to differ in their behaviour, physiology and toxicity
394 from cryptic animals (Pough & Taigen, 1990; Pröhl & Ostrowski, 2011; Santos & Cannatella,
395 2011). Oophaga pumilio is one of the most poisonous species and one with the most
396 complex and costly maternal care among the Dendrobatidae (Daly, Myers & Whittaker, 1987; 17
397 Brust, 1993; Pröhl & Hödl 1999). In this family the cryptic species are ancestral, while the
398 aposematic species are phylogenetically more derived (Santos, Coloma & Cannatella, 2003).
399 Several studies confirmed an evolutionary trend toward the aposematic antipredator
400 strategy from cryptic ancestors in the group and found several co-evolving characters. The
401 aposematic frogs spend more time with active foraging in larger areas, capture more prey
402 items, have a higher aerobic metabolism and narrower dietary niche breath, while
403 specializing their diet mainly on ants, and to a lower degree on mites or termites (Pough &
404 Taigen 1990; Caldwell 1996; Darst, Menéndez-Guerrero, Coloma & Cannatella, 2005). It is
405 possible that territoriality connected to the defence of toxic food is more common in
406 aposematic animals than in more cryptic animals and an integral behavioural element of the
407 aposematic phenotype (see Santos & Cannatella, 2011). How the highly evolved and costly
408 maternal care is correlated to aposematic traits has not been tested for dendrobatid frogs .
409 To our knowledge a possible phenotypic association between territorial behaviour, costly
410 brood care, chemical defence and conspicuous colouration has not been studied in any
411 animal group.
412 4.3 Future directions
413 In summary our study showed a connection between female territorial behaviour and the
414 distribution of an important food resource in a frog species. However in many species the
415 behaviour varies geographically or among different habitat types (e.g. Pröhl, Willink,
416 Hauswaldt, 2013; Foster, 2013). The strawberry poison frog occurs in a variety of habitat
417 types, differing in vegetation, predation regime (Dreher, Cummings & Pröhl, 2015) and
418 probably climate and prey community. Therefore one aim in future research projects should
419 be to verify whether female habitat occupancy and territorial behaviour is influenced by food
18
420 resources in other populations. Furthermore the strawberry poison frog as well as other
421 polymorphic poison frogs, offer the fascinating possibility to find out how the availability and
422 distribution of toxic prey relates to the evolution of the aposematic phenotype.
423
424 ACKNOWLEDGEMENTS
425 We thank the government of Costa Rica (MINAE; Servicio de Parques Nacionales) for permis-
426 sion to conduct research in the Biological Reserve of Hitoy Cerere (MINAE/SINAC permit No.
427 ACLAC-PI-005-2012) and the staff at the Biological Reserve for their support. The German
428 Academic Exchange Service (DAAD) and the German Research Foundation (DFG) supported
429 this work.
19
430 LITERATURE CITED
431 Adams, E.S. (2001). Approaches to the study of territory size and shape. Annual Reviews in Ecology 432 and Systematics, 32, 277-303. https://doi.org/10.1146/annurev.ecolsys.32.081501.114034 433 Adams, E.S. (2016). Territoriality in ants (Hymenoptera: Formicidae): a review. Myrmecological News, 434 23, 101-118. 435 Alcock, J. (2005). Animal Behaviour – An evolutionary approach. (8nd ed.) Sunderland, MA: Sinauer 436 Associates. 437 Baker, R.R. (1983). Insect territoriality. Annual Review of Entomology, 28, 65-89. https://doi.org/ 438 10.1146/annurev.en.28.010183.000433 439 Barlow, GW. (1974). Contrasts in social behaviour between Central American cichlid fishes and coral- 440 reef surgeon fishes. American Zoologist, 14, 9-34. 441 Bonatto, F., Priotto, J., Coda, J., & Steinmann, A.R. (2017). Female intrasexual territoriality and its 442 potential adaptive significance: The Pampean grassland mouse as an ecological model species. 443 Ethology, 123, 230-241. https://doi.org/10.1111/eth.12592 444 Brown, J.L. (1964). The evolution of diversity in avian territorial systems. Wilson Bulletin, 76, 160-169. 445 Brust, D.G. 1990 (unpublished). Maternal brood care by Dendrobates pumilio: a frog that feeds its 446 young. Ph.D. Cornell University. 447 Bunnell, P. (1973). Vocalizations in the territorial behaviour of the frog Dendrobates pumilio. Copeia, 448 1973(2), 277-285. https://doi.org/10.2307/1442966 449 Caldwell, J.P. (1996). The evolution of myrmecophagy and its correlates in poison frogs (Family 450 Dendrobatidae). Journal of Zoology Lond.,240, 75-101. https://doi.org/10.1111/j.1469- 451 7998.1996.tb05487.x 452 Carpenter, F.L., & Mcmillen, R.E. (1976). Threshold model of feeding territoriality and test with a 453 Hawaiian honeycreeper. Science, 194, 639–642. https://doi.org/10.1126/science.194.4265.639 454 Carrillo, J. & González-Dávila, E. (2013). Aggressive behaviour and nest-site defence during the 455 breeding season in an island kestrel population. Journal of Ethology, 31, 211. https://doi.org/ 456 10.1007/s10164-013-0369-x 457 Clutton-Brock, T.H. (1989). Mammalian mating systems. Proceedings of the Royal Society of London B: 458 Biological Sciences, 236, 339-72. https://doi.org/10.1098/rspb.1989.0027 459 Clutton-Brock, T.H. (2007). Sexual selection in males and females. Science, 318, 1882-1885. https:// 460 doi. org/10.1126/science.1133311 461 Clutton-Brock, T.H., & Huchard, E. (2013). Social competition and selection in males and females. 462 Philosophical Transactions of the Royal Society B, 368, 20130074. https://dx.doi.org/ 463 10.1098%2Frstb.2013.0074
20
464 Daly, J.W., Brown, G.B., Mensah-Dwumah, M., & Myers, C.W. (1978). Classification of skin alkaloids 465 from neotropical poison-dart frogs (Dendrobatidae). Toxicon, 16, 163–188. https://doi.org/ 466 10.1016/0041-0101(78)90036-3 467 Daly, J.W., Myers, C.W., & Whittaker, N. (1987). Further classification of skin alkaloids from neotropical 468 poison frogs (Dendrobatidae), with a general survey of toxic/noxious substances in the amphibia. 469 Toxicon 25, 1023-95. https://doi.org/10.1016/0041-0101(87)90265-0 470 Donnelly, M.A. (1989). Effects of reproductive resource supplementation on space-use patterns in 471 Dendrobates pumilio. Oecologia, 81, 212-218. https://doi.org/10.1007/BF00379808 472 Darst, C. R., Menéndez-Guerrero, P. A., Coloma, L. A., & Cannatella, D. (2005). Evolution of dietary 473 specialization and chemical defense in poison frogs (Dendrobatidae): A comparative analysis. 474 American Naturalist, 165, 56-69. https://doi.org/10.1086/426599 475 Davies, N.B. (1991). Mating systems. In: J.R. Krebs & N.B. Davies (Eds), Behavioural ecology: an 476 evolutionary approach (pp. 263–294). Oxford: Blackwell. 477 Donnelly, M.A. (1991). Feeding patterns of the strawberry poison frog, Dendrobates pumilio (Anura: 478 Dendrobatidae). Copeia, 3,723–730. https://doi.org/10.2307/1446399 479 Dreher, C.E., Cummings, M.E., & Pröhl, H. (2015). An analysis of predator selection to affect 480 aposematic coloration in a poison frog species. PLoS ONE, 10(6), e0130571. https://doi.org/ 481 10.1371/journal.pone.0130571 482 Emlen, S.T., Oring, L.W. (1977). Ecology, sexual selection, and the evolution of mating systems. 483 Science, 197, 215–223. https://doi.org/10.1126/science.327542 484 Foster, S.A. (2013). Evolutionary insights from behavioural geography: plasticity, evolution, and 485 responses to rapid environmental change. Evolutionary Ecology Research, 15, 705-731. 486 Guidelines for the Use of Animals (2012). Guidelines for the treatment of animals in behavioural 487 research and teaching. Animal Behaviour, 83, 301-309. https://doi.org/10.1016/j.anbehav. 2011. 488 10.031 489 Haase, A., & Pröhl, H. (2001). Female activity patterns and aggressiveness in the strawberry poison 490 frog Dendrobates pumilio (Anura: Dendrobatidae). Amphibia-Reptilia, 23, 129-140. https://doi. 491 org/10.1163/156853802760061778 492 Halliwell, B., Uller, T., Wapstra, E., & While, G.M. (2017). Resource distribution mediates social and 493 mating behaviour in a family living lizard. Behavioural Ecology, 28, 145-153. https://doi.org 494 /10.1093/beheco/arw134 495 Jakob, E.M., Marshall, S.D., Uetz G.W. (1996). Estimating fitness: a comparison of body condition 496 indices. Oikos, 77, 61–67. https://doi.org/ 10.2307/3545585
21
497 Joseph, M.B., Preston, D., Johnson, P.T.J. (2016). Integrating occupancy models and structural 498 equation models to understand species occurrence. Ecology, 97, 765-775. https://doi.org 499 /10.1890/15-0833.1 500 Kaufmann, J.H. (1983). On the definitions and functions of dominance and territoriality. Biological 501 Reviews, 58, 1–20. https://doi.org/10.1111/j.1469-185X.1983.tb00379.x 502 Kvarnemo, C. (2010) Parental care. In Evolutionary Behavioral Ecology, eds. Westmeat, D.F. & Fox, 503 C.W. Oxford University Press. Pp 451-467. 504 Limerick, S. (1980). Courtship behaviour and oviposition of the poison-arrow frog Dendrobates 505 pumilio. Herpetologica, 36, 69-71. 506 Maher, C.R., & Lott, D.F. (1995) Definitions of territoriality used in the study of variation in vertebrate 507 spacing patterns. Animal Behaviour, 6, 1581-1597. https://doi.org/10.1016/0003-3472(95) 508 90080-2 509 Marler, C.A., Walsberg, G., White, M.L., & Moore, M. (1995). Increased energy expenditure due to 510 increased territorial defence in male lizards after phenotypic manipulation. Behavioural Ecology 511 and Sociobiology, 37, 225–231. https://doi.org/10.1007/BF00177401 512 Meuche, I., Linsenmair, K.E., & Pröhl, H. (2011.) Female Territoriality in the Strawberry Poison Frog (O. 513 pumilio). Copeia, 3, 352-356. https://doi.org/10.1643/CE-08-135 514 Meuche, I., Linsenmair, K.E., & Pröhl, H. (2012). Intrasexual competition, territoriality and acoustic 515 communication in male strawberry poison frogs (Oophaga pumilio). Behavioural Ecology and 516 Sociobiology, 66, 613 – 621. https://doi.org/10.1007/s00265-011-1309-9 517 Miró, A., O’Brien, D., Hall, J., & Jehle, R. (2017) Habitat requirements and conservation needs of 518 peripheral populations: the case of the great crested newt (Triturus cristatus) in the Scottish 519 Highlands. Hydrobiologia, 792, 169-181. http://dx.doi.org/10.1007/s10750-016-3053-7 520 Murphy, C.G. (1994). Determinants of chorus tenure in barking treefrogs (Hyla gratiosa). Behavioural 521 Ecology and Sociobiology, 34, 285–294. 522 Myers, C.W., & Daly, J.W. (1983). Dart-poison frogs. Scientific American, 248, 96-105. https://doi.org/ 523 10.1038/scientificamerican0283-120 524 Ostfeld, R.S. (1990). The ecology of territoriality in small mammals. Trends in Ecology and Evolution, 525 5, 411–415. https://doi.org/10.1016/0169-5347(90)90026-A 526 Poelmann, E.H., & Dicke, M. (2008). Space use of Amazonian Poison frogs: testing the reproductive 527 defence hypothesis. Journal of Herpetology, 42, 270-278. https://doi.org/10.1670/07-1031.1 528 Pough, F.H., & Taigen, T.L. (1990). Metabolic correlates of the foraging and social behaviour of dart- 529 poison frogs. Animal Behaviour, 39, 145-155. https://doi.org/10.1016/S0003-3472(05)80734-1
22
530 Pröhl, H., & Hödl, W. (1999). Parental investment, potential reproductive rates and mating system in 531 the strawberry poison-dart frog Dendrobates pumilio. Behavioural Ecology and Sociobiology, 46, 532 215-220. https://doi.org/10.1007/s002650050612 533 Pröhl, H., & Berke, O. (2001). Spatial distribution of male and female strawberry poison frogs and 534 their relation to female reproductive resources. Oecologia, 129, 534-542. https://doi.org/ 535 10.1007/s004420100751 536 Pröhl, H. (2002). Population differences in female resource abundance, adult sex ratio and male 537 mating success in Dendrobates pumilio. Behavioural Ecology, 13, 175-181. https://doi.org/ 538 10.1093/ beheco/13.2.175 539 Pröhl, H. (2005). Territoriality in dendrobatid frogs. Journal of Herpetology, 39, 354–365. 540 https://doi.org/10.1670/162-04A.1 541 Pröhl, H., Willink, B., & Hauswaldt, S. (2013). Geographic variation in sexual signals and behaviour in 542 two species of poison frogs. Evolutionary Ecology Research, 15, 667–687. 543 R Core Team, (2016). R: A language and environment for statistical computing. R Foundation for 544 Statistical Computing. http://www.R-project.org> (accessed: May. 2016). 545 Roithmair, M.E. (1992). Territoriality and male mating success in the dart-poison frog Epipedobates 546 femoralis (Dendrobatidae, Anura). Ethology, 92, 331-343. https://doi.org/10.1111/j.1439- 547 0310.1992.tb00970.x 548 Rosvall, K.A. (2011). Intrasexual competition in females: evidence for sexual selection. Behavioural 549 Ecology, 22, 1131-1140. https://doi.org/10.1093/beheco/arr106 550 Rousset, F., & Ferdy, J.-B. (2014). Testing environmental and genetic effects in the presence of spatial 551 autocorrelation. Ecography, 37, 781-790. https://doi.org/10.1111/ecog.00566 552 Rudh, A., Rogell, B., Håstad O., & Qvarnström, A. (2011). Rapid population divergence linked with co- 553 variation between coloration and sexual display in Strawberry poison frogs. Evolution, 65, 1271– 554 1282. https://doi.org/10.1111/j.1558-5646.2010.01210.x 555 Santos, J.C., & Cannatella, D.C. (2011). Phenotypic integration emerges from aposematism and scale 556 in poison frogs. Proceedings of the National Academy of Science of the USA, 108, 6175–6180. 557 https://doi.org/10.1073/pnas.1010952108 558 Saporito, R.A., Garraffo, H.M., Donnelly, M.A., Edwards, A.L., Longino, J.T., & Daly, J.W. (2004). 559 Formicine ants: an arthropod source for the pumiliotoxin alkaloids of dendrobatid poison frogs. 560 Proceedings of the National Academy of Science of the USA, 101, 8045-8050. 561 https://doi.org/10.1073/pnas.0402365101 562 Saporito, R.A., Donnelly, M.A., Jain, P., Garraffo, H.M., Spande, T.F., & Daly, J.W. (2007). Spatial and 563 temporal patterns of alkaloid variation in the poison frog Oophaga pumilio in Costa Rica and 564 Panama over 30 years. Toxicon, 50, 757-778. https://doi.org/10.1016/j.toxicon.2007.06.022
23
565 Saporito, R.A., Zuercher, R., Roberts, M., Gerow, K.G., & Donnelly, M.A. (2007). Experimental Evidence 566 for Aposematism in the Dendrobatid Poison Frog Oophaga pumilio. Copeia, 2007(4), 1006–1011. 567 https://doi.org/10.1643/0045-8511(2007)7[1006:EEFAIT]2.0.CO;2 568 Saporito, R.A., Donnelly, M.A., Spande, T.F., & Garraffo, H.M. (2012). A review of chemical ecology in 569 poison frogs. Chemoecology, 22, 159–168. https://doi.org/10.1007/s00049-011-0088-0 570 Savage, J. (2002). The Amphibians and Reptiles of Costa Rica: A Herpetofauna between Two 571 Continents, between Two Seas. Chicago: The University of Chicago Press. 572 Schoener, T.W. (1983). Field experiments on interspecific competition. American Naturalist, 122, 240- 573 285. 574 Staudt, K., Meneses Ospina, S., Pröhl, H. (2010). Foraging behaviour and territoriality of the 575 strawberry poison frog (Oophaga pumilio) in dependence of the presence of ants. Amphibia- 576 Reptilia, 31, 217-227. https://doi.org/10.1163/156853810791069100 577 Steinmann, A.R., Priotto, J.W., & Polop, J. (2009). Territorial behaviour in corn mice, Calomys 578 musculinus (Muridae: Sigmodontinae), with regard to mating system. Journal of Ethology, 27, 51- 579 58. https://doi.org/10.1007/s10164-008-0083-2 580 Stockley, P., & Campbell, A. (2013). Female competition and aggression: interdisciplinary 581 perspectives. Philosophical Transactions of the Royal Society B, 368, 20130073. https://doi.org/ 582 10.1098/rstb.2013.0073 583 Takada, W., Sakata, T., Shimano, S., Enami, Y., Mori, N.,Nishida, R., & Kuwahara, Y. (2005). 584 Scheloribatid mites as the source of pumiliotoxins in dendrobatid frogs. Journal of Chemical 585 Ecology, 31, 2403-2415. https://doi.org/10.1007/s10886-005-7109-9 586 Test, F.H. (1954). Social aggressiveness in a amphibian. Science, 120, 140-141. 587 Webb, J.K., Scott, M.L., Whiting, M.J., & Shine, R. (2015). Territoriality in a snake. Behavioural Ecology 588 and Sociobiology, 69, 1657-1661. https://doi.org/10.1007/s00265-015-1977-y 589 Wells, K.D. (1980a). Social behaviour and communication of a dendrobatid (Colostethus trinitatis). 590 Herpetologica, 36, 189-199. 591 Wells, K.D. (1980b). Behavioural ecology and social organization of a dendrobatid frog (Colostethus 592 inguinalis). Behavioural Ecology and Sociobiology, 6, 199-209. https://doi.org/10.1007/ 593 BF00569201 594 Wells, K.D. (2007). The Ecology and Behaviour of Amphibians. Chicago: The University of Chicago 595 Press. 596 Weygoldt, P. (1980). Complex brood care and reproductive behaviour in captive poison-arrow frogs, 597 Dendrobates pumilio O. Schmidt. Behavioural Ecology and Sociobiology, 7, 329-332. https://doi. 598 org/ 10.1007/BF00300674 599 Wilson, E.O. (1975). Sociobiology: The New Synthesis, Cambridge, MA: Belknap Press.
24
600 Wolff, J.O., & Peterson, J.A. (1998). An offspring-defence hypothesis for territoriality in female 601 mammals. Ethology, Ecology and Evolution, 10, 227-239. https://doi.org/10.1080/08927014. 602 1998.9522854 603 604
25
605
606
607 TABLE 1 Standardised parameter estimates, lower and upper 95% confidence intervals (LCI, UCI), and 608 results of the likelihood-ratio tests (LRT) for predictors of female Oophaga pumilio habitat occupancy 609 as estimated by the spatial generalized mixed effect model. Significant effects are highlighted in bold, 610 SAI: sustained abundance index.
611
612 Predictor Estimate LCI UCI chi2 (LRT) p-value (LRT) 613 Ants SAI 0.854 0.473 1.236 17.897 0.0000 614 Debris clusters 0.393 -0.355 1.140 0.843 0.3585 615 Leaf litter 0.789 0.329 1.256 11.563 0.0007 616 617 Rearing sites 0.538 0.056 1.020 4.352 0.0370 618612 619 613 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
26
614
615 TABLE 2 Ants (Formicidae) genera found in the study area, their taxonomic classification (subfamily) 616 and number captured.
Ants genera Subfamily No. captured Acromyrmex Myrmicinae 1 Atta 3 Cyphomyrmex 2 Pheidole 100 Solenopsis 442 Wasmannia 5 Brachymyrmex Formicinae 8 Camponotus 3 Paratrechina 34 Ectatomma Ectatommine 6 Gnamptogenys 17 Pachycondyla Ponerinae 3 617
618
619
620
27
621 FIGURE 1 Distribution of predictor variables for female habitat occupancy across the grid 622 system in the study plot.
623
624 FIGURE 2 Habitat occupancy of female Oophaga pumilio in the study plot. A) Spatial habitat 625 occupancy: the plot shows the proportion of time spent on each grid cell averaged across the 626 11 females studied. B) Distribution of home ranges (grey areas) and core areas (red grid 627 cells). The location of recorded aggressive events is indicated with dots.
628
629 FIGURE 3 Behavioural and ecological differences between female core areas and home 630 ranges: 3a) Number of ants, 3b) feeding attempts, 3c) clusters of leaf litter, 3d) debris 631 clusters, 3e) tadpole rearing sites.
632
28
633
634 635
636 FIGURE 1
637
638
29
639
640
641
642 FIGURE 2
643
644
645
646
30
647
648
649
650 FIGURE 3
31