Exploring links between personality traits and their social and non-social environments in wild poison frogs

Abstract

An animal's behavioral phenotype comprises several traits, which are hierarchically structured in functional units. This is manifested in measured behaviors often being correlated, partly reflecting the need of a coordinated functional response. Unfortunately, we still have limited understanding whether consistent differences in animal behaviors are due to underlying physiological constraints or a result of plastic adaptation to their current environment. Therefore, characterizing the spatial distribution of behaviors can provide important insights into causes and consequences of behavioral variation. In the present study, we quantified behaviors in a wild, free-ranging population of the Neotropical frog Allobates femoralis. We investigated how these behaviors were linked to the frogs' natural and social environment and quantified the extent to which these behaviors consistently differed among individuals (i.e., animal personality). We assessed levels of aggressiveness, exploration, and boldness by measuring several underlying behaviors expressed in a set of experimental assays, and found evidence for consistent among-individual differences along these axes. Contrary to our expectation, there was no relationship between individual behaviors and their natural environment, but we found a plastic response of males to changes in female density, which might reflect how individuals cope with their socio-ecological environment.

Significance statement: How are behavioral phenotypes distributed across space? Here, we studied an entire free-ranging population of poison frogs, and investigated if the personality traits aggressiveness, exploration, and boldness are linked to the frogs' natural or social environment. We found that behavioral traits were non-randomly distributed across the population, suggesting that the spatial arrangement of behavioral traits reflects how individuals cope with their complex natural and social environment.

Keywords: Animal personality; Behavioral variation; Environment; Non-random distribution; Poison frogs.

Fig. 1. Picture of the territorial defense trial, presenting the speaker and a male on top of the perimeter

Fig. 2

Schematic (a) and picture (b) of the Novel Environment Test. The cooler box measured 50 × 25 × 29 cm, with a 10 cm PVC tube attached on one side of the box. A sliding door separated the shelter from the box. Two LED tubes and a Hero Black 5 Go pro camera were attached to the lid of the box. Three solid PVC tubes (10 cm height, 5 cm diameter) were placed inside the box as visual obstacles at randomized positions every day. A grid was drawn on the floor of the cooler box to help randomize the position of the obstacles. A mesh net was placed in the cooler box at 20 cm height to prevent the frog from jumping on a wall outside of the camera range

Fig. 3

Path diagrams of the best structure equation models (SEMs) (based on difference in Akaike’s information criterion (AIC) values) explaining the covariance structure among four behaviors assessed during an aggressivity test (a), and five behaviors assessed during a Novel Environment Test (b). “HBO” refers to the latency until the first head-body orientation. Squares represent the variances of the different behaviors explained by the SEM structure (R²). Numbers associated with arrows are standardized factor loadings which represent how behavioral responses are predicted to change based on changes to the latent variable. Number in brackets represent variances of residuals or error variances (e) associated to each behavior. All simulated models can be found in the Supplementary material (Supplementary Fig. 1, 2)

Fig. 4. Range of variation in the three behaviors that best represented the latent variables of aggression, exploration and boldness.

The latency until the first jump (a), the time spent in the shelter (b), and the distance travelled in the Novel Environment Test (c) are presented for individual males. All variables have been transformed using a log (a) or constant (b and c) transformation. Boxes indicate the inter quartile range, with the central line depicting the median and the whiskers extending to 1.5*IQR. Dots represent the results of each trial. Males are ordered by their median (represented as a horizontal bold line)

Fig. 5. Maps showing the spatial distribution of individual performance of male frogs on the island in the behavioral essays.

The maps use the Voronoi territories of 7 March 2019 when the most individuals were present at the same time on the island. The maps show the mean value, calculated over all respective trials of (a) the individuals’ latency to jump in the territorial defense trial, (b) the time spent in the shelter during the Novel Environment Test (NET), and (c) the distance travelled in the NET. All color ramps have 20 equal intervals across the full range of the respective value; darker colors represent shorter latency in (a), shorter time spent in the shelter in (b), and longer distances in (c). Black squares indicate the 14 artificial pools that were in place on the island since 2018; thin gray lines show 50 cm elevation isoclines; the blue area shows the river Arataye. The territories of two males that were not tested in the behavioral assays are shown with a hatched white pattern