Development of behavioural profile in the Northern common boa (Boa imperator): Repeatable independent traits or personality?

Abstract

Recent studies of animal personality have focused on its proximate causation and ecological and evolutionary significance in particular, but the question of its development was largely overlooked. The attributes of personality are defined as between-individual differences in behaviour, which are consistent over time (differential consistency) and contexts (contextual generality) and both can be affected by development. We assessed several candidates for personality variables measured in various tests with different contexts over several life-stages (juveniles, older juveniles, subadults and adults) in the Northern common boa. Variables describing foraging/feeding decision and some of the defensive behaviours expressed as individual average values are highly repeatable and consistent. We found two main personality axes-one associated with foraging/feeding and the speed of decision, the other reflecting agonistic behaviour. Intensity of behaviour in the feeding context changes during development, but the level of agonistic behaviour remains the same. The juveniles and adults have a similar personality structure, but there is a period of structural change of behaviour during the second year of life (subadults). These results require a new theoretical model to explain the selection pressures resulting in this developmental pattern of personality. We also studied the proximate factors and their relationship to behavioural characteristics. Physiological parameters (heart and breath rate stress response) measured in adults clustered with variables concerning the agonistic behavioural profile, while no relationship between the juvenile/adult body size and personality concerning feeding/foraging and the agonistic behavioural profile was found. Our study suggests that it is important for studies of personality development to focus on both the structural and differential consistency, because even though behaviour is differentially consistent, the structure can change.

Fig 1. Timeline summarizing the whole experiment.

A time table of particular tests administration at defined lifestages/feeding blocks. The life stages were adjusted to life history of the studied species (the Northern common boa, Boa imperator) to cover the time from birth to reaching sexual maturity of all individuals (more than four years).

Fig 2. Development of Prey acceptance.

Development of variable Prey acceptance measured in the Feeding trials, calculated as an index from regular feeding trials conducted during the whole experiment in six life stages A (0–0.5 years), B (0.5-1years), C₁ (1–1.5 years), C₂ (1.5–2 years), D (2–3 years) and E (3–4 years). The Prey acceptance in early development is significantly different from the subsequent life stages, yet this behaviour is significantly repeatable (R = 0.113, p = 0.014) and fairly consistent as revealed by Kendal’s test (Kendal’s W = 0.367).

Fig 3. Example of context generality in juvenile life stage A (0–0.5 year).

Visualization of the context generality between the Prey acceptance and Catch latency measured in the Feeding trials or Novel prey test, respectively (Spearman’s correlation coefficient = -0.498, p = 0.005). Vertical axis represents rank of the individual ordered according to a particular trait. Catch latency rank order was inverted, so that the lowest Catch latency (i.e. the fastest animals) are in the upper right part of the figure. The lines represent the rank stability/change for each snake in two different contexts. The numbers refer to identity of individual snakes (1–30). Therefore, in the ideal situation where all animals are perfectly consistent, all slopes would be zero (e.g. individual number 30). Snake number six has the third biggest proportion of successful feeding trials per block (Prey acceptance) with the shortest Latency to catch the novel prey (the fastest foraging decision).

Fig 4. Example of context generality in subadults, life stage D (3–4 year).

Visualization of the context generality between the Prey acceptance and Catch latency measured in the Feeding trials or Novel prey test, respectively (Spearman’s correlation coefficient = -0.094, p = 0.619). Vertical axis represents the rank of the individual ordered according to a particular trait. Catch latency rank order was inverted, so that the lowest Catch latency (i.e. the fastest animals) are in the upper right part of the figure. The lines represent the rank stability/change for each snake in two different contexts. The numbers refer to identity of individual snakes (1–30). Therefore, in the ideal situation where all animals are perfectly consistent, all slopes would be zero (e.g. individual number 4). Snake number twenty-seven has the third biggest proportion of successful feeding trials per block (Prey acceptance) with the shortest Latency to catch the novel prey (the fastest foraging decision).

Fig 5. Example of context generality in subadults, life stage D (3–4 year).

Visualization of the context generality between the Index of defensive behaviour and Occurrence of defensive behaviour measured in the Handling test or Restraint test, respectively (Spearman’s correlation coefficient = 0.539, p = 0.002). Vertical axis represents the rank of the individual ordered according to a particular trait. The lines represent the rank stability/change for each snake in two different contexts. The numbers refer to identity of individual snakes (1–30). Therefore, in the ideal situation where all animals are perfectly consistent, all slopes would be zero (e.g. individuals number 1–6). Snake number one has the smallest Index of defensive behaviour as well as the Occurrence of defensive behaviour.

Fig 6. Cluster analysis of variables contributing to Factor 1 (depicting the development of feeding personality).

Visualization of the correlation structure by Cluster analysis (1-Pearsons r was selected as metrics and Ward’s method for clustering). The variables that contributed mainly to the first factor in the overall factor analysis were inserted (Index of catch latency). The difference between two parts of life in the feeding context (development over the ontogeny) are clearly visible. The first two life stages (A and B) are clustering together, while the next stages (C₁, C₂, D and E) comprise the other cluster.

Fig 7. Cluster analysis of variables contributing to Factor 2 (depicting the development of agonistic variables).

Visualization of the correlation structure by Cluster analysis (1-Pearsons r was selected as metrics and Ward’s method for clustering). The variables that contributed mainly to the second factor in the overall factor analysis were inserted (most of the variables from the agonistic context and some from the context of exploration). Notice the clustering of agonistic variables from the first life stages (A, B, C₁ and C₂) delimiting from the second cluster of agonistic variables (the life stages D and E). Such pattern indicates development of this behaviour over the ontogeny.

Fig 8. Cluster analysis depicting mixture of variables contributing to the Factor 3.

Visualization of the correlation structure by Cluster analysis (1-Pearsons r was selected as metrics and Ward’s method for clustering). The variables that contributed mainly to the third factor in the overall factor analysis were inserted (most of the variables from the context of exploration and some of the agonistic and feeding contexts). Notice the mixture of variables contributing to Factor 3.