Biology subjects

Cherono, W.

Publications and source records attributed to Cherono, W..

4 recordsLinked to original sources

bioRxiv2024

Moving towards more holistic validation of machine learning-based approaches in ecology and evolution

Machine-learning (ML) is revolutionizing field and laboratory studies of animals. However, a challenge when deploying ML for classification tasks is ensuring the models are reliable. Currently, we evaluate models using performance metrics (e.g., precision, recall, F1), but these can overlook the ultimate aim, which is not the outputs themselves (e.g. detected species or individual identities, or behaviour) but their incorporation for hypothesis testing. As improving performance metrics has diminishing returns, particularly when data are inherently noisy (as human-labelled, animal-based data often are), researchers are faced with the conundrum of investing more time in maximising metrics versus doing the actual research. This raises the question: how much noise can we accept in ML models? Here, we start by describing an under-reported factor that can cause metrics to underestimate model performance. Specifically, ambiguity between categories or mistakes in labelling validation data produces hard ceilings that limit performance metrics. This likely widespread issue means that many models could be performing better than their metrics suggest. Next, we argue and show that imperfect models (e.g. low F1 scores) can still be useable. Using a case study on ML-identified behaviour from vulturine guineafowl accelerometer data, we first propose a simulation framework to evaluate robustness of hypothesis testing using models that make classification errors. Second, we show how to determine the utility of a model by supplementing existing performance metrics with biological validations. This involves applying ML models to unlabelled data and using the models outputs to test hypotheses for which we can anticipate the outcome. Together, we show that effects sizes and expected biological patterns can be detected even when performance metrics are relatively low (e.g., F1: 60-70%). In doing so, we provide a roadmap for validation approaches of ML classification models tailored to research in animal behaviour, and other fields with noisy, biological data. HighlightsEvaluating machine learning (ML) models must go beyond performance metrics Mislabels in validation data leads to underestimation of models performance Underestimated metrics can cause research delays despite models being useful We propose simulations and biological validations to evaluate model performance Models with low standard metrics can still be powerful for hypothesis testing

ecology↗

bioRxiv2023

The physiological costs of spatial positioning and leadership in collective movements

Individuals can gain substantial benefits from collective actions1-7. However, collective behaviours introduce new challenges, like coordinating actions, maintaining cohesion, and meeting the needs of different individuals. When making collective movements, leaders are typically thought to gain disproportionate benefits through the choice of more beneficial resources3 and/or earlier access to resources8. However, reaping these benefits can also have costs. Being at the front of a group can increase physical exertion4,9,10 and predation risk11,12. Moreover, ending up in a leadership position (i.e., at the front), is a process of negotiation in many animal groups. Within-group differences in directional preferences are typically resolved by some individuals initiating directional movements, after which they are either followed (if they are successful in leading) or return to the group (if they fail)13-30. By combining data on movement initiations (using whole-group GPS tracking31 and individual heart rate from implanted ECG loggers) in wild vulturine guineafowl, we found significant increases in heart rate (and decreases in heart rate variability) during collective movements. We found that attempting--and failing--to initiate directional movement was particularly costly, with the highest costs when consensus among group members was low and when individuals acted against the majority. Increases in heart rate and decreases in its variability can indicate physiological stress, entailing increased energy expenditure and long-term physiological damage. These results suggest that behaviours often thought beneficial to individuals (by influencing group behaviours) are also physiologically costly, representing a constraint on group-living and explaining why sometimes individuals opt out of contributing to leadership.

ecology↗

bioRxiv2023

Breeding alters females' social positions by changing dominance dynamics

Agonistic and affiliative interactions with group members dictate individuals access to resources, and investment in competing for resources is often traded off with other needs. For example, reproductive investment can reduce body condition and, thereby, an individuals ability to win future agonistic interactions. However, group members may also alter their behaviour towards reproductive individuals, such as becoming more or less aggressive. Here, we investigated the social consequences of reproduction in female vulturine guineafowl Acryllium vulturinum, a plural breeder in which females disperse and are subordinate to males. We found opposing patterns for breeders within- and between-sex dominance interactions from before to after breeding. Specifically, while breeding females became far less likely to win dominance interactions with non-breeding females after breeding than before breeding, they received considerably fewer male aggressions than non-breeding females after breeding. Using agent-based models, we then show that hierarchies inferred using the Percolation and Conductance method are robust to such variation in interaction rates, while other common methods may be prone to systematically overestimating or underestimating particular individuals hierarchy positions. Our study highlights reproduction as a driver of dominance dynamics, and illustrates how the study of dominance may benefit from considering variation in interaction rates.

animal behavior and cognition↗

bioRxiv2021

Costs dictate strategic investment in dominance interactions

Dominance is important for access to resources. As dominance interactions are costly, individuals should be strategic in who they interact with. One hypothesis is that individuals should direct costly interactions towards those closest in rank, as they have most to gain--in terms of attaining or maintaining dominance--from winning such interactions. Here, we show that male vulturine guineafowl (Acryllium vulturinum), a gregarious species with steep dominance hierarchies, strategically express higher-cost aggressive interactions towards males occupying ranks immediately below themselves in their groups hierarchy. In contrast, lower-cost aggressive interactions are expressed towards group members further down the hierarchy. By directly evaluating differences in the strategic use of higher- and lower-cost aggressive interactions towards competitors, we show that individuals disproportionately use highest-cost interactions--such as chases--towards males found one to three ranks below themselves. Our results support the hypothesis that the costs associated with different interaction types can determine their expression in social groups with steep dominance hierarchies.

animal behavior and cognition↗