Herding as an emergent behaviour in harem groups of feral Garrano ponies.

Overview

• This research investigates herding behavior in feral Garrano ponies using a quantitative approach called transfer entropy to understand leader-follower dynamics.
• The study provides evidence that herding is an emergent behavior arising from information flow within the harem group, and validates transfer entropy as a reliable method compared to traditional techniques.

Introduction to the Research

• Herding behavior relates to how individuals in groups coordinate movement and make collective decisions, crucial for survival in animal groups and seen in human crowds as well.
• Horses live in multilevel social structures, often forming harems—a group led by a stallion that includes several mares and their offspring.
• Herding within these harems involves gathering the group to move cohesively in a specific direction, but detailed, quantitative understanding of this behavior has been lacking.

Methodology and Data Collection

• The study focuses on a harem group of feral Garrano ponies located in Serra D’Arga, Portugal.
• Researchers applied a metric called transfer entropy, which measures the directed information flow between individuals, to identify who influences whom and when.
• Key parameters quantified include the time lag of influence (average of 1.44 seconds), showing how quickly one pony’s movement affects another’s, and the duration of influence (average of 1.72 seconds), indicating how long this influence lasts during herding.
• Variance in these measurements allowed an understanding of how different social characteristics affect leadership and followership dynamics.

Validation of Transfer Entropy

• To ensure transfer entropy was a valid tool for detecting herding, the researchers compared its results with traditional behavioral analysis techniques based on clustering of group members.
• The scarcity of existing experimental data was addressed by generating synthetic data through a mathematical model specifically designed for feral horse harems.
• On this synthetic dataset, transfer entropy demonstrated superior performance:
• Accuracy of 0.80, indicating the method correctly identified leader-follower relationships 80% of the time.
• F1-score of 0.76, showing a balanced measure of precision and recall for detecting herding events, outperforming traditional clustering and time-series synchrony methods.

Emergent Nature of Herding Behavior

• The analysis revealed herding is not simply a direct reaction from a leader to all members simultaneously, but an emergent process characterized by:
• Indirect propagation of leadership influence: the signal or behavior spreads in a chain-like fashion among individual mares.
• Short bursts of information flow: rapid, intermittent communications or reactions propagate across the harem.
• This emergent property highlights the complexity of collective movement and decision-making in feral horse groups.

Implications and Contributions

• The study enriches scientific understanding of social dynamics and herding behavior in feral horses, an important wild population with natural social structures.
• It offers a robust quantitative tool—transfer entropy—that can be applied to study complex decision-making and movement coordination in other species.
• Findings contribute broadly to fields like behavioral ecology, animal cognition, and possibly inform conservation and management practices for wild or semi-wild horses.