A predator–prey foraging game in experimental aviaries: Effects of the number of goldfish and refuge size

Sundararaj Vijayan, William A. Mitchell, Burt P. Kotler, Michael L. Rosenzweig, Jesse Balaban-Feld, Lotan Tamar Tovelem, Zvika Abramsky

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


Background: In a series of aviary experiments, we showed that little egrets (Egretta garzetta, the predators) and common goldfish (Carassius auratus, their prey) play a behavioural game in which the egrets adjust the timing of their predatory visits to experimental pools, and the fish adjust their exposure to predation by moving between a safe habitat that lacks food and a risky habitat that contains food. But we did not determine whether they do so in response to variation in the density of the fish. Question: Will the egret and the fish respond optimally to variation in prey density? Methods: We used two identical aviaries (7 m diameter), each with three separate pools of water (1.52 m diameter, 0.60 m deep, ∼1000 litres). In the centre of each pool was a single, solid disk (radius of 18.75, 22.75 or 36.75 cm) – the cover – which fish could hide under to protect themselves from egret predation. But fish could feed only in the risky habitat, i.e. the open water beyond the cover. In each 6 hour trial, the three pools had a single cover size and one of three densities of fish (10, 15, or 20) so that we could isolate the effect of number of fish from that of cover size. With high-definition CCTV cameras that yielded a panoramic view of the aviaries, we tabulated the behavioural data of egrets and fish every minute of each 6 hour trial – egret location and fish captures as well as the number of fish in each pool that were in the risky habitat. We also measured the number and percentage of food particles consumed by the fish. We counted the number of surviving fish at the end of each trial. We analysed these data parametrically using Python and SPSS. Results: Fish swam in the open water much less when an egret was patrolling the aviary. Risk-taking by the fish in every control cell (no egret) exceeded risk-taking in its experimental counterpart. Risk-taking was correlated negatively with cover size. The highest number of fish in a pool (i.e. 20) was associated with the least amount of risk-taking. Neither the proportion of fish captured by the egret nor the tendency for risk-taking displayed by the fish varied at different initial numbers of fish in a pool. Cover size, initial number of fish and their interaction affected the number of per-capita food particles that remained at the end of experimental trials in complex ways. Contrary to our prediction, the fish in the pools with small covers left less uneaten food per capita. The egret captured a smaller proportion of fish as cover size increased. But the number of fish captured by the egret rose with cover size and declined with the initial number of fish. The egret spent more time in a single pool if the initial number of fish was higher, and it took longer before returning to such a pool. The more often the egret visited a pool, the less often fish ventured into open water. However, also contrary to our prediction, the fish were most active in the pools with smaller covers (even though they experienced the highest mortality there). All of these findings combine to optimize the egret’s capture of fish during a trial. Conclusion: The fish and the egret were locked in a behavioural game in which each player adjusted its behaviour to that of the other player.

Original languageEnglish
Pages (from-to)333-352
Number of pages20
Issue number3
StatePublished - 1 May 2018

ASJC Scopus subject areas

  • Archaeology
  • History
  • Archaeology


Dive into the research topics of 'A predator–prey foraging game in experimental aviaries: Effects of the number of goldfish and refuge size'. Together they form a unique fingerprint.

Cite this