Predator-dependent functional responses and interaction strengths in a natural food web Timothy E. Essington and Sture Hansson Abstract: Predator-dependent functional responses decouple predation mortality from fluctuations in predator abun- dance and therefore can prevent strong “top-down” interaction strengths in food webs. We evaluated whether contrasts in the functional response of Baltic Sea cod (Gadus morhua) were consistent with the contrasting population dynamics of two prey species, herring (Clupea harengus) and sprat (Sprattus sprattus): sprat abundance increased nearly threefold following a sharp decline in the cod population (a strong interaction), whereas herring abundance failed to increase (a weak interaction). We found striking differences in the functional response of cod on alternative prey, and these were consistent with the observed patterns in interaction strengths. Cod predation was the dominant source of mortality for age-1 and age-2 sprat but was only important for age-1 herring. Moreover, the magnitude of predation mortality on age-1 and age-2 sprat was highly sensitive to cod biomass, whereas predation mortality on herring was only moderately sensitive to cod biomass. These analyses suggest the possibility that food webs are comprised of linkages that vary with respect to the magnitude and importance of predation mortality and how this mortality varies with changes in predator abundance. Résumé : Les réponses fonctionnelles qui dépendent des prédateurs peuvent éliminer le lien entre la mortalité due à la prédation et les fluctuations d’abondance des prédateurs et ainsi empêcher le développement d’importantes forces d’interaction de type «descendant» dans les réseaux trophiques. Nous avons vérifié si les différences de réponses fonc- tionnelles de la morue franche de la Baltique (Gadus morhua) étaient compatibles avec les dynamiques de population distinctes de deux de ses proies, le hareng atlantique (Clupea harengus) et le sprat (Sprattus sprattus) : l’abondance du sprat a augmenté de trois fois après une chute rapide de la population de morues (interaction forte), alors que celle du hareng n’a pas augmenté (interaction faible). Il existe d’importantes différences dans la réponse fonctionnelle de la morue à ces deux proies, différences qui s’accordent aux patterns de forces d’interaction observés. La prédation par les morues est la cause principale de mortalité des sprats d’âges 1 et 2, mais elle n’a d’importance que chez les harengs d’âge 1. De plus, l’intensité de la mortalité due à la prédation chez les sprats d’âges 1 et 2 est très sensible à la bio- masse des morues, alors qu’elle ne l’est que moyennement chez les harengs. Nos analyses indiquent qu’il est possible que les réseaux trophiques soient composés de liens qui varient en fonction de l’ampleur et de l’importance de la mor- talité due à la prédation et de la variation de cette mortalité en fonction des changements d’abondance des prédateurs. [Traduit par la Rédaction] Essington and Hansson 2226 Introduction The study of food webs has proceeded under three largely divergent pathways (Winemiller and Polis 1996). The topo- logical approach focuses on patterns in food web structure, concentrating on the constraints on connectance and com- plexity (Dunne et al. 2002). Second, energetic studies depict the flow of energy or biomass through a food web, focusing on the rates and magnitude of material cycling and identify- ing the trophic source of production (Baird and Ulanowicz 1989). Lastly, interaction webs are used to depict the dy- namic behavior of food webs, focusing on the patterns of in- teraction strength, top-down control, and trophic cascades (Paine 1980) An integration of these disparate approaches is likely to substantially improve our understanding of food web patterns and dynamics (Pickett et al. 1994). Energetic and interaction webs appear, at least initially, to offer the potential for integration, as a strong energetic link might predispose a strong interaction link. Yet, Paine (1980) recognized that strong energetic links and strong interaction links are largely independent, suggesting that quantifying the rate of energy or biomass flow between trophic groups is not sufficient to predict their dynamics. From a trophodynamic perspective, however, we can posit two conditions required for a strong top-down interaction: (i) the mortality rate caused by a predator should represent the dominant component of the prey total mortality rate and (ii) the mortality rate im- posed by this predator should depend strongly on the abun- dance of the predator. If these two conditions are met, a strong top-down interaction can arise provided that there are not any strong compensatory responses (e.g., territoriality, space limitation, etc.) that act to maintain constant prey pro- ductivity. One potential avenue to explore these two conditions de- rives from consideration of the functional response of preda- Can. J. Fish. Aquat. Sci. 61: 2215–2226 (2004) doi: 10.1139/F04-146 © 2004 NRC Canada 2215 Received 7 October 2003. Accepted 2 June 2004. Published on the NRC Research Press Web site at http://cjfas.nrc.ca on 1 February 2005. J17778 T.E. Essington. 1 School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA 98195, USA. S. Hansson. Department of Systems Ecology, Stockholm University, S-106 91 Stockholm, Sweden. 1 Corresponding author (e-mail: essing@u.washington.edu).