A multi-agent ecosystem model for studying changes in a tropical estuarine fish assemblage within a marine protected area

¹ Institut de Recherche pour le Développement (IRD), UMR LEMAR 195/6539 (CNRS/IRD/UBO/Ifremer), BP 1386 Dakar, Senegal
² Institut de Recherche pour le Développement (IRD), UMR 212 EME (IRD, Ifremer, U. Montpellier II), av. Jean Monnet, BP 171, 34203 Sète Cedex, France

^a Corresponding author: timothee.brochier@gmail.com

Received: 14 May 2012
Accepted: 21 September 2012

Abstract

As marine protected areas (MPAs) are increasingly being utilised as a tool for fishery management, their impact on the food web needs to be fully understood. However, little is known about the effect of MPAs on fish assemblages, especially in the presence of different life history and ecological traits. Modelling the observed changes in fish population structures may provide a mechanistic understanding of fish assemblage dynamics. In addition, modelling allows a quantitative estimate of MPA spill-over. To achieve this purpose, we adapted an existing ecosystem model, OSMOSE (Object-oriented simulator of marine biodiversity exploitation), to the specific case of the presence of fish with multiple life histories. The adapted model can manage 4 main categories of life history identified in an estuary MPA: fish that (1) spend their entire life cycle locally, (2) are present only as juveniles, (3) enter the area as juveniles and stay permanently except during reproduction periods, which occur outside the estuary, and (4) are present occasionally and for a short time for foraging purposes. To take into account these specific life-history traits, the OSMOSE code was modified. This modelling approach was developed in the context of the Bamboung Bolong MPA, located in a mangrove area in the Sine-Saloum Delta, Senegal. This was the ideal case to develop our approach as there has been scientific monitoring of the fish population structure inside the MPA before fishery closure, providing a reference state, and continuous monitoring since the closure. Ecologically similar species were pooled by trophic traits into 15 groups that represented 97% of the total biomass. Lower trophic levels (LTL) were represented by 6 compartments. The biomass of the model species was calibrated to reproduce the reference situation before fishery closure. Model predictions of fish assemblage changes after fishery closure corresponding to the Bamboung MPA creation scenario were compared to field observations; in most cases the model reproduces observed changes in biomass (at least in direction). We suggest the existence of a “sanctuary effect”, that was not taken into account in the model, this could explain the observed increase in biomass of top predators not reproduced by the model. Finally, the annual MPA fish spill-over was estimated at 11 tons (~33% of the fish biomass) from the model output, mainly due to diffusive effects.