A Matrix Model to Study the Colonization by Brown Trout of a Virgin Ecosystem in the Kerguelen Islands

We present a matrix model for the study of the population dynamics of brown trout Salmo trutta L., introduced in the '60s in the virgin aquatic ecosystems of the Kerguelen Islands. This species clearly acclimatized very well: a portion of the population became migratory and spent a part of its life cycle in the sea, which allowed the rapid colonization of two rivers close to the stream of origin in the same bay (Baie Norvégienne).These migratory trout can become a smolt at 2, 3, or 4 years of age. The model takes into account age and smolt age structures and in a first step considers the fish from the Baie Norvégienne as belonging to a single population. The transition matrix looks like a 32 × 32 Leslie matrix in which some survival rates are not on the subdiagonal. They represent survival after the first sea migration and are particularly important for the dynamics of the whole population.The estimate of demographic parameters was obtained from a data base containing information collected in the field since 1970. The model was calibrated on the population size estimate and the stock structure of the migratory trout in 1979. Population size was estimated by tagging-recapture and monitoring of the migratory trout in freshwater when they returned to overwinter or reproduce. Under the hypothesis of a constant survival rate for all ages and categories of fish, it was possible to determine relationships between the annual population growth rate and survival rates at first downstream migration, for which no direct estimate was available.These constraints on the model induced paradoxical results. For instance, an increase in survival rate reduced migratory trout numbers in the first years of colonization. These data suggested that the average survival rate should be around 0.3 and not 0.5 as surmised in previous studies. However, the model systematically underestimated numbers of migratory trout during the first years of development. Thus, to improve the model it will be necessary to introduce survival rates varying with time. Another possible approach would be to consider the population as three subunits corresponding to the three colonized rivers of the Baie Norvégienne.