Nonlinear numerical simulation models for assessing tropical fisheries with continuously breeding multicohort populations

Tropical fishery system (TFS) interactions and nonlinearities coupled with poor data and infrastructure exacerbates the dubious utility of applying traditional age-based assessment models, developed for temperate fisheries, in the tropical domain. New insights into TFS stock assessments were required, including: (i) procedures for validating length-based assessments, (ii) parameterizations for yield models, and (iii) development of models with dimensionality comparable to the system. A hierarchy of mathematical models was developed. CORECS (COntinuous RECruitment Simulation) is a generalized age-independent stochastic multicohort population simulation model that incorporates probability relations for recruitment, growth and survivorship. CORECS simulated length-frequency data produced according to a variety of hypotheses, based on empirical evidence from populations exhibiting tropical characteristics. Three circum-tropical marine fish-stock genera encompassing the spectrum of commercial-scale life histories were simulated and the efficacy of traditional assessment techniques were benchmarked against the CORECS model of ensemble conditions. Several length-based mortality methods for estimation of total mortality from average length of the catch data and growth parameter estimates were not efficient nor robust when compared against simulated outputs characterized by errorless data and exact initial conditions. A new statistical algorithm was developed. All estimators were biased with intra-annual trends in recruitment; however, the new estimator was least affected. In a Monte Carlo study, the new formulation was more accurate and precise than any of the previously best methods. Sensitivity analyses using inverse theory and the continuous model facilitated parameter estimation and the understanding of system responses to perturbations. The high dimensionality of the TFS apparently deleteriously influences the results obtained when constant parameterization and density-independent state interactions are assumed. The dynamics of any arbitrary number of population cohorts were modeled by coupled second-order differential equations with realistic nonlinear interactions for age-structured n-dimensional multicohort populations. In a simulation study, even rigidly deterministic systems showed dynamic limit cycles which suggests that traditional assumptions of stable-point equilibria may be unrealistic. Determination of system state space probabilities was discussed and may indicate the framework necessary for moving expert systems for fishery management from diagnostic analysis to optimal decision making