The predicting power of models for eutrophication

Restoration of water-bodies from eutrophication has proved to be extremely difficult. Mathematical models have been used extensively to provide guidance for management decisions. The aim of this paper is to elucidate important problems of using models for predicting environmental changes. First, the necessity for a proper uncertainty assessment of the model, upon calibration, has not been widely recognized. Predictions must not be a single time trajectory; they should be a band, expressing system uncertainty and natural variability. Availability of this information may alter the decision to be taken. Second, even with well-calibrated models, there is no guarantee they will give correct projections in situations where the model is used to predict the effects of measures designed to bring the system into an entirely different ”operating point”, as is typically the case in eutrophication abatement. The concept of educated speculation is introduced to partially overcome this difficulty. Lake Veluwe is used as a case to illustrate the point. Third, as questions become more detailed, such as ”what about expected algal composition”, there is a greater probability of running into fundamental problems that are associated with predicting the behaviour of complex non-linear systems. Some of these systems show extreme initial condition sensitivity and even, perhaps, chaotic behaviour, and are therefore fundamentally unpredictable.

Estimating relative abundance from catch and effort data, using neural networks

We develop and test a method to estimate relative abundance from catch and effort data using neural networks. Most stock assessment models use time series of relative abundance as their major source of information on abundance levels. These time series of relative abundance are frequently derived from catch-per-unit-of-effort (CPUE) data, using general linearized models (GLMs). GLMs are used to attempt to remove variation in CPUE that is not related to the abundance of the population. However, GLMs are restricted in the types of relationships between the CPUE and the explanatory variables. An alternative approach is to use structural models based on scientific understanding to develop complex non-linear relationships between CPUE and the explanatory variables. Unfortunately, the scientific understanding required to develop these models may not be available. In contrast to structural models, neural networks uses the data to estimate the structure of the non-linear relationship between CPUE and the explanatory variables. Therefore neural networks may provide a better alternative when the structure of the relationship is uncertain. We use simulated data based on a habitat based-method to test the neural network approach and to compare it to the GLM approach. Cross validation and simulation tests show that the neural network performed better than nominal effort and the GLM approach. However, the improvement over GLMs is not substantial. We applied the neural network model to CPUE data for bigeye tuna (Thunnus obesus) in the Pacific Ocean.

Gear modifications for fishing octopus, Octopus vulgaris, on live-bottom and adjacent flat bottom habitats in coastal waters off North Carolina

The Common Octopus, Octopus vulgaris, is an r-selected mollusk found off the coast of North Carolina that interests commercial fishermen because of its market value and the cost-effectiveness of unbaited pots that can catch it. This study sought to: 1) determine those gear and environmental factors that influenced catch rates of octopi, and 2) evaluate the feasibility of small-scale commercial operations for this species. Pots were fished from August 2010 through September 2011 set in strings over hard and sandy bottom in waters from 18 to 30 m deep in Onslow Bay, N.C. Three pot types were fished in each string; octopus pots with- and without lids, and conch pots. Proportional catch was modeled as a function of gear design and environmental factors (location, soak time, bottom type, and sea surface water temperature) using binomially distributed generalized linear models (GLM’s); parsimony of each GLM was assessed with Akaike Information Criteria (AIC). A total of 229 octopi were caught throughout the study. Pots with lids, pots without lids, and conch pots caught an average of 0.15, 0.17, and 0.11 octopi, respectively, with high variability in catch rates for each pot type. The GLM that best fit the data described proportional catch as a function of sea surface temperature, soak time, and station; greatest proportional catches occurred over short soak times, warmest temperatures, and less well known reef areas. Due to operating expenses (fuel, crew time, and maintenance), low catch rates of octopi, and high gear loss, a directed fishery for this species is not economically feasible at the catch rates found in this study. The model fitting to determine factors most influential on catch rates should help fishermen determine seasons and gear soak times that are likely to maximize catch rates. Potting for octopi may be commercially practical as a supplemental activity when targeting demersal fish species that are found in similar habitats and depth ranges in coastal waters off North Carolina.