abstract
- Many environmental monitoring programs include an assessment of the health of fish populations using a sentinel species and include an indicator of reproductive potential. Knowledge of the reproductive strategy of the fish species is critical for data interpretation but is not always known. The reproductive strategy of a species can be determined from detailed histological analyses of ovaries throughout the reproductive cycle; however, these studies can be costly and can delay the implementation of a monitoring program. Three quick and cost-effective methods of predicting the reproductive strategy (annual single spawning or annual multiple spawning) are evaluated in this study using predicted probabilities from binary logistic regression models as a means of classifying the reproductive strategies of 18 different fish species in Atlantic Canada. The first method was based on the hypothesis that the variability in the ovary weight-body weight relationship in prespawning females is higher in multiple spawners. This method did not have a good classification rate due to some multiple spawners having low variability. The other two methods involved predictor variables representing the proportion of oocytes in different stages of development and predictor variables representing the distribution of oocyte sizes during the prespawning season for 111 fish (25 different samples for species). Predicted probabilities from these regression models could be used to correctly classify the reproductive strategies of all 25 samples (development stage model) and all but one sample (oocyte size distribution model). These models can be used to estimate the reproductive strategy of a species from a single sample of fish collected during the prespawning period to support species selection and data interpretation in environmental monitoring programs.