7872 - Laserfiche WebLink

Home Browse Search

1 1 1 and the probability of being captured if attacked (Gerritsen and Strickler 1977). Unfortunately, no data were available to estimate the probability that a larval Colorado squawfish would be attacked and eaten by a red shiner, so we conducted laboratory and outdoor mesocosm experiments in order to obtain the necessary information. Laboratory experiments were conducted using 38-L aquaria to estimate the probability that a larval Colorado squawfish would be captured by a red shiner when an attack occurred. In each trial, five larvae were added to an aquarium containing three similar-sized red shiners from one of six different size groups (mean TL = 38, 47, 53, 61, 63 and 69 mm). Larvae in trials had mean TL of 9.8, 11.1, 13.6, 15.3, 17.2, and 18.7 mm. Each of three predators was watched by a different observer, and the number of attacks and successful_captures during two consecutive 10-min intervals were recorded. If three or more larvae were eaten during a trial, an additional five larvae were added to the aquarium. A total of 39 trials were conducted. Predators were assumed to be acting independently because red shiners rarely interacted when capturing Colorado squawfish larvae. The mean number of attacks per trial by an individual was 21; total number of attacks by an individual ranged from 0 to 72. Logistic regression analysis was used to estimate the probability of capture of a larva as a function of predator length, prey length, and their squazed and interaction terms. A global general lineaz model was fit using the method of maximum likelihood, and a model selection tool called Akaike's Information Criterion (AIC) was used to sequentially eliminate predictors that were unimportant for explaining variation in the dependent variable. Model selection suggested that the best predictors of probability of capture (P(capture)) were (prey TL)2 and the 11