Egg distributions and the information a solitary parasitoid has and uses for its oviposition decisions

Approximately three decades ago the question was first answered "whether parasitoids are able to assess the number or origin of eggs in a host" for a solitary parasitoid, Leptopilina heterotoma, by fitting theoretically derived distributions to empirical ones. We extend the set of different theoretically postulated distributions of eggs among hosts by combining searching modes and abilities in assessing host quality. In the models, parasitoids search either randomly (Poisson) (1) or by vibrotaxis (Negative Binomial) (2). Parasitoids are: (a) assumed to treat all hosts equally, (b) able to distinguish them in unparasitised and parasitised hosts only, (c) able to distinguish them by the number of eggs they contained, or (d) able to recognise their own eggs. Mathematically tractable combinations of searching mode (1 and 2) and abilities (a,b,c,d) result in seven different models (M1a, M1b, M1c, M1d, M2a, M2b and M2c). These models have been simulated for a varying number of searching parasitoids and various mean numbers of eggs per host. Each resulting distribution is fitted to all theoretical models. The model with the minimum Akaike's information criterion (AIC) is chosen as the best fitting for each simulated distribution. We thus investigate the power of the AIC and for each distribution with a specified mean number of eggs per host we derive a frequency distribution for classification. Firstly, we discuss the simulations of models including random search (M1a, M1b, M1c and M1d). For M1a, M1c and M1d the simulated distributions are correctly classified in at least 70% of all cases. However, in a few cases model M1b is only properly classified for intermediate mean values of eggs per host. The models including vibrotaxis as searching behaviour (M2a, M2b and M2c) cannot be distinguished from those with random search if the mean number of eggs per host is low. Among the models incorporating vibrotaxis the three abilities are detected analogously as in models with random search. Experiments with two species of solitary parasitoids (L. heterotoma and Asobara tabida) are conducted. All theoretically postulated distributions are separately fitted to the resulting experimental egg distributions. The AIC criterion is used to choose the best fitting theoretical distribution. For both parasitoid species the frequency distribution of best fitting models for experimental data is compared to the classification of distributions generated by simulations. This leads to the conclusion that both L. heterotoma and A. tabida are able to distinguish between parasitised and unparasitised hosts. For L. heterotoma the results point to an ability to assess the number of eggs in a host, whereas A. tabida does not seem to have this ability. This difference suggests that an egg is more valuable for L. heterotoma than for A. tabida.