Search

Knowledge of the genetic make-up and demographic history of invasive populations is critical to understand invasion mechanisms. Commensal rodents are ideal models to study whether complex invasion histories are typical of introductions involving human activities. The house mouse Mus musculus domesticus is a major invasive synanthropic rodent originating from South-West Asia. It has been largely studied in Europe and on several remote islands, but the genetic structure and invasion history of this taxon have been little investigated in several continental areas, including West Africa. In this study, we focussed on invasive populations of M. m. domesticus in Senegal. In this focal area for European settlers, the distribution area and invasion spread of the house mouse is documented by decades of data on commensal rodent communities. Genetic variation at one mitochondrial locus and 16 nuclear microsatellite markers was analysed from individuals sampled in 36 sites distributed across the country. A combination of phylogeographic and population genetics methods showed that there was a single introduction event on the northern coast of Senegal, from an exogenous (probably West European) source, followed by a secondary introduction from northern Senegal into a coastal site further south. The geographic locations of these introduction sites were consistent with the colonial history of Senegal. Overall, the marked microsatellite genetic structure observed in Senegal, even between sites located close together, revealed a complex interplay of different demographic processes occurring during house mouse spatial expansion, including sequential founder effects and stratified dispersal due to human transport along major roads.

A nonlinear model for describing the larval development of the European corn borer, Ostrinia nubilalis (Hübner), was calibrated because previous work had revealed limitations of the degree-day model. The model proposed by Logan et al. in 1976 was used as a starting point. Experimental design was based on destructive sampling in order to record a large number of larvae, and on variable temperatures in order to fit the nonlinear model with few experiments and to avoid artifacts caused by constant temperatures. A calibration method was also proposed. The Logan model and the normal distributions describing the timing of the moult could be calibrated with the proposed method for temperatures ranging between 9 and 42°C. The model was validated under field conditions. Validation revealed that at least 1 environmental variable, probably maize phenology, that was not taken into account by the temperature model had a strong effect on the development rate. Leaving out the infestation where this effect was maximum (infestation during the early whorl stage), the model predicted the 2nd moult with a 48-h accuracy, and the last moult with an accuracy of better than 5 d. However, predictions were only half a day better than those obtained by the degree-day model. Nevertheless, this work helped to determine precisely the range of validity for the degree-day concept. Further investigations are along 2 directions: testing other mathematical formulations of the relationship between development rate and temperature to determine the possible limitations of the Logan formulation, and analyzing the influence of other environmental variables once the influence of temperature has been taken into account.

Nine nonlinear models (degree-day with a 10°C base, degree-day model, polynomial, normal distribution, Stinner model, Sharpe and DeMichele model, Logan model with or without a base, Hilbert and Logan model) were tested to describe the relationship between development rate and temperature for the European com borer, Ostrinia nubilalis Hübner. These models were calibrated with laboratory experiments based on destructive sampling under sinusoidal temperatures from 9 to 42°C. They were then evaluated and compared with field experiments, at temperatures from 4 to 37°C, carried out as artificial infestations in various areas. Laboratory development models with the smallest time shifts between predicted and observed molts were the Sharpe and DeMichele model, polynomial model, normal distribution, and Stinner model. A classification with an adjusted coefficient of determination (taking into account the number of parameters in the model) slightly modified the results and the best models were the polynomial and the normal distribution, followed by the Stinner model and the degree-day model with a 10°C base. The classification was almost reversed under field conditions. Large shifts occurred in all models in some experiments, thereby demonstrating that temperature was not the only factor involved. Models showing the smallest shifts under field conditions were the degree-day model and the Logan model. The experiments that showed the greater shifts corresponded to infestations on early or on late phenological stages of corn. Results confirmed that the experimental design and calibration method were adequate. The laboratory experiments provided a means of calibrating the models and of predicting results under field conditions from laboratory experiments. Future studies will focus on a precise quantitative analysis of factors affecting development other than temperature, notably infestation date, climatic factors, and phenological stage of corn.