The influence of temperature model assumptions on the prognosis accuracy of extinction risk

For a species whose abundance is well-known to correlate on the degree of heat different temperature model assumptions may affect the prognosis accuracy of persistence. Likewise, year-to-year autocorrelations in weather fluctuations are known to decrease extinction risk. Thus, we investigated the grey bush cricket Platycleis albopunctata. For this species is known that growth and reproduction is mainly influenced by temperature. We developed a stochastic individual based model for the bush cricket. This day-degree model described the demographic growth of the species that depends on temperature. Daily temperatures were generated by five different methods: (i) temperatures were sequentially taken from a meteorological database. To analyse the influence of different levels of autocorrelation in temperature records (ii) the day-to-day correlations were reduced by randomly permuting the sequence of days within the months of successive years from the database, (iii) year-to-year correlations were reduced by randomly rearranging the sequence of the years held in the database, (iv) combined day-to-day and year-to-year correlations were reduced according to the submodel (ii) and (iii), and finally (v) temperatures were randomly generated on the basis of Gaussian normal distributions. The mean and the variance of these distributions depended on the date of the year. Distributions were derived from the above mentioned meteorological database. We estimated highly different minimum viable population sizes. These did severely depend on the chosen temperature model. Values ranged from 3000 adults to more than a million adults per generation. High amounts of autocorrelation in temperature values decreased the extinction risk of a bush cricket population. Permuting the sequence of the years in the database increased extinction risk less than reducing day-to-day correlations. A decrease in autocorrelation of temperature records can result in unrealistic phenologies of life stages. Decreasing day-to-day or year-to-year autocorrelation in temperature records resulted in an increase or a decrease of the duration of the egg development or the duration of larval development. For poikilothermic species general implications are presented that are relevant for the design of quantitative models used in conservation biology. (C) 2000 Elsevier Science B.V. All rights reserved.