Abstract: By Monte Carlo method computer simulation is conducted to delineate the overall, annual and monthly fluctuations, and the statistical characteristics of 4 variables of the rodent community in the Chihuahuan Desert of NorthAmerica, including the number of species, biomass, species diversity and evenness. The results show that the means of 4 simulated community variables are 8.91, 1.544 kg/ ha, 1.58 and 0.73. All of the values of the real community variables are between the 95% confidence intervals of 4 simulated community variables. The number of species and biomass of the simulated community show the similar seasonal fluctuation patterns. The number of species, biomass and diversity of the simulated community show the similar interannual fluctuation patterns. When the null hypothesis is tenable: a variable has a normal distribution, X² values show that the number of species in 962 samples, biomass in 109 samples, species diversityin 5529 samples, species evenness in 6654 samples are following the normal distribution, i.e. no one of 4 variables is normally distributed while the confidence level is 95%. The results suggest that when sample size is small, we cannot estimate the confidence intervals of these variables by mean ±1.96 s ( s is the standard deviation of a variable ).　

Key words: Biomass, Community dynamics, Computer simulation, Monte Carlo method, Number of species, Probability distribution

摘要： 利用Monte Carlo 方法模拟了北美CHIHUAHUAN荒漠啮齿动物群落变量的总和、月间和年间变动情况及统计学特征, 这些变量是: 物种数、生物量、物种多样性和均匀性。模拟产生的10 000个样本显示这4个变量的平均值是8.91, 1.544 kg/hm²,1.58和0.73。真实群落4个变量的所有值均在模拟群落变量的95%置信区间内。模拟群落物种数、生物量和物种多样性有相同的年间变动规律。物种数和生物量有相同的季节变动规律。当零假设为这些变量均服从正态分布时, X² 检验及偏度和峭度说明, 在95%的显著性水平上,物种数有962个样本,生物量有109 个样本, 物种多样性有5 529 个样本,均匀性有6 654 个样本服从正态分布,即在95%的显著性水平上, 没有一个变量是服从正态分布的。结果说明, 当样本含量较小时, 我们不能用(平均数+1.96×标准误)的公式来计算。

关键词: 计算机模拟, 生物量, 群落动态, 物种数, 概率分布, Mon t e Ca r l o 法