Burczyk Jaroslaw, Chybicki Igor J
Institute of Biology and Environmental Protection, Department of Genetics, Bydgoszcz University, ul. Chodkiewicza 30, 85-064 Bydgoszcz, Poland.
Evolution. 2004 May;58(5):956-63. doi: 10.1111/j.0014-3820.2004.tb00430.x.
Through simulations we have investigated the statistical properties of two of the main approaches for directly estimating pollen gene flow (m) in plant populations: genotypic exclusion and mating models. When the assumptions about accurately known background pollen pool allelic frequencies are met, both methods provide unbiased results with comparable variances across a range of true m values. However, when presumed allelic frequencies differ from actual ones, which is more likely in research practice, both estimators are biased. We demonstrate that the extent and direction of bias largely depend on the difference (measured as genetic distance) between the presumed and actual pollen pools, and on the degree of genetic differentiation between the local population and the actual background pollen sources. However, one feature of the mating model is its ability to estimate pollen gene flow simultaneously with background pollen pool allelic frequencies. We have found that this approach gives nearly unbiased pollen gene flow estimates, and is practical because it eliminates the necessity of providing independent estimates of background pollen pool allelic frequencies. Violations of the mating model assumptions of random mating within local population affect the precision of the estimates only to a limited degree.
通过模拟，我们研究了植物种群中直接估计花粉基因流（m）的两种主要方法的统计特性：基因型排除法和交配模型法。当满足关于准确已知背景花粉库等位基因频率的假设时，两种方法在一系列真实m值范围内均能提供具有可比方差的无偏结果。然而，当假定的等位基因频率与实际频率不同时（这在研究实践中更有可能发生），两种估计方法都会产生偏差。我们证明，偏差的程度和方向很大程度上取决于假定花粉库与实际花粉库之间的差异（以遗传距离衡量），以及当地种群与实际背景花粉源之间的遗传分化程度。然而，交配模型的一个特点是它能够同时估计花粉基因流和背景花粉库等位基因频率。我们发现这种方法给出的花粉基因流估计几乎是无偏的，并且很实用，因为它无需提供背景花粉库等位基因频率的独立估计值。违反当地种群内随机交配的交配模型假设只会在有限程度上影响估计的精度。
