Sexing weakly/criptically dimorphic birds requires invasive techniques or molecular analysis. Alternatively, sexing can be based upon morphometric analysis, which remains invaluable in remote field conditions such as Polar regions. Nevertheless, discriminatory power may be affected by methodological issues hampering comparison between/within studies, e.g. considering alternative analytical techniques or measurements taken by different researchers. We investigated the sexing potential of bill length, bill depth and body mass in adult Adélie penguins Pygoscelis adeliae of known sex. We performed discriminant analysis on a large dataset of measurements (237 males; 264 females) taken from penguins marked between 1994 and 2001, at Edmonson Point (Ross Sea, Antarctica). In a second step, we accounted for measurements collected by different researchers through permuted discriminant analysis. We found moderate support for sex discrimination through bill measurements (77%). Considering body mass did not improve classification power substantially (78.2%), possibly because of confounding inter/intra-annual oscillations in body condition. Discriminant rate decreased when controlling the researcher effect (72.7% and 71.4%, respectively). Results were also confirmed by logistic (mixed) models. Simulations showed that reduction in sample size markedly increased uncertainty in classification power. Differences with previous studies achieving a slightly greater classification power for sexing Adélie penguins through morphometrics could be related to (i) our higher sample size and/or (ii) geographical, population-specific differences, e.g. lower degree of sexual dimorphism. Our work emphasises key-factors influencing performance of morphometric sexing in avian species, advocating population-specific validation with large sample size as well as the necessity to account for the researcher effect prior to apply morphometric sexing.

对弱/严重二态性鸟类进行性别鉴定需要侵入性技术或分子分析。或者，性别鉴定可以基于形态测量分析，这在极地地区等偏远野外条件下仍然非常宝贵。然而，歧视能力可能会受到阻碍研究之间/内部比较的方法论问题的影响，例如考虑不同研究人员采用的替代分析技术或测量。我们调查了成年阿德利企鹅Pygoscelis adeliae的喙长度、喙深度和体重的性别潜力已知性别。我们对 1994 年至 2001 年间在埃德蒙森角（南极洲罗斯海）标记的企鹅的大型测量数据集（237 只雄性；264 只雌性）进行了判别分析。第二步，我们考虑了不同研究人员通过置换判别分析收集的测量结果。我们通过账单测量发现了对性别歧视的适度支持 (77%)。考虑到体重并没有显着提高分类能力（78.2%），可能是因为身体状况的年际/年内振荡混杂。控制研究者效应时判别率下降（分别为 72.7% 和 71.4%）。结果也得到了逻辑（混合）模型的证实。模拟表明，样本量的减少显着增加了分类能力的不确定性。与之前的研究相比，通过形态测量学对阿德利企鹅进行性别分类的分类能力稍大的差异可能与 (i) 我们的样本量较大和/或 (ii) 地理、种群特异性差异有关，例如性别二态性的程度较低。我们的工作强调影响鸟类物种形态计量性别鉴定性能的关键因素，提倡使用大样本量进行特定人群验证，以及在应用形态计量性别鉴定之前考虑研究人员影响的必要性。