Historical comparisons of body size often lack pertinent details, including information on the sampling protocol and relevant ecological covariates that influence body size. Moreover, historical estimates of body size that rely on museum specimens may be biased towards larger size classes due to collector preferences, and thus size thresholds have been used to focus attention on maximum body size. We tested the consequences of sampling design, ecological covariates, and size thresholds on inferences of body size change using field-contextualized historical records, rather than museum specimens. In 2014-2015, we revisited historical (1947-1963) size-frequency distributions of three gastropods (Tegula funebralis, Lottia digitalis / L. austrodigitalis, Littorina keenae) in the context of population density and tidal height. In general, gastropods declined in size. However, our inferences regarding body size decline were tempered when the variation between sampling units was taken into consideration, resulting in greater uncertainty around the estimate of proportional change in body size. Gastropod size was correlated with population density and tidal height, and these relationships varied over time. Finally, the magnitude and direction of body size change varied with the amount of data available for analysis, demonstrating that the use of size thresholds can lead to incomplete conclusions.

中文翻译：

---

体型的历史比较对数据可用性和生态环境很敏感

体型的历史比较通常缺乏相关的细节，包括采样协议和影响体型的相关生态协变量的信息。此外，由于收藏家的偏好，依赖于博物馆标本的身体尺寸的历史估计可能会偏向更大的尺寸等级，因此尺寸阈值已被用于将注意力集中在最大身体尺寸上。我们使用实地背景化的历史记录，而不是博物馆标本，测试了抽样设计、生态协变量和体型阈值对体型变化推断的影响。在 2014-2015 年，我们在人口密度和潮汐高度的背景下重新审视了三种腹足动物（Tegula funebalis、Lottia digitalis / L. austrodigitalis、Littorina quietae）的历史（1947-1963）大小频率分布。一般来说，腹足动物的体型变小了。然而，当考虑到抽样单位之间的变化时，我们对体型下降的推论被缓和了，导致对体型比例变化的估计存在更大的不确定性。腹足动物的大小与种群密度和潮汐高度相关，这些关系随时间而变化。最后，身体尺寸变化的幅度和方向随着可用于分析的数据量而变化，这表明使用尺寸阈值可能会导致不完整的结论。腹足动物的大小与种群密度和潮汐高度相关，并且这些关系随时间而变化。最后，身体尺寸变化的幅度和方向随着可用于分析的数据量而变化，这表明使用尺寸阈值可能会导致不完整的结论。腹足动物的大小与种群密度和潮汐高度相关，并且这些关系随时间而变化。最后，身体尺寸变化的幅度和方向随着可用于分析的数据量而变化，这表明使用尺寸阈值可能会导致不完整的结论。