Adolph Stephen C, Pickering Trevor
Department of Biology, Harvey Mudd College, 301 Platt Boulevard, Claremont, CA 91711, USA.
J Exp Biol. 2008 Apr;211(Pt 8):1336-43. doi: 10.1242/jeb.011296.
Researchers often estimate the performance capabilities of animals using a small number of trials per individual. This procedure inevitably underestimates maximum performance, but few studies have examined the magnitude of this effect. In this study we explored the effects of intraindividual variation and individual sample size on the estimation of locomotor performance parameters. We measured sprint speed of the lizard Sceloporus occidentalis at two temperatures (20 degrees C and 35 degrees C), obtaining 20 measurements per individual. Speed did not vary temporally, indicating no training or fatigue effects. About 50% of the overall variation in speed at each temperature was due to intraindividual variation. While speed was repeatable, repeatability decreased slightly with increasing separation between trials. Speeds at 20 degrees C and 35 degrees C were positively correlated, indicating repeatability across temperatures as well. We performed statistical sampling experiments in which we randomly drew a subset of each individual's full set of 20 trials. As expected, the sample's maximum speed increased with the number of trials per individual; for example, five trials yielded an estimate averaging 89% of the true maximum. The number of trials also influenced the sample correlation between mean speeds at 20 degrees C and 35 degrees C; for example, five trials yielded a correlation coefficient averaging 90% of the true correlation. Therefore, intraindividual variation caused underestimation of maximal speed and the correlation between speeds across temperatures. These biases declined as the number of trials per individual increased, and depended on the magnitude of intraindividual variation, as illustrated by running sampling experiments that used modified data sets.
研究人员常常通过对每个个体进行少量试验来估计动物的性能能力。这一过程不可避免地会低估最大性能，但很少有研究考察这种影响的程度。在本研究中，我们探讨了个体内部变异和个体样本量对运动性能参数估计的影响。我们在两个温度（20摄氏度和35摄氏度）下测量了西部强棱蜥的冲刺速度，每个个体获得20次测量数据。速度在时间上没有变化，表明不存在训练或疲劳效应。在每个温度下，速度总体变异的约50%归因于个体内部变异。虽然速度具有重复性，但随着试验间隔的增加，重复性略有下降。20摄氏度和35摄氏度下的速度呈正相关，这也表明在不同温度下具有重复性。我们进行了统计抽样实验，在实验中我们从每个个体的20次完整试验中随机抽取一个子集。正如预期的那样，样本的最大速度随着每个个体的试验次数增加而增加；例如，五次试验得出的估计值平均为真实最大值的89%。试验次数也影响了20摄氏度和35摄氏度下平均速度之间的样本相关性；例如，五次试验得出的相关系数平均为真实相关性的90%。因此，个体内部变异导致了最大速度以及不同温度下速度之间相关性的低估。随着每个个体试验次数的增加，这些偏差会减小，并且取决于个体内部变异的程度，这在使用修改数据集的运行抽样实验中得到了说明。
