4.6.5: Sampling Animals in Populations

Nearly all research animal biologists have, at some point, sampled animals from the wild. These researchers strive to obtain samples of individuals from a given population by employing strategies that minimize the possibility for systematic biases. Some strategies include sampling at varying times during the day and in many different areas. Researchers also take care to minimize selectivity for animals of a particular size, age, or maturity (MacKenzie and Kendall 2002; Askey et al. 2007; Cox et al. 2024). However, biases due to the individual characteristics of each animal can affect the likelihood of being captured or detected in a biological study. This is an important issue because it is crucial for biologists to observe animals from the correct population, otherwise conclusions based on the observed data could be misleading (Biro 2013).

For example, consider a biological study of fish in a local lake. Of interest is to get a good idea of how much the average bass weighs. To perform the study the researchers get a net a drag it across the lake. When the net is pulled up the bass are weighed, and the fish are released. The crucial question is whether the bass collected using this method are taken from the whole population of bass in the lake, or if there is some bias in the collection method. Consider, for example, if the holes in the net were rather large. Then one could easily imagine that smaller bass could easily slip through the net and would tend not to be caught. The results from the study would then be biased, concluding that the average bass in the lake weighs more than the true value.

Research has shown that some animals are more likely to be captured or detected than others in a population due to behavioral differences (Crowcroft and Jeffers 1961; Burnham and Overton 1978; Wilson et al. 1993; Gérard et al. 1994; Palencia et al. 2021). In fact, these behavioral differences may not be dependent on the size, age, or maturity of the animals in question (Sih et al. 2004; Reale et al. 2007; Sih and Bell 2008). Active, bold, or exploratory animals are likely to be captured or detected, though this is not universal (Boon et al. 2008; Biro and Dingemanse 2009; Carter et al, 2012; Garamszegi et al. 2009; Guillette et al. 2009; Gabriel and Black 2010; Harrison 2025; Johnstone et al. 2021; Pellegrini et al. 2010; Wilson et al. 2011).

Bias with respect to animal behavior is a concern for any biologist studying animal physiology or ecology because animal behavior directly affects catchability, or whether an individual is observed, and animal behavior is often associated with the traits of interest (Stamps and Groothuis 2010). This can intoduce bias in the conclusions of studies of animal populations. Animal behavior capture bias can also affect the strength of the associations observed between the traits of interest and the associated animal behavior (Garamszegi et al. 2009; Carter et al. 2012).

One study directly researched animal behavior capture bias in a unique way (Biro 2013). In this study four natural lakes with no fish in them and no natural or human predators were stocked with rainbow trout that had three different growth rates (slow, intermediate, and fast). At the end of the season, fish were sampled from the lakes using experimental nets designed to eliminate bias due to the size of the fish. The researcher predicted that the fast-growing fish, independent of their size, would be more likely to be captured in nets. This prediction is based on research that shows that activity rates, boldness, and growth rates are positively associated in fish and other animals (Stamps 2007; Biro and Stamps 2008; Guenther 2018). Hence, the researcher predicted that the activity and the boldness of the faster growing fish would cause them to be more easily captured.

The data collected from the study indicated that the fast-growing fish were most likely to be captured in all lakes included in the study. In two of the lakes the likelihood of being captured increased with growth rate, while in the remaining lakes, the intermediate-growing fish had the lowest likelihood of being captured. Using statistical techniques, the researchers were able to show that these conclusions held even when the accounting for potential confounding effects. The authors further noted that fish were more likely to be captured in two of the lakes but that the data indicated that the mean growth rates in all the lakes were similar. Therefore, while it was easier to capture fish in some of the lakes, this effect is not confounded with the growth rate of the fish.

This study emphasizes the importance of carefully considering the practical aspects of observing individuals from a population. Even though the observation process was very carefully constructed so that there was no size bias in the fish that were sampled, the resulting observations did not correspond to all the fish in the associated lakes. Instead, the true population was fish that were likely to be caught because of their behavior patterns. As the researchers show, these fish can differ in many important ways from the fish in the entire population.