Abstract. Growing evidence that individuals of many generalist animals behave as resource specialists has attracted research interest for its ecological and evolutionary implications. Although variation in resource preferences is critical for developing a general theory of individual specialization, it remains to be shown whether diverging preferences can arise among individuals sharing a similar environment and whether these are stable enough to be ecologically relevant. We addressed these issues by means of common garden experiments in feral pigeons (Columba livia), a species known to exhibit resource specialization in the wild. Food-choice experiments on wild-caught pigeons and their captive-bred descendants showed that variation in food preferences can easily arise within a population and that this variation may represent a substantial fraction of the population niche. However, a cross-fostering experiment revealed that the genetic and early common-environment components of food preferences were low, reducing their stability and eroding niche variation in the long-term.
Keyword: Niche variation; individual differences; decision-making; learning the niche; behavioural plasticity; heritability of behaviour; cross-fostering experiment.
Niche variation—in which individuals consistently use a subset of the resources available for the population (Van Valen 1965)—is a widespread phenomenon in vertebrate and invertebrate taxa (Bolnicket al. 2003). Although much of this variation can be explained by either ontogenetic niche shifts or sexual dimorphism (Schoener 1986; Price 1987), resource-use variation is observed even among individuals of the same age and sex (Araújo et al. 2011). In Brünnich’s guillemots (Uria lomvia ), for example, individuals show foraging specializations in diving technique, which are then reflected in differences in the type of fish and marine invertebrates they mostly capture (Woo et al. 2008). This type of niche variation —known as individual specialization— has attracted considerable interest for its ecological and evolutionary implications (reviewed in Bolnicket al. , 2003, 2011; Dall et al. , 2012). Thus, individual specialization has been related to a wide array of ecologically and evolutionary relevant processes, including skill-pools, population stability, species coexistence, extinction, niche evolution and speciation (Giraldeau 1984; Sol et al. 2005; Tinker et al.2008; Bolnick et al. 2011). Surprisingly, however, how individual variation in resource use is generated and maintained within populations is still insufficiently understood (Araújo et al. 2011; Bolnicket al. 2011). As a consequence, the actual relevance of individual resource specialization in ecology and evolution remains unclear.
The existence of variation among individuals in resource preferences is central for developing a general theory of individual specialization. As pointed by Bolnick et al. (2003), if a population is found to be composed of individuals that consistently prefer different subsets of available foods, few would object to considering such individuals specialists. However, why should individuals sharing a same environment end up using only a subset of all available resources? Classic optimal foraging theory predicts that organisms’ food choices should maximize some currency linked to fitness, like energy intake (Pyke 1984; Shettleworth 1985). It follows that if individuals behave optimally, they should converge in the way they use the resources rather than diverge.
Theoretical models suggest, however, that specialization may still occur if preferences are state-dependent, that is, if the decisions of individuals are modulated by their morphological, physiological and psychological features (Houston & McNamara 1999). Araújo et al. , (2011) discussed two main ways how this can occur. One is when phenotypic variation produces among-individual differences in the ability to detect, capture, handle, or digest alternative prey. The second is when individuals use different optimization criteria in resource-choice, for example because they have different energetic requirements (Schoener 1971). Both mechanisms may not only cause individuals to prefer some resources over others, but they may make preferences to be transmitted to offspring if the state-traits that have driven them are themselves heritable. Thus, although preferences can be extremely plastic, they can change more slowly if optimal choices are connected to a slow-changing state variable (Luttbeg & Sih 2010).
Resource specialization may arise in yet another way, that is, through learning (Tinker et al. 2009). Although behaviour, physiology and morphology may set the limits to what an individual can eat, food preferences within these boundaries may arise through positive feedbacks (sensu Dingemanse & Wolf 2010), where early experiences in the use of certain foods reinforce (if positive) initial choices (Dridi & Lehmann 2015). A variety of situations may make individuals to have distinct early food experiences, including different perception of risk or spatially and/or temporally variation in resource availability. There is indeed growing evidence that learning is crucial for animals to develop their own niche (Slagsvold & Wiebe 2007; Tinker et al. 2009). Given that resource preferences will be environmentally induced rather than constrained by the genetic architecture, however, it is not obvious how learned preferences will be maintained within a population. One would expect that as individuals are exposed to positive food experiences, their preferences will continue to expand rather than remain narrow. Yet, learning may still cause stable resource preferences if adopting new foods is cognitively costly, for example because it requires to learn a new foraging technique (Partridge & Green 1987; Tinker et al. 2009). Learned preferences may also be maintained if they are phase-sensitive, that is, if they occur at a young age and after that individuals acquire an aversion to explore and incorporate new foods (Greenberg & Mettke-hofmann 2001; Slagsvold & Wiebe 2007).
Despite the importance of resource preferences in constructing a general theory of individual specialization, there has been little effort to assess whether individuals within populations consistently vary in resource preferences and to what extent this variation remains stable over time (Araújo et al. 2011; Bolnick et al. 2011). One reason of this neglect is that resource preferences are difficult to quantify. A major confounding effect is resource availability (Moon & Zeigler 1979; De Cáceres et al. 2011). Preference is the likelihood that an individual selects a given item when offered alternative choices on an equal basis (Johnson 1980). If a particular food is relatively scarce or is monopolized by a superior competitor, it may represent only a small proportion of the resources used even if the animal has a high preference for that food. Preference also implies that the likelihood of selecting a given item is higher than expected by chance. This means that resource preferences need to be measured in standardized replicated observations, which may also be challenging (Araújo et al. 2011). On the other hand, although one can measure individual differences in resource preferences, the challenge remains as to how interpret the differences. Given that individuals may use different optimization criteria to rank foods, interpreting preferences requires to consider the energetic and nutritional contents of food (Machovsky-Capuska et al. 2018). It also requires to consider the multi-dimensional nature of foraging behaviour, which includes the search, identification, handling, consumption and digestion of foods.
Perhaps the most tractable way to tackle the difficulties of assessing and interpreting food preferences is the use of common garden experiments. Despite concerns over the extent to which the mechanisms we observe in these experiments tap into those actually used by animals in the wild, food-choice experiments involving a variety of food types in exactly the same amount remain the best way to estimate individual preferences under different optimization criteria and reducing the complexities associated with the multi-dimensional nature of foraging behaviour (Pyke 1984). Combined with cross-fostering breeding experiments —where the effects of rearing environment and genetic influences may be disentangled (Slagsvold & Wiebe 2007)— food-choice experiments offer the opportunity to investigate 1) whether individuals exhibit consistent differences in resource preferences and 2) to what extent these differences have a genetic basis or are environmentally-induced. Here, we address these questions in the Feral pigeon (Columba livia ), a granivorous bird that has been instrumental in the study of decision-making theory (Inman et al.1987; Shettleworth 1987a) and that in the wild is known to exhibit individual differences in food use (Giraldeau & Lefebvre 1985; Johnston & Janiga 1995). Although previous experiments in pigeons have found evidence for a unitary preference for certain seeds (Moon & Zeigler 1979; Inman et al. 1987; Shettleworth 1987a; Biedermann et al. 2012), some authors have suggested that individuals might also vary in their preferences (Moon & Zeigler 1979; Giraldeau & Lefebvre 1985).
To investigate whether resource specialization reflect variation in food preferences, we conducted replicated food-choice assays in a common garden framework using pigeons from two wild populations. In addition, we conducted food-choice assays on their captive-bred descendants—half of them cross-fostered among nests—to estimate the heritability, vertical transmission and reactions norms of the preferences by means of a quantitative genetics approach. Finally, we repeated the food-choice assays one year later to assess the long-term stability of food preferences. Following De Cacéres et al. (2011), the contribution of niche specialization to the total niche width of the populations was assessed based on a comprehensive framework that extends classic information theory by incorporating the energetic and nutritional resemblance between resources into the calculation of resource niche metrics. Characterizing the niche exclusively in terms of quantity and range of different resources would ignore that individuals may use different optimization criteria to rank foods. Moreover, it would bias estimation of niche breadth as the niche of individuals consuming the same number of foods will be considered equivalent even when some individuals are using more different types of foods (Colwell et al. 1971; De Cáceres et al. 2011).