Results
The δ13C values ranged from -17.58 to -11.46‰ and the δ15N values ranged from 6.92 to 8.66‰ (Fig 2., see Supplementary Table 2 for full dataset).
The experimental fish left the release compartment within 10-min in most trials. The results of the GLMM showed that as a group the experimental fish left the compartment quicker with subsequent trials, but activity did not differ across days of the experiment (Table 1). Similarly, there was little variation among individuals in activity but high in latency to explore (Table 1).
The results from the generalized linear model of the estimated intercepts and slopes of behavioral traits (latency to explore and activity) and stable isotope values showed that the intercept of latency to explore was correlated to δ13C values of both fin and muscle. The effect was not large, R2 for both tissue types was 0.16, p = 0.02 (see also Fig. 3). Other correlations were not significant.
The calculated Euclidean distance between fin and muscle stable isotope values varied between fish with a mean value of 0.74 and SD = 0.39. The slope of latency to explore, as well as the interaction between slope and intercept, explained variation in the distance (Table 2, Fig. 4). The model R2 was 0.23 suggesting that the model moderately reflected variation in the stable isotope distance. No significant result was observed for activity.
Discussion The current result show that latency to explore, a common proxy for boldness, assessed in a rapid laboratory array, correlated to retrospective measures of threespine stickleback trophic niche. Specifically, bolder fish had lower δ13C values (Fig. 3) and their stable isotope niche changed more over time (Table 2). These results are among the first to support such a relationship.
The δ13C values of bolder stickleback were consistently lower in both muscle and fins (Fig. 2). Comparing isotopic values of multiple tissues allows estimation of the relative temporal stability of the individual niche (Phillips & Eldridge, 2006; Maruyama et al., 2017). It is currently uncertain how much threespine stickleback fin and muscle differ in turnover rates but estimates for bony fish commonly vary on the scale of a few weeks to months (Willis et al 2013; Busst & Britton 2018; Barton et al., 2019; Winter et al., 2019). The stickleback fin and muscle stable isotope values were highly correlated (Fig. 3). The current study did not include δ13C values of potential prey items within the lake, making inferences on how these patterns relate to the diet of the fish difficult. However, based on previous studies of stable isotope ecology in northern freshwater lakes the lower δ13C values are expected to indicate more pelagic foraging or foraging on invertebrates that rely on pelagic production (France, 1995). Lower δ13C values of bolder individuals could therefore be consistent with a “landscape of fear” (Gaynor et al., 2019), that shyer individuals keep to the more sheltered benthic habitat. It should be noted that in the current experimental setting the intercept will highly reflect initial boldness (latency to explore during the first trial). Both in the current experiment and previous studies stickleback from this population have been found to adjust or habituate to the experiment as many individuals would initially be slow to explore but become quicker in subsequent trials (Ólafsdóttir & Magellan, 2016; Ólafsdóttir et al., 2014). Therefore, the current results should be taken to indicate that initial shyness/boldness is correlated to δ13C values, whether those individuals were consistently shy/bold or adjusted their exploratory speed to context.
Boldness was also linked to change in stable isotope niche over time, that is, to the difference between δ13C and δ15N values in fin and muscle. Both the estimated slope of latency to explore and the interaction of intercept and slope had significant effect on the change in individual stable isotope niche (Table 2). High positive slope (individuals that were initially shy but progressively bold) was correlated to less change in stable isotope niche (Table 2). This is also reflected by the significant interaction effect and is, at least in part, explained by that initially shy individuals will most often have (steep) negative slopes and vice versa. Visualization of these effects can be seen in Fig. 4 and may indicate that the pattern is largely driven by a few individuals. These results highlight that the correlation of trophic niche and behavior can be nuanced but also the need to consider not only the average behavior but also plasticity when relating animal personality to foraging (Coomes et al., 2021).
Based on the current data it is not possible to fully distinguish between three underlying scenarios. First, innate boldness, as a general trait expressed consistently in the laboratory and in the field, results in different foraging patterns and those foraging patterns are reflected in stable isotope values. Second, previous diet was unrelated to innate behavior but induced the observed differences in behavior in the laboratory, for example, through nutrient state or condition. The experimental fish were deliberately chosen to be of approximately the same size and weight, but more detailed variation of condition, fatty deposits, lipid content etc., was not considered. Should previous diet affect behavior the timing of behavioral test could highly determine the results, for example, in the current study behavioral test were conducted with only a single day for fish to adjusting to laboratory conditions. This was necessary to limit the effect of the identical laboratory diet on stable isotope values but could have augmented variation on the boldness-shyness axis, as acclimatization has been found to affect measures of behavioral traits (Biro, 2012; O’Neill et al., 2018) but also resulted in increased variation in nutrient state. Third, both growth rates and standard metabolic rate have been linked to stable isotope turnover rate and are therefore reflected in stable isotope values (Gorokhova, 2018; Scharnweber et al., 2021), although these factors may explain less of the values than diet (Johnston et al., 2021). Varying growth rates of the stickleback in the wild prior to capture could have relevance for the current results, both in that fast-growing individuals would be both bolder and with lower δ13C values, without any causal link, but boldness and growth could also have been correlated in the wild. It has been previously supported that boldness, metabolic rate and growth could be connected in a “pace of life” syndrome (Réale, 2010). Finally, it should be noted that overall the relationship of behavior and stable isotope niche in the current study was small and that no significant correlations were found for activity or δ15N.