Discussion

Using a unique detailed concomitant monitoring of both prey and predators, we showed that predators adjusted their foraging behaviour to diel prey distribution patterns (PDPs) and light levels. We found that while the two predators’ behavioural response to PDPs followed similar curves, there were interesting differences in response to light levels that could be explained through the PDPs (e.g. the difference in timing of effort, period of elevated efficiency and depth of dives). WhileU. aalge dive behaviour was symmetrical around noon, reflecting the depth distribution of prey, the A. torda efforts rather reflected an inverted curve of aggregation numbers, and with next to no dive activity in low light levels despite favourable prey distribution. The predator’s niche partitioning is therefore most likely a result ofU. aalge adaptations to low light conditions (Regular, Hedd and Montevecchi, 2011) which allows them to exploit a wider range of PDPs, both in time and depth, while A. torda appear more dependent on high light levels overall. However, A. torda ’ ability to forage efficiently for an extended period throughout the afternoon and dusk allowed them to meet needs before their efficiency is reduced by late night light conditions. Despite the species differences in response to light, both achieved high efficiency under conditions of shallow prey distributions and intermediate aggregation numbers. Our results highlight shared strategies in adapting to the dynamic prey landscape and niche partitioning in ability to utilize different prey distributions.

1 Predator niche partitioning

Despite their dive capacity (Chimienti et al. , 2017), U. aalge were most efficient when prey was shallow and/or aggregation numbers were low, particularly under low-light conditions. This may be partially explained by the metabolic constraints of longer, deeper dives, which increase recovery times and thus reduce efficiency of dives (Walton, Ruxton and Monaghan, 1998). A. torda , on the other hand, achieved peak efficiency at intermediate prey depths while being more efficient with the deeper of the utilized depth distributions if aggregation numbers were low, indicating that high numbers of aggregations were the main driver for decreased efficiency. This is reflected by their foraging under low azimuth levels, where both species reached peak foraging effort and efficiency well before maximum prey aggregation numbers in the morning, and peaked only after the aggregation numbers had declined in the afternoon. This, along with the dive depth responses suggests that foraging under intermediate and higher number of aggregations during dawn and mornings is likely driven by a trade-off between shallow depths of prey and suitable light levels, rather than by targeting intermediate levels of aggregations. U. aalge in addition tended to dive deeper at higher numbers of aggregations suggesting they adapted hunting tactics based on the PDPs. Potentially, U. aalge has dive- (Schneider and Piatt, 1986; Ponganis, 2015; Chimienti et al. , 2017) and visual (Smith and Clarke, 2012) capacities that allows them to utilize deeper depths with less aggregated prey if needed. Particularly interesting was the finding that in the evenings A. torda foraging effort tended to decrease before efficiency had reached its peak, likely reflecting a combination of state-dependent urgency in foraging (Houston and Rosenström, 2024), lack of predictability of good foraging patches (Bednekoff and Krebs, 1995; Houston and Rosenström, 2024) and the constraints by light availability. Early afternoon foraging efforts could secure resources under suboptimal conditions in anticipation of peak efficiency later in the afternoon, thus avoiding the risk of low-reward, opportunistic foraging after dark (Houston and Rosenström, 2024) when site-to-site orientation may be difficult. Despite both species having the dive capacity needed for all available depths in the study area (Piatt and Nettleship, 1985), A. torda ceased diving at relatively shallow prey depths as compared to theU. aalge and to A. torda in other systems (Barrettet al. , 1990). This is likely influenced by the poor light conditions and turbidity in the Baltic Sea (Murray et al. , 2019), rather than their adaptations regarding pressure at depths (Ponganis, 2015). The predators dove deeper with higher light levels, when prey was highly aggregated and while depth distribution was high. However, if there were more aggregations while prey depth was deep, both species would dive even deeper than before according to the interaction. As aggregations, particularly in high numbers tended to appear at relatively shallow depths, in the period of predator tracking, this response again suggests higher numbers of aggregations were unfavourable to both predator species, leading to strategies specific to aggregations. Much of the deviance in bout length was explained by light levels and prey distributions, where in A. torda bout lengths increased when there were many aggregations or distribution was deep, while for U. aalge the increase was by large attained high numbers of aggregations. It is however not possible to distinguish whether long bout lengths reflect an increased effort due to favourable foraging conditions or decreased success rate, and the two scenarios may not be mutually exclusive. In summary, U. aalge were more flexible in their utilization of prey distribution patterns through exploitation of a broader range of light levels and depths, and may thus be more robust to focal changes within a foraging site or range of colony during breeding.

2 Symmetry and synchrony of PDPs

The twilight bound aggregation peaks observed were likely due to aggregations formed during the vertical migration of clupeids (Zwolinskiet al. , 2007; Solberg and Kaartvedt, 2017). The differences in numbers of aggregations during the morning versus evening (25% reduction) could be explained by state dependent behaviours (Lima and Dill, 1990) during diel vertical migration. During the morning descent phase (<120° azimuth) the fish may have foraged all night (Nilsson et al. , 2003), and should prioritize energy-saving group swimming (Weihs, 1973), or even anti-predation behaviour by aggregating(Brock and Riffenburgh, 1960). This is perhaps particularly true for planktivores that remains in shallower water during the day after the vertical migration of zooplankton to deeper depths (Bollenset al. , 2011). The lower aggregation peak in the afternoon (>250° azimuth) would then reflect prioritized foraging (Lima and Dill, 1990), an activity which is likely to start already during the ascent (See Appendix A4). The asynchrony between aggregation and depth migration supports this interpretation, since numbers of aggregations started before the descent and decreased as vertical migration was finalized. Interestingly, this asymmetry created a temporal window of elevated foraging conditions in the late afternoon and dusk for predators that perform better with low levels of aggregations and better visual conditions. Increased foraging activity in afternoon/dusk (i.e. vespertine preference) compared to dawn (i.e. matutinal preference), as seen here in A. torda, is found in a range of crepuscular taxa (Gupta et al. , 2023). Though this has been explained largely by state dependency, cost of movement and risk of starvation, this study identify nuanced spatiotemporal prey behaviour as an underlying mechanism. Previous studies have shown how U. aalge , utilises prey distributions at different hierarchical scales, but were unable to determine the dynamics between prey distribution and predator foraging site at fine scale (<3km resolution) (Fauchald, Erikstad and Skarsfjord, 2000). We here present an alternative approach to disentangle such small-scale predator-prey dynamics with high temporal resolution. As prey aggregation patterns were highly variable throughout the study period (Appendix A4), seeing an adaptation to handling a specific level of aggregations would be impractical for the predator unless the pattern impacted fecundity drastically. This may partially explain why aggregations seem to have a higher impact on foraging efficiency, effort and strategy across predator species. It should be noted that the number of aggregations predicted on the birds data were modest compared to the average estimates from the initial prey distribution models, which again was modest compared to some of the aggregation patterns recorded. This is due to most bird tracking being performed later in summer when aggregations activity in prey were lower. It has been unclear whether aggregations serve an advantage (e.g. by prey detectability) or a disadvantage (e.g. reduced prey catchability by confusion effect) to diving predators, and in particular small seabirds with limited dive capacity such as the alcids (Lett et al. , 2014; Thiebaultet al. , 2016). Our study strongly suggest hunting under high aggregation is a focal disadvantage for both species, but our measurement for aggregations here were highly simplistic. Future studies should aim to take more complex aggregation characteristics (i.e. density) and presence-absence into account. Further, as we cannot distinguish the prey species or age/size classes in our prey data, the effects could be prey type related.

3 Conclusion

Light is a major driving factor of the predator niche differentiation, where U. aalge have a better ability to hunt in low light conditions as compared to A. torda . We were able to explain the higher dive efficiency and effort by predators during dawn and afternoon/dusk through temporal nuances in prey distribution patterns. Both predators performed better under shallow prey distribution with intermediate numbers of aggregations, but with differences in ability to forage under low-light conditions segregating them in timing of foraging and depth in the water column. U. aalge showed ability to utilize a wider range of light and prey distribution patterns than A. torda. The asymmetry in predator’s behaviour around solar angle (i.e. azimuth) was explained by PDPs both directly and indirectly through timing by light. While the deep diving U. aalge utilized the light levels during prey’s vertical migrations more, the A. torda was more affected by dynamics in the numbers of aggregations. The wide range of conditions utilized by U. aalge could give them an advantage under changes in PDPs, and perhaps even prey type, by season, or population impacts like fisheries and climate change, buffering them for their limitations in flight adaptation as compared to A. torda .