Roi Holzman

Throughout evolution, organisms repeatedly developed elastic elements to power explosive body motions, overcoming ubiquitous limits on the power capacity of fast-contracting muscles. Seahorses evolved such a latch-mediated spring-actuated (LaMSA) mechanism; however, it is unclear how this mechanism powers the two complementary functions necessary for feeding: rapidly swinging the head towards the prey, and sucking water into the mouth to entrain it. Here, we combine flow visualization and hydrodynamic modelling to estimate the net power required for accelerating the suction feeding flows in 13 fish species. We show that the mass-specific power of suction feeding in seahorses is approximately three times higher than the maximum recorded from any vertebrate muscle, resulting in suction flows that are approximately eight times faster than similar-sized fishes. Using material testing, we reveal that the rapid contraction of the sternohyoideus tendons can release approximately 72% of the power needed to accelerate the water into the mouth. We conclude that the LaMSA system in seahorses is powered by two elastic elements, the sternohyoideus and epaxial tendons. These elements jointly actuate the coordinated acceleration of the head and the fluid in front of the mouth. These findings extend the known function, capacity and design of LaMSA systems.

Understanding how organismal traits determine performance and, ultimately, fitness is a fundamental goal of evolutionary eco-morphology. However, multiple traits can interact in non-linear and context-dependent ways to affect performance, hindering efforts to place natural populations with respect to performance peaks or valleys. Here, we used an established mechanistic model of suction-feeding performance (SIFF) derived from hydrodynamic principles to estimate a theoretical performance landscape for zooplankton prey capture. This performance space can be used to predict prey capture performance for any combination of six morphological and kinematic trait values. We then mapped in situ high-speed video observations of suction feeding in a natural population of a coral reef zooplanktivore, Chromis viridis, onto the performance space to estimate the population's location with respect to the topography of the performance landscape. Although the kinematics of the natural population closely matched regions of high performance in the landscape, the population was not located on a performance peak. Individuals were furthest from performance peaks on the peak gape, ram speed and mouth opening speed trait axes. Moreover, we found that the trait combinations in the observed population were associated with higher performance than expected by chance, suggesting that these combinations are under selection. Our results provide a framework for assessing whether natural populations occupy performance optima.