Do oystercatchers (Haematopus ostralegus) select the most profitable limpets (Patella spp.)?

• Eurasian oystercatchers resident on Lundy (SW England) feed mainly on limpets.
• Common prey includes Patella vulgata and Patella depressa, usually attacked anteriorly.
• Methods used include parallel striking, levering, hammering and pulling by weed.
• Smaller (20–30 mm long), deeper, and irregular-bottomed limpets are selectively attacked.
• The birds' preference for deeper limpets reverses in longer size classes.

Eurasian oystercatchers (Haematopus ostralegus) feed on at least 15 different prey species in different habitats along the NW European coast. In each habitat, they specialise mostly on a single molluscan prey and within that prey type, they use various strategies to make foraging more profitable depending upon a number of internal and external factors. Oystercatchers resident on Lundy Island (SW England) feed mainly on limpets (Patella spp.). We found that the oystercatchers attacked both the common limpet, Patella vulgata and the black-footed limpet, Patella depressa, in any of four ways parallel striking, levering, hammering or pulling by weed. Oystercatchers took relatively deeper-shelled limpets in the 20–40 mm length class, but this tendency reversed in limpets > 40 mm long. The oystercatchers showed a selective preference for smaller, deeper, irregular-bottomed limpets, and attacked them on the anterior region. The oystercatchers showed a strong preference for 20–30 mm long limpets (Jacob's preference index relative to other length classes D = 0.85) and increasing avoidance to larger length classes. But the calculated profitability for successfully opened limpets increased monotonically with the length, and the largest limpets had the highest profitability. Hence, the oystercatchers apparently did not select the most profitable limpets according to the simplest rate maximising optimal diet model; however we argue that this is because handling time varies as a function of substrate condition, so that a more complex model is needed.