401 © The Ecological Society of America www.frontiersinecology.org Seabirds, fisheries, and cameras Peer-reviewed letter Who will win the fight over scarce sar- dines in the Benguela ecoregion? In this upwelling system off southern Africa, once vast pelagic (open- ocean) fish stocks have decreased sub- stantially as a result of inherent popu- lation variability, climate change, and fishing pressure (van der Lingen et al. 2006). Off the coast of Namibia, catches of sardines (Sardinops sagax) and anchovies (Engraulis encrasicolus) are now < 2% of those taken during the 1960s, and populations of endemic Cape cormorants (Phalacrocorax capensis), African penguins (Sphenis- cus demersus), and Cape gannets (Morus capensis), all of which depend on these pelagic fish, have declined to < 20% of levels recorded prior to industrial fishing (Crawford 2007). Off the coast of South Africa, sardine stocks declined from 4.2 million to 0.5 million metric tons (t) between 2002 and 2009, and pelagic fish showed substantial distribution shifts (Roy et al. 2007; Coetzee et al. 2008). The remaining seabirds and purse-seine fisheries from South Africa’s west coast are therefore bound to track the same scarce and elusive resources (Grémillet et al. 2008a). In this situa- tion, modern fisheries are usually assumed to outperform other marine predators, given that the former have the logistical capacity to locate and harvest shoaling fish, even at very low densities (Valdemarsen 2001). Marine protected areas (MPAs) closed to fish- eries have been proposed to protect the prey of vulnerable, endemic Benguela seabirds. Localized MPAs are being implemented on an experi- mental basis around some of the pen- guin colonies (Pichegru et al. 2010), yet are currently not considered for the protection of far-ranging Cape gannets, since this would require clos- ing the entire southern Benguela area to fisheries (Pichegru et al. 2009). Alternatively, spatially managed fish- ing quotas may help to keep sufficient fish in the system locally to feed vul- nerable marine predators. Here, we report on a case where compliance with fishing quotas established since the early 1980s allowed a colony of 70 000 Benguela seabirds to compete favorably for increasingly scarce sar- dines with a fleet of 79 purse-seiners. We studied Cape gannets breeding at Malgas Island, South Africa (Figure 1). Once the world’s largest breeding colony for this species, num- bers have declined by 40% since the late 1990s. Because of the recent scarcity of their preferred prey (sar- dines and anchovies) on the west coast, birds now mainly feed on fish- ery waste from hake (Merluccius spp) trawlers (Grémillet et al. 2008b). However, in November 2009, sar- dines and anchovies suddenly reap- peared in the gannets’ diet (78% and 10% of diet by mass, respectively). We tracked the at-sea movements of 21 breeding gannets using miniature global positioning system (GPS) units attached to their backs. Their main feeding areas broadly over- lapped with localized, low-density sardine aggregations detected during a concurrent hydroacoustic research survey (Figure 1). Data from the pelagic fishing fleet showed that, at the time of our investigations, purse- seiners targeted some but not all of the sardine patches, primarily off the fishing harbors of Hout Bay and Gansbaai (Figure 1), catching 356 t of sardines in total. This is < 1% of the maximum, and 3.5% of the aver- age, monthly catch made by the purse-seine fleet fishing for sardines around the south African coastline during the past 23 years. During November 2009, the 70 000 Cape gannets from Malgas Island therefore caught nearly three times as many sardines (equivalent to ~1160 t, assuming that 78% of the birds’ diet comprised sardines and that each bird required 710 g of fish per day; com- pare with Pichegru et al. 2007) as cap- tured by fisheries. To better assess the co-occurrence of gannets and fishing vessels at sea, we equipped five birds WRITE BACK WRITE BACK WRITE BACK Figure 1. Foraging zones (red color scale) of 21 Cape gannets breeding on Malgas Island (yellow star), relative to sardine aggregations (blue color scale) and sardine catches by fishing vessels (black circles) in October–November 2009 (maximum growth phase of the gannet chick; the most appropriate period to test seabird/fishery interactions). White squares indicate where the photographs displayed in Figure 2 (a and b) were taken. Each gannet was tracked for a single foraging trip by way of a GPS device recording position every second. Interruptions in the GPS signal of >1 second were due to plunge-diving behavior, and hence positions associated with these interruptions were used to identify feeding spots. We determined the spatial occurrence of sardines by hydroacoustics, using semi-stratified random sampling across the entire study area. Spatial distribution of sardine catch by purse-seine fishing vessels was compiled through log books, stating the catch volume, time, and position of each haul. Fishing catches (metric tons) <15 15 – 30 30 – 60 60 – 130 Proportion of time spent by foraging gannets per unit area (%) 50% 50 – 75% 75 – 90% Density of sardines (g cm –2 ) 0.01 – 0.1 0.1 – 0.3 0.3 – 0.5 0.5 – 0.7 0.7 – 1 1 – 10 0 100 kilometers