Daniel Pauly coined the expression fishing down the food web in a highly influential paper published in 1998 (Pauly et al. 1998). In it, he and his colleagues described how the average trophic level of fish recorded in FAO landings statistics was in decline in many parts of the world. They argued that fishing down food webs occurred because predatory species higher in food webs tend to be depleted first, causing fishers to switch to other, more abundant species, which tend to come from lower in the food web. Predators are targeted first because they are often large, bold and voracious, and therefore easy to catch, have firm tasty flesh and are valuable. Over time, overfishing necessitates a succession of such switches as ecological communities are stripped of their predators. Paulys idea has been challenged on several grounds. Essington et al. (2006) contend that fishing takes place through food webs rather than down, so fisheries for species lower in food webs are added while those for top predators continue, albeit at lower levels than before. Sethi et al. (2010) suggest that it is more accurate to view fishing down in economic terms with the most valuable species targeted first and then lower value species added as these are depleted. Since many of the most valuable species are top predators the effect also appears as fishing down the foodweb. Finally, some authors contend that FAO landings statistics show a fishing down phenomenon that is less apparent in biomass surveys of fish at sea (Branch et al. 2010). However, abundant historical data indicate that fishing down the food web cannot be explained away so easily. There is compelling evidence from just about every sea and gulf in the world that large-bodied species, usually predators, have fallen to levels far below historical maxima (Jackson et al. 2001, Roberts 2007). In some places, such as Scotlands Firth of Clyde, the same effects have been demonstrated using both landings data and biomass surveys (Thurstan and Roberts 2010, Heath and Speirs 2011). Pauly et al. (1998) expressed fishing down the food web as taking place over time. The same phenomenon can also be observed over space (Figure 1). Where there are geographic gradients in fishing intensity, similar habitats support very different fish assemblages. Within regions like the Caribbean, the differences owe far more to the intensity of fishing than to biogeography or variation in habitat. A few years ago, my wife Julie Hawkins and I counted fish at six islands across the Caribbean. We found that the biomass of predatory fish fell ten-fold from the lightly fished reefs of Bonaire to the intensively exploited reefs of north Jamaica (Hawkins and Roberts 2004, Figure 2). So did the biomass of herbivorous fish, although the decline through sites with intermediate fishing intensities was less steep. Non-target bycatch species like butterfly and angelfish also declined as fishing pressure rose (Hawkins et al. 2007). Within fish families, species declined across the gradient of fishing pressure in sequence of body size. Many had disappeared altogether by the time fishing pressure reached Jamaican intensities, while others were represented only by juveniles that had probably recruited from distant, less-fished reefs (Hawkins and Roberts 2003). Newman et al. (2006) replicated these findings for a different set of study sites in the Caribbean, while Stevenson et al. (2006) showed a similar phenomenon on coral reefs of the Pacific Line Islands. In the Caribbean, Hawkins and Roberts (2004) also saw a gradient of declining coral cover, reduced habitat structural complexity and increasing algal cover as fishing intensities went up. With herbivorous fish ten times less abundant at the most fished sites, this is hardly surprising. But it shows how fishing has effects that cascade through the ecosystem to cause profound differences in structure and function (Estes et al. 2011). Although it was the least fished site in our study, Bonaire was far from unaffected by fishing. A brief glance at the historical record revealed that Bonaire was once home to abundant reef sharks and several large species of grouper, including the goliath grouper (Epinephelus itajara), that were absent by the time of our survey (Hass 1952). They had already fallen victim to overfishing. The baseline from which we inferred the impact of exploitation was not really a baseline at all. The same is true of virtually anywhere else you care to look in the region. Old photographs of fish catches show how big animals like groupers were once far more abundant on Caribbean reefs (McClenachan 2009). In a fascinating piece of historical sleuthing, Loren McClenachan and Andrew Cooper used the extinct Caribbean monk seal to infer how many fish the regions reefs might have supported hundreds of years ago (McClenachan and Cooper 2008). They used a wide range of historical sources, including accounts written by pirates, complemented with population modeling to piece together how many monk seals there were in the 17th century Caribbean. Historical observa-tions showed the seal was widespread from the Gulf of Mexico to South America. The authors estimated there were 233,000 to 338,000 seals distributed throughout this range. Such a large population would have required a lot of fish. Based on food intake rates by the ecologically similar Hawaiian monk seal, McClenachan and Cooper calculated that Caribbean reefs would need to have sustained abundances of fish equivalent to 700 to 1,000 g/m2 to support all these seals. This is higher than the biomass reported for any Caribbean reef today and similar only to levels attained in remote, uninhabited Pacific reefs like Palmyra (McClenachan and Cooper 2008).