Two measures that can be easily estimated from fisheries landings have shown themselves to be highly indicative of the status of the underlying ecosystems, and thus could be used for such monitoring: the mean TL of fish landed and the mean ML of the species in the landings.
Eating and not being eaten is, besides reproduction, the main concern of organisms in ecosystems, and the latter can largely be described, therefore, as a meshing of food chains into complex food webs, within which an organism occupies a given position determined by its size, the anatomy of its mouth parts and its feeding preferences. One dimension of this position is the TL, expressing how many steps away an organism is located away from the base of marine food webs, i.e. phytoplanktonic and benthic algae, assigned a definitional TL of 1, the same as for detritus, mainly derived from ungrazed, dead algae and the excreta of herbivores (Odum & Heald 1975
Phytoplankton is grazed mostly by copepods and other small crustaceans, with a TL of 2, in stark contrast to terrestrial food chains, where the herbivores are often very large. The zooplankton, in turn is consumed mainly by small pelagic fishes (herring, sardine, anchovies), with a TL of approximately 3, the imprecision stemming from the fact that they often consume a variable mix of phytoplankton, herbivorous and carnivorous zooplankton, and detritus. Small pelagics are caught in enormous quantities (38 million tonnes in 2000, i.e. 44% of global marine landings), are either consumed by people (e.g. as canned ‘oil sardine’) or ‘reduced’ to fishmeal and oil, a key component of the chicken and pig feeds, and of farmed salmon (Naylor et al. 2000
). The typical table fish, however (cod, snapper, tuna, halibut, etc.), that restaurants serve whole, or as a steak or fillet, are predators on the small pelagics and other smaller fishes and invertebrates, and tend to have a TL of ca
. 4, with 4.5 an upper limit reached by large sharks (Cortés 1999
), bluefin tuna and other large predators such as some marine mammals (Pauly et al. 1998c
). TLs are variable in space and time, the latter variability referring both to seasons and to the age (size) of fishes. We shall ignore the issue of TL variability here, addressed in some detail in Pauly et al. (2001)
Important, also, is that in the sea, the high TL organisms tend to be larger (typically three to four times in terms of body length) than their prey (Ursin 1973
), and need more time to reach maturity and reproduce (Denney et al. 2002
), which renders them very susceptible to overfishing (e.g. Sadovy & Cheung 2003
In summary, we can conclude that, given the current technical ability to catch whatever marine species are abundant within an ecosystem, and the fact that large fishes are usually more valuable than smaller fishes, increased landings of fishes with lower TL imply a reduction of the abundance of the higher TL species. Or put differently, non-sustainable fishing should manifest itself, at the ecosystem level, in a gradual shift of mean TL towards lower values, even if the individual species for which TACs exist appear to be fished sustainably (Valtysson & Pauly 2003
This process, now known as ‘fishing down marine food webs’ (FD) was originally presented in 1998 based on the global database of landings created and maintained by the FAO, itself relying on data supplied by its member countries, some of them with only rudimentary fisheries monitoring systems (Pauly et al. 1998a
). In particular, these data tend to be over-aggregated in terms of species landed (i.e. many species are combined as ‘mixed fishes’), and areas covered (e.g. fishes caught by a distant-fishing nation in the EEZs of different countries are combined without indication as to their origins). Following a critique by Caddy et al. (1998)
suggesting that these defects of the FAO database may invalidate the conclusion in Pauly et al. (1998a
), several replications of their findings based on disaggregated datasets were published (), establishing the validity of the FD concept and its ubiquity.
Table 1 Some contributions demonstrating the occurrence of FD using locally disaggregated datasets, following the original presentation of this phenomenon by Pauly et al. (1998a), based on the global FAO catch dataset.
In the process, a rule-based mapping technique was developed (Watson et al. 2004
), which allowed assignment for the years 1950–2000 of the FAO fisheries catches to the more than 180 000 half latitude/longtitude degree cells comprising the world ocean, together with the key attributes of these landings (i.e. their species composition, and hence their mean TL and their ML).
This allowed mapping of the FD phenomenon, and simultaneously, elimination of the bias that was caused by fisheries statistics from small islands and some other states, which combine landings of inshore reef fishes with those of adjacent large oceanic, high-TL pelagics such as tunas (Pauly & Palomares 2005
The resulting maps (in Pauly & Watson 2005
) () show how widespread the FD phenomenon is. Indeed, it can be said to occur everywhere it matters, as the shelf areas where TL have strongly declined contribute a large fraction of the world fisheries catches. Indeed, the rate of TL decline has mostly increased since the 1950s, with the strongest rate of decline in the 1980s. Global fisheries were operating, on average, at a TL of 3.37 in the early 1950s; now their mean TL is ca
. 3.29, but this was as low as 3.25 in 1983. Remember: so far, humans do not eat zooplankton (although exceptions exist: there is a market for jellyfish in East Asia, to which some western countries have now begun to export this product).
Figure 2 Differences between the mean ML of fish and invertebrate species in fisheries landing in the 1950s, and that in the 1990s, mapped into 180 000 cells of 1/2 latitude/longitude degrees according to the procedure in Watson et al. (2004). Note areas of strong (more ...)
This analysis is confirmed by , a map of the mean ML reached by the species explicitly mentioned in global landing statistics. As can be seen, declines of up to 1 m over the 50 year period considered here occurred, mainly in the North Atlantic, but also in other areas where highly industrialized fisheries have removed most of the fishes capable of reaching large sizes. shows that this process, viewed globally, is proceeding in a rather smooth fashion, notwithstanding the mesh size and other single-species regulations meant to prevent the size of certain target species from falling below some critical levels (note that this does not consider the well-documented reduction in the average size within species).
In 1950, when the FAO began to assemble the global fisheries dataset analysed here, coastal fisheries had already impacted on inshore populations of fishes and invertebrates of both industrialized and non-industrialized countries (Cushing 1988
; Butcher 1996
). However, the serial depletion induced by the first industrial fisheries in areas such as the North Sea or New England, expanded, after World War II, to deeper waters, especially in the Southern Hemisphere. which shows, by latitude, the mean depth of marine fisheries catches from 1950 to the present, illustrates these trends, and thus explains how the overfishing of local fish populations has been largely masked by landings from new fishing grounds.
Figure 3 Mean depth of global fisheries landings, by latitude, from 1950 to 2000, based on catch data originally mapped into 180 000 cells of 1/2 latitude/longitude degree according to the procedure in Watson et al. (2004). Note the trend toward greater depths, (more ...)