Long-term changes in the mean trophic level of Central Chile fishery landings *

The mean trophic level of fishery landings (TLm) has been proposed by Pauly et al. (1998) as an index of the impact of fishing. On a global scale, TLm has showed a decreasing trend for the last 45 years as a result of a gradual transition in landings from piscivorous, long-lived and high trophic level species to short-lived and low trophic level species (i.e. planktivorous pelagic fish and crustaceans). SCI. MAR., 69 (2): 295-300 SCIENTIA MARINA 2005

This phenomenon is called 'fishing down the food webs' (FDFW) and, in an initial stage the transition in target species produces an increase in landings, followed in a more advanced stage by a decrease in landings that could reflect fishery-induced changes at the community and ecosystem levels (Pauly et al., 1998).
Although FDFW constitutes a controversial issue, even the most reluctant scientists agree that FDFW could be one of the explanations for changes in fish populations on global scales (Caddy et al., 1998;Caddy and Garibaldi, 2000).However, considering the complexity and high regional variability in world fisheries (Botsford et al., 1997), analyses are needed to test whether FDFW is also occurring in more local fisheries.
For the South Pacific Ocean (FAO areas 81 and 87), Pauly et al. (1998) report wide amplitude fluctuations in TL m , with landings continuing to increase as TL m declined in recent years.Both features seem to be an exception to FDFW.A possible explanation is found in the Peruvian anchoveta fishery collapse and the later development of an important offshore fishery based on horse mackerel (Trachurus symmetricus), which is a migratory species at a high trophic level.Therefore, high landings of horse mackerel could have masked changes in more local fisheries in recent years (Pauly et al., 1998).
The Humboldt Current System off Central Chile is a typical eastern boundary current system sustaining one of the most productive fisheries worldwide (FAO, 1995).By the late 1970s, the horse mackerel had become the most important fishery resource of Central Chile thanks to the development and later westward expansion of an important offshore fishery, mainly composed of an industrial purse-seine fleet.Consequently, landings of horse mackerel have influenced total landings in Central Chile in recent decades (Fig. 1).Although the ecosystem impacts of fishing on fish assemblages off Central Chile are poorly understood, evidence indicates that the most important fishery resources seem to have reached their maximum sustainable level, while others show clear signs of overexploitation (Cubillos et al., 1999;Quiñones et al., 2003).
It is acknowledged that El Niño Southern Oscillation (ENSO) is responsible for most of the interannual variability in the whole Eastern Pacific Ocean basin (Strub et al., 1998).In Central Chile, ENSO induces changes in marine populations, especially fishes inhabiting the pelagic domain (Yáñez et al., 1992;Arcos et al., 2001), but also benthic species (Gutiérrez et al., 2000).This is important because landings off Central Chile are dominated mostly by pelagic fish species (Neira et al., 2004).
Consequently, the objectives of the present study are: i) to examine long-term changes in TL m of fishery landings from Central Chile, ii) to determine whether the long term trend of TL m in Central Chile has been masked by landings of horse mackerel, and iii) to determine whether changes in TL m are influenced by ENSO events.

MATERIAL AND METHODS
The study area (Fig. 2) represents the marine zone off Central Chile (33°S to 39°S), which was defined taking into account both fishery management units and the distribution patterns of the most important fishing fleets.The study area is also a biogeographic unit defined by the following oceanographic and topographic features: highly seasonal coastal upwelling events (Arcos and Navarro, 1986); high turbulence during summer due to southwesterly winds, but also during winter due to weather fronts crossing the region from the west; and high offshore transport levels during summer (Ahumada, 1989;Strub et al., 1998).
Landing data by species (in tonnes wet weight) were obtained from the Chilean Fisheries Service Statistics Yearbooks (1979Yearbooks ( -1999;;www.sernapesca.cl).Landings for the 16 selected species (Table 1) correspond to the sum of their landings from 33 to 39°S.The selected species comprise about 95% of the total landings in Central Chile.
The trophic level for species i (TL i ) was estimated using the available information on gut content 296 H. ARANCIBIA and S. NEIRA  .
analysis derived from surveys designed to obtain biomass estimates and diet for the target and bycatch species in fisheries off Central Chile and the Ecopath with Ecosim software version 5.0 (EwE; Christensen and Pauly, 1992;Walters et al., 1997).
Values of TL i for mackerel, elephant fish and snoek were gathered from FishBase (Froese and Pauly, 2002), while TL i for golden conger-eel and red conger-eel were assumed to be similar to TL i estimated for black conger-eel (G.maculatus) using trophic models (see Table 1).Following Pauly et al. (1998), estimates of the mean weighted trophic level of the catch (TL m ) for each year of the time series were calculated as: where TL i is the trophic level of species i; Y i is the landing of species i in a given year; Y T is the total landing of all species in the same year, and n is the number of years.
To account for the effect of horse mackerel on the long-term trend of TL m , two analyses were carried out.First, we calculated TL m considering all selected species.Then, we recalculated TL m excluding landings of horse mackerel.
Spearman's rank correlation (rs) was used to establish a significant relationship between TL m and year and TL m and landings.

RESULTS
The maximum TL i (near TL=4) was found in Pacific sandperch (Prolatilus jugularis), while the lowest value (near TL=2) was shared by planktivorous fish (anchovy, sardine and common sardine) and detritivorous benthic crustaceans (red squat lobster and yellow squat lobster) (Table 1).
Changes in TL m in Central Chile are shown in Figure 3. TL m of all landings did not show a significant decreasing trend (rs=-0.15,p=0.493) (Fig. 3a).The greatest TL m value for the time series was found for 1988 (TL m =3.58).The lowest values are in the first (1979; TL m =3.23) and the last year of the time series (1999; TL m =2.67), and are related mostly to the low horse mackerel landings in those years (Fig. 1).In fact, at the beginning of the time series, the horse mackerel fishery was hardly exploited, while at the end it was heavily over-exploited (Quiñones et al., 2003).However, when landings of horse mackerel were excluded, TL m showed a significant decline (rs=-0.70,p<0.001) (Fig. 3b).ENSO events did not seem to have greatly affected the long-term trend of TLm.In fact, when landings in ENSO years are eliminated, the correlation improves (rs=-0.93,p<0.001) (Fig. 3c).Therefore, a significant negative correlation Table 1.-List of the species included in this study and their corresponding trophic level.(a): estimated in this study using gut content analysis (Lillo et al., 1993;Arancibia, 1992;Arrizaga, 1983;Arrizaga et al., 1993;Cordova et al., 1999) and the Ecopath with Ecosim software; (b): assigned assuming these species have a diet composition similar to G. maculatus, which was estimated in this study; (c): FishBase (Froese and Pauly, 2002).between TL m and time is found.This could reveal the effect of FDFW once the effects of horse mackerel and ENSO are removed.
There was no significant relationship between total landings and corresponding mean trophic levels (rs=0.03;p=0.902;Fig. 4a).Landings increased until 1994 without great changes in TL m .At this point, the relationship tends to revert due to a decline in total landings.When horse mackerel landings are excluded from the analysis (Fig. 4b), a clear inverse relationship between TL m and landings is observed (rs=0.90;p<0.001).In this case, landings are still increasing while mean trophic level of the fishery decreases.The increase of TL m in year 1988 is mainly related to the dramatic drop in landings of the sardine (S. sagax), a species of a low trophic level, from more than 100 thousand tonnes in 1997 to about 50 thousand tonnes in 1998.

DISCUSSION
There are some potential limitations in our analysis.The number of species included in this study is small when compared to similar published analyses (e.g.Pauly et al., 1998;Pinnegard et al., 2003).However, the selected species comprise 95% of landings and therefore we can argue that landings of those species not included in the present analysis are almost insignificant.Therefore, we consider that the potential impact of not including more species on the calculations and observed trends is weak.We consider that the estimation of TL i could be the factor that incorporates the highest uncertainty into our analysis.TL i values were estimated using quantitative information of gut contents obtained from research surveys designed to estimate stock size for the main fishery resources in Central Chile.Unfortunately, gut content analyses for the most of the groups in the study area have been sporadic and short-term.Consequently, it was not possible to estimate the TL i variance.The assumption of assigning a single trophic level to a species is problematic, as pointed out by Caddy et al. (1998), since ontogenetic change in diet, and consequently in trophic level, is common in marine species.However, considering that most species feed at lower trophic levels when they are smaller, and fishing pressure usually reduces the mean size of fish, the effect can be considered as conservative (Pauly et al., 1998;Pinnegard et al., 2003).
We acknowledge that alternative and independent methods, such as using stable isotopes of nitrogen (Post, 2002), are available to establish TL in aquatic organisms.However, although the estimation of TL using gut contents can induce bias, some studies have shown a significant correlation between TLs estimated using gut content and those estimated using radioisotopes (Kline and Pauly, 1998;Pinnegard et al., 2003).
Another factor that could have biased the observed trend in TL m is the relatively low trophic level of small pelagic fish in Central Chile (TL=2.02)compared to TLs of similar species in comparative ecosystems (Jarre-Teichmann et al., 1998).The low TL of small pelagic fish in Central Chile is explained by the fact that, in this system, unlike other upwelling systems, these species seem to feed almost exclusively on phytoplankton (Arrizaga, 1983;Arrizaga et al., 1993).However, when TL m is recalculated under the assumption that small pelagic fish feed 50% on phytoplankton and 50% on zooplankton (i.e.TL=2.5), a significant decreasing trend in TL m is still observed (rs=-0.65;p=0.001).
Since fishing activity in Central Chile started before the period covered by our 21-years data series, it is possible that populations of top predators that are not present in landings today could have been overexploited before 1979.In fact, in this area both sperm whale (Physeter catodon) and sea lion (Otaria flavescens) used to be hunted (Aguayo-Lobo et al., 1998).However, the hunting activities of both species were stopped after their populations had been severely depleted.Therefore, our study can be viewed as summarising the trends in TL of the catches of fish off Central Chile (because harvesting of mammals in early years is excluded), with the exception of swordfish (Xiphias gladius).This species sustained an important inshore artisan fishery from 1952 to 1986 (Barbieri et al., 1998).After that, landings of swordfish have corresponded mostly to individuals captured by an industrial fleet of long-liners operating beyond the Chilean Exclusive Economic Zone (i.e.offshore of 200 nautical miles).
When horse mackerel landings are excluded from the analysis, it is observed that, over time, the longterm TL m shows a decreasing trend, in accordance with what is predicted by the hypothesis of FDFW.Fisheries landings in Central Chile are highly dependent on fish species, especially small-sized pelagic fishes such as common sardine and anchovy, which are located at low trophic levels (Neira et al., 2004).
The increasing importance of small pelagic fish in the Central Chile fishery landings could be explained by a change in fishery preferences (due to the lack of more valuable species), high fishing mortality and good recruitment (SSP, 2001;Quiñones et al., 2003).In addition, both abundance and availability of common sardine and anchovy are also related to interannual and long-term changes in the environment (Yáñez et al., 1992).Fishing removal of both common sardine and anchovy is likely to impact the whole food web, since wasp waist control exerted by small pelagic fishes is hypothesised to be characteristic feature in upwelling ecosystems (Cury et al., 2000).
In conclusion, our analysis confirms that landings of horse mackerel have masked a FDFW process in inshore fisheries off Central Chile, as suggested by Pauly et al. (1998).In addition, we suggest that fisheries in Central Chile have been fishing down the food web as a result of fishery-induced changes rather than environmental changes associated with ENSO events.In Central Chile, TL m has declined at a rate of 0.175 levels per decade, from 2.45 in 1979 to 2.10 in 1999, which is higher than the global rate of 0.10 levels per decade estimated by Pauly et al. (1998).This trend is also higher than the decreasing trend estimated, for instance, for the Gulf of Thailand, an ecosystem in which fisheries resources have been depleted by human activities (Christensen, 1998).
FIG. 3. -Mean trophic level of fishery landings (TLm) in Central Chile (33°S-39°S).a) Total landings; b) excluding landings of horse mackerel; c) excluding landings of horse mackerel and landings in years affected by El Niño Southern Oscillation.