The Bou-Ismail Bay (Algeria) is in the Algerian sub-basin of the western Mediterranean between 2°54’E 36°48’N and 2°24’E 36°38’N. It is considered to be one of the most important bays on the Algerian coast regarding fisheries production. It has three fishing ports: Tipasa, Khemisti and Bouharoun. The ports of Khemisti and Bouharoun are the most important in the area in terms of fleet concentration and quantities landed. There is also a mooring beach (Fouka) and a converted fishing shelter (Bou-Ismail). As of December 2012, the fishing fleet in the ports of Bouharoun and Khemisti consists of 307 units, including 27 trawlers, 108 sardine boats and 172 small traders. The vessel concentration is higher at the port of Bouharoun (73%). There are also a few yachtsmen at Khemisti and two tuna boats registered at Bouharoun.

The statistical analysis was conducted using the R software (R Development Core Team 2013). To evaluate the statistical variability of the parameters studied, we used the bootstrap method (Efron 1979). For curve fitting, we applied the “Loess” locally-weighted regression method (Cleveland 1979), with a smoothing factor (span =1). A confidence interval of 95% obtained from 1000 bootstrap replicas was calculated for the Loess smoothing (see Pennino et al. 2011 for more information).

The species composition of landings (Fig. 1) shows the dominance of sardines (Sardina pilchardus), constituting over 59% of landings, followed by round sardinella (Sardinella aurita), at around 14% of catch. These two small pelagic species make up 73% of the Bou-Ismail Bay fishery’s average annual fishing production.

The total annual landings (Fig. 2) declined constantly during the years studied, falling from over 8000 t to less than 4000 t: a decrease of over 50%. This significant fall in fishing production is due to the decline in sardine catch (Fig. 3), which fell from over 5000 t to around 1000 t between 1999 and 2012. Other species showing declining production included jack and horse mackerels (Trachurus spp.), (Fig. 3), deep-water rose shrimp (Parapenaeus longirostris) and swordfish (Xiphias gladius). In contrast, round sardinella (Sardinella aurita) and the principal demersal fish, namely bogue (Boops boops), pagellus (Pagellus spp.) and red mullets (Mullus spp.), alternated between periods of rising and falling catch (Fig. 3).

The total fishing fleet (Fig. 4A) increased significantly during the years considered, from 215 units in 1999 to 308 units in 2012: an increase of 93 units with a 43% rate. The significant increases were for sardine boats (Fig. 4C) and small traders (Fig. 4D), 46 and 43 additional units respectively, with respective rates of 75% and 33%. Trawler numbers (Fig. 4B) barely changed: from 24 units in 1999 to 29 units in 2012.

The fishing fleet includes boats from Tipasa province, with Bouharoun and Khemisti as landing ports. However, although an official census of fishing boats exists, it is very difficult to define the fishing effort exerted on marine resources in the Bou-Ismail Bay. Indeed, the number of the fleet registered is not truly representative of the real fleet operating in the bay. Several registered vessels are not operational (wrecks, missing boats, etc.) and there is also a significant movement of vessels between fishing grounds of the Algerian coast that is difficult to monitor. Due to this lack of finer data, our main interest is to study the trend of the fishing fleet and how it evolves throughout the study period, and how this can affect the ecosystem.

Regarding the environmental variables, Chl-a (Fig. 5A) increased until 2008, peaking at 0.41 cmg/m³. It then decreased slightly while precipitation levels rose constantly during the period considered, from 692 mm (0.08 mm/h) in 1999 to around 858 mm (0.1 mm/h) in 2012 (Fig. 5B). The SST exhibited a gradual increase from 2007 in Algerian waters, reaching the maximum value of 19.27°C in 2012 (Fig. 5C).

Among the trophic indicators, after reaching 24 in 1999, the P/D index (Fig. 6A) decreased slightly until 2008, to a low of 3.5. A slight increase then began. The annual average calculated for the period in question is estimated at 9.9.

The average MTI for the series in question is estimated at around 3.23. During the first six years, the values for the MTI (Fig. 6B) were at their lowest for the period in question (~3.2). This coincided with high catches of sardines (Sardina pilchardus) from a low TL of 3.1. Subsequently, the MTI rose, whereas sardine catch continued to fall. MTI values were at their highest during the last years, reaching a peak of 3.26 in 2009 and coinciding with the lowest sardine catches. However, at the very end of the series, the MTI began to fall. This trend explains the high catches of round sardinella (Sardinella aurita) with a low TL (TL=3.13). In fact, this species became the dominant species in landings from 2009, constituting almost 47% of overall catch in 2012, and exceeding sardine catch (Fig. 7).

If species of TL<3.25 (salema, golden grey mullet, Norway lobster, saddled seabream, sargo bream, anchovy, sardine, round sardinella and various crustaceans) are excluded from the calculation, the average TL of catch (^3.25MTI) from 1999 to 2012 is estimated at 3.52. It is very close to that of jack and horse mackerels (Trachurus spp.) and pagellus (Pagellus spp.). The ^3.25MTI (Fig. 6C) fell throughout the period studied. It decreased from 3.56 in 1999 to 3.49 in the final year of the series. This decrease of TLs by 0.07 over the 14 years studied is mainly due to the jack and horse mackerel catch declining by over 50%. In fact, if sardine and round sardinella catches are removed from the calculation, jack and horse mackerel, of medium TL (TL=3.5), become dominant, consequently impacting considerably on the ^3.25MTI trend. The increase in Pagellus spp., (TL=0.53) catch during the early years did not alleviate the effects of falling catches of jack and horse mackerel and swordfish (Xiphias gladius), of TL 4.5. From 2005, catches fell for all of the fishery’s major low- and medium-TL species, accentuating the fall in 3.25MTI to its lowest values in the series.

The FiB shows a clear declining trend during the period in question, with negative values from 2002 onwards (Fig. 6D). In fact, a value of 0.05 was recorded in 1999, and a value of -0.45 was recorded in 2012. This decrease was accompanied by a decrease in overall catch.

The average national fishing production is estimated at 126300.5 tonnes. Like the Bou-Ismail fishery, it is dominated by small pelagics, mostly sardine (Sardina pilchardus) at 46% of total landings, followed by round sardinella (Sardinella aurita) at 17%, and jack and horse mackerels (Trachurus spp.) at 11% (Fig. 8). National landings (Fig. 9A) increased to a peak of 138483 in 2006, before falling. This trend coincides with that seen for small pelagic species catches (Fig. 9B). The proportion of these species in landings (80%) clearly influences overall catch. Demersal species (Fig. 9C), molluscs (Fig. 9F) and (to varying degrees) large pelagics (Fig. 9D) contributed to the rise in overall landings during the early years. However, the continuing rise in demersals did not counterbalance the declining trend for total catch during the last few years. Crustaceans (Fig. 9E) fell by around 1000 tonnes until 2007, before catch began to rise again. Among the trophic indicators, after rising slightly from 1999 to 2004, the P/D index fell significantly, reaching its lowest value of around 5 at the end of the series (Fig. 10A). The MTI fell from 3.25 to 3.22 between 1999 and 2003 (Fig. 10B) and then rose, reaching 0.07TL in 2010. The ^3.25MTI fell by around 0.1TL between 1999 and 2007 (Fig. 10C). Following this, it entered a rising trend because of a significant demersal species catch in the later years. The FiB values were positive during the whole period studied, except in the last year, when a value of –0.16 was recorded (Fig. 10D). In fact, following a period of increase with a peak of 0.13 in 2006, the FiB fell, reaching a low of –0.16 in the last year of the series. The trends for landings and the FiB index coincide. The rise in the FiB is accompanied by the rise in overall catch and vice versa.

The Bou-Ismail Bay Fishery, like most Mediterranean fisheries, is characterized by the dominance of small pelagic species. This is reflected in the high values recorded for the P/D index. The average for the ratio in this bay, a value of 9.9, is far higher than those calculated for other Mediterranean regions (Pennino and Bellido 2012, De Leiva Moreno et al. 2000). This could be because of the different spatial-temporal scale and the species taken into consideration when calculating the ratio. It seems useful to carry out micro-level studies for this type of indicator, because of the existence of specific characteristics for each fishery and the related ecosystems, which can be hidden on a larger scale. The P/D ratio is much more heavily influenced by landings of Clupeoid species, primarily sardines (Sardina pilchardus) and round sardinella (Sardinella aurita), which represent over 73% of overall catch. The P/D ratio is characterized by a very large initial decrease, then an increase over the very last years. The initial situation is due to the fall in small pelagic species and the increase in demersal fish within landings. The increasing scarcity of small pelagics, along with an increased fishing effort, suggests that these species are overexploited. In fact, despite rising primary production and nutrient input, which are favourable to these planktivorous species (Caddy 1993), their catch has fallen significantly, particularly in the case of sardine (Sardina pilchardus), whereas catches of the other small pelagic species (round sardinella, or Sardinella aurita) have increased in the latter years.

However, the increase in demersal fish catch is probably due to the increased fishing effort. In fact, over the last decade, several incentives funded by state subsidies have been introduced as part of the national development plan for fishing activities and aquaculture, and other programs have allowed for the growth and modernization of the fishing fleet. This increased fishing pressure is probably the reason behind the increased catches of these species, particularly pagellus (Pagellus spp.) and red mullet (Mullus spp.), in the first few years. However, the last four years of the series saw a fall in demersal fish catch, probably because of overexploitation (given the increased fishing effort) and/or eutrophication. In fact, the latter phenomenon, reflected in the increased primary production indicators (Chl-a and SST), preceded and probably caused the fall in demersal fish catch.

Eutrophication causes hypoxia situations, which have a negative effect on benthic species (Caddy 1993). Moreover, fairly high pollution levels have been observed in this zone (Houma 2009, PNUE 2005). These can also cause hypoxia. The start of a rise in the P/D ratio can indicate FDMW. This situation, as already observed in the Mediterranean (Pennino and Bellido 2012), indicates that the fishery increasingly targets small pelagic fish from the bottom levels of the trophic chain, such as round sardinella, for which catch was high during the latter years.

The ability to identify patterns consistent with FDMW has been discussed in considerable detail in the literature, along with the different conditions and data issues that could lead to similar patterns being identified that do not result from FDMW but from a range of fishery exploitations patterns throughout the food web (Essington et al. 2006, Litzow and Urban 2009, Branch et al. 2010, Sethi et al. 2010). Each of these fishery exploitations patterns should be contrasted to offer alternative explanations to FDMW.

The lowest MTI values were recorded between 1999 and 2003, and correspond to high catches of Clupeidae of low TLs (an estimated 3.1 for sardines, and 3.13 for round sardinella). The high catches of these two species, accounting for over 72% of total landings, are the reason for the lowest recorded MTI figures. However, the increase in the MTI coincides with the lowest landings of sardines (Sardina pilchardus). The increased catches of the main demersal fish species from relatively high TLs during the first years, and their decreases during the last years, did not counterbalance the MTI trend. The predominance of small pelagics can mask a possible effect of fishing on the trophic chain at medium and high TLs. Consequently, this indicator is unreliable for use in semi-enclosed seas, because of coastal phenomena such as eutrophication, which affects the abundance of small pelagic species and consequently impacts on their presence in landings. This can be confused with the effects of fishing (Caddy et al. 1998).

It therefore appears difficult to interpret the evolution of this index in a fishery greatly composed of small pelagics that are very vulnerable to environmental conditions. Consequently, the ^3.25MTI (Pauly and Watson 2004) is an alternative to the MTI that reduces the influence of environmental conditions in the analysis of the effects of fishing on the trophic chain. The falling ^3.25MTI, which indicates a decrease in the average TL of catch, suggests that the fishery is progressively targeting species from lower down in the trophic chain, probably because fish from upper TLs have been overexploited. The fall in the average TL of fish caught, combined with declining catches, where the fishing zone or gear usage has not changed, may indicate a collapse of the food chain (Pauly et al. 2001). The decrease, estimated at around 0.07 TLs for the 14 years studied, is close to that estimated by Pauly et al. (1998) at global level: around 0.1 per decade. However, it is far higher than those found in the Mediterranean by Pinnegar et al. (2003) (~0.07/36 years), which include aquaculture. However, these authors use the MTI indicator and study larger temporal series and scales. The FiB decreased throughout the whole period studied. This decrease was accompanied by an increase in the fishing effort, decreasing catch, and a fall in average TL (^3.25MTI). This indicates a serious problem: it suggests overexploitation on the scale of the ecosystem (Chassot 2005) and the exhaustion of inshore stocks (Bathal 2005), given the coastal nature of this fishery. The decrease in the FiB can be seen as the result of a modification in the trophic structure of the ecosystem, and of an alteration of its operation (Cury et al. 2005b, Pauly and Watson 2005). This is confirmed by the trend in the average TL of catches, which decreased by ~0.07 TL over the 14 years studied.

This trend towards a fall in the average TL of catch has not been accompanied by a rise in catch. Such a result would have justified a deliberate choice to displace efforts towards the bottom of the food chain: towards the more productive lower components of the trophic network. However, this did not happen. In fact, the FiB decreased during the period studied (Freire and Pauly 2010). Moreover, discards are not taken into account in the catch data, which may partly explain this trend in the FiB (Pauly and Watson 2005). This analysis of the FiB suggests that fishing in the Bou-Ismail Bay removes so much biomass from the ecosystem that its functioning is affected (Pauly and Watson 2005).

It is worth noting that despite recent criticisms of the FiB index as an indicator for proving that ecosystems are being degraded by fishing (Hornborg et al. 2013), we understand that it may still be a suitable indicator for showing the changes occurring in the ecosystem, but only if used complementarily with other indices.

The comparison between the Bou-Ismail Bay and the national level is essential since certain phenomena appearing on a small scale can be hidden on a large scale (Bhathal 2005). The average annual landings of the Bou-Ismail bay are estimated at 6889 t/y. They represent nearly 5% of the national landings estimated at 126300 t/y, considering the same period between 1999 and 2010. However, the specific structure resembles and both them are characterized by the dominance of small pelagic species, mainly in the order of importance, sardine (Sardina pilchardus), sardinella (Sardinella aurita) and horse mackerel (Trachurus spp.).

The fall in the P/D ratio is caused by the decline in small pelagic catch despite the increased nominal fishing effort (with a near 70% rise in the number of sardine boats and seiners from 1999 to 2009 (http://www.mpeche.gov.dz) as a result of the sector development policy from 1999 onwards. This situation indicates the overexploitation on the Algerian coasts of small pelagic species, particularly sardines, for which catches are constantly falling (as in the Bou-Ismail Bay). The increased demersal fish catch at national level is due to the near 70% increase in the number of trawlers from 1999 to 2009, and an increase of around 100% in the number of small traders (http://www.mpeche.gov.dz). However, this increase can mask a situation of overexploitation for these species in the coastal zone, and maybe in some particular local sites hypoxia, as indicated for the Bou-Ismail Bay. In fact, new and more powerful trawlers, equipped with full modern navigational equipment, have been acquired with state subsidies with the aim of encouraging fishing in the high seas. The average TL of catch, including species from TL<3.25 (MTI), is highly influenced by landings of small pelagic species, as is the case in the Bou-Ismail Bay. It increases when catches of small pelagic species increase, and vice versa. As a result, a probable FDMW situation can be masked by these low-TL species, which are highly affected by environmental conditions. Consequently, we will use the ^3.25MTI as the basis for our evaluation of the influence of fishing on the food chain in the ecosystem of the Algerian waters.

After eliminating from the calculation all species with a TL<3.25 (^3.25MTI), including small pelagics, detritivores and certain invertebrates, the average TL of catch decreases from 3.67 to 3.59 (~0.08TL/12 years). This decrease is very close to that of the Bou-Ismail Bay (~0.07/14 years), suggesting a situation of FDMW in the upper and middle TLs on the scale of the ecosystems exploited by fishing in the Algerian Mediterranean. An increase during the final period of the time series is seemingly due to the rise in catches of certain demersal species from high TLs, as a probable result of the exploitation of new zones by the new trawlers.

The FiB at national level exhibited positive values almost throughout the period studied, with an initial period of increase, followed by a decrease in the later years. The initial rise in the FiB indicates either an expansion of fishing or a “bottom-up” effect, via the increase of primary productivity (Pauly and Watson 2005). Both scenarios should be envisaged, though the “bottom-up” effect is more likely, since the average TL of catch is falling.

In fact, the expansion of fishing is perhaps caused by the exploitation of new zones by the new, powerful and well-equipped offshore trawlers. These were acquired as part of the plan to incentivize the fishing sector, the first being in 2003. This is reflected particularly by the increase in demersal fish landings. On the other hand, the Mediterranean Sea is a semi-enclosed sea, highly affected by nutrient inputs (Caddy and Garibaldi 2000). In fact this is the case of Algerian coasts, which are enriched not only by soil runoff, but also by nutrients from waters of Atlantic origin. It is likely that this source of nutrients may have favoured the increase in the biomass of small pelagic species decrease in the TL of catches.

Hence, we conclude that the results from Algerian waters are rather similar to those obtained at the level of the Bou-Ismail bay, particularly for the last years, and both show similar trends, suggesting an alteration of the mechanics of the underlying ecosystem, probably caused by overfishing (DasKalov 2002).

Catch-based ecosystem indicators do not necessarily give a reliable representation of the ecosystem (Branch et al. 2010). However, they are useful for providing information about the methods of exploitation which could undoubtedly arise from a change in abundance or from a fishing strategy, as a consequence of the management policy or economic dynamic (Guénette and Gascuel 2012). However, the indicators should be combined and analysed together, in order to obtain an overall picture of the ecosystem dynamic. For example, interpreting a single indicator such as MTI could give misleading results, as demonstrated in this study, where it increased rather than decreasing.

The results of the indicators studied for the Bou-Ismail Bay, combined with the environmental variables and the available data on fisheries (such as landings and fleet data) are summarized in a “trade-off” table (Table 3), to facilitate their interpretation and create a “toolbox” for implementing an EAFM. The decrease in the FiB, accompanied by a fall in the TL of catch (^3.25MTI) and in the P/D, allows the Bou-Ismail Bay’s ecosystem to be classified as an alert zone signifying unsustainable management of this zone. Consequently, this fishery requires immediate planning measures, such as a reduced fishing effort and stronger controls to avoid stock collapse.

This “trade-off” table remains purely indicative. It requires completion by other results concerning fisheries in other regions of the Algerian coast, in order to permit comparisons and probably define other zones that could serve as toolboxes for management needs. This could offer an alternative to the cumbersome model requiring large amounts of data on recruitment, size-weight relationships, and other unavailable and difficult-to-obtain biological data.

In conclusion, the use of trophic indicators in the evaluation of ecosystems subject to fishing offers an alternative to complex models requiring huge amounts of data which are not always available. This is particularly the case is studies of Algerian fisheries (for which data are often scarce). Consequently, we wish to encourage other studies of this kind, evaluating all fisheries at national level using longer historical series and locally-established TLs distributed according to size, in order to obtain the best possible results.