Loggerhead sea turtles (Caretta caretta) are listed as Endangered by the International Union for Conservation of Nature Red List of Threatened Species (IUCN 2011) and they are threatened worldwide by many human-related activities, including by-catch, direct exploitation, and climate change (Lutcavage et al. 1997; Wallace et al. 2010; Witt et al. 2010). In this context, information on population trends is fundamental to assess population status and effects of conservation measures and is considered a priority for sea turtle conservation (Hamann et al. 2010).

In the Mediterranean Sea, as in many other regions, the only available data for detecting possible trends of sea turtle abundance consist of nest counts from a few long-monitored nesting beaches (Casale and Margaritoulis 2010). However, nest counts have several problems. The number of nests is converted to number of nesting females using stochastic parameters estimated with substantial uncertainty, which results in greater uncertainty (often not sufficiently described) in the abundance estimates and associated trends (Mazaris et al. 2008). Nesting females are only a small portion of the population, which is mostly composed of juveniles, and fluctuations of the demographic structure—age and sex ratio—may make adult female abundance a poor indicator of population abundance. Finally, because of the long maturation time of these animals, which in the Mediterranean can take up to 30 yrs (Casale et al. 2009, 2011a, 2011b), trends of the juvenile portion of the population become evident in the adult class only after a long delay, even decades. Thus, the practical use of nest/adult trends is questionable, because an observed collapse of the nests/females means the very end of the population with little possibility of intervention (e.g., Chan and Liew 1996; Spotila et al. 2000).

Indices of abundance at sea are greatly needed because they can provide information of present trends of the bulk of the population, i.e., juveniles. For this reason, abundance data for small juveniles would be particularly valuable. Unfortunately, they are very difficult to obtain. In this respect, incidental catch rates of sea turtles in fishing gear observed during normal fishing activities may represent an index of abundance at sea that does not require a specific sampling effort but only adequate collection and standardization.

As a case study for such an approach, we identified 2 sets of sea turtle by-catch data collected 20 yrs apart from long-liners fishing in the Gulf of Taranto, northern Ionian Sea, a foraging area for small and medium-size juveniles (De Metrio et al. 1983; Deflorio et al. 2005; Casale et al. 2010). Here we present results of a standardized comparison to provide indications of possible changes of sea turtle abundance in this area.

Methods

The Gulf of Taranto is an almost circular area with a diameter of approximately 100 km, located in the south of the Italian peninsula and in the northern part of the Ionian Sea (Fig. 1). Sea turtle catches by long-liners based at Porto Cesareo (Italy) and fishing in the Gulf of Taranto were monitored in 2 periods: 1978–1981 (see De Metrio et al. 1983) and 1998–2003 (for the subperiod 1999–2000 see Deflorio et al. 2005). Fishing area (the whole of the gulf) was the same in the 2 periods. Target species, fishing practices, and vessels remained basically unaltered between the 2 periods, with the exception of slight changes in the fishing gear (Table 1). Monitored longlines belonged to 2 types of fishing gears with different features according to the target species: swordfish (Xiphias gladius) and albacore tuna (Thunnus alalunga) (Table 1). Almost all sea turtles caught were loggerhead turtles (Caretta caretta); therefore only this species was considered in this analysis.

View Figure

• Download as Powerpoint
• Download Figure

Table 1.  Main characteristics of the fishing gear used in the Gulf of Taranto in the 2 studied periods.

View Fullscreen

Catch data from the periods 1998–2003 (for albacore) and 1998–2001 (for swordfish) were directly recorded by on-board observers. In contrast, those from the period 1978–1981 were obtained through a logbook program. Before 1980 sea turtles were not legally protected in Italy and were commonly sold in the local fish markets (De Metrio et al. 1983) and fishermen had no incentive to underreport catches, while they had interest to land and sell turtles. Hence, we have no reasons to believe that data before 1980 are unreliable and included these data (1978–1979) in the analysis. However, logbook information may be biased for a variety of reasons (e.g., Stratoudakis and Marçalo 2002; Hamer et al. 2008) and the actual accuracy of our logbook data cannot be verified.

Data were obtained from 37 boats in the period 1978–1979 and 19 boats in the period 1998–2003. In order to reduce sources of variability, only data from the same months and target species were used for comparison: May–August for swordfish and September–November for albacore. The number of hooks deployed was recorded and catch data were described as presence/absence of a turtle at individual hooks, i.e. as binomial data (0, 1), and comparisons between the 2 periods were performed through the Fisher's Exact test on 2 × 2 contingency tables including total number of hooks with and without a turtle in the 2 periods.

Results and Discussion

In total, 2679 fishing sets, with 3,831,536 hooks, and 653 captured loggerhead turtles were observed. In the period 1999–2000, average turtle size was 43.8 cm curved carapace length (CCL) (n  =  115) and 37.2 cm CCL (n  =  83) for turtles captured by long-liners targeting swordfish and albacore, respectively (Deflorio et al. 2005). In the period 1978–1979 only turtle weight was recorded, with an average of 29.0 kg (n  =  71) and 17.3 kg (n  =  544) for swordfish and albacore long-liners, respectively (De Metrio et al. 1983), which can be converted to 65.3 cm CCL and 54.5 cm CCL through the relationship provided by Deflorio et al. (2005). Although these conversions should be regarded with caution, they suggest a decrease of average turtle size between the 2 periods.

Turtle catch rates were higher in the second period than in the first (Table 2) for both swordfish (Fisher Exact test; p < 0.05; n_hooks  =  1,260,465) and albacore longlines (Fisher Exact test; p < 0.01; n_hooks  =  2,571,071). Increased catch rates and decreased size may suggest an increase of the smaller part of the population. However, it should be noted that catch rate results might be affected by slight differences between the fishing gear used in the 2 periods. For instance, larger hooks are known to reduce turtle catch rate because they are more difficult to swallow, especially by smaller turtles (Stokes et al. 2011). Swordfish long-liners used slightly larger hooks (10 cm) in the first period than in the second one (8–9 cm) (Table 2), and the larger size of turtles in the first period would fit a possible hook size effect. However, larger turtles were also caught in the first period by long-liners for albacore using hooks of the same size in the 2 periods; therefore it is unlikely that a hook size effect is the reason of different turtle size and catch rate. Long-liners for albacore only differed in the distance between floats, with a longer distance in the first period (Table 1). How this might affect turtle by-catch rate is uncertain. Theoretically, a longer distance between floats could allow central hooks to sink deeper, although currents can easily counteract this, and deeper hooks are supposed to catch less turtles (Gilman et al. 2006). All of this considered, the observed increase of catch rates should be interpreted with caution and does not necessarily imply an increase in turtle abundance during the 20-yr period. On the other hand, present results are the opposite of what is expected in a situation of declining abundance, and make a negative trend in the study area and period unlikely.

Table 2.  Monitored fishing effort, loggerhead turtles caught, and turtle catch rates in the Gulf of Taranto in the 2 studied periods.

View Fullscreen

The main identified threats to sea turtles in the basin are destruction or disturbance of reproductive habitats (Casale and Margaritoulis 2010), incidental catch in fishing gear, collision with boats, and intentional killing (Tomás et al. 2008; Casale et al. 2010; Casale 2011), which appear to increase the overall mortality (Casale et al. 2007, 2010) and which, when considered together, represent a high level of threat (Wallace et al. 2011). Therefore, a strong population decline would be expected during 20 yrs. However, due to homing behavior of turtles and consequent population structuring (Carreras et al. 2007), rookeries can have independent dynamics, including different trends, because of different threats affecting them both on land and at sea. At sea, turtles do not distribute uniformly in the Mediterranean; to the contrary, those from different rookeries prefer specific foraging areas (Margaritoulis et al. 2003; Lazar et al. 2004; Carreras et al. 2006; Broderick et al. 2007; Casale et al. 2008; Zbinden et al. 2008). Therefore, trends of turtle abundance at sea should be considered at local levels and associated with the natal rookeries of the turtles frequenting that area. Accordingly, the scarce and mostly anecdotal trend data available for recent periods include decreasing, stable, and increasing patterns in different areas (Casale and Margaritoulis 2010).

Present findings do not show a decline of turtle abundance in the Gulf of Taranto but this should be regarded as a local situation and should not be generalized to the whole Mediterranean population. The Gulf of Taranto is a developing ground for small juveniles, probably from the nearby Greek rookeries, and in particular Zakynthos (Casale et al. 2010), but also from other rookeries such as those in Turkey (Carreras et al. 2006). Therefore, turtle abundance in the study area would primarily reflect these rookeries. A rather stable trend of number of nests was observed at the Zakynthos rookery in the period between 1984 and 2002 (Margaritoulis 2005) and this would be consistent with the present results. However, it cannot be excluded that other factors may affect local abundance and make detecting patterns more difficult. For instance, climate change may affect turtle reproductive phenology (Mazaris et al. 2009) and distribution (Hawkes et al. 2009; Witt et al. 2010) and a possible shifting of turtles between areas may result in local negative trends even if the population is stable, or it may mask real negative trends. Only the availability of trend data from multiple sites can overcome these problems and provide information on overall trends.

In a situation where population status is often assessed by criteria considering past trends inferred from partial data, typically nest counts (Seminoff and Shanker 2008), the availability of different sources of information about population trends is desirable. By-catch data from different periods can be a relatively easy way to obtain trends, provided that fishing effort can be standardized and compared. However, standardizing data from different periods can be challenging and not completely possible, as in our study that compared logbook and observer data. Moreover, technical and operational changes, including improving fishing gear and adapting to changes of target species abundance or demography, are common in the fishery sector, especially in mostly artisanal/small fisheries typical of the Mediterranean (Casale 2011), and this reduces the chances to find suitable data sets to compare. Although the few such opportunities to compare by-catch data sets should be exploited in order to provide insights into possible past population trends, standardized experimental capture programs should be established immediately in the main Mediterranean foraging areas in order to effectively monitor sea turtle trends in the future.