Differential loggerhead by-catch and direct mortality due to surface longlines according to boat strata and gear type

1 Instituto Español de Oceanografía, Centro Oceanográfico de Málaga, Fishing Port s/n pots number 285, 29640 Fuengirola, Málaga, Spain. E-mail: jacaminas@ma.ieo.es 2 Departamento de Biología Animal (Zoología), Facultad de Ciencias, Universidad de Málaga, Campus de Teatinos s/n, E-29071 Málaga, Spain. 3 Instituto Español de Oceanografía, Centro Oceanográfico de Santander, Sant Martin s/n, 39004 Santander, Spain.

Loggerheads can be incidentally captured by surface longlines but information on direct mortality of this or other marine turtles is scarce (Laurent et al., 2001).As a number of loggerheads are caught dead, we consider this number as the direct pre-release mortality, and we will refer to it as direct mortality hereafter.Surface longline fishery is a very important activity for the Spanish fleet in the Mediterranean Sea (Camiñas et al., 1992), which is comprised of different boat strata that target different species by modifying the surface longline (using different lengths, baits, fishing depths, hooks and rollers) according to the season, target species abundance, market conditions and boat length (see Table 1).Since 2003 a new gear type called "american" and also "roller" has been introduced into the fleet, which implies a reduction in the number of hooks and an increase in the main line length and the depth of fishing (Báez et al., 2005).This modified surface longline increases the distance between hooks (21 m in traditional long-line SWB and 90 m in SWBr, abbreviations as in Table 1) and the length of the gear (30-40 nautical miles in traditional long-line and 60-70 nautical miles total length in SWBr), reducing the average number of hooks to 1000 (versus >2500 mean number of hooks in traditional surface longline), and the time needed to board the gear to 6-7 hours, which entails a 2-3 hour reduction with respect to the traditional longline.
An onboard observer program was implemented by the IEO (Spanish Institute of Oceanography) to obtain direct information from the fisheries regarding target and incidental species.In this paper we analyse the data obtained in the above mentioned program, focusing on the comparison between loggerhead incidental by-catch and direct mortality caused by different types of boats and gears, observed from April to December during the period 1999-2004 in the Spanish surface longline fleet fishing in the western Mediterranean Sea.

MATERIAL AND METHODS
The Spanish surface longline fleet from Mediterranean ports consists of 105 vessels from 12 to 27 m in length licensed for surface longline fishing around the year, and more than 2000 shorter boats licensed for artisanal gears, including surface longline, fishing mainly in summer (http://www.mapya.es).The fishing grounds include a large area of the western Mediterranean between 36º and 44ºN and 02ºW and 05ºE.The strongest fishing effort is carried out around the Balearic Islands and in the Ibiza Channel (Camiñas and de la Serna, 1995;Camiñas and Valeiras, 2001;Valeiras and Camiñas, 2003).
A total of 1546 fishing operations were observed onboard from April to December, during the years 1999 to 2004, which represent 3825292 hooks controlled directly.The number of boats with onboard observers and the total number of fishing boats each year can be seen in Table 2.We took data on effort (number of hooks), environmental data (superficial temperature of the water, distance to the coast, depth and weather conditions), standardised fishing parameters (time and spatial co-ordinates, number of attracting lights, and type and size of the bait),  fishing strategies, and fleet dynamics (see Valeiras and Camiñas, 2003).We classified the fleet into six boat strata with some differences in structure and target species (Table 1).Sea turtle by-catch and direct mortality were recorded and classified according to gear type and boat strata.
As the catch per unit effort (CPUE) measure we used the index R (number of incidental catches / number of hooks -3 ), R dead (number of direct turtle mortalities / number of hooks -3 ) (Yeung, 1998;Laurent et al., 2001;Valeiras and Camiñas, 2003;Lewison et al., 2004), F rate (number of incidental catches / number of fishing operations), and F dead (number of direct mortalities / number of fishing operations).We tested the inter-annual differences in R and F within boat strata and gear, and differences in R, R dead , F, and F dead among boat strata and gears with data from all years pooled together, using the chi-squared (χ 2 ) test.Expected valued in the χ 2 tests were calculated according to the number of hooks for R and R dead , and to the number of fishing operations for F and F dead.

RESULTS
The number of fishing operations observed for each boat strata, type of gear, and year, and the corresponding values of loggerhead by-catch and direct mortality can be seen in Table 3. Direct mortality is a low frequency phenomenon.We observed 3480 loggerhead incidental catches, 46 of which the turtle was dead, so the proportion of dead turtles was 1.32% of the incidental catches.Inter-annual differences in R and F were significant for every boat strata except BFTr (Table 3).These inter-year differences may be explained by variations in sea turtle concentrations in feeding areas according to the variations in environmental factors and migratory patterns.Thus, it is preferable to study all catches pooled together to mitigate the effect of inter-annual concentration differences.In addition, the low number of directly dead turtles prevents analysing R and F within a single year.
We observed direct mortality in all boat strata (Table 4), but the relative direct mortality varied among them.Both the by-catch and the direct mortality differed significantly according to the type of boat and gear but the results differed according to whether they were measured in function of the number of hooks or in function of the number of fishing operations (Table 4).With respect to the number of hooks, boats longer than 12 m without a roller that targeted bluefin tuna captured the highest number of Loggerheads, whereas boats longer than 12 m with a roller that targeted swordfish caused the highest direct mortality.With respect to the number of fish-SCI.MAR., 70( 4
According to recent research, the Spanish longline vessels should be classified according to the type of boat and gear (Laurent et al., 2002).Our results imply that it is necessary to study the CPUE of the different boat and gear types.The fishing effort depends not only on the number of hooks but also on other characteristics such as the use of a roller, the economic potential for buying bait and materials, the boat length, and the number of fishermen.Although the CPUE is usually calculated in function of the number of hooks (as an R-index) (Yeng, 1999;Laurent et al., 2002;Lewison et al., 2004), we consider the R rate inadequate for comparing the effort of the different gear types.Table 4 shows that the results of F and F dead are more consistent than those of R and R dead .Therefore, the F rate seems to be more adequate for relating the incidental catch or direct mortality to the fishing effort.Lewison et al. (2004), estimated the number of incidental loggerhead catches in the Mediterranean surface drifting longline fishery as approximately 60000 to 80000, but they assumed that CPUE (measured as an R-index) was homogeneous, which is not exactly so.Differences in incidental catch rates could be due to differential fishing techniques.For example, SWBr had lower F but higher F dead than SWB.The main cause of direct mortality due to longlines is drowning (Work and Balazs, 2002).The "roller" gear has fewer hooks and is less time in the water, and so the number of turtle captures is decreased; however, with the deeper hooks the captured loggerheads cannot come up to the surface to breathe and therefore direct mortality is increased.These results seem to indicate that using a roller may increase Loggerhead direct mortality as it increases the probability of drowning.
Our conclusion is that in order to calculate loggerhead incidental catch or direct mortality due to surface longlines it is necessary to know the CPUE of each boat and gear type.Moreover, it is preferable to estimate the CPUE according to fishing operations.For this reason, the current database in the ICCAT (http://www.iccat.es/)should be improved by including data about boat and gear types for it to be useful for estimating the incidental catch or direct mortality in the Mediterranean due to surface longlines.

TABLE 1 .
-Boat strata used in this paper and their respective abbreviations.

TABLE 2 .
-Number of boats with onboard observers and total number of fishing boats (in parentheses) each year according to boat strata.Abbreviations as in Table1.

TABLE 3
. -Number of hooks in thousands (first number) and fishing operations (second number) observed for each boat strata and year.In brackets, the number of by-catch (first number) and dead turtles (second number).The asterisks indicate significant inter-annual differences.Abbreviations as in Table1.

TABLE 4 .
-Total effort, sea turtle mortality, and mortality rates for boat strata during 1999-2004.Abbreviations as in Table1.R: number of incidental turtle catches / number of hooks (in thousands); F: number of incidental turtle catches / number of fishing operations; R dead : number of dead turtles / number of hooks (in thousands); F dead : number of dead turtles / number of fishing operations.There are two values in each square of the χ 2 -value column, the top value is the χ 2 -value if the distribution of hooks is random, and the bottom value is the χ 2 -value if the distribution of the fishing operations is random.The numbers with two asterisks indicate the highest values.