The objective of the meeting was to conduct the stock assessment for Mediterranean swordfish. Several types of assessment models, including production models and VPA, were applied to the available data. RÉSUMÉ L’objectif de la réunion était d’évaluer le stock d’espadon de la Méditerranée. Plusieurs types de modèles d’évaluation, y compris des modèles de production et VPA, ont été appliqués aux données disponibles. RESUMEN El objetivo de la reunión era realizar la evaluación del stock de pez espada del Mediterráneo. A los datos disponibles se aplicaron varios tipos de modelos de evaluación incluyendo modelos de producción y VPA. KEY WORDS Swordfish, VPA, assessment
1. Opening, adoption of Agenda and meeting arrangements The meeting was held at the ICCAT Secretariat offices in Madrid. Dr. George Tserpes, meeting Chairman, opened the meeting. Mr. Driss Meski, Executive Secretary, welcomed participants (“the Group”).
The Agenda (Appendix 1) was adopted with some changes. The Group noted the reduced number of participants and regretted the lack of participants from countries traditionally involved in the swordfish fisheries. The List of Participants is attached as Appendix 2 and the List of Documents presented at the meeting is attached as Appendix 3. The following participants served as rapporteurs:
1, 9, and 10 P. Pallarés
2 J.M. Ortiz de Urbina
3 A. Di Natale
4 P. Kebe and G. Tserpes
5 P. Peristeraki
6 G. Tserpes
7 G. Tserpes and V Restrepo
8 J. Neilson
2. Description of fisheries Mediterranean swordfish fisheries are characterized by high catch levels. It should be noted that average annual reported catches (on average about 15,177 t from 1984 to 2005; Table 1) are similar to those of the North Atlantic. The Mediterranean is a much smaller body of water compared to the North Atlantic. However, the potential reproductive area in the Mediterranean is probably relatively larger than that in the Atlantic. Further, the productivity of the Mediterranean Sea is thought to be very high.
Swordfish fishing has been carried out in the Mediterranean using harpoons and driftnets (drifting gillnets) at least since Roman times. Currently, swordfish fishing is carried out throughout the Mediterranean Sea. The biggest producers of swordfish in the Mediterranean Sea in the recent years (1997-2005) are Italy (44%), Morocco (23%), Greece (10%), and Spain (9%). Also, Algeria, Cyprus, Malta, Tunisia, and Turkey have fisheries targeting swordfish in the Mediterranean. Incidental catches of swordfish have also been reported by Albania, Croatia, France, Japan, Libya, and Portugal. The Group recognized that there might be additional fleets taking swordfish in the Mediterranean, for example, Israel, Lebanon, Egypt and Monaco, but the data are not reported to ICCAT or FAO. Furthermore, a paper presented at this meeting (SCRS/2007/115, by Orsi Relini et al.) provides information about the activity of a French gillnet fishing fleet which operates in the Ligurian Sea, also within the marine “Pelagos” Mammal Sanctuary. According to this report, the total number of driftnetters has grown constantly, from 46 vessels in year 2000 to more than 100 vessels in 2006, in the studied area, and these vessels are reported to catch also swordfish. The SCRS points out that catches from this fleet have never been reported to ICCAT.
Mediterranean total swordfish landings showed an upward trend from 1965-1972, stabilized between 1973-1977, and then resumed an upward trend reaching a peak in 1988 (20,365 t; Table 1, Figure 1). The sharp increase between 1983 and 1988 may be partially attributed to improvement in the national systems for collecting catch statistics. Since 1988, the reported landings of swordfish in the Mediterranean Sea have declined, and since 1990, they have fluctuated between about 12,000 to 16,000 t. In 2005 catches were 14,601 t.
In recent years, the main fishing gears used are surface longline (56% of the total catch) and gillnet. Most of the previously mentioned countries operate longline fisheries, and in 2005 driftnet fisheries reported were mostly limited to Morocco. There are also other countries known to be fishing with driftnets that do not report their catches. Swordfish are also caught with harpoons and traps, but traps do not target swordfish. It should be noted that since the beginning of 2002 driftnet fishing has been banned in EU countries and this will influence the catch data beginning in 2002.
There is a high demand for swordfish for fresh consumption in most Mediterranean countries.
A description follows for fisheries of those nations that attended the meeting. See Figure 2 for reference to particular locations mentioned below). For additional information about fisheries for some nations not attending the meeting see the 2003 Detailed Report.
EC- Greece The Greek swordfish fleets exclusively use drifting longlines and operate throughout the eastern Mediterranean basin. About 250 vessels were involved in the swordfish fishery in 2006. Most of them entered the fishery occasionally, mainly during the summer months.
The swordfish fishing season lasts from February to the end of September, as there is a closed season in the Greek Seas from October to January, aiming to protect recruits.
Swordfish comprises the main bulk of large pelagic catches in the Greek seas and its production during the 2006 fishing season was estimated at 1,375 t. The Greek swordfish production is rather stable over the last decade.
EC- Italy The Italian swordfish fishery has a long historical tradition. Recent catches usually account for a total between 6,000 to over 7,000 t per year, with slight variability from year to year, according to various factors. The largest fishery, in terms of number of vessels, is the longline fishery with about 1200 vessels from 7 to over 30 meters in length. The fishery is currently carried out from late February to December, in many Mediterranean areas. The most significant changes in the fishing strategies occurred in the last ten years, due to the increase in tuna longlining in the spring, implying a parallel decrease in swordfish longlining. This is changing again since 2006, when the swordfish fishery became more relevant due to different conditions in the bluefin tuna fishery. In the last two years, some vessels have started fishing trials using very deep longline in the southern Tyrrhenian Sea but information on this activity is not available. The swordfish target longline fishery provides the highest catch, while smaller quantities are provided by the tuna longline fishery as by-catch. The driftnet fishery was formerly the most important fishery for swordfish but, according to the EC Regulation, it has been banned since January 2002. Recent catches now come from unclassified nets. The traditional harpoon fishery in the Strait of Messina catches very small quantities of swordfish, while even smaller catches are reported in tuna traps.
The former EC legislation concerning the minimum size for Mediterranean swordfish (120 cm LJFL) was cancelled in 2000 and since then the previous measure, which already existed in the Italian regulation (140 cm UJFL), came into force again.
EC- Spain The Spanish swordfish fishery in the Mediterranean dates back to the early 20th century. Its expansion was initiated in the 1960-1970 period, and it has been stable since the 1980s (SCRS/2003/042). Fishing is carried out mainly by surface longline. Swordfish are also caught occasionally by semi-pelagic longline (“piedri-bola”) and as by-catch of the longline fishery that targets bluefin tuna and albacore.
The Spanish swordfish fishery in the Mediterranean is characterized by the heterogeneity of the fleet and by the composition of the gears, as well as by the changes in fishing strategy. The fleet, which can be comprised by as many as 145 vessels, has evolved and currently presents the following average characteristics: 16 m in length, 166 hp engines, and 28 GRT. In addition, the traditional longline gear is being substituted by the American style longline, which is being used by about 29% of the vessels. The fishing area extends from the Iberian Peninsula to 06ºE and up to the limits of the Moroccan and Algerian territorial waters. The major activity takes place in the summer and autumn months.
In 2006, swordfish catches amounted to 1,592 t (of which 1,190 t were caught by surface longline), similar to the catch level for the period 2000-2002. The American style longlines showed an increase in the mean weight of the catch. As regards fishing effort, it has been stable recently.
Swordfish fishing by surface longline in the Mediterranean is subject to regulation by the Decree of 27 July 2006 (APA/2521/2006) that regulates the technical characteristics of the gears and the minimum size for the catch (90 cm LJFL), among others.
Morocco The Moroccan swordfish fishery in the Mediterranean Sea has been developed since 1983. About 320 vessels are currently involved in this fishery, using mainly driftnet and surface longline (SCRS/2006/125). The boats are, on average, 13 m in length, 13 GRT, and have 110 hp engines.
The most important fishing grounds are located in the Strait of Gibraltar and the southern Alboran Sea. The first area remains the most important in terms of the size of the fleet targeting swordfish. In the Strait of Gibraltar, swordfish fishing takes place throughout the year, with a highest activity from April to October. In the southern Alboran Sea, fishing occurs almost the entire year. Minor catches of swordfish are also taken occasionally by traps and purse seiners.
After the peak landings of 4,900 t registered in 1997, the swordfish catch dropped to stabilize around 3,000 t. In 2005, the catch decreased by about 22% with respect to the previous year. The remarkable change in this fishery during the five last years is the significant reduction in driftnet catches and the increase in longline catches, due to the implementation of the National Plan for banning the driftnet activity following the ICCAT Resolution related to the ban of driftnets in the Mediterranean Sea .
The by-catches of this fishery include mainly small tunas, sharks, billfishes and bluefin tuna.
The size of the landed fish varies according to the fishing area. In the Strait of Gibraltar, the mean size of swordfish is about 145 cm. In the Alboran Sea, the fish have a smaller mean size (106 cm) (SCRS/2003/053).
In Morocco, the regulation of swordfish fishing concerns the establishment of a minimum commercial size of 120cm (25kg) (Decree No.1154-88 of 3 October 1988); the establishment of 2.5 km maximum length for driftnets; the prohibition of mesh size less than 400 mm (Circular No.1232 of 11 March 1991), and a freeze on fishing effort through the suspension of the investments for vessel construction since 1992(Circular note No. 3887 of 18 August 1992).
3. Biological data According to the available information (Anon. 1996), in previous years only a few swordfish from the Mediterranean are reported to exceed 200 kg. In recent times, a slightly higher percentage of large swordfish has been reported from several fisheries. The majority of the Mediterranean catch is comprised of individuals less than 3 years old and the average size is much lower than in the Atlantic. The fact that the fishery is still mostly based on 2-3 young year-classes makes it vulnerable to recruitment changes.
Growth studies of swordfish in the Mediterranean, carried out by several teams, using both anal fin spines and length frequency data, all show a similar pattern of growth. It is also well known that Mediterranean swordfish have sexually dimorphic growth, with males having a lower length-at-age, and achieving a smaller asymptotic size than do females. The growth equations adopted by the GFCM/ICCAT Working Group in 1995 are those published by Tserpes and Tsimenides (1995) and still used as follows:
Linf = 238.60 (1 – e –0.185 (t + 1.404) ) for sexes combined
Linf = 203.08 (1 – e –0.241 (t + 1.205) ) for males
Linf = 226.53 (1 – e –0.210 (t + 1.165) ) for females.
A very recent paper (SCRS/2007/117, by Valeiras et al.) found different growth rates in the swordfish present in the western Mediterranean. This paper provided growth equations by sex, based on anal spines reading. It is important to note that the sampling was conducted in an area where mixing between the Atlantic stock and the Mediterranean stock is possible. According to this paper, the growth rate is lower compared to previous studies, particularly at young ages. SCRS considers that the situation on this crucial area of the Mediterranean should be further investigated, due to its relevance for the understanding of the mixing between the two stocks.
Spawning generally occurs in spring and summer, with peaks in June and July, and variations in timing may be due to a variety of environmental and oceanographic influences. The most important spawning areas in the Mediterranean, according to current knowledge, are around the Balearic Islands, the southern and central Tyrrhenian Sea, the Ionian Sea and the Strait of Messina, and there is a strong indication that spawning areas also exist in the eastern Mediterranean. Juveniles are found throughout the Mediterranean but often tend to concentrate close to the coast, mostly in autumn.
According to a review of the biological information available for the Mediterranean swordfish (SCRS/2001/050), major differences with the Atlantic stock have been noticed. Mature females as small as 110 cm LJFL have been observed and the estimated size at which 50% of the female population is mature occurs at 142 cm (SCRS/95/045). According to the growth curves used by SCRS in the past for Mediterranean swordfish, these two sizes correspond to 2 and 3.5 year-old fish, respectively. At 125 cm about 20% of the females in the Mediterranean would be mature. Males reach sexual maturity at smaller sizes.
Table 2 summarized biological information available for the Mediterranean swordfish.
Environmental factors It is well known that swordfish catches are highly affected by prevailing environmental factors. For example, the catches of swordfish are affected mainly by the presence of a stable thermocline during late spring and summer. In fact, when using driftnets, all catches are done just above the thermocline, where spawning occurs. Swordfish catches are also affected by the moon phase. In the presence of a full moon, higher catches are recorded for traditional longlines, whilst lower catches are recorded for driftnets (SCRS/94/86, SCRS/91/65). Areas characterized by higher turbulence also seem important for swordfish. Higher catches are recorded in the Tyrrhenian and Alboran Seas.
A paper presented at this meeting (SCRS/2007/115 by Orsi Relini et al.) provides a study about the correlation between winter North Atlantic Oscilation (NAO) and the longline CPUE series in the Ligurian Sea over the period 1990-2006. According to this study, an inversely related correlation has been detected, even if other environmental factors should be possibly taken into account. This study also reports that swordfish recruitment in the Ligurian Sea does not show any correlation with the winter NAO index. Temporal differences of SST have been also related to the growth rate of 0 group swordfish in the eastern Mediterranean (Peristeraki et al., 2007).
During this meeting, it was discussed that other possible correlations could be explored and studied, including those between the apparent distribution and concentration of swordfish and the eastern Mediterranean Transient (EMT) index, which seems a very important factor to be taken into account in the Mediterranean, able to induce relevant changes in the pelagic environment, in the spatial and temporal distribution of many pelagic species including swordfish and, then, in the fishery.
It was recommended that more work should be carried out in order to identify better the effects of the environment on swordfish biology, ecology and fishery.
4. Catch data
Two documents related to catch data were presented.
Document SCRS/2007/106 summarized the information on the by-catches and discards data in the Greek swordfish fishery. Landed swordfish represent 84% of the total catch in weight.
SCRS/2007/108 described the discards of undersized swordfish in the Greek longline swordfish fishery. It was noted that between 15 to 17 tons were discarded yearly. After some discussion the Group noted that discards information was not reported in general to the Secretariat in the Mediterranean area, by countries other than Greece.
4.1 Task I In Table 1the Secretariat presented the Mediterranean swordfish catch data reported to ICCAT for the years 1950 to 2006 by flag, fleet and gear. The Group decided to focus only on data up to 2005 as the 2006 reported information is incomplete. During the revisions of the catch table, the lack of the Tunisian catch for 2005 and the Algerian catch for 2003 was noted. The Group was also very concerned about the low level of swordfish reported by Italy for some past years. In order to fill the gaps in the catch time series the Group decided:
− To carry forward the 2004 Tunisian catch (791 t) to 2005.
− To use the 665 t reported in the Algerian Annual Report for 2003 in the assessment and as unofficial data in the Task I database.
− To re-adjust the Italian catches in 1990 (from 5,224 t to 9,104 t), 1991 (from 4,789 t to 8,538 t) and 1995 (from 6,725 t to 7,350 t) following the Italian scientists’ advice.
In addition, the ICCAT Task I and FAO FIGIS data were compared and found to be in agreement for the period after 1967, but showed discrepancies for the period from 1950 to1967. The Group decided to adopt the higher catch figures held in the FIGIS database and recommended that the ICCAT and FAO datasets be harmonized for the period 1950-1967 as they already are for later periods and other species.
Accordingly, Table 1 was updated and presented again by the Secretariat and approved by the Group.
4.2 Size and catch at size The catalog of all Task I and Task II had been distributed earlier (April 2007) to all potentials participants in order to review the gaps and deficiencies and to submit revisions to the Secretariat. Unfortunately, no feedback was received. During the meeting, the same file was updated and presented again to the Group. At the beginning of the meeting, new Task II size information were available for Moroccan gillnet for the years 1999-2005 and also for Italian gillnet and longline by month. According to the new Italian size sample available by fleets, the Group decided to breakdown the Italian catch reported by area in order to match it with the size samples.
The Secretariat presented the substitution rules used (Table 3) to create the catch at size following the rules adopted in 2003 (SCRS/2003/015 and SCRS/2003/050). The catch-at-size file summarized in Table 4 for the years 1985 to 2005 were converted to catch at age (Table 5) by applying the same slicing procedure used in 2003 and the same growth equation (inverse Von-Bertalanffy equation) .
5. Relative abundance indices
Five papers concerning catch rate data were presented.
SCRS/2007/118 presented the updated standardized catch rates for swordfish from the Spanish longline fleet in the Mediterranean Sea for the years 1988-2005. Data included 18,630 observations that were analyzed by means of GLM techniques. The effects of year, area and quarter were considered, and all factors (including interactions) were significant. Annual standardized CPUEs declined rapidly from 1988 to 1992, and more gradually until 1999. It has remained stable thereafter.
SCRS/2007/107 presented annual standardized catch rates from the Italian and Greek fleets operating in the central eastern Mediterranean. The analysis included data from the Greek longline fisheries operating in the eastern Mediterranean and the Sicilian longline fisheries operating in the Tyrrhenian Sea and the Straits of Sicily, for the years 1987-2005. Indices were estimated by means of GLM techniques and results did not demonstrate the presence of any particular trend over time.
SCRS/2007/115 presented a time series of nominal catch rates from the Italian swordfish longline fleets operating in the Ligurian Sea, from 1990 to 2006. No significant trend was observed during the studied period of time. An inverse relation of the swordfish CPUE with the NAO index values was also observed. The Group was pleased to receive this contribution dealing with environmental effects on catch rates, an area where the state of knowledge has been considered deficient. However, the Group encourages further investigation about the effects of environmental and oceanographic factors specifically related to the Mediterranean Sea.
SCRS/2007/116 presented an analysis of the standardized catch rates of the Moroccan driftnet fishery from the Mediterranean Sea. Nominal CPUE, by month and boat, from 1998 to 2006 were analyzed by means of GLM techniques. Standardized catch rates did not show any trend throughout the years. It was commented that high CPUE values were observed compared with other Mediterranean driftnet fisheries, which might be attributed to specific characteristics of the exploited area, which extends around the Gibraltar Straits.
SCRS/2007/119 presented a detailed description of the Italian nominal CPUE time series (1985-2006) from the southern Italian gillnet and longline fleets. The gillnet CPUEs showed a slightly increasing trend, while the longline CPUEs showed a moderately decreasing trend. The exploitation pattern of the fleets was considered stable for the last two decades.
6. Stock status results 6.1 Production model evaluations 6.1.1 ASPIC The non-equilibrium surplus production model (ASPIC, cataloged version 5.16) was applied to catch and effort data for Mediterranean swordfish. The input data used in these analyses are presented in Table 6 In order to better inform the model, recorded catches from 1950 to 2005 were used. The ICCAT Task I and FAO FIGIS data were compared and found to be in agreement for the period after 1967, but showed discrepancies for the period from 1950 to1967. The Group decided to adopt the higher catch figures held in the FIGIS database and recommended that the ICCAT and FAO datasets be harmonized for the period 1950-1967 (they are already harmonized for later periods and other species). For these analyses, a composite CPUE pattern was developed as the weighted average of the Italian longline (SCRS/2007/107), Greek longline (SCRS/2007/107), Moroccan gillnet (SCRS/2007/116), Italian Gillnet (Anon. 2004), Spanish longline (SCRS/2007/118), and Japanese longline (see Anon, 1996) catch rate time series. It was noted that the Japanese data are related to the swordfish by-catch in the tuna longline fishery. Weighting was used in this case, due to concerns that some of the fleets from which time series were available represented a relatively small area of fishing and/or typically represented small volumes of the total Mediterranean catch of swordfish. In this case, a weight of 4 was assigned to the Italian and Moroccan indices, a weight of 2 to the Greek and Spanish time-series, and a weight of 1 to the Japanese time-series. The resulting CPUE pattern is shown in Figure 3.
The production model was first fit to catch and effort for the period 1968-2005 (reflecting the assessment conducted in 2003). In this case, there was insufficient information in the data with which to freely estimate all model parameters. The model convergence could be achieved by fixing the initial biomass ratio, but the Group considered that assuming the biomass was at an unfished level (K) in 1968 was an incorrect one in light of the reported catches since 1950. The Group decided to fix B1968 at .75K for this run as a better representation of the situation at that time. In order to inform the model and possibly improve estimates of stock productivity and current status, the data from 1950 through 2005 were then fit. In this case, it was possible to freely estimate all model parameters, although when doing so the modeled stock dynamics prior to the mid 1980s was unexpected, showing a build-up of biomass from very low levels in the early 1950s. Therefore, the Groupdecided to also conduct a run assuming the stock was at an unfished level in 1950 even though some catches are known to have occurred before that time. Across the models, (see Figure 4 and Appendix 4 for details supporting the model fits), the estimates of population status in the most recent year indicated a stock that was at or somewhat below the ICCAT Convention objective while recent fishing mortality was somewhat above the level that would permit the stock to attain the level necessary to attain the Convention objective (MSY levels). While the uncertainty in these results based on bootstrapping is large (Figure 5), the weight of the evidence supports these conclusions (Table 7). The median results of the model outcomes (Figures 6a, b) indicate that the fishery underwent a rapid expansion in the 1980s resulting in F’s likely at or above FMSY and a slowly declining stock which has recently likely fallen below the level which can support MSY over the long-run (Figures 7a, b).
6.1.2 Tserpes and McAllister Method (TSM) A non-equilibrium production model was applied based on the approach followed in SCRS/2007/109. The model used total catch data for the 1987-2005 period and a combined CPUE series, the same as used in ASPIC. The XSA estimates of average F were used to estimate the harvest rate at the beginning of the examined period and consequently the initial biomass fraction
Based on the ICCAT XSA assessment the values of F and M for the beginning of the period were fixed to 0.42 and 0.20 respectively. The best fit was provided for r = 0.67 and k=90547mt. Observed and predicted indices are shown in Figure 8. Based on the above estimates equilibrium MSY was found to be equal to 15166mt. The corresponding rates for fishing mortality and biomass are: and mt. Annual catches in the latest years are around to MSY, while stock biomass levels are stable but about 12% lower than Bmsy (Figure 9).
6.2 Age structured models 6.2.1 XSA assessment The XSA model was implemented using the code developed in R-language (see Appendix 5) under the auspices of the FLR-project (Kell et al., 2007; http://www.flr-project.org/). Catch-at-age tables included ages 0 to 10 (plus group) and six tuning data sets were available from the following fleets: Italian longliners–ITLL (SCRS/2007/107), Greek longliners-GLL (SCRS/2007/107), Moroccan gillnetters-MODN (SCRS/2007/116), Italian gillnetters-ITDN (SCRS/2003/040), Spanish longline-SPLL (SCRS/2007/118), and Japanese longliners-JALL (see Anon, 1996). Greek, Italian, and Spanish longline CPUE series were considered as representative of 2-9 age-group abundances, while for the rest, the 3-9 age-group was assumed. Full maturity was assumed from age 4 onwards and 50% at age 3. Zero maturity was assumed for the younger ages.
A series of preliminary runs with different parameterization were performed and based on the fleet catchability diagnostics the final model was based on runs assuming q independent of year-class size for all ages except 0 and 1, constant q after age 6, as well as population and F shrinkage. Natural mortality was considered equal to 0.2. Figure 10 llustrates the catchability residuals by fleet and age. In general, residuals do not show any specific pattern for the younger more abundant age-classes, while they are positively biased in the older ages.
Tables 8 and 9 present the estimates of fishing mortality and population numbers-at-age, respectively. As in the 2003 assessment, recruitment appears to be consistent without any especially strong or weak year classes. The mean Fs for ages 2-5 are plot against year in Figure 11.
Both total and spawning stock biomass estimates remained stable during the last decade (Figure 12).
6.2.2 VPA The software VPA-2BOX was used to conduct a sequential population analysis using the same input data as in the XSA analyses. The following assumptions were made:
− Indices: A lognormal error structure was assumed for all indices. The index selectivities were estimated using the partial catches (including all ages, from 0 to 10+).
− F ratios: F10+/F9 = 1.0 in all years.
− Terminal year Fs: Estimated for ages 5 and 8. Assumed values: F0=0.01*F5, F1=0.4*F5, F2=F5, F3=F5, F4=F5, F6=F8, F7=F8, F9=F8.
The fit to the data was poor. The coefficients of variation for the estimates of F5 and F8 in 2005 were 15% and 81% respectively. The fits to the individual indices were rather poor in some cases (see Figure 13).
The Group examined the estimated selectivities at age for the various indices (Figure 14) and concluded that they seemed reasonable, given the size composition of the catches for the corresponding fleets.
The overall results of the model fit are given in Table 10 and Appendix 6. In terms of stock size, the estimated recruitment (age 0) trend is rather constant at slightly above one million fish per year, and the trend for ages 2 to 8 is a declining one (Figure 15). The estimated trend in spawning biomass shows an overall decline of 40% between 1985 and 2005, but the level of exploitable biomass is estimated to have varied without a trend (Figure 16).
The estimates of fishing mortality (apical values) are shown in Figure 17, suggesting that it has varied without a trend during the last decade. The estimated selectivities at age by 5-year time periods are shown in Figure 18, suggesting that selectivity has remained relatively constant since 1990.
6.2.3 Comparison of the age-structured results XSA and VPA-2BOX are different implementations of sequential population analyses. For this Mediterranean swordfish assessment, they were used with the same data and similar, although not identical, assumptions.
The Group compared the results obtained with both methods in terms of recruitment, fishing mortality, and biomass trends, and found them to be very similar (Figure 19). Small discrepancies are evident only for older ages (e.g., F at age 9). But, overall, both methods provide a very similar perception of stock status.
6.2.4 Equilibrium yield-per-recruit analyses The VPA-2Box and XSA results were used as the basis for yield-per-recruit analyses which are a form of long-term projection. The input age-specific vectors are given in Table 11. The resulting equilibrium estimates for several biological reference points are given in Table 12. These are per-recruit results, scaled to a recruitment level of 1,059,533 age-0 swordfish (the mean 1985-2002 level estimated in the VPA-2box). Figure 20 shows the equilibrium yield levels obtained for different fishing mortality multipliers of the current selectivity vector.
Bootstraping (1,000 iterations) was used to characterize the uncertainty in the VPA-2BOX assessment of recent status for Mediterranean swordfish. There is a high degree of variability in the estimates of recent status, but all of the bootstrap outcome indicate this stock is both overfished and undergoing overfishing (Figure 21).
6.2.5 Summary of age-structured assessment results. The two age-structured models used in the assessment give very consistent results. During the past 20 years, fishing mortality has fluctuated at high levels, usually doubling the value of natural mortality. The value of natural mortality is sometimes used as a proxy for FMSY in data-poor situations, which would suggest that overfishing has occurred during this time. Both models estimate that spawning biomass has declined between 1985 and 2005 (a decline between 24% and 38%, depending on the model). On the other hand, recruitment has varied (CV 12%) during this time period without a trend.
Results of equilibrium yield-per-recruit analyses that are based on age-structured assessments also indicate that growth overfishing is taking place. Depending on the model used, current (2005) fishing mortality is 1.7 or 2.1 times higher than the value that would maximize yield per-recruit. In the case of the VPA, assuming a constant level of recruitment, a 69% reduction in fishing mortality to the FMAX level would result in a modest (7%) increase in long-term yield and a substantial (more than double) increase in spawning biomass.
In addition, current F is expected to result in a spawning stock biomass per recruit (SPR) at about 8% of the unfished condition, a level which is considered to result in a non-negligible risk of rapid stock decline. Fishing at FMAX given the current selectivity would be expected to result in an SPR of 20%.
Using FMAX as a proxy for FMSY, the Convention's objective, suggests that the Mediterranean swordfish stock is in an overfished condition and that overfishing is taking place (Figure 22). Note, however, that these conclusions are based on deterministic analyses of the available data. The level of uncertainty in these estimates has not been evaluated.
6.3 Stock status summary Two forms of assessment, both with high degree of uncertainty, gave a consistent view of declining stock abundance, but differed in the extent of the decline, in the sense that some models suggested little changes in the last decade. Estimates of population status from production modeling using a longer time-series of catch and effort for which we have less confidence indicated a stock level that was most likely about 10% below that necessary to achieve the ICCAT Convention objective while recent fishing mortality was about 25% above the level that would permit the stock to attain MSY levels. The results of the production model assessment indicate that the fishery underwent a rapid expansion in the 1980s resulting in Fs likely at or above FMSY and a slowly declining stock biomass which has recently most likely fallen below the level which can support MSY. Estimates of stock status from virtual population analysis using a shorter time series of catch and effort data for which we have more confidence, indicated a relatively stable spawning stock level and stable recruitment over the past 20 years, but that level is less than half that necessary to achieve the ICCAT Convention objective and estimates of recent fishing mortality rates from this form of assessment are about twice that which, if continue into the future, is expected to drive the spawning biomass to a very low level (about 8% SPR) within a generation. Those low levels are considered to give rise to non negligible risks of rapid declines in the stock, although no such a signal has yet been observed in the Mediterranean swordfish fisheries.
7. Projections 7.1 Production model projections The combined production model bootstrap outcomes were projected forward under several different future constant catch scenarios. Catches in years 2006 and 2007 were assumed to have been at the 2005 level (14,600 t) and catches in subsequent years through 2015 were assumed to equal either 10,000, 12,000, 14,300 (the approximate MSY) or 16,000 t. The projections indicate that catches in excess of 12,000 t starting in 2008 are likely to result in future decline in stock status (Figure 23).
7.2 VPA scenarios The VPA-2BOX model bootstrap outcomes were projected forward under several different future constant catch scenarios. Projected recruitment was taken as a random draw from the 1985-2002 time series and the recruitments (and cohort strengths for the corresponding ages) for 2003-2005 year classes were also assumed to be of the same dimension. Catches in years 2006 and 2007 were assumed to have been at the 2005 level (14,600 t) and catches in subsequent years through 2015 were assumed to equal either 10,000, 12,000, 14,300 (the production model approximate MSY) or 16,000 t. The projections indicate that catches in excess of 14,000 t starting in 2008 are likely to result in future decline in stock status (Figure 24).
Using the Fisheries Library in R (FLR) framework that was developed in the frames of the European Fisheries Management System (EFIMAS) project (http://www.flr-project.org/), four exploitation scenarios were applied.
The operational model used in all scenarios was based on the estimated (through XSA) stock population at age at the beginning of the year 2000. This was used as a starting population and each projection scenario was simulated 250 times for a period of 20 years by assuming:
a) Natural mortality equal to 0.2
b) An empirically estimated Beverton-Holt S/R model. As VPA results has not allowed estimation of model parameters (due to low contrasts), those were estimated empirically (Hilborn & Walters, 1992) assuming that half of the current stock sized will produce about half of the maximum recruitment.
The initial vector of abundance at age and the predicted recruitment from the S/R model, were modelled by assuming normally distributed errors with CVs equal to 15% and 10%, respectively. Thus, the corresponding values were drawn randomly from the assumed distributions.
All scenarios apart from the first one, attempt to examine the effects of global fishery closures during the recruitment period. Recruitment extends from September to February, with its peak from October-January. Such closures have been extensively discussed in the past (Di Natale et al., 2002) and it has been assumed that they will mainly affect the fishing mortality of age 0 fish (up to 71cm of LJFL in the catch-at-age table).
Scenario 1: The current exploitation pattern It was assumed that Fs at age will be equal to the average values estimated for the 1995-1999 period for the whole of the projection period. Based on YPR, this F level is about twice that which would permit the stock to attain MSY level and in the long-run, the expected spawning stock biomass would be around 50% of BMSY or around 10% of the unfished biomass, which is considered very low and resulting in non-negligible risk of rapid stock decline although such a signal has not yet been observed in the Mediterranean swordfish fisheries.
Results are summarized in Figure 24. The scenario indicates a stable production pattern with annual catches being around to 14,000-15,000 t (probability > 50%) of which juveniles equal 5,800-6,200 t. The number of juvenile fish in the catch will mostly vary from 380,000-420,000 fish corresponding to 72-74% of the total catch number. In general, the reported rates for the period 2000-2004 are in the range of the model predicted values.