Fishing strategy choices of purse seines in the Mediterranean: implications for management

Authors


*Tel: 30-210-985-6703.
Fax: 30-210-981-1713. Email: etsitsik@ath.hcmr.gr

Abstract

ABSTRACT:  An observer's sampling scheme, that employed fisheries scientists onboard fleet vessels was used to examine temporal fishing tactics and strategies affecting catches of the purse seine fishery in the Mediterranean. The month, water depth and the fishers' behavior were found to have an effect on total and Trachurus spp. retained catches, with fishers' behavior explaining the largest percentage of the data variation. The distance of the fishing ground from port and the market price modulated both the retained catches and the fishing location choice. Results confirmed that fishers while in a specific fishing ground developed strategies that would allow them to retain, and thus land, the best possible combination of landings × market value. The current findings also revealed that, when constrained by physical and economic conditions, fishers preferred to minimise risk rather than maximize landings. The observed major switches in fishing strategy were attributed to fishers' risk attitude response towards higher profitability. As the Mediterranean fishery system is mainly based on control effort and technical measures regimes, the current findings are discussed in the light of the need to consider additional information to management plans and decisions.

INTRODUCTION

The Mediterranean Sea is considered to be one of the most interesting semi-enclosed seas because of the great range of processes and interactions occurring within. The north-east Mediterranean, i.e. the Aegean Sea, is dominated by important basin and sub-basin features such as gyres, jets, eddies and meandering currents, reflecting its complex geometry, bathymetry and highly variable atmospheric forcing. The Mediterranean fisheries are extremely diverse, targeting a great number of species, and have an extensive scope of fishing gear and methods. Catches are highly multi-specific. Fishing in the Mediterranean is a major economic activity in terms of jobs, revenues and food supply. The management of fishing effort alongside technical measures is the main tool responsible for delivering sustainable fisheries in the Mediterranean Sea. Catches of many species peaked in the late 1980s and early 1990s and have declined since. Although fishing in the Mediterranean has not undergone any dramatic event, some overfishing symptoms are evident for the most important commercial species.

The knowledge of how fishers allocate their fishing effort in space and time is essential to understand how a fishery develops, and the relationship between catch rate and abundance.1–3 Understanding fishing strategies is also vital in predicting how a fishery might respond to proposed management changes such as effort or area restrictions, and in the formulation of management policy.3,4 Nonetheless, the study of fishers' behavior has generally received little attention so far. Some authors suggest that fishers' behavior depends on economic incentives, i.e. expected returns and its variability,1,3–7 while others argue that in addition to economics, other stimuli influence behavior (such as weather, vessel competitions), and should thus be taken into account.8,9

Fishers' behavior has not been used widely in catch analyses and fish abundance estimations. In the absence of research surveys, these estimations rely heavily on collecting representative data from the commercial fishing fleet. The main drawback of several national fisheries statistical systems, in particular catch and effort statistics, is that they use a problematic census method, i.e. fishers are asked to complete a monthly statistical questionnaire referring to their catches and activities during the previous month. This system possesses deficiencies such as memory errors, inappropriate survey procedures, biased estimations and processing operations, and poor coverage of fishing effort.10 To this end and in order to relax the serious shortcomings of the reported catches, it was decided to place scientific observers onboard a purse seine fleet. This allowed for more realistic retained catch estimates to be included in the present analysis.

Purse seining constitutes one of the most important fishing methods in the Mediterranean Sea. Total retained catches alongside retained catches of the two most commercially important fish species (i.e. Mediterranean horse mackerel Trachurus mediterraneus and horse mackerel Trachurus trachurus) of a purse seine fishery were examined during 2002 and 2003 in the north-east Mediterranean. Greek fleets dominate this fishery (>90% of total production). Trachurus spp. are target species for purse seining with an average catch of more than 6.5% of the total yield of the country.11 In Greece, the purse seine fleet consists of approximately 317 units with an average gross tonnage (GT) of 44.5 and an average horsepower of 167.7 (Kapantagakis A, pers. comm., 2006). Each purse seine is responsible for fish searching, catching and transporting the catches to port. Fishing operations are carried out exclusively during the night (20:00–05:00 hours) with each vessel occupying 8–10 persons. The fish in the upper water column are attracted by lamps scattered at the surface, and are finally caught by the encircling net. All vessels conducted daily trips. Management regulations currently in force for the purse seine fishery include mesh size regulations (>14 mm), technical measures such as closed seasons (December–February), closed areas, and fishing prohibitions within specific distances from the coast (100 m).

In this study we analyze factors affecting the retained catches of a typical multi-species Mediterranean purse seine fishery incorporating aspects of fishers' behavior and strategy. Interviews with fishers onboard each vessel were conducted by questionnaires developed to allow researchers to describe the most important variables affecting fishers' behavior. Further, fishers' behavior was simplified and calibrated empirically with survey fisheries data. Although fishers' behavior has been successfully studied in other works,1,3–6,12–14 this is the first attempt to model this factor in a typical Mediterranean multispecies purse seine fishery.

MATERIALS AND METHODS

Onboard observations

The data used for the analysis were the total and the Mediterranean horse mackerel and horse mackerel retained catches (kg per day at sea) of a typical purse seine fleet operating during two fishing periods: March–November 2002 and March–November 2003. Purse seines operating in the study area share similar vessel characteristics: length of 17.2–23.2 m (standard deviation, SD ± 1.86), 295–300 horsepower (SD ± 2.89) and 52.09–59.51 GT (SD ± 3.8). The data collection scheme included eight days of monthly sampling (randomly chosen) by scientific observers onboard the purse seine fleet. Each scientific observer attended a fishing trip once a week so as to ensure objectivity in the sampling procedure. At the end of the study period a total of 144 fishing trips had been conducted obtaining a 29% monthly coverage of the total fleet.

Reasonable cooperation was developed with the captains that enabled the collection of representative data. For each trip the data recorded were date of the trip, fishing area, water depth, species composition and weight of each species, number of boxes per species and total weight of the species retained (i.e. landings). The factor of water depth corresponds to the depth of the water column and not to the depth of the sampling area.

Interview scheme

Fishers generally target either the most abundant species at a fishing ground or the species providing a satisfactory combination of landings × market value.4 Notwithstanding fluctuations in fish abundance, the variability of landings is also affected by fishers' response to economic and regional incentives. In the Mediterranean purse seine fishery, anecdotal evidence suggests that the market prices drive not only the choice of fishing location but also the retained catches.

Interviews onboard each vessel were conducted in order to identify and incorporate important variables affecting fishers' behavior in the present analysis. From fishers' responses it became obvious that both fishing operation and location choice is a multilevel decision process. Random selection was rarely mentioned. Most of them referred to market prices, weather, distance of the fishing areas from the port, and species abundance. Based on these answers and on the assumption that fishers would always try to achieve high revenues from their catches, we selected a subset of variables and constructed an arithmetic scale questionnaire which was further verified for its accuracy from all fishers (Table 1). Apart from abundance, fishers' behavior was found to be modulated on two variables: distance of the fishing area from the port and market prices; that is, fishers' behavior = distance from the port + market prices. The distance from the port is the distance between the port and the fishing area, measured as a straight line. This variable was scaled from 1 to 3, where: (i) 1 represented a fishing area away from the port (∼3 h from the port); (ii) 2, a fishing area a medium distance from the port (∼1.5 h); and (iii) 3, a fishing area near the port (∼40 min). Market price is the price that would be attained in the port for each catch. This variable was scaled from 1 to 3, where: (i) 1 represented low market price (1 Euro/kg); (ii) 2, medium market price (4 Euro/kg); and (iii) 3, high market price (7 Euro/kg). Aiming to attain more representative catch values, it was decided to place two scientific observers onboard purse seines in each fishing trip.

Table 1.  Scaled questionnaire including key factors modulating fishers' behavior
Distance from portMarket priceFishers' behaviorExplanation of fishers' behavior
112Fishing areas away from the port and low market price
123a) Fishing areas away from the port and medium market price
213b) Fishing areas in medium distance from the port and low market price
134a) Fishing areas away from the port and high market price
224b) Fishing areas in medium distance from the port and medium market price
314c) Fishing areas near the port and low market price
235a) Fishing areas in medium distance from the port and high market price
325b) Fishing areas near the port and medium market price
336Fishing areas near the port and high market price

Species selection

The two Trachurus species were chosen to be analyzed because: (i) the relative frequency of these two species was high (15.459 and 14.976%, respectively) (Table 2); and (ii) through the interviews, it was revealed that all fishers operating their vessels in the study area consider them as two of the most important target species. This was also verified by our personal experience gained through the scientific observers who followed these vessels throughout the study period.

Table 2.  Taxonomic composition, relative frequency and relative abundance (% weight) of commercial purse seine retained catches during 2002 and 2003
 FamilyScientific nameRelative frequency (%)
Cephalopods
 LologinidaeLoligo vulgaris4.831
Todarodes sagittatus1.448
Fish
 CarangidaeCaranx rhonchus7.729
Seriola dumerili0.489
Trachurus mediterraneus15.459
Trachurus trachurus14.976
ClupeidaeSardinella aurita10.628
Sardina pilchardus0.965
EngraulidaeEngraulis encrasicolus5.314
MugilidaeMugil cephalus1.932
PomatomidaePomatomus saltator6.763
ScombridaeScomber colias8.213
SerranidaeEpinephelus aeneus0.482
SparidaeBoops boops6.280
Oblada melanura2.415
Sarpa salpa1.449
Sparus aurata1.449
SphyraenidaeSphyraena sphyraena9.178

Data analysis

Retained catch values were analyzed as a function of month of fishing, water depth and fishers' behavior using generalized linear models (GLMs).15,16 The analyses were performed by applying the routines contained in the SPLUS programming environment.17 A GLM provides a way to estimate a function of the mean response (called a link function) as a linear function of the values of some set of predictors. The general form of the GLM is given by equation 1:15

image(1)

where E[y] is the expected value E of the response of a random variable y (here retained catch), g[] is the link function defining the relationship between the response and the linear predictor inline image, β0 is an intercept term, xk is the value of the kth spatial covariate and βk[] represents linear functions of the k covariates.15,16 In GLMs the standard linearity assumption is extended to include any underlying probability distribution from the exponential family (including Poisson, gamma, normal and binomial distributions). We used Gaussian error distribution on the logarithmic transformation of the total and the two species catch rates following an examination of diagnostic plots of the deviance residuals.15 For regression models, residuals are used to assess the importance and relationship of a term in the model, as well as to search for anomalous values. For GLMs there are two additional tasks: assessing how well a model fits, and verifying the form of the variance as a function of the mean response.17 There was no indication of serious violations of the model assumptions; thus, the choice of error distributions was considered adequate.17 The density of points for different covariate values is shown by the rug under the single covariate effects plots (Figs 1–3).

Figure 1.

(a) Diagnostic plot and (b) main effects of month, (c) water depth and (d) fishers' behavior of the model fitted to purse seine total retained catches (log10). Model plots (b-d) represent the contribution of the corresponding covariate to the fitted linear predictor. The ‘rug’ on the x-axis shows the density of points for each covariate included in the model. Dashed lines represent two standard error ranges around the covariate main effect. Error bars, ± two standard errors.

Figure 2.

(a) Diagnostic plot and (b) main effects of month, (c) water depth and (d) fishers' behavior of the model fitted in Mediterranean horse mackerel retained catches (log10). The ‘rug’ on the x-axis shows the density of points for each covariate included in the model. Dashed lines represent two standard error ranges around the covariate main effect. Error bars, ± two standard errors.

Figure 3.

(a) Diagnostic plot and (b) main effects of month, (c) water depth and (d) fishers' behavior of the model fitted in horse mackerel retained catches (log10). The ‘rug’ on the x-axis shows the density of points for each covariate included in the model. Dashed lines represent two standard error ranges around the covariate main effect. Error bars, ± two standard errors.

Model selection

All available variables and their first order interactions were initially included in all three models. Backward stepwise elimination was used to select a set of significant covariates. Analysis of deviance was used to compare models by analyzing changes in deviances relative to the changes in degrees of freedom. To evaluate these differences and compare the models, approximate F-tests based on the approximate degrees of freedom and the corresponding percentage point of the F-distribution were computed, as suggested by Chambers and Hastie.16 The final model for the total and the two-species retained catches was of the form GLM (log10[total/two-species retained catches] = month + water depth + fishers' behavior).

RESULTS

Exploratory data analysis

In total, 18 species were recorded during the study from the purse seines: 16 fish and two cephalopods (Table 2). The same tendency in retained catch values were evidenced between the two studied years. Total retained catches differed considerably between months with the highest values in April and the lowest in November. Highest values of Mediterranean horse mackerel and horse mackerel retained catches were recorded in September and June, respectively (Fig. 4).

Figure 4.

Averaged monthly variations of total, Mediterranean horse mackerel and horse mackerel retained catches (kg/day at sea) during the study period 2002–2003. Error bars, ± standard deviation.

Interview scheme

Fishers' behavior is mainly modulated by two factors: distance of the fishing area from the port, and market prices attained for the species fished (Table 1). Alternative choices were recorded in three cases (fishers' behavior 3, 4 and 5). Each of these choices corresponds to different explanations of fishers' behavior (explanation of fishers' behavior 3a,b, 4a–c and 5a,b, respectively, Table 1).

Model selection

In Table 3 the significance values of GLM covariates for all analyses are given alongside parameter estimates of the final models. The analysis of deviance indicates that the differences observed in total and the two Trachurus spp. retained catches between months, water depth and fishers' behavior were significant (P < 0.05). The final model reduces the null deviance by 68.1, 66.3 and 61.4% for total, Mediterranean horse mackerel and horse mackerel, respectively. Most of the variance is explained by the differences in factors of fishers' behavior and month.

Table 3.  Analysis of deviance and significance values (P-levels) for all GLM covariates of the final models fitted to the logarithm transformation of total, Mediterranean horse mackerel Trachurus mediterraneus and horse mackerel Trachurus trachurus retained catches
Source of variationDevianced.f.Residuals devianceResiduals d.f.FPDeviance Explained
  • Value in parentheses, % Deviance Explained.

  • d.f., degrees of freedom

Total retained catches
 Null  128.2143   
 Month34.7893.513544.5330.00027.1 (39.8)
 Water depth4.1189.41346.9790.0093.2 (4.7)
 Fishers' behavior48.5440.913062.3860.00037.8 (55.5)
Total explained  87.3   68.1 (100)
Mediterranean horse mackerel retained catches
 Null  103.1111   
 Month28.7874.310328.3740.00027.8 (41.9)
 Water depth6.4167.910256.6880.0006.3 (9.5)
 Fishers' behavior33.2434.798128.7860.00032.2 (48.6)
Total explained  68.4   66.3 (100)
Horse mackerel retained catches
 Null  45.6131   
 Month11.3834.31237.8730.00024.7 (40.2)
 Water depth1.1133.21220.6800.042.4 (3.9)
 Fishers' behavior15.6417.611824.1780.00034.3 (55.9)
Total explained  28   61.4 (100)

The contribution of each of the main effects included in the GLM to the variation of total and the two species retained catch accounting for the other factors is given in Figures 1–3.

Assessment of model's fit through diagnostic plots revealed the errors to be normally distributed since the ordered residuals are clustered along the superimposed normal quantile line (Figs 1a–3a).

Results indicate the presence of a seasonal trend in total and horse mackerel retained catch in the area during the study period. Levels of total and horse mackerel retained catch were higher during spring and summer than autumn (Figs 1b and 3b). The Mediterranean horse mackerel retained catch showed a different seasonal trend, with species retained catch increasing with month. Higher values were recorded during September and October (Fig. 2b).

The total retained catch decreased with increasing water depth while the two Trachurus spp. retained catch followed the opposite pattern. Nonetheless, most of the retained catch values were recorded at seabed depths near 30–40 m (Figs 1c–3c).

The effect of fishers' behavior followed a similar pattern in both total and two Trachurus spp. retained catch, showing increased preferences for choices 5 and 6 (Table 1). Fishers preferred to conduct their fishing operations in areas near or at a medium distance from port, achieving high or medium market price (Table 1, Figs 1d–3d).

DISCUSSION

The large variation shown in all three models by month and fishers' behavior reflected the observed differences in retained catches of a typical Mediterranean purse seine fishery, and confirmed the importance of considering these factors when attempting to conduct effort standardizations. The water depth explained only a small part of the observed variation of the retained catches.

The observed monthly reduction in both total and horse mackerel retained catches during the study period can partly be attributed to the lower availability of fish resources as the fishing season progressed. Because of a closed season prohibition (December–February), fishing is more intense during the first months following the opening of the area. Furthermore, bad weather is known to reduce the possibility of a secure fishing activity.18 In the study area strong winds and currents prevailing after September impeded the setting of the nets and thus increased fish species' escapement during fishing activity. The Mediterranean horse mackerel retained catches progressively increased during the examined period. The highest values were observed during late summer–early autumn, coinciding with the peak species' reproductive season. These observations are in accordance with other results19 in the same study area.

The present findings indicated that depth was an important factor affecting both total and the two Trachurus spp. retained catches. According to our observations, most catches were recorded at seabed depths near 30–40 m. As the Mediterranean purse seine fishing strategy is heavily dependent on visual stimuli in the water column (the light from the lamps decreases with increasing water depth), the species that are concentrated in the upper water column near the surface lamps will be mainly caught.

Observations made by the fisheries scientists onboard the vessels confirmed that in the Mediterranean purse seine fishery, the retained catches were modulated by both market price and distance from port. While in a specific fishing ground, fishers were found to retain and thus land the best possible combination of landings × market value. Furthermore, fishers adapted their fishing operations so as to return to port to sell their catches at the best possible price (‘profit maximizers’). The present results provided direct evidence to support the role of the distance from port in the development of this fishing strategy. Fishers' decision regarding the quantity and quality of retained catches/landings was affected by the distance of the specific fishing ground from the landing place. For example, when fishing in distant grounds, fishers preferred to retain larger but fewer fish so as to return to port and land them earlier, attaining a more competitive price because of higher fish quality.

The market prices depended on several factors such as species quality, freshness and the hour of arrival at the port (i.e. whether other fishers reached the port earlier and had already sold their catches). The current analysis revealed that fishers preferred to conduct their fishing operations near the port (choice 6). By doing so, they were reducing operating costs whilst obtaining higher profits, as in this way, they could return and land their catches earlier. Alternative choices in fishers' behavior (choices 3, 4 and 5) were also recorded. These were partly caused by competitive interactions that are known to exist among fishers.20 On several occasions fishers elected to exploit distant fishing areas. Fishing operations conducted in these areas aimed to provide higher market prices (e.g. larger fish) as a compensation for increased operating costs.

Fishers' interviews of the entire fleet conducted prior to the statistical analysis were critical as the data enabled us to specify a behavioral model able to capture the key aspects of individual heterogeneity with sufficient simplicity to facilitate statistical estimation with a large data set. It was found that constraints imposed by the physical, environmental and economic environment compel the purse seine fishers to minimise risks rather than to maximise retained catches by selecting specific fishing areas and achieving a lower but more secure market price (choices 3b, 4b,c and 5b than choices 3a, 4a and 5a, respectively). On several occasions fishers counted on knowledge from earlier experience (even the catch rate values from the previous day) and avoided risky or uncertain decisions. As evidenced, this specific strategy was followed particularly during the last months of the fishing period when fishing operations were strongly influenced by weather conditions. The current findings are in agreement with those reported in other areas of the world,2,13,21 and are the first, to our knowledge, concerning the Mediterranean Sea.

The present results also provided evidence that the observed major switches in fishing strategy (e.g. the exploration and fishing activity at distant areas, choices 3a, 4a and 5a) were due to fishers' risk attitude response towards higher profitability. It was observed that depending on seasonal price fluctuations, fishers relied heavily on their ability to locate particular species (i.e. two Trachurus spp.), adjusting fishing strategy accordingly. These findings are consistent with those reported in other coastal fishing communities1,3,6,22–24 (Tsitsika EV, unpubl. data, 2004).

Implications for management

In this work, we developed a model of multi-species harvesting for a typical purse seine fishery in the Mediterranean by explicitly integrating the dynamic nature of the fisher behavior and the temporal and bathymetric interactions of the fishing fleet. Unlike other fisher behavior studies, the model is implemented using a large amount of data collected on a monthly basis by onboard scientific observers during a two-year period. Thus, the present study represents the first analysis of a Mediterranean purse seine fishery that suggests how such detailed fishery production data could also be of use to fishery evaluation plans and decision-making. The analytical results demonstrate the urgent need for a multidisciplinary approach to describe fishers' behavior, if realistic predictions on future developments in Mediterranean management schemes are to be made. Regulations that restrict entry or reduce the profitability of some areas relative to others result in a redirection of fishing effort. A better understanding of fishers' behavior will help policy makers better predict and accommodate these changes. As conditions and regulations in the Mediterranean purse seine fishery alter, the ability to predict fishers' behavior might prove useful in designing management strategies to accommodate changes in effort distribution. A further practical application of the outcomes of this study would be to assess and advance fishing activity by region on the basis of expected prices.

For regulators, permanent or temporary closures of fishing grounds are currently among the most widely implemented regulations. This is also the case in Mediterranean fisheries. Area closures serve as a useful management measure in some regions since fishing mortality may be reduced, stocks could be protected from collapsing and fisheries catches in contiguous areas would increase. The fishery effort in the study region is rapidly approaching levels that would make fisheries for the most valuable species unsustainable in the long term. Since fishers respond to seasonal, regional and economic incentives, the simple analytical approach of the current study might be useful in technical measures and management considerations in analogous cases. Fishers targeting the same species typically have dramatically different net returns depending on their location choice. The critical dimension of both fishers' behavior and area closures, mentioned above, is space. Therefore, it is likely that the current results could have implications in other areas where similar technical measures and fisheries coexist. For example, they may be able to pre-empt potential overfishing in areas to which effort is displaced and better predict the results of the regulations on the area at which they are targeted.

CONCLUSION

This study is the first attempt to include aspects of fishers' behavior in a typical Mediterranean purse seine fishery analysis. As the current statistical system suffers from unrealistic catch estimates, obtaining retained catches through scientific observers improved the quality of the data, and thus, the reliability of the analysis performed. Furthermore, onboard observations revealed the necessity of integrating additional quantitative information (e.g. interactions among species, weather conditions and competition among vessels) to stock abundance estimations for improved fisheries management scenarios. In helping to provide a sound basis for management advice of the purse seine fishery, future studies should focus on the amalgamation of fishers' and scientists' knowledge.

ACKNOWLEDGMENTS

We thank the fishers of the purse seines and their crew who accepted the scientists onboard and offered their help during the study, and to A. Kapantagakis, Hellenic Center of Marine Research, for providing purse seine fleet statistics.

Ancillary