Yellowfin tuna catch opportunities in Cape Verde coping with uncertainties of local CPUEs Foto Péricles Silva Ivanice Monteiro, Heino Fock and Péricles silva Lanzarote, April,19th 2018
Introdution Yellowfin tuna (Thunnus albacares, YFT)- tropical and subtropical epipelagic species at all three oceans. Sizes exploited range 30cm - to over 170 cm; maturity at about 100 cm. In the Atlantic, YFT migrate seasonally avoiding areas of lower SST high YFT catches obtained in the ETA in quarters 2,3, and 4, increasing around Cabo Verde Local catch CPUE is treated as a function of stock size, N s, and environmental factors, V i. For tuna stocks, no fisheries independent information is available- limiting unbiased abundance and distribution indices calculations Catch data from artisanal and industrial fisheries from 10-20 N lat. and 10-30 W long. Were used to indicate catch opportunities in Cabo Verde waters. In this presentation, we emphasize on the assumptions underlying the calculation of local CPUE and model weighting to take account of for instance aggregation behaviour both of fishermen and tuna.
Methodology Artisanal catches are very important on the total Cabo Verde catches (high representativeness before 80 s), (Belhabib et al, 2016) Domestic catches around 1.9 times the landings supplied to FAO (Belhabib et al, 2016) YFT catches representing ~ 35% of total artisanal catches in CV (INDP Statistics, 2018) Average Catches 1950-2010 2 % 5 % 40 % 53 % Artisanal Industrial Subsistence Recriational 50's 3 % 60's 9; 9 % 2; 2 % 70's 4; 4 % 2; 2 % 7 % 15 % 75 % 18; 18 % 71; 71 % 31; 31 % 63; 63 % Artisanal catches composition in Cabo Verde
Artisanal fishery data analysis in Cabo Verde has not been easy because of different reasons Artisanal fishery characteristics: small boats (3.5-6.5m) with ~4 fishermen each different gears in a single fishery (hand line, 80%) no target specie catches disturbed in small pelagic, large pelagic (tuna and similar), demersal and others Effort not well defined (days at sea- seasons at the fishing site) Hand line Purse seine Gill net beach net diving Foto Carlos Monteiro Foto Péricles Silva Foto Péricles Silva
Nevertheless, much work has been done to include artisanal fisheries information in fisheries analyzes. Effort on artisanal fishery are not clear, days at sea were used but is known that the period at sea changes depending of different aspects ( catch, weather, behavior). Using the available artisanal data, raw CPUE were calculated and the trend from artisanal and commercial data from ICCAT were compared. (decreasing for artisanal and increasing for LL. Stock Index. The index.4 doesn't show substantial changes).
A negative correlation (-0.431) was found between Art. CPUE and LLindextotStock and not significant (0,068) between Art.CPUE and Index.4, showing the difficulty to use one of the data set in representation on the other. Very little agreement were find between Art.cpue, cpue.nr and cpue.kg and even negative- all from the same area and fisheries target YFT cpue.nr cpue.kg LLindextotStock Index.4 Art.cpue cpue.nr 1 0.128767 0.3286415 0.247562 0.168429 cpue.kg 0.128767 1 0.383138-0.26574-0.38288 LLindextotStock 0.328642 0.383138 1 0.931896-0.43077 Index.4 0.247562-0.26574 0.931896 1 0.068484 Art.cpue 0.168429-0.38288-0.43077361 0.068484 1
Fisheries comercial data are affected by fleet efficiency shift in target species (big problem at CV) environmental factors fleet dynamics Fishermen behavoir (significant effect on local CPUE) efficiency of search interaction between fishermen, and depletion of the fish stock at local scales.
Comercial Data and variables input Commercial data from foreign fleets and CV Industrial fleet were used to show the YFT catch opportunities at Cabo Verde region. Long-line data from 1964 to 2014, from 10-20 N lat. and 10-30 W long were used from ICCAT "t2ce" database. CPUE by number caught (CPUE.nr) or by catch weight (CPUE.kg) (CPUE l ) is treated as a function of stock size, N s, and environmental factors, V i, (V i at local scale or at stock scale in terms of climate indices) (Eq. a) Log-transformed CPUE values were used (Eq. b) ln(cpue l ) ~ f(n s ) + Σf(V i ) + ε (Eq. a) ln(cpue l ) ~ f(log(n s )) + Σf(V i ) + ε (Eq. b)
YFT stock indices Japanese long line stock index (LLindex)- recommended by ICCAT Index4 series- combined index using a GLM approach Environmental variables Local SST and wind-noaa-esrl Physical Sciences Division TNA, TSA, NTA- (SST) ENSO, NAO, PDO -large scale atmospheric pressure Models multiple linear regression model Linear model-h1 H_AIC model generalized additive model- GAM Bayesian Model Averaging - BMA
Results Of the two stock indices used, the LLindex had higher correlations with both local CPUE indices The LLIndex shows a considerable decline in stock size since 1967 (with small peaks), and an upward trend for recent years after a steady low fase. negative correlation (-0.27) between raw CPUE.kg and Index4 CPUE models were evaluated in 2*2*2 categories, (CPUE by weight or by number, standardised vs. raw CPUE estimates, and type II vs. type III models). Standardised CPUE.kg negative coefficients were obtained for LL stock index in both type II and III models CPUE Nr. CPUE Kg Model type Raw Stand. Raw Stand. II 0.2 0.25 0.31 0.57 III 0.52 0.38 0.59 0.45
Results CPUE Kg Data from 1995, SST was included in the model evaluation process. CPUE.kg resembles the trend of Llindex; reached in the H_AIC model, but not in the H1 and GAM models. CPUE.kg trend is not different between H AIC and H 1 models
Results CPUE Nr. Time series was 50 years, shorter SST time series was excluded from this analysis. CPUE.nr reached its highest value in 1979, consistent with the smaller peak in the LLindex; H AIC and H 1 overestimate CPUE.nr before 1970, and underestimate the 1979 peak and the increase 2001-2004. Overall goodness-of-fit was smaller than for CPUE.kg models
Summary Taking into account the behaviour of the fleet and fishes, the raw CPUE Type III models appear apropriate and that effort is a significant parameter in longline CPUE. Both YFT CPUE by weight (Maury et al., 2001; Doray et al., 2009)and by numbers caught (Lan et al., 2011) have been applied. Congruency between CPUE.kg Type II and Type III results and the easy interpretation of the resp. beta coefficients indicates that CPUE.kg models enable a better understanding of CPUE dynamics. In particular, CPUE.kg is able to reflect the increase in YFT stock after 2000, which was also seen in Taiwanese long-line data (Lan et al., 2011).