Mh Griffiths

An ecosystem approach to fisheries in the southern Benguela context

The 2001 Reykjavík Declaration on Responsible Fisheries in the Marine Ecosystem and the Plan of Implementation of the 2002 World Summit on Sustainable Development highlighted the need in fisheries to look beyond considering only the target species and to consider in fisheries management the impacts of fishing on the ecosystem as a whole, as well as the impacts of the ecosystem on fisheries. This paper examines the practical implications of progressing towards ecosystem approaches by reference, in particular, to the FAO technical guidelines on the topic. It goes on to examine the major fishery types in South Africa and the southern Benguela, and to consider the probable impacts of those fisheries on target species, bycatch species and the ecosystem, as well as the indirect impacts on other affected species. The review reveals that all fisheries have impacts beyond the target species and that an ecosystem approach is required in order to ensure the long-term sustainability of the living marine resources of the southern Benguela and the ecosystem as a whole. Finally, the likely obstacles to successful implementation of an ecosystem approach to fisheries in the southern Benguela are discussed.

DISTRIBUTION PATTERNS OF KEY FISH SPECIES OF THE SOUTHERN BENGUELA ECOSYSTEM: AN APPROACH COMBINING FISHERY-DEPENDENT AND FISHERY-INDEPENDENT DATA

Within the context of an ecosystem approach for fisheries, there is a need for quantitative information on distributions of key marine species. This information is valuable input for modelling species interactions in the southern Benguela ecosystem. In the present study, a method is described for mapping the density distribution of 15 key species: anchovy Engraulis encrasicolus, sardine Sardinops sagax, round herring Etrumeus whiteheadi, chub mackerel Scomber japonicus, horse mackerel Trachurus trachurus capensis, lanternfish Lampanyctodes hectoris, lightfish Maurolicus muelleri, albacore Thunnus alalunga, bigeye tuna Thunnus obesus, yellowfin tuna Thunnus albacares, silver kob Argyrosomus inodorus, snoek Thyrsites atun, Cape hake Merluccius spp., kingklip Genypterus capensis and chokka squid Loligo vulgaris reynaudi. The purpose was to make use of all available sources of data to extend the spatial and temporal coverage of the southern Benguela. Six sources of data were combined on a 10′ × 10′ cell grid in a Geographical Information System: acoustic and demersal surveys conducted by Marine and Coastal Management (MCM), and pelagic, demersal (including midwater trawl), hake-directed and tuna-directed longline commercial landings data collected by MCM. Comparisons of distributions between two periods (1980s and 1990s) and between two semesters (April – September and October – March) were conducted, but biases as a result of major differences in sampling strategy prevented detailed analysis for certain species. Maps of density distributions are nevertheless presented here and the method to determine them is discussed.

Effects of fishing on the size and dominance structure of linefish of the Cape region, South Africa

A dataset of linefish catch, effort and fish size distribution records has been assembled from archives to cover three short periods over the 100 years from 1897 to 1998 in four regions of the former Cape Colony, South Africa. Linefish catch and effort have increased several-fold over the period. Aggregate catch per unit effort (cpue) declined by more than 80% from values in the 1890s, but the cpue of several species within that aggregate have declined much more. Analysis of historical mean size and modern length frequency data shows that in seven of 12 species considered, the mean length of fish declined substantially along with the increased fishing pressure. Multivariate analysis of cpue shows that the years 1897–1906 cluster quite close to the years 1927–1931, but a major change by the years 1986–1998, revealing a large change in abundances of linefish between the 1930s and the 1990s, which is also the period when fishing effort increased most. A related dataset was used to calculate the combined distribution of fish sizes of the 12 species in logarithmic size-classes in the same years. The negative slope of that size spectrum indicates the decline in numbers of large size-classes compared with small ones; the more negative the slope, the greater the relative decline in numbers of large fish. Slopes become significantly more negative in the modern period, showing that the modern linefish catch has fewer large fish and relatively more small ones than previously. Changes in linefish assemblages, implied by changes in catch composition, are different in the four regions studied. The cool-temperate upwelling regions differ from the warm-temperate ones, particularly with regard to the influence of the fast-growing, nomadic, pelagic snoek Thyrsites atun. Inclusion of snoek gives the size spectrum of the cool-temperate regions a shallower slope than the warm-temperate ones. A new method of plotting the size spectrum is believed to free the intercept (height) from dependence on the slope and simplifies interpretation of the relative values of height, which reflect overall fish abundance. Dominance curves reflect the distribution of biomass among species. The cool waters of the Western Cape show a trend towards increasing dominance with increased effort, whereas the warm-temperate regions show decreased dominance with increased fishing pressure. These findings have important consequences for fisheries management, because not only are several stocks badly overfished, but the linefish considered are predators at different trophic levels that influence the tropho-dynamic functioning of whole ecosystems.

Identity and distribution of southern African sciaenid fish species of the genus Umbrina

Two Umbrina species, U. canariensis Valenciennes 1843 and U. robinsoni Gilchrist and Thompson 1908, are recognised from southern Africa. The latter species was hitherto believed to be a synonym of Umbrina ronchus Valenciennes 1843 (type locality Canary Islands). U. canariensis is distributed along the South Africa eastern seaboard from Cape Point to Sodwana Bay and U. robinsoni is known from False Bay to Madagascar and Oman. African Umbrina taxonomy has, however, been hindered by geographic samples that were either too few or consisted of specimens of disparate length; and as a result the identification and distribution of South African Umbrina species was confused. Morphological comparison of a large number of South African Umbrina with specimens from the type locality (Canary Islands) confirmed the identity of South African U. canariensis and allowed for an expanded description of the species. However, differences between specimens of U. ronchus and those of the second South African species (n = 251) led us to resurrect U. robinsoni (Gilchrist and Thompson 1908) as a valid name for this species. U. robinsoni differs from U. ronchus in having a smaller supraoccipital crest and thus a less steep pre-dorsal profile; a shallower preorbital bone (13–21% head length [HL] vs 21% HL); and a shorter nostril-orbit distance (2.4–6.9% HL vs 7.8–8.5% HL). Colour patterns also differ between the two species, with U. ronchus lacking the oblique, wavy, white stripes evident on the flanks of U. robinsoni. U. ronchus does not occur in South African waters, and is an eastern Atlantic species occurring from Gibraltar to Angola. Specimens from the east coast of Africa (Moçambique to Gulf of Oman) that were previously identified as U. ronchus are U. robinsoni. Differences between U. robinsoni and U. canariensis include: a lower modal number of soft dorsal fin rays, (22–27 vs 24–30); less deep body depth, (26–36% standard length [SL] vs 33–39% SL); shorter pectoral fin length (15–21% SL vs 20–25% SL); longer caudal peduncle length (26–34% SL vs 21–28% SL) and snout length (27–38% HL vs 23–32% HL); and smaller orbit diameter (14–33% HL vs 23–34% HL). Otoliths of U. robinsoni differ from those of U. canariensis in being smaller, less elongate, lacking a massive post-central umbo and having a post-dorsal spine remnant. The body colour and nature of the striping pattern on the flanks differs markedly between the species: in U. robinsoni the oblique stripes are thin, wavy, white lines; in U. canariensis the oblique stripes are thicker, nearly straight and brown; U. robinsoni also lacks the triangle-shaped mark on the outer operculum and the dark pigmentation of the inner operculum that is found on U. canariensis. Spatial analysis of South African specimens collected with a variety of gear revealed U. robinsoni to be a shallow-water species found from the surf-zone to 40m, whereas U. canariensis occurs predominantly from 40 to 100m depth. Although both species occur throughout the South African eastern seaboard, U. canariensis is most common west of the Kei River, where the shelf is wider. Examination of three specimens of U. steindachneri Cadenat 1950 confirmed the presence of a fourth sub-Saharan Umbrina species that is limited to tropical West African waters from Senegal to Angola. U. steindachneri differs from the other African Umbrina in having a high number of soft dorsal rays (28–29), a greater 3rd dorsal spine length (25–27% SL) and a very pronounced and convoluted striping pattern on the flanks.

Stock separation and life history of Argyrozona argyrozona (Pisces: Sparidae) on the South African east coast

Fishery independent biomass surveys and commercial linefish catch returns were used to elucidate the spatial patterns of carpenter Argyrozona argyrozona distributed along the South African continental shelf. Two distinct areas of abundan ce ere determined, one on the central and the other on the eastern Agulhas Bank. Tagging studies revealed little exchange between them. Nurseries were identified in Algoa Bay on he eastern Agulhas Bank and on the central Agulhas Bank (CAB). Early juveniles (<100mm total length) on the CAB were found offshore in the vicinity of the Alphard Bank. They were found to move inshore with growth and then back offshore as they approached maturity. Juveniles in Algoa Bay dispersed both eastwards and westwards with growth. Otoith readability and growth rates varied between regions, with fish from the Eastern Cape (Port Elizabeth and Port Alfred combined) having the lowest average percentage error (4.82 vs 5.33 and 7.03) and the slowest growth rates. Size-at-50 % mturity (L50) varied regionally, female fish in the Eastern Cape maturing at a smaller size (L50 = 206mm fork length) than in the Tsitsikamma National Park (L50 = 292mm) or the CAB (L50 = 267mm). Mass-at-length varied between regions, with carpenter in the Park having the highest mass-at-length and those in the Eastern Cape having the lowest. Based on the distribution of carpenter, variability in otolith readability, mass-at-length, variation in growth and size-at-maturity, it is concluded that carpenter exist as two separate stocks and should be managed accordingly.

Influence of sample size and sampling frequency on the quantitative dietary descriptions of a predatory fish in the Benguela ecosystem

Spatial and temporal variation in the diet and feeding intensity of snoek, Thyrsites atun, a top predator of the southern Benguela, was investigated to provide information on data requirements for accuracy of annual diets as inputs to ecosystem models. Appropriate sample sizes to produce accurate daily or event-scale diet descriptions were investigated by means of a posteori tests. Cumulative prey diversity curves showed that 55 (± 25) stomachs containing food are required to accurately quantify presence and absence data of prey species. Differences in the percentage contribution of dominant prey in consecutive and cumulative 10-stomach classes, indicates that a minimum of 75–80 (± 25) stomachs containing food are necessary to describe the proportion by weight of primary prey. Diets of snoek (proportions by weight) sampled inshore (shallower than 50 m and within 25 km of the coast) off the Cape Peninsula during six consecutive weeks (i.e.one day per week) in autumn 2001 were highly variable. Comparing diet for the six-week period with that of diets from consecutive 10-sample weekly units revealed 95% similarity at 70 samples per week. Seasonal diets between Cape Columbine and Cape Hangklip also varied, but there was no evidence of a predictable seasonal pattern in diet that could be related to prey life history; two-way nested ANOSIM revealed that seasonal prey proportions across years were statistically less similar than those within years. Snoek spawn offshore in winter/spring. Statistical differences between inshore (< 50m) and offshore (>150m) prey composition were largely influenced by the absence of anchovy Engraulis encrasicolus and much larger proportions of Cape hake Merluccius spp. and lanternfish Lampanyctodes hectoris in the offshore diet. Feeding intensity (in terms of proportions of fish with prey and mean stomach fullness) was strongly seasonal and highest during the spawning season. Sampling programmes for the southern Benguela should account for spatial and temporal variation in diet and feeding intensity of predators if accurate annual dietary descriptions are to be achieved. The results of this study indicate that a minimum of 70 snoek stomachs (containing food) should be collected at each sampling event.

Do dart tags suppress growth of dusky kob Argyrosomus japonicus

Growth rates of dusky kob Argyrosomus japonicus calculated from mark-recapture and otolith-reading methods were compared. Mark-recapture data showed that A. japonicus are resident in an area between the Breede River Estuary and Cape Agulhas on the south-east coast of South Africa. Maximum recapture length was 1 150mm. A von Bertalanffy growth curve, fitted to length-at-age data derived from otolith sections of 168 specimens (1–42 years) from the Breede Estuary and adjacent marine environment, was differentiated to provide instantaneous length-based growth rates. Comparison of mark-recapture and otolith-based growth rates revealed that external dart tags suppress growth of A. japonicus <750mm, but do not affect the growth of larger fish.

Life-history strategies of Umbrina robinsoni (Sciaenidae) in warm-temperate and subtropical South African marine reserves

Biological data for Umbrina robinsoni were obtained from fish sampled monthly during 2001–2002 by shore-angling in the warm-temperate De Hoop Marine Protected Area (MPA) (n = 312), and by means of spearfishing during three trips (May, September and January 2001–2002) in the Kosi Bay region of the subtropical Maputaland MPA (n = 354). Annuli in otolith sections were validated by means of marginal increment analyses and fluorochrome marking (oxytetracycline). Maximum ages recorded were 12 and 16 years at Kosi Bay and De Hoop respectively. Kosi Bay fish obtained a significantly greater asymptotic length than De Hoop fish and mean length-at-age (for ages 2–10 years) was significantly greater. The fitted von Bertalanffy growth equations for combined sexes were: Lt = 594 (1—e−0.183 (t + 2.42)) for De Hoop and Lt = 875 (1—e−0.151 (t + 2.49)) for Kosi Bay. Trends in mean monthly gonadosomatic indices and proportions of histologically validated macroscopic gonad stages indicated a summer spawning season (November–February) at De Hoop and year-round spawning at Kosi Bay. Relative condition peaked in both spring and autumn at De Hoop but showed little seasonal variation in the Kosi Bay region. Kosi Bay females attained 50% sexual maturity at a significantly larger size (48 vs 39 cm) but at younger age (2.8 vs 3.4 years) than those at De Hoop. Analysis of maturity schedules indicates that current sizes at maturity are plastic responses that maximise lifetime fecundity within local regimes of somatic growth and natural mortality. The instantaneous rate of natural mortality was substantially higher at Kosi Bay than at De Hoop (M = 0.35 vs M = 0.26).

Management of Argyrozona argyrozona (Pisces: Sparidae) in South Africa based on per-recruit models

Carpenter Argyrozona argyrozona is an endemic sparid that constitutes an important component of the South African linefishery between Cape Agulhas and Port Alfred, where it exists as two stocks; one on the central Agulhas Bank and the other on the eastern Agulhas Bank. Spawner biomass-per-recruit (SB/R), fecundity-per-recruit (Egg/R) and yield-per-recruit (Y/R) models were used to model both South African carpenter stocks. Owing to the allometric relationship between annual fecundity and individual size, Egg/R ratios were between 40% and 74% of SB/R at equivalent fishing mortality (F). Egg/R ratios account for allometric increases in fecundity with size/age, and are therefore regarded as better estimators of reproductive potential. It is shown that the current length at first capture (Lc) of 250mm TL and F (at M = 0.1) will reduce Egg/R to 6.41% of the pristine value in the eastern Agulhas Bank population and to between 6.06% and 14.15% on the central Agulhas Bank, indicating that both stocks are heavily overfished. An increase in Lc from 250mm to 350mm TL and a 70% reduction in commercial fishing effort is recommended to attain a target reference point of 40% Egg/RF=0. Bag frequencies indicate that a reduction in daily bag limit from 10 fish person−1 day−1 to four fish person−1 day−1 would effect an equivalent reduction in recreational fishing mortality. The trawl bycatch of carpenter is only 3% of the reported line catch, consequently restrictions to this fishery are not recommended.

An investigation of the effect of temporal variation in growth rate of Umbrina robinsoni on biological reference point estimates calculated using per-recruit models

The effect of temporal variation in growth rate on per-recruit model outputs was investigated by comparing biological reference points obtained using growth curves derived for Umbrina robinsoni populations from False Bay, on the south coast of South Africa, sampled 10 years apart (1991–1993 and 2001–2003), and applying the two different age–length keys to length frequency data collected over the period 1991–1993 (n = 1 389) for the estimation of total mortality (Z), fishing mortality (F) and spawner biomass per recruit (SB/R). The SB/R and yield per-recruit (Y/R) curves constructed using biological data collected during the two different time periods were very similar, with target and threshold reference points differing only moderately (11%). This suggests that for long-lived species, the frequency of age and growth studies can be at least the lifespan of a cohort if they are to be assessed with a per-recruit approach. The use of the more recent length-at-age data, however, resulted in lower estimates of F (0.61 y−1 vs 1.05 y−1) and a slightly higher spawner biomass per-recruit ratio (19.5% vs 14.5%) for the earlier period. The reduced estimate of F when using the more recent dataset was likely due to a bias towards older fish caused by small sample size, the philopatric nature of U. robinsoni, and possible recruitment collapse. This highlights the importance of having a large, representative sample of length-at-age data for construction of age–length keys.