Despite high primary productivity. the yield of pelagic fish in the southern Benguela is relatively low compared to that in the Humboldt system. Part of the constraint may be the ability of pelagic fish to reproduce successfully in a strongly pulsed upwelling environment, where enrichment, retention and concentration mechanisms are less compatible than in Peru-Chile. Anchovy Engraulis capensis spawn upstream of the main upwelling centres on the food-poor, thermally stratified western Agulhas Bank, over a protracted summer season (October-February) when high wind speeds of 7 -8 m·s-I are prevalent. Eggs spawned farther east, on the central or eastern Agulhas Bank, may be subject to increased cannibalism and advective losses, whereas those spawned farther west could be susceptible to heavy advective losses offshore during periods of strong southerly winds. Copepod concentrations are negatively correlated with spawner biomass on the western Bank and are inversely linked to high rates of gonad atresia in anchovy and reduced subsequent recruitment. There is a restricted area of suitable spawning temperatures (l6-19°C) on the western Bank, and fish outside that range appear to have less successful spawning, affecting recruitment; a reduction of the duration of spawning appears to decrease the chances of good recruitment. Oil content of pelagic catches on the west coast of South Africa suggests that good feeding conditions for adult fish are also favourable for growth of young fish in the nursery area. An inverse correlation between the numbers of recruits estimated in June and the strength of upwelling-favourable winds over the previous summer indicates the importance of advective loss of eggs and larvae in the transport and nursery areas. The presence of Agulhas Rings close to the shelfedge, predators on juvenile anchovy, and the strength of upwelling on the West Coast could also be important factors affecting recruitment success. There seem to have been changes in the inshore-offshore distributions of spawning anchovy and/or their eggs over the past three decades, resulting in different recruitment responses to south-easterly winds.

Environmental and biological sampling and monitoring have been carried out in the southern Benguela since 1988. The overall goal of this research is to investigate environmental factors affecting anchovy recruitment and to develop the ability to forecast anchovy recruitment from year‐to‐year using field data obtained during the spawning season (August to March). Sampling was conducted at three different temporal and spatial scales: during annual (November) broad‐scale hydro‐acoustic surveys to determine spawner biomass on the entire spawning ground and in the core transport and recruitment areas; during monthly surveys in the core spawning, transport and recruitment regions over two entire spawning seasons (1993/94 and 1994/95); and during weekly sampling (since 1995) along a single transect downstream from the spawning area. Annual surveys provide the best spatial coverage, but are inadequate for representing environmental conditions and anchovy spawning success over a prolonged season. Weekly sampling provides the best temporal coverage, but logistical constraints restrict information to a limited portion of the spawning area and a reduced number of variables. Monthly surveys provide intermediate coverage in time and space, but are expensive and labour‐intensive. Forecasting anchovy recruitment has been based on two different approaches: the establishment of empirical relationships, and the development of rule‐based expert systems. Forecasts from deterministic expert systems have compared well with final estimates of recruitment strength, and indicate that environmental and biological variables may be used in a structured way to forecast anchovy recruitment.

Understanding long-term variability of pelagic fish populations is important in developing forecasting strategies for fisheries management and planning. However, many current fisheries models have only shortterm datasets available, whereas those of suitable duration often lack reliability. As resources are placed under increasing pressure, all available information should be used to assist management. Two simple rulebased deterministic modelling approaches are described, which use semi-quantitative and qualitative rules to relate recruitment success of South African anchovy Engraulis capensis to physical and biological indices. The first model relates recruitment success to indices of wind and sea surface temperature by way of a rulebased decision support system. In the second model, significant environmental and biological factors were identified and related to anchovy recruitment by way of an expert system approach. These two approaches are evaluated and compared. It is suggested that these types of models, when satisfactorily validated, have great potential in supporting the future management of the South African anchovy fishery in the dynamic environment of the Benguela Current.

No abstract

Clupeoids form the basis of important commercial fisheries off Namibia and South Africa's Western Cape. Studies on the abundance of pilchard (sardine) and anchovy populations subject to heavy exploitation have revealed the possible succession of sardine by anchovy as the dominant species. In investigating the evidence for species replacement, a simple, age-structured computei simulation model was developed, which assumes that the system is initially in a steady state. The model's sensitivity to starting biomass, natural mortality, harvesting, the level at which density-dependence is initiated and environmental perturbation is consideJ:Cd. Important factors determining the results are the present estimates of natural mortality and the level at which density-dependence is initiated. A lower natural mortality in pilchard results in dominance by anchovy. High threshold values of densitydependence produce dynamic oscillation in the model populations. Harvesting affects the natural balance between populations in a system at steady-state, causing the dominant stock to decline, resulting in species replacement. This situation is particularly evident in scenarios employing harvesting numbers-at-age. The intensity and the duration of harvesting determines whether replacement is temporary or permanent. The model populations show resistance toward environmental perturbation; indeed, in scenarios including it, species replacement is delayed. Klupeo'ides II!tielangrike handelsvisserye te~no?r Nami~ie en Suid-A,frika "': Wes-Kaap len gron?slag. Sru~ies oor die talrykheid van sardyn (pelser)-en ansJovlsbevolkings wat hewlg oRlgm word, het moontlike opvolgmg van sardyn deur ansjovis as die oorheersende spesie aan die lig gebring. Met die ondersoek van die bewyse vir spesievervanging is 'n eenvoudige, ouderdomsgegronde rekenaarsimulasiemodel ontwikkel waarin die aanname gemaak word dat die stelsel aanvanklik in 'n ewewigstoestand verkeer. Die model se gevoeligheid vir aanvanklike biomassa, natuurlike sterftesyfer, ontginning, die vlak waarby digtheidsafuanklikheid inlree en omgewingsversteuring word in ag geneem. Belangrike faktore wat die resultate bepaal, is huidige ramings van natuurlike sterftesyfer en die vlak waarby digtheidsafuanklikheid begin word. 'n Laer natuurlike sterftesyfer by sardyn het oorheersing deur ansjovis tot gevolg. Hoe drempelwaardes vir digtheidsafbanklikheid lei tot dinamiese skommelings in die modelbevolkings. Ontginning be'invloed die natuurlike ewewig tussen bevolkings in 'n stelsel wat in ewewig verkeer en laat so die oorheersende stapel agteruitgaan tot spesieverplasing plaasvind. Hierdie situasie is veral opvallend in scenario's wat van ontginning volgens getal per ouderdomsklas gebruik maak. Die hewigheid en duur van ontginning bepaal of verplasing tydelik of blywend is. Die modelbevolkings toon weerstand teen omgewingsversteuring; inderdaad, in scenario's wat dit insluit, word spesieverplasing vertraag.