Mwelu mukata mukandanshe - the wide waters that the locust cannot cross is the full name of Lake Mweru, the lake situated on the border of Northern-Zambia and the Democratic Republic of Congo in the Luapula valley. Its fish provides the basis for food, employment and income for the estimated 400 000 people that live there. Lake Mweru and the Luapula River with its floodplains, swamps and lagoons have a long history of fishing in connection with cassava farming as main economic and subsistence activities (Aarnink, 1997). Since the beginning of last century, fishing has been, and still is, closely linked by trade to the towns of the Copperbelt and the diamond mines in Zambia and Congo (Musambachime, 1981; Gordon, 2000). Around 1944, after the decimation of the crocodile population instigated by the Belgian colonial authorities, the river and lake area became fully accessible for fishing activities (Musambachime, 1987). It facilitated the development of a fishing pattern, mainly conducted by European fishermen that caused the decline and virtual destruction of a once important fishery on the cyprinid Labeo altivelis. At its height in the 1940s this species contributed 40-60 percent of the commercial catch, but was fished down within four years during its spawning migrations upstream the Luapula river to less than three percent of the total catch, never to recover again (Kimpe, 1964; Gordon, 2003). With the increase in population in the valley, roughly following the demographic rate of increase in both Zambia and the Congo, fishing pressure increased as well. Between 1965 and 1970 total catches and catch rates of the cichlid Oreochromis mweruensis, declined severely, which led to concern about the sustainability of the fishery on this now most important stock (Aarnink, 1999a; Aarnink, 1999b). Two decades later, despite an enormous increase in fishing effort by any measure, it returned as one of the most important stocks in the fishery and allowed two new fish freezing factories to thrive in the 1990s. Oreochromis and other cichlids remained the mainstay of the fishery to this day, roughly taking 60-70 percent of the long-term average catch of 8 300 tonnes that the Zambian part of the fishery has produced since 1955 (Figure 1).
Nevertheless, fishing patterns did change over this period. Large traps and cotton gillnets, still used in the 1950s, were replaced by multifilament nylon gillnets. In 30 years the dominant mesh size decreased from 102 mm to 63 mm stretched mesh and active methods in combination with gillnets became prevalent. Diversification of methods take place continually, the latest addition being the development of Fish Aggregating Devices (Kapasa, 1998) targeting the hitherto underexploited pelagic stocks of Alestes macrophthalmus. The largest new development in fishing patterns, however, took place in the early 1970s, when the fishery on a small pelagic freshwater herring Microthrissa moeruensis started its rapid development to become, in terms of production, the most important fishery of Zambia. The light fishery on chisense is now estimated to produce 25 000-45 000 tonnes per year (Zwieten et al., 1996), far exceeding the production of all previous and present fisheries of Mweru-Luapula.
FIGURE 1. Zambian catch from Lake Mweru and the total catch of Zambia, in both cases excluding the catch of the pelagic fishery on Microthrissa moeruensis (chisense). The monitored (demersal) catch of Lake Mweru decreases slightly over 45 years. The long-term mean catch is 8 350 tonnes. The pelagic light fishery is not monitored regularly: estimates are based on daily logbooks of 21 fishermen (Zwieten et al., 1996) except for the 1985 estimate that is taken from Scullion, 1985.
This short account of the developments in the Mweru-Luapula fisheries already reveals the difficulty to put into perspective legitimate concerns about the effects of the fast increasing fishing effort over the past 50 years. The total production of the demersal trap and gillnet fisheries has remained relatively stable over a long period of time, whereas total fish production of the lake has increased tremendously since the onset of the pelagic light fishery (Figure 1). One species effectively has disappeared from the fishery, other stocks have declined or exhibited tremendous fluctuations even with increased effort, and despite being an intensively fished lake for at least a century (Musambachime, 1981; Gordon, 2000) some stocks are known to have been subject to very low exploitation rates until very recently. Though, the fish community and with that the ecology of the lake has changed, as will be shown later, many stocks prove to be surprisingly resilient to todays high exploitation levels. Mweru-Luapula is a productive system but with highly fluctuating stock levels of important species. Directional changes have taken place in the contribution and size of species in the catch as well that could reflect changes in the fish community. Fishermen have reacted to changes in stocks by adapting their fishing pattern. How can such adaptive changes in effort be characterized and what can be said about the effects of increased effort in a system like Mweru-Luapula? These are complex issues that cannot be treated here exhaustively. In the following overview we will intend to describe the complexity and how changes can be interpreted and possibly quantified.
Lake Mweru is an allotrophic riverine lake (Kolding, 1994) meaning that production is to a significant extent dependent on nutrient pulses brought in with the floods. In such a system susceptibility to increased fishing effort is thought to be low, recovery potential rapid and yield potential high but variable. Our observations on Lake Mweru suggest a differing impact of increased and changing fishing effort on the recovery potential of different species after periods of low water levels. After a description of the lake and surroundings we will describe the developments in fishing effort based on statistical frame surveys of the fishery and then characterize these changes. The picture that emerges is that access to and utilization of the fish resources is highly dynamic involving many people that enter and exit the fishery within short time spans, next to a large resident fishing and farming community. Next we will discuss the selectivity of the various fishing methods employed in the fishery. Though the fishery is based on three gear categories - gillnets, traps and liftnets - many different fishing patterns are employed with these gears that target the fish community in different ways. These patterns combined possibly create an overall, more or less unselective, fishing pattern. In the last paragraph we will describe the changes in stocks from a system perspective with the aid of biomass-size distributions of the demersal stocks constructed from experimental fishing surveys with gillnets, and relate this to dynamic pulses in productivity as a result of changes in water levels. In the discussion we will address the central question of all biological case studies of this research: what is the effect of increased fishing effort on the stocks in these pulsed systems? How clear are such effects, or in other words, how are the effects of fishing effort obscured by the variability in stocks due to changes in productivity? Ultimately this will have to lead to answers on what monitoring data need to be collected and whether present data collections through fishery dependent and independent surveys provide the information needed to recognize in time and conclusively the magnitude of changes taking place through changes in effort.
FIGURE 2. Map of the Mweru-Luapula fishery indicating the main trading center (Kashikishi), the district administrative center (Nchelenge) and islands, lagoons and rivers. The four areas, strata I to IV, are statistical areas of the Catch and Effort Data Recording System (CEDRS) of Zambia, roughly coinciding with the main ecological areas of the fishery. The northern border of stratum I is formed by the Lunchinda River. The border between Zambia and the Democratic Republic of Congo follows the main course of the Luapula River. The broken line gives and approximate indication of the border. From the Mambilima falls to the Luvua River is approximately 230 km as the crow flies.
 See Appendix for local and
 Formerly known as a subspecies of Oreochromis macrochir but elevated to the status of species (Schwanck, 1994).
 A fishing pattern is a fishing method including its spatio-temporal allocation, scale of operation and labour input. This to a large extent determines the selection of species (see Jul-Larsen et al., 2003).