Limnol. Oceanogr., 44(2), 1999, 459–465

Nonconsumptive mortality of mesozooplankton in Lake Constance was directly estimated by collecting dead zooplankton with sediment traps. Patterns of zooplankton sedimentation observed in the sediment trap reflected the population dynamics in the pelagic zone. The migration of Cyclops vicinus toward the sediment (where they spend their diapause) at the end of May resulted in a pronounced occurrence in the traps (migration rate .106 individuals (ind.) m22 d21). Other copepods and Daphnia hyalina had only sedimentation losses of 0.5 and 0.2% of the standing stock per day, respectively, demonstrating the minor role of nonconsumptive mortality for these species. In contrast, nonconsumptive mortality had a high significance on the population dynamics of nonmigrating epilimnetic Daphnia galeata. High sedimentation rates (up to 3 3 103 ind. m22 d21; in total, 38% of the total population decline from 16 3 104 to 1 3 104 ind. m22 within 21 d) in June 1993 were attributed to an unidentified infection. From April to November, losses of D. galeata as a result of nonconsumptive mortality (average of 2.3% of standing stock per day) accounted on average for 23% of the estimated production (10% of the standing stock per day). Zooplankton are important members of aquatic food webs. Numerous articles deal with standing stock, production, grazing impact on lower trophic levels, and nutrient regeneration by zooplankton. The control of zooplankton by invertebrate predators and fish has also been the subject of many studies (Gliwicz and Pijanowska 1989). While most studies have focused on the production of different groups of zooplankton and their interactions within the food web, comparatively little is known about the impact of nonconsumptive mortality on zooplankton population dynamics (Andersen 1997). Few articles address zooplankton mortality per se (Clarke and Carter 1974; Prepas and Rigler 1978; Ghilarov 1985; Gabriel et al. 1987; Brett et al. 1992). There is general agreement that consumptive mortality is the most important factor determining zooplankton population dynamics (Gliwicz and Pijanowska 1989). However, there are several potential reasons for nonconsumptive mortality: e.g., senescence, starvation (due to shortage of food quantity and quality), unsuccessful predator attack, and illness. The importance of these factors relative to mortality rates has yet not been examined in detail in nature (Green 1974; Schwartz and Cameron 1993). In culture experiments, mortality rates can be measured relatively easily (Brett et al. 1992). Under natural conditions, mortality is usually estimated as the difference between a calculated estimate of birth rate (b) and estimated population growth rate (r) (Clarke and Carter 1974; Prepas and Rigler 1978; Rigler and Downing 1984; Ghilarov 1985). Like all field-estimated parameters, both r and b are subject to methodological errors (see Prepas and Rigler 1978). In particular, heterogeneous population distributions (horizontal and vertical) (Patalas 1969; Patalas and Salki 1993) and egg mortality (Green 1974; Threlkeld 1979; Boersma and Vijverberg 1995) may cause overor underestimates of r and b. Estimates of death rate (d), calculated as d 5 b 2 r, are confounded by errors in r and b and are therefore less precise than either (Taylor and Slatkin 1981). Consequently, it is difficult to obtain reliable estimates of mortality under natural conditions (Prepas and Rigler 1978; Brett et al. 1992). In addition to problems associated with the estimation of d, these estimates do not address the mechanistic causes of mortality. In this paper we describe a direct in situ estimate of nonconsumptive mortality, one that is based on sediment trap collection of dead or morbid zooplankton. There are few articles dealing with the sedimentation of zooplankton. Rigler and MacCallum (1974), for example, found a nearly perfect fit, over the course of 2 yr, between the calculated production rates and production estimates based on cast exuviae that were collected from sediment traps. Frequently, however, sediment trap data are thought to overestimate sedimentation rates because of the use of poisons (Coale 1990; Michaels et al. 1990; Bathmann et al. 1991). A swimming zooplankter that randomly enters such a trap can be poisoned, immobilized, and added to the settled zooplankton. As a consequence, trap-collected mesozooplankton are often removed prior to analysis to avoid an overestimation of the sedimentation rate (Coale 1990; Michaels et al. 1990; Lee et al. 1992; von Bodungen et al. 1995). Therefore, we used unpoisoned traps and kept the exposure time short (3 to 4 d) to minimize the degradation of the settled material (Bloesch and Burns 1980). Our study area was Lake Constance, a large, deep, warmmonomictic, mesotrophic, prealpine lake (area, 500 km2; Zmax 5 253 m; Zmean 5 95 m; Braun and Schärpf 1994). The sampling station was located at the center of Überlinger See, a fjordlike northwestern sidearm with a maximum depth of 147 m. Zooplankton were collected twice per week from the end of March to November 1993 using a net and a sediment trap. Swimming zooplankton were collected within the upper 50 m with a Clarke–Bumpus sampler (30-cm diameter, 140-mm mesh). The samples were stored in lake water at 48C and were transported within 4 h to the laboratory, where they were filtered again (140 mm), transferred into preweighed scintillation vials, and frozen at 2258C (Shapiro and Wright 1989). After lyophilization (Berberovic and Pinto-Coelho 1989), an aliquot of each sample (200–1,000 ind.) was counted under a dissecting microscope (316 magnification). To avoid a counting bias, samples were counted in arbitrary order, beginning with pelagic samples and ending with trap samples (see below). Three different taxa were distinguished: copepods, of which only Cyclops vicinus was indentified to the species level, Daphnia galeata, and Daphnia