Behavior and aquatic plants as factors affecting predation by three species of larval predaceous diving beetles (Coleoptera: Dytiscidae)

Predation among aquatic invertebrate predators can have important effects on patterns of exclusion and coexistence in aquatic habitats, especially if these predators also act as intraguild predators. Such patterns may be explained by variation in predator foraging mode and in the extent and overlap of habitat use. Predaceous diving beetles (Coleoptera: Dytiscidae) are abundant in isolated bodies of water and are effective predators on many aquatic organisms, including other dytiscids. The under-investigated role of hunting behavior and habitat use in altering outcomes of predation under different plant densities may offer insights into patterns of coexistence among larval dytiscids. I performed experiments that quantified behavior of larvae of three common genera of dytiscids that share common prey and then measured predation among genera in the presence or absence of aquatic plants. Behavioral analyses concluded that there were significant differences in foraging modes, with Dytiscus primarily exhibiting sit-and-wait tactics, Graphoderus engaging in active, open water searching, and Rhantus displaying combinations of these behaviors. Predation among larvae was common and occurred when predators were larger than the prey, with no indication of prey preference. Incidence of predation among generic combinations depended on the presence of plants and appeared to be related to behavioral differences among genera. The presence or absence of plants and differences in larval behavior may help to mitigate predation by reducing negative interactions in natural aquatic systems. These results have implications for IGP interactions and may be one of the explanations for the observed richness of this group of predators within aquatic habitats.     

Predation by Pterostichus melanarius (Illiger) (Coleoptera: Carabidae) on immature Rhagoletis mendax Curran (Diptera: Tephritidae) in semi-field and field conditions

Practices that enhance abundance and diversity of generalist predators are often employed with the objective of improving biological control of insect pests. Ground beetles and other predators can prey on blueberry maggot, an important pest of blueberries, when mature larvae pupate in the ground. We conducted mesocosm and field experiments to determine if Pterostichus melanarius, a common predatory ground beetle, lowers maggot numbers in compost mulch or when predator and alternative prey abundances are manipulated. At background (field) densities of alternative prey, increasing densities of P. melanarius did not significantly reduce pest numbers in mesocosms containing compost or soil. When alternative prey were removed from compost, beetles reduced pest numbers by up to 35%. In field experiments, maggot numbers were higher when beetles and other predators were excluded from soil plots, but beetle exclusion had no effect in compost plots where both predator and alternative prey numbers were high. Our results indicate that there can be some reduction of blueberry maggot by P. melanarius and other potential predators when there are few alternative prey. However, despite attracting large numbers of predators compost mulch did not lead to a significant reduction in blueberry maggot; in fact, the high abundance of alternative food associated with compost appeared to interfere with beetle predation on blueberry maggot.     

Spatial and temporal fluctuations in presence and use of chironomid prey by shorebirds in the Odiel saltpans, south-west Spain

We studied the seasonal variation in abundance and distribution of shorebirds and chironomid Chironomus salinarius larvae in both traditional and industrial salines in the Odiel marshes, south-west Spain, in 2001. We selected 12 ponds that were representative of the different phases of the salt production process. The benthic chironomids were sampled in each pond every 2 months, and the birds were counted weekly. Chironomid larvae were most abundant in spring and autumn, and in the ponds of lower salinity. The density of larvae averaged 7023 ± 392 m⁻² (±SE) over the six sampling events. Shorebirds were always more abundant at high tide than at low tide, and were especially abundant during the spring and autumn migration periods when up to 20,775 birds were counted. A total of 24 species were recorded, six of which were present in internationally important numbers. The salines were especially important as foraging and roosting habitat during migration. The percentage of birds that were feeding in the ponds was positively correlated with the abundance of chironomid larvae at accessible depths. The number of feeding birds was also higher in ponds with more chironomid larvae available. Despite more intensive management, industrial salines held higher densities of birds and a similar abundance of chironomids when compared with traditional salines.     

Invertebrate predation predicts variation in an autotomy-related trait in larval damselfly

Autotomy, the discarding of a prey appendage grasped by a predator, may evolve when the benefits of immediate survival outweigh the costs of appendage loss. In larval damselflies, joints connecting lamellae to the abdomen vary in size and shape within and among taxa suggesting that they may evolve under selection by invertebrate predators, such as dragonfly larvae. Assuming that joint width is proportional to the force required for autotomy, we tested if invertebrate predation favours smaller lamellar joints for autotomy or larger joints for structural support of lamellae for swimming. We compared the maximum joint widths of larval Lestes and Enallagma among ponds that varied in risk of invertebrate predation. Relative predation risk estimated as the frequency of regenerated lamellae within ponds was weakly and positively related to the relative abundance of larval dragonflies. The allometry of lamellar joint size decreased with increasing risk of invertebrate predation across ponds after controlling for variation in body size in Lestes congener but not in Enallagma species. Both species of Lestes had larger joint sizes than the five species of Enallagma, suggesting that the ancestral divergence of lamellar joints between these genera may influence contemporary responses to invertebrate predation.     

Diving beetles (Dytiscidae) as predators of mosquito larvae (Culicidae) in field experiments and in laboratory tests of prey preference

Field experiments were performed in artificial ponds to evaluate how the density of predatory diving beetles (Dytiscidae) would affect the population levels of mosquito larvae (Culicidae). Mosquitoes colonizing the ponds were predominantly species of the genus Culex. In 2000, most of the dytiscids colonizing the ponds were small (Hydroporus spp.), and these predators had no impact on the size of larval mosquito populations, not even in ponds with added dytiscids. In 2001, larger beetles (Ilybius, Rhantus, and Agabus spp.) were more common, and there were significantly fewer mosquito larvae in ponds with the highest numbers of dytiscids. There was a negative correlation between numbers of diving beetles in the ponds and the mean body length of mosquito larvae. In neither year could dytiscid densities be maintained above a certain level owing to emigration. In laboratory tests, there were marked differences between three common dytiscid species in regard to preferences for Daphnia and Culex species as prey: Colymbetes paykulli Erichson chose mosquito larvae more often, whereas both Ilybius ater (De Geer) and I. fuliginosus (Fabricius) preferred Daphnia spp. All of the tested dytiscids consumed large numbers of prey. Since some dytiscid species can efficiently decrease populations of mosquito larvae, they are probably important in the natural control of these dipterans.     

Spatially correlated recruitment of a marine predator and its prey shapes the large-scale pattern of density-dependent prey mortality

Patterns of predator dispersal can be critical to the dynamics of prey metapopulations. In marine systems, oceanic currents may shape the dispersal of planktonic larvae of both predators and prey, producing spatial correlations in the recruitment of both species and distinctive geographic patterns of prey mortality. I examined the potential for this phenomenon in two fishes, a wrasse and its grouper predator, at a Caribbean island where the near-shore oceanographic regime produces a temporally consistent spatial pattern of fish recruitment. I found that recruitment and adult abundance of groupers were spatially correlated with recruitment of wrasse prey. Furthermore, the local abundance of predators strongly affected the nature of density-dependent prey mortality. At sites with few predators, wrasse mortality was inversely density-dependent, while mortality was positively density-dependent at sites with higher predator densities. This phenomenon could be important to the dynamics of any metacommunity in which physical forces produce correlated dispersal.     

Predation and disease: understanding the effects of predators at several trophic levels on pathogen transmission

相似文献   

Do nymphs of Parameletus chelifer and P. minor (Ephemeroptera) reduce mortality from predation by occupying temporary habitats?

Summary The possible effect of predation on the habitat selection of nymphs of the two mayfly species Parameletus chelifer Bengtsson and P. minor (Bengtsson) was investigated for different habitats of a North Swedish boreal river. In spring, these two species migrate to river margins and subsequently, in different proportions, to adjacent seasonal ponds. Feeding rates on mayfly nymphs and mosquito larvae by some dytiscid beetles and larvae, a corixid waterbug and a caddisfly larva were studied in the laboratory. In the seasonal pond, the consumption rate of invertebrate predators on Parameletus nymphs was below that expected from their relative abundance, and mosquito larvae were preferred. At the river margin, where mosquito larvae were absent, Parameletus nymphs were consumed in agreement with their relative abundance. The absolute abundance of macroinvertebrate predators was similar at the river margin and in the pond. Fish predators visited the seasonal pond only during a short time of flood, while at the river margin their presence was continuous. In the seasonal pond, minnow and roach consumed nymphs of both P. chelifer and P. minor in a lower proportion than expected. At the river margin, nymphs of P. chelifer were consumed by minnow and ten-spined stickleback in a higher proportion than expected. Nymphs of P. minor were consumed according to expectance. Imagines of P. minor emerged early, and nymphs were equally abundant in the river and in the seasonal pond. In contrast, almost all nymphs of P. chelifer migrated to the seasonal pond. The later emergence of P. chelifer would probably result in a very high mortality from predation if it remained at the river margin.     

Habitat complexity reduces prey vulnerability: An experimental analysis using aquatic insect predators and immature dipteran prey

The effects of alternative prey and structural complexity of habitat on the selection of mosquito larvae by aquatic insect predators were evaluated in the laboratory. The water bugs Anisops bouvieri, Diplonychus (= Sphaerodema) rusticus, and D. annulatus, and the odonate nymphs, Ceriagrion coromandelianum and Brachydiplax chalybea chalybea, selected mosquito larvae based on their abundance relative to chironomid larvae and on the levels of habitat complexity. The effect of one prey species on the other was asymmetrical, as indicated through prey selectivity values. Compared to open habitat, the presence of macrophytes reduced the vulnerability of mosquito larvae while the effect was reverse in the presence of sediments. When both sediment and macrophytes were present in habitats, all the predators except D. annulatus consumed more mosquito larvae than chironomid larvae. The clearance rate, an indicator of predatory efficiency, varied among the predator species and habitat types. The results suggest that the outcome of the interactions between insect predators and mosquito immatures was context-dependent and that it was mediated by the presence of alternative controphic species and the habitat complexity.     

Effects of predator-induced thinning and activity changes on life history in a damselfly

We investigated how the lethal and non-lethal presence and absence of a fish predator, perch (Perca fluviatils), influenced behaviour, numbers emerging, size at emergence, and development rate of the damselfly Lestes sponsa. The experiment was carried out in outdoor artificial ponds and spanned from the egg stage to emergence of the damselflies. During the experiment food resources for the damselflies were continuously monitored. Damselflies exposed to a lethal predator showed a significantly lower activity level than those in the absence of predators or subjected to a non-lethal predator. Half-way through the larval stage the reduction in activity level was correlated with the presence of lethal predators, and at the end of the larval stage with higher zooplankton densities. Though larvae decreased activity level, size at emergence was larger and development time faster for individuals in the lethal predator treatment. Since fewer larvae emerged from that treatment we interpret the larger size at emergence to be an effect of a combination of thinning and higher zooplankton densities.     

Manipulating carabid beetle abundance alters prey removal rates in corn fields

Generalist natural enemies such as carabid beetles have the potential to maintain a variety of pests below outbreak levels in annual crops. To assess the relationship between carabid beetle abundance and field rates of prey removal, we created plots surrounded by different boundaries that selectively affected dispersal of edaphic arthropods, primarily carabids. Three treatments were established: (1) naturally occurring communities, (2) augmented communities using ingress boundaries, and (3) reduced communities using egress boundaries. Selective boundaries altered carabid communities with minimal habitat alteration and without use of insecticides. Three times during the growing season, a fixed number of onion fly pupae were placed in plots to evaluate the impact of carabid abundance on predation rates. A combination of vertebrate and invertebrate exclosures allowed us to evaluate prey removal by invertebrates alone. In comparison to the no boundary treatment, carabids increased 54.2% and decreased 83.1% in plots surrounded by ingress and egress boundaries respectively. Predation rates were positively correlated with carabid abundance (r² = 0.70, p < 0.0001). Significantly more pupae were removed from exclosures allowing access to invertebrates alone than from total exclosures, suggesting that invertebrates represented an important group of predators. Laboratory trials tested the feeding potential of the four most abundant carabid species and showed that they readily consumed onion fly pupae, supporting our hypothesis that carabids were the main predators in field tests. This study corroborates and extends previous observations of the importance of carabid beetles as generalist predators of insect pests in agricultural fields.     

Some notes on the ecology of a GermanBranchipus schaefferi population (Crustacea: Anostraca)

Habitat preference, seasonal occurrence, starvation resistance, hatching eggs ofBranchipus schaefferi, and effects of predation onB. schaefferi were studied.Branchipus was only present in turbid, unvegetated ponds and absent in ponds which contain higher aquatic vegetation and theSpirogyra sp. The first individuals ofB. schaefferi appeared in April when water temperature was 10 °C and the last adults in November at a water temperature of 3.5 °C. Up to 6 reproducing generations were observed during this period. Abundance ofB. schaefferi was higher in temporary ponds than in permanent ponds. Sex ratio was close to unity for most of the year. Body size ofB. schaefferi males and females was significantly positively correlated with pond volume. Without foodB. schaefferi could survive for 1.5 to 2 days at 20 °C and 4 to 5 days at 10 °C. Hatching success of eggs decreased when eggs were dried for 7 months. Freezing of eggs had no effect on hatching success. From] the predators tested,Chaoborus sp. larvae clearly selected smallB. schaefferi; one consumed approximately 6Branchipus d–1 at a density of 6 to 12 prey 1–1. The other predators, dragonfly larvae, and larvae and adults ofTriturus alpestris selected alternative prey types, for exampleTubifex sp. and ostracods.     

Selective predation by larval Agabus (Coleoptera: Dytiscidae) on mosquitoes: support for conservation-based mosquito suppression in constructed wetlands

1. Wetland insect predators can structure aquatic prey communities via selective predation, but receive considerably less attention than vertebrate predators. We conducted laboratory experiments to test selective predation by two species of larval dytiscid beetles ( Agabus ; Coleoptera: Dytiscidae) and the potential contribution of these beetles to suppression of mosquito populations in constructed wetlands.
2.  Agabus consumed copepods, ostracods and mosquito larvae in no-choice tests. When offered a choice, 76% of all prey consumed were mosquito larvae, indicating selective predation. Subsequent experiments revealed this preference was due to ease of capture of mosquito larvae over alternative prey.
3. Cannibalism and intraguild predation were common within and between species of Agabus , which may reduce the overall impact of the observed selective predation.
4.  Agabus larvae selectively preyed on mosquito larvae over alternative prey, which is not characteristic of some fish used as biological control agents for mosquitoes. Predator exclusion or similar experiments in the field could document how these results translate into a natural setting.
5. The findings of this study suggest developing mosquito suppression strategies focused on conservation of native wetland predators. These strategies are preferable to introducing non-native generalist predators, or applying pesticides.     

Direct and indirect effects of predation on tadpole community structure in the Amazon rainforest

Abstract The relationship between the distribution of predators (fish, odonates and water beetles) and prey assemblages (amphibian larvae) was investigated in the tropical rainforest of central Amazonas, Brasil. The anuran community uses a variety of waterbodies for reproduction, ranging from streams and streamside ponds to isolated forest ponds. Predators in this system include fish in streams and streamside ponds, and invertebrates (primarily odonate naiads and beetles) in forest ponds. Tadpole species richness and assemblage structure were related to fish density and species richness. No relationships between tadpole assemblages and abiotic pond characteristics were detected. The presence offish explained much of the variation in both species composition and species richness within and among ponds. Some species of tadpole were consistently found to coexist with high densities of fish. Path analyses suggest that while fish have a strong direct effect on tadpole associations and species richness, they also have an indirect effect through invertebrate predators (odonate larvae and coleopteran beetles). Prey survival-strategies such as palatability and behaviour may explain how tadpole species composition is affected by predators at the community level. These findings suggest that the observed patterns of habitat use by larval anurans may be structured in response to the distribution of key predators (fish) in this system.     

Context-Dependent Plastic Response during Egg-Laying in a Widespread Newt Species

Previous research on predator-induced phenotypic plasticity mostly focused on responses in morphology, developmental time and/or behaviour during early life stages, but the potential significance of anticipatory parental responses has been investigated less often. In this study I examined behavioural and maternal responses of gravid female smooth newts, Lissotriton vulgaris, in the presence of chemical cues originating from invertebrate predators, Acilius sulcatus water beetles and Aeshna cyanea dragonfly larvae. More specifically, I tested the extent of oviposition preference, plasticity in egg-wrapping behaviour and plasticity in egg size when females had the possibility to lay eggs at oviposition sites with and without predator cues during overnight trials. I found that individuals did not avoid laying eggs in the environment with predator cues; however, individuals that deposited eggs into both environments adjusted the size of the laid eggs to the perceived environment. Females deposited larger eggs earlier in the season but egg size decreased with time in the absence of predator cues, whereas individuals laid eggs of average size throughout the investigated reproductive period when such cues were present. Also, egg size was found to be positively related to hatching success. Individuals did not adjust their wrapping behaviour to the presence of predator cues, but females differed in the extent of egg-wrapping between ponds. Females’ body mass and tail depth were also different between ponds, whereas their body size was positively associated with egg size. According to these results, female smooth newts have the potential to exhibit activational plasticity and invest differently into eggs depending on temporal and environmental factors. Such an anticipatory response may contribute to the success of this caudate species under a wide range of predator regimes at its natural breeding habitats.     

Time‐restricted flight ability influences dispersal and colonization rates in a group of freshwater beetles

Variation in the ability to fly or not is a key mechanism for differences in local species occurrences. It is increasingly acknowledged that physiological or behavioral mechanisms rather than morphological differences may drive flight abilities. However, our knowledge on the seasonal variability and stressors creating nonmorphological differences in flight abilities and how it scales to local and regional occurrences is very limited particularly for small, short‐lived species such as insects. Here, we examine how flight ability might vary across seasons and between two closely related genera of freshwater beetles with similar geographical ranges, life histories, and dispersal‐related morphology. By combining flight experiments of >1,100 specimens with colonization rates in a metacommunity of 54 ponds in northern and eastern Europe, we have analyzed the relationship between flight ability and spatio‐environmental distribution of the study genera. We find profound differences in flight ability between the two study genera across seasons. High flight ability for Acilius (97% of the tested individuals flew during the experiments) and low for Graphoderus (14%) corresponded to the different colonization rates of newly created ponds. Within a 5‐year period, 81 and 31% of the study ponds were colonized by Acilius and Graphoderus, respectively. While Acilius dispersed throughout the season, flight activity in Graphoderus was restricted to stressed situations immediately after the emergence of adults. Regional colonization ability of Acilius was independent of spatial connectivity and mass effect from propagule sources. In contrast, Graphoderus species were closely related to high connectivity between ponds in the landscape. Our data suggest that different dispersal potential can account for different local occurrences of Acilius and Graphoderus. In general, our findings provide some of the first insights into the understanding of seasonal restrictions in flight patterns of aquatic beetles and their consequences for species distributions.     

Opportunistic foraging by heteropteran mosquito predators

Tropical aquatic environments host a large number of predatory insects including heteropteran water bugs Anisops bouvieri Kirkaldy, 1704 (Heteroptera: Notonectidae), Diplonychus (=Sphaerodema) rusticus Fabricius, 1781 (Heteroptera: Belostomatidae), and Diplonychus (=Sphaerodema) annulatus Fabricius, 1781 (Heteroptera: Belostomatidae) feeding on a range of organisms. In tropical and subtropical wetlands, ponds, and temporary pools these predators play a role in regulation of dipteran populations, particularly mosquitoes and chironomids. Their relative abilities to control mosquitoes depend in part on predator preference for mosquitoes in relation to other natural prey, and the predators’ propensities to switch to mosquitoes as mosquito density increases. The prey electivity and switching dynamics of these predatory water bugs were evaluated in the laboratory under various prey densities, using two instars (II and IV) of chironomid and mosquito larvae as prey. Studies of electivity at relatively high densities (20 prey L⁻¹) in small (5 L) vessels demonstrated that all predators showed opportunistic foraging as the mosquito:chironomid ratio changed, with some evidence that mosquito larvae were positively selected over chironomids. In particular, Anisops showed strong electivity for mosquitoes when presented with any ratio of large mosquito and chironomid prey in the high density experiment, although the preference was not expressed in lower density (2.5 prey L⁻¹) treatments executed in 40 L vessels. In these lower density treatments, D. rusticus demonstrated higher electivity for mosquitoes when the mosquito:chironomid ratio was high, consistent with non-significant trends observed in the higher density experiment. The positive electivity of D. rusticus for mosquitoes was reinforced in an experiment executed over 16 days at varying prey ratios, in which D. rusticus mosquito electivity was high and consistent while D. annulatus showed slight avoidance of mosquito larvae, and Anisops remained largely opportunistic in foraging on prey in proportion with availability. Anisops and D. rusticus are potentially good biocontrol agents for mosquito larvae, in that they preferentially consume mosquitoes under many circumstances but can readily forage on other prey when mosquito density is low.     

Feeding ecology of larval and juvenile American shad (Alosa sapidissima) in a small pond

Release ponds are used as part of a multifacet effort to restore American shad ( Alosa sapidissima ) in the upper Susquehanna River basin. Little information exists, however, on the feeding ecology of young shad in small ponds. Consequently, we examined feeding ecology and prey selection of 299 larval and 299 juvenile American shad in a small pond during spring and summer. Larval shad mainly consumed copepods (37.7%) and cladocerans (37.4%) whereas juvenile shad ate chironomids (43.1%) and ostracods (28.4%). Larval and juvenile shad exhibited diel variation in diet composition and feeding periodicity. Food consumption by shad was minimal at night; feeding activity was highest during the day, peaking at 2000 h for both larvae and juveniles. Electivity values of shad larvae for prey taxa were highest for cladocerans (+0.27) and lowest for ostracods (−0.07). Electivity values of juvenile shad were highest for chironomids (+ 0.21) and ostracods (+ 0.09), and lowest for copepods (− 0.08) and baetids (− 0.14). Our data indicate differences in diet composition, prey preference and, to a lesser extent, feeding patterns between larval and juvenile American shad in small pond environments.     

An Apparent Trade-Off between Direct and Signal-Based Induced Indirect Defence against Herbivores in Willow Trees

Signal-based induced indirect defence refers to herbivore-induced production of plant volatiles that attract carnivorous natural enemies of herbivores. Relationships between direct and indirect defence strategies were studied using tritrophic systems consisting of six sympatric willow species, willow leaf beetles (Plagiodera versicolora), and their natural predators, ladybeetles (Aiolocaria hexaspilota). Relative preferences of ladybeetles for prey-infested willow plant volatiles, indicating levels of signal-based induced indirect defence, were positively correlated with the vulnerability of willow species to leaf beetles, assigned as relative levels of direct defence. This correlation suggested a possible trade-off among the species, in terms of resource limitation between direct defence and signal-based induced indirect defence. However, analyses of volatiles from infested and uninfested plants showed that the specificity of infested volatile blends (an important factor determining the costs of signal-based induced indirect defence) did not affect the attractiveness of infested plant volatiles. Thus, the suggested trade-off in resource limitation was unlikely. Rather, principal coordinates analysis showed that this ‘apparent trade-off’ between direct and signal-based induced indirect defence was partially explained by differential preferences of ladybeetles to infested plant volatiles of the six willow species. We also showed that relative preferences of ladybeetles for prey-infested willow plant volatiles were positively correlated with oviposition preferences of leaf beetles and with the distributions of leaf beetles in the field. These correlations suggest that ladybeetles use the specificity of infested willow plant volatiles to find suitable prey patches.     

Riparian ground beetles (Coeloptera, Carabidae) preying on aquatic invertebrates: a feeding strategy in alpine floodplains

The food and feeding habits of riparian ground beetles were studied in four alpine floodplains (Bavaria, Germany): a 5th-order stream (the Isar) and three 3rd-order streams. The riparian fauna along the streams mainly consists of predaceous species. Riparian ground beetle densities were much higher along the Isar than along the small streams. Aquatic invertebrates composed 89% of the potential prey for carnivorous terrestrial insects along the Isar. Besides aquatic organisms washed ashore, stoneflies emerging on land are of considerable importance as potential prey for terrestrial predators. In contrast, only 34% of the potential prey organisms collected along the small streams were of aquatic origin. Food abundance was 9 times higher in the shore region of the Isar compared to the small streams. Surface drift in the Isar, a potentially important food source for riparian organisms, was about 10⁶ organisms and exuviae per meter stream width in 24 h. The drift density in the Isar was 59 times higher than that in a small stream. Terrestrial organisms provided only 3% of the drifting particles in the Isar, but 50% in the small stream. Gut content analysis reveals, that riparian ground beetles in the Isar floodplain mainly feed on aquatic organisms washed ashore or emerging on land. While small Bembidion species prefer chironomids (larvae and adults) the larger species Nebria picicornis feeds on emerging stoneflies, terrestrial riparian organisms and aquatic organisms accumulating along the shoreline. The prey of riparian ground beetles in the floodplain of the three small streams mainly consists of terrestrial species some of which may have been washed ashore. Received: 2 September 1996 / Accepted: 26 February 1997