Helen L. Hooper, School of Animal and Microbial Sciences, University of Reading, PO Box 228, Reading RG6 6AJ, UK (fax +44 1189 310180; e-mail firstname.lastname@example.org).
1Chemical effects on organisms are typically assessed using individual-level endpoints or sometimes population growth rate (PGR), but such measurements are generally made at low population densities. In contrast most natural populations are subject to density dependence and fluctuate around the environmental carrying capacity as a result of individual competition for resources. As ecotoxicology aims to make reliable population projections of chemical impacts in the field, an understanding of how high-density or resource-limited populations respond to environmental chemicals is essential.
2Our objective was to determine the joint effects of population density and chemical stress on the life history and PGR of an important ecotoxicological indicator species, Chironomus riparius, under controlled laboratory conditions. Populations were fed the same ration but initiated at different densities and exposed to a solvent control and three concentrations of 14C-cypermethrin in a sediment–water test system for 67 days at 20 ± 1 °C.
3Density had a negative effect on all the measured life-history traits, and PGR declined with increasing density in the controls. Exposure to 14C-cypermethrin had a direct negative effect on juvenile survival, presumably within the first 24 h because the chemical rapidly dissipated from the water column. Reductions in the initial larval densities resulted in an increase in the available resources for the survivors. Subsequently, exposed populations emerged sooner and started producing offspring earlier than the controls. 14C-cypermethrin had no effect on estimated fecundity and adult body weight but interacted with density to reduce the time to first emergence and first reproduction. As a result, PGR increased with cypermethrin concentration when populations were initiated at high densities.
4Synthesis and applications. The results showed that the effects of 14C-cypermethrin were buffered at high density, so that the joint effects of density and chemical stress on PGR were less than additive. Low levels of chemical stressors may increase carrying capacity by reducing juvenile competition for resources. More and perhaps fitter adults may be produced, similar to the effects of predators and culling; however, toxicant exposure may result in survivors that are less tolerant to changing conditions. If less than additive effects are typical in the field, standard regulatory tests carried out at low density may overestimate the effects of environmental chemicals. Further studies over a wide range of chemical stressors and organisms with contrasting life histories are needed to make general recommendations.
In most chemical regulatory schemes, effects on populations of aquatic invertebrates are generally inferred from chronic or long-term studies (e.g. a Daphnia sp. 21-day reproduction study, OECD 1998; and a Chironomus sp. 28-day emergence study, OECD 2001) that examine the effects of a toxicant on the survival, growth and/or reproduction of an individual. However, the results of these studies are not usually expressed in terms of population growth rate (PGR), the usual ‘currency’ of population ecologists. Some researchers use life table response experiments (LTRE; Allen & Daniels 1982; Bechmann 1994; Levin et al. 1996) to examine toxicant effects on PGR, but in the majority of these studies organisms are exposed individually or at low density and are provided with adequate food. In contrast, most natural populations are regulated by density dependence and fluctuate around the environmental carrying capacity as a result of individual competition for resources. Because ecotoxicology aims to make reliable population projections of toxicant impacts in the field, an understanding of how resource-limited or high-density populations respond to environmental chemicals is essential and should be considered in ecological risk assessments (Sibly 1999a).
In general, increasing density and increasing chemical concentration reduce PGR when applied separately (Sibly & Hone 2002) but few studies have investigated how populations respond to both stressors when applied simultaneously. Concerns arise because populations regulated by density dependence have fewer resources and therefore less energy available for defence against environmental stress, and so may be more susceptible to toxicant stress (Sibly 1999b). LTRE carried out at low densities may therefore underestimate the population response to chemical stress. However, simulation studies indicate that when survival or production rates are density dependent, mortalities caused by the toxicant may alleviate the intensity of the density dependence because more resources per capita would be available to the survivors (Calow, Sibly & Forbes 1997; Grant 1998; Hansen, Forbes & Forbes 1999). Consequently, resource-limited or high-density populations may compensate for the negative effects of a chemical stressor. LTRE may therefore overestimate toxicant effects on field populations.
In a recent review, Forbes, Sibly & Calow (2001) considered the possible ways in which density dependence (operating mainly via food limitation) and toxicants may affect PGR. They described three general types of interaction: additive, less than additive and more than additive. In the additive model, the combined effects of density dependence and toxicant concentration are independent (i.e. no interaction). In the less than additive model, toxicant effects are buffered or compensated. Conversely, in the more than additive model toxicant effects are exacerbated at high density.
The aim of this study was to measure the joint effects of population density and chemical stress on the life history and PGR of an important ecotoxicological indicator species, Chironomus riparius Meigen, under controlled laboratory conditions. The synthetic pyrethroid cypermethrin was selected as the chemical stressor. Pyrethroids have been widely used for more than 20 years to control insect pests in a variety of crops. The toxicity, behaviour and fate of these lipophilic insecticides have been well studied in both the laboratory (Stephenson 1982; Sakata et al. 1986; Maund et al. 2002) and in the field (Crossland, Shires & Bennett 1982; Hill, Shaw & Maund 1994; Farmer, Hill & Maund 1995). Pyrethroids have a short half-life in the water column (generally less than 2 days) and rapidly adsorb to suspended particulate material and sediments, which greatly reduces their bioavailability to water column organisms. However, the potential exists for benthic organisms like C. riparius to be exposed by direct contact or ingestion of contaminated sediment particles or via the interstitial (pore) water. We hypothesized that high-density C. riparius populations exposed to cypermethrin in a sediment–water test system would compensate for the initial mortalities caused by the toxicant (as in the less than additive model) because more resources (food and space) would be available to the survivors.
Materials and methods
life cycle of c. riparius
Most of the life cycle of chironomids is spent in the juvenile phase, the non-feeding adult stage being brief and largely concerned with reproduction and dispersal (Oliver 1971; Pinder 1986; Armitage, Cranston & Pinder 1995). Chironomus riparius mates on the wing and the female deposits an egg rope into freshwater. The larvae hatch within a couple of days and initially disperse in the water column before settling on the sediment. After constructing a protective case, the larvae frequently emerge from the open end to feed on detritus on the sediment surface (Rasmussen 1984). Once the larvae are mature pupation is triggered by environmental cues, and the adults emerge a day or so later. Organisms for the study were obtained from a culture maintained at Syngenta, Jealott's Hill International Research Centre, UK (see Experimental animals for culturing details).
Laboratory populations of newly hatched larvae were initiated at four different densities to span the range often reported for field populations of C. riparius (Davis & Hawkes 1981; Rasmussen 1984; Groenendijk et al. 1998). The densities chosen were 0·5, 1, 2 and 4 cm−2, corresponding to 64, 128, 256 and 512 larvae vessel−1. Each density was exposed to a solvent control and three concentrations of 14C-cypermethrin in a sediment–water experimental system.
Treatments were selected to provide nominal concentrations of 0, 0·0150, 0·0225 and 0·0300 µg l−1 overlying water (control, low, medium and high, respectively). On the basis that the bioavailability of lipophilic compounds adsorbed to sediments is best predicted from the concentration in the water phase at equilibrium (DiToro et al. 1991; Maund et al. 2002), low-density pilot experiments were carried out to predict the aqueous concentrations that would result in 50% (low), 70% (medium) and 90% (high) juvenile mortality. For each density and chemical treatment there were two replicates, giving 32 test populations in total. All populations were fed a fixed ration (25 mg day−1) of Tetramin® fish flakes (J & K Aquatics, Taunton, UK), so as density increased per capita food and space availability decreased.
Radiolabelled cypermethrin was used to facilitate the analysis of the chemical in the experimental systems and in the adult tissues. 14C-phenoxy-labelled cypermethrin in acetone with a specific activity of 2·1 GBq mmol−1 and a purity of > 95% was obtained from Syngenta, Jealott's Hill International Research Centre. The water solubility of cypermethrin at 20 °C and pH 7 is 4 µg l−1 (Tomlin 1994).
For each of the four chemical treatments (control, low, medium and high), a test solution was prepared using known volumes of acetone and 14C-cypermethrin to give the required nominal concentration per system in 90 µl. The radioactivity of the test solutions was quantified by liquid scintillation counting (LSC) before and immediately after spiking the test systems (see Chemical analyses).
The experiment was carried out at 20 ± 1 °C. Photoperiod was controlled to provide 16 h light and 7 h dark, with two 30-min transitional periods representing dawn and dusk to stimulate swarming. The light intensity directly above the experimental systems was between 800 and 1000 lux at the start of the experiment.
Each population was contained in a plastic aquarium (length × width × height = 15 × 8·5 × 10·5 cm3) covered with fine netting. A measured amount (41·32 g) of standardized, formulated sediment (consisting of 70% sand, 20% kaolin clay and 10% peat; OECD 1984) and 0·9 l reconstituted water were added to each vessel (see Spiking procedure) to give a sediment and water depth of 0·5 cm and 6 cm, respectively. Thus the top 4 cm of the test vessel was available for the adults and exceeded the volume (2–3 cm3) reported to be necessary for swarming (Caspary & Downe 1971). The aqueous medium was reconstituted water, which was prepared by dissolving 122·5 mg MgSO47H2O, 96 mg NaHCO3, 60 mg CaSO42H2O and 4 mg KCl in 1 l of de-ionized water. Total hardness (CaCO3) and initial pH ranged from 70 to 85 mg l−1 and 7·8–8·2, respectively. A day after adding the test organisms to the experimental systems, the overlying water was gently aerated to minimize fluctuations in dissolved oxygen. After 3 days of exposure, the aqueous phase was continuously renewed at a rate of 38 ml h−1 to avoid the accumulation of excreted material. Water temperature, measured twice weekly in the control systems, was within the range 20 ± 1 °C.
For each population, 41·32 g (dry weight) of formulated sediment and 0·2 l of aerated reconstituted water were added to a 500-ml glass jar. The jars were sealed and vigorously shaken to ensure thorough mixing before leaving the sediments to soak for a week. An additional four jars (one per concentration) were prepared in the same way specifically for chemical analyses at the start of the experiment (see Chemical analyses). Two days before the populations were initiated, each jar was shaken again and the slurry was immediately spiked with the appropriate test solution using a 100-µl Hamilton syringe. The jars were resealed and mixed on a roller at 0·008 g for 30 min. The treated slurries were poured into separate test vessels and were allowed to settle for 24 h. The remaining 0·7 l of overlying water was added to each vessel by trickling over a plastic disc floating on the settled mixture. The experimental systems were allowed to equilibrate for a further 24 h before initiating the populations. In our pilot studies this method proved suitable to obtain the nominal concentrations in the overlying water at day 0.
Chironomus riparius egg ropes were obtained from a culture maintained at Jealott's Hill International Research Centre. The strain originated from a culture maintained at Cardiff University, Cardiff, UK, which was identified 20 years ago by Dr M. A. Learner. Culturing methods are based on those described by McCahon & Pascoe (1988). Larvae were reared at low density with unlimited food and completed their life cycle in 3–4 weeks at 20 ± 1 °C. Three days before the start of the experiment, 40 egg ropes (deposited within the previous 24 h) were removed from the cultures using a plastic pipette and were divided equally between four hatching dishes containing 200 ml of reconstituted water. Three days later (day 0) newly hatched larvae were randomly selected from each dish and transferred to the treated experimental systems using a Pasteur pipette. After a 2-h settling period and daily thereafter, each population was fed 25 mg of Tetramin® fish flakes. The food was ground to a fine powder, prepared as a suspension and administered using a digital pipette.
biological parameters and analyses
Populations were monitored for 67 days and the following parameters were recorded daily: the number and sex of the pupal exuviae (these were removed from the experimental systems and used to indicate the number of male and female flies that emerged); the number of egg ropes; the number of eggs per rope. The latter was estimated using the ring count method (Benoit et al. 1997), which predicted actual counts with a very high confidence in a pilot experiment (regression analysis R2 = 97%, n= 10, P < 0·001). Chironomus riparius egg ropes consist of a single strand of eggs, typically in a coil-like arrangement, surrounded by a gelatinous matrix. Using the ring count method, the number of eggs per rope is estimated by multiplying the mean number of eggs per coil or ring, derived from five rings at equal distances along the length of the rope, by the total number of rings in the rope. Individual egg ropes were stored for up to 1 week in a sealed glass vial containing 30 ml of reconstituted water at 20 ± 1 °C to measure hatching success.
For each population, PGR was calculated as 1/t loge (N2/N1), where N1 is the initial larval density, N2 is the number of offspring and t is the time in weeks when the most offspring were produced or the modal reproduction time. Using the mode is an objective way of assessing generation time that is not affected by the time the experiment was stopped. Juvenile survival to adult emergence was calculated as the number of adults divided by the initial larval density. The time to first adult emergence and the time to first reproduction were determined from when the populations were initiated (day 0), and fecundity was estimated as the number of eggs per population divided by the number of females that emerged. The adults were removed from the experimental systems once they had died, and a random sample of each sex (n = 10) was oven-dried at 60 °C for 48 h. Individuals were weighed on a Cahn 29 Automatic Electrobalance (Cahn Instruments Inc., Cerritos, California, USA) to the nearest µg before combusting the samples (see Chemical analyses) to determine the radioactivity in the adult tissues. The joint effects of larval density and chemical stress on the life history and PGR of C. riparius were analysed using a two-way anova (MINITAB 13). The residuals were examined for normality and homogeneity of variances and were found to comply reasonably well with anova assumptions.
The radioactivity in the sediments and overlying water was measured by destructively sampling one replicate of each chemical treatment. The additional replicate systems set up specifically for chemical analyses were sampled at the start of the study, while systems initiated at high population density were sampled at the end of the study. For each replicate, duplicate aliquots of 245 ml of the overlying water were transferred to 250-ml glass volumetric flasks using a digital pipette. Each sample was extracted in 5 ml hexane and aliquots were taken for LSC analyses. The remainder of the overlying water was removed and sediment samples of roughly 1·5 g (four per replicate) were transferred to 50-ml Teflon centrifuge tubes. Acetonitrile (30 ml) was added to each tube, which was sealed and placed on an end-over-end shaker at 22 r.p.m. for ≥ 1 h. The tubes were centrifuged (4000 r.p.m. for 15 min at room temperature) and the resulting supernatants were transferred to 100-ml glass volumetric flasks. This extraction process was repeated twice more, the supernatants from the same tubes were combined and their volumes adjusted to 100 ml with acetonitrile. Aliquots were taken for LSC analyses. The residual sediment pellets were dried in a fume cupboard and weighed to the nearest mg. The radioactivity remaining on the pellets was determined by combustion followed by LSC.
Combustions were carried out using a Harvey OX500 Biological Oxidizer (Lab Impex Teddington, UK). Samples (sediment and adults) were weighed into a cellulose combustion cone (Packard Instrument Co., Meriden Conn, USA), wrapped inside a second cone and compressed into a pellet. The 14CO2 that evolved from the combusted samples was trapped in 2-methoxyethylamine, part of a combustion cocktail consisting of Optiphase Safe™ (Fisher Scientific Loughborough, UK): 2-methoxyethylamine : de-ionized water (50 : 25 : 2). The efficiency of the oxidizer (> 96%) was determined by combustion of a known quantity of radioactivity (Spec-Chec™; Packard Limited) spiked into a combustion cone.
The radioactivity in the test solutions, sediments, overlying water and adult samples was quantified by LSC using a Wallac 1409 Liquid Scintillation Counter (Perkin Elmer, Cambridge, UK), and Optiphase Safe™ scintillation cocktail. Each group of samples was preceded by two blank samples from which a mean was calculated. This mean background count was automatically subtracted from subsequent samples to provide net disintegrations per second values (Bq).
Aqueous and sediment extracts were further analysed by thin-layer chromatography (TLC) to determine what proportion of the extracted radioactivity was parent cypermethrin as opposed to breakdown products. TLC was carried out using Macherey-Nagel (SIL-G-25 UV 254) pre-coated plates (Fisher Scientific). Aliquots were applied to the plates in 1-cm bands using a digital pipette. Samples were chromatographed in parallel with an unlabelled standard of cypermethrin in a toluene : hexane : chloroform : acetonitrile (60 : 20 : 4 : 1) solvent system. The radioactive areas on the plate were quantified using a Fuji BAS 2000 phosphorimager (Raytex, Sheffield, UK).
Chemical concentrations in the sediments and overlying water were measured at the start of the study (day 0) by destructively sampling the extra replicate system for each chemical treatment. TLC analysis confirmed that the extracted radioactivity was > 87% and > 80% parent cypermethrin in the sediment and water samples, respectively. Measured values (expressed as equivalent 14C-cypermethrin concentrations) were compared with the expected nominal concentrations (Fig. 1a). While there was good agreement between the sediment values (within 11·5% of nominal), the measured overlying water concentrations were c. 50% lower than nominal. This may have been due to an error in the spiking concentrations; however, obtaining accurate measurements of lipophilic compounds in the aqueous phase is notoriously difficult. One of the advantages of using a radiolabelled toxicant is that it allows body burdens to be measured, which provide an indication of exposure concentrations (see below). In addition we decided to sample the overlying water (20 ml) of the two replicate systems, initiated at high population density and spiked with the high chemical treatment (small volumes of the low and medium chemical treatments were below the limits of detection). Aqueous samples were taken 48 h, 72 h, 96 h and 168 h after initiating the populations and the measured values were expressed as equivalent 14C-cypermethrin concentrations (Fig. 1b). These results show that the chemical rapidly dissipated from the water column with a half-life of c. 2 days and suggest that the overlying water concentrations in systems containing larvae may have been similar to nominal on day 0.
At the end of the study (day 67) no radioactivity was detected in the overlying water and that in the sediments had declined by c. 60%. Measured sediment values expressed as equivalent 14C-cypermethrin were 0·049, 0·073 and 0·086 mg kg−1 for the low, medium and high chemical treatments, respectively, although on average only 50% of the extracted radioactivity was parent cypermethrin.
The radioactivity recovered from the adult tissues expressed as equivalent 14C-cypermethrin concentrations is shown in Fig. 2. The effects of larval density and chemical concentration were analysed using a two-way anova (which excludes the control populations). In both sexes, tissue concentrations increased with chemical concentration (males F2,12 = 8·01, P= 0·006; females F2,12 = 19·79, P ≤ 0·001), indicating that the larvae were actually exposed to increasing amounts of the pesticide. Body burdens also increased with larval density (males F3,12 = 4·02, P = 0·034; females F3,12 = 20·55, P ≤ 0·001). This may have been due to the longer exposure time at higher densities (see Biological results), although it is clear from Fig. 1b that the chemical rapidly dissipated from the water column and thus was only bioavailable for a short time. The initial lack of per capita resources may have resulted in less energy to metabolize the chemical, or food-limited larvae may have ingested more sediment (relative to fish flake food) and were exposed via this route. Alternatively, higher densities may have led to greater flushing of the pore water, which may have increased the bioavailability of the chemical. Note that within each population, males had slightly higher tissue concentrations than females (paired t23 = 3·04, P= 0·006). It is possible that some of the chemical may have adsorbed to the gelatinous matrix that surrounds the eggs and was therefore excreted when the females deposited their egg ropes.
The joint effects of larval density and 14C-cypermethrin on the life history and PGR of C. riparius were analysed using a two-way anova (Table 1). Density affected all nine parameters and chemical concentration affected five parameters. The effects of the toxicant on these five parameters differed with density and always resulted in a significant interaction between the two stressors (Table 1).
Table 1. Results of statistical analyses (two-way anova) showing the effects of larval density and 14C-cypermethrin on the life history and population growth rate of the midge C. riparius under constant laboratory conditions. Juvenile survival data were transformed prior to analysis using arcsin (square root)
Juvenile survival to emergence
Time to first emergence
Time to first reproduction
Modal reproduction time
Number of offspring per population
Fecundity (number of eggs per female)
Male body weight
Female body weight
Population growth rate (week−1)
Juvenile survival to adult emergence (Fig. 3) declined with increasing density. In the control populations initiated at high density, the proportion that emerged in 67 days was reduced by 94%. The effects of 14C-cypermethrin differed with larval density and resulted in a significant interaction between the two stressors (Table 1). At low density, juvenile survival declined with increasing concentration; conversely, at high density, survival to emergence increased threefold with increasing concentration (Fig. 3). Survival at low density was up to 48% higher than expected on the basis of pilot studies. However, predictions of toxicity using the concentration of the chemical in the water phase at equilibrium are typically only accurate within a factor of two to three (DiToro et al. 1991; Maund et al. 2002).
The time between population initiation and first emergence (Fig. 4a) increased with increasing density. In the control populations initiated at high density, emergence was delayed by c. 3 weeks. The chemical stressor had little effect at low density: adults started to emerge 15–17 days after the experiment started, which may suggest a minimum development period is necessary. At higher densities, exposure to 14C-cypermethrin had a positive effect by reducing the time to first emergence, generally with increasing concentration. This resulted in a significant interaction between the two stressors (Table 1 and Fig. 4a).
The time between population initiation and first reproduction increased with increasing density (Fig. 4b). In the high-density controls, reproduction was delayed by c. 6 weeks, which was due to the delay in female emergence. At low density females emerged a couple of days after males but at high density females emerged up to 4 weeks later, although egg ropes were generally deposited within a few days of female emergence. Exposure to 14C-cypermethrin had little effect at low density, but as density increased the chemical had a positive effect by reducing the time to first reproduction, generally with increasing concentration (Fig. 4b). This resulted in a significant interaction between the two stressors (Table 1).
The modal reproduction time (i.e. from population initiation to when the most offspring were produced) increased with increasing density (Fig. 4c). Note that the experiment was stopped when the high-density control populations produced their first and consequently most offspring (67 days). In contrast, the high-density exposed populations were increasingly producing offspring as adult emergence (Fig. 4a) and reproduction (Fig. 4b) occurred sooner. The modal reproduction time for populations initiated at the highest density was therefore the same in all treatments within the time frame of this study. However, it is clear from Fig. 4c that exposure to increasing 14C-cypermethrin concentrations had a positive effect when populations were initiated at 1 and 2 larvae cm−2. Modal reproduction time declined with increasing concentration and resulted in a significant interaction between the two stressors (Table 1).
In all populations > 85% of the egg ropes hatched. The number of offspring per population (Fig. 5a) was significantly affected by population density (Table 1). Exposure to 14C-cypermethrin had marginal effects: offspring number declined with increasing concentration at low density and increased with increasing concentration at high density and appeared to be related to the number of females that emerged. However, there was no significant (P < 0·05) toxicant effects or interaction between the two stressors (Table 1), probably due to the high variability and low replication of treatments. Estimated fecundity (Fig. 5b) declined with increasing density, in part because at high densities not all females deposited an egg rope. Exposure to 14C-cypermethrin had no effect on estimated fecundity (Table 1). Similarly, adult body weight (Fig. 6) generally declined with increasing density but exposure to the pesticide had no effect (Table 1).
Weekly per capita PGR (Fig. 7a), calculated using the number of offspring and the modal reproduction time, declined with increasing density and in the controls was negative at the highest density. The effects of 14C-cypermethrin differed with density and resulted in a significant interaction between the two stressors (Table 1). At low density, exposure to the chemical stressor appeared to have little effect, while at higher densities PGR increased with increasing concentration (Fig. 7a).
PGR contours plotted against initial larval density and 14C-cypermethrin concentrations (MINITAB 13) (Fig. 7b) tended to be concave which suggested that the combined effect of the stressors may be to increase the PGR of C. riparius. The PGR = 0 contour is of particular importance because it indicates carrying capacity; however, interpolation was limited as only the two replicate control populations initiated at high density had an estimated PGR < 0 (Fig. 7a).
We aimed to determine the joint effects of population density and 14C-cypermethrin on the life history and PGR of C. riparius under controlled laboratory conditions. Increasing density had a negative effect on PGR, whereas exposure to increasing chemical stress had a positive effect on PGR when populations were initiated at high density. These results support the hypothesis that high-density or resource-limited populations compensate for toxicant effects, as in the less than additive model outlined by Forbes, Sibly & Calow (2001).
In agreement with a previous study (Hooper et al. 2003), increasing density had a negative effect on all the measured life-history traits of C. riparius. Exposure to 14C-cypermethrin had a direct negative effect on juvenile survival, presumably within the first 24 h of exposure, because the chemical rapidly dissipated from the water column. A reduction in the initial larval densities resulted in an increase in the available resources (food and space) for the survivors, which emerged sooner and started producing offspring earlier than the control populations. The chemical stressor therefore had an indirect positive effect on juvenile development time when populations were initiated at high densities. Regardless of density, exposure to 14C-cypermethrin had no significant effect on estimated fecundity, the number of offspring per population and adult body weight.
At low density, juvenile development time was unaffected by the pesticide and the survivors emerged at a similar time (day 15–17). However, toxicant-related mortalities were compensated for by breeding earlier, which resulted in similar estimates of PGR at low density. It has been reported that in most chironomid species females emerge with eggs that are one-third to one-half developed (Oliver 1971). As the exposed females reached the critical body size (Hooper et al. 2003) within the hypothesized minimum juvenile development period, the increase in resources and therefore energy (as a result of toxicant-related mortalities) may have been diverted towards egg maturation, thus allowing reproduction to occur sooner.
The data reported here are in broad agreement with observations of chironomid populations in mesocosm studies. In a study where cypermethrin was sprayed over the surface of mesocosms, equivalent to spray drift following application at the typical field rate, concentrations in water samples taken 1 h after application were around 0·03 µg l−1 and after 24 h had declined to about 0·009 µg l−1 (Farmer, Hill & Maund 1995). These measured concentrations and the reported effects on chironomid populations are comparable with the present laboratory study in that exposure to cypermethrin initially reduced the number of larvae but recovery was rapid and 8 weeks after application estimated densities were up to three times higher in the treated mesocosms than the controls (Farmer et al. 1995).
Previous work has shown that when food alone is limiting, laboratory populations of C. riparius buffer the effects of a chronic stressor, cadmium (Postma et al. 1994). Exposed food-limited populations attained a higher PGR than the food-limited controls because mortalities occurred during the early larval stages, leaving more food for the survivors as reported here. In contrast to our results, food-limited populations compensated for the negative effects of cadmium by an increase in the fecundity of the surviving females (Postma et al. 1994), not by a reduction in juvenile development time. Interestingly, PGR declined with increasing cadmium concentrations when populations were well fed, as mortalities occurred during the late juvenile stages and juvenile development time increased with cadmium concentration (Postma et al. 1994).
Results from the present study and Postma et al. (1994) suggest that high-density or food-limited C. riparius populations buffer or compensate for the effects of both acute (cypermethrin) and chronic (cadmium) chemical stressors, although different life-history traits are affected. Perhaps this is not surprising as Forbes & Calow (1999) found no consistent pattern in the data obtained from 41 experimental studies (mostly on cladocerans) regarding which life-history traits were most or least sensitive to toxicant exposure. Because our prime interest was in determining the population response using an efficient experimental design, our method did not allow detailed schedules of individual life-history traits to assess directly their relative sensitivities across treatments and their contribution to PGR.
Cadmium has been the model chemical stressor in a number of laboratory studies examining the response of food-limited cladoceran populations. Less than additive effects on PGR were reported for Daphnia galeata mendotae (Marshall 1978) and Moinodaphnia macleayi (Barata, Baird & Soares 2002). In both studies, food-limited populations compensated for the cadmium-related mortalities in the same way as C. riparius (Postma et al. 1994), by an increase in the reproductive rate of the survivors. However, cadmium effects on other food-limited cladocerans, Daphnia magna (Kluttgen & Ratte 1994) and Echinisca triserialis (Chandini 1988), were additive and more than additive, respectively. We have some confidence in ignoring the Kluttgen & Ratte (1994) study, as the daphnids were exposed individually so mortality effects at the population-level were not incorporated in the experimental design. In Chandini's (1988) study, cadmium concentration had little effect on PGR when food was available but PGR declined with increasing concentration when food was limited, mainly due to a decrease in the reproductive rate. Although toxicant-related mortalities resulted in more food per surviving individual, Chandini (1988) concluded that cadmium stress suppressed the feeding rate of the survivors. The joint effects of density dependence and chemical stress may therefore differ among closely related species and may also differ with toxicant concentration, as observed in the marine polychaete Capitella sp., exposed to fluoranthene (Linke-Gamenick, Forbes & Sibly 1999). The joint effects were less than additive at low concentrations and more than additive at high fluoranthene concentrations, as reproduction was completely inhibited (Linke-Gamenick, Forbes & Sibly 1999).
From an increasing number of recent studies of the joint effects of density and chemical stress it is beginning to appear that population density may modify toxicant effects and often, although not always, toxicant effects are compensated for when resources are initially limiting, as in the less than additive model. In general, low levels of chemical stressors may paradoxically increase carrying capacity (density when PGR = 0) by reducing juvenile competition for resources, so more and perhaps better quality or fitter adults are produced. This is similar to the effects of predators and culling, which may directly improve the quality of prey populations (Cox 1993). However, toxicant exposure may result in survivors that are less tolerant to changing environmental conditions. More specifically, if less than additive effects are typical in the field, standard regulatory laboratory studies that are carried out at low densities may overestimate the effects of environmental chemicals. However, further studies over a wide range of chemical stressors and life-history types are needed before any general recommendations can be made.
This research was funded by Syngenta and AstraZeneca UK Ltd. Many thanks to Una Goggin (Syngenta) for her skilled technical assistance and to Professor Valerie Forbes and two anonymous referees for their valuable comments on the manuscript.