Effects of air pollution on natural enemies of the leaf beetle Melasoma lapponica

Authors


Elena L. Zvereva, Section of Ecology, Department of Biology, University of Turku, FIN-20014 Turku, Finland (fax 358 2 3336550; e-mail ELENA.ZVEREVA@utu.fi).

Summary

1.  Air pollution might have differential effects on herbivores and their natural enemies, thus changing population dynamics. Therefore, from 1993 to 1998 we studied mortality caused by parasitoids and predators to the willow-feeding leaf beetle Melasoma lapponica in the impact zone of the Severonikel nickel–copper smelter (Kola Peninsula, north-western Russia).

2.  Densities of M. lapponica were very low at clean forest sites (below five beetles per 10-min count) but higher in polluted areas (10–340 beetles per count). There were, however, variations between study years.

3.  Egg predation, mainly by syrphid larvae and zoophagous bugs, was higher at relatively clean sites (55·3%) than at polluted sites (22·2%). Similarly, predation on larvae by zoophagous bugs and wood ants was higher at clean sites (68·4%) than at polluted sites (32·9%).

4.  In contrast to predation, mortality caused by the parasitoid flies Megaselia opacicornis (Phoridae) and Cleonice nitidiuscula (Tachinidae) was lower at clean sites (12·3%) than at polluted sites (35·3%). Total parasitism levels increased significantly with pollution load.

5.  Total mortality caused by natural enemies was higher at clean sites (93·7%) than at polluted sites (79·4%) due to higher predation rates, which may partly explain increased leaf beetle density within the smelter's impact zone. The effects of predators in clean forests were confirmed by the extinction of adults of M. lapponica introduced to one of the forest sites.

6.  Although some individual sources of mortality appeared to be density dependent (direct or inverse), the joint effect of all natural enemies was not.

7.  Our data show that a decrease in predation can contribute to increased leaf beetle density at polluted sites. However, the overall effects of natural enemies in this case were not sufficient to account for all density variations between sites. To our knowledge this is the first study to assess how pollution affects the partitioning of mortality in herbivorous insects between predators and parasitoids.

Introduction

Pollution affects populations of herbivorous insects in different ways, and can cause both increase and decrease of their numbers (Alstad, Edmunds & Weinstein 1982; Riemer & Whittaker 1989; Kozlov 1990, 1997; Heliövaara & Väisänen 1993). Herbivore outbreaks are often associated with polluted areas (Baltensweiler 1985; Führer 1985; Zvereva, Kozlov & Neuvonen 1995a), and some authors attribute this association to a decrease in the activity of parasitoids (Edmunds 1973; Heliövaara, Terho & Koponen 1982; Selikhovkin 1985) or predators (Petal 1980; Kataev, Golutvin & Selikhovkin 1983). However, the data supporting this hypothesis remain surprisingly scarce (Kozlov 1990), and some well-designed field studies have failed to detect any effect of pollution on the natural enemies of herbivores (Chlodny 1976; Heliövaara & Väisänen 1986; Koricheva 1994; Koricheva, Lappalainen & Haukioja 1995). Furthermore, the majority of studies have involved a single species, or closely related groups, of parasitoids or predators, whereas herbivores are usually attacked by several enemies. Each may respond to pollution in its own way (Villemant 1980; Selikhovkin 1985).

Between 1993 and 1998 we studied population dynamics of the willow-feeding leaf beetle Melasoma lapponica L. (Coleoptera: Chrysomelidae) along two pollution gradients originating at the Severonikel nickel–copper smelter in Russia, which is one of the largest sources of aerial emissions in Europe (Anonymous 1994). Melasoma lapponica was infrequent in the undisturbed habitats of the Kola Peninsula but abundant in the area surrounding the smelter (Zvereva, Kozlov & Neuvonen 1995a; Zvereva, Kozlov & Haukioja 1997b). In our earlier studies (Zvereva, Kozlov & Neuvonen 1995a; Zvereva, Kozlov & Haukioja 1997b) we suggested that this distribution could be explained partly by higher activity of leaf beetle enemies in clean forest sites, but at that time we did not have enough data to verify this hypothesis statistically. Within the impact zone of the smelter we observed density fluctuations that were not synchronized between study sites. Outbreak densities of M. lapponica (64–340 beetles per 10-min count) were recorded in 1989 close to the smelter (E. Zvereva & M. Kozlov, personal observations), and at two moderately polluted sites situated 5 km north and 14 km south of the smelter from 1993 to 1996 (Zvereva, Kozlov & Neuvonen 1995a; Zvereva, Kozlov & Haukioja 1997b). During the outbreaks, leaf beetles caused nearly complete defoliation of Salix borealis (Fries.) Nazar., the species most preferred by M. lapponica. A population decline following outbreaks could be at least partly due to the decrease in host-plant quality after severe defoliation (Zvereva, Kozlov & Haukioja 1997a; Zvereva et al. 1997). The role of natural enemies in these population fluctuations remains unknown.

The aim of this study was to assess the contribution of natural enemies to (i) differences in population density of M. lapponica between unpolluted habitats and the zone impacted by the Severonikel smelter, and (ii) density variations within the impact zone. In addition, we aimed to compile a list of the natural enemies of M. lapponica for our study area and to collect the quantitative data necessary to reveal any direct effects of pollution on parasitism and predation of M. lapponica. As mortality caused by predators and parasitoids often depends on host population densities, density variations of M. lapponica between study sites were accounted for in our study.

Materials and methods

Study sites and pollution loads

The Severonikel nickel–copper smelter is located in the city of Monchegorsk on the Kola Peninsula, north-west Russia (67°56′ N, 32°49′ E). The area surrounding this smelter is one of the most extreme examples of terrestrial pollution in the boreal forest zone (Kozlov & Haukioja 1995). The smelter produces aerial emissions consisting mostly of SO2 and heavy metals (Ni, Cu, Co). The annual amount of SO2 emitted in 1993–98 was 1·00–1·61 × 108 kg; emissions of nickel were 3·2–4·0 × 106 kg (Gipronikel Institute, official data for 1993 and 1994; Anonymous 1996; V. Barkan, personal communication). The distribution of SO2 and heavy metals follow similar spatial (Barkan 1993) and temporal patterns (Kozlov et al. 1995).

Sulphur dioxide and heavy metals emitted by the smelter have caused widespread destruction of soils and vegetation (Kryuchkov 1993; Kozlov & Haukioja 1995; Tikkanen & Niemelä 1995). Near the smelter coniferous forests have vanished (Table 1) and the most polluted habitats have been transformed to industrial barrens (Kryuchkov 1993).

Table 1.  Characteristics of study sites
Site codeDistance from the smelter (km)Direction from the smelterForest stateFoliar Ni in Salix borealis (µg g−1)*
  • *

    Concentrations of foliar nickel are averaged for the years 1993–98 (mean±S.E.).

16N16North-eastWeakened44·5±11·0
11N11North-eastDamaged56·5±9·6
5N5NorthVanished127·3±13·4
1N1North-westVanished177·8±33·0
1S1SouthVanished184·0±24·0
5S5SouthVanished146·9±23·6
14S14SouthDamaged123·8±16·1
20S20SouthDamaged75·5±9·2
29S29SouthWeakened30·6±3·7
36S36SouthWeakened23·9±0·6

Ten sites (nine in 1993 and 1994) were chosen 1–16 km north-east and 1–36 km south of the smelter along the road from Murmansk to St Petersburg, at least 50 m from the road to minimize the effects of traffic-borne pollutants (Table 1). However, to make the study feasible we used a smaller number of sites for some of the labour-intensive experiments. Subsequently, the sites were named with the distance and direction from the smelter (e.g. 16N for the site located 16 km north of the smelter). The larger extent of the impact zone and, consequently, of the study area to the south, compared with the north (Table 1), results from the spread of aerial pollution by predominantly north winds during the summer. Study plots were set up in localities representing different stages of pollution-induced deterioration of spruce-dominated forests (Koroleva 1993) (Table 1).

Pollution loads at the study sites were evaluated by measuring foliar concentrations of nickel in S. borealis. Fifty mature leaves were randomly sampled at the end of July every year from 1993 to 1998. Leaves were taken from the mid-crown of five bushes in each study site. Unwashed leaves were dried at +80 °C for 12 h, then ground in a plant mill. The concentration of nickel was determined in bush-specific samples by X-ray fluorescence (Spectrace 5000 spectrometer, Tracor X-Ray, Naarden, the Netherlands; resolution 5 µg g−1) in 1993, 1994, 1997 and 1998, and by ARL 3580 vacuum ICP-emission spectrometer (Applied Research Laboratories, EcuBlens, Switzerland) in 1995 and 1996 (for a description of the intercalibration procedure, see Kozlov et al. 1995), and averaged to give site-specific values. Due to technical problems, some site-specific data were missed.

Biology and density of the leaf beetle

Melasoma lapponica is a medium-sized (5–8 mm length) black-and-red patterned beetle. Adults hibernate in the soil and appear on willow bushes at the time of leaf flush, which in our study area occurs around mid-June. Larvae feed on mature leaves from early July to early August and pupate on leaves. Beetles of the next generation emerge in mid-August, feed for several days and then dig into the soil for overwintering. In the Kola Peninsula, M. lapponica was recorded on six willow species but was most abundant on S. borealis, which is common over the entire study area, including heavily polluted sites (Zvereva, Kozlov & Neuvonen 1995b).

Densities of M. lapponica were estimated every year from 1993 to 1998 at the end of June by 10-min counts of overwintered beetles, three counts per site (for more details, see Zvereva, Kozlov & Neuvonen 1995a,b; Zvereva, Kozlov & Haukioja 1997b). These counts correlated positively with the numbers of both beetles (1994 data, r = 0·99, n = 10 sites, P < 0·0001) and larvae (r = 0·95, n = 10 sites, P < 0·0001) recorded within the 2 × 25-m plot (mean of four replicates), and thus served as a reliable estimate of population density of M. lapponica. The larval density (LD; individuals m−2) was calculated from the results of 10-min counts of beetles (BC; individuals) as follows: LD = 0·0235BC + 0·0122 (linear regression model, F1,9 = 10·2, P < 0·0001).

Search for parasitoids and predators

All evidence of predation on M. lapponica was recorded and predators were collected whenever possible. Collected specimens were preserved in alcohol or pinned. Each year approximately 100 egg batches and 500 prepupae and pupae were collected to assess parasitism rates.

Egg mortality

Egg predation was evaluated in 1998 in six study sites. Egg batches of M. lapponica were collected at the time of mass oviposition (the first 10 days of June), 12–40 batches per site depending on beetle density. Normal and damaged (empty chorions) eggs were counted under a dissecting microscope. Eggs of the syrphid fly on leaf beetle batches were also counted. We considered a batch with even a single syrphid egg to be completely eliminated, as in a laboratory test one fly larva (which hatched earlier than the beetle larvae) ate all beetle eggs in the batch in 1 day.

Enemy exclusion experiment

In order to assess larval mortality due to predation, we conducted an enemy exclusion experiment in six study sites. In each site, we selected five individuals of S. borealis (each situated at least 2 m from the nearest conspecific neighbour) from which we removed all adults and egg batches of M. lapponica. At the time of egg hatching (7 July 1997), two branches of each plant were each infested with 20 first instar larvae collected at site 14S. All larvae in a group were descended from different batches to minimize possible effects of parental genotypes. The number of larvae per branch corresponded approximately to an average number of larvae hatching from one egg batch (Zvereva, Kozlov & Neuvonen 1995a). One of the infested branches was covered with a mesh bag (‘sleeve cage’) so that neither vertebrate nor invertebrate predators could get in; the other branch was left as an uncaged control. The experimental plants were visited each third day to remove beetle immigrants and to eliminate new batches in uncaged controls, and to record dispersal of uncaged larvae. After 20 days we collected the survivors, most of which had pupated by this date. The difference in the numbers of caged and uncaged larvae was attributed to the activity of predators. Surviving pupae were kept in the laboratory until beetles hatched, to estimate parasitism rate.

Introduction experiment

On 3 July 1997 we introduced adult beetles, collected at site 14S, to four study sites: 1N, 5N, 14S and 36S. Beetles were marked by a spot of coloured nail polish on their elytra. Ten beetles with identical marks were placed on small (c. 50 cm in height) S. borealis bushes, with five replicates per site. The original and surrounding willow bushes were searched every day during 8 days after the release to record the location of marked beetles.

Pupal mortality

Mortality due to parasitism was determined annually in all study sites. All prepupae and pupae were collected from five to seven randomly chosen S. borealis bushes in each study site at the time of mass pupation (end of July). This yielded 50–400 prepupae and pupae per site each year (6925 individuals in total), which were kept in the laboratory until adult beetles hatched. At this time beetle individuals were sorted to the following categories: alive (beetle hatched); containing large parasitoid pupae; containing small parasitoid pupae; and dead for unknown reasons.

Total pupal mortality was assessed only once on 14–16 August 1997. All alive (beetles hatched) and dead pupae were counted on five small to medium-sized (height 0·5–1·5 m) bushes in each of the 10 sites. The tip of the pupal abdomen is tightly attached to the leaf and remains on the leaf if a predator removes the pupa, thus allowing estimation of predator activity. Pupae predated by bugs usually bear no visible signs of damage, and therefore those pupae that died of unknown reasons were considered to have been killed by sucking predators. Site-specific rates of pupal mortality due to predation were estimated as the difference between total mortality and total parasitism rate.

Data analysis

All data were averaged for site-specific values. Total pupal mortality and larval mortality due to predation were distributed normally; data on parasitism by Cleonice nitidiuscula (Zett.) were square-root transformed; beetle density and foliar nickel concentrations were log-transformed, prior to calculation of the Pearson correlation coefficient (SAS corr procedure). We failed to find appropriate transformations for parasitism by Megaselia opacicornis Schmitz and total parasitism, and therefore non-parametric methods (Spearman correlation and Kruskal–Wallis test) were used in the analysis of these data (SAS Institute 1990).

Results

Parasites and predators of m. lapponica

Three predatory groups, wood ants (Formica spp.), bugs (four species) and larvae of the syrphid fly Parasyrphus nigritarsus (Zett.), along with two species of parasitoid flies, caused most mortality in populations of M. lapponica. Crab spiders, mites and ladybird beetle larvae were observed to prey upon M. lapponica only occasionally (Table 2). We have not observed birds feeding on M. lapponica.

Table 2.  Predators and parasitoids of Melasoma lapponica in the Monchegorsk region
Predator/parasitoid Attacked host stage*Abundance
  • *

    E, eggs; L, larvae; P, pupae; I, imago.

  • Occasional, < 5 individuals collected; low, 6–50 individuals collected; high, > 50 individuals collected.

  • Parasitoid.

Arachnida
 Araneae
   Thomisidae
    Xysticus audax (Schrank)L, IOccasional
    X. obscurus CollettL, IOccasional
 Acari
   Erythraeidae
    Balaustium sp.LOccasional
Insecta
 Hemiptera
   Miridae
    Lygocoris rugicollis (Fall.)E, L, PHigh
    Psallus aethiops (Zett.)E, L, PHigh
    Orthotylus boreellus (Zett.)E, L, PHigh
   Anthocoridae
    Anthocoris nemorum (L.)E, L, PHigh
 Coleoptera
   Coccinellidae LOccasional
 Diptera
   Syrphidae
    Parasyrphus nigritarsus (Zett.)E, LHigh
   Phoridae
    Megaselia opacicornis SchmitzPHigh
   Tachinidae
    Cleonice nitidiuscula (Zett.)LHigh
 Hymenoptera
   Formicidae
    Formica spp.L, P, IHigh
   Pteromalidae
    Schizonotus sieboldy (Ratz.)PLow

Wood ants caused high larval and adult mortality of the leaf beetle in some forest sites; on most of the bushes infested with larvae of M. lapponica we found foraging wood ants. In the study area we recorded 10 species of wood ants but leaf beetle mortality could not be attributed to a particular ant species.

Both nymphs and adults of four bug species were observed preying upon M. lapponica. In several cases, two to three bugs managed to enter a sleeve cage, where they eliminated all 20 larvae. Bug nymphs and adults successfully attack the larvae from the ventral side of the abdomen, which has no defensive glands.

Larvae of the syrphid fly P. nigritarsus, a specialist predator of chrysomelid beetles (Schneider 1953), were repeatedly found preying upon leaf beetle eggs and larvae. Flies laid eggs on (or sometimes near) leaf beetle egg batches (1–4 per batch). Hatching larvae ate all beetle eggs in a batch, before moving to the next batch. In some sites these larvae eliminated up to 45% of beetle egg batches. Fly larvae also fed on leaf beetle larvae, but this level of mortality due to P. nigritarsus could not be assessed quantitatively in the field.

Females of the tachinid fly C. nitidiuscula infested first or second instar larvae of the leaf beetle, and the tachinid larvae pupated when the host larvae had completed feeding, killing the host at the stage of prepupa. Females of phorid fly Megaselia opacicornis mostly laid eggs on prepupae; larvae developed in the pupae of M. lapponica and pupated before the beetles hatched from intact pupae; one to five (usually two) parasitoid larvae developed in each host pupa.

The chalcidoid wasp Schizonotus sieboldy (Ratz.), a specialist parasitoid of the genus Melasoma, as well as two other species of hymenopteran parasitoids, Mesochorus confusus Holmgr. (Ichneumonidae) and Aspilota sp. (Braconidae), were reared from pupae of M. lapponica. It is presumed that the two latter species parasitized flies developing in the leaf beetle (M. Koponen, personal communication; Gauld 1991). No egg parasitoids were recorded in spite of intensive collecting of egg batches.

Pupal mortality for unknown reasons (including death from diseases) was low (average 3·3%). About 2% of beetles died due to infestation by Penicillium sp. (V. Serebrov, personal communication). Although Penicillium is usually a saprophytic fungi, it can sometimes also be a pathogen (Sen, Jolly & Jammy 1970). The pathogenic nature of Penicillium found in our beetles was confirmed experimentally (V. Glupov, personal communication).

Mortality due to predators

Total egg predation was higher in relatively clean sites (55·3%) than in polluted sites (22·9%) (Fig. 1). Larvae of P. nigritarsus were the principal egg predators, responsible for 33·3–79·9% of egg mortality. In relatively clean forest sites 45·5% of leaf beetle batches contained syrphid eggs, whereas in polluted sites the infestation rate was lower (13·1%). However, egg predation correlated with neither pollution load (rs = 0·03, n = 6 sites, P = 0·96) nor host population density (rs = −0·38, n = 6 sites, P = 0·46).

Figure 1.

Mortality of different stages of Melasoma lapponica caused by predators and parasitoids in relatively clean (16N and 29S) and heavily polluted (1N and 1S) sites with low leaf beetle density. Bars indicate standard errors.

Larval predation rates (estimated from the enemy exclusion experiment) in two relatively clean sites was about twice as high as in four sites situated in the impact zone (68·4% and 32·9%, respectively; χ2 = 3·43, d.f. = 1, P = 0·06). At the same time, mortality did not vary either among polluted sites (χ2 = 3·68, d.f. = 3, P = 0·30) or between relatively clean sites (χ2 = 0, d.f. = 1, P = 0·99). Predation rate did not correlate with pollution load (rs = −0·49, n = 6 sites, P = 0·33) but decreased with an increase in the site-specific densities of M. lapponica (Fig. 2).

Figure 2.

Larval mortality of Melasoma lapponica (site-specific means based on five bush-specific values each) due to predation in the enemy exclusion experiment (rs = −1·00, n = 6, P = 0·005) plotted against leaf beetle density (measured by numbers of M. lapponica beetles per 10-min count). Bars indicate standard errors.

Total pupal mortality determined after beetle emergence in 1997 did not correlate with pollution load (r = 0·44, n = 8 sites, P = 0·28) but demonstrated a positive correlation with M. lapponica population density (r = 0·80, n = 10, P = 0·005). Pupal mortality due to predation was density independent (rs = 0·36, n = 9, P = 0·34).

Adult beetles introduced to the clean forest site (36S) disappeared 2 days after release, whereas in the impact zone 38–44% of beetles were recorded on this date, and 8–16% were still found 8 days later (Fig. 3). The difference in survival between clean site 36S and the polluted sites was significant the day after the beetle release (χ2 = 10·1, d.f. = 1, P = 0·0015) but no differences in beetle number were found among the polluted sites on any day of the experiment (χ2 = 0·11–3·00, d.f. = 2, P > 0·20).

Figure 3.

Decrease in numbers (means based on five bush-specific values) of adult beetles of Melasoma lapponica after introduction to four study sites.

Mortality due to parasitoids

Parasitism of M. lapponica varied between study sites in most of the study years (χ2 = 14·9–26·5, d.f. = 8–9, P = 0·09–0·0009); between year variation was also significant (χ2 = 14·1, d.f. = 5, P = 0·02). If the data from 6 study years were combined, total parasitism was positively correlated with the pollution load (rs = 0·26, n = 52, P = 0·04; Fig. 4); mortality from parasitoids in the impact zone was on average three times as high as in relatively clean sites (35·3% and 12·3%, respectively; χ2 = 11·2, d.f. = 1, P = 0·0008). Higher parasitism in the polluted sites (14·1 vs. 2·9%; χ2 = 3·43, d.f. = 1, P = 0·06) was also detected in the enemy exclusion experiment.

Figure 4.

Total parasitism in Melasoma lapponica (annual site-specific means based on five to seven bushes each) in relation to pollution load (correlations calculated for the 6-year data: rs = 0·29, n = 52, P = 0·04).

Parasitism by Megaselia opacicornis was density dependent (P = 0·05) in 4 of 6 study years (Fig. 5a); parasitism by C. nitidiuscula was density independent in all study years. However, when the data from all 6 years were combined, parasitism by C. nitidiuscula demonstrated non-linear density dependence (Fig. 5b); exclusion of the lowest (less than five beetles per count) host densities resulted in an inverse density dependence (r = −0·40, n = 43, P = 0·008), whereas at densities below 20 beetles per count we failed to detect density dependence (r = 0·26, n = 32, P = 0·15). Total mortality due to parasitism was density independent (Fig. 5c), except for the year 1998 when we detected direct density dependence (r = 0·87, n = 10, P = 0·001).

Figure 5.

Mortality of Melasoma lapponica in relation to leaf beetle population density (measured by numbers of M. lapponica beetles per 10-min count). (a) Caused by Megaselia opacicornis (rs = 0·43, n = 58, P = 0·0009); (b) caused by Cleonice nitidiuscula (r = −0·08, n = 58, P = 0·55); (c) due to total parasitism (rs = 0·19, n = 58, P = 0·15).

Total mortality caused by parasitoids and predators

The data on both larval predation and pupal mortality due to parasitoids and predators obtained in 1997 enabled the calculation of mortality of M. lapponica due to the complex of natural enemies. This mortality correlated neither with pollution load (rs = 0·00, n = 5 sites, P = 1·00) nor with M. lapponica population density (rs = 0·30, n = 5, P = 0·62).

To exclude possible confounding effects of density dependence in assessing the effect of pollution, we compared two heavily polluted sites (1N, 1S) with two relatively clean sites (16N, 29S) with similar densities of M. lapponica (means for the 5 study years: 14·9 and 3·4 beetles per 10-min count, respectively; F3,16 = 3·09, P = 0·10). Predation of eggs, larvae and pupae was lower at polluted sites, whereas parasitism was either higher at polluted sites (phorid fly) or similar at clean and polluted sites (tachinid fly) (Fig. 1). Summing up these mortality factors, we calculated that total mortality caused by natural enemies was higher at clean sites (93·7%) than at polluted sites (79·4%). Thus, if we do not take other mortality sources into account, the estimated number of hatching beetles at polluted sites will be 3·25 times as high as at clean sites.

Discussion

Natural enemies of m. lapponica

The list of natural enemies of M. lapponica generally includes the same groups of predators and parasites as for other leaf beetles (Kanervo 1946; Topp, Beracz & Zimmerman 1989; Topp & Bell 1992; Rank 1994). However, there are two exceptions. The tachinid fly C. nitidiuscula, which caused up to 66% mortality of M. lapponica in our study area (Zvereva, Kozlov & Neuvonen 1995a), was previously considered a very rare species reared from only one host, Melasoma saliceti (Richter & Zvereva 1996). Another parasitoid, the phorid fly Megaselia opacicornis (erroneously reported as Megaselia rubricornis Schmitz by Zvereva, Kozlov & Neuvonen 1995a; Zvereva, Kozlov & Haukioja 1997b) was earlier known by only a few specimens (R.H.L. Disney, personal communication). The biology of this fly is similar to Megaselia rubricornis, which infested 30–50% of pupae of Melasoma aenea in Finland (Kanervo 1946).

It is difficult to be certain that birds do not prey upon M. lapponica. However, there are observations of avoidance or rejection of beetle larvae (Topp & Bell 1992) and adults (Lundvall, Neuvonen & Halonen 1998) by insectivorous birds, presumably due to chemical defences. Moreover, bird numbers in the impact area of the Severonikel smelter were only 10–20% of those in the background areas (Gilyazov 1993). In summary, we have no reason to assume that bird predation significantly affected populations of M. lapponica.

Small insectivorous mammals living in the study area, such as shrews (Sorex araneus and S. caecutiens) and some voles (e.g. Clethrionomus glareolus), are common in unpolluted forest habitats, but their densities decline strongly with an increase in pollution (Kataev 1989; Kataev, Suomela & Palokangas 1994). Although these rodents have the opportunity to prey on hibernating adults of M. lapponica, there are no records of mammals predating chemically defended leaf beetles (Kanervo 1946; Topp & Bell 1992; Rank 1994; but see Schwenke 1974). It is unlikely that mammals are important natural enemies of M. lapponica in the study area.

Effects of pollution

With some exceptions (Krzysztofiak 1991; Koricheva, Lappalainen & Haukioja 1995), the decrease in populations of wood ants in polluted areas has been reported frequently (Petal 1980; Chlodny & Styfi-Bartkiewicz 1982; Kataev, Golutvin & Selikhovkin 1983; Stary & Kubiznakova 1987). In our study area, the abundance of Formica spp. (as estimated by pitfall traps) was higher in relatively clean habitats than in the impact zone of the smelter (M. Kozlov, personal observation). This supports the view that there is higher predation mortality of M. lapponica in forest sites.

Densities of zoophagous bugs and syrphid flies may increase in moderately and even heavily polluted areas (Chlodny & Styfi-Bartkiewicz 1982). Although quantitative studies of the abundance of bugs preying upon M. lapponica were not carried out along the Severonikel pollution gradient, our observations indicate that bug populations do not decline with an increase in pollution. On the other hand, the abundance of P. nigritarsus was higher in the clean localities, as indicated by a higher proportion of beetle egg batches containing syrphid eggs. Thus, lower predation of M. lapponica in the polluted areas may be explained mainly by the decrease in wood ant and P. nigritarsus populations.

In contrast to predation, parasitism was lower in clean forest sites than in polluted habitats. This indicates that pollution does not cause any adverse effects on the two studied parasitoid flies but may even favour them, probably via an increase in host population densities. A positive association between parasitism rate and pollution load has not been reported before (cf. Riemer & Whittaker 1989; Kozlov 1990; Heliövaara & Väisänen 1993), possibly because the majority of studies have been concerned with hymenopteran parasitoids. Parasitic hymenopterans seem to be sensitive to pollution (Riemer & Whittaker 1989; Kozlov 1990; Heliövaara & Väisänen 1993) whereas many fly groups can withstand environmental contamination and frequently are even more abundant in polluted areas than in clean sites (Dabrowska-Prot 1980; Zvereva 1993).

Thus, predators and parasitoids of M. lapponica responded differentially to pollution, but generally predators were responsible for more mortality and the cumulative effect was therefore determined by the activity of predators. The low population density of M. lapponica in unpolluted localities may be maintained by severe predation pressure. This suggestion was confirmed by the failure of the introduced M. lapponica in the clean forest site. Lower predation in the entire impact zone favours an increase in leaf beetle densities; however, in some of polluted sites density remained low, suggesting that other factors must limit population growth (such as direct toxicity of pollutants or host-plant quality). Within the impact zone predation did not depend on the pollution load. This may indicate that predators are affected by pollution-induced environmental deterioration (e.g. forest decline and destruction of ground vegetation) rather than by toxicity of pollutants.

Effects of host (prey) density

The population density of M. lapponica differed significantly between our study sites, indicating that density-dependent processes could interfere with the effects of pollution on leaf beetle mortality due to natural enemies. Direct density dependence, inverse density dependence and density independence are all commonly observed in predator–prey and parasitoid–host systems, although different authors refer to different frequencies of these cases (Lessels 1985; Stiling 1987; Walde & Murdoch 1988; Cappuccino 1991). In our study, we found all these types. Leaf beetle infestation by Phoridae showed direct density dependence. Infestation by Tachinidae at high host densities and larval predation showed inverse density dependence. Infestation by Tachinidae at low host densities, egg and pupal predation, total parasitism and cumulative mortality were density independent.

In general, patterns of population dynamics are the results of both density-dependent and density-independent processes (Hassell 1986), and our data confirm that pooling of different mortality factors may produce misleading results. In addition, our results indicate how study of a complex pollution problem can help to test ecological theory (Ormerod, Pienkowski & Watkinson 1999). In our system, density-independent parasitism resulted from the combination of the density-dependent parasitism by Phoridae with the inversely density-dependent parasitism by Tachinidae. Similarly, density-independent predation resulted from the combination of inverse density-dependent predation on larvae and density-independent predation on eggs and pupae. Moreover, density-independent egg predation may have determined the total mortality of M. lapponica, making detection of density dependence on older stages a difficult task (Van Hamburg & Hassell 1984).

Lessels (1985) suggested that, under some circumstances (e.g. egg or time limitation in parasitoid), density dependence may be dome-shaped. The pattern of density dependence in the parasitism rate of C. nitidiuscula is very close to the domed relationship (Fig. 5b), being highest at moderate beetle densities and demonstrating inverse density dependence at high densities. A similar pattern may be suggested for another tachinid fly, Doryphorophaga doryphorae, a parasitoid of the Colorado potato beetle Leptinotarsa decemlineata. This parasitoid demonstrated direct density dependence in tomato fields with relatively low densities of the beetle (Latheef & Harcourt 1974), and inverse density dependence in potato fields with higher beetle densities (Harcourt 1971).

Various mechanisms can underlie the observed differences in spatial density dependence. For the parasitic flies, our data fit the conclusion by Walde & Murdoch (1988), who found more frequent density dependence in small parasitoids (like Phoridae) and inverse density dependence in larger parasitoids (like Tachinidae), and attributed the effect of size to different migratory abilities of small and large insects. According to the predictions of Lessels (1985), egg or time limitation in a parasitoid, or imperfect information on patch quality, constrains parasitoid foraging which may result in inverse density-dependent parasitism. Inverse density dependence in larval predation, observed both for M. lapponica and another leaf beetle, Plagiodera versicolora (Crowe 1995), may result from the higher efficiency of larval chemical defence against the generalist predators at higher larval densities. The same mechanism may explain lower predation rates recorded for aggregated pupae (six to eight on one leaf) of M. lapponica compared with the dispersed pupae (one per leaf) (E. Zvereva, personal observation).

Although we discovered various types of density dependence for different groups of natural enemies or different stages of the leaf beetle, cumulative mortality due to the complex of natural enemies demonstrated no spatial density dependence and, thus, observed among-site variation in mortality due to natural enemies may be attributed mainly to the effects of pollution. On the other hand, we could not exclude the possibility that high abundance of M. lapponica in the impact zone depends on factors other than enemy exclusion (e.g. habitat changes), and low densities in clean forest sites are determined by inverse density dependence of predation on larvae. However, the comparison of polluted and clean sites with low densities of the leaf beetle (Fig. 1) confirmed that the observed difference in predation between clean forests and the impact zone was related to pollution effects rather than to host density.

Thus, the detrimental effects of pollution on the predators of M. lapponica can contribute to a higher abundance of this herbivore in polluted habitats. However, within the impact zone neither pollution load nor density-dependent mortality due to natural enemies explained the observed density variations of M. lapponica, suggesting the existence of other factors governing population density fluctuations of the leaf beetle in polluted territories.

Acknowledgements

We are grateful to V. Zverev, E. Melnikov and A. Blashkevich for assistance with field work, and to A. Bakhtiarov for conducting the analysis of metal contamination. We thank R. Disney, R. Jussila, I. Kerzhner, V. Kipyatkov, S. Koponen, M. Koponen, S. Kuznetsov, J. Makol, V. Richter, V. Serebrov, A. Sharkov and I. E. Sääksjärvi for identification of parasitoids and predators. S. Koponen, S. Neuvonen, M. Sabelis and an anonymous referee made valuable comments on an earlier draft of the manuscript. The study was financially supported by Maj and Tor Nessling Foundation and the Academy of Finland.

Received 16 November 1998; revision received 27 November 1999

Ancillary