1. Introduction
An increased soil nutrient availability affects plant nutrient content, stoichiometry and biomass, and therefore indirectly the resource availability and quality to herbivorous insects [
1]. Many studies also show that insect performance is increased on nutrient rich plants and that insect numbers are typically higher in areas with nutrient rich soils [
2,
3,
4], even though this response varies between taxa in a community [
5,
6]. Beside the effects on herbivore numbers, theory suggests that a high nutrient availability should also destabilize population dynamics [
7,
8,
9]. Population dynamics are assumedly destabilized because high resource availability may allow for very high population growth rates in years when other conditions are suitable. In predator-prey system, theory suggests that time lags in the predator numerical response would similarly cause prey populations to fluctuate more at high productivity. The probability of unstable dynamics may however decrease due to interference among consumers [
10], or spatial heterogeneity [
11].
Many empirical studies have examined the effect from varying resource availability on consumer number and dynamics. These studies, however, are mostly done at a spatial and temporal scale where observations of population fluctuations are not relevant measures of population dynamics. There are a few exceptions involving model communities of limnic systems, and these studies provide little support to the destabilizing effect of nutrient enrichment. For instance, McCauley
et al. [
7] found that the presence of inedible prey reduced the destabilizing effect of nutrient enrichment on
Daphnia dynamics. At the same time, modeling has progressed such that we are now able to provide more precise predictions on the temporal dynamics for specific systems. This development has reduced the interest in the quite broad predictions provided by the original theories. Nevertheless, most real world systems will never be covered by realistic models and broad generalizations will be valuable for predicting the dynamics in these systems. For instance, the dynamics of terrestrial arthropod populations are seldom studied across gradients of nutrient enrichment at a scale where population level mechanisms become relevant, and at the same time, destabilizations due to nutrient enrichment are often implied in various studies involving pest insects [
4].
Islands with nesting seabirds provide one potential way to study effects of increased nutrient availability on arthropod population dynamics at a population level. Seabird nesting colonies often occur on islands and their feces often affect nutrient availability for plants across the whole island. Consequently, the effect on arthropod numbers and densities will occur through population level mechanisms rather than through movement processes, where individuals would congregate in resource rich areas. Moreover, the range of nutrient availabilities for islands with and without nesting seabirds is larger than is typically achieved in traditional fertilization experiments. In extreme cases, nutrient loads are so high that plant growth is compromised due to ammonium toxicity. A drawback from using seabird islands is that the treatment (fertilization) is not tightly controlled. However, by working along a range of seabird nesting activity, we believe that realistic effects from increased nutrient availabilities on plant-insect interactions can be observed. The natural experiment provided by seabird defecation provides an opportunity to test fertilization effects at a scale where traditional experiments are not possible.
In this study, we used the plant
Lythrum salicaria and its associated herbivore fauna (
Galerucella spp.,
Chrysomelidae) to quantify effects of increased soil nutrients on plant nutrient availability, herbivore density and herbivore dynamics, on shores and islands in the Stockholm archipelago (
Figure 1). Unfortunately, we were unable to separate two closely related
Galerucella species (
G. pusilla and
G. calmariensis) and the densities and dynamics are therefore joint measures of two species. Other studies in the same area, however, show that
G. calmariensis is by far the most common species in coastal sites, typically more than 80%. In a pot experiment, we also experimentally quantified stoichiometric variation in the herbivores and their performance along a gradient of plant nutrient content to relate the dynamic responses to theories of ecological stoichiometry (ES). The general pattern for heterotrophic organisms is that N and P contents show little variation with food nutrient content. ES suggests that such homeostasis may cause individual growth and reproductive rates to decrease when the resource stoichiometry deviates from the consumer stoichiometry. For this system, we know that pupal mass affects female fecundity in the following years suggesting a potential link between plant nutrient content, individual growth rates and population growth rates depending on the relationship between plant nutrient content and pupal mass.
2. Study Organism
Purple loosestrife,
Lythrum salicaria L. (
Lythraceae), is a perennial, insect-pollinated herb that is native to Eurasia and has been introduced to North America [
12]. It is abundant on islands and the shore-line in most areas of the northern Baltic Sea. Reproducing plants are on average 50 cm tall and produce one to several flowering shoots. Flower buds develop in leaf nodes in the upper part of the flowering shoot. In Fennoscandia,
L. salicaria flowers for six to eight weeks in July–August. The seeds mature six to eight weeks after flowering.
The beetles
Galerucella calmariensis L. and
G. pusilla L. (
Coleoptera:
Chrysomelidae) are monophagous herbivores on
L. salicaria [
13], feeding on leaves and flower buds. Their life cycle is univoltine and the adults overwinter in the litter [
14]. Around Stockholm, adults appear in May when the first host plant leaves appear. The adults feed on leaves and lay their eggs in batches on the stem and on the lower leaf surface. The larvae hatch seven to 10 days later, feed on leaves and flower buds for two to three weeks and then pupate in the ground. Previous studies suggest that larval growth is limited by both temperature and intraspecific density [
15,
16]. Similar to several other herbivore systems [
17,
18], low larval densities reduce pupal mass and thereby also reduce fecundity of females in the next generation.
4. Discussion
Population dynamic responses to nutrient enrichment are rarely observed in the field, because nutrients are seldom manipulated at a scale where population level mechanisms matter. Previous studies therefore mainly involve movement mechanisms where individuals may have aggregated in fertilized plots, rather than being born into them. In this study, we observed density fluctuations of two closely related herbivorous chrysomelid beetles on islands along a natural gradient in soil nutrient levels caused by defecation from nesting cormorants and roosting seabirds. These islands varied in soil nutrient contents, with an increased N content on active nesting islands and an increased P content on three out of five nesting islands (see also Ref. [
19]). Surprisingly, the variation in soil nutrient content did not translate into a corresponding variation in leaf N and P contents, contrary to previous measures on the same islands for a broader set of plant species [
5]. Nevertheless, the islands showed substantial variation in leaf nutrient contents and the
Galerucella number was in fact correlated with both leaf N content and plant height. On the other hand, and contrary to the expectation, the variability in
Galerucella densities did not increase with either leaf N or P contents. In fact, the coefficient of variation in
Galerucella densities tended to decrease with increased leaf P content. This effect is not likely due to a too short time series, as the pattern of fluctuations on control sites is similar to 13–15 year long time series collected from other coastal sites (Hambäck unpublished data). It should be remembered that densities are joint measures of two
Galerucella species and therefore that different dynamics in the two species may have affected the result. We cannot exclude this possibility, but note that one species (
G. calmariensis) normally dominates coastal sites in the area (>80%).
Nitrogen limitation and density responses to increased N are commonly observed in arthropod populations [
3,
6], whereas P limitation is more seldom observed, similar to our study. The mechanism underlying the observed response is less clear and the pot experiment provided no further information as pupal weights and survival did not respond to the nitrogen fertilization treatments, suggesting that other growth stages or interactions with other organisms may have been more important in the field. Previous studies on the same cormorant islands suggest strong responses in some taxa to cormorant nesting but also that some groups increase while other groups decrease (for a similar example in another system, see Ref. [
6]). Among herbivore groups,
Lepidoptera and aphids always had the highest density on active nesting islands, islands where the leaves from a set of collected plant species other than
L. salicaria had extremely high N and P contents [
5]. Herbivorous
Coleoptera, on the other hand, had higher densities on abandoned islands, where phosphorus but not nitrogen content was increased in the soil, than on active nesting islands. Why some plants increase in N or P content but not others is currently unknown, but may be related to either difference in plant physiological responses or to differences in microsite conditions. Among predatory groups, parasitic
Hymenoptera had the highest density on active nesting islands while web spiders had the highest density on abandoned nesting islands. In contrast, wolf spiders had the lowest density on active nesting islands. No doubt, responses by specific taxa may depend on responses in other groups and the strength of top-down and bottom-up processes may vary with fertilization [
2], and this may underlie the observed responses by herbivores on
L. salicaria.
The observation of reduced fluctuations in high-P conditions was surprising and contrasted with our expectations, and we are currently unable to explain the underlying mechanism. The pot experiment did indicate that larval survival may decrease with an increased P fertilization, while there was no effect on pupal mass. On the other hand, the leaf P gradient in the pot experiment was much larger than observed in the field, and the reduced larval survival was only observed at the highest fertilization levels. It is also notable that the larvae showed a fairly strict homeostasis in relation to N:C, P:C and N:P ratios, similarly to observations in other terrestrial insect systems [
19,
20]. An alternative mechanism for the reduced fluctuations in high-P conditions could be that some other life history trait is non-linearly related to leaf P content. If that is the case, a high quality resource may buffer population growth in years when conditions are otherwise poor, e.g., in the case of very dry weather.
To conclude, we found limited support for the suggestion that arthropod populations are destabilized by nutrient enrichment within the Lythrum-Galerucella system using islands with and without nesting cormorants. Even though egg densities of Galerucella increased in relation to both N content and plant height, fluctuations in density did not increase with a higher N and P. Whether this is a general pattern is unclear, but the results do suggest that increased density fluctuations at a population level should not necessarily be the outcome of fertilizing plants.