Multi-factor climate change effects on insect herbivore performance

Abstract

The impact of climate change on herbivorous insects can have far-reaching consequences for ecosystem processes. However, experiments investigating the combined effects of multiple climate change drivers on herbivorous insects are scarce. We independently manipulated three climate change drivers (CO2, warming, drought) in a Danish heathland ecosystem. The experiment was established in 2005 as a full factorial split-plot with 6 blocks × 2 levels of CO2 × 2 levels of warming × 2 levels of drought = 48 plots. In 2008, we exposed 432 larvae (n = 9 per plot) of the heather beetle (Lochmaea suturalis Thomson), an important herbivore on heather, to ambient versus elevated drought, temperature, and CO2 (plus all combinations) for 5 weeks. Larval weight and survival were highest under ambient conditions and decreased significantly with the number of climate change drivers. Weight was lowest under the drought treatment, and there was a three-way interaction between time, CO2, and drought. Survival was lowest when drought, warming, and elevated CO2 were combined. Effects of climate change drivers depended on other co-acting factors and were mediated by changes in plant secondary compounds, nitrogen, and water content. Overall, drought was the most important factor for this insect herbivore. Our study shows that weight and survival of insect herbivores may decline under future climate. The complexity of insect herbivore responses increases with the number of combined climate change drivers.

Keywords: Chrysomelidae; FACE experiment; climaite; condensed tannins; multiple climate change drivers; multitrophic interactions; plant secondary metabolites.

Figure 1

Study site and study organism. (A) Aerial view of the Climaite experiment, showing the 12 octagons with drought and warming curtains in action. Curtains were drawn over plots for illustrative purposes only. (B) A single octagon, surrounded by CO₂ tubes, split into four subsectors where factorial combinations of drought and warming were applied. (C) Larva and (D) imago of the heather beetle, Lochmaea suturalis (Thomson, 1866) feeding on Calluna vulgaris L. (Ericaceae). Image credits: (A) T. N. Mikkelsen; (B) D. Gladbach; (C) and (D) C. Scherber.

Figure 2

Temporal dynamics of climate change treatments applied in this experiment. (A) Nighttime air temperature at a height of 20 cm above ground in warmed (red) and unwarmed plots (blue); lines show a non-parametric smoothing spline ± 1 SE (B) Daytime air CO₂ concentration (ppm) in elevated CO₂ octagons (red) and ambient CO₂ octagons (blue); lines are from a locally weighted polynomial regression smoother; (C) Soil water content (percent) from TDR probes at 0–20 cm depth in prolonged drought plots (grey) and non-drought plots (black); lines are exact mixed-effects model predictions for each day; the grey shaded area in (C) shows the period in which the drought treatment was applied. Bars in (C) show precipitation events, red bars indicate the precipitation events that were excluded in the drought treatment.

Figure 3

Weight (mg) of Lochmaea suturalis larvae over time for plots with elevated treatments (N = 24 per time point; filled circles, solid lines) and controls (N = 24 per time point; open circles, dashed lines). (A) Ambient versus Elevated CO₂; (B) No Drought versus Drought; (C) Warming versus No Warming. Lines in (A) and (B) show model predictions from minimal adequate mixed-effects models; lines in (C) derived from a mixed model with all fixed effects terms included. The time:CO₂ effect in (A) was significant in interaction with drought (P = 0.001; see Table 1); the drought:time effect in (B) had P = 0.003; warming (C) had no significant effect on weight.

Figure 4

Kaplan–Meier survivorship of Lochmaea suturalis larvae over time for plots with elevated treatments (N = 24 per time point, solid lines) and ambient plots (N = 24 per time point, broken lines). Grey shaded areas show 95% confidence intervals of the Kaplan–Meier estimator. Significance of interactions with time: (A) P = 0.057; (B) P < 0.0001. (C) Warming was only significant in a three-way interaction with CO₂ and drought (P = 0.019).

Figure 5

Effects of combinations of climate change treatments on (A) mean larval weight and (B) mean larval survival. Black bars indicate the two most extreme conditions applied in this experiment – either “fully ambient” or “full climate change”, that is, a combination of all three climate change treatments. Sample sizes are given in brackets.

Figure 6

Structural equation model showing how treatment effects of CO₂, warming and drought (bottom) affect survival and weight (top), mediated via changes in plant chemistry (latent variable) and soil/plant water content (latent variable).Variables e3-e8 are error terms. For clarity, standardized path coefficients are shown. All variables with error terms were scaled before analysis. CN, carbon/nitrogen ratio; Tannins, condensed tannins; TDR, soil water content; LWC, leaf water content.

Figure 7

Weight (open circles) and survival (filled circles) of beetle larvae (N = 48) as a function of the number of climate change drivers, scaled to [0;1] using a ranging transformation. Survival (solid line) and growth (dashed line) decreased significantly (P < 0.05) with the number of climate change drivers. Lines show local smoothing splines (for illustrative purposes only).