The shape of the predator biomass distribution affects biological pest control services in agricultural landscapes
Data files
Sep 10, 2020 version files 109.06 KB
-
FE-2019-01367_FuncEcol_files.xlsx
Abstract
1. Understanding how community composition of service-providing organisms affects ecosystem functioning is a key challenge in ecology. Although it has been proposed that taxonomic diversity and functional traits mediate this relationship, how several facets of community structure affect the delivery of key ecosystem services remains to be explored.
2. In this study, we investigated how abundance, taxonomic richness as well as the shape of biomass distribution in predator communities affect biological pest control services in vineyard landscapes. Our analyses were based on a dataset combining samples of arthropod predators, measures of predation rates of grape pests and characterization of environmental covariables for 42 fields located in South-Western France.
3. We found that beside the abundance or the taxonomic richness of predators, the shape of biomass distribution (mean, variance, skewness and kurtosis of the distribution) influences the level of biological control. Predator communities largely dominated by low biomass species provided the bulk of biological control services. Lower levels of predation resulted from increased proportions of large biomass species and more evenly distributed biomass values in the communities.
4. Our results indicate that the top-down control provided by low biomass species decreases as the relative proportion of large biomass species increases in the predator community. This suggests that biological control may be affected by negative interactions (e.g., intraguild predation, behavioural interactions) between predators arising from the recruitment of large individuals in the community.
5. Our study revealed that the shape of biomass distribution is a major aspect of functional diversity in predator communities providing insights into the mechanisms that link biodiversity and ecosystem services. While our study focuses on biomass, considering other traits involved in trophic interactions may increase our ability to predict the level of biological control in ecosystems.
Methods
(a) Experimental design
Our study sites were located in a vineyard-dominated region in south-western France (44°81’0°14′W). Our study design consisted of 21 pairs of organic and conventional vineyards (42 fields). The pairs were selected along two uncorrelated landscape gradients: a gradient of proportion of organic farming (ranging from 2% to 25%) and a gradient of proportion of semi-natural habitats (ranging from 1% to 75%) in a 1-km radius around the centre of the field. Such an experimental design allows to disentangle the relative effects of local farming systems from the proportion of semi-natural habitats and farming systems at the landscape scale. Landscape variables were calculated using ArcGIS 10.1 (ESRI).
(b) Predator community sampling
Predator communities of the vineyard foliage were sampled three times in 2015 (between June and September) (N = 124: 3 × 42 vineyards and 2 plots with missing values). At each sampling date, predators were sampled by beating 30 vine stocks in each vineyard (Muneret et al., 2019a). All predators were stored in 70% ethanol and individuals were identified at the lowest possible taxonomic resolution. Araneidae and Opiliones were identified at the species level while Dermaptera and Neuroptera were identified at the family level.
(c) Community metrics
Several metrics were calculated to characterize each predator community at each date: the overall abundance of individuals, the taxonomic richness as well as the four moments of the biomass distribution. Biomass distribution of the community was obtained by multiplying the number of individuals of each taxonomic unit by its average dry biomass (hereafter biomass distribution refers to the weighted biomass distribution). To calculate average values of dry biomass for each taxonomic unit, stage (adult or juvenile for spiders) and sex (for adult spiders and harvestmen), we randomly selected 10 individuals among all the individuals sampled (all sampling dates combined) and measured their dry biomass. For spiders with a dry body mass lower than 0.01 mg, we estimated the dry body mass using length–mass regression of the form: body mass = exp (a + b *log(length of cephalothorax)) (Barnes et al., 2016, see supplementary materials Fig. S1). When the abundance of each taxonomic unit, stage and sex was lower than 10 individuals, we measured all the specimens available to calculate the average biomass values. For each vineyard and sampling date, the mean, variance, skewness and kurtosis of each biomass distribution, were calculated following the formula in Gross et al. (2017). All the four moments were calculated on log-transformed data for each field at each sampling date.
(d) Biological pest control service
We used a sentinel approach to measure levels of biological pest control at three dates in 2015 concomitantly with the predator sampling (Birkhofer et al., 2017). Sentinel cards consisted of 10 eggs of the grape berry moth, L. botrana, which is the most damaging insect pest in the studied region. Eggs of L. botrana laid on parchment paper (1 × 3 cm card previously glued on felt) by laboratory-reared females were cut and glued on plastic cards (1 × 8 cm). Each card was attached to a vine shoot on one vine stock. In each field, 10 sentinel cards were exposed to predation for 4- or 5-days, depending of the sampling date. All the cards were settled at least 10 m away from the edge or from any other card. At the end of the 5-day exposure, the cards were collected (N=1207) and the number of remaining eggs per card was assessed using a microscope (Muneret et al., 2019b). We then estimated predation rates for each card as the ratio of the number of eggs predated to the total number of eggs initially exposed. In very few cases, eggs were damaged due to climatic conditions and the ratio was therefore calculated on the number of eggs initially exposed minus the number of damaged eggs. This variable was used as a proxy for biological pest control services.