Effect of forest stand composition on OPM moth captures
Experimental design – We used pheromone trapping to compare OPM population density in three forest stand types (henceforth, ‘triplets’) consisting of Q. robur and Q. petraea pure stands and of mixtures of both oak species. According to the definition of the National Forest Inventory of France, a forest stand is considered a pure stand if the dominant tree species represents more than 75% of the canopy cover. In mixed oak stands, both oak species were present in similar proportions. The trial was carried out in France, in 33 forests spread over 30 administrative departments (Fig. 1A). Forest stands were all in the public domain and managed by the French Forest National Office (ONF).
Stands were at least 500 m apart from each other within a triplet to avoid trapping individuals from the same OPM population. Triplets were separated from each other by several tens of kilometres and covered the entire French territory where the two oak species co-occurs.
In each stand, we used two funnel traps located at least 50 m apart from each other. We baited each trap with one different pheromone lure, obtained from Temmen GmbH Biotechnologie (Germany) (lure 1) and Pherobank (the Netherlands) (lure 2). Each trap was positioned in the upper canopy (10 – 15 m) to maximize captures (Breuer et al. 2003, Williams et al. 2013) at the end of June, before the expected flight period. The two traps were removed at the end of August 2018 and moths were returned to the laboratory for identification and counting.
Statistical analysis – We tested the effect of stand composition, pheromone type and their interaction with Generalised Linear Mixed-effects Models (GLMM with Poisson error distribution) using triplet and stand identity as a random factor. We estimated model fit as the proportion of variance explained by fixed effects (marginal R², Rm²) and fixed plus random effects (conditional R², Rc²) (Nakagawa and Schielzeth 2013).
Effect of forest stand composition on presence of OPM nests and defoliation
Stand selection – We tested the effect of host species and stand species composition on OPM nest density and defoliation in natural mature forests in North Eastern France. We selected 25 forest stands with Q. robur or Q. petraea as the main species with a sampling design consisting in five replicates of the following composition treatments: Q. robur pure stands, Q. petraea pure stands, mixtures of Q. robur and Q. petraea, mixtures of Q. robur and another broadleaved species (European beech Fagus sylvatica (L.)), mixtures of Q. petraea and another broadleaved species (hornbeam Carpinus betulus (L.)). Sampled forest stands were all in public forests and managed by ONF. In order to standardize pedoclimatic conditions among composition treatments, all stands were selected in the SER (sylvo-eco region) C30 corresponding to “plains and clay depressions of the North-Eastern France” (Inventaire forestier national 2011).
Field survey – We visited each stand in early summer (June-July 2018) before the production of lammas growth (i.e., right before the second leaf flush have obscured initial defoliation rate). We sampled 10 oaks of each species (i.e., 10 oaks in pure stands, 20 oaks in mixtures, 300 sampled oak trees in total) by walking in the stand along a U-shaped transect and selecting the closest oak tree to the transect line, every 10 m. For each individual oak, we measured the diameter at breast height (dbh). We described the tree species composition around five of the focal oaks per species and per stand (i.e. every 20 m along the transect) by identifying and measuring the dbh of every living tree (with dbh > 22.5 cm) in a buffer of 9 m radius (i.e., in a 254 m² buffer). We calculated the basal area of OPM host trees (i.e., summed basal area of Q. robur and Q. petraea) and the basal area of OPM non-host trees in each of these 254 m² plots.
Two observers examined the crown of each sampled oak tree with binoculars for five minutes; they counted the number of OPM larval nests and estimated their size using surface classes. However, the size of OPM nests appeared to vary strongly among oak trees, particularly for the bigger nests, and it was therefore impossible to provide an accurate and reliable estimate of the number of OPM caterpillars per nest. This information was not retained in further analyses. The number of OPM nests could therefore not be used with confidence as a proxy for OPM abundance. Instead, the response variable used for statistical analysis was made binomial, using presence or absence on the sampled tree. We estimated crown defoliation by assigning each tree to a defoliation class, on a scale of 10 % (Eichhorn et al. 2016). Feeding damage by OPM larvae is easily recognized, as the damaged leaves are left with only the veins (skeletonizer like). Defoliation was found to be greater than 10 % only for oaks with at least one OPM larval nest. We therefore assumed that OPM larvae were the main cause of defoliation in surveyed plots.
Statistical analysis – We tested the effect of stand composition on the presence of OPM nests and OPM defoliation at the level of individual focal oak trees by using (Generalised) Linear Mixed effect Models (G)LMMs with forest stand as a random factor. We tested the effect of oak species and plot composition on OPM using a two-step approach, for presence/absence data and defoliation separately. First, we modelled OPM response to oak species and forest stand composition described as a three-level factor (pure stand, mixtures of oaks, mixture of oak plus another broadleaved species). Second, we replaced the ‘composition’ three-level factor by the basal area of host and non-host species in plots, to address more explicitly the effect of host density vs host frequency on herbivores. In each model, we used the dbh of focal oak trees as a covariate. For each model, we simplified the initial full model by sequentially removing non-significant fixed effects, starting with the least significant, and estimated model coefficient parameters for the simplified model. We analysed OPM presence/absence data using a GLMM with binomial error distribution. We analysed defoliation with a LMM, using a square root transformation of the response variable to satisfy model assumptions. Defoliation data were analysed on the subset of trees that had been attacked by OPM (i.e., trees with at least one OPM nest, n = 195).
All analyses and figures were performed in R v3.5.1 (R Core Team 2018) with the following packages: multcomp, car, lme4, ggplot2, cowplot, lmerTest, sciplot, MuMIn, dplyr, DHARMa, doBy and tidyr (Hothorn et al. 2008, Fox and Weisberg 2011, Bates et al. 2015, Wickham 2016a, Kuznetsova et al. 2017, Morales et al. 2017, Wilke 2017, Barton 2018a, Hartig 2018, Højsgaard and Halekoh 2018, Wickham et al. 2018, Wickham and Henry 2018).
Effect of forest stand composition on OPM moth captures
We caught a total of 2,756 OPM males in the 185 pheromone traps. The mean number of captures per trap (± se) was 15 ± 3, ranging from 0 to 388 (median: 2). Captures were more concentrated in North Eastern France (Fig. 3A). Although most of variance was explained by site location (Rm² = 0.02; Rc² = 0.79), there was a significant effect of forest stand species composition on male captures (² = 8.27, Df = 2, P = 0.016) (Fig. 3). Specifically we captured, on average, 1.7 times more OPM in pure Q. petraea stands (15.28 ± 3.44) than in pure Q. robur stands (8.87 ± 2.36). We also found a significant interaction between stand composition and pheromone lure type (² = 38.12, Df = 2, P < 0.001). The number of captures differed between pheromone types only in pure Q. petraea stands (the number of captures was 2.0 times higher with the lure 1) and in mixed stands (the number of captures was 1.3 times higher with the lure 2) (Fig. 3B).
Effect of forest stand composition on presence of OPM nests and defoliation
Among the 300 oak trees sampled in our survey, 195 had at least one OPM nest (i.e., 65 %). Neither oak species, nor stand composition, nor basal area of neighbours (host or non-host), nor the dbh of focal oaks had a statistically clear effect on the probability of an oak being attacked by OPM (Table 1). By contrast, stand location (i.e. forest identity) explained 86 % of variability in presence of OPM nests on oaks.
Defoliation of oak trees by OPM larvae was on average (± se) 23.36 ± 1.79 %. Quercus petraea was significantly more defoliated (31.78 ± 3.37 %) than Q. robur (16.78 ± 1.62 %) (Fig 4, Table 1). OPM defoliation was significantly higher in pure stands (31.21 ± 3.83 %) than in mixed stands with other broadleaved species (15.00 ± 1.85 %) (Fig 4, Table 1). Neither basal areas of neighbours (host or non-host) nor dbh of focal tree significantly explained the percentage of defoliation (Table 1). Stand location (i.e., forest identity) explained 34 % of variability in OPM defoliation.
Effect of focal and neighbour tree species identity on OPM performance, leaf traits and plant phenology
OPM performance significantly differed between focal oak species (Table 1). In particular, OPM larval mortality was on average twice higher and OPM growth rate was on average twice lower on Q. robur than on Q. petraea (Fig. 5A, 5B), indicating that Q. petraea is a more suitable host for this herbivore species. We did not find any significant effects of neighbour tree species identity nor Focal × Neighbour interaction on OPM larval mortality or growth rate (Fig. 5, Table 2).
Some leaf traits significantly differed among focal oak species (Table 2). In particular, concentrations of hydrolysable tannins and flavonoids were on average 1.7-fold higher and lower (respectively) in Q. petraea than in Q. robur (Fig. 6). We did not find significant effects of neighbour tree species identity on leaf traits (Table 2). However, the Focal × Neighbour interaction significantly affected leaf flavonoid concentration and C:N (Table 2). Specifically, the concentration of flavonoids was on average 1.5 times higher in Q. robur in presence of heterospecific neighbours, whereas the C:N ratio was on average 1.2 time higher in Q. petraea in presence of conspecific neighbours (Fig. 6). Phenology was not significantly affected by focal or neighbour species (Table 2).
Leaf traits associated with effects of focal and neighbour tree species identity on OPM performance
Bud phenology, but not C:N nor concentrations of any type of phenolic compounds, had a significant effect on OPM growth and mortality rate (Table 3). OPM larvae performed better (lower mortality and better growth) when neonate had access to open buds with expanding leaves (Fig. 7AB).
The significant effect of focal oak species identity on OPM performance (growth and mortality) remained significant after including bud phenology as covariate (Table 4), indicating that phenology did not determine all the observed differences in herbivore performance between the two oak species.
Effects of oak species identity on OPM performance and leaf traits
Our results showed tree species-specific differences in OPM performance. In particular, OPM grew faster and suffered lower mortality rates when feeding on Q. petraea in comparison with Q. robur. Noteworthy, although our study was not designed to survey OPM development time, a greater proportion of Q. petraea than Q. robur seemed to have OPM larvae that had reached the third instar at the end of our experiment (26 days) (Fig. 8). These findings are consistent with two previous studies from our group. In a field experiment with mature oak trees, (Damestoy et al. under review) found that Q. petraea was consistently more attractive to OPM moths (i.e. more captures of moths in Q. petraea stands by pheromone trapping) and more defoliated than Q. robur (Damestoy et al. under review). Similarly, in a greenhouse experiment with one-year-old oak saplings Moreira et al. (2018a) found that leaf damage by gypsy moth larvae (Lymantria dispar) was significantly greater on Q. petraea than on Q. robur.
Table of contents :
1.1. Les insectes ravageurs : une problématique écologique et économique
1.2. Traits foliaires impliqués dans la résistance des arbres aux insectes défoliateurs
1.2.1. Qualité nutritive et métabolites spécialisés : des défenses contre les herbivores défoliateurs
1.3. Spécialiste vs généraliste : deux stratégies alimentaires
1.4. Résistance et susceptibilité par association dans les peuplements mélangés : patterns et mécanismes
1.4.1. Effet de la dilution de la ressource sur la susceptibilité vis-à-vis des herbivores
1.4.2. Effet indirect des plantes voisines sur les traits de résistance aux herbivores
1.5. Contribution de la diversité génétique aux effets d’association
1.6. Problématique et système d’étude
1.6.1. La chenille processionnaire du chêne : problème de santé des forêts et humaine
1.6.2. Problématique et hypothèses
2. Quels effets ont l’espèce hôte et la composition sur les captures des adultes et les défoliations de la processionnaire ?
2.2. Materials and methods
2.2.1. Effect of forest stand composition on OPM moth captures
2.2.2. Effect of forest stand composition on presence of OPM nests and defoliation .
2.3.1. Effect of forest stand composition on OPM moth captures
2.3.2. Effect of forest stand composition on presence of OPM nests and defoliation .
3. Quels effets ont l’espèce hôte et le voisinage sur les traits foliaires et les performances de la processionnaire ?
3.2. Materials and methods
3.3.1. Effect of focal and neighbour tree species identity on OPM performance, leaf traits and plant phenology
3.3.2. Leaf traits associated with effects of focal and neighbour tree species identity on OPM performance
3.4.1. Effects of oak species identity on OPM performance and leaf traits
3.4.2. Leaf traits associated with effects of focal tree species identity on OPM performance
3.4.3. Effects of oak neighbour species identity on OPM performance and leaf traits 61
4. Quels effets ont le génotype du chêne et les composés phénoliques sur les performances de deux insectes herbivores ?
4.2. Materials and methods
4.3.1. Effect of phenolics on performance of GM and OPM larvae
4.3.2. Effect of genetic variation in oaks on herbivore performance and oak phenolics
4.3.3. Genetic correlations between GM and OPM performance
4.4.1. Oak genotype influences herbivore consumption and digestion, but not growth
4.4.2. Leaf phenolics have contrasting effects on generalist and specialist herbivore species
4.4.3. Herbivore response to oak genotype is not primarily mediated by leaf phenolics
5.1. Quelle place ont les défenses chimiques dans la résistance aux herbivores ?
5.1.1. Les traits chimiques jouent-ils un rôle dans les défenses de la plante ?
5.1.2. Quelles autres défenses sont impliquées dans la résistance ?
5.2. Quel est le rôle du voisinage dans les préférences et performances des insectes herbivores ?
5.3. Comment les plantes font elles face à la grande diversité d’ennemies ?
5.4. La résistance aux herbivores est-elle contrôlée génétiquement ?
5.5. Implications pour la gestion des chênaies pour la résistance à la chenille processionnaire
5.5.1. La diversité est-elle une barrière face aux ravageurs ?
5.5.2. Les programmes de gestion des chênaies doivent-ils favoriser le chêne pédonculé au détriment du chêne sessile ?