Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2020 May 28;126(5):865–872. doi: 10.1093/aob/mcaa101

Leaf chemical defences and insect herbivory in oak: accounting for canopy position unravels marked genetic relatedness effects

Elena Valdés-Correcher 1,✉, Audrey Bourdin 1, Santiago C González-Martínez 1, Xoaquín Moreira 2, Andrea Galmán 2, Bastien Castagneyrol 1,#, Arndt Hampe 1,#
PMCID: PMC7539359  PMID: 32463869

Abstract

Background and Aims

Highly controlled experiments document that plant genetic diversity and relatedness can shape herbivore communities and patterns of herbivory. Evidence from the field is, however, scarce and inconsistent. We assessed whether a genetic signal underlying herbivory can be detected in oak woodlands when accounting for variation at smaller (within-tree) and larger (among-stand) scales.

Methods

We tested relationships between tree genetic relatedness, leaf chemical defences and insect herbivory for different canopy layers in 240 trees from 15 pedunculate oak (Quercus robur) forest stands. We partitioned sources of variability in herbivory and defences among stands, individuals and branches.

Key Results

Leaf defences, insect herbivory and their relationship differed systematically between the upper and the lower tree canopy. When accounting for this canopy effect, the variation explained by tree genetic relatedness rose from 2.8 to 34.1 % for herbivory and from 7.1 to 13.8 % for leaf defences. The effect was driven by markedly stronger relationships in the upper canopy.

Conclusions

Our findings illustrate that considerable effects of the host plant genotype on levels of leaf chemical defences and associated insect herbivory can be detected in natural tree populations when within-individual variation is properly accounted for.

Keywords: Genetic relatedness, chemical defences, herbivory, Quercus robur, plant–insect interactions

INTRODUCTION

A rapidly growing number of studies have shown that plant genetic diversity and genetic relatedness can influence herbivore communities and associated patterns of invertebrate herbivory (Crutsinger et al., 2006; McArt and Thaler, 2013; Kagiya et al., 2018). It has been proposed that the composition and activity of herbivore communities are heritable traits of the host plant that are partly driven by the heritability of its anti-herbivore chemical defences (Wimp et al., 2005; Bangert et al., 2006; Bustos-Segura et al., 2017; Barker et al., 2019). Indeed, plant families vary considerably in their edibility and resulting herbivore damage (Donaldson and Lindroth, 2007; Fernandez-Conradi et al., 2017; Barker et al., 2018; Damestoy et al., 2019). However, most previous research has been performed using highly controlled experiments (e.g. common gardens), often with juvenile plants and minimized spatial and environmental effects, settings that could lead to overemphasis on the putative role of genetics in nature (Tack et al., 2012; Lämke and Unsicker, 2018). Accordingly, more research in natural plant populations is needed for understanding the extent to which genetically based variation in plant chemical defences determines insect herbivory in real-world contexts (Wimp et al., 2005; Carmona et al., 2011).

Diverse biological mechanisms can contribute to the blurring of links between plant genotype, plant chemical defences and herbivory patterns in mature trees in the wild. Many secondary metabolites exhibit low heritability because their production is controlled by multiple genes and their interactions (Külheim et al., 2011; Büchel et al., 2016). Different plant parts experience different microclimates (e.g. irradiation, temperature, humidity) that can trigger extensive within-individual variation in leaf morphology and chemistry, especially along vertical gradients. For instance, upper canopy leaves are typically thicker, tougher, smaller and drier and contain higher levels of chemical leaf defences than lower canopy leaves (Murakami and Wada, 1997; Le Corff and Marquis, 1999; Murakami et al., 2005; Ruhnke et al., 2009; De Casas et al., 2011; Castagneyrol et al., 2019). More specifically, differences in microclimate should directly affect the expression of genes that encode the production of leaf chemical defences (Lämke and Unsicker, 2018). In turn, vertical gradients in insect herbivory can result from differences in herbivore dispersal (e.g. flying insects concentrated in the upper canopy; Ulyshen, 2011) or herbivore exposure to predators (e.g. lower predation rates in the upper canopy; Aikens et al., 2013), which are not driven by leaf chemistry. The plethora of potential confounding factors underpins the need for careful study designs (from the tree level to the within-individual level) to thoroughly assess the effects of genetically based variation in leaf chemical defences on herbivory in natural plant populations.

This study investigated the relationships between tree genetic relatedness, leaf defences and insect herbivory in natural forest stands of pedunculate oak (Quercus robur). For this purpose, we genotyped 703 trees from 15 stands and quantified the concentration of leaf phenolic compounds and herbivory by leaf-chewing insects in the intermediate and upper canopy layers for a subset of 235 trees. Specifically, we addressed the following questions: (1) To what extent do leaf phenolics and insect leaf herbivory vary among trees within stands and between canopy layers within trees? (2) Do leaf phenolics and herbivory show a genetic signal when accounting for their scale-dependent variation? (3) To what extent does variation in leaf phenolics explain patterns of leaf herbivory? By addressing these questions, we aim to combine a thorough description of cross-scale patterns typical of natural systems with insights into the biological mechanisms that underlie plant–insect herbivore relationships in non-experimental contexts.

MATERIALS AND METHODS

Study system

We performed this study in the Landes de Gascogne region (south-western France) ~40 km south of Bordeaux (44°41′ N, 00°51′ W). The area is dominated by extensive maritime pine (Pinus pinaster) plantations with scattered small stands of broadleaf forest. These are usually dominated by pedunculate oak and contain other tree species, like birch (Betula pendula), Pyrenean oak (Quercus pyrenaica), holly (Ilex aquifolium) and willows (Salix spp.) in minor abundance. Such stands are not subjected to intensive forest management and many are actively expanding (Gerzabek et al., 2017), favoured by a recent change in silvicultural management that tends to conserve oaks recruiting within adjacent pine plantations in order to increase biological pest management (Dulaurent et al., 2012). Pedunculate oak supports a large community of specialist and generalist herbivore insects in these stands (Giffard et al., 2012). Leaf chewers, skeletonizers, miners and gallers are the principal guilds responsible for background herbivory (i.e. the damage caused by a community of herbivores whose populations are at normal densities) that amounts to around 17.8 % (Giffard et al., 2012).

Forest stands, sampling and herbivory measurements

We selected 15 forest stands of variable size and spatial isolation within the landscape (Fig. 1). All stands were second-growth forests that have established since the 1950s through natural tree regeneration (Valdés-Correcher et al., 2019). They were strongly dominated by pedunculate oak and contained a variable but often rather sparse woody understorey vegetation. The number of established oak trees ranged from 16 to 124 individuals and their surface (as derived from the minimum polygon including all trees) from 0.04 to 0.5 ha. Further information can be found in Supplementary Data Table S1 (see also Valdés-Correcher et al., 2019). Within each stand, we mapped and tagged every oak tree with a diameter at breast height >3 cm and collected leaf material that was stored in silica gel until DNA isolation for genotyping. This exhaustive sampling included a total of 703 individuals (Supplementary Data Table S1).

Fig. 1.

Fig. 1.

Open in a new tab

Map of the study area in the Aquitaine region, south-western France, showing the 15 oak stands.

In September 2018 (i.e. late growing season), we randomly selected 16 individuals with a diameter at breast height >6 cm within each stand (total n = 240). On each tree, we randomly chose and cut two south-facing branches situated 4 and 8 m above ground level, respectively, which corresponds to the intermediate (shaded) and upper (sun-exposed) tree canopies in most of our trees (see also Castagneyrol et al., 2019). In the smallest stands (B, L and P; Supplementary Data Table S1) five trees were excluded as they did not have enough leaves to perform the measurements of leaf phenolics and insect herbivory. Three of the 235 sampled trees did not reach 8 m, so we shifted the position of the intermediate and upper tree canopy layers 2 m downward (i.e. 2 and 6 m, respectively). Operators unaware of the study design systematically picked the 30 most apical leaves from each branch, resulting in a total of 60 leaves per tree. Samples were stored at −18 °C until insect herbivory measurement.

For each leaf, we visually estimated the percentage of leaf area removed by chewing insects using the following scale: 0, 0 %; A, 1–5 %; B, 6–15 %; C, 16–25 %; D, 26–50 %; E, 51–75 %; F, >75 %. We used pre-established templates mimicking known levels of insect herbivory on oak leaves to increase the reliability and repeatability of herbivory measurements. Herbivory levels were always estimated by the same observer (A.B.) blind to leaf origin in order to maximize consistency of the estimates and reduce unconscious bias. We averaged values across all leaves to obtain a mean value per branch, and then used the median of each percentage class for statistical analyses (Castagneyrol et al., 2019).

We also collected ten fully expanded leaves with no signs of herbivory or pathogen infection from each branch for quantification of phenolic compounds. We immediately oven-dried these leaves for 48–72 h at 45 °C and ground them to a thin powder before further chemical analyses (see below).

Molecular analyses

We isolated genomic DNA from the leaves using the Invisorb® DNA Plant HTS 96 Kit/C and the standard protocol and genotyped all trees using 141 single-nucleotide polymorphism (SNP) markers from the sets described in Guichoux et al. (2013) and Gerzabek et al. (2017). The list of loci is available via an open-access repository (https://doi.org/10.5281/zenodo.3786180). Loci were multiplexed using an iPLEX Gold kit on a MassARRAY Typer 4.0.26.75 (Agena Biosciences) at the Genomic and Sequencing Facility of Bordeaux (France), as described in detail in Gerzabek et al. (2017). We obtained high-quality data with a low proportion of missing calls for all markers and individuals, which we assembled into multilocus genotypes for the exhaustive sample (n = 703).

Chemical analyses

We extracted phenolic compounds from 20 mg of dry leaf tissue with 1 mL of 70 % methanol in an ultrasonic bath for 20 min, followed by centrifugation (Moreira et al., 2014). Samples were centrifuged and transferred to chromatographic vials. We performed the chromatographic analyses in an ultra-high-performance liquid chromatograph (UHPLC Nexera LC-30AD; Shimadzu Corporation, Kyoto, Japan) equipped with a Nexera SIL-30AC injector and an SPD-M20A UV/VIS photodiode array detector.

For compound separation, we used a Kinetex™ 2.6-µm C18 82-102 Å LC Column 100 × 4.6 mm (Phenomenex, Torrance, CA, USA), protected with a C18 guard cartridge. The flow rate was established at 0.4 mL min−1 and the oven temperature was set at 25°C. The mobile phase consisted of two solvents: water–formic acid (0.05 %) (A) and acetonitrile–formic acid (0.05 %) (B), starting with 5 % B and using a gradient to obtain 30 % B at 4 min, 60 % B at 10 min, 80 % B at 13 min and 100 % B at 15 min. The injection volume was 15 µL. We recorded chromatograms at 330 nm and processed data with LabSolutions software (Shimadzu, Kyoto, Japan). For phenolic compound identification, we used ultra-performance liquid chromatography coupled with electrospray ionization quadrupole (Thermo Dionex Ultimate 3000 LC; Thermo Fisher Scientific, Waltham, MA, USA) time-of-flight mass spectrometry (UPLC-Q-TOF-MS/MS; Bruker Compact™, Bruker, Billerica, MA, USA). We identified four groups of phenolic compounds: flavonoids; ellagitannins and gallic acid derivatives (‘hydrolysable tannins’ hereafter); proanthocyanidins (‘condensed tannins’ hereafter); and hydroxycinnamic acid precursors to lignins (‘lignins’ hereafter). We quantified flavonoids as rutin equivalents, condensed tannins as catechin equivalents, hydrolysable tannins as gallic acid equivalents, and lignins as ferulic acid equivalents (Moreira et al., 2018). We quantified these phenolic compounds with external calibration using calibration curves at 0.25, 0.5, 1, 2 and 5 μg mL−1. We calculated total phenolics for each branch as the sum of flavonoids, lignins, condensed tannins and hydrolysable tannins, and expressed concentrations of each phenolic group in mg g−1 tissue on a dry weight basis.

Statistical analyses

Prior to the analysis of genetic relatedness, we first examined and visualized the landscape-scale genetic structure of our oak stands by calculating pairwise FST between stands (Weir and Cockerham, 1984) and applying principal components analysis (PCoA) to the obtained values with the dudi.pca function in the R package adegenet (Jombart and Ahmed, 2011). Overall low values (mean FST = 0.041, range 0.006–0.111) (Supplementary Data Table S2) and the absence of population structure (Supplementary Data Fig. S1) confirmed that the 15 stands can be considered a single gene pool and eventual confounding effects due to population genetic structure are negligible. Then, we quantified the level of genetic relatedness between each pair of trees relative to the full sample (n = 703). For this, we computed a kinship matrix using Nason’s kinship coefficient (Loiselle et al., 1995) with the software SPAGeDi version 1.2 (Hardy and Vekemans, 2002). We extracted the values for our 16 target trees per stand from the global matrix and used this information as a quantitative estimate of their genetic relatedness in the subsequent analyses (Van Horn et al., 2008). Note that kinship-based estimates of relatedness, while commonly used in population genetics (Pemberton, 2008), are not directly comparable to those obtained through pedigree analyses.

We modelled patterns of insect leaf herbivory and leaf phenolics at the whole-tree level (i.e. pooling the two canopy layers) and at the canopy layer level (i.e. for each layer separately) by means of linear mixed-effect models (LMMs). At the whole-tree level, we built two independent LMMs with stand ID and the kinship values of the target trees as random factors in order to estimate the variance and the percentage of the overall variance explained by the local environment (stand ID) and by the genetic relatedness among trees (the kinship matrix). The first model was an intercept-only model with (tree-level mean) concentration of leaf phenolics as response variable (eqn 1). The second model included (tree-level mean) insect herbivory as response variable and leaf phenolics as an additional explanatory variable (eqn 2).

Model 1: Phenolicsi,j = β0 + Si + Tj + εi,j (1)

Model 2: Herbivoryi,j = β0 + β1 × Phenolicsi,j+ Si + Tj + εi,j (2)

where β0 is the model intercept, β1 the fixed effect of leaf phenolics, Si the random effect of stand i, Tj the random effect of tree genetic similarity j (entered as the kinship matrix) and εi,j the error, with Si ∈ N(0, σS2), Tj ∈ N(0, σT2) and εi,j ∈ N(0, σE2). For each model, we computed the variance of the fixed effects (if any, σF2) and calculated the percentage of variance explained by each random factor (e.g. 100 × σT2/(σS2 + σT2 + σE2) for the random effect of tree genetic similarity in model 2).

The second group of models, performed at canopy layer level, adopted the same approach with two independent LMMs modelling the response of total leaf phenolics and herbivory, respectively (eqns 3 and 4):

Model 3: Phenolicsi,j,k = β0 + β1 × Canopy layeri,j,k + Si + Tj + εi,j,k (3)

Model 4: Herbivoryi,j,k = β0 + β1 × Phenolicsi,j,k + β2 + Canopy layeri,j,k + Si + Tj + εi,j,k (4)

where k indicates the branch, β0 the intercept, β1 and β2 the coefficient parameters of the fixed effects and Si, Tj and εi,j,k as above. Again, we computed the variance of the fixed effects (if any, σF2) and calculated the percentage of variance explained by each random factor. Prior analyses revealed that including axes 1, 2 and 3 from our PCoA on the landscape-scale genetic structure (which explained 1.05 % of the total variation) did not change the outcomes, so we decided not to include the axes in the final models.

All analyses were done in R version 3.5.2 (R Core Team, 2018). LMMs including a kinship matrix as random factor were fitted with the function lmekin in package coxme (Therneau, 2018).

RESULTS

Leaf phenolics and insect herbivory at whole-tree level

Leaf phenolic concentration was on average (± s.e.) 14.69 ± 0.39 mg g−1 (Fig. 2). The random factors collectively explained 26.9 % of the overall variation, with stand ID accounting for 19.7 % and tree genetic relatedness for 7.1 %. Insect leaf herbivory was on average 12.27 ± 0.29 % and decreased significantly with increasing leaf phenolic concentration (model coefficient parameter estimate −0.12 ± 0.05, z = −2.48, P = 0.013). However, the effect size was small (2.0 %). Stand ID explained 38.1 % of the overall variation in herbivory and genetic similarity among trees only accounted for another 2.9 %.

Fig. 2.

Fig. 2.

Open in a new tab

Percentage of insect herbivory and concentration of total leaf phenolics in the intermediate and upper tree canopy. Symbols and error bars represent means ± s.e. aggregated at the level of oak stands (A–Q). Note that stands are ordered according to mean insect herbivory and the same order is used to display stand-level variability in leaf phenolics.

Leaf phenolics and insect herbivory at canopy layer level

Leaf phenolic concentration was significantly lower in the intermediate than in the upper canopy layer (mean ± s.e. 13.63 ± 0.50 versus 15.79 ± 0.59 mg g−1) (Fig. 2, Table 1). Stand ID accounted for 13.7 % and tree genetic relatedness for 13.9 % of the overall variation (Table 1). Insect leaf herbivory did not differ significantly between tree canopies (12.53 ± 0.39 versus 11.99 ± 0.43 %) and was independent of leaf phenolic concentration (Table 1). Stand ID and tree genetic relatedness accounted for 32.0 and 34.7 % of the overall variability in insect herbivory, respectively (Table 1).

Table 1.

Summary of LMM testing the effect of canopy layer (upper versus intermediate) on leaf phenolics or insect herbivory. For insect herbivory, the effect of leaf phenolics was also included in the model. Significant variables are indicated in bold σ2 (%), variance and percentage of variance explained by the random factors of stand ID, genetic similarity introduced as a kinship matrix, and the residuals.R2m and R2c, variance explained by fixed and fixed plus random factors, respectively.

Response Predictors Coefficient ± s.e. z- value P- value σ 2 (%) R 2m (R2c)
Stand ID σS2 Genetic relatedness σT2 Residuals σE2
Phenolics Canopy layer 2.15 ± 0.66 3.22 0.001 9.27 (19.7) 9.64 (13.8) 50.22 (70.9) 1.63 (29.1)
Herbivory Phenolics −0.458 ± 0.24 −1.9 0.057 13.44 (32.0) 14.57 (34.7) 13.95 (33.3) 0.79 (66.6)
Canopy layer −0.585 ± 0.36 −1.62 0.1
Open in a new tab

In the intermediate canopy layer, stand ID and genetic relatedness accounted for 13.5 and 0.03 % of the overall variation in leaf phenolics, respectively. Leaf phenolics had no significant effect on herbivory (Fig. 3). Stand ID explained 40.5 % of the overall variation in herbivory, while tree genetic relatedness accounted for less than 0.02 %.

Fig. 3.

Fig. 3.

Open in a new tab

Effects of total leaf phenolics on insect herbivory in the upper (A) and intermediate (B) canopy layer. Data points represent individual trees. The thick solid line and the thin dashed lines in (A) represent model predictions and corresponding standard errors, respectively.

In the upper canopy layer, stand ID and genetic relatedness explained 17.4 and 24.8 % of the overall variation in leaf phenolics, respectively. There was a significant, albeit weak, negative effect of leaf phenolic concentration on herbivory (coefficient parameter estimate ± s.e. −0.12 ± 0.05, z = −2.63, P = 0.009) (Fig. 3). Leaf phenolics accounted 2.8 % of the overall variation in herbivory while stand ID and tree genetic relatedness accounted for 25.3 and 14.5 %, respectively.

DISCUSSION

Tree genetic relatedness explained a noteworthy part of the overall variation in leaf phenolics and associated insect leaf herbivory. However, this effect was only evident in the upper tree canopy, where concentrations of leaf phenolics were consistently higher. To our knowledge, our work represents one of the first confirmations of genotype–phenotype–herbivory links in natural tree populations and highlights that increased consideration of canopy effects is needed to improve our understanding of ecological and evolutionary factors driving plant–herbivore interactions in long-lived plants.

Trees lost between 7 and 22 % of their leaf area to insect herbivores, a range of defoliation similar to previous estimates (Giffard et al., 2012; Castagneyrol et al., 2019; Valdés-Correcher et al., 2019). Our analysis at the whole-tree level attributed most of the overall variation in leaf herbivory to differences among forest stands, whereas the contributions of tree genetic relatedness and leaf phenolics were very weak. This result might suggest that insect leaf herbivory in our system would be basically driven by the nature of the forest stand, which encapsulates diverse environmental drivers acting at the local (e.g. stand size, tree density and species composition, vegetation structure; Maguire et al., 2016; Fuller et al., 2018; van Schrojenstein Lantman et al., 2018) to landscape (e.g. stand isolation, nature of matrix habitats; Morante-Filho et al., 2016) scale. Valdés-Correcher et al. (2019) actually reported for the same study stands that their size and isolation affected patterns of herbivory by different insect guilds. Limiting our analyses to the whole-tree level would hence have led to the conclusion that insect herbivory is primarily determined by extrinsic drivers and shaped by the ecological neighbourhood of the focal tree.

While tree genetic relatedness had little effect on herbivory (2.9 %), it was somewhat more influential in the case of leaf phenolics (7.1 %) (Fig. 4). Together with the likewise weak but statistically significant negative association between leaf phenolics and herbivory, one might argue that our results mirror – albeit extremely faintly – experimental studies that have consistently identified plant chemistry as the phenotypic link between the host plant genotype and the structure of associated arthropod communities (Bangert et al., 2006; Barbour et al., 2009, 2015) or patterns of herbivory (Bailey et al., 2006; Andrew et al., 2007; Donaldson and Lindroth, 2007). But consistent empirical support for this linkage from natural tree populations remains very scarce. In one of the few available studies, Kagiya et al. (2018) found that genetic relatedness of alder (Alnus hirsuta) trees largely influenced associated arthropod communities, although the effect was stronger for herbivore predators than for the herbivores themselves. Maldonado-López et al. (2015) observed that tree genetic relatedness of Quercus castanea trees was significantly associated with chemical defences but not with insect herbivory. In turn, Tack et al. (2012) and Gossner et al. (2015) failed to detect any relationships between tree genetic relatedness and herbivory in Q. robur populations and concluded that genetic effects tend to be overwhelmed by environmental and spatial factors.

Fig. 4.

Fig. 4.

Open in a new tab

Summary of variance partitioning among random effects and model residuals for leaf phenolics (left) and insect herbivory (right). Box length is proportional to the percentage of variance explained by each component.

In line with the predominant trend reported in the literature (e.g. Yamasaki and Kikuzawa, 2003; Poorter et al., 2006; Lämke and Unsicker, 2018; but see Roslin et al., 2006), we observed that upper canopy leaves systematically contained higher concentrations of leaf phenolics than those from the intermediate canopy. Extensive within-individual variation in leaf morphological and chemical traits is an inherent feature of plants (Niklas et al., 2009; Herrera, 2017). For leaf phenolics, the phenomenon has been primarily explained as an ecophysiological, enzymatic and transcriptomic consequence of the higher irradiance that upper canopy leaves receive, given that diverse phenolic compounds are involved in protection from UV-B damage (reviewed in Jenkins and Brown, 2018; see also Lämke and Unsicker, 2018). This vertical gradient in leaf phenolics could have important consequences for plant–insect herbivore interactions. Under experimental settings, herbivores tend to forage preferentially on upper canopy leaves owing to their higher nutritive value (Fortin and Mauffette, 2002; Oishi et al., 2006). Yet field surveys typically report higher levels of leaf removal in lower canopy layers (e.g. Yamasaki and Kikuzawa, 2003; Stiegel et al., 2017; Castagneyrol et al., 2019), which is in line with a higher abundance and diversity of herbivores in these layers (reviewed in Ulyshen, 2011). The present study did not, however, detect differences in herbivory between the upper and the intermediate canopy layer. This could be due to the combined effects of different vertical gradients in bottom-up (e.g. leaf nutritional quality and defences) and top-down (e.g. predation and parasitism) forces that act simultaneously upon insect herbivores. For instance, in upper canopy layers herbivores are favoured by high leaf nutrient quality (Fortin and Mauffette, 2002) but constrained by high levels of leaf defences (Moreira et al., 2017), while in low canopy layers herbivores are also negatively influenced by higher abundances of predators (Aikens et al., 2013). On the other hand, differences in herbivore community composition might also contribute to the blurring of vertical trends in herbivory (Murakami et al., 2005). Extensive research on within-individual variation in leaf traits and associated herbivory has given rise to the hypothesis that variance in nutritional quality itself could act as a defence mechanism that reduces insect herbivore performance by forcing herbivores to actively forage for suitable food (e.g. Wetzel et al., 2016; Wetzel and Meek, 2019). Yet few if any studies have addressed the implications of this within-individual variability for genotype–phenotype–herbivory relationships.

The effect of tree genetic relatedness on leaf phenolics and insect herbivory was contingent on the canopy layer. Effects were considerable in the upper canopy but negligible in the lower canopy. The higher average levels of phenolic compounds in the upper canopy layer indicate that the genes encoding the production of leaf defences are generally more strongly expressed in this part, probably owing to their stronger exposure to solar radiation (Lämke and Unsicker, 2018). This effect would exacerbate phenotypic differences between more and less effective gene variants. In contrast, light transmittance tends to be lower and more variable in the intermediate compared with the upper canopy layer (Parker et al., 2002), which would weaken genotype–phenotype relationships in this part. Sun leaves are far more productive in terms of carbon fixation than shade leaves (Poorter et al., 2006) and their defence against herbivores is therefore disproportionately important for overall tree performance. Our finding that tree genotypes with high phenolic compound contents in the upper canopy systematically experience lower herbivory hence suggests that such genotypes could have a non-negligible fitness advantage. In turn, leaf defence levels in the intermediate canopy layer would be overall too low and too much driven by small-scale variation in light transmittance to generate notable among-genotype differences in insect herbivory. On the other hand, the extent of intra-individual variability in phenolic compounds can be heritable (Herrera, 2017) and might act as an indirect defensive trait (Wetzel et al., 2016; Wetzel and Whitehead, 2020). If this were the case in our study system, we would expect that trees with large differences in defence allocation between upper and lower canopy leaves would tend to experience reduced herbivory. Our data did not, however, confirm such a trend (results not shown), suggesting that the strength of within-individual variation in leaf defences either lacks a genetic basis or has no effect on (tree-level) herbivore activity. Finally, the genetic signal in leaf herbivory that we detected suggests that leaf defences may differentially drive herbivory community heritability across different parts of the canopy. The phenomenon has been thoroughly documented at the whole-plant level in common garden experiments (e.g. Andrew et al., 2007; Robinson et al., 2012), whereas studies in natural populations have reported lower or non-significant levels of genetic variation and heritability. One important reason may be that most previous studies investigating the role of tree genetics in defences and associated herbivory have not explicitly addressed the role of the canopy layer (but instead pooled leaf samples from different heights; e.g. Gossner et al., 2015; Maldonado-López et al., 2015; Kagiya et al., 2018).

Conclusions

Our study shows that failing to properly take within-individual variability in defences and herbivory into account can easily mask underlying genetic signals in natural populations. Our study therefore calls for a re-interpretation of the debated role of plant genetics in determining herbivore damage. Based on our findings, we recommend that future studies adopt hierarchical sampling designs and properly consider within-individual variability in both plant traits and insect herbivory when exploring their genetic basis in real-world contexts. Finally, we also recommend that further studies include other defence traits (e.g. physical defences such as trichomes and toughness or indirect defences such as volatile organic compounds) and strategies (e.g. induced defences or tolerance). Distinguishing between all these traits or strategies would make it possible to fully characterize multivariate defensive phenotypes (i.e. syndromes) and to better understand within and among-individual variation in genotype–phenotype–herbivory relationships.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following.

Figure S1: results of principal components analysis depicting axes PC1 versus PC2 and PC2 versus PC3 for the 15 stands. Table S1: information about the location and size of the oak stands included in the study. Table S2: pairwise FST values of the 15 oak stands.

mcaa101_suppl_Supplementary_Figure_A1
Click here for additional data file. (219.1KB, pptx)
mcaa101_suppl_Supplementary_Table_A1
Click here for additional data file. (13.1KB, docx)
mcaa101_suppl_Supplementary_Table_A2
Click here for additional data file. (14.4KB, docx)

ACKNOWLEDGEMENTS

We are most indebted to Xavier Capdevielle, Yannick Mellerin, Inge van Halder, Gilles Saint-Jean, Victor Rébillard, Martine Martin-Clotte and Patrick Leger for their technical assistance in the field, Gabriel Gerzabek, Erwan Guichoux, Marie Massot and Adline Delcamp for performing the molecular analyses, and Benjamin Brachi for his help with the data analyses. We also thank two anonymous reviewers for helpful comments. E.V.C., B.C. and A.H. conceived the study. E.V.C. and A.B. acquired the data. A.B., X.M. and A.G. performed the chemical analysis. E.V.C, A.B. and B.C analysed the data with help from S.C.G.M. E.V.C. and A.H. drafted the first version of the manuscript. All authors edited the final version of the manuscript and approved its submission.

FUNDING

This work was funded by The BiodivERsA project SPONFOREST (BiodivERsA3-2015-58). Genotyping was performed at the Genome Transcriptome Facility of Bordeaux (grants from Investissements d’Avenir, Convention attributive d’aide EquipEx Xyloforest ANR-10-EQPX-16-01).

LITERATURE CITED

  1. Aikens KR, Timms LL, Buddle CM. 2013. Vertical heterogeneity in predation pressure in a temperate forest canopy. PeerJ  2013: 1–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Andrew RL, Wallis IR, Harwood CE, Henson M, Foley WJ. 2007. Heritable variation in the foliar secondary metabolite sideroxylonal in Eucalyptus confers cross-resistance to herbivores. Oecologia  153: 891–901. [DOI] [PubMed] [Google Scholar]
  3. Bailey JK, Wooley SC, Lindroth RL, Whitham TG. 2006. Importance of species interactions to community heritability: a genetic basis to trophic-level interactions. Ecology Letters  9: 78–85. [DOI] [PubMed] [Google Scholar]
  4. Bangert RK, Turek RJ, Rehill B, et al.  2006. A genetic similarity rule determines arthropod community structure. Molecular Ecology  15: 1379–1391. [DOI] [PubMed] [Google Scholar]
  5. Barbour MA, Rodriguez-Cabal MA, Wu ET, et al.  2015. Multiple plant traits shape the genetic basis of herbivore community assembly. Functional Ecology  29: 995–1006. [Google Scholar]
  6. Barbour RC, Baker SC, O’Reilly-Wapstra JM, Harvest TM, Potts BM. 2009. A footprint of tree-genetics on the biota of the forest floor. Oikos  118: 1917–1923. [Google Scholar]
  7. Barker HL, Holeski LM, Lindroth RL. 2018. Genotypic variation in plant traits shapes herbivorous insect and ant communities on a foundation tree species. PLoS ONE  13: e0200954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Barker HL, Holeski LM, Lindroth RL. 2019. Independent and interactive effects of plant genotype and environment on plant traits and insect herbivore performance: a meta‐analysis with Salicaceae. Functional Ecology  33: 422–435. [Google Scholar]
  9. Büchel K, Fenning T, Gershenzon J, Hilker M, Meiners T. 2016. Elm defence against herbivores and pathogens: morphological, chemical and molecular regulation aspects. Phytochemistry Reviews  15: 961–983. [Google Scholar]
  10. Bustos-Segura C, Poelman EH, Reichelt M, Gershenzon J, Gols R. 2017. Intraspecific chemical diversity among neighbouring plants correlates positively with plant size and herbivore load but negatively with herbivore damage. Ecology Letters  20: 87–97. [DOI] [PubMed] [Google Scholar]
  11. Carmona D, Lajeunesse MJ, Johnson MTJ. 2011. Plant traits that predict resistance to herbivores. Functional Ecology  25: 358–367. [Google Scholar]
  12. De Casas RR, Vargas P, Pérez-Corona E, Manrique E, García-Verdugo C, Balaguer L. 2011. Sun and shade leaves of Olea europaea respond differently to plant size, light availability and genetic variation. Functional Ecology  25: 802–812. [Google Scholar]
  13. Castagneyrol B, Giffard B, Valdés-Correcher E, Hampe A. 2019. Tree diversity effects on leaf insect damage on pedunculate oak: the role of landscape context and forest stratum. Forest Ecology and Management  433: 287–294. [Google Scholar]
  14. Le Corff J, Marquis RJ. 1999. Differences between understorey and canopy in herbivore community composition and leaf quality for two oak species in Missouri. Ecological Entomology  24: 46–58. [Google Scholar]
  15. Crutsinger GM, Collins MD, Fordyce JA, Gompert Z, Nice CC, Sanders NJ. 2006. Plant genotypic diversity predicts community structure and governs an ecosystem process. Science  313: 966–968. [DOI] [PubMed] [Google Scholar]
  16. Damestoy T, Brachi B, Moreira X, Jactel H, Plomion C, Castagneyrol B. 2019. Oak genotype and phenolic compounds differently affect the performance of two insect herbivores with contrasting diet breadth. Tree Physiology  39: 615–627. [DOI] [PubMed] [Google Scholar]
  17. Donaldson JR, Lindroth RL. 2007. Genetics, environment, and their interaction determine efficacy of chemical defense in trembling aspen. Ecology  88: 729–739. [DOI] [PubMed] [Google Scholar]
  18. Dulaurent AM, Porté AJ, van Halder I, Vétillard F, Menassieu P, Jactel H. 2012. Hide and seek in forests: colonization by the pine processionary moth is impeded by the presence of nonhost trees. Agricultural and Forest Entomology  14: 19–27. [Google Scholar]
  19. Fernandez-Conradi P, Jactel H, Hampe A, Leiva MJ, Castagneyrol B. 2017. The effect of tree genetic diversity on insect herbivory varies with insect abundance. Ecosphere  8: e01637. [Google Scholar]
  20. Fortin M, Mauffette Y. 2002. The suitability of leaves from different canopy layers for a generalist herbivore (Lepidoptera: Lasiocampidae) foraging on sugar maple. Canadian Journal of Forest Research  32: 379–389. [Google Scholar]
  21. Fuller L, Fuentes-Montemayor E, Watts K, Macgregor NA, Bitenc K, Park KJ. 2018. Local-scale attributes determine the suitability of woodland creation sites for Diptera. Journal of Applied Ecology  55: 1173–1184. [Google Scholar]
  22. Gerzabek G, Oddou-Muratorio S, Hampe A. 2017. Temporal change and determinants of maternal reproductive success in an expanding oak forest stand. Journal of Ecology  105: 39–48. [Google Scholar]
  23. Giffard B, Corcket E, Barbaro L, Jactel H. 2012. Bird predation enhances tree seedling resistance to insect herbivores in contrasting forest habitats. Oecologia  168: 415–424. [DOI] [PubMed] [Google Scholar]
  24. Gossner MM, Brändle M, Brandl R, Bail J, Müller J, Opgenoorth L. 2015. Where is the extended phenotype in the wild? The community composition of arthropods on mature oak trees does not depend on the oak genotype. PLoS ONE  10: e0115733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Guichoux E, Garnier-Géré P, Lagache L, Lang T, Boury C, Petit RJ. 2013. Outlier loci highlight the direction of introgression in oaks. Molecular Ecology  22: 450–462. [DOI] [PubMed] [Google Scholar]
  26. Hardy OJ, Vekemans X. 2002. SPAGeDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Molecular Ecology Notes  2: 618–620. [Google Scholar]
  27. Herrera CM. 2017. The ecology of subindividual variability in plants: patterns, processes, and prospects. Web Ecology  17: 51–64. [Google Scholar]
  28. Van Horn RC, Altmann J, Alberts SC. 2008. Can’t get there from here: inferring kinship from pairwise genetic relatedness. Animal Behaviour  75: 1173–1180. [Google Scholar]
  29. Jenkins GI, Brown BA. 2018. UV-B perception and signal transduction. Annual Plant Reviews online  155–182. [Google Scholar]
  30. Jombart T, Ahmed I. 2011. adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics  27: 3070–3071. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kagiya S, Yasugi M, Kudoh H, Nagano AJ, Utsumi S. 2018. Does genomic variation in a foundation species predict arthropod community structure in a riparian forest?  Molecular Ecology  27: 1284–1295. [DOI] [PubMed] [Google Scholar]
  32. Külheim C, Yeoh SH, Wallis IR, Laffan S, Moran GF, Foley WJ. 2011. The molecular basis of quantitative variation in foliar secondary metabolites in Eucalyptus globulus. New Phytologist  191: 1041–1053. [DOI] [PubMed] [Google Scholar]
  33. Lämke JS, Unsicker SB. 2018. Phytochemical variation in treetops: causes and consequences for tree-insect herbivore interactions. Oecologia  187: 377–388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Loiselle BA, Sork VL, Nason J, Graham C. 1995. Spatial genetic structure of a tropical understory shrub, Psychotria officinalis (Rubiaceae). American Journal of Botany  82: 1420–1425. [Google Scholar]
  35. Maguire DY, Buddle CM, Bennett EM. 2016. Within and among patch variability in patterns of insect herbivory across a fragmented forest landscape. PLoS ONE  11: 1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Maldonado-López Y, Cuevas-Reyes P, González-Rodríguez A, Pérez-López G, Acosta-Gómez C, Oyama K. 2015. Relationships among plant genetics, phytochemistry and herbivory patterns in Quercus castanea across a fragmented landscape. Ecological Research  30: 133–143. [Google Scholar]
  37. McArt SH, Thaler JS. 2013. Plant genotypic diversity reduces the rate of consumer resource utilization. Proceedings of the Royal Society B: Biological Sciences  280: 20130639. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Morante-Filho JC, Arroyo-Rodríguez V, Lohbeck M, Tscharntke T, Faria D. 2016. Tropical forest loss and its multitrophic effects on insect herbivory. Ecology  97: 3315–3325. [DOI] [PubMed] [Google Scholar]
  39. Moreira X, Mooney KA, Rasmann S, et al.  2014. Trade-offs between constitutive and induced defences drive geographical and climatic clines in pine chemical defences. Ecology Letters  17: 537–546. [DOI] [PubMed] [Google Scholar]
  40. Moreira X, Glauser G, Abdala-Roberts L. 2017. Interactive effects of plant neighbourhood and ontogeny on insect herbivory and plant defensive traits. Scientific Reports  7: 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Moreira X, Castagneyrol B, Abdala-Roberts L, et al.  2018. Latitudinal variation in plant chemical defenses drives latitudinal patterns of leaf herbivory. Ecography  41: 1124–1134. [Google Scholar]
  42. Murakami M, Wada N. 1997. Difference in leaf quality between canopy trees and seedlings affects migration and survival of spring-feeding moth larvae. Canadian Journal of Forest Research  27: 1351–1356. [Google Scholar]
  43. Murakami M, Yoshida K, Hara H, Toda MJ. 2005. Spatio-temporal variation in lepidopteran larval assemblages associated with oak, Quercus crispula: the importance of leaf quality. Ecological Entomology  30: 521–531. [Google Scholar]
  44. Niklas KJ, Cobb ED, Spatz HC. 2009. Predicting the allometry of leaf surface area and dry mass. American Journal of Botany  96: 531–536. [DOI] [PubMed] [Google Scholar]
  45. Oishi M, Yokota T, Teramoto N, Sato H. 2006. Japanese oak silkmoth feeding preference for and performance on upper-crown and lower-crown leaves. Entomological Science  9: 161–169. [Google Scholar]
  46. Parker GG, Davis MM, Chapotin SM. 2002. Canopy light transmittance in Douglas-fir–western hemlock stands. Tree Physiology  22: 147–157. [DOI] [PubMed] [Google Scholar]
  47. Pemberton JM. 2008. Wild pedigrees: the way forward. Proceedings. Biological Sciences  275: 613–621. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Poorter H, Pepin S, Rijkers T, de Jong Y, Evans JR, Körner C. 2006. Construction costs, chemical composition and payback time of high- and low-irradiance leaves. Journal of Experimental Botany  57: 355–371. [DOI] [PubMed] [Google Scholar]
  49. R Core Team  2018. R: A Language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
  50. Robinson KM, Ingvarsson PK, Jansson S, Albrectsen BR. 2012. Genetic variation in functional traits influences arthropod community composition in aspen (Populus tremula L.). PLoS ONE  7: e37679. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Roslin T, Gripenberg S, Salminen JP, et al.  2006. Seeing the trees for the leaves – oaks as mosaics for a host-specific moth. Oikos  113: 106–120. [Google Scholar]
  52. Ruhnke H, Schädler M, Klotz S, Matthies D, Brandl R. 2009. Variability in leaf traits, insect herbivory and herbivore performance within and among individuals of four broad-leaved tree species. Basic and Applied Ecology  10: 726–736. [Google Scholar]
  53. van Schrojenstein Lantman IM, Hertzog LR, Vandegehuchte ML, et al.  2018. Leaf herbivory is more impacted by forest composition than by tree diversity or edge effects. Basic and Applied Ecology  29: 79–88. [Google Scholar]
  54. Stiegel S, Entling MH, Mantilla-Contreras J. 2017. Reading the leaves’ palm: leaf traits and herbivory along the microclimatic gradient of forest layers. PLoS ONE  12: e0169741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Tack AJM, Johnson MTJ, Roslin T. 2012. Sizing up community genetics: it’s a matter of scale. Oikos  121: 481–488. [Google Scholar]
  56. Therneau TM. 2018. coxme: Mixed Effects Cox Models.  R package version 2.2–10; https://CRAN.R-project.org/package=coxme. [Google Scholar]
  57. Ulyshen MD. 2011. Arthropod vertical stratification in temperate deciduous forests: Implications for conservation-oriented management. Forest Ecology and Management  261: 1479–1489. [Google Scholar]
  58. Valdés-Correcher E, van Halder I, Barbaro L, Castagneyrol B, Hampe A. 2019. Insect herbivory and avian insectivory in novel native oak forests: divergent effects of stand size and connectivity. Forest Ecology and Management  445: 146–153. [Google Scholar]
  59. Weir BS, Cockerham C. 1984. Estimating F-statistics for the analysis of population structure. Evolution  38: 1358–1370. [DOI] [PubMed] [Google Scholar]
  60. Wetzel WC, Meek MH. 2019. Physical defenses and herbivory vary more within plants than among plants in the tropical understory shrub Piper polytrichum. Botany  97: 113–121. [Google Scholar]
  61. Wetzel WC, Whitehead SR. 2020. The many dimensions of phytochemical diversity: linking theory to practice. Ecology Letters  23: 16–32. [DOI] [PubMed] [Google Scholar]
  62. Wetzel WC, Kharouba HM, Robinson M, Holyoak M, Karban R. 2016. Variability in plant nutrients reduces insect herbivore performance. Nature  539: 425–427. [DOI] [PubMed] [Google Scholar]
  63. Wimp GM, Martinsen GD, Floate KD, Bangert RK, Whitham TG. 2005. Plant genetic determinants of arthropod community structure and diversity. Evolution  59: 61–69. [PubMed] [Google Scholar]
  64. Yamasaki M, Kikuzawa K. 2003. Temporal and spatial variations in leaf herbivory within a canopy of Fagus crenata. Oecologia  137: 226–232. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mcaa101_suppl_Supplementary_Figure_A1
Click here for additional data file. (219.1KB, pptx)
mcaa101_suppl_Supplementary_Table_A1
Click here for additional data file. (13.1KB, docx)
mcaa101_suppl_Supplementary_Table_A2
Click here for additional data file. (14.4KB, docx)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES