Does insect herbivory suppress ecosystem productivity? Evidence from a temperate woodland

Our current understanding of the relationship between insect herbivory and ecosystem productivity is limited. Previous studies have typically quantified only leaf area loss, or have been conducted during outbreak years. These set-ups often ignore the physiological changes taking place in the remaining plant tissue after insect attack, or may not represent typical, non-outbreak herbivore densities. Here, we estimate the amount of carbon lost to insect herbivory in a temperate deciduous woodland both through leaf area loss and, notably, through changes in leaf gas exchange in non-consumed leaves under non-outbreak densities of insects. We calculate how net primary productivity changes with decreasing and increasing levels of herbivory, and estimate what proportion of the carbon involved in the leaf area loss is transferred further in the food web. We estimate that the net primary productivity of an oak stand under ambient levels of herbivory is 54 - 69% lower than that of a completely intact stand. The effect of herbivory quantified only as leaf area loss (0.1 Mg C ha−1 yr−1) is considerably smaller than when the effects of herbivory on leaf physiology are included (8.5 Mg C ha−1 yr−1). We propose that the effect of herbivory on primary productivity is non-linear and mainly determined by changes in leaf gas exchange. We call for replicated studies in other systems to validate the relationship between insect herbivory and ecosystem productivity described here.


76
In this study, we quantify the effects of insect herbivory on forest-level net primary 77 productivity in a temperate maritime woodland in southern England, UK. In a previous study 78 in this system (Visakorpi et al. 2018), we found that photosynthesis of oak (Quercus robur, 79 L.) was substantially lower in leaves subjected to herbivory by winter moth (Operophtera 80 brumata, L.) caterpillars than in intact leaves surrounded only by other intact leaves. 81 Moreover, a similar reduction in photosynthetic rate was seen in intact leaves on the same 82 shoots as the damaged leaves, resulting in an estimated 50% reduction in canopy-level 83 photosynthesis. How these changes in carbon assimilation affect tree and stand level 84 productivity remains unknown. Here, we combine our previous measurements of herbivory- Wytham Woods in which our study took place has been a woodland at least since the 18 th 101 century, and most likely since the last ice age (Savill 2011).

102
Our focal study area is an 18-ha forest dynamics monitoring plot, part of the    Table S3).

139
Estimating oak NPP. Figure 1 summarizes our approach and Appendix 1 provides more 140 detail on the methodology. In brief, we used three methods to estimate oak NPP. First, NPP 141 was estimated as the difference between canopy net photosynthesis and woody (stem + root) 142 respiration ("NPP through canopy upscaling"). Second, we used census data on woody 143 growth and combined these data with allometric equations to estimate leaf production and 144 belowground production ("NPP through tree growth census"). Third, we used earlier 145 estimates of oak above-and belowground NPP at the site obtained from biometric 146 measurements ("NPP through biometric measurements") (Fenn 2010. For 147 each of the three methods, we estimate oak NPP for a hypothetical stand comprising only of mature oak trees, assigning the total basal area (33 m 2 ha -1 ) of trees at the site to oaks only, 149 and for the actual per ha of oak of the site (oak basal area 6.7 m 2 ha -1 ). For each case, we 150 estimate oak NPP for two scenarios: 1) for an intact canopy (hereafter "Intact canopy") and rate was further scaled with the daytime photosynthetic rate of the previous day (Whitehead et al. 2004). To estimate woody respiration, we used data from previous measurements of oak 174 stem respiration at the site (Walker 2017). We assumed that stem respiration followed the 175 same pattern as leaf respiration in response to herbivory, since both leaf and stem respiration 176 rates have a similar, positive relationship with photosynthesis (Wertin and Teskey 2008).

177
Stem respiration was scaled to hourly temperature data. Root respiration was assumed to be 178 13% of stem respiration, based on earlier measurements at the site . Canopy  Table S2).

191
To estimate the effect of herbivory on oak NPP, we estimated NPP in the absence of 192 herbivory assuming that the total NPP under normal herbivory situation was reduced to the 193 same extent (ca. 56%, Appendix 1, Table S2) as canopy gas exchange in the canopy  Table S2). The biometric estimates 200 were based on dendrometer measurements (woody production), measures of leaf production 201 (litter traps) and of root production (soil respiration and inputs). We combined these 202 relationships for an estimate of whole oak NPP as Mg C yr -1 per ha of the actual study site.

203
To estimate the effect of herbivory on oak NPP, we again assumed that the total NPP was 204 reduced to the same extent as canopy gas exchange in the canopy upscaling calculations least one damaged leaf. After this, the proportion of damaged leaves was set to increase until 214 all leaves were damaged. Finally, we set the proportion of leaf area loss per leaf to increase, 215 until the tree was completely defoliated. In other words, we assumed that the herbivory first 216 spreads evenly to all shoots, then to all leaves, and then increases per leaf. LAI was assumed 217 to decrease as leaf area loss increased, increasing the amount of light reaching lower canopy 218 levels ( Figure S7, Appendix 2). We assumed a linear relationship between winter moth 219 caterpillar density and leaf area loss, with 5 individuals m -2 in 2015 (Lionel Cole, 220 unpublished data) corresponding to 5.9% leaf area loss (across the canopy) and to peak-221 season LAI of 6.5, and maximum reported density of 1200 individuals m -2 corresponding to 222 complete defoliation (Feeny 1970). We then estimated daytime canopy net photosynthesis with herbivory levels ranging from 0% to 100% leaf area loss. To test how sensitive the 224 estimated relationship between the level of herbivory and its effect on canopy assimilation 225 was to the assumptions of our models, we estimated the same relationship under four 226 alternative scenarios: 1) the difference in photosynthetic rate between intact and damaged 227 leaves is smaller than our field measurements suggest, 2) the photosynthetic rate of intact 228 leaves increases with increasing level of herbivory in the canopy, 3) herbivory spreads 229 through the canopy following a different spatial pattern than in our initial assumptions and 4) 230 a second leaf flush compensates for some of the leaf area loss. For details, see Appendix 3.

232
Results and discussion 233 We estimate that insect herbivores remove on average 0.12 (± 0.02) Mg C ha yr -1 in an oak  Figure S5). Depending on the method, we estimate the NPP of a 238 forest consisting of only mature oak trees to be between 3.4 (± 0.6; biometric measurements) 239 and 4.6 (± 1.1; tree census) Mg C ha -1 yr -1 under the observed level of herbivory (5.9% leaf 240 area loss, Table 2, Figure 3). This is between 54 ± 21% to 69 ± 42% lower than the NPP of a 241 completely intact canopy (  shown that herbivore outbreaks can result in a considerable loss of carbon from the 244 ecosystem, our study suggests that insect herbivores can have a large impact on the forest 245 primary productivity even at a low density. 246 We suggest that the relationship between the intensity of herbivory and its effect on

291
In our study, the largest effect of herbivory on NPP (69%) was estimated through the 292 canopy upscaling approach. This is most likely because stem and root respiration were 293 assumed to respond to herbivory similarly to leaf respiration, and because the difference 294 between the two scenarios (intact and with herbivory) was smaller for leaf respiration than for 295 photosynthesis. One of the biggest uncertainties in our calculations is how woody respiration 296 responds to leaf herbivory. This is a clear knowledge gap that needs to be addressed in future 297 studies. Had we assumed that a change in photosynthesis would not result in a change in stem respiration, or that stem respiration would increase after herbivory, our estimate for the effect 299 of herbivory on oak NPP would have been even larger. Thus, our current assumptions on the 300 response of woody respiration to herbivory provide a more conservative estimate on the 301 effect of herbivory on forest NPP than if we had assumed constant stem respiration or an 302 increased respiration after herbivory.    The fate of carbon contained in the leaf area lost to herbivory. Previous studies on the 398 patterns of leaf area loss and insect energetics allows us to estimate the pathways of carbon 399 once it is removed from the canopy. Based on previous studies on the amounts of greenfall 400 (falling leaf fragments due to herbivore feeding), roughly 75% of the lost leaf area is ingested 401 by the caterpillars . Of the ingested carbon, approximately 15% will be 402 respired, 60% will be turned into frass and transported into the soil, and the rest will turn into 403 insect tissue (of lepidopteran caterpillars; Wiegert and Petersen 1983). Based on estimated 404 predation rates of winter moth caterpillars and cadavers, most of this insect tissue will be 405 eaten by predators (East 1974). Thus, we estimate that 18% of the carbon contained in the 406 observed leaf area lost to herbivory (0.02 Mg C ha -1 yr -1 ) is likely to be transferred to higher 407 trophic levels in this system, and 70% of the leaf area loss (0.08 Mg C ha -1 yr -1 ) is directly 408 deposited in the soil, partly as greenfall and partly as frass ( Figure 5). Data on the amount of           Early-season moth caterpillars 6% -0.1 Leaf area loss NPP -net primary productivity, NEP -net ecosystem productivity, NBP -net biome productivity 708 709