Background invertebrate herbivory on dwarf birch (Betula glandulosa-nana complex) increases with temperature and precipitation across the tundra biome

Chronic, low intensity herbivory by invertebrates, termed background herbivory, has been understudied in tundra, yet its impacts are likely to increase in a warmer Arctic. The magnitude of these changes is however hard to predict as we know little about the drivers of current levels of invertebrate herbivory in tundra. We assessed the intensity of invertebrate herbivory on a common tundra plant, the dwarf birch (Betula glandulosa-nana complex), and investigated its relationship to latitude and climate across the tundra biome. Leaf damage by defoliating, mining and gall-forming invertebrates was measured in samples collected from 192 sites at 56 locations. Our results indicate that invertebrate herbivory is nearly ubiquitous across the tundra biome but occurs at low intensity. On average, invertebrates damaged 11.2% of the leaves and removed 1.4% of total leaf area. The damage was mainly caused by external leaf feeders, and most damaged leaves were only slightly affected (12% leaf area lost). Foliar damage was consistently positively correlated with mid-summer (July) temperature and, to a lesser extent, precipitation in the year of data collection, irrespective of latitude. Our models predict that, on average, foliar losses to invertebrates on dwarf birch are likely to increase by 6–7% over the current levels with a 1 °C increase in summer temperatures. Our results show that invertebrate herbivory on dwarf birch is small in magnitude but given its prevalence and dependence on climatic variables, background invertebrate herbivory should be included in predictions of climate change impacts on tundra ecosystems.


INTRODUCTION 95
The role of invertebrate herbivores in tundra ecosystems has been understudied (Haukioja 1981). Admittedly, the 96 proportion of herbivore taxa among invertebrates is lower in Arctic regions than at lower latitudes (Danks 1986), 97 and invertebrate herbivores generally occur at relatively low abundances in tundra (Haukioja 1981). However, 98 outbreaks of invertebrate herbivores have been well documented in the forest-tundra ecotone (Jepsen et al. 2008; 99 Kaukonen et al. 2013 internally feeding herbivores as they provide safer shelter against pathogens and reduce levels of desiccation 154 (Carneiro et al. 2005). We propose that the same distinction between external and internal feeders will drive 155 differences in the patterns of invertebrate herbivory in tundra.

156
Temperatures and precipitation are predicted to continue increasing in the Arctic (Cook et al. 2014), and warming 157 in tundra is expected to occur at a higher rate than the global average (IPCC 2013). The rapid pace of 158 environmental changes in the Arctic underscores the urgency of studying the responses of fundamental ecological 159 processes, such as herbivory, to varying climatic conditions. Insects living at higher latitudes are highly responsive 160 to climate changes (Hodkinson and Bird 1998), and warming-induced increases in insect herbivory are expected to 161 be stronger at higher latitudes (Wolf et al. 2008;Kozlov et al. 2015a

165
The objective of this study is to assess the intensity of background invertebrate herbivory and characterise its 166 relationships with latitude and climatic variables in tundra. To achieve this goal, we measured leaf damage by invertebrate herbivory by different feeding guilds will correspond with different climatic variables, given their 176 sensitivity to different environmental cues. Specifically, we expect leaf damage by externally-feeding defoliators to 177 be more strongly associated with summer temperature than damage by internally feeding herbivores (leaf miners 178 and gallers), and conversely that the latter will be more affected by climatic variables that determine leaf 179 toughness, such as precipitation. Alaska to Newfoundland, as well as the southern part of Greenland (Feilberg 1984

251
To investigate the effects of latitude and climatic variables on invertebrate herbivory in tundra we built Linear 252 Mixed Effects Models for total herbivory and for each feeding group separately. In all models, sampling protocol 253 (2008)(2009)(2010)(2011)(2012)(2013)2014 or 2015) was included as a random effect to account for potential confounding effects of year of 254 sampling, person scoring leaf damage and/or protocol design. Nearly half of the sampling locations (25 out of 62) 255 sampled one site only, so location could not be included in the models as a random factor; therefore, 256 measurements of invertebrate herbivory for locations with more than one site were averaged across sites, and the 257 number of sites sampled at each location was included as weights in the models to account for differences in 258 sampling effort.

9
We extracted the following indices as potential predictors of background herbivory: mean July temperature in the 260 year of sampling, annual temperature, temperature seasonality (standard deviation of annual temperature), 261 maximum temperature of the warmest month, minimum temperature of the coldest month, mean temp of 262 warmest quarter (Jun-Aug), total July precipitation in the year of sampling, annual precipitation and precipitation 263 seasonality (coefficient of variation). Initial correlation analyses indicated that mean July temperature and total

291
The vast majority of damage (98.6%) was caused by defoliators. Damage by internally feeding herbivores (leaf 10 miners and gallers) was found on relatively few leaves: 31 were mined by larvae of several moth species and only 293 24 bore galls (see Online Resource S6 for identification of mines and galls).

294
Total herbivory 295 Both the percentage of leaves with signs of invertebrate damage and the percentage of total leaf area damaged 296 were positively associated with July temperature and precipitation (Table 2a) (Figure 2b). Relative to observed current 309 levels of leaf area damaged (1.4%), these figures imply predicted increases of 6.7% in leaf area damaged by 310 invertebrate herbivores per degree C increase in mean July temperature, at locations with July temperature values 311 within the centre of the observed temperature range (mean July temperature observed across sites = 11.4° C).

312
The potential effects of increased precipitation followed similar trends, albeit a much weaker modelled effect than 313 temperature. With a 10 mm increase in July precipitation, the percentage of leaves damaged by invertebrate 314 herbivores increased by 0.3% in locations with the lowest observed precipitation (10.8 mm). In contrast, at 315 locations with the highest observed mean July precipitation measured in our study (136.3 mm), the model 316 estimated a 0.6% absolute increase in the percentage of leaves damaged (Figure 2c); the increase in the 317 percentage of leaf area damaged ranged between 0.05% and 0.12% in locations with drier and wetter summers 318 (Figure 2d). Relative to current levels of invertebrate herbivory, at sites with intermediate observed levels of July 319 precipitation (mean total July precipitation observed across sites = 53.2 mm), the models predicted a 3.6% relative 320 increase in the percentage of leaves damaged and 4.5% increase in percentage of leaf area damaged per 10 mm of 321 increased precipitation.

322
None of the covariates (birch taxa or collection date) included in the models for total herbivory were associated 323 with the percentage of leaves damaged at each location or with the percentage leaf area damaged ( Table 2a). The 11 average damage per damaged leaf was not associated with latitude, temperature, precipitation, collection date or 325 birch taxa (Table 2a).

327
The distribution of damage by free-living defoliators within a site, as measured with the percentage of leaves 328 damaged was associated with higher July temperature and precipitation (Table 2b), but no latitudinal pattern was 329 apparent. When looking at foliar loss, the percentage of leaf area affected by defoliators was positively, albeit 330 weakly, related to July precipitation and temperature ( Table 2b). None of the covariates explained variation in on 331 the percentage of leaves damaged by defoliators at each location or the percentage of leaf area damaged (Table   332 2b). On average, defoliators consumed 11.09 ± 1.26% of leaf area on damaged leaves (median = 8.56%), and this 333 value was not associated with latitude, temperature, precipitation, collection date or birch taxa (Table 2b).

335
The mean percentage of leaves damaged by leaf miners at each location was 0.06 ± 0.02% and, when present, leaf 336 miners affected on average 11.77 ± 3.05% of leaf area. Galls were found on 0.08 ± 0.05% leaves per location, and 337 affected 35.78 ± 8.29% of the leaf area of damaged leaves (excluding 2 galled leaves with petiole galls). The 338 percentage of leaves damaged by leaf miners increased with July precipitation and collection date (

390
Our models predicted that changes in invertebrate herbivory in response to temperature and precipitation will 391 differ along the range of climates sampled. It must be kept in mind that our approach represents a space-for-time 392 substitution, where we infer changes in herbivory from locations with different climatic variables. Despite its 393 limitations, this approach provides the best solution given the virtual lack of long-term trend data in patterns of 394 invertebrate herbivory in tundra over time. Given that climate models project warming of 6-10 degree C over the 395 next 100 years (IPCC 2013), the influence of temperatures on invertebrate background herbivory could be 396 important. According to the logarithmic relationship indicated by our models, increases in invertebrate herbivory 397 in locations with higher summer temperatures would be more pronounced than at locations with colder summers.

398
The effect of precipitation followed similar trends but was not as pronounced and did not differ as much between 399 the ends of the precipitation gradient. Precipitation is predicted to increase in the Arctic as a result of climate