Effects of temperature of plant cultivation on plant palatability modify species response to novel climate

Climate warming is expected to strengthen the plant-herbivore interactions and thus increase the plant consumption rate. However, indirect impacts of temperature (acting via changes in host plant quality) on herbivore performance have only rarely been studied, and therefore, the net effect of temperature change is difficult to predict. We thus tested the effects of temperature on plant palatability and assessed whether the effects can be explained by changes in leaf traits. We conducted multi-choice feeding experiments with six species of the genus Impatiens cultivated at three different temperatures in the growth chambers and in the experimental garden and also studied changes in leaf morphology and chemistry. The leaves of Impatiens species were most eaten when cultivated in the temperature predicted by climate warming scenario. We found the traits related to leaf morphology (SLA, LDMC and leaf size) partly mediated the effects of temperature on leaf herbivore damage. Herbivores preferred smaller leaves with lower SLA and higher LDMC values. Results of our study suggested that elevated temperature will lead to change in leaf traits and increase their palatability. This will further enhance the levels of herbivory caused by increased herbivore pressure under climate warming.


Introduction 35
Insect herbivores are one of the most important drivers of performance of plant populations 1,2 . Under 36 upcoming global climate change, increase in temperature is expected 3 . The general effects of global 37 warming on insects are relatively well documented 4 . Higher temperatures are predicted to increase 38 insect population densities, cause alterations in their body size, genetic composition, duration of life 39 cycles or exploitation of host plants 5 . Warming is thus expected to strengthen various herbivore-plant 40 interactions 5,6 . However, these predictions do not take into account that increased temperature might 41 also affect the plant traits that determine their palatability to the herbivores 7 . 42 Plants are growing along a wide range of biotic and abiotic conditions 8 and variation in herbivore 43 damage largely depends on such conditions 9-11 . The significant relationship between the plant damage 44 by herbivores and the environmental conditions was shown in many previous studies. Plants tend to 45 suffer higher herbivore damage in wetter, nutrient richer and more shaded habitats 12-14 , i.e., in 46 habitats in which they often grow more vigorously as suggested by Plant vigor hypothesis 15 . Increased 47 leaf quality also contributes to greater levels of herbivory in nutrient richer habitats 7,16 . Leaves of 48 plants growing in benign conditions habitats often also have higher specific leaf area, lower tissue 49 density, thinner leaf lamina and weaker veins 17 and are less tough 18 , which makes them more 50 palatable to the herbivores. On the other hand, environmental stress can reduce plant resistance to 51 herbivores making them more palatable according to Plant stress hypothesis 19 . For example, 20 52 demonstrated that drought stressed plants had fewer secondary metabolites and were consequently 53 more affected by some of their herbivores. 54 The effects of temperature on plant palatability are often studied along altitudinal or latitudinal 55 gradients 21,22 . More palatable plants at higher altitudes/latitudes suggest lower palatability at higher 56 temperatures 7 . The plants at lower altitudes/latitudes might be better defended 10,23 , produce leaves 57 morphologically harder to consume (with low specific leaf area or with trichomes 24 ) or have lower 58 nitrogen and phosphorus leaf content 25 . However, most of the above described studies are from 59 (4°C) until germination. Triplets of germinating seeds of the same species were transplanted into 5 x 5 132 x 8.5 cm pots filled with a mixture of common garden soil and sand (1:2) and placed to the three growth 133 chambers (Vötch 1014) differing in their temperature regimes. After two weeks, the seedlings were 134 weeded to keep only one seedling per pot. There were five individuals of each of the six Impatiens 135 species in each of the three growth chambers, i.e. 90 individuals in total. The temperature regimes 136 were set to represent the present and future temperatures at localities where Impatiens species 137 naturally grow in their native range in Nepal. Temperature regimes were set as follows:  The leaves were individually fresh weighted and scanned both before and after the herbivory (e.g. 48 ). 179 Leaf area was estimated using ImageJ software (version 1.52a, Java 1.8.0_112, Wayen Rasband, U.S. 180 National Institutes of Health, Bethesda, MD, USA; website: http://rsb.info.nih.gov/ij/download.html).

9
After herbivory, the leaves were dried to a constant weight and weighted again. This information, 182 together with information on leaf size, was used to calculate specific leaf area (SLA; mm 2 mg −1 dry 183 mass) and leaf dry matter content (LDMC; g dry mass g −1 fresh mass) for each leaf. Leaves were evenly 184 eaten by herbivores and no leaf parts were preferred. In two cases the whole leaf was eaten. Then we 185 used mean SLA and LDMC values of the other nine leaves from respective temperature regime. We 186 also analyzed total carbon (C), nitrogen (N) and phosphorus (P) content in leaf biomass. Since it was 187 not possible to analyze leaf nutrient content in the leaves used directly in multi-choice feeding 188 experiments due their small size, we used mixed sample of ten randomly chosen leaves per each 189 species and growth chamber temperature regime/common garden which were not exposed to Second, we tested how the leaf traits (SLA, LDMC, initial leaf area) are affected by the temperature 206 regime, the Impatiens species identity and their interaction using ANOVA. Differences in leaf nutrient 207 content were not tested since these data were not replicated within growth chambers and species 208 variant due to lack of the plant material available for analyses. All the traits were, however, used to 209 explain leaf herbivory (see below). Since leaf nutrient contents and their ratios were largely correlated, 210 we only used those which were not highly correlated with each other (r < 0.9). We thus only used 211 content of C and ratios of C:N, N:P and C:P. 212 Further, we explored the effect of leaf traits on leaf herbivory, either alone or after accounting for 213 species identity and temperature regime. For these tests, we added combination of species identity 214 and temperature regime as another random factor to account for the fact that the traits measured 215 within one species and temperature regime are not independent and in case of leaf nutrient content 216 they were only measured once. Usefulness of adding the traits into the models was assessed using AIC 217 criteria leading to identification of optimal model. 218 219 Experiment 2 220 The tests largely followed the logic described above in Experiment 1. First, we tested the effect of the 221 species identity, environment and their interaction on leaf herbivory. We used linear mixed effects 222 model with arena code as a random factor for this test. Second, we tested how the leaf traits differed 223 among the six Impatiens species and the two environments (growth chamber and common garden) 224 using ANOVA. We also explored the effect of leaf traits on leaf herbivory, either alone or after 225 accounting for species identity and environment. Optimal model explaining leaf herbivory was 226 constructed in the same way as in Experiment 1. 227 228

Effect of temperature and species identity on leaf herbivory 230
Results of Experiment 1 showed that leaf herbivory differed among the three temperature regimes (P 231 = 0.003, Fig. 1A) and among the six Impatiens species (P < 0.001, Online Resource 2). Moreover, there 232 was a significant interaction between species and temperature regime (P = 0.007, Online Resource 4A). 233 Leaves of five species were the most eaten in the warm2050 regime and most of them the least eaten 234 in the cold one. The only exception was I. balsamina with the highest herbivore damage when taken 235 from the cold temperature regime (Online Resource 4A). 236 In Experiment 2, leaf herbivory also differed among the six Impatiens species (P = 0.022) but was not 237 affected by the environment (P = 0.194) or the interaction between species identity and environment 238 (P = 0.236). I. balsamina and I. racemosa tended to be the least damaged by the herbivores (Fig. 2,  239 Online Resource 3). identity. The optimal model explaining leaf herbivory included SLA, LDMC and initial leaf area (Table  262 1A). Herbivores preferred smaller leaves but their response to SLA and LDMC was not consistent across 263 the six Impatiens species (Fig. 3). When the null model did not include temperature regime and species 264 identity, the optimal model included LDMC, initial leaf area and C:N ratio. SLA and the other leaf 265 nutrient contents were not included in the optimal models explaining variation in leaf herbivory (Table  266 1A). 267 Similarly to Experiment 1, SLA, LDMC and initial leaf area significantly contributed to explaining leaf 268 herbivory compared to the null model including only species and environment in Experiment 2. The 269 optimal model explaining leaf herbivory included all these three traits (Table 1B). Herbivores preferred 270 smaller leaves and leaves with smaller SLA and higher LDMC (Fig. 3). When null model did not include 271 species and environment, the optimal model only included SLA and LDMC. Leaf nutrient contents did 272 not explain any additional variation in leaf herbivory in both models in Experiment 2 (Table 1B). Global warming is expected to strengthen the plant-herbivore interactions and thus increase the plant 276 consumption rate especially by ectotherm omnivores 27,58 . However, indirect impacts of temperature 277 13 (acting via changes in host plant quality) on herbivore performance have only rarely been studied, and 278 therefore, the net effect is difficult to predict. Our study is one of the very few exploring both effects 279 of temperature of plant cultivation and leaf traits on herbivore damage. We found strong effect of 280 temperature on herbivore damage which can be partly explained by temperature effects on leaf traits 281 mainly related with leaf morphology such as SLA, LDMC and leaf size. We also showed that the effect 282 of elevated temperature on plant palatability strongly differ among the Impatiens species and 283 conclusions about the effects of climate warming must be done specifically for each species. 284 285

Effect of temperature on leaf herbivory 286
Results of our study suggest that plant palatability increases with temperature. This is in contrast with 287 previous study of 7 . They found that rising temperatures (15, 20 and 25°C) significantly decreased plant 288 palatability in aquatic plant Potamogeton lucens, which could be explained by changes in the 289 underlying leaf nutrient content (decreasing content of N and P in the leaves). However, they did not 290 find any effect of temperature on palatability in two other aquatic plants included in their study. The 291 absence of an apparent relationship between temperature and leaf nutrient content in our study could 292 explain contrasting pattern. Other studies on the effect of elevated temperature on herbivore damage 293 did not find any significant result at Salix myrsinifolia 59  It may be because the highest temperature is out of the species optima leading to changes in leaf 307 structure. LDMC showed similar but inversed pattern to SLA, which is in agreement with other studies 308 62,64,65 . We also found that leaf size decreased with increasing temperature, especially in warm2050 309 regime. This might be at least partly related to non-linear relationship between SLA/LDMC and 310 temperature. Larger leaves are not advantageous in elevated temperatures as they increase 311 transpiration area 66 . Surprisingly, we did not find any consistent effect of temperature on leaf nutrient 312 content as for example 25 , 67 and 7 who found lower nitrogen and phosphorus leaf content in plants 313 growing at higher temperatures. Impatiens seems to respond to elevated temperature by change in 314 leaf morphological structures while keeping the nutrient content in the leaves at the similar level. 315 However, more data are needed to confirm this as our leaf nutrient content data were not replicated 316 and thus could not be formally tested. Similarly to our study, meta-analysis of 27 showed no effects of 317 temperature increase on either nitrogen concentrations or C/N ratio and they suggested that not 318 nutrients but defense chemicals are the main reason for lower palatability of plants at elevated 319 temperatures. 320 321

Importance of leaf traits for leaf palatability 322
We demonstrated that herbivores prefer plants with lower SLA and higher LDMC. This contrasts with 323 the conclusions of several previous studies demonstrating strong positive effect of SLA on leaf 324 herbivory e.g. 34,68,69 . High SLA is primarily related to growth rate and resource acquisition, but it is also 325 expected to contribute to greater palatability to herbivores 68,70 . The difference in the results may be 326 caused by the fact that the previous studies used wider ranges of species with more variable SLA as 327 suggested by 32 who also found negative correlation between herbivore damage and SLA. They further 328 argued that the negative correlation might be due to leaf sampling throughout range of different 329 habitats which was not the case in our study when all the plants were grown in the same conditions 330 only differing in temperature. Very high SLA values at Impatiens species might be another reason for 331 non-positive relationship between SLA and herbivory. Compared to other studies (such as 32,34,68 ), 332 gradient of SLA started at higher values (indicating thin leaves) and the highest SLA values recorded at 333 our study (50 m 2 kg -1 ) probably are not attractive for herbivores any more due to being too thin. Thus, 334 the direction of the relationship between palatability and SLA may in fact be unimodal on large scale 335 and the direction detected in the different studies depends on the exact range of the SLA values. 336 Our results also indicate that herbivores preferred smaller leaves. This is in contrast with many other 337 studies which found that more vigorous plants suffered more leaf damage e. g. 71,72 as predicted by 338 Plant Vigour Hypothesis 15 . Similarly to our study, 73 found higher herbivory of gall forming insect at 339 smaller and not larger leaves. They suggested that smaller leaves should possess higher concentrations 340 of resources essential for larval development. However, we found higher nitrogen and phosphorus 341 content (and less carbon) in leaf biomass of larger leaves even though these relationships are based 342 on means at the population and environment level (Fig. S3). Negative relationship between leaf size 343 and extent of herbivore damage might be due to increase in concentration of substances decreasing 344 leaf palatability at larger leaves as found by 74 . They found that decrease in palatability at larger willow 345 seedlings was positively correlated with an increase in condensed tannin concentration. 75 also 346 suggested that young (i.e. small) leaves of Phytolacca americana may be more nutritious and less tough 347 than mature leaves explaining greater herbivory. 348 Even though ratios C:N and N:P were included in the optimal model explaining leaf herbivory, they did The only exception is I. balsamina, seeds of which were collected at 1330 m a. s. l., which is altitude 369 with temperatures simulated in warm2050 regime in our study. Arena and combination of environment and species were used as random factors in all the tests. 587 Asterisks indicate leaf traits included in the optimal model.