Research Collection

Doctoral Thesis

Functional Characterization of Plant-Herbivore Interactions and the Response of Alpine Plants to Climate Change

Author(s): Descombes, Patrice

Publication Date: 2018

Permanent Link: https://doi.org/10.3929/ethz-b-000304133

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library Functional Characterization of Plant-Herbivore Interactions and the Response of Alpine Plants to Climate Change

Patrice Descombes

DISS. ETH NO. 25269

DISS. ETH NO. 25269

FUNCTIONAL CHARACTERIZATION OF PLANT- HERBIVORE INTERACTIONS AND THE RESPONSE OF ALPINE PLANTS TO CLIMATE CHANGE

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH ZURICH

(Dr. sc. ETH Zurich)

presented by

PATRICE DESCOMBES

MSc in Behaviour, Evolution and Conservation, University of Lausanne

born on 29.05.1987

citizen of Lignières NE

accepted on the recommendation of

Prof. Dr. Loïc Pellissier Prof. Dr. Sergio Rasmann Prof. Dr. Konrad Fiedler

2018

- 2 -

TABLE OF CONTENTS

SUMMARY …………………………………………………………………………………… 5

RESUME ……………………………………………………………………………………… 7

INTRODUTION ……………………………………………………………………………… 11

CHAPTER 1 …………………………………………………………………………………... 35 Uneven rate of plant turnover along elevation in grasslands

CHAPTER 2 …………………………………………………………………………………... 65 Community‐level plant palatability increases with elevation as insect herbivore abundance declines

CHATPER 3 …………………………………………………………………………………... 115 Alpine plant palatability is associated with physical and chemical traits in situ and under a warming treatment

CHATPER 4 …………………………………………………………………………………... 155 Trophic conservatism predicts alpine plants responses to herbivore ecosystem incursion

CHATPER 5 …………………………………………………………………………………... 213 Simulated shifts in trophic niche breadth modulate range loss of alpine butterflies under climate change

CONCLUSION ……………………………………………………………………………….. 245

ACKNOWLEDGEMENTS ………………………………………………………………….. 261

CURRICULUM VITAE ……………………………………………………………………… 263

- 3 -

- 4 -

SUMMARY

Climate change drives rapid altitudinal and latitudinal range shifts of species worldwide. Insects, which have a higher sensitivity to thermal changes and greater ability to disperse, exhibit a faster response to climate change than less mobile organisms such as plants. Consequently, asynchronous migration rates between insects and plants might lead to reshuffled communities and new ecological interactions between species that never co-occurred. Herbivores, which occupy primary consumer positions in trophic chains, can have a major influence on ecosystem processes by altering the nutrient cycling and plant species assemblages. Therefore, modifications of plant- herbivore interactions and herbivory rates under climate change might have important consequences on plant communities, especially on high-elevation grasslands, which are generally less defended against herbivores. Changes in top-down herbivore pressure on high-elevation grasslands might favor plant community turnover and profoundly impact extant alpine grasslands. However, shifts in trophic interactions, such as those between insect herbivores and plants, have been poorly investigated and may play a primary role in how plant and insect herbivore communities will respond to climate change. Forecasting future changes in ecosystems under climate change requires a deeper understanding of the biological mechanisms shaping trophic interactions.

This thesis aimed at providing a better understanding on how plant and insect herbivore communities will respond to changes in trophic interactions under climate change. To tackle this question, this thesis focused on the plant-orthopteran and plant-butterfly bi-trophic networks, and combined several field experiments and statistical modelling techniques together with phylogenetic and functional trait information of the species. The first three chapters of this thesis investigated how plant communities are structured and defended against insect herbivores along an elevational gradient and tested the effect of temperature warming on plant functional traits. The fourth chapter experimentally simulated climate-driven insect herbivore incursion on alpine grasslands by translocating a lowland community of orthopteran herbivores on high elevation grasslands. This experiment was contrasted to a temperature warming treatment by using open-top chamber greenhouses, and the effect of both treatments on plant community composition was analyzed. The fifth chapter combined contemporary modelling techniques with phylogenetic and functional trait information to simulate trophic shifts and assess herbivore population persistence under future climate change.

Chapters 1 and 2 revealed a strong structuration of plant communities along the elevation gradient. In particular, the first chapter identified a transitional zone near the treeline with important rates of grassland compositional changes that delimits lowland and alpine ecosystems. Chapter 2

- 5 -

found a lower resistance of alpine plant communities to herbivores. Chapter 3 determined that plant-herbivore interactions are mostly driven by chemical profiles and physical leaf traits specific to plant families. Chapter 4 revealed a strong shift in the top-down selective pressure of translocated lowland herbivores on high-elevation plant communities due to different feeding behaviors between lowland and native insects on the recipient site. In particular, we found a phylogenetic and functional conservatism in the feeding behavior of insect herbivores when moved to higher elevation, resulting in predictable responses of plants to herbivore incursion. In addition, herbivore incursion induced a stronger effect than temperature warming on the plant community composition by increasing plant species richness and strengthening community dissimilarity. Finally, Chapter 5 revealed that dietary shifts may favor herbivore population persistence under future climate change. Taken together, the results suggest that insect herbivore incursion on high elevation grasslands might be a stronger driver of plant community turnover than temperature warming. This effect might be particularly strong near and above the treeline, where extant plant communities show higher rates of compositional turnover and an overall lower defence against herbivores.

This thesis contributed to a better understanding of the chemical and physical factors influencing plant-herbivore interactions. In particular, this thesis provides the experimental evidence for a diet conservatism during climate-driven range shifts of insect herbivores on high- elevation plant communities, resulting in predictable responses of plants to new herbivore incursions. Together, the predictability of reshuffled plant-herbivore interactions using functional and phylogenetic approaches has important implications for future research in ecological community modelling, as it allows to forecast future community trends under climate change. Possible shifts in trophic interactions should be carefully considered in future research aimed at understanding and predicting the dynamics of plants.

- 6 -

RESUME

Sous l’influence des changements climatiques actuels, les espèces migrent à de plus hautes altitudes et latitudes. Les insectes tendent à répondre plus rapidement aux changements climatiques que les espèces moins mobiles tel que les plantes, car ils disposent d’une plus forte sensibilité aux variations thermiques et une plus grande capacité de dispersion. Par conséquent, des migrations asynchrones entre insectes et plantes risquent de réorganiser les communautés et engendrer de nouvelles interactions entre espèces qui n’ont jamais été en contact auparavant. Les herbivores sont des consommateurs primaires dans les chaines trophiques. Ils peuvent avoir un impact très important sur les processus écosystémiques tel que les cycles nutritifs et les assemblages d’espèces de plantes. Ainsi, des modifications dans la nature et l’intensité des interactions trophiques sous l’impulsion des changements climatiques pourraient avoir de sérieuses conséquences sur les communautés de plantes, et plus particulièrement sur les prairies alpines qui sont généralement moins défendues contre les herbivores. Un changement dans la pression de sélection des plantes par les herbivores pourrait favoriser le remplacement de certaines espèces de plantes dans les communautés et profondément modifier les prairies alpines telles que nous les connaissons de nos jours. Cependant, les changements dans les interactions plantes-insectes n’ont été que rarement étudiés et pourraient jouer un rôle fondamental dans la réponse des communautés de plantes et d’incsectes herbivores aux changements climatiques. Afin de prédire les effets des changements climatiques sur les écosystèmes, une meilleure connaissance des processus biologiques influençant les interactions trophiques est nécessaire.

La présente thèse a pour but de fournir une meilleure compréhension de la réponse des communautés de plantes et d’insectes herbivores à des modifications dans les interactions trophiques sous l’effet des changements climatiques. Afin de répondre à cette question, cette thèse utilise les réseaux trophiques plantes-lépidoptères et plantes-orthoptères, ainsi que différentes expérimentations sur le terrain et des analyses statistiques combinant des informations sur les traits fonctionnels et les relations phylogénétiques entre espèces. Les trois premiers chapitres de cette thèse ont étudié comment les communautés de plantes sont structurées et défendues contre les herbivores le long d’un gradient altitudinal et analysé l’effet d’une augmentation de température sur les traits fonctionnels des plantes. Le quatrième chapitre a expérimenté et simulé l’effet de la colonisation des insectes herbivores sur les prairies de hautes altitudes. Pour cela, une communauté d’orthoptères herbivores de basse altitude a été déplacée dans des cages disposées sur des sites à plus haute altitude. Les changements dans la composition des communautés de plantes ont été analysés et contrastés à un traitement de température en utilisant des serres. Le cinquième chapitre a simulé des changements d’interactions trophiques en utilisant des informations phylogénétiques et

- 7 -

des traits fonctionnels afin d’évaluer leurs incidences sur la persistance des populations d’herbivores sous l’effet des changements climatiques.

Les chapitres 1 et 2 révèlent une importante structuration des communautés de plantes le long du gradient altitudinal. En particulier, le Chapitre 1 a permis d’identifier une zone the transition à proximité de la limite forestière avec d’importantes modifications compositionnelles, délimitant les écosystèmes de basse et de haute altitude. Le Chapitre 2 révèle que les communautés de plantes de haute altitude sont en moyenne moins défendues contre les herbivores. Le Chapitre 3 a permis de montrer que les interactions entre plantes et insectes herbivores sont principalement définies par le profil chimique spécifique de certaines familles de plantes ainsi que des traits physiques. Le Chapitre 4 révèle une forte modification de la pression sélective des herbivores de basse altitude déplacé sur les communautés de plantes alpines. Cette modification est due à des préférences différentes dans la diète entre les insectes de basse et de haute altitude. En particulier, cette étude montre que les préférences dans la diète des herbivores transférés en altitude sont fonctionnellement et phylogénétiquement conservées. Les nouvelles interactions entre plantes et herbivores ont induit un impact plus fort sur la composition des communautés de plantes que le traitement température en augmentant la richesse spécifique et la dissimilarité compositionnelle des communautés. Finalement, le Chapitre 5 a permis de révéler que des modifications dans la diète pourraient favoriser la persistance des populations d’insectes herbivores sous l’effet des changements climatiques. En conclusion, des modifications dans les interactions trophiques pourraient engendrer des modifications plus importantes dans les communautés de plantes alpines qu’une augmentation de température. Ces effets devraient être particulièrement importants au niveau et en-dessus de la limite forestière où les communautés de plantes sont généralement moins défendues contre les herbivores et présentes une forte dissimilarité compositionnelle.

L’ensemble de cette thèse contribue à une meilleure compréhension des facteurs chimiques et physiques influençant les interactions entres plantes et herbivores. En particulier, cette thèse démontre par évidence expérimentale un conservatisme de la diète des herbivores durant leur migration sur des communautés de plantes de haute altitude, résultant en des effets prédictibles sur les plantes. La prédictibilité des interactions plantes-herbivores en utilisant des informations fonctionnelles et phylogénétiques permet de prédire la réponse des communautés de plantes sous l’effet des changements climatiques. Ces résultats ont d’importantes répercussions dans les domaines de la recherche sur la dynamique des communautés et apportent de nouvelles perspectives dans la prédiction des communautés sous l’effet des changements climatiques. Les recherches visant à comprendre et prédire la dynamique des plantes doivent soigneusement considérer les nouvelles interactions trophiques.

- 8 -

- 9 -

- 10 -

INTRODUTION

Ecosystems under climate change

Ecosystems are under the influence of biotic (i.e., living organisms) and abiotic (i.e., chemical and physical, non-living components such as air temperature, water supply and nutrient availability in the soil) components that interact trough nutrient cycles and energy flows (Chapin et al., 2002; Tansley, 1935). Living organisms, such as primary producers and consumers, constantly interact and respond to their abiotic environment (Guisan and Thuiller, 2005). However, disturbances such as changes in abiotic (e.g., warming, precipitation) or biotic (e.g., species abundances, trophic interactions) properties can affect ecosystems by breaking the current state of equilibrium (Chapin et al., 2002; DeLucia et al., 2012). Such changes may have a profound effect on ecosystem functions and services to humans (de Groot et al., 2002), especially if ecosystems are not resistant or show low resilience to disturbances (Chapin et al., 2002). In this sense, climate change is currently driving rapid elevational and latitudinal range shifts of species worldwide (Chen et al., 2011; Parmesan and Yohe, 2003), leading to novel species assemblages (Williams and Jackson, 2007; Wing et al., 2005). This may, in turn, affect ecosystem processes. Insects exhibit strong and fast responses to thermal changes, which influence their development rates, reproduction potential and survival (Bale et al., 2002). Their higher sensitivity to thermal changes and greater ability to disperse leads them to respond faster to climate change through range shift than less mobile and longer-lived organisms such as plants (Berg et al., 2010; Rasmann et al., 2014; de Sassi and Tylianakis, 2012). Thus, asynchronous migration rates between insects and plants may lead to reshuffled communities and new ecological interactions in their new ranges (Berg et al., 2010; Rasmann et al., 2014). However, the consequences of those shifts on plant communities, such as between insect herbivores and their host plants, have been poorly investigated and may play a primary role in how plant communities and ecosystems will be impacted by climate change (Van der Putten et al., 2010).

Plant - herbivore interactions

Plant-herbivore interactions play a major role in regulating plant community structure, functions and processes (Belovsky and Slade, 2000; Duffy et al., 2007; Olff and Ritchie, 1998). Changes in those interactions can alter plant diversity (Borer et al., 2014; Kaarlejärvi et al., 2017; Olff and Ritchie, 1998), plant primary productivity (Blumer and Diemer, 1996; Carson and Root, 2000; Coupe and Cahill, 2003; Wardle et al., 2004), nutrient cycling (Belovsky and Slade, 2000; Blumer and Diemer, 1996; DeAngelis, 2012; Metcalfe et al., 2014; Nitschke et al., 2014; Seastedt

- 11 -

and Crossley, 1984), colonization, extinction and competition processes (Olff and Ritchie, 1998). Herbivores as a whole have been estimated to consume up to 20% of the terrestrial annual net primary productivity (Agrawal, 2011) and to strongly reduce the biomass of terrestrial and marine primary producers (Poore et al., 2012). By exerting a strong selective pressure on plants, they can profoundly affect plant community structure and succession (Brown, 1985; Brown and Gange, 1992; Gibson et al., 1987; Hulme, 1996a; del-Val et al., 2004), plant fitness and physiology (Bigger and Marvier, 1999; Hulme, 1996b; Karban and Strauss, 1993), and plant evolution (Holeski et al., 2010; Rasmann and Agrawal, 2011; Züst et al., 2012). The levels of herbivory generally differ between plant species (Tanentzap et al., 2011), suggesting that herbivores have feeding preferences and target specific plant species. Hence, herbivores may be limited in their capacity to feed on some plant species due to digestive (Ibanez et al., 2012) or biomechanical (i.e., mandibular strength) constraints (Ibanez et al., 2013). In addition, phylogenetically conserved interaction patterns between herbivores and plants (Pearse and Altermatt, 2013; Pellissier et al., 2013a) suggest that specific sets of plant functional traits drive the feeding choice of herbivores. Thus, plant-herbivore interactions, which are non-random in nature, might be mediated by evolutionary and functional characteristics (i.e., traits) (Violle et al., 2007). As climate change is expected to reshape plant- herbivore interactions (DeLucia et al., 2012), there is a need to characterize and understand which traits modulate extant trophic interactions.

Plant defence strategies

Because plants are sessile and unable to escape from herbivory, they have evolved a variety of strategies to protect themselves against herbivore attacks (Agrawal and Fishbein, 2006; Coley et al., 1985; Grime et al., 1968; Hanley et al., 2007; Mithöfer and Boland, 2012; Rhoades, 1979). Plants have a basal constitutive defence that is always expressed (Núñez-Farfán et al., 2007), but can also induce transient defences in response to pathogen or herbivore damage (Karban, 2011; Karban et al., 1999), or be tolerant to herbivory by compensating biomass loss with higher re-growth capacities, increased reproduction and photosynthetic rates, and changes in nutrient uptake and allocation (Carmona et al., 2010; Pratt and Mooney, 2013; Strauss and Agrawal, 1999). Constitutive and induced defences consist of physical (e.g., leaf toughness, trichomes, thorns, hairs or silica content) (Awmack and Leather, 2002; Brizuela et al., 1986; Hanley et al., 2007; Massey and Hartley, 2009; Massey et al., 2006) and chemical (e.g., alkaloids, terpenoids and phenolics coumpounds) plant traits (Becerra, 1997; Mithöfer and Boland, 2012; Salazar et al., 2018) affecting herbivore performance by decreasing leaf palatability and digestibility (Agrawal, 2007). Those defences can act directly on the fitness of herbivores (e.g., leaf toughness, thorns, hairs, carbon- and

- 12 - nitrogen-based compounds) (Agrawal, 2007) or indirectly by recruiting natural enemies such as predators and parasitoids (e.g., domatia, extrafloral nectar, chemical volatiles) (Heil, 2008; Kessler and Heil, 2011; Paré and Tumlinson, 1999; Rasmann and Turlings, 2007; Turlings et al., 1990). In parallel, herbivores also counter-adapt to plant defences by developing various strategies such as avoidance, excretion, sequestration or degradation of toxins (Ibanez et al., 2012). This race between co-evolutionary arms might have driven the co-diversification of plants and insects (Berenbaum, 2001; Edger et al., 2015; Ehrlich and Raven, 1964; Speed et al., 2015). Together, all those traits act in concert in the form of syndromes (Agrawal and Fishbein, 2006). However, the level of physical and chemical defences displayed may largely vary among plant species and depend on the growing strategy and the ecological context (e.g., abiotic conditions, herbivore pressure) influencing the relative costs and benefits (i.e., optimal defence theory) (Coley et al., 1985; Rhoades, 1979). Studies aiming at identifying the traits driving herbivory are very diverse, but principally focus on a few commonly used functional traits (e.g., SLA, LDMC, nitrogen content) (Descombes et al., 2017; Hanley et al., 2007; Massey et al., 2006; Peeters et al., 2007; Pérez‐Harguindeguy et al., 2003). The multidimensional nature of trait syndromes in plant-herbivore interactions calls for a multivariate approach of traits (Loranger et al., 2012), which should ideally combine physical, chemical and phenological traits. Because of the diversity of chemical secondary metabolites in plants (Mithöfer and Boland, 2012; Rhoades, 1979) and the technical challenge for quantifying and documenting them for large samples of species, it remains unclear whether the chemical signature of plants is associated to herbivore performance across a wide range of plant species and plant families. Recent developments in analytical chemistry and bioinformatics (e.g., untargeted metabolomics) provide new insights into plant-herbivore interaction research and could be used to assess herbivore performance for a large range of species (Coley et al., 2018; Kergunteuil et al., 2018; Richards et al., 2015; Salazar et al., 2018). Co-occurring herbivores in ecosystems might shape plant communities via differential selective pressure on plants, which will be modulated by specific plant functional trait preferences.

Impact of herbivores on plant communities

Herbivores can have profound and variable impacts on plant communities by altering plant colonization, extinction and competition processes (Kim et al., 2013; Olff and Ritchie, 1998), plant primary productivity (Blumer and Diemer, 1996; Carson and Root, 2000; Coupe and Cahill, 2003; Wardle et al., 2004) or nutrient cycling (Belovsky and Slade, 2000; Blumer and Diemer, 1996; DeAngelis, 2012; Metcalfe et al., 2014; Nitschke et al., 2014; Seastedt and Crossley, 1984). Hence, empirical evidence suggests that herbivores can increase or decrease plant community richness and

- 13 -

that the direction of this effect depends on the feeding behavior of the herbivore (Hulme, 1996b; Olff and Ritchie, 1998) and the growth form of the plant (Borer et al., 2014). A preferential herbivory on rare plants should hinder plant fitness and strengthen the competitive advantage of the dominant plant species in the plant community, resulting in a decrease of species richness (Olff and Ritchie, 1998). For instance, the geographical range of the rare perennial plant Arnica montana has been suggested to be constrained by selective herbivory (Bruelheide and Scheidel, 2001; Scheidel and Bruelheide, 2001). In contrast, a preferential herbivory on dominant plants should stabilize species richness (Hulme, 1996b; Long et al., 2003; Mortensen et al., 2018). The reduced biomass and competitive ability of dominant plants increases light availability at ground level, which favors the establishment of short-stature and new plant species (Borer et al., 2014; Kaarlejärvi et al., 2017). On the other hand, species richness should be stable if all plants show proportional herbivory rates. In addition, herbivores can also favor plant biomass by speeding up nutrient cycling via faster decomposition rates of plant litter and concentration of nutrients in rapidly decomposable faeces and body tissues (Belovsky and Slade, 2000; Blumer and Diemer, 1996). Nevertheless, the positive effect of herbivory on plant productivity may only be possible if herbivore consumption does not exceed the biomass gained through the enhancement of nutrient cycling (Belovsky and Slade, 2000). The effect of herbivores on plant communities has been widely investigated by manipulating herbivore communities, principally through herbivore exclusion or removal (Borer et al., 2014; Fraser and Madson, 2008; Kaarlejärvi et al., 2017), as well as manipulation or addition (e.g. Bruelheide and Scheidel, 2001; Deraison et al., 2014). Those studies were principally conducted on closed systems involving local plant and herbivore species. Hence, studies investigating novel trophic interactions (Pearse and Altermatt, 2013) and their effect on plant communities are scarce (Bruelheide, 2003). Because climate change is expected to modify trophic interactions via asynchronous range shifts of species (Berg et al., 2010; Rasmann et al., 2014; Van der Putten et al., 2010), and since herbivores can have profound and variable impact on plant communities, there is an urgent need to account for novel arising interactions between resident and range-shifting species in community ecology. In this sense, manipulative experiments are necessary to validate hypotheses and further allow predictions of community changes under climate change.

Plant - herbivore interactions under climate change

Warming has been shown to increase herbivory rates, especially at higher elevation, through an increase in insect herbivore abundance (Rasmann et al., 2014), performance (Lemoine and Burkepile, 2012; Lemoine et al., 2013; O’Connor, 2009; Zvereva and Kozlov, 2006) and additional generation within a season (Altermatt, 2010; Bale et al., 2002; DeLucia et al., 2012). This will

- 14 - irremediably promote biomass removal, but show contrasting outcomes on species richness, as mentioned previously, if the target species are rare or abundant (Hulme, 1996b; Kaarlejärvi et al., 2017; Long et al., 2003; Mortensen et al., 2018; Olff and Ritchie, 1998). While consumption rate and herbivore performance have been shown to be favored by temperature warming (Lemoine et al., 2013; Zvereva and Kozlov, 2006), climate warming and water stress might also alter the plant phenology, reproductive success, growth, and phenotypic selection (Hochachka and Somero, 2002; Totland, 1999), which might in turn affect the efficacy of plant defences against herbivores (Coley et al., 1985; Gutbrodt et al., 2011; Veteli et al., 2002). Hence, considerable variation in herbivore performance has been observed among different plant species subjected to warming (Lemoine et al., 2013, 2014). For instance, Braschler and Hill (2007) found that warming decreased the survival of Polygonia c‐album caterpillars, while Chong et al. (2004) observed opposite responses of Phenacoccus madeirensis. Thus, warming induces stress responses in plants (Melillo et al., 2002), which are likely species-specific (Bidart‐Bouzat and Imeh‐Nathaniel, 2008; Gutbrodt et al., 2011) and might exacerbate or impair insect herbivore performance by altering plant palatability (Dury et al., 2002; Evans and Burke, 2013; Gutbrodt et al., 2011; Zvereva and Kozlov, 2006). In addition, climate change might also favor the dispersal or growth of plant species with high anti-browsing properties, thereby deteriorating pasture quality and negatively affecting herbivore performance (Fauchald et al., 2017). Due to the inconsistent responses of plants to warming, manipulative experiments of climate warming on larger sets of species and different plant functional types (i.e., evergreen or deciduous shrubs, forbs, grasses and sedges) are necessary to understand and forecast how climate change might affect plants and reshape plant-herbivore interactions (Hudson et al., 2011).

Forecasting the effect of climate change on plant - herbivore interactions

Climate change is currently driving range shifts of species tracking favorable environments in the landscape to higher altitude and latitude (Chen et al., 2011; Devictor et al., 2012; Parmesan, 2006; Parmesan and Yohe, 2003; Schweiger et al., 2012; Walther et al., 2002). An elevational shift of up to 300 m has been already observed between 1967 and 2005 (Merrill et al., 2008; Wilson et al., 2007), and future projections under expected climate change predict a further elevational shift of 650 m by 2100 (Merrill et al., 2008). The assumption that the range distribution of species is constrained by their environmental conditions fostered the development of niche-based models (e.g., species distribution models), which relate species occurrences or abundances to ecological conditions (Guisan and Thuiller, 2005). Niche-based models allow the prediction of species distribution in a spatial context under current and future climate conditions (Thuiller et al., 2005).

- 15 -

However, models including only abiotic predictors in future forecasts have been criticized because they oversimplify the estimation of species distribution (Davis et al., 1998; Guisan and Thuiller, 2005; Wisz et al., 2013). Biotic interactions play a primarily role in how species will be affected by climate change (Araújo and Luoto, 2007; Eskildsen et al., 2015; Pellissier et al., 2012a; Schweiger et al., 2012). For instance, herbivores are frequently specialized in a restricted number of host plants, as they evolved detoxification mechanisms against their secondary metabolites (Becerra, 1997). In turn, asynchronous range shifts between host plants and specialized herbivores under climate change (Berg et al., 2010) might cause herbivore population declines by reducing their ability to colonize new geographical ranges. Alternatively, ecological or evolutionary shifts in host plant use may also occur and provide a possible mechanism for herbivore population persistence and expansion (Hódar et al., 2003; Pateman et al., 2012). For instance, the brown argus butterfly (Aricia agestis) shifted its diet from its historic host plant Helianthemum nummularium to Geranium molle, a more widespread species, which facilitated the rapid range expansion of the butterfly (Pateman et al., 2012). Therefore, biotic interactions and potential shifts in trophic interactions should be considered when modelling the response of species to climate change.

In addition, the characterization of trophic interactions in ecological networks, such as between plants and insect herbivores (butterflies, grasshoppers, etc.), remained mainly descriptive, and only few studies have attempted to predict the interactions based on functional traits and phylogenies (Loranger et al., 2012; Pearse and Altermatt, 2013; Pearse et al., 2013; Pellissier et al., 2013b). For instance, Pearse and Altermatt (2013) accurately predicted the majority of Lepidoptera interactions with non-native host plants by using information about the phylogenetic relatedness of plants and a native ecological network. In addition, phylogenetically conserved interaction patterns between herbivores and plants (Becerra, 1997; Pearse and Altermatt, 2013; Pellissier et al., 2013a) suggest that specific sets of plant functional traits drive the feeding choice of herbivores. While functional traits and phylogenies have proven their relevance for modelling trophic interactions, studies predicting new interactions in ecological networks affected by climate change (e.g., new incursions of herbivores from lower elevation or latitude) are missing.

Elevation as a laboratory to understand climate change impacts

Elevation gradients, which have stimulated research in ecology for centuries (Bonnier and Flahault, 1878; von Humboldt, 2009; Körner, 2003, 2007), represent unique natural laboratories for studying the ecological and evolutionary responses of ecosystems to climate change (Körner, 2003, 2007; Rogora et al., 2018). Abiotic factors rapidly vary along elevation gradients and over short horizontal distances (Körner, 2007). Increasing elevation is generally associated with an increased

- 16 - variability in climatic conditions, a drop in temperature of 0.55 K per 100 m of altitude (Barry, 1992), a reduced growing season (in non-tropical regions), and higher exposure to UV-B solar radiations and wind (Körner, 2007). The variation of abiotic factors along the elevation gradient influences the survival and reproduction of species, resulting in a constrained species range distribution within delimited abiotic conditions (Hutchinson, 1957; Soberón, 2007). As a result, strong compositional changes of species have been observed along the elevation gradient in plants (de Bello et al., 2012; Gentry, 1988; Pellissier et al., 2010), animals (Kergunteuil et al., 2016; Pellissier et al., 2012b; Sanders, 2002) or microorganisms (Pellissier et al., 2014a). Because they endure stressful abiotic conditions (e.g., strong wind, ground instability, frost during the growing season), high-elevation plant communities are generally functionally distinct and characterized by the presence of small-stature plant species with slow growth rates (Diaz and Cabido, 2009; Körner, 2003). In addition, harsh environments have also strong physiological constraint on ectothermic organisms (Buckley et al., 2013; Dillon et al., 2006), such as insects herbivores, leading to strong differences in species richness and abundances along the elevation gradient. Hence, herbivore communities are generally more diverse and abundant in warmer conditions (Descombes et al., 2017; Pellissier et al., 2014b; Rasmann et al., 2014; but see Kergunteuil et al., 2016). It is, therefore, expected that the degree of herbivory generally increases towards lower elevation or latitude (Galmán et al., 2018; Garibaldi et al., 2011; Hargreaves et al., 2018; Hülber et al., 2015; Metcalfe et al., 2014; Pellissier et al., 2014b; Pennings et al., 2009; Reynolds and Crossley, 1997; Scheidel et al., 2003; Zhang et al., 2016), despite recent doubts against the generality of this pattern (Moles and Ollerton, 2016; Moreira et al., 2017, 2018 and references herein).

On the whole, the variation of abiotic and biotic factors along the elevation gradient is recognized to affect the expression of plant functional traits (Linhart and Grant, 1996), which may in turn influence plant-herbivore interactions. Hence, variations in plant functional traits in response to harsh climate conditions along the elevation gradient may indirectly confer increased resistance to herbivores. For instance, harsh climate conditions (e.g., wind, frost) favor plants with higher leaf toughness (Körner, 2003), which may decrease leaf palatability to herbivores due to biomechanical constraints (Ibanez et al., 2013). Secondary metabolites involved in protection against UV-B solar radiations (e.g., flavonoids) may also indirectly confer anti-herbivore resistance (Close and McArthur, 2002). In response to the lower herbivore pressure at high elevation, and because plant defences are costly (Gulmon and Mooney, 1986), alpine plants have been shown to invest less resources in anti-herbivore defences, resulting in lower resistance to herbivores (Pellissier et al., 2012b, 2014b). High elevation areas represent refuges for some palatable alpine plants which do not support high levels of herbivory (e.g., Arnica montana) (Bruelheide and Scheidel, 2001;

- 17 -

Bruelheide, 2003; Galen, 1990; Scheidel and Bruelheide, 2001). However, palatable alpine and endangered rare plants may suffer from forecasted new plant-herbivore interactions and increased herbivory under climate change (Bruelheide and Scheidel, 2001; Bruelheide, 2003; Galen, 1990; Maze, 2009; Rasmann et al., 2014). So far, only a few studies have evaluated the effect of changes in trophic interactions and herbivore abundance on alpine plant communities (Bruelheide, 2003). Because mountain ecosystems are expected to be strongly affected by climate change (Pauli et al., 2012; Rogora et al., 2018), manipulative experiments of plant or herbivore communities along the elevation gradient might provide new insights to predict the effects of climate change on species- rich alpine plant communities. Unravelling the consequences of climate change on mountain ecosystems is of particular importance, as they are centers of plant diversity, including many endemic plant species and specialized vascular plants (Barthlott et al., 1996; Dirnböck et al., 2011; Myers et al., 2000).

Thesis outline

This thesis aimed at providing a better understanding on how plant and insect communities will respond to changes in trophic interactions under climate change. Specifically, I investigated how plant communities are structured (Chapter 1) and defended against insect herbivores along the elevation gradient at the species and community level (Chapter 2 and 3), and how plant defences may respond to climate warming (Chapter 3). Second, I experimented the effect of climate-driven range shifts of insect herbivores from lowland on high-elevation alpine plant communities (Chapter 4). Finally, I simulated changes in plant-herbivore interaction networks under future climate change and assessed herbivore survival (Chapter 5). Unravelling the consequences of altered trophic interactions on plants will allow us to understand and forecast future responses and turnover of plant communities under climate change.

This thesis relies on data collection and field experiments performed in the Western Alps of Switzerland (Fig. 1a). The study area exhibits an elevational gradient ranging from 375 m to 3200 m, a calcareous bedrock, and displays a temperate climate (mean annual temperature: 8 °C at 375 m and -5°C at 3200 m; annual sum of precipitations: 1200-2600 mm; Bouët, 1985). In the lowland and below 800 m, most of the open areas are urbanized or used for pastures and agriculture (e.g., vineyards, crops; Fig. 1a-b). Between 800 m and 1800 m, human activities are mostly concentrated around small localities, and open areas are used for hay meadows and pastures (Fig. 1c). Above the treeline (approx. 1800-1900 m), the highlands are occupied by species-rich subalpine and alpine grasslands (Fig. 1d-e), rock cliffs, screes and glaciers. In the subalpine and alpine belt, human

- 18 - disturbances are very low and cattle grazing occurs only during a few months (July-September) in the most accessible areas.

The focus of this thesis lays on plant and insect herbivore communities along the elevation gradient. In particular, I used the plant-orthoptera and the plant-butterfly bi-trophic networks; plants being placed at the first level of the trophic chain and grasshoppers and butterflies holding the lead roles in herbivore plant damage. Insect herbivores, such as grasshoppers, can reach very important densities in grasslands (up to 140 individuals per square meter) (Miao et al., 2018; Onsager, 1991) and have important effects on aboveground plant productivity, composition and nutrient cycling (Blumer and Diemer, 1996). Their well-known sensitivity to both plant physical and chemical defences makes them perfect candidates for investigating the mechanisms regulating plant- herbivore interactions.

Figure 1. (a) Location of the study area in the western Alps of Switzerland. The green areas represent forest ecosystems and the dark grey line shows the 800 m isoline. (b-e) The study area exhibits a strong elevation gradient ranging from 375 m (colline belt) to 3200 m (nival belt). (e) Subalpine and alpine grasslands display a rich flora and fauna. On the pictures: Primula auricula, Primula farinosa, Dracocephalum ruyschiana, Parnassius apollo. Photo credit: P. Descombes, ETH Zürich, Switzerland.

- 19 -

The thesis is structured as follows:

Chapter 1 addressed how plant communities are structured along an elevational gradient. I investigated changes in species richness, and taxonomic and phylogenetic turnover of plant species assemblages by using 900 plant inventories performed between 400 m and 3200 m (Fig. 1a). This study allowed to determine the rate of grassland compositional changes and the detection of transitional zones along the elevation gradient.

Chapter 2 investigated the variation in plant species and community level defences along an elevational gradient, which might vary in response to relaxed in situ herbivory and changes in abiotic conditions. To this aim, I collected 172 plant species along an elevational gradient (Fig. 1a) and assessed their palatability to a highly polyphagous chewing insect herbivore commonly used in bioassay experiments (Spodoptera littoralis). Species-level and community-level (i.e. community weighted mean of the 900 plant inventories used in Chapter 1) plant palatability were related to elevation and plant functional traits (e.g., SLA, LDMC, C:N) measured for the 172 species in the study area. Changes in plant palatability were finally associated to variation in orthopteran herbivore pressure along the elevation gradient. This study allowed to identify plant traits associated to herbivore performance and plant anti-herbivore defence relaxation in some parts of the elevation gradient.

Chapter 3 explored the contribution of plant leaf physical and chemical functional traits in mediating herbivore performance, both in situ and in a bioassay. To this purpose, I assessed the rate of natural herbivory on plants and measured plant palatability using Spodoptera littoralis caterpillars as a bioassay for all plant species occurring in three high elevation grasslands in the Swiss Alps (1800 m, 2070 m and 2270 m; Fig. 1a). In situ herbivory and caterpillar performance were associated to a set of leaf physical traits and axes of ordination of chemical signature. Furthermore, I assessed the effect of climate warming on leaf physical and chemical profiles together with leaf caterpillar performance by artificially altering the in situ abiotic conditions with open-top chamber (OTC) greenhouses. This study allowed to identify plant functional traits associated to herbivore performance (in situ and in a bioassay) and plant functional traits affected by climate warming, which might potentially reshape plant-herbivore interactions.

Chapter 4 experimentally simulated the effect of the climate-driven ecosystem incursion of herbivores from lower elevation on alpine plant communities in the Swiss Alps (Fig. 1a). I translocated a representative density of orthopteran herbivores from lower elevation (1400 m) into cages placed on three high elevation grasslands (1800 m, 2070 m and 2270 m) and evaluated the compositional changes in plant communities between inventories performed in 2014 and 2017.

- 20 -

Furthermore, I contrasted changes in herbivore pressure to a direct effect of temperature by modifying the in situ abiotic conditions with open-top chamber (OTC) greenhouses. This study provides new insights into the potential impact of range shifting herbivores on alpine grasslands, which might reshape plant communities and represent a stronger driver of ecosystem modification than temperature.

Chapter 5 assessed whether adjustments of trophic interactions via diet expansion may provide a mechanism for population persistence of host plant-specialized butterfly species in the Swiss Alps (Fig. 1a). To this aim, I modelled the present and future distribution of 60 butterfly and 298 plant species by using species distribution models, constrained the butterfly abiotic distribution with their associated host plant abiotic distribution, and allowed the progressive inclusion of new host plant species into the butterfly diet by using both a phylogenetic and a trait-based approach. This study will reveal if ecological or evolutionary shifts in host plant use may provide a possible mechanism for population persistence under future climate change.

References

Agrawal, A.A. (2007). Macroevolution of plant defense strategies. Trends in Ecology & Evolution 22, 103–109. Agrawal, A.A. (2011). Current trends in the evolutionary ecology of plant defence. Functional Ecology 25, 420–432. Agrawal, A.A., and Fishbein, M. (2006). Plant Defense Syndromes. Ecology 87, S132–S149. Altermatt, F. (2010). Climatic warming increases voltinism in European butterflies and moths. Proceedings of the Royal Society of London B: Biological Sciences 277, 1281–1287. Araújo, M.B., and Luoto, M. (2007). The importance of biotic interactions for modelling species distributions under climate change. Global Ecology and Biogeography 16, 743–753. Awmack, C.S., and Leather, S.R. (2002). Host plant quality and fecundity in herbivorous insects. Annual Review of Entomology 47, 817–844. Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer, T.M., Brown, V.K., Butterfield, J., Buse, A., Coulson, J.C., Farrar, J., et al. (2002). Herbivory in global climate change research: direct effects of rising temperature on insect herbivores. Global Change Biology 8, 1–16. Barry, R.G. (1992). Mountain Weather and Climate (Psychology Press). Barthlott, W., Lauer, W., and Placke, A. (1996). Global Distribution of Species Diversity in Vascular Plants: Towards a World Map of Phytodiversity (Globale Verteilung der

- 21 -

Artenvielfalt Höherer Pflanzen: Vorarbeiten zu einer Weltkarte der Phytodiversität). Erdkunde 50, 317–327. Becerra, J.X. (1997). Insects on Plants: Macroevolutionary Chemical Trends in Host Use. Science 276, 253–256. de Bello, F., Lavorel, S., Lavergne, S., Albert, C.H., Boulangeat, I., Mazel, F., and Thuiller, W. (2012). Hierarchical effects of environmental filters on the functional structure of plant communities: a case study in the French Alps. Ecography 36, 393–402. Belovsky, G.E., and Slade, J.B. (2000). Insect herbivory accelerates nutrient cycling and increases plant production. PNAS 97, 14412–14417. Berenbaum, M.R. (2001). Chemical Mediation of Coevolution: Phylogenetic Evidence for Apiaceae and Associates. Annals of the Missouri Botanical Garden 88, 45–59. Berg, M.P., Kiers, E.T., Driessen, G., Heijden, M.V.D., Kooi, B.W., Kuenen, F., Liefting, M., Verhoef, H.A., and Ellers, J. (2010). Adapt or disperse: understanding species persistence in a changing world. Global Change Biology 16, 587–598. Bidart‐Bouzat, M.G., and Imeh‐Nathaniel, A. (2008). Global Change Effects on Plant Chemical Defenses against Insect Herbivores. Journal of Integrative Plant Biology 50, 1339–1354. Bigger, D.S., and Marvier, M.A. (1999). How different would a world without herbivory be?: A search for generality in ecology. Integrative Biology: Issues, News, and Reviews 1, 60–67. Blumer, P., and Diemer, M. (1996). The Occurrence and Consequences of Grasshopper Herbivory in an Alpine Grassland, Swiss Central Alps. Arctic and Alpine Research 28, 435–440. Bonnier, G., and Flahault, C. (1878). Observations sur les modifications des végétaux. Ann. Sci. Nat. 7, 93–125. Borer, E.T., Seabloom, E.W., Gruner, D.S., Harpole, W.S., Hillebrand, H., Lind, E.M., Adler, P.B., Alberti, J., Anderson, T.M., Bakker, J.D., et al. (2014). Herbivores and nutrients control grassland plant diversity via light limitation. Nature 508, 517–520. Bouët, M. (1985). Climat et Météorologie de La Suisse Romande (Lausanne, Switzerland: Payot). Braschler, B., and Hill, J.K. (2007). Role of larval host plants in the climate-driven range expansion of the butterfly Polygonia c-album. Journal of Animal Ecology 76, 415–423. Brizuela, M.A., Detling, J.K., and Cid, M.S. (1986). Silicon Concentration of Grasses Growing in Sites With Different Grazing Histories. Ecology 67, 1098–1101. Brown, V.K. (1985). Insect Herbivores and Plant Succession. Oikos 44, 17–22. Brown, V.K., and Gange, A.C. (1992). Secondary plant succession: how is it modified by insect herbivory? Vegetatio 101, 3–13. Bruelheide, H., and Scheidel, U. (2001). Slug herbivory as a limiting factor for the geographical range of Arnica montana. Journal of Ecology 87, 839–848.

- 22 -

Bruelheide, H. (2003). Translocation of a montane meadow to simulate the potential impact of climate change. Applied Vegetation Science 6, 23–34. Buckley, L.B., Miller, E.F., and Kingsolver, J.G. (2013). Ectotherm Thermal Stress and Specialization Across Altitude and Latitude. Integr Comp Biol 53, 571–581. Carmona, D., Lajeunesse, M.J., and Johnson, M.T.J. (2010). Plant traits that predict resistance to herbivores. Functional Ecology 25, 358–367. Carson, W.P., and Root, R.B. (2000). Herbivory and Plant Species Coexistence: Community Regulation by an Outbreaking Phytophagous Insect. Ecological Monographs 70, 73–99. Chapin, S.F., Matson, P.A., and Mooney, H.A. (2002). Terrestrial Decomposition. In Principles of Terrestrial Ecosystem Ecology, (Springer New York.), pp. 151–175. Chen, I.-C., Hill, J.K., Ohlemüller, R., Roy, D.B., and Thomas, C.D. (2011). Rapid Range Shifts of Species Associated with High Levels of Climate Warming. Science 333, 1024–1026. Chong, J.-H., van lersel, M.W., and Oetting, R.D. (2004). Effects of Elevated Carbon Dioxide Levels and Temperature on the Life History of the Madeira Mealybug (Hemiptera: Pseudococcidae). Journal of Entomological Science 39, 387–397. Close, D.C., and McArthur, C. (2002). Rethinking the role of many plant phenolics – protection from photodamage not herbivores? Oikos 99, 166–172. Coley, P.D., Bryant, J.P., and Chapin, F.S. (1985). Resource Availability and Plant Antiherbivore Defense. Science 230, 895–899. Coley, P.D., Endara, M.-J., and Kursar, T.A. (2018). Consequences of interspecific variation in defenses and herbivore host choice for the ecology and evolution of Inga, a speciose rainforest tree. Oecologia 1–16. Coupe, M.D., and Cahill, J.F. (2003). Effects of insects on primary production in temperate herbaceous communities: a meta-analysis. Ecological Entomology 28, 511–521. Davis, A.J., Jenkinson, L.S., Lawton, J.H., Shorrocks, B., and Wood, S. (1998). Making mistakes when predicting shifts in species range in response to global warming. Nature 391, 783–786. DeAngelis, D.L. (2012). Dynamics of Nutrient Cycling and Food Webs (New York: Chapman and Hall). DeLucia, E.H., Nabity, P.D., Zavala, J.A., and Berenbaum, M.R. (2012). Climate Change: Resetting Plant-Insect Interactions. Plant Physiology 160, 1677–1685. Deraison, H., Badenhausser, I., Börger, L., and Gross, N. (2014). Herbivore effect traits and their impact on plant community biomass: an experimental test using grasshoppers. Functional Ecology 29, 650–661.

- 23 -

Descombes, P., Marchon, J., Pradervand, J.-N., Bilat, J., Guisan, A., Rasmann, S., and Pellissier, L. (2017). Community-level plant palatability increases with elevation as insect herbivore abundance declines. Journal of Ecology 105, 142–151. Devictor, V., Swaay, C. van, Brereton, T., Brotons, L., Chamberlain, D., Heliölä, J., Herrando, S., Julliard, R., Kuussaari, M., Lindström, Å., et al. (2012). Differences in the climatic debts of birds and butterflies at a continental scale. Nature Climate Change 2, 121–124. Diaz, S., and Cabido, M. (2009). Plant functional types and ecosystem function in relation to global change. Journal of Vegetation Science 8, 463–474. Dillon, M.E., Frazier, M.R., and Dudley, R. (2006). Into thin air: Physiology and evolution of alpine insects. Integr Comp Biol 46, 49–61. Dirnböck, T., Essl, F., and Rabitsch, W. (2011). Disproportional risk for habitat loss of high-altitude endemic species under climate change. Global Change Biology 17, 990–996. Duffy, J.E., Cardinale, B.J., France, K.E., McIntyre, P.B., Thébault, E., and Loreau, M. (2007). The functional role of biodiversity in ecosystems: incorporating trophic complexity. Ecology Letters 10, 522–538. Dury, S.J., Good, J.E.G., Perrins, C.M., Buse, A., and Kaye, T. (2002). The effects of increasing CO2 and temperature on oak leaf palatability and the implications for herbivorous insects. Global Change Biology 4, 55–61. Edger, P.P., Heidel-Fischer, H.M., Bekaert, M., Rota, J., Glöckner, G., Platts, A.E., Heckel, D.G., Der, J.P., Wafula, E.K., Tang, M., et al. (2015). The butterfly plant arms-race escalated by gene and genome duplications. PNAS 112, 8362–8366. Ehrlich, P.R., and Raven, P.H. (1964). Butterflies and Plants: A Study in Coevolution. Evolution 18, 586–608. Eskildsen, A., Carvalheiro, L.G., Kissling, W.D., Biesmeijer, J.C., Schweiger, O., and Høye, T.T. (2015). Ecological specialization matters: long-term trends in butterfly species richness and assemblage composition depend on multiple functional traits. Diversity and Distributions 21, 792–802. Evans, S.E., and Burke, I.C. (2013). Carbon and Nitrogen Decoupling Under an 11-Year Drought in the Shortgrass Steppe. Ecosystems 16, 20–33. Fauchald, P., Park, T., Tømmervik, H., Myneni, R., and Hausner, V.H. (2017). Arctic greening from warming promotes declines in caribou populations. Science Advances 3, e1601365. Fraser, L.H., and Madson, E.B. (2008). The interacting effects of herbivore exclosures and seed addition in a wet meadow. Oikos 117, 1057–1063. Galen, C. (1990). Limits to the Distributions of Alpine Tundra Plants: Herbivores and the Alpine Skypilot, Polemonium viscosum. Oikos 59, 355–358.

- 24 -

Galmán, A., Abdala‐Roberts, L., Zhang, S., Teran, J.C.B.-M. y, Rasmann, S., and Moreira, X. (2018). A global analysis of elevational gradients in leaf herbivory and its underlying drivers: Effects of plant growth form, leaf habit and climatic correlates. Journal of Ecology 106, 413– 421. Garibaldi, L.A., Kitzberger, T., and Chaneton, E.J. (2011). Environmental and genetic control of insect abundance and herbivory along a forest elevational gradient. Oecologia 167, 117–129. Gentry, A.H. (1988). Changes in Plant Community Diversity and Floristic Composition on Environmental and Geographical Gradients. Annals of the Missouri Botanical Garden 75, 1– 34. Gibson, C.W.D., Brown, V.K., and Jepsen, M. (1987). Relationships between the effects of insect herbivory and sheep grazing on seasonal changes in an early successional plant community. Oecologia 71, 245–253. Grime, J.P., MacPherson-Stewart, S.F., and Dearman, R.S. (1968). An Investigation of Leaf Palatability Using the Snail Cepaea Nemoralis L. Journal of Ecology 56, 405–420. de Groot, R.S., Wilson, M.A., and Boumans, R.M.J. (2002). A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecological Economics 41, 393–408. Guisan, A., and Thuiller, W. (2005). Predicting species distribution: offering more than simple habitat models. Ecology Letters 8, 993–1009. Gulmon, S.L., and Mooney, H.A. (1986). Costs of defense and their effects on plant productivity. In On the Economy of Plant Form and Function, (Cambridge, UK: Cambridge University Press), pp. 681–698. Gutbrodt, B., Mody, K., and Dorn, S. (2011). Drought changes plant chemistry and causes contrasting responses in lepidopteran herbivores. Oikos 120, 1732–1740. Hanley, M.E., Lamont, B.B., Fairbanks, M.M., and Rafferty, C.M. (2007). Plant structural traits and their role in anti-herbivore defence. Perspectives in Plant Ecology, Evolution and Systematics 8, 157–178. Hargreaves, A., Suarez, E., Mehltreter, K., Myers-Smith, I., Vanderplank, S.E., Slinn, H.L., Vargas- Rodriguez, Y.L., Haeussler, S., David, S., Munoz, J., et al. (2018). Seed predation increases from the Arctic to the Equator and from high to low elevations. BioRxiv 304634. Heil, M. (2008). Indirect defence via tritrophic interactions. New Phytologist 178, 41–61. Hochachka, P.W., and Somero, G.N. (2002). Biochemical adaptation: Mechanism and process in physiological evolution (Oxford: Oxford University Press).

- 25 -

Hódar, J.A., Castro, J., and Zamora, R. (2003). Pine processionary caterpillar Thaumetopoea pityocampa as a new threat for relict Mediterranean Scots pine forests under climatic warming. Biological Conservation 110, 123–129. Holeski, L.M., Chase‐Alone, R., and Kelly, J.K. (2010). The Genetics of Phenotypic Plasticity in Plant Defense: Trichome Production in Mimulus guttatus. The American Naturalist 175, 391– 400. Hudson, J.M.G., Henry, G.H.R., and Cornwell, W.K. (2011). Taller and larger: shifts in Arctic tundra leaf traits after 16 years of experimental warming. Global Change Biology 17, 1013– 1021. Hülber, K., Haider, J.A., Hager, T.E., Dullinger, S., and Fiedler, K. (2015). Insect herbivory in alpine grasslands is constrained by community and host traits. Journal of Vegetation Science 26, 663–673. Hulme, P.E. (1996a). Herbivores and the Performance of Grassland Plants: A Comparison of Arthropod, Mollusc and Rodent Herbivory. Journal of Ecology 84, 43–51. Hulme, P.E. (1996b). Herbivory, Plant Regeneration, and Species Coexistence. Journal of Ecology 84, 609–615. von Humboldt, A. (2009). Ideen zu einer Geographie der Pflanzen (BoD – Books on Demand). Hutchinson, G.E. (1957). Cold spring harbor symposium on quantitative biology. Concluding Remarks 22, 415–427. Ibanez, S., Gallet, C., and Després, L. (2012). Plant Insecticidal Toxins in Ecological Networks. Toxins 4, 228–243. Ibanez, S., Lavorel, S., Puijalon, S., and Moretti, M. (2013). Herbivory mediated by coupling between biomechanical traits of plants and grasshoppers. Functional Ecology 27, 479–489. Kaarlejärvi, E., Eskelinen, A., and Olofsson, J. (2017). Herbivores rescue diversity in warming tundra by modulating trait-dependent species losses and gains. Nature Communications 8, 419. Karban, R. (2011). The ecology and evolution of induced resistance against herbivores. Functional Ecology 25, 339–347. Karban, R., and Strauss, S.Y. (1993). Effects of Herbivores on Growth and Reproduction of their Perennial Host, Erigeron Glaucus. Ecology 74, 39–46. Karban, R., Agrawal, A.A., Thaler, J.S., and Adler, L.S. (1999). Induced plant responses and information content about risk of herbivory. Trends in Ecology & Evolution 14, 443–447. Kergunteuil, A., Campos-Herrera, R., Sánchez-Moreno, S., Vittoz, P., and Rasmann, S. (2016). The Abundance, Diversity, and Metabolic Footprint of Soil Nematodes Is Highest in High Elevation Alpine Grasslands. Front. Ecol. Evol. 4, 84.

- 26 -

Kergunteuil, A., Descombes, P., Glauser, G., Pellissier, L., and Rasmann, S. (2018). Plant physical and chemical defence variation along elevation gradients: a functional trait-based approach. Oecologia 1–11. Kessler, A., and Heil, M. (2011). The multiple faces of indirect defences and their agents of natural selection. Functional Ecology 25, 348–357. Kim, T.N., Underwood, N., and Inouye, B.D. (2013). Insect herbivores change the outcome of plant competition through both inter- and intraspecific processes. Ecology 94, 1753–1763. Körner, C. (2003). Alpine Plant Life: Functional Plant Ecology of High Mountain Ecosystems; with 47 tables (Springer Science & Business Media). Körner, C. (2007). The use of ‘altitude’ in ecological research. Trends in Ecology & Evolution 22, 569–574. Lemoine, N.P., and Burkepile, D.E. (2012). Temperature-induced mismatches between consumption and metabolism reduce consumer fitness. Ecology 93, 2483–2489. Lemoine, N.P., Drews, W.A., Burkepile, D.E., and Parker, J.D. (2013). Increased temperature alters feeding behavior of a generalist herbivore. Oikos 122, 1669–1678. Lemoine, N.P., Burkepile, D.E., and Parker, J.D. (2014). Variable effects of temperature on insect herbivory. PeerJ 2, e376. Linhart, Y.B., and Grant, and M.C. (1996). Evolutionary Significance of Local Genetic Differentiation in Plants. Annual Review of Ecology and Systematics 27, 237–277. Long, Z.T., Mohler, C.L., and Carson, W.P. (2003). Extending the Resource Concentration Hypothesis to Plant Communities: Effects of Litter and Herbivores. Ecology 84, 652–665. Loranger, J., Meyer, S.T., Shipley, B., Kattge, J., Loranger, H., Roscher, C., and Weisser, W.W. (2012). Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology 93, 2674–2682. Massey, F.P., and Hartley, S.E. (2009). Physical defences wear you down: progressive and irreversible impacts of silica on insect herbivores. Journal of Animal Ecology 78, 281–291. Massey, F.P., Ennos, A.R., and Hartley, S.E. (2006). Silica in grasses as a defence against insect herbivores: contrasting effects on folivores and a phloem feeder. Journal of Animal Ecology 75, 595–603. Maze, D.M. (2009). Effect of Terrestrial Mollusc Herbivory on Holocarpha macradenia (Asteraceae) Seedlings in California Coastal Prairie under Different Clipping Regimes. Madroño 56, 1–7. Melillo, J.M., Steudler, P.A., Aber, J.D., Newkirk, K., Lux, H., Bowles, F.P., Catricala, C., Magill, A., Ahrens, T., and Morrisseau, S. (2002). Soil Warming and Carbon-Cycle Feedbacks to the Climate System. Science 298, 2173–2176.

- 27 -

Merrill, R.M., Gutiérrez, D., Lewis, O.T., Gutiérrez, J., Díez, S.B., and Wilson, R.J. (2008). Combined effects of climate and biotic interactions on the elevational range of a phytophagous insect. Journal of Animal Ecology 77, 145–155. Metcalfe, D.B., Asner, G.P., Martin, R.E., Espejo, J.E.S., Huasco, W.H., Amézquita, F.F.F., Carranza‐Jimenez, L., Cabrera, D.F.G., Baca, L.D., Sinca, F., et al. (2014). Herbivory makes major contributions to ecosystem carbon and nutrient cycling in tropical forests. Ecology Letters 17, 324–332. Miao, H.-T., Liu, Y., Shan, L.-Y., and Wu, G.-L. (2018). Linkages of plant-soil interface habitat and grasshopper occurrence of typical grassland ecosystem. Ecological Indicators 90, 324– 333. Mithöfer, A., and Boland, W. (2012). Plant defense against herbivores: chemical aspects. Annual Review of Plant Biology 63, 431–450. Moles, A.T., and Ollerton, J. (2016). Is the notion that species interactions are stronger and more specialized in the tropics a zombie idea? Biotropica 48, 141–145. Moreira, X., Castagneyrol, B., Abdala‐Roberts, L., Teran, J.C.B.-M. y, Timmermans, B.G.H., Bruun, H.H., Covelo, F., Glauser, G., Rasmann, S., and Tack, A.J.M. (2017). Latitudinal variation in plant chemical defences drives latitudinal patterns of leaf herbivory. Ecography 40, 1–10. Moreira, X., Petry, W.K., Mooney, K.A., Rasmann, S., and Abdala-Roberts, L. (2018). Elevational gradients in plant defences and insect herbivory: recent advances in the field and prospects for future research. Ecography 00, 1–12. Mortensen, B., Danielson, B., Harpole, W.S., Alberti, J., Arnillas, C.A., Biederman, L., Borer, E.T., Cadotte, M.W., Dwyer, J.M., Hagenah, N., et al. (2018). Herbivores safeguard plant diversity by reducing variability in dominance. Journal of Ecology 106, 101–112. Myers, N., Mittermeier, R.A., Mittermeier, C.G., Fonseca, G.A.B. da, and Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature 403, 853–858. Nitschke, N., Wiesner, K., Hilke, I., Eisenhauer, N., Oelmann, Y., and Weisser, W.W. (2014). Increase of fast nutrient cycling in grassland microcosms through insect herbivory depends on plant functional composition and species diversity. Oikos 124, 161–173. Núñez-Farfán, J., Fornoni, J., and Valverde, P.L. (2007). The Evolution of Resistance and Tolerance to Herbivores. Annual Review of Ecology, Evolution, and Systematics 38, 541– 566. O’Connor, M.I. (2009). Warming strengthens an herbivore–plant interaction. Ecology 90, 388–398. Olff, H., and Ritchie, M.E. (1998). Effects of herbivores on grassland plant diversity. Trends in Ecology & Evolution 13, 261–265.

- 28 -

Onsager, J.A. (1991). Toward an Optimum Strategy for Estimating Density of Grasshoppers (Orthoptera: Acrididae) from a Poisson Distribution. Environ Entomol 20, 939–944. Paré, P.W., and Tumlinson, J.H. (1999). Plant Volatiles as a Defense against Insect Herbivores. Plant Physiology 121, 325–332. Parmesan, C. (2006). Ecological and Evolutionary Responses to Recent Climate Change. Annual Review of Ecology, Evolution, and Systematics 37, 637–669. Parmesan, C., and Yohe, G. (2003). A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, 37–42. Pateman, R.M., Hill, J.K., Roy, D.B., Fox, R., and Thomas, C.D. (2012). Temperature-Dependent Alterations in Host Use Drive Rapid Range Expansion in a Butterfly. Science 336, 1028– 1030. Pauli, H., Gottfried, M., Dullinger, S., Abdaladze, O., Akhalkatsi, M., Alonso, J.L.B., Coldea, G., Dick, J., Erschbamer, B., Calzado, R.F., et al. (2012). Recent Plant Diversity Changes on Europe’s Mountain Summits. Science 336, 353–355. Pearse, I.S., and Altermatt, F. (2013). Predicting novel trophic interactions in a non-native world. Ecology Letters 16, 1088–1094. Pearse, I.S., Harris, D.J., Karban, R., and Sih, A. (2013). Predicting novel herbivore–plant interactions. Oikos 122, 1554–1564. Peeters, P.J., Sanson, G., and Read, J. (2007). Leaf biomechanical properties and the densities of herbivorous insect guilds. Functional Ecology 21, 246–255. Pellissier, L., Fournier, B., Guisan, A., and Vittoz, P. (2010). Plant traits co-vary with altitude in grasslands and forests in the European Alps. Plant Ecol 211, 351–365. Pellissier, L., Pradervand, J.-N., Pottier, J., Dubuis, A., Maiorano, L., and Guisan, A. (2012a). Climate-based empirical models show biased predictions of butterfly communities along environmental gradients. Ecography 35, 684–692. Pellissier, L., Fiedler, K., Ndribe, C., Dubuis, A., Pradervand, J.-N., Guisan, A., and Rasmann, S. (2012b). Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecology and Evolution 2, 1818–1825. Pellissier, L., Ndiribe, C., Dubuis, A., Pradervand, J.-N., Salamin, N., Guisan, A., and Rasmann, S. (2013a). Turnover of plant lineages shapes herbivore phylogenetic beta diversity along ecological gradients. Ecology Letters 16, 600–608. Pellissier, L., Rohr, R.P., Ndiribe, C., Pradervand, J.-N., Salamin, N., Guisan, A., and Wisz, M. (2013b). Combining food web and species distribution models for improved community projections. Ecology and Evolution 3, 4572–4583.

- 29 -

Pellissier, L., Niculita‐Hirzel, H., Dubuis, A., Pagni, M., Guex, N., Ndiribe, C., Salamin, N., Xenarios, I., Goudet, J., Sanders, I.R., et al. (2014a). Soil fungal communities of grasslands are environmentally structured at a regional scale in the Alps. Molecular Ecology 23, 4274– 4290. Pellissier, L., Roger, A., Bilat, J., and Rasmann, S. (2014b). High elevation Plantago lanceolata plants are less resistant to herbivory than their low elevation conspecifics: is it just temperature? Ecography 37, 950–959. Pennings, S.C., Ho, C.-K., Salgado, C.S., Więski, K., Davé, N., Kunza, A.E., and Wason, E.L. (2009). Latitudinal variation in herbivore pressure in Atlantic Coast salt marshes. Ecology 90, 183–195. Pérez‐Harguindeguy, N., Díaz, S., Vendramini, F., Cornelissen, J.H.C., Gurvich, D.E., and Cabido, M. (2003). Leaf traits and herbivore selection in the field and in cafeteria experiments. Austral Ecology 28, 642–650. Poore, A.G.B., Campbell, A.H., Coleman, R.A., Edgar, G.J., Jormalainen, V., Reynolds, P.L., Sotka, E.E., Stachowicz, J.J., Taylor, R.B., Vanderklift, M.A., et al. (2012). Global patterns in the impact of marine herbivores on benthic primary producers. Ecology Letters 15, 912–922. Pratt, J.D., and Mooney, K.A. (2013). Clinal adaptation and adaptive plasticity in Artemisia californica: implications for the response of a foundation species to predicted climate change. Global Change Biology 19, 2454–2466. Rasmann, S., and Agrawal, A.A. (2011). Evolution of Specialization: A Phylogenetic Study of Host Range in the Red Milkweed Beetle (Tetraopes tetraophthalmus). The American Naturalist 177, 728–737. Rasmann, S., and Turlings, T.C.J. (2007). Simultaneous feeding by aboveground and belowground herbivores attenuates plant-mediated attraction of their respective natural enemies. Ecology Letters 10, 926–936. Rasmann, S., Pellissier, L., Defossez, E., Jactel, H., and Kunstler, G. (2014). Climate-driven change in plant–insect interactions along elevation gradients. Functional Ecology 28, 46–54. Reynolds, B.C., and Crossley, D.A. (1997). Spatial Variation in Herbivory by Forest Canopy Arthropods Along an Elevation Gradient. Environ Entomol 26, 1232–1239. Rhoades, D.F. (1979). Evolution of plant chemical defence against herbivores. In Herbivores: Their Interaction with Secondary Plant Metabolites, (New York, NY, USA: Academic Press: London), pp. 3–54. Richards, L.A., Dyer, L.A., Forister, M.L., Smilanich, A.M., Dodson, C.D., Leonard, M.D., and Jeffrey, C.S. (2015). Phytochemical diversity drives plant–insect community diversity. PNAS 112, 10973–10978.

- 30 -

Rogora, M., Frate, L., Carranza, M.L., Freppaz, M., Stanisci, A., Bertani, I., Bottarin, R., Brambilla, A., Canullo, R., Carbognani, M., et al. (2018). Assessment of climate change effects on mountain ecosystems through a cross-site analysis in the Alps and Apennines. Science of The Total Environment 624, 1429–1442. Salazar, D., Lokvam, J., Mesones, I., Vásquez, M.P., Ayarza, J.Z., De, P.V., and Fine, P.V.A. (2018). Origin and maintenance of chemical diversity in a species-rich tropical tree lineage. Nat Ecol Evol 2, 983–990. Sanders, N.J. (2002). Elevational gradients in ant species richness: area, geometry, and Rapoport’s rule. Ecography 25, 25–32. de Sassi, C., and Tylianakis, J.M. (2012). Climate Change Disproportionately Increases Herbivore over Plant or Parasitoid Biomass. PLOS ONE 7, e40557. Scheidel, U., and Bruelheide, H. (2001). Selective slug grazing on montane meadow plants. Journal of Ecology 87, 828–838. Scheidel, U., Röhl, S., and Bruelheide, H. (2003). Altitudinal gradients of generalist and specialist herbivory on three montane Asteraceae. Acta Oecologica 24, 275–283. Schweiger, O., Heikkinen, R.K., Harpke, A., Hickler, T., Klotz, S., Kudrna, O., Kühn, I., Pöyry, J., and Settele, J. (2012). Increasing range mismatching of interacting species under global change is related to their ecological characteristics. Global Ecology and Biogeography 21, 88–99. Seastedt, T.R., and Crossley, D.A. (1984). The Influence of Arthropods on Ecosystems. BioScience 34, 157–161. Soberón, J. (2007). Grinnellian and Eltonian niches and geographic distributions of species. Ecology Letters 10, 1115–1123. Speed, M.P., Fenton, A., Jones, M.G., Ruxton, G.D., and Brockhurst, M.A. (2015). Coevolution can explain defensive secondary metabolite diversity in plants. New Phytologist 208, 1251–1263. Strauss, S.Y., and Agrawal, A.A. (1999). The ecology and evolution of plant tolerance to herbivory. Trends in Ecology & Evolution 14, 179–185. Tanentzap, A.J., Lee, W.G., Dugdale, J.S., Patrick, B.P., Fenner, M., Walker, S., and Coomes, D.A. (2011). Differential responses of vertebrate and invertebrate herbivores to traits of New Zealand subalpine shrubs. Ecology 92, 994–999. Tansley, A.G. (1935). The Use and Abuse of Vegetational Concepts and Terms. Ecology 16, 284– 307. Thuiller, W., Lavorel, S., Araújo, M.B., Sykes, M.T., and Prentice, I.C. (2005). Climate change threats to plant diversity in Europe. PNAS 102, 8245–8250.

- 31 -

Totland, Ø. (1999). Effects of temperature on performance and phenotypic selection on plant traits in alpine Ranunculus acris. Oecologia 120, 242–251. Turlings, T.C.J., Tumlinson, J.H., and Lewis, W.J. (1990). Exploitation of Herbivore-Induced Plant Odors by Host-Seeking Parasitic Wasps. Science 250, 1251–1253. del-Val, E., Crawley, M.J., and Bruelheide, H. (2004). Importance of tolerance to herbivory for plant survival in a British grassland. Journal of Vegetation Science 15, 357–364. Van der Putten, W.H., Macel, M., and Visser, M.E. (2010). Predicting species distribution and abundance responses to climate change: why it is essential to include biotic interactions across trophic levels. Philosophical Transactions of the Royal Society B: Biological Sciences 365, 2025–2034. Veteli, T.O., Kuokkanen, K., Julkunen‐Tiitto, R., Roininen, H., and Tahvanainen, J. (2002). Effects of elevated CO2 and temperature on plant growth and herbivore defensive chemistry. Global Change Biology 8, 1240–1252. Violle, C., Navas, M.-L., Vile, D., Kazakou, E., Fortunel, C., Hummel, I., and Garnier, E. (2007). Let the concept of trait be functional! Oikos 116, 882–892. Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T.J.C., Fromentin, J.-M., Hoegh-Guldberg, O., and Bairlein, F. (2002). Ecological responses to recent climate change. Nature 416, 389–395. Wardle, D.A., Bardgett, R.D., Klironomos, J.N., Setälä, H., Putten, W.H. van der, and Wall, D.H. (2004). Ecological Linkages Between Aboveground and Belowground Biota. Science 304, 1629–1633. Williams, J.W., and Jackson, S.T. (2007). Novel climates, no‐analog communities, and ecological surprises. Frontiers in Ecology and the Environment 5, 475–482. Wilson, R.J., Gutiérrez, D., Gutiérrez, J., and Monserrat, V.J. (2007). An elevational shift in butterfly species richness and composition accompanying recent climate change. Global Change Biology 13, 1873–1887. Wing, S.L., Harrington, G.J., Smith, F.A., Bloch, J.I., Boyer, D.M., and Freeman, K.H. (2005). Transient Floral Change and Rapid Global Warming at the Paleocene-Eocene Boundary. Science 310, 993–996. Wisz, M.S., Pottier, J., Kissling, W.D., Pellissier, L., Lenoir, J., Damgaard, C.F., Dormann, C.F., Forchhammer, M.C., Grytnes, J.-A., Guisan, A., et al. (2013). The role of biotic interactions in shaping distributions and realised assemblages of species: implications for species distribution modelling. Biological Reviews 88, 15–30. Zhang, S., Zhang, Y., and Ma, K. (2016). Latitudinal variation in herbivory: hemispheric asymmetries and the role of climatic drivers. Journal of Ecology 104, 1089–1095.

- 32 -

Züst, T., Heichinger, C., Grossniklaus, U., Harrington, R., Kliebenstein, D.J., and Turnbull, L.A. (2012). Natural Enemies Drive Geographic Variation in Plant Defenses. Science 338, 116– 119. Zvereva, E.L., and Kozlov, M.V. (2006). Consequences of simultaneous elevation of carbon dioxide and temperature for plant–herbivore interactions: a metaanalysis. Global Change Biology 12, 27–41.

- 33 -

- 34 -

CHAPTER 1

Uneven rate of plant turnover along elevation in grasslands

Patrice Descombes1,2, Pascal Vittoz3, Antoine Guisan3,4, Loïc Pellissier1,2

1Landscape Ecology Institute of Terrestrial Ecosystems, ETH Zürich, Zürich, Switzerland 2Swiss Federal Research Institute WSL Birmensdorf, Switzerland 3Institute of Earth Surface Dynamics, University of Lausanne, Lausanne, Switzerland 4Department of Ecology and Evolution University of Lausanne, Lausanne, Switzerland

Published in Alpine Botany (2017), 127, 53-63 doi: https://doi.org/10.1007/s00035-016-0173-7 Post-print version

- 35 -

Abstract

Plant taxonomic and phylogenetic composition of assemblages are known to shift along environmental gradients, but whether the rate of species turnover is regular or not (e.g., accelerations in particular sections of the gradient) remains poorly documented. Understanding how rates of assemblage turnover vary along gradients is crucial to forecast where climate change could promote the fastest changes within extant communities. Here we analysed turnover rates of plant assemblages along a 2500 m elevation gradient in the Swiss Western Alps. We found a peak of turnover rate between 1800 and 2200 m indicating an acceleration of grassland compositional changes at the transition between subalpine and alpine belts. In parallel, we found a peak in phylogenetic turnover rate in Poales between 1700 m and 1900 and Super-Rosids between 1900 and 2300 m. Our results suggest that changes in abiotic or biotic conditions near the human-modified treeline constitute a strong barrier for many grassland plant species, which share analogous elevation range limits. We propose that this vegetation zone of high ecological transitions over short geographical distances should show the fastest community responses to climate change from the breakdown of barrier across ecotones.

Introduction

Since species survive and reproduce within bounded abiotic conditions (Hutchinson 1957; Soberón 2007), the composition of species assemblages changes along environmental gradients. Clines in temperature or moisture are generally associated with strong species compositional changes, such as in plants (Gentry 1988; Pellissier et al. 2010; de Bello et al. 2013), animals (Sanders 2002; Graham et al. 2009; Longino and Colwell 2011; Pellissier et al. 2012) and even microorganisms (Pellissier et al. 2014a). However, few studies so far investigated the rate of assemblage turnover along environmental gradients (Mena and Vázquez-Domínguez 2005; Bach et al. 2007; Jankowski et al. 2009, 2013). If all species show idiosyncratic response to abiotic conditions, a constant rate of compositional turnover is expected along the entire environmental gradient (Gleason 1926; Bach et al. 2007). In contrast, if a large proportion of species shares similar environmental limits, referred as “range boundary clumping” (Clements 1916; Leibold and Mikkelson 2002), a peak in the turnover component of beta-diversity would be expected in this section of the gradient (Mena and Vázquez-Domínguez 2005).

Elevation gradients are among the most studied environmental clines in ecology since they provide large variations in abiotic conditions over very short distances (Körner 2007). In the Alps and other mountain ranges in temperate climate, the climate shifts toward more stressful conditions

- 36 - for plant growth with increasing elevation (Körner 2003) and clines in species alpha- and beta- diversity can be observed (Körner 2000; Dubuis et al. 2011; Pellissier et al. 2013a). Species richness decreases (Körner 2000; Vittoz et al. 2010) and lowland species are replaced by high elevation specialists (Körner 2000; Theurillat et al. 2003). Gradients in environmental conditions along elevation can also be confounded with change in human land uses and disturbances, and in the intensity of biotic processes, especially around the treeline (Pottier et al. 2013). Yet, the rate of compositional changes along elevation remains poorly studied (Odland and Birks 1999; Jankowski et al. 2013). Jankowski et al. (2013) investigated compositional changes of trees and birds in tropical mountains of the Peruvian Andes and showed distinct peaks in plant and bird turnover rates along elevation. At a functional level, Ndiribe et al. (2013b) demonstrated the importance of climate and land use factors in shaping patterns of functional and phylogenetic beta-diversity, and Pellissier et al. (2010) highlighted that dominant functional traits expressed in communities in the Swiss Alps change more rapidly around the treeline. Repeated beta-diversity comparisons between pairs of plots experiencing a small difference in elevation allow us to evaluate the evenness of the nestedness and turnover components of beta-diversity along elevation.

Complementing measures of taxonomic beta-diversity, phylogenetic beta-diversity provides additional insights into the mechanisms underlying diversity patterns along environmental gradients by considering phylogenetic relatedness among species (Graham and Fine 2008; Pellissier et al. 2013b). Phylogenetic community ecology is tightly linked to the concept of niche conservatism, the tendency of closely related species to retain the same environmental niche (Wiens et al. 2010). Phylogenetic turnover of assemblages along environmental gradients is expected to reflect niche- related processes, especially environmental filtering of lineages (Graham and Fine 2008). Phylogenetic diversity patterns have been observed to change along elevation (Culmsee and Leuschner 2013; Ndiribe et al. 2013a; Pellissier et al. 2013b), but the absence of relationships were also documented in other studies (Bryant et al. 2008; Chalmandrier et al. 2015). Large phylogenetic distances associated to low trait conservatism between species in assemblages can blur ecological signals (Kembel and Hubbell 2006; Godoy et al. 2014). Thus, focusing on patterns within specific clades might provide more detailed information. If niche boundaries are phylogenetically conserved, or clumped in particular clades, the rate of phylogenetic turnover along elevation should not be constant but present irregularities in sections of the elevation gradient (Ndiribe et al. 2013a).

Taxonomic beta-diversity variation along environmental gradients can be further decomposed into a turnover and a nestedness component (Baselga 2010). Nestedness of species assemblages occurs when the species composition of sites with smaller numbers of species are subsets of the composition at richer sites (Wright and Reeves 1992; Ulrich and Gotelli 2007), and reflects species

- 37 -

loss as a consequence of processes promoting the disaggregation of assemblages into its subsets (Gaston et al. 2000). In contrast, assemblage turnover implies the replacement of some species by others across community pairs, and results predominantly from a shift in environmental conditions (Qian et al. 2005). Decomposing beta-diversity into turnover and nestedness components provides complementary information on the processes shaping the assembly of communities (Baselga and Leprieur 2015).

Here, we investigate the gamma- and alpha-diversity, and the turnover and nestedness components of beta-diversity of plant communities in grasslands along a 2500 m elevation gradient in the Western Swiss Alps with a special emphasis on the evenness of the turnover rate. We computed beta-diversity of pairs of plots separated by an elevation smaller than 20 m and decomposed it into turnover and nestedness components (Baselga 2010). Using a species-level phylogeny of the regional flora, we further investigated phylogenetic turnover of Poales, Super- Asterids and Super-Rosids clades along elevation. In the case of idiosyncratic response of species to shifting environmental conditions along the elevation gradient, we should observe a flat relationship between turnover rate and elevation. In contrast, if species share similar range limits, a higher turnover rate should be observed in portions of the gradient. Documenting turnover rates along elevation gradients is particularly relevant in the context of climate change. Sections of the gradient with higher turnover rate indicate the presence of a strong barrier across different ecosystems, which might be lifted by climate change.

Methods

Study area and data collection

The study area, covering approximately 700 km2, is located in the Western Alps of Switzerland (canton de Vaud) and exhibits an elevational gradient ranging from 375 to 3200 m with a soil parent material that is mainly calcareous (46°10′–46°30′N; 6°50′–7°10′E; Fig. 1). The region has a temperate climate with mean annual temperature between 8 °C at 375 m and −5 °C at 3200 m and with annual sum of precipitation between 1200 and 2600 mm (Bouët 1985). Below the treeline (i.e. 1900 m; lowered by a few 100 m through centuries of human activities; Gehrig-Fasel et al. 2007), most of the open vegetation areas are used for grazing and/or mowing, often with regular fertilisation, whereas the areas in alpine belt are occupied by alpine grasslands and glaciers with much lower levels of human disturbance, except cattle grazing in summer on the most accessible areas. The species data have been collected in open and non-woody vegetation areas only (i.e. grasslands, meadows, rocks and screes; see Fig. S1 to see how open areas and selected plots are

- 38 - distributed along the elevation gradient) using a balanced random stratified sampling design (Hirzel and Guisan 2002) relying on slope, elevation and aspect (see Fig. S2 to see how the selected plots are distributed in the ecological space). Since slope, elevation and aspect are proxies for contrasted ecological conditions, this design allows us to collect data from the full range of vegetation types present along the elevation gradient. The vegetation sampling includes 912 sites surveyed exhaustively on 2 × 2 m squares across the whole gradient between 2002 and 2009 (for more details see Dubuis et al. 2011) and which had similar topography. Species cover was visually estimated according to a 7-level scale.

Figure 1. Location of the study area in the western Alps of Switzerland. The dots represent vegetation sampling sites and green areas represent forests ecosystems. The light grey line shows the limits of study area. The dark grey line shows the 800 m isoline.

- 39 -

Alpha- and gamma-diversity

To investigate how the local species pools vary along elevation gradient, we computed the total number of species encountered in sampled plant communities within 20 m elevation bands centred on each 10 m elevation steps. This represents the local gamma-diversity defined for each elevation section. In this study, we only used elevation as ecological gradient (see Körner 2007) since this gradient is highly correlated with abiotic factors such as degree-days (Spearman’s correlation: r = −0.997; Fig. S2) or precipitation (Spearman’s correlation: r = 0.959; Fig. S2), and also associated with shifts in biotic conditions such as reduced competition at higher elevation (e.g., Michalet et al. 2006, 2015), a decrease in herbivore pressure (e.g., Reynolds and Crossley 1997; Garibaldi et al. 2011; Pellissier et al. 2014b) or a gradient of land use by humans (see above).

In addition, within each of the 20 m elevation sections, we computed the mean number of species found in communities as a measure of local average alpha-diversity in each elevation band. Comparing alpha- and gamma-diversity and their deviation along elevation, a measure of beta- diversity (Tuomisto 2010), provide information on the intensity of environmental and human mediated filtering processes within each elevation band. The range of 20 m was selected because it constitutes a good compromise between resolution of the elevation bands and number of possible comparisons between pairs of vegetation plots. However, to ensure that our conclusions were non sensitive to the choice of threshold, we also ran the analyses with a range of 10 m and a range of 50 m. Because land use might influence plant alpha- and gamma-diversity patterns (e.g. Fischer et al. 2008; Niedrist et al. 2009), we also related number of open areas (see Fig. S1) and diversity of vegetation types (see Fig. S3) to the alpha- and gamma-diversity within each of the 20 m elevation bands. Plots were grouped with a hierarchical clustering and the groups were attributed to a vegetation type, according to the classification of Delarze and Gonseth (2008), on the basis of their respective differential species.

Community taxonomic turnover and nestedness of beta-diversity

We computed beta-diversity of all plant species, Poales (i.e. Cyperaceae, Juncaceae and Poaceae), Super-Asterids (i.e. Apiales, Asterales, Caryophyllales, Dipsacales, Ericales, Gentianales, Lamiales and Santalales) and Super-Rosids (i.e. Brassicales, Celastrales, Fabales, Geraniales, Malpighiales, Malvales, Rosales and Saxifragales) between all pairs of communities with an elevation difference lower than 20 m and partitioned the total beta-diversity (Jaccard dissimilarity index, βjac) into turnover (Turnover component of Jaccard dissimilarity, βjtu) and nestedness- resultant dissimilarity (Nestedness-resultant component of Jaccard dissimilarity, βjne) by using the

- 40 - package “betapart” (Baselga 2012; Baselga and Orme 2012) in R (R Development Core Team, www.R-project.org). βjtu and βjne vary between 0 and 1, where high values indicate greater dissimilarity in species composition and low values indicate greater proportion of shared species. In the absence of nestedness (i.e. species between pairs of communities are completely different), βjtu is equal to βjac and equal to 1. The difference between βjtu and βjac is a measure of the nestedness component of beta-diversity. In the absence of turnover (i.e. species of a community are a subset of a richer community), βjne is equal to βjac and is influenced by differences in species richness. We related the βjtu and βjne values to the mean elevation of each pair of plots using a linear model including both a linear and a quadratic term. Any deviation from an intercept-only model, either with a linear or non-linear slope, would indicate a non-constant turnover and nestedness rate along elevation. We also investigated how the spatial distance between pairs of plots varies between elevation bands along the elevation gradient by relating the horizontal distance separating each pair of plots with their mean elevation and their taxonomic turnover (βjtu). Because land use might influence plant beta-diversity patterns (Ndiribe et al. 2013b), we also related elevation and habitat variables (i.e. number of open areas, diversity of vegetation types) to the mean taxonomic turnover

(βjtu) within 20 m elevation bands by using an ordinary least squares regression (OLS) model and quantified the relative importance of elevation vs. habitat variables for explaining beta-diversity variation with a variance partitioning analyses (see Appendix S1 for methodological details on the OLS models and variable partition analyses). Finally, we extracted the elevation minima and maxima of each plant species from the 912 plots. We related these range limits to the elevation gradient to explain how range boundaries could influence dissimilarities in species composition.

Community phylogenetic turnover of beta-diversity

We tested the phylogenetic signal in species distribution along the elevation gradient (the median elevation at which the species occurred), by pruning from a published phylogeny of the 231 most frequent and abundant plant species in our study area (Ndiribe et al. 2013a). We calculated Blomberg’s K statistic with the “phylosignal” function as implemented in the “picante” R package (Blomberg et al. 2003; Kembel et al. 2010), as our measure of phylogenetic signal. We calculated Blomberg’s K across all species and in three angiosperm clades: Poales, Super-Asterids and Super- Rosids. Blomberg’s K statistic compares the observed distribution of the trait values to expectations under a Brownian motion model of trait evolution. K values close to 1 indicate trait evolution consistent with a Brownian motion model of evolution, while K values close to 0 indicate a random distribution of trait values with respect to the phylogeny (Blomberg et al. 2003). We tested the significance of this test by comparing the observed K value to a null distribution generated by

- 41 -

comparing 999 randomizations of trait values across the tips of the phylogenetic tree (Kembel et al. 2010).

We computed phylogenetic turnover of beta-diversity of all plant species, Poales, Super- Asterids and Super-Rosids between pairs of plots with an elevation difference lower than 20 m, using the mean pairwise distance (MPD) implemented in the “comdist” function in the “picante” R package (Kembel et al. 2010). We related the mean elevation of each pair of plots with their phylogenetic turnover value and tested for the existence of shifts in rates of phylogenetic turnover using a linear model including quadratic terms. To visualise the contribution of families to communities along elevation, we calculated the proportion of species occurrences and the proportion of species cover of the dominant plant clades of Poales, Super-Asterids and Super- Rosids (i.e. Poaceae, Cyperaceae, Asteraceae, Fabaceae, Apiaceae, Saxifragaceae) for 200 m elevation bands.

Results

Alpha- and gamma-diversity

We found that the gamma-diversity within the 20-m elevation bands showed a hump-shaped curve, with a peak between 1500 and 1900 m (Fig. 2a). Similarly, the mean alpha-diversity of communities within each elevation band showed a hump-shaped curve, but with a peak between 1100 and 1500 m (Fig. 2a). We observed a positive relationship between the gamma- and the mean alpha-diversity (Spearman’s correlation: r = 0.632), indicating that a larger gamma-diversity is associated with a higher mean alpha-diversity of communities, but the relationship showed higher discrepancies in some section of the gradient (Fig. 2a). Differences between gamma- and mean alpha-diversities are overall larger at lower elevation, but with a peak around 1400–2100 m (maximum difference 222 at 1820 m). Note that the difference between gamma and alpha-diversity in elevation bands of 20 m is another measure of beta-diversity (Tuomisto 2010), and showed only weak correlation with the turnover component of beta-diversity calculated between pairs of plots with less than 20 m of difference in elevation (Spearman’s correlation r = 0.212). Finally, we also observed a strong relationship between the gamma-diversity and the number of open areas (Spearman’s correlation r = 0.750) and the diversity of vegetation types (r = 0.786), indicating that larger open areas and areas with diverse vegetation types sustain higher gamma-diversity. The relationships with alpha-diversity were weaker for number of open areas (r = 0.585) and the diversity of vegetation types (r = 0.285).

- 42 -

Community taxonomic turnover and nestedness of beta-diversity

Overall, beta-diversity between pairs of plots along elevation is mainly generated by species turnover (βjtu; mean 0.80, sd 0.17; Fig. 2b), while the contribution of nestedness to overall dissimilarity was much lower (βjne; mean 0.06, sd 0.10; Fig. 2b). We found that the turnover (βjtu) and nestedness (βjne) components of beta-diversity were uneven along elevation (βjtu: linear s = 4.8 −4 −7 −4 −8 × 10 , quadratic s = −1.2 × 10 , βjne: linear s = −2.6 × 10 , quadratic s = 7.9 × 10 ; Fig. 2b). Including quadratic terms improved the models for turnover (AIC difference −229.8) and nestedness (AIC difference −260.7) supporting the existence of a hump-shaped relationship. The rate of community turnover was highest (greatest dissimilarity in proportion of unshared species) in the elevation section between 1800 and 2200 m, while elevation ranges with the highest nestedness (greatest dissimilarity in proportion of shared species) were below 1200 m and above 2200 m. We found a weak correlation between the Euclidian geographic distance separating pairs of plots with their mean elevation (Spearman’s correlation r = −0.232) and with their taxonomic turnover (βjtu; Spearman’s correlation r = 0.080), suggesting that shift in spatial distance along elevation does not explain the observed pattern of beta-diversity. We also found a weak relationship between mean beta-diversity (βjtu) and the number of open areas (Spearman’s correlation r = 0.363), alpha- diversity (r = −0.231) and gamma-diversity (r = 0.165) in elevation bands of 20 m along the elevation gradient. Elevation was the only significant parameter in the model (OLS model: estimate

= 0.460, t value = 4.093, p value <0.001) explaining the variation in taxonomic turnover (βjtu). When partitioning the explained variance of mean beta-diversity (R2 = 0.21) among elevation and habitat variables (i.e. number of open areas and diversity of vegetation types) in the OLS model, elevation displayed the strongest independent effect, with a higher independent proportion of explained variance (11.7 %) than habitat variables (0.0 %).

The turnover (βjtu) component of beta-diversity was also uneven along elevation for Poales −4 −7 −4 (βjtu: linear s = 6.7 × 10 , quadratic s = −1.8 × 10 ; Fig. 3a), Super-Rosids (βjtu: linear s = 3 × 10 , −8 −4 quadratic s = −6.2 × 10 ; Fig. 3a) and Super-Asterids (βjtu: linear s = 6.5 × 10 , quadratic s = −1.9 × 10−7; Fig. 3a). Including quadratic terms improved the models for Poales (AIC difference −154.9), Super-Rosids (AIC difference −10.8) and Super-Asterids (AIC difference −292), supporting the existence of a hump-shaped relationship. However, Super-Rosids show a weaker hump-shaped relationship than Poales and Super-Asterids (Fig. 3a). The rate of community turnover was highest in the elevation section between 1700 and 2000 m for Poales, between 2000 and 2400 m for Super- Rosids and between 1600 and 1900 m for Super-Asterids (Fig. 3a).

- 43 -

Figure 2. Diversity changes along elevation gradients obtained by comparing plant communities within 20 m elevation bands as a measure of (a) mean community diversity (black points; mean alpha-diversity), total species richness (green points; gamma-diversity) and (b) proportion of species turnover (black points; turnover component of beta-diversity) and nestedness (green points; nestedness component of beta- diversity). Curves represent the quadratic relationships. Dashed lines represent the 5 and 95 percentiles.

Figure 3. Relationship between elevation and (a) proportion of plant species turnover (turnover component of beta-diversity) and (b) phylogenetic plant relatedness calculated as the mean pairwise distance (MPD) separating taxa in pairs of plant inventories of the same elevation (range 20 m) for Super-Asterids (black points; dashed line), Super-Rosids (blue points; dot-dashed line) and Poales (red points; solid line) clades. Curves represent the quadratic relationships.

- 44 -

The distribution of the species elevation maxima follows a hump-shaped curve with a peak between 2000 and 2200 m (Fig. 4). In contrast, the distribution of the species elevation minima shows a plateau between lowland and 1600 m and decreases rapidly between 1800 and 2400 m (Fig. 4). The important decrease of the minimum range values around 2000 m indicates that many species have their lower range limit around this elevation. As a result, many high and low elevation species have their lower and higher elevation limit near 2000 m, respectively.

Figure 4. Distribution of the minimum (black) and maximum (grey) elevation of the species ranges in elevation bands of 20 m for each 10 m along the elevation gradient in the study area. Curves were fitted with a GAM function

Community phylogenetic turnover of beta-diversity

We found a weak phylogenetic signal of niche conservatism (i.e. species-specific median of elevation distribution) across all plant species from the phylogeny (Blomberg’s K: K = 0.095, n = 231, Z score = −2.411, p value = 0.002), in Poales (Blomberg’s K: K = 0.174, n = 48, Z score = −1.649, p value = 0.022), Super-Rosids (Blomberg’s K: K = 0.251, n = 50, Z score = −1.589, p value = 0.003) and Super-Asterids (Blomberg’s K: K = 0.101, n = 122, Z score = −1.194, p value =

- 45 -

0.1). Only Super-Asterids showed a non-significant difference of Blomberg’s K compared to a null distribution across the phylogeny, suggesting that the plant niche of Super-Asterids does not follow patterns of phylogenetic inertia.

The turnover rate in phylogenetic beta-diversity was more even along elevation across all plant species from the phylogeny (All plant species: linear s = 2.6 × 10−2, quadratic s = −7.2 × 10−6) than the turnover rates in phylogenetic beta-diversity of Poales, Super-Rosids and Super-Asterids which were more uneven along elevation (Poales: linear s = 0.188, quadratic s = −5.5 × 10−5, Super- Rosids: linear s = 0.138, quadratic s = −3.3 × 10−5, Super-Asterids: linear s = −4.5 × 10−2, quadratic s = 1.5 × 10−5; Fig. 3b). Including quadratic terms improved the models of phylogenetic beta- diversity for all clades (AIC difference; All plant species = −179.3, Poales = −1646.4, Super-Rosids = −257.6, Super-Asterids = −233.6), supporting the existence of a non-linear relationship. The turnover rate in phylogenetic beta-diversity showed a strong hump-shaped pattern in Poales (peak between 1700 and 1900 m) and Super-Rosids (peak between 1900 and 2300 m), while turnover rate in Super-Asterids was more even along elevation (Fig. 3b). The explained variance in the relationship between phylogenetic beta-diversity and elevation was higher in Poales (R2 = 0.236) and Super-Rosids (R2 = 0.164) than Super-Asterids (R2 = 0.056) or across all plant species from the phylogeny (R2 = 0.035).

Plant clades show different patterns of distribution and dominance along elevation (Fig. 5, S4, S5, S6). Plant communities between 800 and 2800 m show a global decrease in the species proportion of Poales (from 28.6 % at 1300 m to 11.6 % at 2700 m; Fig. 5) along elevation, with an increase in Super-Rosids (from 21 % at 1500 m to 65.7 % at 2900 m; Fig. 5) and a contrasting pattern for Super-Asterids (minimum = 14.8 % at 2900 m, maximum = 47.3 % at 2100 m; Fig. 5). However, Poales species dominate plant communities in term of relative cover (i.e. bare soil and rock excluded; range 42.8–58.7 %; Fig. S5) compared to Super-Rosids (range 14.5–25 %; Fig. S5) and Super-Asterids (range 19.9–32.5 %; Fig. S5), except in the highest bands. In Poales, Poaceae species dominate plant communities of low elevation and show a constant decrease of their relative cover from 900 m (57.3 %) to 2700 m (31.3 %; Fig. S5), while Cyperaceae species show a constant increase of their relative cover from 900 m (1.4 %) to 2700 m (16.6 %; Fig. S5). In Super-Rosids and Super-Asterids, Fabaceae species have a higher relative cover at low elevations (10.4 % at 900 m; Fig. S5), Asteraceae and Apiaceae species at mid-elevations (11.8 % at 1900 m and 5.9 % at 1500 m, respectively; Fig. S5) and Saxifragaceae species dominate in relative cover plant communities of very high elevations (17.8 % at 2700 m, 91.4 % at 2900 m; Fig. S5). Yet, the net breakpoint in the dominance of Saxifragaceae species above 2800 m is partly due to the smaller plot

- 46 - sampling and the small number of species occurring at this elevation, with a more regular transition when the real cover (i.e. bare soil and rock included) is considered (Fig. S6).

Figure 5. Mean proportion of occurrences of the main vascular plant clades in plant communities in elevation bands of 200 m. Each colour represents the mean proportion of the plant clade in plant communities for the corresponding elevation band.

Overall, we found that the results were not sensitive to the resolution of elevation section considered, whether it is 10, 20 or 50 m (Figs. S7, S8).

Discussion

Measures of species turnover are essential tools to investigate assemblage shifts along environmental gradients (Williams 1996), and particularly along elevation gradients which are commonly used as proxies of shifts in abiotic (see Körner 2007) and biotic conditions (Reynolds and Crossley 1997; Michalet et al. 2006; Körner 2007; Pellissier et al. 2014b; Michalet et al. 2015). Using this analytical tool, we showed that turnover rate along elevation in grassland communities is not constant but peaks around 1800–2200 m, corresponding approximately to the regional treeline (1900 m, Gehrig-Fasel et al. 2007). The juxtaposition of highly dissimilar assemblages, large difference between gamma- and mean alpha-diversity (maximum between 1400 and 2100 m), and the high diversity of vegetation types (Fig. S3) indicate a singular ecological transition in this narrow elevation band (Figs. 2, 3, S3). This supports the hypothesis that across regional species pool in the Western Swiss Alps, many plant species share the same upper or lower elevation range limit (Fig. 4) and display “range boundary clumping” (Leibold and Mikkelson 2002). Human

- 47 -

disturbances and current land management have probably favoured the mosaic of habitats occurring at mid-elevation. However, from the habitat data available, we found no evidence that the larger open areas or the higher diversity of vegetation types observed at mid elevation explain the observed taxonomic and phylogenetic turnover near the treeline. Together, our results document the singularity of grassland taxonomic and phylogenetic turnover at the border between the subalpine and alpine belts. Our study suggests that, like trees, this ecotone (Theurillat et al. 2003) constitutes a strong barrier for some herbaceous plant clades, even after centuries of land use (Tinner and Theurillat 2003). Despite grazing having shaped a continuum of open vegetation types across the subalpine and alpine belts, which should have since long allowed species exchanges along the elevation gradient (Vittoz et al. 2009), our results indicate the persistence of this ecotone.

The transition from the subalpine to the alpine belts, where the turnover was the most acute, is associated with several changes in the abiotic environment, including more stressful temperatures and a shorter growing season (Körner 2007). Enduring stressful abiotic conditions in the alpine belt requires particular and often convergent adaptations (Pellissier et al. 2010), including a lower stature (Körner 2003), a decreasing specific leaf area (Salinas et al. 2011) associated to slower growth rates (Whittaker 1956). Plant distribution may also be limited by other abiotic factors in the alpine belt, including strong wind, ground instability, the type of underground rock (i.e. calcareous or siliceous), or frost during the growing season, which can result in functionally distinct plant communities at high compared to low elevations (Diaz and Cabido 1997). While clumped minima and maxima elevation range values observed around the treeline is expected to be partially controlled by abiotic factors (Diaz and Cabido 1997), changes in biotic interactions might also modulate plant range limit at the subalpine-alpine ecotone. In symmetry to the stressful abiotic conditions limiting lowland species in the alpine belt, higher competition among plant species (Choler et al. 2001; Alexander et al. 2015) and higher insect herbivore pressure on poorly defended alpine species in the subalpine belt likely limits the growth of small alpine plant species (Galen 1990; Bruelheide and Scheidel 1999; Pellissier et al. 2012, 2014b). However, high grazing pressure by cows or sheep in subalpine pastures can also reduce plant competition and favor the establishment of alpine plants at lower elevation (Vittoz et al. 2009). Moreover, higher species richness could also be favoured at intermediate position along environmental severity gradients as a result of decreasing inter-specific competition and increasing stress-tolerance with elevation (Michalet et al. 2006; Holmgren and Scheffer 2010; Verwijmeren et al. 2013; Michalet et al. 2015) favouring the mixture of low competitive species and high elevation stress-tolerant species (Michalet et al. 2015). Our results contrast with a study in Norway where no major discontinuity in

- 48 - species richness, composition or turnover was observed at the forest-limit ecotone (Odland and Birks 1999).

Taxonomic turnover component of beta-diversity was associated to uneven phylogenetic turnover rate along elevation. Poales and Super-Rosids showed a significant but weak phylogenetic conservatism of species range and an acceleration of phylogenetic turnover with a peak reached around 1900 m for both clades (Fig. 3b). This corroborates the finding of Ndiribe et al. (2013a, b) showing singular phylogenetic diversity patterns in Liliopsida (i.e. incl. Poales) along elevation. Several Poales and Super-Rosids lineages showed a preference either for the montane-subalpine or for the alpine environment, shaping the higher lineage turnover at the subalpine-alpine ecotone. For instance, the species in the genus Carex (Cyperaceae) show a preference for mid-elevation environments (i.e. 1300–2500 m; Fig. 5, Fig. S5), the species in the genus Saxifraga (Saxifragaceae) show a preference to colder environment above the treeline (i.e. >2100 m; Fig. 5, S5) and the species in the Fabaceae family show a preference for lower elevation environments (i.e. <2300 m; Fig. 5, S5). Many Carex species are tolerant to low temperatures (Körner 2003) and can be dominant and diversified in communities above the treeline (Grabherr 1989; Körner 2003). Conversely, Poaceae generally dominate grasslands below treeline (Fig. S5). In different regions, phylogenetic patterns in Liliopsida distinct from coexisting Magnolopsida have been reported along environmental gradients (Silvertown et al. 2001; Cahill et al. 2008). For instance, Cahill et al. (2008) observed that the intensity of competition showed a stronger phylogenetic signal in Liliopsida than Magnolopsida, as a consequence of higher niche conservatism in Liliopsida. Moreover, phylogenetic turnover is not always associated to functional turnover, due to possible convergence of traits between phylogenetically distinct species groups (Godoy et al. 2014). The lower phylogenetic conservatism of Super-Asterids range suggests that most lineages contain species that are distributed both above and below the treeline, explaining the constant lineage turnover rate along elevation (Fig. 3b; Chalmandrier et al. 2015). This constant phylogenetic turnover contrasts with the observed peak of taxonomic turnover at mid-elevation for Super- Asterids (Fig. 3), indicating that taxonomic turnover is not always associated to phylogenetic turnover. Similarly, Ndiribe et al. (2013a) found prevailing patterns of phylogenetic overdispersion in three families of the Super-Asterids clade (i.e. Apiaceae, Lamiaceae and Asteraceae families), indicating that closely related species diversified to occupy communities in contrasting environmental conditions, or that close relatives co-occur less often than expected. The low niche conservatism observed in Super-Asterids could be due to strong rates of evolution occurring in this clade favouring niche differentiation (Cooper et al. 2010) or the lack of high/low elevation specialised clades, which may in part be attributed to their life-history traits (Ndiribe et al. 2013a).

- 49 -

For instance, species of the Lamiaceae family have evolved phenolic compounds providing herbivore resistance and favouring their persistence in communities in contrasted environmental conditions (Grøndahl and Ehlers 2008).

Compared to gamma-diversity, the peak in alpha-diversity occurs at a lower elevation around 1100–1500 m, indicating that plant community richness is not necessary strictly associated to a higher species richness in a local pool. Alpha-diversity seems to reflect the disturbances occurring on plant communities at both ends of the elevation gradient, with more intensive land use (pastures, fertilisation) at low elevation increasing plant exclusion by competition (Eriksson et al. 1995; Foster and Gross 1998) and limiting the diversity of vegetation types (Fig. S3) and severe environmental conditions at high elevation allowing the growth of few stress-tolerant species. Before the intensification of agriculture, beginning around 60 years ago, dry and oligotrophic grasslands were more frequent below 1100 m (Lachat et al. 2010) and a similar analysis would have probably not resulted in so steep decline of alpha-diversity at low elevations. These grasslands are very species rich but are now very rare in the landscape at low elevations. A higher intensity of land use may also explain the lower gamma-diversity occurring at low elevation, which tends to homogenise the composition of plant communities. The land use at low elevation and the high elevation stress could explain the higher nestedness pattern below 1200 m and above 2200 m. In alpine habitat, communities with lower species richness are more frequently a subset of richer alpine communities. Since alpine habitats are supposed to be more stochastic due to stronger temporal variations in environmental conditions such as solifluction or landslides, some communities may suffer random loss of species shaping nestedness in the alpine belt (Körner 2003). Nestedness below 1200 m is probably the result of the intensive land use (i.e. pasturing, grazing, mowing, and fertilisation), which limits plant composition to the more competitive species, subset of richer lowland communities.

Climate change is currently increasing temperature in the Alps with rapid detectable changes in alpine plant communities (Pauli et al. 2012). Based on the present study, we can expect that the same temperature rise along the elevation gradient may not trigger the same amount of turnover rate in communities. In the transition between the subalpine and alpine belts, distinct flora are juxtaposed and only a strong ecological barrier appears to keep them apart. Climate change may lift the existing barrier across the subalpine-alpine ecotone, allowing for the upward movement and invasion of more competitive subalpine plants in the alpine grasslands, shaping novel assemblages and potentially causing local extinction of species in those communities (Alexander et al. 2015). Monitoring scheme investigating plant community changes along wide elevation gradients are required to evaluate the speed of changes (Vittoz et al. 2010). While the absence of change in the

- 50 - turnover rate of phylogenetic beta-diversity along elevation were documented (Bryant et al. 2008; Chalmandrier et al. 2015), our study reports a strong species turnover between the subalpine and alpine vegetation belts and suggests that climate change might, in turn, have an uneven impact on species’ range shifts across the elevation gradient.

Acknowledgments

We thank all the people involved in collecting the vegetation data over the years and two anonymous reviewers for their constructive comments on the manuscript. This project was supported by the Swiss National Science Foundation (SNSF) Grant Nos. 31003A-162604 and 31003A-1528661.

References

Alexander JM, Diez JM, Levine JM (2015) Novel competitors shape species’ responses to climate change. Nature 525:515–518. doi: 10.1038/nature14952 Bach K, Kessler M, Gradstein SR (2007) A simulation approach to determine statistical significance of species turnover peaks in a species-rich tropical cloud forest. Divers Distrib 13:863–870. doi: 10.1111/j.1472-4642.2007.00357.x Baselga A (2010) Partitioning the turnover and nestedness components of beta diversity. Glob Ecol Biogeogr 19:134–143. doi: 10.1111/j.1466-8238.2009.00490.x Baselga A (2012) The relationship between species replacement, dissimilarity derived from nestedness, and nestedness. Glob Ecol Biogeogr 21:1223–1232 Baselga A, Leprieur F (2015) Comparing methods to separate components of beta diversity. Methods Ecol Evol 6:1069–1079 Baselga A, Orme CDL (2012) betapart: an R package for the study of beta diversity. Methods Ecol Evol 3:808–812. doi: 10.1111/j.2041-210X.2012.00224.x Blomberg SP, Garland T, Ives AR (2003) Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57:717–745. doi: 10.1111/j.0014- 3820.2003.tb00285.x Bouët M (1985) Climat et météorologie de la Suisse romande. Payot, Lausanne Bruelheide H, Scheidel U (1999) Slug herbivory as a limiting factor for the geographical range of Arnica montana. J Ecol 87:839–848. doi: 10.1046/j.1365-2745.1999.00403.x Bryant JA, Lamanna C, Morlon H et al (2008) Microbes on mountainsides: contrasting elevational patterns of bacterial and plant diversity. Proc Natl Acad Sci USA 105:11505–11511. doi: 10.1073/pnas.0801920105

- 51 -

Cahill JF, Kembel SW, Lamb EG, Keddy PA (2008) Does phylogenetic relatedness influence the strength of competition among vascular plants? Perspect Plant Ecol Evol Syst 10:41–50. doi: 10.1016/j.ppees.2007.10.001 Chalmandrier L, Münkemüller T, Lavergne S, Thuiller W (2015) Effects of species’ similarity and dominance on the functional and phylogenetic structure of a plant meta-community. Ecology 96:143–153. doi: 10.1890/13-2153.1 Choler P, Michalet R, Callaway RM (2001) Facilitation and competition on gradients in alpine plant communities. Ecology 82:3295–3308. Clements FE (1916) Plant succession: an analysis of the development of vegetation. Carnegie Institution, Washington Cooper N, Jetz W, Freckleton RP (2010) Phylogenetic comparative approaches for studying niche conservatism. J Evol Biol 23:2529–2539 Culmsee H, Leuschner C (2013) Consistent patterns of elevational change in tree taxonomic and phylogenetic diversity across Malesian mountain forests. J Biogeogr 40:1997–2010. doi: 10.1111/jbi.12138 de Bello F, Lavorel S, Lavergne S et al (2013) Hierarchical effects of environmental filters on the functional structure of plant communities: a case study in the French Alps. Ecography (Cop) 36:393–402. doi: 10.1111/j.1600-0587.2012.07438.x Delarze R, Gonseth Y (2008) Guide des milieux naturels de Suisse. Ecologie—Menaces—Espèces caractéristiques. Rossolis, Bussigny Diaz S, Cabido M (1997) Plant functional types and ecosystem function in relation to global change. J Veg Sci 8:463–474. doi: 10.2307/3237198 Dubuis A, Pottier J, Rion V et al (2011) Predicting spatial patterns of plant species richness: a comparison of direct macroecological and species stacking modelling approaches. Divers Distrib 17:1122–1131 Eriksson Å, Eriksson O, Berglund H (1995) Species abundance patterns of plants in Swedish semi- natural pastures. Ecography 18:310–317 Fischer M, Rudmann-Maurer K, Weyand A, Stöcklin J (2008) Agricultural land use and biodiversity in the Alps—how cultural tradition and socioeconomically motivated changes are shaping grassland biodiversity in the Swiss Alps. Mt Res Dev 28:148–155 Foster BL, Gross KL (1998) Species richness in a successional grassland: effects of nitrogen enrichment and plant litter. Ecology 79:2593–2602 Galen C (1990) Limits to the distributions of alpine tundra plants: herbivores and the alpine skypilot, Polemonium viscosum. Oikos 59:355–358

- 52 -

Garibaldi LA, Kitzberger T, Chaneton EJ (2011) Environmental and genetic control of insect abundance and herbivory along a forest elevational gradient. Oecologia 167:117–129. doi: 10.1007/s00442-011-1978-0 Gaston KJ, Blackburn TIMM, Greenwoodx JD et al (2000) Abundance–occupancy relationships. J Appl Ecol 37:39–59. doi: 10.1046/j.1365-2664.2000.00485.x Gehrig-Fasel J, Guisan A, Zimmermann NE (2007) Tree line shifts in the Swiss Alps: climate change or land abandonment? J Veg Sci 18:571–582. doi: 10.1111/j.1654- 1103.2007.tb02571.x Gentry AH (1988) Changes in plant community diversity and floristic composition on environmental and geographical gradients. Ann Missouri Bot Gard 75:1–34CrossRefGoogle Scholar Gleason HA (1926) The individualistic concept of the plant association. Bull Torrey Bot Club 53:7– 26 Godoy O, Kraft NJ, Levine JM (2014) Phylogenetic relatedness and the determinants of competitive outcomes. Ecol Lett 17:836–844 Grabherr G (1989) On community structure in high alpine grasslands. Vegetatio 83:223–227. doi: 10.1007/BF00031694 Graham CH, Fine PVA (2008) Phylogenetic beta diversity: linking ecological and evolutionary processes across space in time. Ecol Lett 11:1265–1277. doi: 10.1111/j.1461- 0248.2008.01256.x Graham JH, Krzysik AJ, Kovacic DA et al (2009) Species richness, equitability, and abundance of ants in disturbed landscapes. Ecol Indic 9:866–877. doi: 10.1016/j.ecolind.2008.10.003 Grøndahl E, Ehlers BK (2008) Local adaptation to biotic factors: reciprocal transplants of four species associated with aromatic Thymus pulegioides and T. serpyllum. J Ecol 96:981–992 Hirzel A, Guisan A (2002) Which is the optimal sampling strategy for habitat suitability modelling. Ecol Modell 157:331–341 Holmgren M, Scheffer M (2010) Strong facilitation in mild environments: the stress gradient hypothesis revisited. J Ecol 98:1269–1275. doi: 10.1111/j.1365-2745.2010.01709.x Hutchinson EG (1957) Concluding remarks. Cold Spring Harb Symp Quant Biol 22:415–427 Jankowski JE, Ciecka AL, Meyer NY, Rabenold KN (2009) Beta diversity along environmental gradients: implications of habitat specialization in tropical montane landscapes. J Anim Ecol 78:315–327. doi: 10.1111/j.1365-2656.2008.01487.x Jankowski JE, Merkord CL, Rios WF et al (2013) The relationship of tropical bird communities to tree species composition and vegetation structure along an Andean elevational gradient. J Biogeogr 40:950–962. doi: 10.1111/jbi.12041

- 53 -

Kembel SW, Hubbell SP (2006) The phylogenetic structure of a neotropical forest tree community. Ecology 87:S86–S99 Kembel SW, Cowan PD, Helmus MR et al (2010) Picante: R tools for integrating phylogenies and ecology. Bioinformatics 26:1463–1464. doi: 10.1093/bioinformatics/btq166 Körner C (2000) Why are there global gradients in species richness? Mountains might hold the answer. Trends Ecol Evol 15:513–514. doi: 10.1016/S0169-5347(00)02004-8 Körner C (2003) Alpine plant life: functional plant ecology of high mountain ecosystems, 2nd edn. Springer Körner C (2007) The use of “altitude” in ecological research. Trends Ecol Evol 22:569–574. doi: 10.1016/j.tree.2007.09.006 Lachat T, Pauli D, Gonseth Y et al (2010) Wandel der Biodiversität in der Schweiz seit 1900. Ist die Talsohle erreicht ?. Haupt Verl, Berne Leibold MA, Mikkelson GM (2002) Coherence, species turnover, and boundary clumping: elements of meta-community structure. Oikos 97:237–250. doi: 10.1034/j.1600-0706.2002.970210.x Longino JT, Colwell RK (2011) Density compensation, species composition, and richness of ants on a neotropical elevational gradient. Ecosphere 2:art29. doi: 10.1890/ES10-00200.1 Mena JL, Vázquez-Domínguez E (2005) Species turnover on elevational gradients in small rodents. Glob Ecol Biogeogr 14:539–547. doi: 10.1111/j.1466-822X.2005.00189.x Michalet R, Brooker RW, Cavieres LA et al (2006) Do biotic interactions shape both sides of the humped-back model of species richness in plant communities? Ecol Lett 9:767–773. doi: 10.1111/j.1461-0248.2006.00935.x Michalet R, Maalouf JP, Choler P et al (2015) Competition, facilitation and environmental severity shape the relationship between local and regional species richness in plant communities. Ecography (Cop) 38:335–345. doi: 10.1111/ecog.01106 Ndiribe C, Pellissier L, Antonelli S et al (2013a) Phylogenetic plant community structure along elevation is lineage specific. Ecol Evol 3:4925–4939. doi: 10.1002/ece3.868 Ndiribe C, Pellissier L, Dubuis A et al (2013b) Plant functional and phylogenetic turnover correlate with climate and land use in the Western Swiss Alps. J Plant Ecol. doi: 10.1093/jpe/rtt064 Niedrist G, Tasser E, Luth C et al (2009) Plant diversity declines with recent land use changes in European Alps. Plant Ecol 202:195–210 Odland A, Birks HJB (1999) The altitudinal gradient of vascular plant richness in Aurland, western Norway. Ecography (Cop) 22:548–566 Pauli H, Gottfried M, Dullinger S, et al (2012) Recent plant diversity changes on Europe’s mountain summits. Science 336:353–355. doi: 10.1126/science.1219033

- 54 -

Pellissier L, Fournier B, Guisan A, Vittoz P (2010) Plant traits co-vary with altitude in grasslands and forests in the European Alps. Plant Ecol 211:351–365. doi: 10.1007/s11258-010-9794-x Pellissier L, Fiedler K, Ndribe C et al (2012) Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecol Evol 2:1818–1825. doi: 10.1002/ece3.296 Pellissier L, Alvarez N, Espindola A et al (2013a) Phylogenetic alpha and beta diversities of butterfly communities correlate with climate in the western Swiss Alps. Ecography (Cop) 36:541–550. doi: 10.1111/j.1600-0587.2012.07716.x Pellissier L, Ndiribe C, Dubuis A et al (2013b) Turnover of plant lineages shapes herbivore phylogenetic beta diversity along ecological gradients. Ecol Lett 16:600–608. doi: 10.1111/ele.12083 Pellissier L, Niculita-Hirzel H, Dubuis A et al (2014a) Soil fungal communities of grasslands are environmentally structured at a regional scale in the Alps. Mol Ecol 23:4274–4290. doi: 10.1111/mec.12854 Pellissier L, Roger A, Bilat J, Rasmann S (2014b) High elevation Plantago lanceolata plants are less resistant to herbivory than their low elevation conspecifics: is it just temperature? Ecography (Cop) 37:950–959. doi: 10.1111/ecog.00833 Pottier J, Dubuis A, Pellissier L et al (2013) The accuracy of plant assemblage prediction from species distribution models varies along environmental gradients. Glob Ecol Biogeogr 22:52– 63. doi: 10.1111/j.1466-8238.2012.00790.x Qian H, Ricklefs RE, White PS (2005) Beta diversity of angiosperms in temperate floras of eastern Asia and eastern North America. Ecol Lett 8:15–22. doi: 10.1111/j.1461-0248.2004.00682.x Reynolds BC, Crossley DAJ (1997) Spatial variation in herbivory by forest canopy arthropods along an elevation gradient. Environ Entomol 26:1232–1239 Salinas N, Malhi Y, Meir P et al (2011) The sensitivity of tropical leaf litter decomposition to temperature: results from a large-scale leaf translocation experiment along an elevation gradient in Peruvian forests. New Phytol 189:967–977. doi: 10.1111/j.1469- 8137.2010.03521.x Sanders NJ (2002) Elevational gradients in ant species richness: area, geometry, and Rapoport’s rule. Ecography (Cop) 25:25–32. doi: 10.1034/j.1600-0587.2002.250104.x Silvertown J, Dodd M, Gowing D (2001) Phylogeny and the niche structure of meadow plant communities. J Ecol 89:428–435. doi: 10.1046/j.1365-2745.2001.00553.x Soberón J (2007) Grinnellian and Eltonian niches and geographic distributions of species. Ecol Lett 10:1115–1123

- 55 -

Theurillat J, Schlüssel A, Geissler P, et al (2003) Vascular plant and bryophyte diversity along elevation gradients in the Alps. In: Alpine biodiversity in Europe. Springer, Berlin, pp 185– 193 Tinner W, Theurillat JP (2003) Uppermost limit, extent, and fluctuations of the timberline and treeline ecocline in the Swiss central Alps during the past 11,500 years. Arct Antarct Alp Res 35:158–169. Tuomisto H (2010) A diversity of beta diversities: straightening up a concept gone awry. Part 1. Defining beta diversity as a function of alpha and gamma diversity. Ecography (Cop) 33:2– 22. doi: 10.1111/j.1600-0587.2009.05880.x Ulrich W, Gotelli NJ (2007) Null model analysis of species nestedness patterns. Ecology 88:1824– 1831. doi: 10.1890/06-1208.1 Verwijmeren M, Rietkerk M, Wassen MJ, Smit C (2013) Interspecific facilitation and critical transitions in arid ecosystems. Oikos 122:341–347. doi: 10.1111/j.1600-0706.2012.00111.x Vittoz P, Randin C, Dutoit A et al (2009) Low impact of climate change on subalpine grasslands in the Swiss Northern Alps. Glob Chang Biol 15:209–220. doi: 10.1111/j.1365- 2486.2008.01707.x Vittoz P, Camenisch M, Mayor R et al (2010) Subalpine-nival gradient of species richness for vascular plants, bryophytes and lichens in the Swiss Inner Alps. Bot Helv 120:139–149. doi: 10.1007/s00035-010-0079-8 Whittaker RH (1956) Vegetation of the Great Smoky Mountains. Ecol Monogr 26:1–80 Wiens JJ, Ackerly DD, Allen AP et al (2010) Niche conservatism as an emerging principle in ecology and conservation biology. Ecol Lett 13:1310–1324. doi: 10.1111/j.1461- 0248.2010.01515.x Williams PH (1996) Mapping variations in the strength and breadth of biogeographic transition zones using species turnover. Proc R Soc London B Biol Sci 263:579–588 Wright DH, Reeves JH (1992) On the meaning and measurement of nestedness of species assemblages. Oecologia 92:416–428. doi: 10.1007/BF00317469

- 56 -

Supplementary materials

Appendix S1. Methodological details about OLS models and variable partition analyses

OLS models

We compared the turnover component of beta-diversity to elevation and habitat variables (frequency of open areas and diversity of vegetation types) by using an ordinary least squares regression (OLS) model and including all predictor variables and quadratic terms to account for non-linear relationships. The proportion of variation in the turnover component of beta-diversity explained by the OLS models was quantified with the coefficient of determination (R2).

Variable partition analyses

In addition, we quantified the relative importance of elevation vs. habitat variables (i.e. frequency of open areas and diversity of vegetation types) for explaining beta-diversity variation by using a variance partitioning analyses (Borcard et al., 1992). This analysis decomposes the proportion of variation in beta-diversity explained by the full OLS model (R2) into two sources of variation by means of partial regressions (Legendre, 2012): (i) variation due to the independent effect of the elevation variable, (ii) variation due to the independent effect of habitat variables, and (iii) variation due to the combined effect of elevation and habitat variables. When partitioning the explained variance of beta-diversity (R2 = 0.21), among elevation and habitat variables in OLS full models, elevation displayed the strongest independent effect, with a higher independent proportion of explained variance (11.7 %) than habitat variables (0.0%).

References

Borcard D, Legendre P, & Drapeau P (1992) Partialling out the Spatial Component of Ecological Variation. Ecology 73:1045–1055. Legendre L (2012) Numerical Ecology. Elsevier, Amsterdam.

- 57 -

Figure S1. Number of open areas (black points) and plots (red points) in elevation bands of 20 m for each 10 m along the elevation gradient in the study area. The unit of open areas is the number of pixels in elevation bands at a resolution of 25 m.

- 58 -

Figure S2. Principal Components Analysis based on plot location (points) related to six environmental variables (blue arrows): elevation, annual sum of precipitation, annual sum of degree-days (better factor than mean temperature to explain plan distribution; Zimmermann and Kienast 1999), annual sum of solar radiation, curvature (related to shape of land, with negative values for concave areas, positive values for convex areas and 0 for flat areas or regular slopes) and slope. The two first axes of the initial PCA explained 73.4 % of the total variance (axis 1, 54.9 %; axis 2, 18.5 %). The length of the vectors represents the magnitude of the correlation between the variables and the axes.

- 59 -

Figure S3. Number of vegetation types among inventoried plots, within 20 m elevation bands centered on each 10 m elevation steps along the elevation gradient in the study area. The plots were grouped with a hierarchical clustering and the groups were attributed to a vegetation type, according to the classification of Delarze and Gonseth (2008), on the basis of their respective differential species. The curve was fitted with a GAM function.

- 60 -

Figure S4. Proportion of the main vascular plant clades in the species pool (gamma-diversity) in elevation bands of 200 m. The different shades of red-orange correspond to Poales, the blue shades to Super-Rosids and the grey-black shades to Super-Asterids.

Figure S5. Mean relative cover (bare soil and rock excluded) of the main vascular plant clades in plant communities in elevation bands of 200 m. See Fig. S5 for the real mean cover of the main vascular plant clades in plant communities when taking in account bare soil and rock. The different shades of red-orange correspond to Poales, the blue shades to Super-Rosids and the grey-black shades to Super-Asterids.

- 61 -

Figure S6. Mean cover of the main vascular plant clades and cover of bare soil and rock in plant communities in elevation bands of 200 m. The different shades of red-orange correspond to Poales, the blue shades to Super-Rosids and the grey-black shades to Super-Asterids.

- 62 -

Figure S7. Diversity changes along the elevation gradient obtained by comparing plant communities of the same elevation (range; a and b: 10 m; c and d: 50 m) as a measure of (a and c) mean community diversity (black points; mean alpha-diversity), total species richness (green points; gamma-diversity) and (b and d) proportion of species turnover (black points; turnover component of beta-diversity) and nestedness (green points; nestedness component of beta-diversity). Curves represent the quadratic relationships. Dashed lines represent the 5 and 95 percentiles.

- 63 -

Figure S8. Relationship between elevation and (a and c) proportion of plant species turnover (turnover component of beta-diversity) and (b and d) phylogenetic plant relatedness calculated as the mean pairwise distance (MPD) separating taxa in pairs of plant inventories of the same elevation (range; a and b: 10 m, c and d: 50 m) for Super-Asterids (black points), Super-Rosids (blue points) and Poales (red points) clades. Curves represent the quadratic relationships.

References

Delarze R, Gonseth Y (2008) Guide des milieux naturels de Suisse. Ecologie – Menaces – Espèces caractéristiques. Rossolis, Bussigny Zimmermann NE, Kienast F (1999) Predictive mapping of alpine grasslands in Switzerland : Species versus community approach. J Veg Sci 10:469–482. doi: 10.2307/3237182

- 64 -

CHAPTER 2

Community‐level plant palatability increases with elevation as insect herbivore abundance declines

Patrice Descombes1,2,3, Jeremy Marchon1, Jean-Nicolas-Pradervand4, Julia Bilat1, Antoine Guisan4,5, Sergio Rasmann6, Loïc Pellissier2,3

1Unit of Ecology & Evolution, University of Fribourg, Fribourg, Switzerland 2Landscape Ecology Institute of Terrestrial Ecosystems, ETH Zürich, Zürich, Switzerland 3Swiss Federal Research Institute WSL Birmensdorf, Switzerland 4Department of Ecology and Evolution University of Lausanne, Lausanne, Switzerland 5Institute of Earth Surface Dynamics, University of Lausanne, Lausanne, Switzerland 6 Laboratory of Functional Ecology, Institute of Biology, University of Neuchâtel, Neuchâtel, Switzerland

Published in Journal of Ecology (2017), 105, 142-151 doi: https://doi.org/10.1111/1365-2745.12664 Post-print version

- 65 -

Summary

Plants protect themselves against herbivore attacks through a myriad of physical structures and toxic secondary metabolites. Together with abiotic factors, herbivores are expected to modulate plant defence strategies within plant assemblages. Because the abundance of insect herbivore decreases in colder environments, the palatability of plants in communities at higher elevation should shift in response to both abiotic and biotic factors. We inventoried grasshopper communities to document changes in herbivore abundance along elevation gradients and quantified associated shifts in plant palatability. We measured plant palatability by measuring the growth of Spodoptera littoralis generalist caterpillars fed with the leaves of 172 plant species. We related plant palatability to leaf traits and elevation at the species and community levels. In congruence with the decrease in grasshopper abundance with elevation, we found that the mean palatability level of plant communities increases with elevation. In addition, plant palatability was negatively associated with the community‐weighted mean of leaf dry matter content. At the species level, plants with high carbon‐to‐nitrogen ratio were less palatable, while we found no effect of species mean elevation on plant palatability. Our results suggest that plant communities at higher elevation are composed of species that are generally more palatable for insect herbivores. Shift in plant palatability with elevation may thus be the outcome of a relaxation of the in situ herbivore pressure and changes in abiotic conditions.

Introduction

Changes in plant functional properties along environmental gradients are influenced by both abiotic conditions (e.g. temperature, edaphic factors; e.g. Asner et al. 2014) and biotic interactions (e.g. plant–animal interactions; Gentry 1988; Sundqvist, Sanders & Wardle 2013). Along elevation gradients, the degree of herbivory generally decreases (Reynolds & Crossley 1997; Garibaldi, Kitzberger & Chaneton 2011; Metcalfe et al. 2014; Pellissier et al. 2014), concomitantly with changes in climatic and soil factors (Körner 2007), which should influence functional composition of plant assemblages. For instance, it can be postulated that plants at higher elevations should invest their limited resources more in resistance against abiotic stressors (e.g. cold, wind, UV radiation), other than the unnecessary production of defences against herbivores (Pellissier et al. 2014). While examples of intraspecific and interspecific defence relaxation at high elevations or latitude exist (Pennings, Siska & Bertness 2001; Scheidel & Bruelheide 2001; Salgado & Pennings 2005; Pellissier et al. 2012, 2014; Rasmann et al. 2014a; but see Moles et al. 2011a,b), it remains unclear whether those genotypic‐ or species‐specific patterns can be generalized to entire communities.

- 66 -

Plants have evolved a large diversity of defences to protect themselves against herbivores (Grime, MacPherson‐Stewart & Dearman 1968; Rhoades 1979; Coley, Bryant & Chapin 1985; Agrawal & Fishbein 2006; Hanley et al. 2007). Chemical defences primarily rely on carbon‐ and nitrogen‐based secondary compounds that act as toxins or digestibility reducers (Mithöfer & Boland 2012). In complement, physical defences act via traits that physically decrease the potential for food acquisition (e.g. leaf toughness, trichomes; Awmack & Leather 2002; Hanley et al. 2007). The different components of the plants’ defence arsenal are expressed together in the form of syndromes to counter‐attack a wide range of herbivore guilds (Agrawal & Fishbein 2006). How elevation gradients shape the deployment of plant defence syndromes has not been addressed across many species, partly because of the high endeavour required for identifying and quantifying defences of plant species belonging to a wide range of families, each one with a unique set of phytochemicals and forms (Wink 2003). An alternative approach for measuring the net outcome of plant defence traits is therefore to use bioassays that quantify the palatability of each plant species against a unique, but highly polyphagous herbivore (e.g. Grime, MacPherson‐Stewart & Dearman 1968; Edwards, Wratten & Cox 1985; Pellissier et al. 2012).

Evidence suggests that herbivore communities can influence plant species composition (e.g. Carson & Root 2000; Fine, Mesones & Coley 2004) and related plant functional traits. For instance, Becerra (2007) or Kursar et al. (2009) detected a signal of chemical overdispersion within communities of closely related plant species and suggested that plant–insect co‐evolution modulate plant communities’ structure. Pellissier et al. (2013) found a correlation between herbivore and plant phylogenetic beta diversity, suggesting a link between plant and insect community assembly, while Richards et al. (2015) showed that the richness of herbivores correlated with phytochemical diversity in Piper species assemblages. In accordance, we therefore predict that in habitats where the abundance and richness of herbivores is large, the relative pressure of distinct herbivore taxa will impose a top‐down control on plant composition and maintain a relatively strong and diverse defence syndrome (e.g. Kursar & Coley 2003; Becerra 2007). In contrast, in habitats with lower herbivory pressure, such as higher elevation, plant species should invest less in herbivore resistance and plant communities should be on average more palatable.

Abiotic factors could also shape community‐level trait variation, in turn also modifying plant responses to herbivory (Rasmann et al. 2014b). For instance, at high elevation, plant species may produce leaves with increased leaf dry matter content (LDMC) as an adaptation to severe climatic conditions (Körner 2003; Dubuis et al. 2013). Because LDMC is positively related to leaf toughness, where high LDMC plants are more resistant to physical hazards (i.e. wind, hail; Cornelissen et al. 2003), it might also indirectly confer increased resistance to insect herbivores

- 67 -

(Schuldt et al. 2012; Ibanez et al. 2013). Therefore, the selective abiotic forces might promote functional changes in plant communities, which could indirectly confer more or less resistance to herbivores. However, the link between herbivore abundance and richness, abiotic conditions and plant palatability at the level of communities has not been thoroughly investigated so far.

In this study, we assessed plant palatability for chewing insect herbivores, physical defence traits and nutritional composition of plant species growing between 400 and 3200 m of elevation, and comprising common plant community composition at each site. Plant palatability was quantified by measuring the growth of the larvae of the generalist moth Spodoptera littoralis (Brown & Dewhurst 1975) on leaves collected in their natural growing locations. The weight of the larvae should be considered as the integrated measure of the physical and chemical defence for each plant species, where high values of larval weight indicate lower defence against herbivores (Bossdorf et al. 2004; Schädler et al. 2007; Ruhnke et al. 2009; Pellissier et al. 2012). Additionally, we surveyed for natural grasshopper abundance and richness along the same elevation gradient to document the expected decrease in abundance and richness in colder environments. We chose grasshoppers as indicators of herbivore abundance since these insects are among the most important herbivores in calcareous grasslands, where they can remove up to 30% of the above‐ground phanerogam biomass (Blumer & Diemer 1996). Using this data set, we investigated how plant palatability is simultaneously related to leaf functional traits and elevation, at the species and at the community level, and how it is likely associated to a shift in herbivore abundance.

Materials and methods

Plant and orthopteran communities

The study area is located in the Western Swiss Alps (Switzerland, 46°10′–46°30′ N; 6°50′– 07°10′ E) and covers about 700 km2, with elevations ranging between 375 and 3210 m a.s.l (see Fig. S1 in Supporting Information). It is characterized by a temperate climate with annual average temperatures and precipitations varying, respectively, between 8 °C and 1200 mm at 600 m, and −5 °C and 2600 mm at 3000 m (Bouët 1985). The vegetation along the elevation gradient is typical of the calcareous Alps, but also strongly influenced by the human land use and pasture.

In order to sample vegetation data along elevation gradients, a total of 912 plots of 4 m2 were selected between 400 and 3210 m using a random stratified sampling design with regard to elevation, slope and orientation (Hirzel & Guisan 2002), which were inventoried between May and September 2002–2010 (see Fig. S1). All plots were selected among open, non‐woody areas by counting with a minimal distance of 200 m between plots in order to limit the potential spatial

- 68 - autocorrelation resulting from field inventories performed in the same meadow and presenting a very similar plant composition. All vascular plant species present in the plots were inventoried, and their relative abundance estimated using the simplified cover scheme based on Vittoz & Guisan (2007): <0.1, 0.1–1, 1–5, 5–15, 15–25, 25–50, 50–75 and >75%. The median values of these classes (0.05, 0.5, 3, 10, 20, 37.5, 62.5 and 82.5%) were used in all further analyses.

To assess the abundance of orthopteran along elevation gradients, we randomly selected 175 plots above 1000 m out of the initial 912 plots (see Fig. S1). Between 20th July and 20th September 2012, sites were visited from low elevation to high elevation by following the species phenology. Most of the sites below 1800 m a.s.l. (corresponding approximately to the lower limit of the treeline ecotone in the study area; Gehrig‐Fasel, Guisan & Zimmermann 2007) were visited a second time later in the season to collect data on late emerging species due to a longer growing season. The sampling took place between 10 a.m. and 5 p.m. with optimal temperature conditions for insects, and within a 50 × 50 m area. Most individuals were identified directly on the field to the species level by net catching and/or by their songs. Unknown species and larvae were collected for identification in the laboratory and by experts from the Swiss Center from Faunal Cartography (http://www.cscf.ch/). We inventoried a total of 36 orthopteran species, including members of the Caelifera and Ensifera suborder. The abundance of each species was then estimated for a 10 × 10 m area using four classes of abundance: 1–5, 5–20, 20–50, 50–100 individuals. This value of abundance was estimated by means of four 10 × 10 m subplots situated in each cardinal directions (North, South, East and West) at 10 m from the central point and sampled circularly by walking towards the centre. For sites visited a second time later in the season, only the abundance of the additional species was assessed (see Table S1 for the species list and abundance observed on each sampling site). A global measure of abundance for each plot was obtained for Caelifera and Ensifera separately, by summing the median values of the classes of abundance (3, 13, 35 and 75, respectively) of all species. We omitted members of the Tetrigidae family, since these species are principally feeding on algae, lichens, mosses and detritus (e.g. Paranjape & Bhalerao 1985; Kuřavová & Kočárek 2015). We assumed that the measure of global grasshopper abundance is correlated with overall herbivore pressure on the plant community at each site. We extracted elevation information for all sampled plots from a Digital Elevation Model at a resolution of 25 m.

Plant palatability bioassay and plant traits

In order to evaluate overall plant palatability for chewing insect herbivores, we performed a bioassay experiment on 172 plant species commonly growing within the study area, using larvae of the African cotton leafworm S. littoralis (Lepidoptera, Noctuidae; Brown & Dewhurst 1975)

- 69 -

obtained from Syngenta (Switzerland). We used S. littoralis as a non‐adapted species to remove the confounding effect of possible local adaptation to plants. In addition, S. littoralis is a highly generalist insect herbivore, reported to feed on more than 40 families of plants (Brown & Dewhurst 1975), therefore commonly used in similar bioassays (e.g. Edwards, Wratten & Cox 1985; Bossdorf et al. 2004; Schädler et al. 2007; Ruhnke et al. 2009; Pellissier et al. 2012). While herbivores are frequently specialized on a restricted range of host plants due to possible pre‐adaptations to quantitative and qualitative leaf traits of a given plant species (e.g. Mopper 1996; Pellissier et al. 2013; Rasmann et al. 2014b), the assessment of differences in leaf quality by using a generalist herbivore such as S. littoralis should provide a more unbiased measure (Ruhnke et al. 2009). The 172 plant species were selected to represent the diversity of families (number of sampled species relative to total number in each family; Spearman correlation: 0.942) and to represent diversity variation along the elevation gradient (number of sampled versus total species in each elevation bands of 200 m; Spearman correlation: 0.991). Three individuals of each plant species were collected, by selecting plants in sites with contrasting environmental conditions along an elevation gradient transect in the study area (see Fig. S1) so as to cover as much of the total distribution range of each species as possible. Eggs were hatched on wet paper at 20 °C without food to ensure a standard size. Once hatched, three larvae were placed on each individual plant, by placing leaves of each species in distinct Petri dishes for 5 days in a climatic chamber at 24 °C (L) and 18 °C (D), 55 ± 5% RH and a 14:10 L:D photoperiod. Completely eaten or dried leaves were replaced during this period with leaves that were stored at 4 °C. At the end of the experiment, all the larvae were dried for 72 h at 50 °C and weighed. We retained dead larvae into the analyses only if replicated experiments showed the same results (i.e. larvae died in the Petri dishes). Finally, we estimated the average palatability of the 172 plant species by averaging the dry weight of the larvae in and through replicates of the same plant species.

We next measured three traits related to physical leaf defence and leaf nutrient content: specific leaf area (SLA), leaf dry matter content (LDMC) and carbon‐to‐nitrogen ratio (C:N). SLA and LDMC were estimated for 245 plant species, while C:N was estimated for 251 plant species. We collected 4–20 individuals in sites with contrasting environmental conditions along an elevation gradient transect (see Fig S1) so as to cover as much of the total distribution range of each species as possible (see Dubuis et al. 2013). Species individuals were sampled at the same phenological stage whenever possible by following the growing season according to altitude, stored in moist bags in a cool box (10 °C) and rehydrated previous to measurements by using the partial rehydration method described in Vaieretti et al. (2007). One well‐developed entire leaf was then collected per individual (Cornelissen et al. 2003; Pérez‐Harguindeguy et al. 2013) for trait measurement and

- 70 - dried at 60 °C for minimum 4 days. SLA (mm2 mg-1) was calculated as the area of the leaf divided by its dry mass. SLA is correlated with the potential relative growth rate or mass‐based photosynthetic rate of a plant, where lower values are associated to higher investments in structural defence strategies, and long leaf lifespan (Cornelissen et al. 2003). LDMC (mg g-1) was calculated as the ratio of the leaf dry mass to its saturated fresh mass. LDMC is positively related to leaf toughness, where high LDMC plants present higher resistance to physical hazards (i.e. herbivory, wind, hail; Cornelissen et al. 2003) and lower digestibility (Gardarin et al. 2014). Leaf nitrogen (mg g-1) and carbon (mg g-1) contents were analysed on one sample of mixed ground leaves per species by using an elemental analyser (NC‐2500 from CE Instruments). C and N are linked to plant photosynthetic rates and nutrient cycling processes (Cornelissen et al. 2003). We used the average trait value among all sampled individuals for each species for further analyses.

Statistical analyses

First, we evaluated phylogenetic signal of all plant functional traits described above and plant palatability, by pruning from the published phylogeny of the 231 most frequent and abundant plant species for the study area (Ndiribe et al. 2013), and by calculating Blomberg's K statistic with the ‘phylosignal’ function as implemented in the ‘picante’ r package (Blomberg, Garland & Ives 2003; Kembel et al. 2010) in r (R Development Core Team 2014, r version 3.2.2). Blomberg's K statistic compares the observed distribution of the trait values to expectations under a Brownian motion model of trait evolution. K values close to one indicate trait evolution consistent with a Brownian motion model of evolution, while K values close to zero indicate a random distribution of trait values with respect to the phylogeny (Blomberg, Garland & Ives 2003). We tested the significance of this test by comparing the observed K value to a null distribution generated by comparing 999 randomizations of trait values across the tips of the phylogenetic tree (Kembel et al. 2010).

Next, we related species‐level variation in plant palatability to plant trait (SLA, LDMC and C:N) and plant mean elevation, by fitting a phylogenetic least squares model (PGLS) with the package ‘caper’ (Orme et al. 2013) in r, with λ transformation for the phylogenetic tree optimized through maximum likelihood, and by square root transforming the response variable and rescaling all variables around their mean. We ensured that the model residuals did not deviate from a normal distribution. PGLS model allows correcting for phylogenetic non‐independence among species while correlating two variables. Since collinearity can bias parameter estimation in regression‐type models, we calculated a variance inflation factor (VIF; Quinn & Keough 2002) for our predictors by using the ‘vifstep’ function in the package ‘usdm’ (Naimi 2015). This function calculates a VIF for all variables and excludes highly correlated variables from the set through a stepwise procedure

- 71 -

based on a threshold. An ideal VIF has a value of one. While VIF values higher than 10 are clearly indicative of collinearity issues (Quinn & Keough 2002), it has also been suggested that values higher than three could also be indicative of potential collinearity issues (Zuur, Ieno & Elphick 2010; Mundry 2014). We kept all predictors since we found low collinearity (VIF values; SLA = 1.859, C:N = 1.343; LDMC = 1.269, species mean elevation = 1.341). Finally, to ensure that our conclusions were non‐sensitive to the choice of the elevation metric at the species level, we run additional analysis using elevation range limits with quantile 5% and 95% in addition to the mean elevation.

To investigate the relationships between the plants traits and the environment at the community level, we computed the community‐weighted mean (CWM), which represents the mean of a trait for a whole plant community, weighted by the abundance of each species that occur in the community (Garnier et al. 2004). We retained only plots (i.e. 307 plots) whose cover was composed with more than 70% of species for which species trait measurement was available. For each 307 plots, we computed CWM of plant palatability, SLA, LDMC and C:N by using the ‘functcomp’ function provided by the package ‘FD’ (Laliberté, Legendre & Shipley 2014). We related CWM of plant palatability to the plant traits and community elevation using an ordinary least squares regression model (OLS), by square root transforming the response variable and by rescaling all variables around their mean. We calculated a variance inflation factor (VIF; Quinn & Keough 2002) for our predictors following the same procedure as above. We excluded SLA from the regression model since this variable showed high VIF value (SLA = 4.440; C:N = 2.589; community elevation = 2.263; LDMC = 2.241). After removing SLA, VIF values reached 1.842 for C:N, 1.813 for LDMC and 1.025 for elevation. We also ensured that the model residuals did not deviate from a normal distribution and that the model residuals were not spatially autocorrelated. To ensure that our conclusions were non‐sensitive to the choice of cover threshold used for selecting plot (Pakeman & Quested 2007), we run additional analyses when considering only plots whose cover was composed with more than 80% (n = 167) or 90% (n = 72) of species for which species trait measurement was available.

Finally, we related orthopteran (Caelifera and Ensifera suborder separately) richness (i.e. number of species) and abundance (i.e. counts of individuals) to elevation using a linear regression model (i.e. 175 plots). The proportion of variation in Caelifera and Ensifera richness or abundance explained by the model was quantified with the coefficient of determination (R2).

- 72 -

Results

Species‐level analyses

We found a weak phylogenetic signal for LDMC (Blomberg's K: K = 0.335, n = 218, Z‐score = −4.981, P‐value = 0.001), SLA (K = 0.124, n = 218, Z‐score = −2.702, P‐value = 0.001), C:N (K = 0.138, n = 221, Z‐score = −2.217, P‐value = 0.001) and plant palatability (K = 0.122, n = 133, Z‐ score = −0.627, P‐value = 0.265; Fig. 1), indicating that the variation of these plant traits is labile across the phylogeny. Despite general trait lability, we observed family‐level differences in plant palatability (Fig. 1, see Fig. S2). Some plant families such as Apiaceae, Cyperaceae and Poaceae present lower palatability levels, while Polygonaceae and Salicaceae exhibited higher palatability. Other families such as Asteraceae and Saxifragaceae showed high variability (Fig. 1, see Fig. S2).

Figure 1. Phylogeny of angiosperms palatability (n = 133). The trait mapped correspond to the mean larval weight (square root transformed) of Spodoptera littoralis after 5 days of feeding on leaf plant samples, as a measure of plant palatability. Bigger circles indicate a higher palatability of the plant to generalist chewing insect herbivores (see Fig. S2 for a complete phylogeny with plant species names).

- 73 -

At the species level and among all predictor variables, only C:N was significantly associated with plant palatability in the PGLS model, which showed a low explained deviance of the relationship (Table 1a; Fig. 2). We found no relationship between plant palatability and LDMC, SLA and mean elevation of sites where the species was found (Table 1a). The results were consistent, whether we used the mean elevation at the species level, the quantile 5% or the quantile 95% representing niche limits (see Table S2).

Table 1. (a) Relationships between plant palatability and four predictor variables at the plant species level (n = 129) estimated from phylogenetic least squares model (PGLS) including all predictor variables and bivariate linear regressions. (b) Relationships between community‐weighted mean (CWM) of plant palatability and four predictor variables at the plant community level estimated from ordinary least squares multiple regressions (OLS) and bivariate linear regressions. CWM was estimated for communities whose plant cover was composed with more than 70% of species for which species trait measurement was available (n = 307). SLA was not considered in the OLS model since it was highly correlated to the other predictors. The table (a and b) shows the coefficient of determination (R2), the t‐value and the standardized regression coefficients (Estimate). SLA, specific leaf area; LDMC, leaf dry matter content; C:N, carbon‐to‐nitrogen content; ELEV, mean elevation. *P < 0.05, ***P < 0.001.

(a) Species level PGLS Bivariate model 2 Estimate t-value Estimate t-value R SLA 0.009 0.081 ns 0.097 1.093 0.002 ns LDMC -0.090 -0.698 ns -0.178 -2.037 0.024 * C:N -0.256 -2.536 * -0.201 -2.312 0.033 * ELEV -0.062 -0.650 ns 0.009 0.101 -0.008 ns 2 R 0.054 * lamda 0.416

(b) Community level OLS Bivariate model 2 Estimate t-value Estimate t-value R SLA -0.028 -0.489 -0.002 ns LDMC -0.397 -6.132 *** -0.429 -8.298 0.182 *** C:N -0.058 -0.893 ns -0.279 -5.079 0.075 *** ELEV 0.343 7.034 *** 0.327 6.042 0.104 *** 2 R 0.292 ***

- 74 -

Figure 2. Partial regression plot (i.e. added variable plot) calculated from PGLS model (n = 129) showing the independent contribution of C:N in explaining plant species palatability variation (i.e. solid black line; Table 2a) at the species level. The axis represents the residuals of the models (x‐axis: C:N~ SLA + LDMC + ELEV; y‐axis: Palatability~SLA + LDMC + ELEV). Before fitting the PGLS model, plant palatability was square root transformed and all variables were rescaled around their mean. The dashed line corresponds to the mean of the y‐axis. SLA (specific leaf area); LDMC (leaf dry matter content); C:N (carbon‐to‐nitrogen content); ELEV (mean elevation). Plants are more palatable when their leaves are more nutrient rich (i.e. low C:N).

Community‐level analyses

We found that the CWM of plant palatability was significantly associated to the CWM of LDMC and community elevation in OLS multiple regressions (Table 1b, Fig. 3). Plant communities are more palatable at higher elevation, where they are composed of plant species affording lower levels of LDMC (Table 1b, Fig. 3). In contrast to the results at the species level, the CWM of plant palatability was not associated to the CWM of C:N in OLS multiple regressions (Table 1b). We found no autocorrelation in the residuals of the CWM model (Moran's I = 0.009; P‐value = 0.209). Overall, we found that the results were not sensitive to the threshold considered for selecting plot communities, whether it is 70, 80 or 90% of cover in plots for which species trait measurement was available (see Table S3).

- 75 -

Figure 3. Partial regression plots (i.e. added variable plots) calculated from OLS model showing the independent contribution of (a) LDMC and (b) community elevation in explaining plant community palatability variation (i.e. solid black line; Table 2b). The axis represents the residuals of the models. CWM of plant palatability was estimated for communities whose plant cover was composed with more than 70% of species for which species trait measurement was available (n = 307). Before fitting the OLS model, plant palatability was square root transformed and all variables were rescaled around their mean. The grey area corresponds to the 95% confidence interval around the mean. The dashed line corresponds to the mean of the y‐axis. LDMC (leaf dry matter content); C:N (carbon‐to‐nitrogen content); ELEV (mean elevation). Communities are more palatable at high elevation sites and/or when they are composed of plant species presenting lower LDMC.

Finally, orthopteran richness (i.e. number of species) and abundance (i.e. counts of individuals) were both negatively correlated to elevation in Caelifera (linear regression; richness: n = 175, R2 = 0.407, t‐value = −10.970, slope = −0.004, P‐value <0.001; abundance: n = 175, R2 = 0.212, t‐value = −6.925, slope = −0.093, P‐value <0.001; Fig. 4) and Ensifera (linear regression; richness: n = 175, R2 = 0.528, t‐value = −13.99, slope = −0.002, P‐value <0.001; abundance: n = 175, R2 = 0.257, t‐value = −7.828, slope = −0.017, P‐value <0.001; Fig. 4). We also found a strong correlation between orthopteran richness and abundance in Caelifera (Spearman correlation: ρ = 0.845, R2 = 0.714, P‐value <0.001) and Ensifera (Spearman correlation: ρ = 0.823, R2 = 0.678, P‐ value <0.001), as a result of the contrast between low elevation rich and abundant orthopteran communities versus high elevation poor communities (Fig. 4).

- 76 -

Figure 4. Changes in (a) orthopteran richness measured as the number of species and (b) orthopteran abundance (i.e. counts of individuals) along the elevation gradient for Caelifera (black dots) and Ensifera (grey triangles) suborder. Each dot and each triangle represent a sampled community where orthopteran abundance and richness were estimated (n = 175; see Fig. S1). The grey and black areas correspond to the 95% confidence interval around the mean for Ensifera and Caelifera, respectively. Orthopteran Caelifera and Ensifera are more abundant and diverse on low elevation sites.

Discussion

Species‐level plant palatability along elevation gradients

Plant defence levels may vary along an elevation gradient by following two expectations. The first one relies on the fact that producing defensive chemicals is costly (Gulmon & Mooney 1986; Gershenzon 1994; Cipollini, Purrington & Bergelson 2003) and predicts a decrease of defence with elevation, in line with the concomitant decrease in herbivore abundance. The second one relies on the resource availability hypothesis (Coley, Bryant & Chapin 1985; Endara & Coley 2011), which states that plants with slow growth rates occurring in environments with low resources should be highly defended against herbivory because of the high cost of tissue loss in resource‐poor environments. Accordingly, as environmental harshness and low resources availability limit plant growth rate at high elevation, we would expect high elevation plants to exhibit high levels of defence against herbivory. The fact that we found no effect of elevation on plant palatability at the species level (see similar results in Rasmann et al. 2014b) suggests that different species may show dissimilar sensitivities to herbivore abundance and abiotic conditions, leading to a lack of a clear trend when considering all species individually (Rasmann et al. 2014b).

- 77 -

In accordance with the common rule of insects being mostly deficient in nitrogen, larvae performed slightly better on plants with a low carbon‐to‐nitrogen content (Table 1a, Fig. 2; White 1984; Elser et al. 2000). Other leaf variables such as leaf toughness (represented here by LDMC and SLA) were not associated to plant palatability, which is in contrast with previous results in which SLA was associated with investment in physical leaf defence (Coley 1983; Cornelissen et al. 2003; but see Louda & Rodman 1996). Since values of SLA and LDMC are highly plastic across sites, it is likely that more samples would be required to capture intraspecific variability and to accurately catch the multifaceted dimensions of plant defences based on physical traits only (Ackerly 2009; Cornwell & Ackerly 2009). In addition, this study did not take into account plant secondary metabolites specifically, which are known to play a preponderant role in driving insect performance (Fraenckel 1959; but see Carmona, Lajeunesse & Johnson 2011). Plants chemical defences are difficult to study across such a large plant taxonomic scale because the secondary metabolites involved are extremely diverse (Mithöfer & Boland 2012; Kant et al. 2015), but may largely explain the observed difference in palatability among plant species.

We found a weak phylogenetic signal for plant leaf traits such as SLA, LDMC and C:N, even if the latter were significantly different from a random distribution of plant traits across the phylogeny. This suggests that variation of several plant functional traits across species shows a pattern of adaptation to the environment, with limited phylogenetic inertia (Wiens & Graham 2005; Losos 2008). Similarly, we found no phylogenetic signal for plant palatability suggesting that mechanisms of plant defence (physical and chemical combined) are probably species‐specific, and independent of the phylogeny at the scale we performed our analysis. Indeed, palatability could show phylogenetic conservatism at deeper (e.g. family level) nodes of the phylogeny. For instance, most of the species in the Apiaceae, Poaceae and Cyperaceae show lower palatability to insect herbivores (Fig. 1), suggesting that family‐specific defence traits, such as secondary metabolites (e.g. furanocoumaris for Apiaceae; Berenbaum, Zangerl & Nitao 1986), or silica content in grasses and sedges (O'Reagain & Mentis 1989; Vicari & Bazely 1993; Massey, Ennos & Hartley 2006; Massey & Hartley 2009) might drive conservatism of plant–herbivore interaction (Futuyma & Agrawal 2009). However, testing for family differences would require a larger species sampling than considered here.

Community‐level plant palatability along elevation gradients

At the community level, we observed an increase of the overall plant palatability with elevation, suggesting that the most dominant species in plant communities growing at higher elevations are generally more palatable to herbivores than their low elevation counterparts. This

- 78 - result is in line with the observed decrease of orthopteran abundance and richness with elevation and suggests that dominant plants respond to reduced herbivore abundance by relaxing their defences overall (Pellissier et al. 2012), supporting the hypothesis of defence relaxation with elevation (Gulmon & Mooney 1986; Gershenzon 1994). Previous studies showed a similar relationship between plant chemical and biomechanical defence traits at the community-level and herbivore abundance and richness (Becerra 2007; Peeters, Sanson & Read 2007; Richards et al. 2015). However, those study at the level of communities never extended at higher latitude or elevation.

Together with elevation, we found that the mean community‐level value of LDMC was also related to plant community palatability (Table 1b, Fig. 3). In other words, plant communities situated at higher elevation and/or presenting higher mean community‐level value of LDMC are generally less palatable to herbivores. This result parallels finding at the species level in which leaf toughness was associated with investment in physical leaf defence (Coley 1983; Choong 1996; Hochuli 2001; Cornelissen et al. 2003; Clissold et al. 2009; Ibanez et al. 2013; but see Louda & Rodman 1996). Abiotic factors could also shape community‐level trait variation, in turn also modifying plant responses to herbivory (Rasmann et al. 2014b). For instance, plant species at high elevation may first produce leaves with increased leaf dry matter content (LDMC) as an adaptation to severe climatic conditions (Körner 2003; Dubuis et al. 2013), thus increasing resistance to physical hazards (i.e. wind, hail; Cornelissen et al. 2003). In turn, this might indirectly confer increased resistance to arthropod herbivores (Schuldt et al. 2012; Ibanez et al. 2013).

In sum, we found contrasted results in the species‐ and community‐level analyses. Plant palatability was related to LDMC and elevation at the community level, while only C:N showed a trend in species‐based analyses driven by a few species (Fig. 2). While in species‐level analyses each species is given the same weight, at the community level, trait values are weighed by the dominance of species (i.e. higher weight to dominant plant species; Garnier et al. 2004; Pellissier et al. 2012). This suggests that the relationships between plant palatability, LDMC and elevation are driven predominantly by species with a higher cover. Dominant plant species are expected to be more frequently targeted by herbivores and to invest more energy in physical defence to generalist herbivore (Pellissier et al. 2015). For plant species predominantly relying on physical defences, a decrease in LDMC would directly result in higher palatability. In contrast, for species also relying on chemical defences, the relationship would be less clear. Moreover, the increase in plant palatability at higher elevation communities might suggest that dominant plants better modulate defences to shifting ecological conditions, such as biotic interaction with herbivores. Together, our

- 79 -

results indicate that plant dominance should be included when addressing the ecology and evolution of plant defence strategies.

Our findings indicate that the measure of community‐level plant palatability could be one of the processes explaining how climate change will affect plant community composition in the future. Many insects are currently limited in their distribution by temperature and length of the growing season (Maclean 1983; Strathdee et al. 1993; Whittaker & Tribe 1996; Miles, Bale & Hodkinson 1997; Bird & Hodkinson 1999; Hodkinson et al. 1999; Ritchie 2000). However, the higher temperatures associated with climate change will allow invertebrate herbivores to track climatic changes (Wilson et al. 2007) to an extent that plants cannot (Grabherr, Gottfried & Pauli 1994 but see Cannone, Sgorbati & Guglielmin 2007). For instance, Bässler et al. (2013) found an upslope shift of the upper range margin for insects that exceeded expectations based on climatic changes. The predicted future increase in herbivore abundance at high elevations under climate changes (Rasmann et al. 2014b) will increase the pressure on palatable plant communities. Herbivory could have in turn a disproportionate negative impact on the cover of these species (Brown & Gange 1989) and may cause severe shifts in the pattern of plant and soil carbon cycling (Metcalfe et al. 2014). Ultimately, palatable species could disappear from high elevation communities because insects may selectively feed on them, facilitating their replacement by lower elevation plant species (Rasmann et al. 2014b). Moreover, increasing stress conditions through climate change (e.g. drought) could also alter the defence of plants against insect herbivory by reducing investments in secondary metabolites (Gutbrodt, Mody & Dorn 2011).

In conclusion, our study showed that the decrease in plant palatability at higher elevation can be scaled up to entire communities. In contrast, we found only weak trends when looking at species individually. Future work using community‐wide metabolomics approaches for obtaining more information on leaf chemistry, and identifying the dominant chemical defence strategies along the elevation gradient might help overcoming these obstacles (Van Dam & Poppy 2008; Jansen et al. 2009). High elevation areas represent refuges from herbivory for some alpine plants: they are excluded from low elevations because of the high herbivory pressure in lowland (Galen 1990; Bruelheide & Scheidel 1999; Bruelheide, 2003). However, climate change is expected to lift current climate barriers and herbivore colonization at higher elevation may promote fast plant communities turnover, in which less palatable high elevation plants will be selected.

- 80 -

Acknowledgements

We thank all the people who helped with field work, Oliver Kindler and Roland Reist (Syngenta, Stein, Switzerland) for providing S. littoralis eggs, and the editor and two anonymous referees who provided valuable comments to improve the manuscript. The project was financed by the Swiss National Fund grant ‘Lif3web’ number 31003A‐162604 to L.P.

References

Ackerly, D. (2009) Conservatism and diversification of plant functional traits: evolutionary rates versus phylogenetic signal. Proceedings of the National Academy of Sciences of the United States of America, 106, 19699–19706. Agrawal, A.A. & Fishbein, M. (2006) Plant defense syndromes. Ecology, 87, S132–S149. Asner, G.P., Martin, R.E., Tupayachi, R., Anderson, C.B., Sinca, F., Carranza‐Jiménez, L. & Martinez, P. (2014) Amazonian functional diversity from forest canopy chemical assembly. Proceedings of the National Academy of Sciences of the United States of America, 111, 5604–5609. Awmack, C.S. & Leather, S.R. (2002) Host plant quality and fecundity in herbivorous insects. Annual Review of Entomology, 47, 817–844. Bässler, C., Hothorn, T., Brandl, R. & Müller, J. (2013) Insects overshoot the expected upslope shift caused by climate warming. PLoS ONE, 8, e65842. Becerra, J.X. (2007) The impact of herbivore‐plant coevolution on plant community structure. Proceedings of the National Academy of Sciences of the United States of America, 104, 7483–7488. Berenbaum, M.R., Zangerl, A.R. & Nitao, J.K. (1986) Constraints on chemical coevolution: wild parsnips and the parsnip webworm. Evolution, 40, 1215–1228. Bird, J.M. & Hodkinson, I.D. (1999) Species at the edge of their range: the significance of the thermal environment for the distribution of congeneric Craspedolepta species (Sternorrhyncha: Psylloidea) living on Chamerion angustifolium (Onagraceae). European Journal of Entomology, 96, 103–109. Blomberg, S.P., Garland, T. & Ives, A.R. (2003) Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution, 57, 717–745. Blumer, P. & Diemer, M. (1996) The occurrence and consequences of grasshopper herbivory in an alpine grassland, Swiss Central Alps. Arctic and Alpine Research, 28, 435–440.

- 81 -

Bossdorf, O., Schröder, S., Prati, D. & Auge, H. (2004) Palatability and tolerance to simulated herbivory in native and introduced populations of Alliaria petiolata (Brassicaceae). American Journal of Botany, 91, 856–862. Bouët, M. (1985) Climat et Météorologie de La Suisse Romande. Payot, Lausanne, Switzerland. Brown, E.S. & Dewhurst, C.F. (1975) The genus Spodoptera (Lepidoptera: Noctuidae) in Africa and the Near East. Bulletin of Entomological Research, 65, 221–262. Brown, V.K. & Gange, A.C. (1989) Differential effects of above and below ground herbivory during early plant succession. Oikos, 54, 67–76. Bruelheide, H. (2003) Translocation of a montane meadow to simulate the potential impact of climate change. Applied Vegetation Science, 6, 23–34. Bruelheide, H. & Scheidel, U. (1999) Slug herbivory as a limiting factor for the geographical range of Arnica montana. Journal of Ecology, 87, 839–848. Cannone, N., Sgorbati, S. & Guglielmin, M. (2007) Unexpected impacts of climate change on alpine vegetation. Frontiers in Ecology and the Environment, 5, 360–364. Carmona, D., Lajeunesse, M.J. & Johnson, M.T.J. (2011) Plant traits that predict resistance to herbivores. Functional Ecology, 25, 358–367. Carson, W.P. & Root, R.B. (2000) Herbivory and plant species coexistence: community regulation by an outbreaking phytophagous insect. Ecological Monographs, 70, 73–99. Choong, M.F. (1996) What makes a leaf tough and how this affects the pattern of Castanopsis fissa leaf consumption by caterpillars. Functional Ecology, 10, 668–674. Cipollini, D., Purrington, C.B. & Bergelson, J. (2003) Costs of induced responses in plants. Basic and Applied Ecology, 4, 79–85. Clissold, F.J., Sanson, G.D., Read, J. & Simpson, S.J. (2009) Gross vs. net income: how plant toughness affects performance of an insect herbivore. Ecology, 90, 3393–3405. Coley, P.D. (1983) Herbivory and defensive characteristics of tree species in a lowland tropical forest. Ecological Monographs, 53, 209–233. Coley, P.D., Bryant, J.P. & Chapin, F.S. (1985) Resource availability and plant antiherbivore defense. Science, 230, 895–899. Cornelissen, J.H.C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D.E., Reich, P.B., Ter Steege, H., Morgan, H.D., Van Der Heijden, M.G.A., Pausas, J.G. & Poorter, H. (2003) A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany, 51, 335–380. Cornwell, W.K. & Ackerly, D.D. (2009) Community assembly and shifts in plant trait distributions across an environmental gradient in coastal California. Ecological Monographs, 79, 109–126.

- 82 -

Descombes, P., Marchon, J., Pradervand, J.‐N., Bilat, J., Guisan, A., Rasmann, S. & Pellissier, L. (2016) Data from: Community‐level plant palatability increases with elevation as insect herbivore abundance declines. Figshare, http://dx.doi.org/10.6084/m9.figshare.3803955. Dubuis, A., Rossier, L., Pottier, J., Pellissier, L., Vittoz, P. & Guisan, A. (2013) Predicting current and future spatial community patterns of plant functional traits. Ecography, 36, 1158–1168. Edwards, P.J., Wratten, S.D. & Cox, H. (1985) Wound‐induced changes in the acceptability of tomato to larvae of Spodoptera littoralis: a laboratory bioassay. Ecological Entomology, 10, 155–158. Elser, J.J., Fagan, W.F., Denno, R.F., Dobberfuhl, D.R., Folarin, A., Huberty, A., Interlandi, S., Kilham, S.S., McCauley, E., Schulz, K.L., Siemann, E.H. & Sterner, R.W. (2000) Nutritional constraints in terrestrial and freshwater food webs. Nature, 408, 578–580. Endara, M.J. & Coley, P.D. (2011) The resource availability hypothesis revisited: a meta‐analysis. Functional Ecology, 25, 389–398. Fine, P.V.A., Mesones, I. & Coley, P.D. (2004) Herbivores promote habitat specialization by trees in Amazonian forests. Science, 305, 663–665. Fraenckel, G.S. (1959) The raison d’être of secondary plants substances. Science, 129, 1466–1470. Futuyma, D.J. & Agrawal, A.A. (2009) Macroevolution and the biological diversity of plants and herbivores. Proceedings of the National Academy of Sciences of the United States of America, 106, 18054–18061. Galen, C. (1990) Limits to the distributions of alpine tundra plants: herbivores and the alpine skypilot, Polemonium viscosum. Oikos, 59, 355–358. Gardarin, A., Garnier, É., Carrère, P., Cruz, P., Andueza, D., Bonis, A. et al. (2014) Plant trait‐ digestibility relationships across management and climate gradients in permanent grasslands. Journal of Applied Ecology, 51, 1207–1217. Garibaldi, L.A., Kitzberger, T. & Chaneton, E.J. (2011) Environmental and genetic control of insect abundance and herbivory along a forest elevational gradient. Oecologia, 167, 117–129. Garnier, E., Cortez, J., Billès, G., Navas, M.L., Roumet, C., Debussche, M., Laurent, G., Blanchard, A., Aubry, D., Bellmann, A., Neill, C. & Toussaint, J.P. (2004) Plant functional markers capture ecosystem properties during secondary succession. Ecology, 85, 2630–2637. Gehrig‐Fasel, J., Guisan, A. & Zimmermann, N.E. (2007) Tree line shifts in the Swiss Alps: climate change or land abandonment? Journal of Vegetation Science, 18, 571–582. Gentry, A.H. (1988) Changes in plant community diversity and floristic composition on environmental and geographical gradients. Annals of the Missouri Botanical Garden, 75, 1– 34.

- 83 -

Gershenzon, J. (1994) The cost of chemical defense against herbivory: a biochemical perspective. Insect‐Plant Interactions (ed. E.A. Bernays), pp. 105–173. CRC Press, Boca Raton, FL, USA. Grabherr, G., Gottfried, M. & Pauli, H. (1994) Climate effects on mountain plants. Nature, 369, 448. Grime, J.P., MacPherson‐Stewart, S.F. & Dearman, R.S. (1968) An investigation of leaf palatability using the snail Cepaea nemoralis L. Journal of Ecology, 56, 405–420. Gulmon, S.L. & Mooney, H.A. (1986) Costs of defense and their effects on plant productivity. On the Economy of Plant Form and Function (ed. T.J. Givnnish), pp. 681–698. Cambridge University Press, Cambridge, UK. Gutbrodt, B., Mody, K. & Dorn, S. (2011) Drought changes plant chemistry and causes contrasting responses in lepidopteran herbivores. Oikos, 120, 1732–1740. Hanley, M.E., Lamont, B.B., Fairbanks, M.M. & Rafferty, C.M. (2007) Plant structural traits and their role in anti‐herbivore defence. Perspectives in Plant Ecology, Evolution and Systematics, 8, 157–178. Hirzel, A. & Guisan, A. (2002) Which is the optimal sampling strategy for habitat suitability modelling. Ecological Modelling, 157, 331–341. Hochuli, D.F. (2001) Insect herbivory and ontogeny: how do growth and development influence feeding behaviour, morphology and host use? Austral Ecology, 26, 563–570. Hodkinson, I.D., Bird, J., Miles, J.E., Bale, J.S. & Lennon, J.J. (1999) Climatic signals in the life histories of insects: the distribution and abundance of heather psyllids (Strophingia spp.) in the UK. Functional Ecology, 13, 83–95. Ibanez, S., Lavorel, S., Puijalon, S. & Moretti, M. (2013) Herbivory mediated by coupling between biomechanical traits of plants and grasshoppers. Functional Ecology, 27, 479–489. Jansen, J.J., Allwood, J.W., Marsden‐Edwards, E., van der Putten, W.H., Goodacre, R. & van Dam, N.M. (2009) Metabolomic analysis of the interaction between plants and herbivores. Metabolomics, 5, 150–161. Kant, M.R., Jonckheere, W., Knegt, B., Lemos, F., Liu, J., Schimmel, B.C.J. et al. (2015) Mechanisms and ecological consequences of plant defence induction and suppression in herbivore communities. Annals of Botany, 115, 1015–1051. Kembel, S.W., Cowan, P.D., Helmus, M.R., Cornwell, W.K., Morlon, H., Ackerly, D.D., Blomberg, S.P. & Webb, C.O. (2010) Picante: R tools for integrating phylogenies and ecology. Bioinformatics, 26, 1463–1464. Körner, C. (2003) Alpine Plant Life: Functional Plant Ecology of High Mountain Ecosystems. Springer, Berlin, Germany.

- 84 -

Körner, C. (2007) The use of “altitude” in ecological research. Trends in Ecology & Evolution, 22, 569–574. Kuřavová, K. & Kočárek, P. (2015) Seasonal variation in the diet of Tetrix tenuicornis (Orthoptera: Tetrigidae). Entomological Science, 18, 489–501. Kursar, T.A. & Coley, P.D. (2003) Convergence in defense syndromes of young leaves in tropical rainforests. Biochemical Systematics and Ecology, 31, 929–949. Kursar, T.A., Dexter, K.G., Lokvam, J., Pennington, R.T., Richardson, J.E., Weber, M.G., Murakami, E.T., Drake, C., McGregor, R. & Coley, P.D. (2009) The evolution of antiherbivore defenses and their contribution to species coexistence in the tropical tree genus Inga. Proceedings of the National Academy of Sciences of the United States of America, 106, 18073–18078. Laliberté, E., Legendre, P. & Shipley, B. (2014) FD: measuring functional diversity from multiple traits, and other tools for functional ecology. R package version 1.0‐12. Losos, J.B. (2008) Phylogenetic niche conservatism, phylogenetic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecology Letters, 11, 995–1007. Louda, S.M. & Rodman, J.E. (1996) Insect herbivory as a major factor in the shade distribution of a native crucifer (Cardamine cordifolia A. Gray, bittercress). Journal of Ecology, 84, 229–237. Maclean, S.F. (1983) Life cycles and the distribution of psyllids (Homoptera) in arctic and subarctic Alaska (willows Salix). Oikos, 40, 445–451. Massey, F.P., Ennos, A.R. & Hartley, S.E. (2006) Silica in grasses as a defence against insect herbivores: contrasting effects on folivores and a phloem feeder. Journal of Animal Ecology, 75, 595–603. Massey, F.P. & Hartley, S.E. (2009) Assessing the accuracy of species distribution models to predict amphibian species richness patterns. Journal of Animal Ecology, 78, 281–291. Metcalfe, D.B., Asner, G.P., Martin, R.E., Silva Espejo, J.E., Huasco, W.H., Farfán Amézquita, F.F. et al. (2014) Herbivory makes major contributions to ecosystem carbon and nutrient cycling in tropical forests. Ecology Letters, 17, 324–332. Miles, J.E., Bale, J.S. & Hodkinson, I.D. (1997) Effects of temperature elevation on the population dynamics of the upland heather psyllid Strophingia ericae (Curtis) (Homoptera: Psylloidea). Global Change Biology, 3, 291–297. Mithöfer, A. & Boland, W. (2012) Plant defense against herbivores: chemical aspects. Annual Review of Plant Biology, 63, 431–450.

- 85 -

Moles, A.T., Bonser, S.P., Poore, A.G.B., Wallis, I.R. & Foley, W.J. (2011a) Assessing the evidence for latitudinal gradients in plant defence and herbivory. Functional Ecology, 25, 380–388. Moles, A.T., Wallis, I.R., Foley, W.J., Warton, D.I., Stegen, J.C., Bisigato, A.J. et al. (2011b) Putting plant resistance traits on the map: a test of the idea that plants are better defended at lower latitudes. New Phytologist, 191, 777–788. Mopper, S. (1996) Adaptive genetic structure in phytophagous insect populations. Trends in Ecology & Evolution, 11, 235–238. Mundry, R. (2014) Statistical issues and assumptions of phylogenetic generalized least squares. Modern Phylogenetic Comparative Methods and Their Application in Evolutionary Biology (ed. L.Z. Garamszegi), pp. 131–153. Springer‐Verlag, Berlin, Germany. Naimi, B. (2015) Uncertainty Analysis for Species Distribution Models. R package version 1.1‐15., https://cran.r-project.org/web/packages/usdm/index. Ndiribe, C., Pellissier, L., Antonelli, S., Dubuis, A., Pottier, J., Vittoz, P., Guisan, A. & Salamin, N. (2013) Phylogenetic plant community structure along elevation is lineage specific. Ecology and Evolution, 3, 4925–4939. O'Reagain, P.J. & Mentis, M.T. (1989) Leaf silicification in grasses – a review. Journal of the Grassland Society of Southern Africa, 6, 37–43. Orme, D., Freckleton, R., Thomas, G., Petzoldt, T., Fritz, S., Isaac, N. & Pearse, W. (2013) CAPER: Comparative Analyses of Phylogenetics and Evolution in R. R package version 0.5.2. Pakeman, R.J. & Quested, H.M. (2007) Sampling plant functional traits: what proportion of the species need to be measured? Applied Vegetation Science, 10, 91–96. Paranjape, S.Y. & Bhalerao, A.M. (1985) Bioecological observations on a Pigmy Locust, Potua sabulosa Hancock (Tetrigidae: Orthoptera). Psyche, 92, 331–336. Peeters, P.J., Sanson, G. & Read, J. (2007) Leaf biomechanical properties and the densities of herbivorous insect guilds. Functional Ecology, 21, 246–255. Pellissier, L., Fiedler, K., Ndribe, C., Dubuis, A., Pradervand, J.N., Guisan, A. & Rasmann, S. (2012) Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecology and Evolution, 2, 1818–1825. Pellissier, L., Ndiribe, C., Dubuis, A., Pradervand, J.‐N., Salamin, N., Guisan, A. & Rasmann, S. (2013) Turnover of plant lineages shapes herbivore phylogenetic beta diversity along ecological gradients. Ecology Letters, 16, 600–608.

- 86 -

Pellissier, L., Roger, A., Bilat, J. & Rasmann, S. (2014) High elevation Plantago lanceolata plants are less resistant to herbivory than their low elevation conspecifics: is it just temperature? Ecography, 37, 950–959. Pellissier, L., Litsios, G., Fishbein, M., Salamin, N., Agrawal, A.A. & Rasmann, S. (2015) Different rates of defense evolution and niche preferences in clonal and nonclonal milkweeds (Asclepias spp.). New Phytologist, 209, 1230–1239. Pennings, S.C., Siska, E.L. & Bertness, M.D. (2001) Latitudinal differences in plant palatability in Atlantic coast salt marshes. Ecology, 82, 1344–1359. Pérez‐Harguindeguy, N., Diaz, S., Garnier, E., Lavorel, S., Poorter, H., Jaureguiberry, P. et al. (2013) New handbook for standardized measurement of plant functional traits worldwide. Australian Journal of Botany, 61, 167–234. Quinn, G.P. & Keough, M.J. (2002) Experimental Designs and Data Analysis for Biologists. Cambridge University Press, Cambridge, UK. R Development Core Team (2014) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Rasmann, S., Buri, A., Gallot‐Lavallée, M., Joaquim, J., Purcell, J. & Pellissier, L. (2014a) Differential allocation and deployment of direct and indirect defences by Vicia sepium along elevation gradients. Journal of Ecology, 102, 930–938. Rasmann, S., Pellissier, L., Defossez, E., Jactel, H. & Kunstler, G. (2014b) Climate‐driven change in plant‐insect interactions along elevation gradients. Functional Ecology, 28, 46–54. Reynolds, B.C. & Crossley, D.A.J. (1997) Spatial variation in herbivory by forest canopy arthropods along an elevation gradient. Environmental Entomology, 26, 1232–1239. Rhoades, D. (1979) Evolution of plant chemical defence against herbivores. Herbivores: Their Interaction With Secondary Plant Metabolites (eds G.A. Rosenthal & D.H. Janzen), pp. 3–54. Academic Press, New York, NY, USA. Richards, L.A., Dyer, L.A., Forister, M.L., Smilanich, A.M., Dodson, C.D., Leonard, M.D. & Jeffrey, C.S. (2015) Phytochemical diversity drives plant – insect community diversity. Proceedings of the National Academy of Sciences of the United States of America, 112, 10973–10978. Ritchie, M.E. (2000) Nitrogen limitation and trophic vs. abiotic influence on insect herbivores in a temperate grassland. Ecology, 81, 1601–1612. 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.

- 87 -

Salgado, C.S. & Pennings, S.C. (2005) Latitudinal variation in palatability of salt‐marsh plants: are differences constitutive? Ecology, 86, 1571–1579. Schädler, M., Roeder, M., Brandl, R. & Matthies, D. (2007) Interacting effects of elevated CO2, nutrient availability and plant species on a generalist invertebrate herbivore. Global Change Biology, 13, 1005–1015. Scheidel, U. & Bruelheide, H. (2001) Altitudinal differences in herbivory on montane compositae species. Oecologia, 129, 75–86. Schuldt, A., Bruelheide, H., Durka, W., Eichenberg, D., Fischer, M., Kröber, W. et al. (2012) Plant traits affecting herbivory on tree recruits in highly diverse subtropical forests. Ecology Letters, 15, 732–739. Strathdee, A.T., Bale, J.S., Block, W.C., Coulson, S.J., Hodkinson, I.D. & Webb, N.R. (1993) Effects of temperature elevation on a field population of Acyrthosiphon svalbardicum (Hemiptera: Aphididae) on Spitsbergen. Oecologia, 96, 457–465. Sundqvist, M.K., Sanders, N.J. & Wardle, D.A. (2013) Community and ecosystem responses to elevational gradients: processes, mechanisms, and insights for global change. Annual Review of Ecology, Evolution, and Systematics, 44, 261–280. Vaieretti, M.V., Díaz, S., Vile, D. & Garnier, E. (2007) Two measurement methods of leaf dry matter content produce similar results in a broad range of species. Annals of Botany, 99, 955– 958. Van Dam, N.M. & Poppy, G.M. (2008) Why plant volatile analysis needs bioinformatics‐detecting signal from noise in increasingly complex profiles. Plant Biology, 10, 29–37. Vicari, M. & Bazely, D.R. (1993) Do grasses fight back? The case for antiherbivore defences. Trends in Ecology & Evolution (Personal edition), 8, 137–141. Vittoz, P. & Guisan, A. (2007) How reliable is the monitoring of permanent vegetation plots? A test with multiple observers. Journal of Vegetation Science, 18, 413–422. White, T.C.R. (1984) The abundance of invertebrate herbivores in relation to the availability of nitrogen in stressed food plants. Oecologia, 63, 90–105. Whittaker, J.B. & Tribe, N.P. (1996) An altitudinal transect as an indicator of responses of insects to climate change. European Journal of Entomology, 93, 319–324. Wiens, J.J. & Graham, C.H. (2005) Niche conservatism: integrating evolution, ecology, and conservation biology. Annual Review of Ecology, Evolution, and Systematics, 36, 519–539. Wilson, R.J., Gutiérrez, D., Gutiérrez, J. & Monserrat, V.J. (2007) An elevational shift in butterfly species richness and composition accompanying recent climate change. Global Change Biology, 13, 1873–1887.

- 88 -

Wink, M. (2003) Evolution of secondary metabolites from an ecological and molecular phylogenetic perspective. Phytochemistry, 64, 3–19. Zuur, A.F., Ieno, E.N. & Elphick, C.S. (2010) A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution, 1, 3–14.

- 89 -

Supplementary materials

Table S1. Species list observed on the sampling sites with their estimated abundance.

Plot identity Elevation Species Abundance 687 1074.4 CHORTHIPPUS BRUNNEUS 35 687 1074.4 MECOSTETHUS PARAPLEURUS 3 687 1074.4 PHOLIDOPTERA GRISEOAPTERA 3 687 1074.4 EUTHYSTIRA BRACHYPTERA 3 687 1074.4 NEMOBIUS SYLVESTRIS 3 687 1074.4 TETTIGONIA CANTANS 3 661 1077.2 CHORTHIPPUS PARALLELUS 75 661 1077.2 METRIOPTERA ROESELII 13 661 1077.2 OMOCESTUS VIRIDULUS 35 661 1077.2 POLYSARCUS DENTICAUDA 3 661 1077.2 STAURODERUS SCALARIS 13 661 1077.2 TETTIGONIA CANTANS 3 661 1077.2 GOMPHOCERIPPUS RUFUS 13 653 1079.9 CHORTHIPPUS APRICARIUS 3 653 1079.9 CHORTHIPPUS PARALLELUS 35 653 1079.9 EUTHYSTIRA BRACHYPTERA 13 653 1079.9 METRIOPTERA ROESELII 35 653 1079.9 METRIOPTERA SAUSSURIANA 13 653 1079.9 MIRAMELLA ALPINA 3 653 1079.9 OMOCESTUS VIRIDULUS 35 653 1079.9 PHOLIDOPTERA GRISEOAPTERA 3 653 1079.9 STAURODERUS SCALARIS 3 653 1079.9 TETTIGONIA VIRIDISSIMA 3 653 1079.9 STENOBOTHRUS LINEATUS 3 689 1087.2 CHORTHIPPUS PARALLELUS 35 689 1087.2 CHRYSOCHRAON DISPAR 3 689 1087.2 DECTICUS VERRUCIVORUS 3 689 1087.2 EUTHYSTIRA BRACHYPTERA 13 689 1087.2 METRIOPTERA ROESELII 3 689 1087.2 PLATYCLEIS ALBOPUNCTATAALB 3 689 1087.2 STAURODERUS SCALARIS 3 689 1087.2 STENOBOTHRUS LINEATUS 13 689 1087.2 CHORTHIPPUS BIGUTTULUS 35 689 1087.2 GOMPHOCERIPPUS RUFUS 13 670 1101.4 CHORTHIPPUS BIGUTTULUS 13 670 1101.4 CHORTHIPPUS PARALLELUS 13 670 1101.4 CHRYSOCHRAON DISPAR 3 670 1101.4 DECTICUS VERRUCIVORUS 13 670 1101.4 GRYLLUS CAMPESTRIS 3 670 1101.4 METRIOPTERA ROESELII 13 670 1101.4 PLATYCLEIS ALBOPUNCTATAALB 3 670 1101.4 STAURODERUS SCALARIS 3 670 1101.4 STENOBOTHRUS LINEATUS 3

- 90 -

670 1101.4 GOMPHOCERIPPUS RUFUS 13 673 1102.5 CHORTHIPPUS BIGUTTULUS 75 673 1102.5 CHORTHIPPUS PARALLELUS 75 673 1102.5 CHRYSOCHRAON DISPAR 13 673 1102.5 DECTICUS VERRUCIVORUS 13 673 1102.5 EUTHYSTIRA BRACHYPTERA 13 673 1102.5 GRYLLUS CAMPESTRIS 13 673 1102.5 METRIOPTERA ROESELII 13 673 1102.5 STAURODERUS SCALARIS 35 673 1102.5 STENOBOTHRUS LINEATUS 75 673 1102.5 TETTIGONIA CANTANS 3 676 1105.7 EUTHYSTIRA BRACHYPTERA 3 676 1105.7 METRIOPTERA ROESELII 13 676 1105.7 METRIOPTERA SAUSSURIANA 13 676 1105.7 MIRAMELLA ALPINA 3 676 1105.7 OMOCESTUS VIRIDULUS 3 676 1105.7 PHOLIDOPTERA GRISEOAPTERA 3 674 1119.8 CHORTHIPPUS BIGUTTULUS 35 674 1119.8 CHORTHIPPUS PARALLELUS 35 674 1119.8 DECTICUS VERRUCIVORUS 13 674 1119.8 METRIOPTERA ROESELII 13 674 1119.8 OMOCESTUS VIRIDULUS 13 674 1119.8 STENOBOTHRUS LINEATUS 13 890 1125 CHORTHIPPUS BIGUTTULUS 35 890 1125 CHORTHIPPUS PARALLELUS 13 890 1125 DECTICUS VERRUCIVORUS 3 890 1125 EUTHYSTIRA BRACHYPTERA 75 890 1125 METRIOPTERA ROESELII 3 890 1125 PHOLIDOPTERA GRISEOAPTERA 13 890 1125 STAURODERUS SCALARIS 3 890 1125 CHORTHIPPUS BRUNNEUS 35 890 1125 GOMPHOCERIPPUS RUFUS 3 890 1125 NEMOBIUS SYLVESTRIS 3 662 1133.9 CHORTHIPPUS DORSATUS 3 662 1133.9 CHORTHIPPUS PARALLELUS 35 662 1133.9 CHRYSOCHRAON DISPAR 3 662 1133.9 METRIOPTERA ROESELII 3 662 1133.9 METRIOPTERA SAUSSURIANA 3 662 1133.9 MIRAMELLA ALPINA 3 662 1133.9 OMOCESTUS VIRIDULUS 3 662 1133.9 STAURODERUS SCALARIS 3 662 1133.9 TETTIGONIA CANTANS 13 662 1133.9 CHORTHIPPUS BIGUTTULUS 13 659 1134.8 CHORTHIPPUS DORSATUS 75 659 1134.8 CHORTHIPPUS PARALLELUS 75 659 1134.8 METRIOPTERA ROESELII 35 659 1134.8 MIRAMELLA ALPINA 3 659 1134.8 PHOLIDOPTERA GRISEOAPTERA 3

- 91 -

659 1134.8 STETHOPHYMA GROSSUM 13 659 1134.8 TETTIGONIA CANTANS 3 683 1146.9 CHORTHIPPUS BIGUTTULUS 13 683 1146.9 CHORTHIPPUS PARALLELUS 75 683 1146.9 DECTICUS VERRUCIVORUS 3 683 1146.9 EUTHYSTIRA BRACHYPTERA 3 683 1146.9 METRIOPTERA ROESELII 3 683 1146.9 NEMOBIUS SYLVESTRIS 3 683 1146.9 OMOCESTUS RUFIPES 35 683 1146.9 PHOLIDOPTERA GRISEOAPTERA 3 683 1146.9 PLATYCLEIS ALBOPUNCTATAALB 13 683 1146.9 STAURODERUS SCALARIS 3 683 1146.9 STENOBOTHRUS LINEATUS 13 683 1146.9 GOMPHOCERIPPUS RUFUS 13 683 1146.9 METRIOPTERA SAUSSURIANA 3 666 1172.7 ARCYPTERA FUSCA 3 666 1172.7 CHORTHIPPUS BRUNNEUS 3 666 1172.7 CHORTHIPPUS PARALLELUS 13 666 1172.7 DECTICUS VERRUCIVORUS 3 666 1172.7 EUTHYSTIRA BRACHYPTERA 75 666 1172.7 GRYLLUS CAMPESTRIS 13 666 1172.7 METRIOPTERA ROESELII 13 666 1172.7 STAURODERUS SCALARIS 13 666 1172.7 STENOBOTHRUS LINEATUS 13 666 1172.7 CHORTHIPPUS BIGUTTULUS 35 666 1172.7 MECOSTETHUS PARAPLEURUS 13 666 1172.7 TETTIGONIA CANTANS 3 678 1176.6 CHORTHIPPUS PARALLELUS 75 678 1176.6 METRIOPTERA SAUSSURIANA 13 678 1176.6 MIRAMELLA ALPINA 3 678 1176.6 OMOCESTUS VIRIDULUS 35 678 1176.6 STAURODERUS SCALARIS 3 678 1176.6 METRIOPTERA ROESELII 3 655 1197 MIRAMELLA ALPINA 3 655 1197 OMOCESTUS VIRIDULUS 3 655 1197 POLYSARCUS DENTICAUDA 3 655 1197 STAURODERUS SCALARIS 3 655 1197 CHORTHIPPUS BIGUTTULUS 13 655 1197 METRIOPTERA ROESELII 3 655 1197 STENOBOTHRUS LINEATUS 3 677 1217.1 CHORTHIPPUS PARALLELUS 35 677 1217.1 METRIOPTERA ROESELII 13 677 1217.1 METRIOPTERA SAUSSURIANA 35 677 1217.1 MIRAMELLA ALPINA 13 677 1217.1 OMOCESTUS VIRIDULUS 3 677 1217.1 STAURODERUS SCALARIS 13 677 1217.1 TETTIGONIA CANTANS 13 677 1217.1 CHORTHIPPUS DORSATUS 13

- 92 -

677 1217.1 CHRYSOCHRAON DISPAR 3 656 1223.3 CHORTHIPPUS BIGUTTULUS 3 656 1223.3 CHORTHIPPUS MONTANUS 3 656 1223.3 CHORTHIPPUS PARALLELUS 75 656 1223.3 CHRYSOCHRAON DISPAR 13 656 1223.3 DECTICUS VERRUCIVORUS 3 656 1223.3 EUTHYSTIRA BRACHYPTERA 35 656 1223.3 METRIOPTERA ROESELII 13 656 1223.3 MIRAMELLA ALPINA 3 656 1223.3 OMOCESTUS VIRIDULUS 13 656 1223.3 PHOLIDOPTERA GRISEOAPTERA 3 656 1223.3 STAURODERUS SCALARIS 3 656 1223.3 TETTIGONIA VIRIDISSIMA 3 656 1223.3 CHORTHIPPUS DORSATUS 75 656 1223.3 TETTIGONIA CANTANS 3 679 1241.5 CHORTHIPPUS PARALLELUS 35 679 1241.5 EUTHYSTIRA BRACHYPTERA 3 679 1241.5 GOMPHOCERIPPUS RUFUS 75 679 1241.5 METRIOPTERA SAUSSURIANA 3 679 1241.5 PHOLIDOPTERA GRISEOAPTERA 3 679 1241.5 STAURODERUS SCALARIS 3 679 1241.5 GRYLLUS CAMPESTRIS 3 679 1241.5 STENOBOTHRUS LINEATUS 13 889 1258.5 MIRAMELLA ALPINA 3 889 1258.5 PHOLIDOPTERA GRISEOAPTERA 3 889 1258.5 METRIOPTERA SAUSSURIANA 3 686 1260.6 CHORTHIPPUS BIGUTTULUS 13 686 1260.6 CHORTHIPPUS PARALLELUS 75 686 1260.6 DECTICUS VERRUCIVORUS 3 686 1260.6 EUTHYSTIRA BRACHYPTERA 13 686 1260.6 GOMPHOCERIPPUS RUFUS 35 686 1260.6 METRIOPTERA ROESELII 13 686 1260.6 PLATYCLEIS ALBOPUNCTATAALB 13 686 1260.6 STAURODERUS SCALARIS 13 686 1260.6 STENOBOTHRUS LINEATUS 13 686 1260.6 CHORTHIPPUS DORSATUS 3 686 1260.6 TETTIGONIA CANTANS 3 663 1263.8 CHORTHIPPUS MONTANUS 3 663 1263.8 CHORTHIPPUS PARALLELUS 35 663 1263.8 CHRYSOCHRAON DISPAR 35 663 1263.8 DECTICUS VERRUCIVORUS 13 663 1263.8 EUTHYSTIRA BRACHYPTERA 13 663 1263.8 GRYLLUS CAMPESTRIS 3 663 1263.8 METRIOPTERA ROESELII 13 663 1263.8 OMOCESTUS VIRIDULUS 3 663 1263.8 STAURODERUS SCALARIS 35 663 1263.8 STENOBOTHRUS LINEATUS 35 663 1263.8 TETTIGONIA CANTANS 3

- 93 -

663 1263.8 CHORTHIPPUS BIGUTTULUS 35 724 1278.1 EUTHYSTIRA BRACHYPTERA 13 724 1278.1 METRIOPTERA SAUSSURIANA 3 724 1278.1 MIRAMELLA ALPINA 13 724 1278.1 OMOCESTUS VIRIDULUS 13 724 1278.1 PHOLIDOPTERA GRISEOAPTERA 13 724 1278.1 STAURODERUS SCALARIS 13 724 1278.1 TETRIX BIPUNCTATA BIPUNCTA 3 724 1278.1 CHORTHIPPUS DORSATUS 13 724 1278.1 METRIOPTERA ROESELII 3 715 1309.8 CHORTHIPPUS PARALLELUS 35 715 1309.8 EUTHYSTIRA BRACHYPTERA 13 715 1309.8 METRIOPTERA ROESELII 3 715 1309.8 METRIOPTERA SAUSSURIANA 3 715 1309.8 MIRAMELLA ALPINA 13 715 1309.8 OMOCESTUS VIRIDULUS 13 715 1309.8 STAURODERUS SCALARIS 3 715 1309.8 TETTIGONIA VIRIDISSIMA 3 715 1309.8 CHORTHIPPUS APRICARIUS 13 715 1309.8 CHORTHIPPUS BIGUTTULUS 3 715 1309.8 GOMPHOCERIPPUS RUFUS 3 715 1309.8 STENOBOTHRUS LINEATUS 35 715 1309.8 TETTIGONIA CANTANS 3 698 1353.1 CHORTHIPPUS PARALLELUS 35 698 1353.1 EUTHYSTIRA BRACHYPTERA 3 698 1353.1 METRIOPTERA ROESELII 3 698 1353.1 METRIOPTERA SAUSSURIANA 13 698 1353.1 MIRAMELLA ALPINA 13 698 1353.1 OMOCESTUS VIRIDULUS 35 698 1353.1 TETTIGONIA CANTANS 3 698 1353.1 CHORTHIPPUS BIGUTTULUS 3 693 1380.1 CHORTHIPPUS PARALLELUS 3 693 1380.1 EUTHYSTIRA BRACHYPTERA 75 693 1380.1 METRIOPTERA ROESELII 3 693 1380.1 MIRAMELLA ALPINA 3 693 1380.1 OMOCESTUS VIRIDULUS 13 693 1380.1 STAURODERUS SCALARIS 3 693 1380.1 CHORTHIPPUS DORSATUS 35 693 1380.1 CHORTHIPPUS MONTANUS 75 693 1380.1 DECTICUS VERRUCIVORUS 3 693 1380.1 METRIOPTERA SAUSSURIANA 13 697 1385.3 CHORTHIPPUS BRUNNEUS 3 697 1385.3 CHORTHIPPUS PARALLELUS 35 697 1385.3 EUTHYSTIRA BRACHYPTERA 3 697 1385.3 METRIOPTERA SAUSSURIANA 35 697 1385.3 MIRAMELLA ALPINA 13 697 1385.3 STAURODERUS SCALARIS 3 697 1385.3 METRIOPTERA ROESELII 3

- 94 -

697 1385.3 OMOCESTUS VIRIDULUS 3 697 1385.3 TETTIGONIA CANTANS 3 694 1389.8 CHORTHIPPUS PARALLELUS 13 694 1389.8 METRIOPTERA SAUSSURIANA 13 694 1389.8 MIRAMELLA ALPINA 35 694 1389.8 OMOCESTUS VIRIDULUS 13 694 1389.8 POLYSARCUS DENTICAUDA 3 694 1389.8 CHORTHIPPUS MONTANUS 13 694 1389.8 METRIOPTERA ROESELII 13 694 1389.8 STETHOPHYMA GROSSUM 35 694 1389.8 CHORTHIPPUS PARALLELUS 13 726 1390.6 CHORTHIPPUS BRUNNEUS 3 726 1390.6 CHORTHIPPUS PARALLELUS 13 726 1390.6 METRIOPTERA SAUSSURIANA 13 726 1390.6 MIRAMELLA ALPINA 3 726 1390.6 STAURODERUS SCALARIS 3 726 1390.6 TETRIX BIPUNCTATA BIPUNCTA 13 726 1390.6 CHORTHIPPUS BIGUTTULUS 3 726 1390.6 TETTIGONIA CANTANS 3 708 1394.3 CHORTHIPPUS PARALLELUS 75 708 1394.3 DECTICUS VERRUCIVORUS 3 708 1394.3 EUTHYSTIRA BRACHYPTERA 3 708 1394.3 METRIOPTERA ROESELII 75 708 1394.3 MIRAMELLA ALPINA 13 708 1394.3 OMOCESTUS VIRIDULUS 75 708 1394.3 STAURODERUS SCALARIS 13 708 1394.3 STENOBOTHRUS LINEATUS 35 699 1396.6 CHORTHIPPUS PARALLELUS 75 699 1396.6 METRIOPTERA SAUSSURIANA 3 699 1396.6 MIRAMELLA ALPINA 3 699 1396.6 OMOCESTUS VIRIDULUS 35 699 1396.6 STAURODERUS SCALARIS 13 699 1396.6 METRIOPTERA ROESELII 3 699 1396.6 TETTIGONIA CANTANS 3 720 1396.6 EUTHYSTIRA BRACHYPTERA 3 720 1396.6 METRIOPTERA ROESELII 3 720 1396.6 METRIOPTERA SAUSSURIANA 3 720 1396.6 MIRAMELLA ALPINA 13 720 1396.6 STAURODERUS SCALARIS 3 720 1396.6 CHORTHIPPUS PARALLELUS 13 720 1396.6 STENOBOTHRUS LINEATUS 3 705 1403.9 CHORTHIPPUS BIGUTTULUS 75 705 1403.9 CHORTHIPPUS BRUNNEUS 3 705 1403.9 CHORTHIPPUS PARALLELUS 75 705 1403.9 DECTICUS VERRUCIVORUS 13 705 1403.9 EUTHYSTIRA BRACHYPTERA 75 705 1403.9 METRIOPTERA ROESELII 3 705 1403.9 STAURODERUS SCALARIS 3

- 95 -

705 1403.9 STENOBOTHRUS LINEATUS 75 705 1403.9 TETRIX TENUICORNIS 3 705 1403.9 CHORTHIPPUS DORSATUS 3 705 1403.9 MECOSTETHUS PARAPLEURUS 13 692 1415.9 CHORTHIPPUS PARALLELUS 35 692 1415.9 OMOCESTUS VIRIDULUS 13 692 1415.9 MIRAMELLA ALPINA 13 717 1417.7 CHORTHIPPUS PARALLELUS 75 717 1417.7 EUTHYSTIRA BRACHYPTERA 35 717 1417.7 OMOCESTUS VIRIDULUS 35 717 1417.7 STENOBOTHRUS LINEATUS 75 717 1417.7 CHORTHIPPUS BIGUTTULUS 75 717 1417.7 METRIOPTERA ROESELII 3 717 1417.7 METRIOPTERA SAUSSURIANA 3 723 1422.8 EUTHYSTIRA BRACHYPTERA 13 723 1422.8 GOMPHOCERIPPUS RUFUS 75 723 1422.8 PHOLIDOPTERA GRISEOAPTERA 3 723 1422.8 PODISMA PEDESTRIS 3 723 1422.8 STAURODERUS SCALARIS 3 723 1422.8 CHORTHIPPUS PARALLELUS 35 723 1422.8 GRYLLUS CAMPESTRIS 3 723 1422.8 METRIOPTERA SAUSSURIANA 13 723 1422.8 PSOPHUS STRIDULUS 13 723 1422.8 STENOBOTHRUS LINEATUS 3 690 1426.2 CHORTHIPPUS PARALLELUS 35 690 1426.2 METRIOPTERA ROESELII 3 690 1426.2 METRIOPTERA SAUSSURIANA 13 690 1426.2 MIRAMELLA ALPINA 3 690 1426.2 OMOCESTUS VIRIDULUS 35 691 1442.4 CHORTHIPPUS PARALLELUS 13 691 1442.4 EUTHYSTIRA BRACHYPTERA 3 691 1442.4 METRIOPTERA ROESELII 3 691 1442.4 MIRAMELLA ALPINA 3 691 1442.4 OMOCESTUS VIRIDULUS 13 691 1442.4 POLYSARCUS DENTICAUDA 3 691 1442.4 STAURODERUS SCALARIS 3 691 1442.4 CHORTHIPPUS BIGUTTULUS 3 691 1442.4 CHORTHIPPUS MONTANUS 13 691 1442.4 METRIOPTERA SAUSSURIANA 3 710 1442.4 CHORTHIPPUS PARALLELUS 13 710 1442.4 EUTHYSTIRA BRACHYPTERA 13 710 1442.4 METRIOPTERA SAUSSURIANA 3 710 1442.4 MIRAMELLA ALPINA 3 710 1442.4 OMOCESTUS VIRIDULUS 13 710 1442.4 POLYSARCUS DENTICAUDA 3 710 1442.4 STAURODERUS SCALARIS 35 710 1442.4 CHORTHIPPUS BIGUTTULUS 35 710 1442.4 CHORTHIPPUS DORSATUS 75

- 96 -

710 1442.4 METRIOPTERA ROESELII 3 710 1442.4 TETTIGONIA CANTANS 13 696 1443.7 CHORTHIPPUS BIGUTTULUS 3 696 1443.7 CHORTHIPPUS PARALLELUS 75 696 1443.7 DECTICUS VERRUCIVORUS 3 696 1443.7 EUTHYSTIRA BRACHYPTERA 3 696 1443.7 METRIOPTERA ROESELII 13 696 1443.7 MIRAMELLA ALPINA 35 696 1443.7 OMOCESTUS VIRIDULUS 75 696 1443.7 STENOBOTHRUS LINEATUS 13 696 1443.7 CHORTHIPPUS DORSATUS 3 696 1443.7 CHORTHIPPUS MONTANUS 75 696 1443.7 METRIOPTERA SAUSSURIANA 13 696 1443.7 STETHOPHYMA GROSSUM 13 718 1455.9 CHORTHIPPUS PARALLELUS 75 718 1455.9 METRIOPTERA SAUSSURIANA 13 718 1455.9 MIRAMELLA ALPINA 35 718 1455.9 OMOCESTUS VIRIDULUS 75 718 1455.9 POLYSARCUS DENTICAUDA 3 718 1455.9 CHORTHIPPUS BRUNNEUS 13 718 1455.9 STENOBOTHRUS LINEATUS 35 718 1455.9 STETHOPHYMA GROSSUM 3 700 1464.3 ARCYPTERA FUSCA 3 700 1464.3 CHORTHIPPUS APRICARIUS 3 700 1464.3 CHORTHIPPUS BIGUTTULUS 75 700 1464.3 CHORTHIPPUS PARALLELUS 75 700 1464.3 PSOPHUS STRIDULUS 13 700 1464.3 STAURODERUS SCALARIS 35 700 1464.3 STENOBOTHRUS LINEATUS 35 700 1464.3 TETTIGONIA CANTANS 13 700 1464.3 EUTHYSTIRA BRACHYPTERA 3 700 1464.3 METRIOPTERA SAUSSURIANA 3 729 1479.4 CHORTHIPPUS BIGUTTULUS 75 729 1479.4 CHORTHIPPUS PARALLELUS 13 729 1479.4 EUTHYSTIRA BRACHYPTERA 35 729 1479.4 METRIOPTERA SAUSSURIANA 35 729 1479.4 OMOCESTUS VIRIDULUS 13 729 1479.4 STAURODERUS SCALARIS 13 728 1482.3 EUTHYSTIRA BRACHYPTERA 3 728 1482.3 TETRIX BIPUNCTATA BIPUNCTA 3 712 1494.8 CHORTHIPPUS PARALLELUS 13 712 1494.8 MIRAMELLA ALPINA 35 712 1494.8 OMOCESTUS VIRIDULUS 35 712 1494.8 STETHOPHYMA GROSSUM 3 722 1497.1 CHORTHIPPUS BIGUTTULUS 35 722 1497.1 CHORTHIPPUS BRUNNEUS 3 722 1497.1 PHOLIDOPTERA GRISEOAPTERA 3 722 1497.1 PSOPHUS STRIDULUS 3

- 97 -

722 1497.1 STAURODERUS SCALARIS 3 722 1497.1 TETTIGONIA CANTANS 3 722 1497.1 GOMPHOCERIPPUS RUFUS 3 722 1497.1 METRIOPTERA SAUSSURIANA 3 722 1497.1 OEDIPODA GERMANICA 13 714 1498.1 CHORTHIPPUS PARALLELUS 13 714 1498.1 METRIOPTERA SAUSSURIANA 13 714 1498.1 MIRAMELLA ALPINA 35 714 1498.1 OMOCESTUS VIRIDULUS 35 714 1498.1 STAURODERUS SCALARIS 75 714 1498.1 CHORTHIPPUS BRUNNEUS 3 703 1512.8 CHORTHIPPUS PARALLELUS 75 703 1512.8 MIRAMELLA ALPINA 3 703 1512.8 OMOCESTUS VIRIDULUS 35 703 1512.8 STAURODERUS SCALARIS 35 703 1512.8 STENOBOTHRUS LINEATUS 3 703 1512.8 CHORTHIPPUS BIGUTTULUS 13 703 1512.8 METRIOPTERA SAUSSURIANA 13 703 1512.8 TETTIGONIA CANTANS 3 732 1520.1 MIRAMELLA ALPINA 13 725 1526.9 CHORTHIPPUS BIGUTTULUS 3 725 1526.9 CHORTHIPPUS PARALLELUS 3 725 1526.9 EUTHYSTIRA BRACHYPTERA 75 725 1526.9 LEPTOPHYES PUNCTATISSIMA 3 725 1526.9 METRIOPTERA SAUSSURIANA 3 725 1526.9 MIRAMELLA ALPINA 13 725 1526.9 POLYSARCUS DENTICAUDA 13 725 1526.9 STAURODERUS SCALARIS 13 725 1526.9 TETTIGONIA CANTANS 3 727 1529.4 CHORTHIPPUS BIGUTTULUS 75 727 1529.4 EUTHYSTIRA BRACHYPTERA 3 727 1529.4 OMOCESTUS VIRIDULUS 3 727 1529.4 PHOLIDOPTERA GRISEOAPTERA 3 727 1529.4 STAURODERUS SCALARIS 13 727 1529.4 STENOBOTHRUS LINEATUS 3 727 1529.4 DECTICUS VERRUCIVORUS 3 727 1529.4 METRIOPTERA ROESELII 3 746 1534.7 EUTHYSTIRA BRACHYPTERA 35 746 1534.7 METRIOPTERA SAUSSURIANA 13 746 1534.7 OMOCESTUS VIRIDULUS 13 746 1534.7 STAURODERUS SCALARIS 3 746 1534.7 CHORTHIPPUS DORSATUS 75 746 1534.7 CHORTHIPPUS PARALLELUS 75 746 1534.7 METRIOPTERA ROESELII 13 746 1534.7 MIRAMELLA ALPINA 3 742 1543.4 CHORTHIPPUS PARALLELUS 35 742 1543.4 DECTICUS VERRUCIVORUS 13 742 1543.4 MIRAMELLA ALPINA 13

- 98 -

742 1543.4 OMOCESTUS VIRIDULUS 13 742 1543.4 STAURODERUS SCALARIS 75 750 1567.2 CHORTHIPPUS PARALLELUS 13 750 1567.2 METRIOPTERA ROESELII 13 750 1567.2 METRIOPTERA SAUSSURIANA 13 750 1567.2 MIRAMELLA ALPINA 35 750 1567.2 OMOCESTUS VIRIDULUS 75 750 1567.2 CHORTHIPPUS DORSATUS 3 750 1567.2 STAURODERUS SCALARIS 3 750 1567.2 STENOBOTHRUS LINEATUS 3 757 1580 CHORTHIPPUS APRICARIUS 35 757 1580 CHORTHIPPUS BIGUTTULUS 13 757 1580 CHORTHIPPUS PARALLELUS 35 757 1580 EUTHYSTIRA BRACHYPTERA 75 757 1580 METRIOPTERA ROESELII 3 757 1580 METRIOPTERA SAUSSURIANA 3 757 1580 PHOLIDOPTERA GRISEOAPTERA 13 757 1580 STAURODERUS SCALARIS 13 757 1580 TETTIGONIA CANTANS 3 754 1580.9 EUTHYSTIRA BRACHYPTERA 3 754 1580.9 METRIOPTERA SAUSSURIANA 3 754 1580.9 OMOCESTUS VIRIDULUS 3 759 1589.2 BARBITISTES SERRICAUDA 3 759 1589.2 CHORTHIPPUS BIGUTTULUS 35 759 1589.2 EUTHYSTIRA BRACHYPTERA 75 759 1589.2 METRIOPTERA SAUSSURIANA 13 759 1589.2 MIRAMELLA ALPINA 35 759 1589.2 PODISMA PEDESTRIS 3 759 1589.2 PSOPHUS STRIDULUS 3 759 1589.2 STAURODERUS SCALARIS 75 759 1589.2 TETRIX BIPUNCTATA BIPUNCTA 3 752 1590.1 CHORTHIPPUS PARALLELUS 13 752 1590.1 EUTHYSTIRA BRACHYPTERA 35 752 1590.1 METRIOPTERA SAUSSURIANA 35 752 1590.1 PODISMA PEDESTRIS 3 752 1590.1 POLYSARCUS DENTICAUDA 3 752 1590.1 PSOPHUS STRIDULUS 13 752 1590.1 STAURODERUS SCALARIS 3 752 1590.1 TETTIGONIA CANTANS 3 752 1590.1 CHORTHIPPUS BIGUTTULUS 3 752 1590.1 MIRAMELLA ALPINA 3 740 1592.7 CHORTHIPPUS PARALLELUS 35 740 1592.7 METRIOPTERA SAUSSURIANA 3 740 1592.7 MIRAMELLA ALPINA 13 740 1592.7 OMOCESTUS VIRIDULUS 35 740 1592.7 STAURODERUS SCALARIS 13 740 1592.7 TETTIGONIA CANTANS 3 749 1604.9 CHORTHIPPUS PARALLELUS 35

- 99 -

749 1604.9 METRIOPTERA ROESELII 3 749 1604.9 METRIOPTERA SAUSSURIANA 13 749 1604.9 MIRAMELLA ALPINA 13 749 1604.9 OMOCESTUS VIRIDULUS 75 749 1604.9 STAURODERUS SCALARIS 35 749 1604.9 CHORTHIPPUS BRUNNEUS 3 749 1604.9 GOMPHOCERIPPUS RUFUS 3 736 1629.1 NA 0 735 1634.8 EUTHYSTIRA BRACHYPTERA 35 735 1634.8 METRIOPTERA SAUSSURIANA 35 735 1634.8 MIRAMELLA ALPINA 75 735 1634.8 POLYSARCUS DENTICAUDA 13 735 1634.8 STAURODERUS SCALARIS 3 735 1634.8 CHORTHIPPUS DORSATUS 3 735 1634.8 CHORTHIPPUS PARALLELUS 35 735 1634.8 OMOCESTUS VIRIDULUS 3 735 1634.8 STENOBOTHRUS LINEATUS 3 765 1640.6 CHORTHIPPUS PARALLELUS 75 765 1640.6 METRIOPTERA SAUSSURIANA 35 765 1640.6 OMOCESTUS VIRIDULUS 75 765 1640.6 STAURODERUS SCALARIS 13 738 1647.7 CHORTHIPPUS PARALLELUS 3 738 1647.7 METRIOPTERA ROESELII 3 738 1647.7 METRIOPTERA SAUSSURIANA 35 738 1647.7 MIRAMELLA ALPINA 13 738 1647.7 OMOCESTUS VIRIDULUS 35 738 1647.7 STAURODERUS SCALARIS 35 730 1649.8 CHORTHIPPUS PARALLELUS 13 730 1649.8 METRIOPTERA ROESELII 3 730 1649.8 OMOCESTUS VIRIDULUS 13 730 1649.8 STAURODERUS SCALARIS 3 745 1664.7 CHORTHIPPUS PARALLELUS 35 745 1664.7 DECTICUS VERRUCIVORUS 3 745 1664.7 GOMPHOCERUS SIBIRICUS 3 745 1664.7 METRIOPTERA SAUSSURIANA 13 745 1664.7 MIRAMELLA ALPINA 35 745 1664.7 OMOCESTUS VIRIDULUS 35 745 1664.7 STAURODERUS SCALARIS 13 745 1664.7 CHORTHIPPUS BIGUTTULUS 35 745 1664.7 CHORTHIPPUS DORSATUS 3 768 1674.7 MIRAMELLA ALPINA 3 737 1678.6 CHORTHIPPUS PARALLELUS 75 737 1678.6 CHRYSOCHRAON DISPAR 3 737 1678.6 EUTHYSTIRA BRACHYPTERA 13 737 1678.6 METRIOPTERA SAUSSURIANA 13 737 1678.6 MIRAMELLA ALPINA 75 737 1678.6 OMOCESTUS VIRIDULUS 35 737 1678.6 POLYSARCUS DENTICAUDA 3

- 100 -

743 1683.4 ARCYPTERA FUSCA 3 743 1683.4 CHORTHIPPUS PARALLELUS 75 743 1683.4 EUTHYSTIRA BRACHYPTERA 13 743 1683.4 METRIOPTERA ROESELII 3 743 1683.4 MIRAMELLA ALPINA 3 743 1683.4 OMOCESTUS VIRIDULUS 3 743 1683.4 POLYSARCUS DENTICAUDA 3 743 1683.4 STAURODERUS SCALARIS 35 743 1683.4 CHORTHIPPUS BIGUTTULUS 3 743 1683.4 CHORTHIPPUS DORSATUS 75 743 1683.4 STENOBOTHRUS LINEATUS 13 743 1683.4 TETTIGONIA CANTANS 3 758 1686.7 ARCYPTERA FUSCA 3 758 1686.7 CHORTHIPPUS BIGUTTULUS 3 758 1686.7 CHORTHIPPUS PARALLELUS 35 758 1686.7 EUTHYSTIRA BRACHYPTERA 75 758 1686.7 METRIOPTERA SAUSSURIANA 13 758 1686.7 MIRAMELLA ALPINA 35 758 1686.7 OMOCESTUS VIRIDULUS 3 758 1686.7 POLYSARCUS DENTICAUDA 13 758 1686.7 PSOPHUS STRIDULUS 3 758 1686.7 STAURODERUS SCALARIS 13 758 1686.7 TETTIGONIA CANTANS 3 753 1697.7 CHORTHIPPUS PARALLELUS 75 753 1697.7 METRIOPTERA SAUSSURIANA 3 753 1697.7 MIRAMELLA ALPINA 13 753 1697.7 OMOCESTUS VIRIDULUS 35 753 1697.7 EUTHYSTIRA BRACHYPTERA 13 731 1698.2 CHORTHIPPUS PARALLELUS 75 731 1698.2 METRIOPTERA SAUSSURIANA 35 731 1698.2 MIRAMELLA ALPINA 75 731 1698.2 OMOCESTUS VIRIDULUS 35 731 1698.2 STETHOPHYMA GROSSUM 75 748 1711.3 ARCYPTERA FUSCA 13 748 1711.3 CHORTHIPPUS BIGUTTULUS 75 748 1711.3 CHORTHIPPUS PARALLELUS 75 748 1711.3 DECTICUS VERRUCIVORUS 3 748 1711.3 EUTHYSTIRA BRACHYPTERA 75 748 1711.3 METRIOPTERA ROESELII 35 748 1711.3 METRIOPTERA SAUSSURIANA 3 748 1711.3 MIRAMELLA ALPINA 13 748 1711.3 OMOCESTUS VIRIDULUS 35 748 1711.3 STAURODERUS SCALARIS 35 748 1711.3 STENOBOTHRUS LINEATUS 13 744 1713.2 CHORTHIPPUS BIGUTTULUS 13 744 1713.2 CHORTHIPPUS PARALLELUS 75 744 1713.2 DECTICUS VERRUCIVORUS 13 744 1713.2 EUTHYSTIRA BRACHYPTERA 13

- 101 -

744 1713.2 METRIOPTERA ROESELII 13 744 1713.2 MIRAMELLA ALPINA 3 744 1713.2 OMOCESTUS VIRIDULUS 35 744 1713.2 STAURODERUS SCALARIS 13 744 1713.2 STENOBOTHRUS LINEATUS 13 744 1713.2 CHORTHIPPUS DORSATUS 13 744 1713.2 METRIOPTERA SAUSSURIANA 13 751 1760.1 CHORTHIPPUS PARALLELUS 35 751 1760.1 EUTHYSTIRA BRACHYPTERA 13 751 1760.1 METRIOPTERA SAUSSURIANA 13 751 1760.1 MIRAMELLA ALPINA 13 751 1760.1 OMOCESTUS VIRIDULUS 35 751 1760.1 POLYSARCUS DENTICAUDA 3 893 1764.7 METRIOPTERA SAUSSURIANA 3 893 1764.7 MIRAMELLA ALPINA 3 787 1767.8 CHORTHIPPUS PARALLELUS 75 787 1767.8 EUTHYSTIRA BRACHYPTERA 35 787 1767.8 METRIOPTERA SAUSSURIANA 13 787 1767.8 OMOCESTUS VIRIDULUS 75 787 1767.8 STAURODERUS SCALARIS 3 787 1767.8 TETRIX BIPUNCTATA BIPUNCTA 3 897 1770 CHORTHIPPUS PARALLELUS 75 897 1770 METRIOPTERA SAUSSURIANA 35 897 1770 MIRAMELLA ALPINA 13 897 1770 OMOCESTUS VIRIDULUS 3 767 1770.9 CHORTHIPPUS PARALLELUS 75 767 1770.9 EUTHYSTIRA BRACHYPTERA 13 767 1770.9 METRIOPTERA SAUSSURIANA 13 767 1770.9 MIRAMELLA ALPINA 3 767 1770.9 OMOCESTUS VIRIDULUS 35 767 1770.9 PODISMA PEDESTRIS 13 767 1770.9 STAURODERUS SCALARIS 3 778 1781.3 CHORTHIPPUS PARALLELUS 13 778 1781.3 EUTHYSTIRA BRACHYPTERA 13 778 1781.3 METRIOPTERA SAUSSURIANA 13 778 1781.3 MIRAMELLA ALPINA 13 778 1781.3 OMOCESTUS VIRIDULUS 3 778 1781.3 POLYSARCUS DENTICAUDA 3 776 1795 MIRAMELLA ALPINA 3 776 1795 OMOCESTUS VIRIDULUS 13 776 1795 CHORTHIPPUS PARALLELUS 13 776 1795 METRIOPTERA SAUSSURIANA 13 804 1812.1 CHORTHIPPUS BIGUTTULUS 35 804 1812.1 CHORTHIPPUS PARALLELUS 75 804 1812.1 EUTHYSTIRA BRACHYPTERA 75 804 1812.1 METRIOPTERA SAUSSURIANA 3 804 1812.1 MIRAMELLA ALPINA 3 804 1812.1 OMOCESTUS VIRIDULUS 75

- 102 -

804 1812.1 PHOLIDOPTERA GRISEOAPTERA 13 804 1812.1 PODISMA PEDESTRIS 3 804 1812.1 STAURODERUS SCALARIS 75 806 1814.3 CHORTHIPPUS PARALLELUS 35 806 1814.3 METRIOPTERA SAUSSURIANA 3 806 1814.3 MIRAMELLA ALPINA 35 806 1814.3 OMOCESTUS VIRIDULUS 13 792 1818.6 CHORTHIPPUS DORSATUS 3 792 1818.6 CHORTHIPPUS PARALLELUS 35 792 1818.6 EUTHYSTIRA BRACHYPTERA 13 792 1818.6 METRIOPTERA ROESELII 3 792 1818.6 METRIOPTERA SAUSSURIANA 3 792 1818.6 MIRAMELLA ALPINA 35 792 1818.6 OMOCESTUS VIRIDULUS 35 792 1818.6 POLYSARCUS DENTICAUDA 3 795 1824.3 METRIOPTERA SAUSSURIANA 3 795 1824.3 MIRAMELLA ALPINA 3 795 1824.3 OMOCESTUS VIRIDULUS 13 795 1824.3 POLYSARCUS DENTICAUDA 3 784 1830.2 CHORTHIPPUS APRICARIUS 3 784 1830.2 CHORTHIPPUS BIGUTTULUS 3 784 1830.2 CHORTHIPPUS PARALLELUS 3 784 1830.2 METRIOPTERA SAUSSURIANA 3 784 1830.2 OMOCESTUS VIRIDULUS 13 784 1830.2 POLYSARCUS DENTICAUDA 3 784 1830.2 MIRAMELLA ALPINA 3 779 1831.5 CHORTHIPPUS PARALLELUS 13 779 1831.5 EUTHYSTIRA BRACHYPTERA 35 779 1831.5 GOMPHOCERUS SIBIRICUS 75 779 1831.5 METRIOPTERA SAUSSURIANA 3 779 1831.5 MIRAMELLA ALPINA 3 779 1831.5 OMOCESTUS VIRIDULUS 75 779 1831.5 STAURODERUS SCALARIS 13 779 1831.5 TETTIGONIA CANTANS 3 772 1834.2 CHORTHIPPUS PARALLELUS 13 772 1834.2 METRIOPTERA SAUSSURIANA 35 772 1834.2 MIRAMELLA ALPINA 35 772 1834.2 OMOCESTUS VIRIDULUS 13 794 1843.8 CHORTHIPPUS PARALLELUS 75 794 1843.8 GOMPHOCERUS SIBIRICUS 35 794 1843.8 METRIOPTERA SAUSSURIANA 35 794 1843.8 MIRAMELLA ALPINA 75 794 1843.8 OMOCESTUS VIRIDULUS 35 794 1843.8 POLYSARCUS DENTICAUDA 3 794 1843.8 STAURODERUS SCALARIS 3 794 1843.8 CHORTHIPPUS BIGUTTULUS 35 794 1843.8 METRIOPTERA ROESELII 3 809 1843.9 ARCYPTERA FUSCA 13

- 103 -

809 1843.9 CHORTHIPPUS BIGUTTULUS 35 809 1843.9 CHORTHIPPUS PARALLELUS 35 809 1843.9 EUTHYSTIRA BRACHYPTERA 75 809 1843.9 GOMPHOCERUS SIBIRICUS 13 809 1843.9 METRIOPTERA SAUSSURIANA 3 809 1843.9 PSOPHUS STRIDULUS 13 809 1843.9 STAURODERUS SCALARIS 13 801 1846.6 CHORTHIPPUS APRICARIUS 35 801 1846.6 CHORTHIPPUS BIGUTTULUS 35 801 1846.6 CHORTHIPPUS PARALLELUS 35 801 1846.6 EUTHYSTIRA BRACHYPTERA 75 801 1846.6 METRIOPTERA SAUSSURIANA 13 801 1846.6 MIRAMELLA ALPINA 3 801 1846.6 OMOCESTUS VIRIDULUS 35 801 1846.6 PSOPHUS STRIDULUS 13 801 1846.6 STAURODERUS SCALARIS 35 801 1846.6 STENOBOTHRUS LINEATUS 35 807 1872 CHORTHIPPUS PARALLELUS 35 807 1872 EUTHYSTIRA BRACHYPTERA 3 807 1872 GOMPHOCERUS SIBIRICUS 35 807 1872 METRIOPTERA SAUSSURIANA 13 807 1872 MIRAMELLA ALPINA 13 783 1873.1 NA 0 895 1877.2 MIRAMELLA ALPINA 3 780 1884.9 CHORTHIPPUS BIGUTTULUS 13 780 1884.9 CHORTHIPPUS PARALLELUS 13 780 1884.9 EUTHYSTIRA BRACHYPTERA 35 780 1884.9 METRIOPTERA SAUSSURIANA 3 780 1884.9 PSOPHUS STRIDULUS 13 780 1884.9 STAURODERUS SCALARIS 13 800 1887.2 CHORTHIPPUS PARALLELUS 13 800 1887.2 EUTHYSTIRA BRACHYPTERA 35 800 1887.2 METRIOPTERA SAUSSURIANA 13 800 1887.2 MIRAMELLA ALPINA 13 800 1887.2 OMOCESTUS VIRIDULUS 3 800 1887.2 POLYSARCUS DENTICAUDA 3 800 1887.2 STAURODERUS SCALARIS 13 782 1888.5 CHORTHIPPUS PARALLELUS 35 782 1888.5 GOMPHOCERUS SIBIRICUS 3 782 1888.5 METRIOPTERA SAUSSURIANA 13 782 1888.5 MIRAMELLA ALPINA 3 782 1888.5 OMOCESTUS VIRIDULUS 13 782 1888.5 PODISMA PEDESTRIS 13 894 1892.7 OMOCESTUS VIRIDULUS 3 769 1905.2 CHORTHIPPUS PARALLELUS 75 769 1905.2 METRIOPTERA SAUSSURIANA 13 769 1905.2 MIRAMELLA ALPINA 35 769 1905.2 OMOCESTUS VIRIDULUS 13

- 104 -

769 1905.2 STAURODERUS SCALARIS 3 789 1919.7 METRIOPTERA SAUSSURIANA 3 789 1919.7 MIRAMELLA ALPINA 13 805 1928.1 ARCYPTERA FUSCA 13 805 1928.1 CHORTHIPPUS BIGUTTULUS 35 805 1928.1 CHORTHIPPUS PARALLELUS 75 805 1928.1 DECTICUS VERRUCIVORUS 13 805 1928.1 EUTHYSTIRA BRACHYPTERA 75 805 1928.1 METRIOPTERA SAUSSURIANA 13 805 1928.1 MIRAMELLA ALPINA 3 805 1928.1 OMOCESTUS VIRIDULUS 3 805 1928.1 POLYSARCUS DENTICAUDA 3 805 1928.1 PSOPHUS STRIDULUS 13 805 1928.1 STAURODERUS SCALARIS 75 805 1928.1 STENOBOTHRUS LINEATUS 13 805 1928.1 TETRIX BIPUNCTATA BIPUNCTA 3 790 1936.7 ANONCONOTUS ALPINUS 3 790 1936.7 CHORTHIPPUS PARALLELUS 75 790 1936.7 DECTICUS VERRUCIVORUS 35 790 1936.7 EUTHYSTIRA BRACHYPTERA 35 790 1936.7 GOMPHOCERUS SIBIRICUS 13 790 1936.7 METRIOPTERA SAUSSURIANA 13 790 1936.7 MIRAMELLA ALPINA 13 790 1936.7 OMOCESTUS VIRIDULUS 35 790 1936.7 PODISMA PEDESTRIS 13 790 1936.7 STAURODERUS SCALARIS 35 775 1937.3 OMOCESTUS VIRIDULUS 3 803 1938.1 CHORTHIPPUS BIGUTTULUS 3 803 1938.1 CHORTHIPPUS PARALLELUS 35 803 1938.1 EUTHYSTIRA BRACHYPTERA 35 803 1938.1 GOMPHOCERUS SIBIRICUS 13 803 1938.1 METRIOPTERA ROESELII 13 803 1938.1 METRIOPTERA SAUSSURIANA 13 803 1938.1 MIRAMELLA ALPINA 35 803 1938.1 OMOCESTUS VIRIDULUS 35 803 1938.1 POLYSARCUS DENTICAUDA 3 777 1941.5 CHORTHIPPUS PARALLELUS 35 777 1941.5 METRIOPTERA SAUSSURIANA 13 777 1941.5 MIRAMELLA ALPINA 35 777 1941.5 OMOCESTUS VIRIDULUS 35 896 1945.4 CHORTHIPPUS PARALLELUS 13 896 1945.4 METRIOPTERA SAUSSURIANA 3 896 1945.4 MIRAMELLA ALPINA 3 896 1945.4 PODISMA PEDESTRIS 3 791 1953.4 CHORTHIPPUS PARALLELUS 75 791 1953.4 EUTHYSTIRA BRACHYPTERA 3 791 1953.4 GOMPHOCERUS SIBIRICUS 35 791 1953.4 METRIOPTERA SAUSSURIANA 13

- 105 -

791 1953.4 MIRAMELLA ALPINA 35 791 1953.4 OMOCESTUS VIRIDULUS 75 791 1953.4 POLYSARCUS DENTICAUDA 3 791 1953.4 STAURODERUS SCALARIS 3 770 1954.1 CHORTHIPPUS PARALLELUS 75 770 1954.1 METRIOPTERA SAUSSURIANA 13 770 1954.1 MIRAMELLA ALPINA 3 770 1954.1 OMOCESTUS VIRIDULUS 13 770 1954.1 POLYSARCUS DENTICAUDA 35 843 1972.6 CHORTHIPPUS APRICARIUS 75 843 1972.6 CHORTHIPPUS BIGUTTULUS 13 843 1972.6 CHORTHIPPUS PARALLELUS 3 843 1972.6 EUTHYSTIRA BRACHYPTERA 35 843 1972.6 METRIOPTERA SAUSSURIANA 13 843 1972.6 MIRAMELLA ALPINA 13 843 1972.6 POLYSARCUS DENTICAUDA 3 846 1983.5 METRIOPTERA SAUSSURIANA 3 846 1983.5 MIRAMELLA ALPINA 13 846 1983.5 OMOCESTUS VIRIDULUS 3 797 1988.8 CHORTHIPPUS PARALLELUS 35 797 1988.8 METRIOPTERA SAUSSURIANA 35 797 1988.8 MIRAMELLA ALPINA 13 797 1988.8 PODISMA PEDESTRIS 3 821 2005.1 CHORTHIPPUS PARALLELUS 35 821 2005.1 METRIOPTERA SAUSSURIANA 3 821 2005.1 MIRAMELLA ALPINA 3 821 2005.1 OMOCESTUS VIRIDULUS 35 828 2011.9 CHORTHIPPUS BIGUTTULUS 13 828 2011.9 CHORTHIPPUS PARALLELUS 75 828 2011.9 METRIOPTERA SAUSSURIANA 13 828 2011.9 MIRAMELLA ALPINA 13 828 2011.9 OMOCESTUS VIRIDULUS 35 828 2011.9 PODISMA PEDESTRIS 3 828 2011.9 STAURODERUS SCALARIS 35 837 2015.8 METRIOPTERA SAUSSURIANA 3 837 2015.8 MIRAMELLA ALPINA 13 837 2015.8 OMOCESTUS VIRIDULUS 3 837 2015.8 POLYSARCUS DENTICAUDA 3 820 2015.9 CHORTHIPPUS PARALLELUS 35 820 2015.9 METRIOPTERA SAUSSURIANA 13 820 2015.9 MIRAMELLA ALPINA 35 820 2015.9 PODISMA PEDESTRIS 3 814 2016.2 EUTHYSTIRA BRACHYPTERA 3 814 2016.2 MIRAMELLA ALPINA 3 814 2016.2 OMOCESTUS VIRIDULUS 13 814 2016.2 PODISMA PEDESTRIS 13 813 2021 METRIOPTERA SAUSSURIANA 3 813 2021 MIRAMELLA ALPINA 13

- 106 -

830 2024.5 CHORTHIPPUS PARALLELUS 35 830 2024.5 METRIOPTERA SAUSSURIANA 13 830 2024.5 MIRAMELLA ALPINA 13 817 2035.3 METRIOPTERA ROESELII 3 817 2035.3 METRIOPTERA SAUSSURIANA 13 817 2035.3 MIRAMELLA ALPINA 13 817 2035.3 POLYSARCUS DENTICAUDA 3 902 2039.5 CHORTHIPPUS PARALLELUS 75 902 2039.5 METRIOPTERA SAUSSURIANA 35 902 2039.5 MIRAMELLA ALPINA 13 902 2039.5 PODISMA PEDESTRIS 3 838 2046.6 CHORTHIPPUS PARALLELUS 13 838 2046.6 EUTHYSTIRA BRACHYPTERA 35 838 2046.6 METRIOPTERA SAUSSURIANA 13 838 2046.6 MIRAMELLA ALPINA 13 838 2046.6 PODISMA PEDESTRIS 3 825 2053.7 CHORTHIPPUS PARALLELUS 75 825 2053.7 EUTHYSTIRA BRACHYPTERA 75 825 2053.7 METRIOPTERA SAUSSURIANA 3 825 2053.7 MIRAMELLA ALPINA 75 825 2053.7 OMOCESTUS VIRIDULUS 75 825 2053.7 POLYSARCUS DENTICAUDA 3 825 2053.7 STAURODERUS SCALARIS 13 822 2055.2 CHORTHIPPUS APRICARIUS 13 822 2055.2 CHORTHIPPUS BIGUTTULUS 13 822 2055.2 CHORTHIPPUS PARALLELUS 35 822 2055.2 METRIOPTERA SAUSSURIANA 13 822 2055.2 OMOCESTUS VIRIDULUS 3 822 2055.2 POLYSARCUS DENTICAUDA 3 822 2055.2 STAURODERUS SCALARIS 13 818 2063.6 NA 0 812 2067.3 CHORTHIPPUS BIGUTTULUS 13 812 2067.3 CHORTHIPPUS PARALLELUS 13 812 2067.3 OMOCESTUS VIRIDULUS 13 812 2067.3 PODISMA PEDESTRIS 13 827 2080.6 CHORTHIPPUS PARALLELUS 35 827 2080.6 METRIOPTERA SAUSSURIANA 13 827 2080.6 MIRAMELLA ALPINA 3 845 2090.1 CHORTHIPPUS PARALLELUS 35 845 2090.1 EUTHYSTIRA BRACHYPTERA 3 845 2090.1 METRIOPTERA SAUSSURIANA 13 845 2090.1 MIRAMELLA ALPINA 3 845 2090.1 STAURODERUS SCALARIS 13 900 2100.8 CHORTHIPPUS BIGUTTULUS 3 900 2100.8 CHORTHIPPUS PARALLELUS 3 900 2100.8 EUTHYSTIRA BRACHYPTERA 13 900 2100.8 GOMPHOCERUS SIBIRICUS 13 900 2100.8 MIRAMELLA ALPINA 3

- 107 -

900 2100.8 OMOCESTUS VIRIDULUS 3 901 2109.8 ANONCONOTUS ALPINUS 3 901 2109.8 CHORTHIPPUS PARALLELUS 13 901 2109.8 METRIOPTERA SAUSSURIANA 13 901 2109.8 MIRAMELLA ALPINA 13 901 2109.8 OMOCESTUS VIRIDULUS 13 901 2109.8 PODISMA PEDESTRIS 13 901 2109.8 STAURODERUS SCALARIS 3 833 2117.6 ANONCONOTUS ALPINUS 3 833 2117.6 CHORTHIPPUS PARALLELUS 13 833 2117.6 DECTICUS VERRUCIVORUS 13 833 2117.6 EUTHYSTIRA BRACHYPTERA 35 833 2117.6 GOMPHOCERUS SIBIRICUS 3 833 2117.6 METRIOPTERA SAUSSURIANA 35 833 2117.6 MIRAMELLA ALPINA 35 833 2117.6 PODISMA PEDESTRIS 13 833 2117.6 STAURODERUS SCALARIS 13 831 2130.6 CHORTHIPPUS APRICARIUS 3 831 2130.6 CHORTHIPPUS PARALLELUS 3 831 2130.6 EUTHYSTIRA BRACHYPTERA 3 831 2130.6 METRIOPTERA SAUSSURIANA 3 831 2130.6 MIRAMELLA ALPINA 13 831 2130.6 STAURODERUS SCALARIS 3 831 2130.6 CHORTHIPPUS BIGUTTULUS 3 831 2130.6 OMOCESTUS VIRIDULUS 3 841 2136.1 CHORTHIPPUS PARALLELUS 35 841 2136.1 MIRAMELLA ALPINA 13 841 2136.1 POLYSARCUS DENTICAUDA 13 811 2137.6 NA 0 898 2146.5 CHORTHIPPUS PARALLELUS 13 898 2146.5 EUTHYSTIRA BRACHYPTERA 13 898 2146.5 METRIOPTERA SAUSSURIANA 3 898 2146.5 MIRAMELLA ALPINA 13 898 2146.5 PODISMA PEDESTRIS 3 898 2146.5 POLYSARCUS DENTICAUDA 3 810 2170.5 EUTHYSTIRA BRACHYPTERA 13 810 2170.5 GOMPHOCERIPPUS RUFUS 3 810 2170.5 GOMPHOCERUS SIBIRICUS 35 810 2170.5 METRIOPTERA SAUSSURIANA 3 810 2170.5 STAURODERUS SCALARIS 35 826 2176 CHORTHIPPUS PARALLELUS 75 826 2176 PODISMA PEDESTRIS 3 826 2176 STAURODERUS SCALARIS 13 899 2184.2 CHORTHIPPUS BIGUTTULUS 3 899 2184.2 MIRAMELLA ALPINA 3 899 2184.2 OMOCESTUS VIRIDULUS 3 899 2184.2 STENOBOTHRUS LINEATUS 3 824 2226.1 CHORTHIPPUS PARALLELUS 13

- 108 -

824 2226.1 PODISMA PEDESTRIS 13 832 2227.8 CHORTHIPPUS PARALLELUS 3 842 2242.8 ARCYPTERA FUSCA 3 842 2242.8 CHORTHIPPUS PARALLELUS 75 842 2242.8 EUTHYSTIRA BRACHYPTERA 35 842 2242.8 METRIOPTERA SAUSSURIANA 3 842 2242.8 MIRAMELLA ALPINA 35 842 2242.8 POLYSARCUS DENTICAUDA 13 842 2242.8 STAURODERUS SCALARIS 13 906 2255 MIRAMELLA ALPINA 3 906 2255 OMOCESTUS VIRIDULUS 3 905 2255.8 ANONCONOTUS ALPINUS 13 905 2255.8 CHORTHIPPUS PARALLELUS 75 905 2255.8 METRIOPTERA SAUSSURIANA 13 905 2255.8 MIRAMELLA ALPINA 13 853 2257 NA 0 904 2259.4 CHORTHIPPUS PARALLELUS 13 904 2259.4 MIRAMELLA ALPINA 13 857 2261.4 NA 0 850 2275.5 NA 0 844 2282.9 CHORTHIPPUS PARALLELUS 35 844 2282.9 MIRAMELLA ALPINA 13 877 2283.8 NA 0 856 2291.4 EUTHYSTIRA BRACHYPTERA 3 856 2291.4 GOMPHOCERUS SIBIRICUS 75 856 2291.4 OMOCESTUS VIRIDULUS 3 862 2293.1 CHORTHIPPUS BIGUTTULUS 3 862 2293.1 EUTHYSTIRA BRACHYPTERA 3 862 2293.1 STENOBOTHRUS LINEATUS 3 860 2310.2 GOMPHOCERUS SIBIRICUS 3 860 2310.2 METRIOPTERA SAUSSURIANA 3 860 2310.2 MIRAMELLA ALPINA 13 852 2322 NA 0 861 2331.7 CHORTHIPPUS PARALLELUS 75 861 2331.7 METRIOPTERA ROESELII 3 861 2331.7 MIRAMELLA ALPINA 35 861 2331.7 OMOCESTUS VIRIDULUS 13 861 2331.7 STAURODERUS SCALARIS 3 869 2339.7 CHORTHIPPUS PARALLELUS 3 847 2342.1 NA 0 849 2355.2 NA 0 848 2373.1 CHORTHIPPUS BIGUTTULUS 13 848 2373.1 EUTHYSTIRA BRACHYPTERA 3 848 2373.1 GOMPHOCERUS SIBIRICUS 3 848 2373.1 MIRAMELLA ALPINA 13 854 2383.8 NA 0 855 2404.8 CHORTHIPPUS PARALLELUS 35 859 2437.1 NA 0

- 109 -

887 2443.2 NA 0 858 2487.2 NA 0 870 2492.2 NA 0 884 2515.5 NA 0 868 2524 NA 0 873 2535.4 NA 0 871 2588.9 NA 0 867 2601.5 OMOCESTUS VIRIDULUS 3 909 2632.4 NA 0 866 2662.4 NA 0 879 2679.1 NA 0 880 2689.7 NA 0 912 2796.4 NA 0 878 2804.1 NA 0 911 2836.8 NA 0 886 3041.6 NA 0

- 110 -

Table S2. Relationships between plant palatability and four predictor variables at the plant species level (n=129) estimated from phylogenetic least squares model (PGLS) including all predictor variables, and when using either quantile 5% or 95% of species elevation as predictor instead of species mean elevation. The PGLS allows correcting for non-independence of the observations due to phylogenetic relationships. The table shows the coefficient of determination (R2), the t-value and the standardized regression coefficients (Estimate). SLA (specific leaf area); LDMC (leaf dry matter content); C:N (carbon-to-nitrogen content); ELEV q5 (quantile 5% of species elevation); ELEV q95 (quantile 95% of species elevation). *P < 0.05, **P < 0.01, ***P < 0.001.

Species-level PGLS PGLS Estimate t-value Estimate t-value SLA 0.004 0.032 ns SLA 0.033 0.293 ns LDMC -0.087 -0.679 ns LDMC -0.083 -0.641 ns C:N -0.259 -2.566 * C:N -0.252 -2.493 * ELEV q5 -0.080 -0.848 ns ELEV q95 -0.013 -0.134 ns

2 2 R 0.056 * R 0.049 * lamda 0.420 lamda 0.399

- 111 -

Table S3. Relationships between community weighted mean (CWM) of plant palatability and four predictor variables at the plant community level estimated from ordinary least squares multiple regressions (OLS) and bivariate linear regressions, when considering only plots whose cover was composed with more than 80% (n = 167) or 90% (n = 72) of species for which species trait measurement were available. SLA was not considered in the OLS model since it was highly correlated to the other predictors. The table shows the coefficient of determination (R2), the t-value and the standardized regression coefficients (Estimate). SLA (specific leaf area); LDMC (leaf dry matter content); C:N (carbon-to-nitrogen content); ELEV (mean elevation). *P < 0.05, **P < 0.01, ***P < 0.001.

Community level OLS - 80% OLS - 90% Estimate t-value Estimate t-value SLA SLA

LDMC -0.484 -5.398 *** LDMC -0.482 -3.797 *** C:N 0.064 0.715 ns C:N 0.090 0.683 ns ELEV 0.345 5.422 *** ELEV 0.362 3.543 ***

2 2 R 0.337 *** R 0.352 ***

- 112 -

Figure S1. Location of the study area in the western Alps of Switzerland. The dots (green and red) represent vegetation sampling sites. The red dots correspond to the vegetation sites where grasshoppers were also sampled. The gray line shows the limits of study area and the dark gray line shows the 1000 m isoline. The two black lines represent the approximate location of two elevation gradient transects where plants were collected for trait analyses.

- 113 -

Figure S2. Larval weight of Spodoptera littoralis plotted onto a plant phylogeny (square root transformed). The trait mapped correspond to the mean larval weight of Spodoptera littoralis after 5 days of feeding on leaf plant samples, as a measure of plant palatability. Bigger circles indicate a high larval weight and a lower defence against herbivores in the plant.

- 114 -

CHAPTER 3

Alpine plant palatability is associated with physical and chemical traits in situ and under a warming treatment

Patrice Descombes1,2*, Alan Kergunteuil3, Gaëtan Glauser3, Sergio Rasmann3, Loïc Pellissier1,2

1 Landscape Ecology, Institute of Terrestrial Ecosystems, ETH Zürich, CH-8092 Zürich, Switzerland 2 Swiss Federal Research Institute WSL, CH-8903 Birmensdorf, Switzerland 3 Laboratory of Functional Ecology, Institute of Biology, University of Neuchâtel, CH-2000 Neuchâtel, Switzerland

In preparation (target: Global Change Biology)

- 115 -

Abstract

Plants protect themselves against herbivore attacks through various physical structures and toxic secondary metabolites. Climate warming may affect plant species phenotypes, which in turn may reshape plant-herbivore interactions. Here, we investigated how plant physical traits and chemical profile relate to herbivore performance in subalpine and alpine grasslands of the Swiss Alps. We assessed the rate of natural herbivory, and measured plant palatability using Spodoptera littoralis caterpillars as a bioassay. We related natural herbivory and the growth of S. littoralis to a set of leaf physical traits and axes of an ordination of the species chemical profile measured using ultra-high pressure liquid chromatography (UHPLC). We showed that both natural herbivory and plant palatability were associated to an ordination axis of the plant chemical profile distinguishing Gentianaceae, Brassicaceae, Euphorbiaceae, Primulaceae and Violaceae species (lower herbivory) to Geraniaceae, Asteraceae, Ericaceae and Plantaginaceae species (higher herbivory). Next, we assessed the effect of temperature warming on leaf physical and chemical profiles together with plant palatability using open top chambers greenhouses. We found that the warming treatment increased SLA by 9.9 % and decreased chemical richness by 1.1% on average, but we found no evidence that changes in leaf functional traits were systematically associated to changes in plant palatability. Overall, we provide evidence for a relationship between family specific chemical profiles on natural herbivory and plant palatability. Our results further suggest that leaf physical and chemical properties of alpine plants may be altered under climate warming, but with possible weak effects on herbivore fitness.

Introduction

Plants have evolved a wide array of defence traits, such as physical structures and toxic secondary metabolites, to protect themselves against herbivore attack (Agrawal & Fishbein, 2006; Farmer, 2014; Rhoades, 1979; Schoonhoven, Van Loon, van Loon, & Dicke, 2005). Physical defences, such as leaf toughness, trichomes or silica content, affect herbivore performance by decreasing leaf palatability and digestibility (Awmack & Leather, 2002; Brizuela, Detling, & Cid, 1986; Hanley, Lamont, Fairbanks, & Rafferty, 2007; Massey, Ennos, & Hartley, 2006; Massey & Hartley, 2009). Chemical defences, such as alkaloids, terpenoids and phenolic compounds, act as toxins or digestibility reducers (Mithöfer & Boland, 2012). The current paradigm indicates that chemical and physical defences act together in the form of syndromes to counteract a wide array of herbivores (Agrawal & Fishbein, 2006; K. Callis-Duehl, Vittoz, Defossez, & Rasmann, 2017; Kursar et al., 2009). In addition to the selective effect of herbivores on plant defence traits (Agrawal, 1998; Kessler & Baldwin, 2001), abiotic factors, especially temperature, can change the

- 116 - expression of plant phenotypes (Gutbrodt, Mody, & Dorn, 2011; Totland, 1999) and in turn influence interaction with herbivores (Lemoine, Drews, Burkepile, & Parker, 2013). Hence, to understand how climate change may reshape plant-herbivore interactions, the relation between plant phenotype and herbivory should not only be documented under extant temperature, but also under warmer conditions (DeLucia, Nabity, Zavala, & Berenbaum, 2012).

The amount and diversity of physical and chemical defences can vary among species (Agrawal & Fishbein, 2006). Physical traits can be a combination of leaf toughness, abrasive compounds such as silica, and other appendices such as trichomes (Awmack & Leather, 2002; Brizuela et al., 1986; Hanley et al., 2007; Massey et al., 2006; Massey & Hartley, 2009). Beyond physical traits, the elemental composition of leafs has a large influence on their palatability. Loranger et al. (2012) identified that leaf nitrogen and lignin concentration were the two most important determinants of herbivory for a pool of plant species under natural herbivory. Nitrogen is generally scarce in plants and is a limiting nutrient for many herbivores (Mattson, 1980), so that tender leaves with higher nutritional quality are preferred by herbivores (Pérez‐Harguindeguy et al., 2003). In addition to primary metabolites, plant evolved a myriad of secondary metabolites to counteract the effect of herbivores (Mithöfer & Boland, 2012; Rhoades, 1979). Because quantifying and documenting such complex plant-based chemical profile has been generally technically challenging, researchers have often measured the relative importance of chemical traits in mediating plant herbivore interaction in natural communities using controlled palatability bioassays with highly polyphagous insect herbivores, whose response is considered as a proxy of plant chemical toxicity (Pellissier et al., 2012; Pérez‐Harguindeguy et al., 2003). Therefore, the current challenge in plant-herbivore interaction research is to combine recent developments in analytical chemistry and bioinformatics with the direct assessment on herbivory and herbivore performance for a large range of naturally-growing species (Phyllis D. Coley, Endara, & Kursar, 2018; Kergunteuil, Descombes, Glauser, Pellissier, & Rasmann, 2018; Richards et al., 2015; Diego Salazar et al., 2018).

In addition of using insect herbivore bioassays, to circumvent chemical analytical challenges for quantifying the effect of phytochemical diversity, researchers have used plant phylogenetic positions as proxy (Futuyma & Agrawal, 2009). The general assumption behind such approach is that plant chemical compounds and other defensive traits are phylogenetically conserved (Ehrlich & Raven, 1964). For instance, Wink (2003) used molecular phylogenies of the Fabaceae, Solanaceae and Lamiaceae to map the distribution of defence compounds that are typical for the respective plant families and showed that classes of secondary metabolites are generally conserved. Moreover, phylogenetically conserved plant-herbivore interactions networks suggest that plant phylogenies

- 117 -

can be used as proxies of plant physical and chemical profile (Farrell & Mitter, 1998; Janz & Nylin, 1998; Pellissier et al., 2013; Rasmann & Agrawal, 2011; Rønsted et al., 2012). For instance, Rasmann & Agrawal (2011) showed that the expression of secondary metabolites in Asclepias responded to both ecological conditions and phylogeny. Nevertheless, low phylogenetic signal of leaf secondary chemicals were also found in the genus Piper (Diego Salazar, Jaramillo, & Marquis, 2016) or in the tropical tree Inga (Kursar et al., 2009), where co-occurring species tended to diverge in chemical composition. Hence, whether plant chemical profiles have a strong phylogenetic signal and whether the latter is stronger than physical traits remains to be evaluated across systems. Identifying phylogenetic signals in traits is of main importance for large-scale comparative studies which should take into account the phylogenetic non-independence into their analyses (Agrawal et al., 2009).

Because plant physical and chemical expression may be altered by abiotic factors (Hochachka & Somero, 2002), climate warming might reshape plant–herbivore interactions (Gutbrodt et al., 2011; Pellissier et al., 2018). Increased temperature might enhance the metabolism and growth of some, but not all, plant species (Veteli, Kuokkanen, Julkunen‐Tiitto, Roininen, & Tahvanainen, 2002), which might in turn affect the nutritional content or the concentrations of defensive compounds in plants, thereby affecting herbivore feeding preferences (Coley, 1985; Evans & Burke, 2013; Gutbrodt et al., 2011). To date, most studies quantifying the impact of warming on plant–herbivore interactions have focused on how warming affects foliar damage (Lemoine, Burkepile, & Parker, 2014; Lemoine et al., 2013) or growth rates of whole plants (O’Connor, 2009; Richardson, Press, Parsons, & Hartley, 2002). Yet, there is little information on how climate warming might affect plant traits and, subsequently, alter plant–herbivore interactions (DeLucia et al., 2012). Rising temperatures can alter the efficacy of plant defences against herbivores, which become either more or less susceptible at higher temperatures (Lemoine et al., 2013; Stamp & Yang, 1996). Generally, while consumption rate and herbivore performance are favoured by warming (Lemoine et al., 2013), considerable variation exists across different plant species (Braschler & Hill, 2007; Chong, van lersel, & Oetting, 2004; Lemoine et al., 2013). Warming might impact plant chemistry and has for instance been shown to decrease leaf nitrogen or sugar contents, both influencing plant palatability to herbivores (Zvereva & Kozlov, 2006). Warming may further induce stresses in plants (Melillo et al., 2002), which might affect plant susceptibility to herbivory (Evans & Burke, 2013; Gutbrodt et al., 2011). Thus, whether warming exacerbates or diminishes insect herbivore performance should be further investigated (Bidart‐Bouzat & Imeh‐Nathaniel, 2008; Gutbrodt et al., 2011).

- 118 -

In this study, we assessed plant palatability for chewing insect herbivores, natural herbivory, physical defence traits, nutritional composition, and chemical traits of all plant species growing in three high elevation grasslands at 1800 m, 2070 m and 2270 m in the Swiss-Alps (Fig. S1). We further quantified the effect of temperature change on plant palatability and leaf traits by warming those alpine plant communities using open-top chambers (OTC) (Henry & Molau, 1997; Marion et al., 1997). We had the following hypotheses:

1) We expected a stronger phylogenetic signal in chemical traits driven by plant families presenting specific chemical profiles, than physical traits, which can be more environmentally driven.

2) We expected natural herbivory and plant palatability to be associated with physical and chemical traits due to biomechanical and chemical feeding constraints.

3) We expected that a boost in plant metabolism under the warming treatment should alter plant phenotypes with a negative effect on plant palatability to herbivores.

Methods

Study sites and plot selection

In 2014, we selected three typical unmanaged (no mowing and no pasture grazing) alpine, species rich, calcareous grasslands at 1800 m (82 plant species over a 100 m2 surface), at 2070 m (75 plant species) and at 2270 m (59 plant species) in the Swiss-Alps (Fig. S1), and presenting similar dominant floristic composition (Seslerion vegetation type community; Delarze, Gonseth, & Galland, 1998). On each field site, we randomly selected 16 plant communities of 50 cm x 50 cm as similar as possible in floristic composition, canopy structure and plant biomass.

Measurements of plant physical traits

We measured for all plant species recorded on the field sites a set of functional traits relating to (i) competition (i.e. plant height, leaf area), (ii) the leaf economic spectrum (i.e. leaf dry matter content, specific leaf area) and (iii) physical and chemical resistance against abiotic stresses and/or herbivores (i.e. leaf toughness, silica content, carbon to nitrogen ratio, chemical richness, chemical diversity, total chemical abundance, leaf metabolomics components content).

Plant height and leaf area are related to the plant competitive vigour (Cornelissen et al., 2003; Gaudet & Keddy, 1988). SLA is correlated with the potential relative growth rate of a plant,

- 119 -

investment in structural defence strategies, resource acquisition and leaf life span (Cornelissen et al., 2003). High SLA leaves have higher protein and minerals content, but tend to have low concentrations of other expensive compounds such as lipids or lignin (Villar & Merino, 2001). High LDMC plants have higher leaf toughness, present higher resistance to physical hazards (i.e. herbivory, wind, hail) (Cornelissen et al., 2003) and lower digestibility (Gardarin et al., 2014) and are mostly associated to low productive and low disturbed environments (Cornelissen et al., 2003). Leaf toughness is an indicator of carbon investment in structural protection of the photosynthetic tissues, where high values confer better protection against abiotic conditions (i.e. wind, hail) and biotic damages (Cornelissen et al., 2003; Pérez-Harguindeguy et al., 2013). Plant toughness is also correlated to the orthopteran incisive strength and may work as a barrier trait preventing insects from chewing the leaves (Ibanez, Lavorel, Puijalon, & Moretti, 2013; Santamaría & Rodríguez- Gironés, 2007). Finally, silica is known to affect herbivore performance by decreasing leaf digestibility and increasing leaf abrasiveness for insect’s mandibles (Brizuela et al., 1986; Massey et al., 2006; Massey & Hartley, 2009).

Plant height was assessed on each site in 2014 by using a vegetation height survey performed with the point-intersect method (Jonasson, 1988; Mueller-Dombois & Ellenberg, 1974) on 32 plots (50 x 50 cm). We recorded all plant contacts with a 6 mm steel rod needle that was vertically placed through the plant community on each intersection of an equally spaced grid pattern placed over each plot (36 measurements per plot). We recorded plant contacts within five vertical height categories (i.e. 0-2 cm, 2-5 cm, 5-10 cm, 10-20 cm and 20-50 cm) and calculated for each plant species and site the mean height. We measured leaf area (LA), specific leaf area (SLA) and leaf dry matter content (LDMC) by following standard protocols (Cornelissen et al., 2003; Pérez- Harguindeguy et al., 2013). In 2015, we randomly collected on each field sites 10 individuals per species at the same phenological stage whenever possible. We stored plant individuals in moist bags and in a cool box (10°C) and rehydrated them previous to measurements (Vaieretti, Díaz, Vile, & Garnier, 2007). One well-developed and undamaged leaf per individual was chosen for leaf measurements. Leaves were scanned, weighed wet and dried during 4 days at 50°C. LA was estimated from the scanned leaf by using custom codes in R (in mm2). SLA was measured as LA divided by its leaf dry mass (mm2 mg-1) and LDMC (mg g-1) was calculated as the ratio of the leaf dry mass to its saturated fresh mass. We also multiplied the total number of leaves counted in the plot by the mean dry mass of ten leaves of the plant species collected on the field site in order to obtain the total leaf mass of each species in plots representing total biomass per species, a measure of biomass availability for herbivores.

- 120 -

We assessed leaf toughness with a punching test machine (Imada Inc) measuring the force required to pierce a hole through the lamina of the leaf (Aranwela, Sanson, & Read, 1999; Sanson, Read, Aranwela, Clissold, & Peeters, 2001). The device consists of a flat-ended cylindrical steel rod (2 mm diameter) mounted onto a moving head which can be applied through a sharp-edged hole with a 0.15 mm clearance located on a stationary base (Aranwela et al., 1999; Sanson et al., 2001). In 2016, we randomly collected on each field sites 10 individuals per species at the same phenological stage whenever possible. We stored plant individuals in moist bags and in a cool box (10°C) and rehydrated them previous to measurements (Vaieretti et al., 2007). One well-developed and undamaged leaf per individual was chosen for the punch test. We measured the leaf thickness prior to punch test with a digital caliper gauge (0.01 mm precision) by avoiding primary and secondary veins. From those measurements, we calculated the specific punch strength (i.e. the punch strength per unit leaf thickness at the point of testing expressed in GN m-2 m-1, a measure of leaf toughness.

Finally, we quantified plant leaf silica content (Si g-1) on one sample of 10 mixed dry leaves per species by using an alkaline extraction of biogenic silica with a sodium carbonate solution (K. L. Callis-Duehl, McAuslane, Duehl, & Levey, 2017; Hallmark, Wilding, & Smeck, 1982). We used the average trait value among all sampled individuals for each species and sites for further analyses.

Measurements of plant chemical traits

We performed untargeted metabolomics analyses for estimating chemical richness, chemical diversity, and total chemical abundance of each species collected in 2015. We sampled and mixed 10 leaves per species, and extracted 20 mg of dry ground tissue with 0.5 ml extraction solution (MeOH: MilliQ water: formic acid; 80:19.5:0.5) and analysed the sample via ultra-high pressure liquid chromatography - quadrupole time-of-flight mass spectrometry (UHPLC-QTOFMS) using an Acquity UPLCTM coupled to a Synapt G2 MS (Waters). A volume of 2.5 μL of extract was injected on an Acquity UPLCTM C18 column (50x2.1mm, 1.7μm). A binary solvent system consisting of H2O and acetonitrile, both supplemented with 0.05% formic acid was employed. The chromatographic separation was carried out at a flow rate of 0.6 mL/min and a temperature of 40°C using a linear gradient from 2-100% acetonitrile in 6.0 min. MS detection was done in positive electrospray ionization over a mass range of 85-1200 Da. The MS source was cleaned before each of the 3 batches running over 3 consecutive days and peak piking was performed in Markerlynx XS (Waters) (Gaillard, Glauser, Robert, & Turlings, 2018). We used the full metabolite profile to assess the number of individual chemical compounds per species (chemical richness), the summed abundance of chemical compounds per species (total chemical abundance) and the inverse Simpson

- 121 -

diversity index based on the abundance of individual peaks per species (chemical diversity) using the package Vegan in R (Oksanen et al., 2007). We applied the binarized metabolite profile (presence and absence of metabolites) to a correspondence analysis using the ade4 package (Dray & Dufour, 2007) and retained the four first axis (CA Axis 1-4) representing 5.8% of the total variability (Fig. S2).

We measured leaf nitrogen (mg g-1) and carbon (mg g-1) contents on the same sample as for the chemical leaf metabolomics analyses by using an elemental analyser (NC-2500 from CE Instruments). We calculated the Carbon to Nitrogen ratio (C:N) which indicates plant nitrogen availability per unit of carbon to herbivores, which is indicative of the plant nutritional quality to the consumers in food webs (Cornelissen et al., 2003; Mattson, 1980). We used the average trait value among all sampled individuals for each species and sites for further analyses.

Phylogenetic signal

We evaluated phylogenetic signal of plant physical and chemical traits, by pruning the phylogeny and calculating Blomberg's K statistic with the ‘phylosignal’ function implemented in the ‘picante’ package (Blomberg, Garland, Ives, & Crespi, 2003; Kembel et al., 2010) in R version 3.4.1 (R Development Core Team, 2014). Phylogenetic relationships between plants were retrieved from a well resolved dated phylogeny of European plant species (Durka & Michalski, 2012). We pruned the tree with the plant species pool found in all our sites. We used the mean plant trait calculated across sites (plant functional traits and plant palatability: N = 132, plot herbivory: N = 101). K values close to one indicate phylogenetic signal under a Brownian motion model of evolution with respect to the phylogeny (i.e. closely related species share more similar trait values), while K values close to zero indicate that trait values are randomly distributed within the phylogeny (Blomberg et al., 2003). When K values are greater than one, close relatives are more similar than expected under a Brownian motion model of evolution. The statistical significance of the observed K value was tested and compared to a null distribution by permuting trait values 999 times across the tips of the phylogeny (Kembel et al., 2010).

Relating natural herbivory and bioassay to traits

We counted the number of plant leaves with and without herbivory marks in all control ambient plots (see warming experiment below) for each plant species at the end of season 2016. The percentage of leaf eaten was visually estimated according to a 7-level scale: : <1%, 1-5%, 5- 13%, 13-25%, 25-50%, 50-75%, 75-100%. We estimated herbivory damage only for chewing damage, because sap sucking, leaf mining, and rasping were rarely observed. We estimated the

- 122 - eaten dry leaf mass of each species in plots by multiplying the proportion of the leaf eaten by the mean dry mass of ten leaves of the plant species. We investigated the relationship between eaten dry leaf mass from natural herbivory measured on the field sites and plant traits (LDMC, toughness, SLA, C:N, plant height, available biomass, chemical richness and diversity, CA Axis 1-4 and silica content) and with a Monte Carlo Markov Chain generalised linear mixed model (Hadfield, 2010), by taking into account the phylogenetic relatedness of the plant species, by setting traits and available biomass as fixed effect, and by setting site and plot identity as random effect with uninformative priors (V=1 and nu=0.002). Because the response variable (eaten leaf mass) is zero- inflated, we multiplied it by 1000, rounded it to the nearest integer and used a MCMCglmm with zero-inflated hurdle Poisson distribution following the recommendation of Hadfield (Hadfield, 2010, 2015). We square root transformed leaf toughness, silica content and available biomass and rescaled all selected variables around there mean. Except for total chemical abundance highly correlated to chemical richness (Spearman correlations: r = 0.78, p-value < 0.001; Table S1), variables showed low correlation between each other (Spearman correlations: r < |0.72|; Table S1). We run the model with 200000 iterations and checked for convergence of posteriors in the models. Finally, we performed a principal component analysis by using the ade4 package (Dray & Dufour, 2007) to represent the major sources of variability among plant species based on the 12 functional traits and available plant biomass measured in plots of the study sites.

We evaluated palatability against chewing insect herbivores for all plant species present in the field sites (i.e. over a 100 m2: 82 plant species at 1800 m, 75 plant species at 2070 m and 59 plant species at 2270 m) by performing a bioassay experiment using caterpillars of the African cotton leafworm Spodoptera littoralis (Lepidoptera, Noctuidae) (Brown & Dewhurst, 1975) obtained from Syngenta (Switzerland). We used S. littoralis as a non‐adapted species to remove the confounding effect of possible local adaptation to plants. This species is highly generalist, reported to feed on more than 40 families of plants (Brown & Dewhurst, 1975) and is commonly used in similar bioassays (Bossdorf, Schröder, Prati, & Auge, 2004; Descombes et al., 2017; Edwards, Wratten, & Cox, 1985; Pellissier et al., 2012; Ruhnke, Schädler, Klotz, Matthies, & Brandl, 2009; Schädler, Roeder, Brandl, & Matthies, 2007). In August 2016, we randomly collected 5-7 plant leaves from different plant individuals for all plant species on the field site. Eggs were hatched on wet paper at 20 °C without food to ensure a standard size. Once hatched, ten larvae were placed on each individual plant species, by placing 2-3 leaves of each species in distinct Petri dishes for 5 days in a climatic chamber at 24 °C (L) and 18 °C (D), 55 ± 5% RH and a 14:10 L:D photoperiod. Completely eaten or dried leaves were replaced during this period with fresh leaves that were stored at 4 °C. At the end of the experiment, all the larvae were dried for 72 h at 50 °C and weighed. We

- 123 -

investigated the relationship between plant palatability (i.e. larval weight) and plant traits (LDMC, toughness, SLA, C:N, plant height, chemical richness and diversity, CA Axis 1-4 and silica content) with a Monte Carlo Markov Chain generalised linear mixed model with Gaussian distribution and implemented in the MCMCglmm package (Hadfield, 2010), by taking into account the phylogenetic relatedness of the plant species, by setting traits as fixed effect and site as random effect with uninformative priors (V=1 and nu=0.002). We square-root transformed plant palatability, leaf toughness and silica content to reach a normal distribution. We retained only uncorrelated variables (Spearman correlations: r < |0.65|; Table S2) and scaled them around their mean. We run the model with 50000 iterations and checked for convergence of posteriors in the models. Finally, we used a principal component analysis by using the ade4 package (Dray & Dufour, 2007) to represent the major sources of variability among plant species based on the 12 functional traits measured on the study sites.

Effect of OTC on traits and herbivore resistance

To simulate climate warming on alpine plant communities, we randomly allocated a warming treatment and an ambient control to the 16 plot on each site, leading to 8 replicates per site and treatment. The warming treatment consisted of a hexagonal open-top chambers (OTC; Fig. S1) following ITEX (The International Tundra Experiment) standards (Henry & Molau, 1997; Marion et al., 1997). OTCs provide an effective and simple method of climate change simulation and consisted of a hexagonal enclosure built of clear transparent 2 mm thick polymethylmethaacrylate material (PMMA-XT transparent clear, Angst+Pfister SA). The walls of the OTC have a 60° inclination relative to the ground, a ground diameter of 111 cm, a top opening of 60 cm in diameter and a height of 38 cm (Fig. S1). We carefully set up the greenhouses over the 50 cm x 50 cm plot in spring as soon as possible after snowmelt and removed them before snowfalls in October. Greenhouses were fixed to the ground with steel tent pegs and secured with tensioned ropes. Using high resolution temperature loggers (DS1922L-F5 HomeChip) placed in the middle and outside of each OTC greenhouses, the warming treatment increased the mean aboveground temperature (20 cm) by 1.1 K (linear mixed-effects model: slope = 1.06, P < 0.001) and strengthened diurnal and nocturnal extremes (mean diurnal temperature 11h-17h = +3.8 K, mean nocturnal temperature 23h- 5h = -0.6 K) during the summer season (July-August period in 2017).

To investigate the effect of the warming treatment on plant leaf traits, we randomly collected 3-5 plant leaves of 16 plant species in 4-5 plots of each treatment and control (i.e. 4-5 replicates per treatment and plant species) in August 2016. Leaves were collected around the central 50 x 50 cm plots whenever possible in order to reduce the potential impact of biomass removal on the plot

- 124 - experiment. The selected plant species correspond to the most frequent plant species observed overall (4 species at 1800 m, 5 species at 2070 m and 7 species at 2270 m) and were selected to represent as much as possible the diversity of plant families. We measured LA, SLA, LDMC and leaf toughness for the 16 plant species following the same protocols as described above and calculated the mean trait value per species and plot. We performed nutrient content (C:N) and chemical content analyses (chemical richness, chemical diversity and total chemical abundance) on one sample of dry mixed ground leaves per species and plot. All trait measurements followed the same protocols as described above. Note that we did not measure plant height and silica content in this experiment. We then related the warming treatment (binary: ambient = 0, warming =1) to plant traits (LA, LDMC, toughness, SLA, C:N, chemical richness, diversity and sum) of the 16 selected plant species collected inside and outside the OTC greenhouses with a Monte Carlo Markov Chain generalised linear mixed model with binomial distribution (Hadfield, 2010). We took into account the phylogenetic relatedness of the plant species, set traits as fixed effect, and set uninformative priors for the random effects (V=1 and nu=0.002) and strong priors for the residuals (V=1, fix=1) (Hadfield, 2015). We square root transformed leaf toughness and rescaled all selected variables around their mean. We retained only uncorrelated variables (Spearman correlations: r < |0.71|; Table S3). We run the model with 400000 iterations, reduced autocorrelation of consecutive samples by setting thinning factor to 200, and checked for convergence of posteriors in the models.

To investigate the effect of the warming treatment on plant palatability via changes in plant leaf traits properties, we performed a bioassay experiment on 7 plant species, using S. littoralis caterpillars. The 7 selected plant species (i.e. Alchemilla conjuncta aggr, Carex sempervirens, Helianthemum nummularium subsp grandiflorum, Polygonum viviparum, Potentilla crantzii, Prunella grandiflora, Sesleria caerulea) correspond to the most frequent plant species on the field sites and were selected to represent as much as possible the diversity of plant families. In 2016, we randomly collected 3-5 plant leaves of the selected 7 plant species in 3-4 plots of each treatment and control (i.e. 3-4 replicates per treatment and plant species). Leaves were collected around the central 50 x 50 cm plots whenever possible in order to reduce the potential impact of biomass removal on the plot experiment. The bioassay experiment followed the same protocol as described above. In addition, we collected 3-5 plant leaves in the same plot and measured plant traits (LA, LDMC, toughness, SLA, C:N, chemical richness, diversity and sum) following the same protocol as described above. Then, we investigated how changes in plant palatability (i.e. larval weight) are related to changes in plant leaf traits (LA, LDMC, toughness, SLA, C:N, chemical richness, diversity and sum) under the warming treatment for the 7 selected plant species with a Monte Carlo Markov Chain generalised linear mixed model with Gaussian distribution implemented in the

- 125 -

MCMCglmm package (Hadfield, 2010) and considering interaction between traits and the treatment. We took into account the phylogenetic relatedness of the plant species and set traits as fixed effect and site as random effect with uninformative priors (V=1 and nu=0.002). We square root transformed plant palatability and leaf toughness to reach a normal distribution, and rescaled all selected variables around their mean. We retained variables with low correlation between each other (Spearman correlations: r < |0.75|; Table S4). We run the model with 200000 iterations. We checked for convergence of posteriors in the models.

Results

Metabolomics ordination

UHPLC-QTOFMS profiling detected 15’668 metabolic peaks across the 248 taxa (i.e. samples) collected on the three field sites, corresponding to 154 plant species. When applying this metabolite profile to a correspondence analyses, the first axis (CA Axis 1; 1.81% of total variance) opposed Gentianaceae, Caprifoliaceae, Orobanchaceae and Plantaginaceae (high scores) to other plant families (low scores; Fig. S2). The second axis (CA Axis 2; 1.66% of total variance) distinguished Gentianaceae, Brassicaceae, Euphorbiaceae, Primulaceae and Violaceae species (low scores) to Geraniaceae, Asteraceae, Ericaceae and Plantaginaceae species (high scores; Fig. S2). The third axis (CA Axis 3; 1.22% of total variance) principally distinguished Orobanchaceae and Plantaginaceae (high scores) to other plant families (low scores; Fig. S2). The fourth axis (CA Axis 4; 1.14% of total variance) opposed the taxonomic group of Poales (high scores; i.e. Poaceae, Cyperaceae and Juncaceae) to Rosaceae, Cystaceae, Ericaceae and Fabaceae species (low scores; Fig. S2).

Phylogenetic signal in plant leaf functional traits

We found a significant but weak phylogenetic signal for most of the plant functional traits (K < 0.7, p-value < 0.05; Table S5 and Fig. 1), indicating that the variation of these plant traits is relatively labile across the plant phylogeny. Only the fourth axis of the metabolomics correspondence analyses showed a higher phylogenetic signal (CA Axis 4: K = 0.893, n = 132, Z‐ score = −5.381, p‐value = 0.001; Table S5 and Fig. 1). The phylogenetic signal was on average 2.3 times higher for chemical traits (i.e. chemical richness, chemical diversity, total chemical abundance, CA Axis 1-4; mean K value = 0.45) than traits relating to competition, leaf structure and nutritional value (i.e. LA, SLA, LDMC, toughness, C:N, plant height, Silica; mean K value = 0.19).

- 126 -

Figure 1. Phylogenetic signal in plant functional traits assessed with Blomberg's K statistic. K values close to one indicate that trait values follow a Brownian motion model of evolution with respect to the phylogeny (i.e. closely related species chare more similar trait values). The statistical significance of the observed K value was tested and compared to a null distribution by permuting trait values 999 times across the tips of the phylogeny. The phylogenetic signal is generally higher in chemical traits (i.e. ChemRich, ChemDiv, ChemSum, CA Axis 1-4) compared to traits relating to competition, leaf structure and nutritional content (i.e. toughness, LDMC, Silica, plant height, LA, SLA, C:N). On this figure, each trait was log+1 transformed and normalized. The colour scale represents the size of the trait value. LA = leaf area, SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, ChemRich = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, ChemDiv = chemical diversity based on the abundance of individual peaks per species, ChemSum = total chemical abundance based on the abundance of individual peaks per species , CA Axis 1-4 = four first axes of the correspondence analyses of the untargeted metabolomics analyses, * p-value < 0.05,** p-value < 0.01, *** p-value < 0.001.

Herbivory and bioassay in relation to traits

Regarding natural herbivory, among all predictor variables, herbivores have a preference for tall (MCMCglmm: slope = 0.606, p-value < 0.001; Table 1a and Fig. S4) and abundant plants (slope = 0.841, p-value < 0.001; Table 1a and Fig. S4), which are nitrogen rich (slope = -0.714, p- value < 0.001; Table 1a and Fig. S4), present low SLA (slope = -0.397, p-value = 0.001; Table 1a

- 127 -

Table 1. (a) Relationship between in situ herbivory (i.e. dry leaf mass eaten by a natural community of herbivores) and 12 plant functional traits and available biomass estimated with MCMCglmm under zero- inflated hurdle Poisson distribution. Each datum in Hurdle zero-inflated model are associated with two latent variables (intercept = the mean parameter of a zero-truncated Poisson distribution; Hurdle = the probability on the logit scale that the response variable is zero or not). (b) Relationship between plant palatability (i.e. larval weight of Spodoptera littoralis assessed with a bioassay experiment) and 12 plant functional traits estimated with MCMCglmm under Gaussian distribution. Significant variables are highlighted in bold. CA = correspondence analyses of the leaf metabolomics, *** p-value < 0.001, ** p- value < 0.01, * p-value < 0.05, . p-value < 0.1.

(a) post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) 2.864 -0.997 6.579 2988 0.121 Hurdle -1.527 -1.886 -1.170 2779 0.000 *** Plant biomass 0.841 0.630 1.056 4038 0.000 *** LDMC 0.243 -0.211 0.694 2857 0.283 Toughness 0.292 -0.105 0.708 3839 0.164 SLA -0.397 -0.644 -0.142 3975 0.001 ** C:N -0.714 -1.075 -0.363 2631 0.000 *** Chemical richness 0.255 -0.091 0.595 4508 0.147 Chemical diversity 0.180 -0.070 0.435 4396 0.154 CA Axis 1 0.153 -0.122 0.431 3416 0.277 CA Axis 2 0.559 0.172 0.915 3550 0.002 ** CA Axis 3 0.173 -0.057 0.396 3830 0.137 CA Axis 4 -0.176 -0.559 0.217 2783 0.377 Height 0.606 0.385 0.824 4597 0.000 *** Silica -0.171 -0.450 0.098 5206 0.214

(b) post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) 0.417 0.079 0.788 4700 0.026 * LDMC -0.022 -0.065 0.013 4343 0.240 Toughness -0.016 -0.053 0.021 4700 0.389 SLA 0.035 0.011 0.060 4700 0.006 ** C:N -0.014 -0.041 0.017 4508 0.357 Chemical richness 0.033 0.000 0.068 4700 0.054 . Chemical diversity 0.008 -0.016 0.031 4700 0.501 CA Axis 1 -0.035 -0.076 0.007 4505 0.100 CA Axis 2 0.047 0.005 0.091 4700 0.034 * CA Axis 3 0.008 -0.032 0.044 4700 0.697 CA Axis 4 -0.068 -0.120 -0.017 4700 0.009 ** Height 0.019 -0.003 0.042 4700 0.081 . Silica -0.010 -0.042 0.020 4700 0.519

and Fig. S4) and have high scores on the second axis of the metabolic ordination (CA Axis 2: slope = 0.559, p-value = 0.002; Table 1a, Fig. S2 and Fig. S4). Regarding the bioassay experiment, S.

- 128 - littoralis caterpillars preferred plants with high SLA (slope = 0.035, p-value = 0.006; Table 1b and Fig. 2), high scores on the second axis of the metabolic ordination (CA Axis 2: slope = 0.047, p- value = 0.034; Table 1b, Fig. 2 and Fig. S2) and low scores on the fourth axis of the metabolic ordination (CA Axis 4: slope = -0.068, p-value = 0.009; Table 1b, Fig. 2 and Fig. S2).

Figure 2. Principal component analysis (PCA) representing major sources of variability among plant species based on the 12 functional traits measured on the study sites. Species scores are represented along axes 1 and 2, and coloured according to their affiliation to diverse or note‐worthy plants clades (see also Fig. S3). The size of the points represents the caterpillar dry weight in the bioassay experiment after five days of feeding on the plant species and reflects the palatability of the plant to a generalist insect herbivore. The lengths of the vectors in the correlation circle represent the magnitude of the correlation between the variables and the axes. Abbreviations are explained in Fig. 1.

Effect of warming on plant functional traits and plant palatability

The warming treatment significantly affected SLA and chemical richness of the 16 investigated plant species (Table 2). On average, the warming treatment increased SLA (MCMCglmm: slope = 1.540, p-value = 0.009; Table 2) by approximately 9.3 % (mean ± sd = 9.3 ± 8.0) and decreased the chemical richness (slope = -1.867, p-value = 0.033; Table 2) in plant leaves by 1.1 % (mean ± sd = -1.1 ± 3.6). At the species level, all species increased their SLA values on average under the warming treatment, except Anthyllis alpestris (Fig. S5), but we found contrasted responses of chemical richness to warming (Fig. S6). Plant palatability was not related to plant

- 129 -

functional traits of the 7 investigated plant species under the warming treatment (MCMCglmm: all traits with p-value > 0.05; Table S6). At the species level, we found contrasted responses of S. littoralis to the warming treatment (Fig. S7).

Table 2. Warming effect on 8 plant functional traits estimated with MCMCglmm under binomial distribution (ambient control = 0, warming treatment = 1). Significant variables are highlighted in bold. *** p-value < 0.001, ** p-value < 0.01, * p-value < 0.05. post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) 0.482 -5.196 5.886 1985 0.888 LA 1.103 -0.376 3.348 755 0.164 SLA 1.540 0.118 2.981 1342 0.009 ** LDMC -1.012 -3.669 1.195 1140 0.373 Toughness -0.174 -1.613 1.455 1764 0.752 C:N 0.829 -0.256 2.140 1043 0.128 Chemical richness -1.867 -3.734 -0.207 1985 0.033 * Chemical diversity 0.581 -0.548 1.727 1985 0.266 Chemical sum 0.801 -0.642 2.349 1985 0.265

Discussion

Our analysis of herbivory and palatability in alpine grasslands provides a more comprehensive understanding of the role of physical and chemical traits in modulating herbivory, and how those may shift under climate warming. In particular, using untargeted metabolomics applied to all species in the sampled communities, we observed a consistent effect of plant species chemical composition on the degree of natural herbivory and the bioassay with S. littoralis. Hence, beyond traditional measures of physical traits in herbivory studies (Cornelissen et al., 2003; Pérez- Harguindeguy et al., 2013), we stress the necessity to characterize the chemical profiles of plants to understand and predict their susceptibility to herbivory within communities. We further showed that the warming treatment modified both physical and chemical plant phenotypes, but these changes did not result in a significant altered plant palatability to S. littoralis. Hence, while plant leaf physical and chemical properties may be altered under climate warming, shifts in phenotypes may not systematically reshape the plant susceptibility to herbivores.

Beyond the effect of physical and nutritional traits documented in many studies (Descombes et al., 2017; Hanley et al., 2007; Massey et al., 2006; Peeters, Sanson, & Read, 2007; Pérez‐ Harguindeguy et al., 2003), we found that both in situ herbivory and the performance of S. littoralis were associated to the plant chemical profile summarized along ordination axes. We identified

- 130 - major trends related to plant taxonomy: overall resistance was higher and herbivory lower on Gentianaceae and Euphorbiaceae plant species, which have been shown to harbour toxic or repellent chemical compounds (e.g. bitterness in Gentianaceae, toxic latex in Euphorbiaceae; Singh & Agarwal, 1988). In contrast, Asteraceae, Rosaceae, Campanulaceae and some Fabaceae plant species were more palatable. In addition, the performance of S. littoralis (but not in situ herbivory) was also lower when feeding on Cyperaceae and Poaceae plant species which have distinct chemical profiles compared to other plant families. So far, studies investigating the relationship between chemical traits and herbivore performance are typically measuring one or a few groups of chemical compounds, such as flavonoids, phenolics or cardenolides (e.g. K. Callis-Duehl et al., 2017; Pellissier et al., 2016; Rasmann & Agrawal, 2011). Because plants have evolved a myriad of chemical secondary metabolites to counteract herbivory (Mithöfer & Boland, 2012; Rhoades, 1979), the overall chemical arsenal in plants is unlikely to be restricted to one single compound class. Our results provide the evidence that untargeted metabolomics analyses are highly powerful for identifying family patterns in plant defences for a large range of naturally-growing species. Yet, a major challenge of this approach is to deal with the high dimensionality of the plant chemical profile, which may contain thousands of metabolomics peaks (i.e. here 15’668) and to explain a fraction of the total variance when summarizing chemical profiles along ordination axes (i.e. here 5% for the four first axes). Future studies should develop pipelines to better identify metabolites based on metabolomics peaks and use adequate tools to reduce the dimensionality of the data (Barker & Rayens, 2003; Kuhl, Tautenhahn, Böttcher, Larson, & Neumann, 2012; Diego Salazar et al., 2018; Tibshirani, 1996).

Physical traits, classically assessed in plant herbivore-interaction studies (Descombes et al., 2017; Hanley et al., 2007; Massey et al., 2006; Peeters et al., 2007; Pérez‐Harguindeguy et al., 2003), were also associated to natural herbivory and S. littoralis performance (Table 1). In particular, we found a negative relationship between C:N and natural herbivory in grasslands (Table 1a), suggesting a preference for plants with high nutritional content. Nitrogen is generally scarce in plants and is a limiting nutrient for many herbivores (Mattson, 1980), so that leaves with higher nutritional quality are preferred by herbivores (Pérez‐Harguindeguy et al., 2003). Moreover, we found that S. littoralis had a higher performance on plants with high leaf SLA values (Table 1b). Plants with high SLA are typically fast growing species and tend to be richer in nitrogen, proteins and minerals, and to display lower concentrations of other cost-intensive secondary compounds such as lipids, tannins or lignin (Pérez-Harguindeguy et al., 2013; Pérez‐Harguindeguy et al., 2003; Villar & Merino, 2001). Herbivorous insects may thus perform better on high SLA plant species because they confer a combined higher nutritional value and a higher digestibility due to lower

- 131 -

concentrations of deleterious secondary chemical compounds. However, in contrast to the bioassay experiment, we found that the in situ herbivore community preferred to eat low SLA species (Table 1a). Hence, natural herbivory may also be influenced by other traits, such as biomechanical limitation in mandibular strength (Ibanez et al., 2013), or different resource acquisition strategies between guilds of herbivores. Furthermore, leaf accessibility to herbivores in natural grasslands can be context dependent and varies between plant species (Pérez‐Harguindeguy et al., 2003). For instance, plants may attract natural enemies of the herbivores (Rasmann & Turlings, 2007; Turlings, Tumlinson, & Lewis, 1990), induce chemical defence (Karban, Agrawal, Thaler, & Adler, 1999), associate with unpalatable plants to increase their resistance (Olff et al., 1999), or simply be more attractive to herbivores due to their higher abundance on the field site. This is supported by the relationship between herbivory and the height and available biomass of plants in our analyses (Table 1a). Hence, while we found some similarities between the bioassay and natural herbivory, predicting the response of plants in a community context is likely to be more complex.

Phylogeny is frequently used as a proxy for plant chemical profiles in studies of plant-insect interactions (Futuyma & Agrawal, 2009) assuming a conservatism in plant chemical expression (Ehrlich & Raven, 1964; Wink, 2003). Because secondary metabolites have been shown to be generally conserved along the angiosperm phylogeny (Wink, 2003), they may be very good candidate traits to explain the structure of plant herbivore interactions. For instance, Pellissier et al. (2013) observed a strong pattern of phylogenetic matching between butterflies and plants, suggesting that butterflies are responding to phylogenetically conserved plant phenotypes. Supporting this general assumption, we found a stronger phylogenetic signal for plant chemical profiles compared to physical traits (Fig. 1). The phylogenetic signal of chemical traits was on average 2.3 times higher than traits relating to competition, leaf structure and nutritional value, which suggests that plant metabolomics are discriminating well plant families in their chemical profiles and that plant chemistry follows a stronger phylogenetic inertia. In contrast, we found only weak phylogenetic signal for almost all plant leaf traits, principally physical traits relating to competition, leaf structure and nutritional value (Fig. 1). This suggests that variation of several plant physical functional traits across species shows a pattern of adaptation to the environment, with limited phylogenetic inertia (Wiens & Graham, 2005).

Climate warming increased leaf SLA and decreased plant chemical richness across the investigated plant species. While most studies investigating changes in leaf traits are performed at the species level or at the plant functional type level (i.e. such as evergreen or deciduous shrubs, forbs, grasses and sedges) (Hudson, Henry, & Cornwell, 2011), our study found a consistent change in SLA and chemical richness across 16 forbs, grasses and sedges. Hudson et al. (2011) found that

- 132 -

SLA was decreased in shrubs under warming, that grasses and forbs responded less to long-term warming than evergreen or deciduous shrubs, and that traits related to growth are more affected by warming than chemical traits. Contrastingly, our result indicates that warming increases plant chemical richness, which might affect plant palatability to herbivores (Gutbrodt et al., 2011). Anti- herbivore phenolic compounds or terpenoids has been shown to decrease or increase in responses to warming (Veteli et al., 2002; Zvereva & Kozlov, 2006). In contrast, we found no significant change in plant palatability associated to changes in SLA across the 7 plant species investigated, suggesting that climate warming might not systematically alter herbivore fitness or reshape plant-herbivore interactions. This result may be due to different and species-specific responses of S. littoralis to increased temperature and changes in plant leaf physical or chemical properties, as observed at the species level (Fig. S7). In agreement with our results, O’Connor (2009) reported no effect of warming on plant palatability, while other studies reported contrasting responses of herbivores to warming when feeding on different plant species (Bidart‐Bouzat & Imeh‐Nathaniel, 2008; Gutbrodt et al., 2011). Thus, warming might exacerbate or negate insect herbivore performance by altering plant palatability, and responses to warming are likely to be species-specific (Bidart‐Bouzat & Imeh‐Nathaniel, 2008; Gutbrodt et al., 2011; Lemoine et al., 2014).

Together, our study reports evidence of a strong association of plant physical and chemical phenotypes with herbivory under both in situ herbivory and S. littoralis bioassay performance. Nevertheless, despite altered physical and chemical properties in a warming treatment, we found no consistent directional response of plant palatability. Hence, the response of plant-herbivore interaction to climate change might be species-specific and cannot be easily generalized (Lemoine et al., 2014). Because warming can also influence herbivore consumption rates and dietary preferences (Lemoine et al., 2014; Lemoine & Shantz, 2016), forecasted changes in plant-herbivore interactions under climate change should investigate the combined effect of warming on plants defences and insects performance. Understanding how plants and their natural enemies will be affected by climate change is crucial to predict future community changes.

Acknowledgements

Thanks to the people who helped for fieldwork, Oliver Kindler and Roland Reist (Syngenta, Stein, Switzerland) for providing S. littoralis eggs. Thanks to Hélène Blauenstein and Andreas Zurlinden for the technical support on the palatability bioassay experiments with S. littoralis.

- 133 -

References

Agrawal, A. A. (1998). Induced responses to herbivory and increased plant performance. Science, 279(5354), 1201–1202. Agrawal, A. A., & Fishbein, M. (2006). Plant Defense Syndromes. Ecology, 87, S132–S149. Agrawal, A. A., Fishbein, M., Halitschke, R., Hastings, A. P., Rabosky, D. L., & Rasmann, S. (2009). Evidence for adaptive radiation from a phylogenetic study of plant defenses. Proceedings of the National Academy of Sciences, 106(43), 18067–18072. Aranwela, N., Sanson, G., & Read, J. (1999). Methods of assessing leaf-fracture properties. The New Phytologist, 144(2), 369–383. Awmack, C. S., & Leather, S. R. (2002). Host plant quality and fecundity in herbivorous insects. Annual Review of Entomology, 47(1), 817–844. Barker, M., & Rayens, W. (2003). Partial least squares for discrimination. Journal of Chemometrics, 17(3), 166–173. Bidart‐Bouzat, M. G., & Imeh‐Nathaniel, A. (2008). Global Change Effects on Plant Chemical Defenses against Insect Herbivores. Journal of Integrative Plant Biology, 50(11), 1339–1354. Blomberg, S. P., Garland, T., Ives, A. R., & Crespi, B. (2003). Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution, 57(4), 717–745. Bossdorf, O., Schröder, S., Prati, D., & Auge, H. (2004). Palatability and tolerance to simulated herbivory in native and introduced populations of Alliaria petiolata (Brassicaceae). American Journal of Botany, 91(6), 856–862. Braschler, B., & Hill, J. K. (2007). Role of larval host plants in the climate-driven range expansion of the butterfly Polygonia c-album. Journal of Animal Ecology, 76(3), 415–423. Brizuela, M. A., Detling, J. K., & Cid, M. S. (1986). Silicon Concentration of Grasses Growing in Sites With Different Grazing Histories. Ecology, 67(4), 1098–1101. Brown, E. S., & Dewhurst, C. F. (1975). The genus spodoptera (Lepidoptera, Noctuidae) in Africa and the Near East. Bulletin of Entomological Research, 65(2), 221–262. Callis-Duehl, K. L., McAuslane, H. J., Duehl, A. J., & Levey, D. J. (2017). The Effects of Silica Fertilizer as an Anti-Herbivore Defense in Cucumber. Journal of Horticultural Research, 25(1), 89–98. Callis-Duehl, K., Vittoz, P., Defossez, E., & Rasmann, S. (2017). Community-level relaxation of plant defenses against herbivores at high elevation. Plant Ecology, 218(3), 291–304. Chong, J.-H., van lersel, M. W., & Oetting, R. D. (2004). Effects of Elevated Carbon Dioxide Levels and Temperature on the Life History of the Madeira Mealybug (Hemiptera: Pseudococcidae). Journal of Entomological Science, 39(3), 387–397.

- 134 -

Coley, P. D., Bryant, J. P., & Chapin, F. S. (1985). Resource Availability and Plant Antiherbivore Defense. Science, 230(4728), 895–899. Coley, Phyllis D., Endara, M.-J., & Kursar, T. A. (2018). Consequences of interspecific variation in defenses and herbivore host choice for the ecology and evolution of Inga, a speciose rainforest tree. Oecologia, 1–16. Cornelissen, J. H. C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D. E., … Poorter, H. (2003). A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany, 51(4), 335–380. Delarze, R., Gonseth, Y., & Galland, P. (1998). Guide des milieux naturels de Suisse: écologie, menaces, espèces caractéristiques. Delachaux et niestlé Lausanne. DeLucia, E. H., Nabity, P. D., Zavala, J. A., & Berenbaum, M. R. (2012). Climate Change: Resetting Plant-Insect Interactions. Plant Physiology, 160(4), 1677–1685. Descombes, P., Marchon, J., Pradervand, J.-N., Bilat, J., Guisan, A., Rasmann, S., & Pellissier, L. (2017). Community-level plant palatability increases with elevation as insect herbivore abundance declines. Journal of Ecology, 105(1), 142–151. Dray, S., & Dufour, A.-B. (2007). The ade4 Package: Implementing the Duality Diagram for Ecologists. Journal of Statistical Software, 22(4). Durka, W., & Michalski, S. G. (2012). Daphne: a dated phylogeny of a large European flora for phylogenetically informed ecological analyses. Ecology, 93(10), 2297–2297. Edwards, P. J., Wratten, S. D., & Cox, H. (1985). Wound-induced changes in the acceptability of tomato to larvae of Spodoptera littoralis: a laboratory bioassay. Ecological Entomology, 10(2), 155–158. Ehrlich, P. R., & Raven, P. H. (1964). Butterflies and Plants: A Study in Coevolution. Evolution, 18(4), 586–608. Evans, S. E., & Burke, I. C. (2013). Carbon and Nitrogen Decoupling Under an 11-Year Drought in the Shortgrass Steppe. Ecosystems, 16(1), 20–33. https://doi.org/10.1007/s10021-012-9593-4 Farmer, E. E. (2014). Leaf Defence. OUP Oxford. Farrell, B. D., & Mitter, C. (1998). The timing of insect/plant diversification: might Tetraopes (Coleoptera: Cerambycidae) and Asclepias (Asclepiadaceae) have co-evolved? Biological Journal of the Linnean Society, 63(4), 553–577. Futuyma, D. J., & Agrawal, A. A. (2009). Evolutionary history and species interactions. Proceedings of the National Academy of Sciences, 106(43), 18043–18044. Gaillard, M. D. P., Glauser, G., Robert, C. A. M., & Turlings, T. C. J. (2018). Fine-tuning the ‘plant domestication-reduced defense’ hypothesis: specialist vs generalist herbivores. New Phytologist, 217(1), 355–366.

- 135 -

Gardarin, A., Garnier, É., Carrère, P., Cruz, P., Andueza, D., Bonis, A., … Kazakou, E. (2014). Plant trait–digestibility relationships across management and climate gradients in permanent grasslands. Journal of Applied Ecology, 51(5), 1207–1217. Gaudet, C. L., & Keddy, P. A. (1988). A comparative approach to predicting competitive ability from plant traits. Nature, 334(6179), 242–243. Gutbrodt, B., Mody, K., & Dorn, S. (2011). Drought changes plant chemistry and causes contrasting responses in lepidopteran herbivores. Oikos, 120(11), 1732–1740. Hadfield. (2010). MCMC Methods for Multi-Response Generalized Linear Mixed Models: The MCMCglmm R Package. Journal of Statistical Software, 33(2), 1–22. Hadfield. (2015). MCMCglmm: Course notes. Retrieved from https://cran.r- project.org/web/packages/MCMCglmm/vignettes/CourseNotes.pdf Hallmark, C. T., Wilding, L. P., & Smeck, N. E. (1982). Silicon. Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties (Methodsofsoilan2), 263–273. Hanley, M. E., Lamont, B. B., Fairbanks, M. M., & Rafferty, C. M. (2007). Plant structural traits and their role in anti-herbivore defence. Perspectives in Plant Ecology, Evolution and Systematics, 8(4), 157–178. Henry, G. H. R., & Molau, U. (1997). Tundra plants and climate change: the International Tundra Experiment (ITEX). Global Change Biology, 3(S1), 1–9. Hochachka, P. W., & Somero, G. N. (2002). Biochemical adaptation: Mechanism and process in physiological evolution. Oxford: Oxford University Press. Hudson, J. M. G., Henry, G. H. R., & Cornwell, W. K. (2011). Taller and larger: shifts in Arctic tundra leaf traits after 16 years of experimental warming. Global Change Biology, 17(2), 1013–1021. Ibanez, S., Lavorel, S., Puijalon, S., & Moretti, M. (2013). Herbivory mediated by coupling between biomechanical traits of plants and grasshoppers. Functional Ecology, 27(2), 479– 489. Janz, N., & Nylin, S. (1998). Butterflies and Plants: A Phylogenetic Study. Evolution, 52(2), 486– 502. Jonasson, S. (1988). Evaluation of the Point Intercept Method for the Estimation of Plant Biomass. Oikos, 52(1), 101–106. Karban, R., Agrawal, A. A., Thaler, J. S., & Adler, L. S. (1999). Induced plant responses and information content about risk of herbivory. Trends in Ecology & Evolution, 14(11), 443– 447.

- 136 -

Kembel, S. W., Cowan, P. D., Helmus, M. R., Cornwell, W. K., Morlon, H., Ackerly, D. D., … Webb, C. O. (2010). Picante: R tools for integrating phylogenies and ecology. Bioinformatics, 26(11), 1463–1464. Kergunteuil, A., Descombes, P., Glauser, G., Pellissier, L., & Rasmann, S. (2018). Plant physical and chemical defence variation along elevation gradients: a functional trait-based approach. Oecologia, 1–11. Kessler, A., & Baldwin, I. T. (2001). Defensive Function of Herbivore-Induced Plant Volatile Emissions in Nature. Science, 291(5511), 2141–2144. Kuhl, C., Tautenhahn, R., Böttcher, C., Larson, T. R., & Neumann, S. (2012). CAMERA: An Integrated Strategy for Compound Spectra Extraction and Annotation of Liquid Chromatography/Mass Spectrometry Data Sets. Analytical Chemistry, 84(1), 283–289. Kursar, T. A., Dexter, K. G., Lokvam, J., Pennington, R. T., Richardson, J. E., Weber, M. G., … Coley, P. D. (2009). The evolution of antiherbivore defenses and their contribution to species coexistence in the tropical tree genus Inga. Proceedings of the National Academy of Sciences, 106(43), 18073–18078. Lemoine, N. P., Burkepile, D. E., & Parker, J. D. (2014). Variable effects of temperature on insect herbivory. PeerJ, 2, e376. Lemoine, N. P., Drews, W. A., Burkepile, D. E., & Parker, J. D. (2013). Increased temperature alters feeding behavior of a generalist herbivore. Oikos, 122(12), 1669–1678. Lemoine, N. P., & Shantz, A. A. (2016). Increased temperature causes protein limitation by reducing the efficiency of nitrogen digestion in the ectothermic herbivore Spodoptera exigua. Physiological Entomology, 41(2), 143–151. Loranger, J., Meyer, S. T., Shipley, B., Kattge, J., Loranger, H., Roscher, C., & Weisser, W. W. (2012). Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology, 93(12), 2674–2682. Marion, G. M., Henry, G. H. R., Freckman, D. W., Johnstone, J., Jones, G., Jones, M. H., … Virginia, R. A. (1997). Open-top designs for manipulating field temperature in high-latitude ecosystems. Global Change Biology, 3(S1), 20–32. Massey, F. P., Ennos, A. R., & Hartley, S. E. (2006). Silica in grasses as a defence against insect herbivores: contrasting effects on folivores and a phloem feeder. Journal of Animal Ecology, 75(2), 595–603. Massey, F. P., & Hartley, S. E. (2009). Physical defences wear you down: progressive and irreversible impacts of silica on insect herbivores. Journal of Animal Ecology, 78(1), 281– 291.

- 137 -

Mattson, W. J. (1980). Herbivory in Relation to Plant Nitrogen Content. Annual Review of Ecology and Systematics, 11, 119–161. Melillo, J. M., Steudler, P. A., Aber, J. D., Newkirk, K., Lux, H., Bowles, F. P., … Morrisseau, S. (2002). Soil Warming and Carbon-Cycle Feedbacks to the Climate System. Science, 298(5601), 2173–2176. Mithöfer, A., & Boland, W. (2012). Plant defense against herbivores: chemical aspects. Annual Review of Plant Biology, 63, 431–450. Mueller-Dombois, D., & Ellenberg, H. (1974). Aims and methods of vegetation ecology. New York: Wiley O’Connor, M. I. (2009). Warming strengthens an herbivore–plant interaction. Ecology, 90(2), 388– 398. Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M. H. H., Oksanen, M. J., & Suggests, M. (2007). The vegan package. Community Ecology Package, 10, 631–637. Olff, H., Vera, F. W. M., Bokdam, J., Bakker, E. S., Gleichman, J. M., Maeyer, K. de, & Smit, R. (1999). Shifting Mosaics in Grazed Woodlands Driven by the Alternation of Plant Facilitation and Competition. Plant Biology, 1(2), 127–137. Peeters, P. J., Sanson, G., & Read, J. (2007). Leaf biomechanical properties and the densities of herbivorous insect guilds. Functional Ecology, 21(2), 246–255. Pellissier, L., Descombes, P., Hagen, O., Chalmandrier, L., Glauser, G., Kergunteuil, A., … Rasmann, S. (2018). Growth-competition-herbivore resistance trade-offs and the responses of alpine plant communities to climate change. Functional Ecology, 0(0). Pellissier, L., Fiedler, K., Ndribe, C., Dubuis, A., Pradervand, J.-N., Guisan, A., & Rasmann, S. (2012). Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecology and Evolution, 2(8), 1818–1825. Pellissier, L., Moreira, X., Danner, H., Serrano, M., Salamin, N., Dam, N. M. van, & Rasmann, S. (2016). The simultaneous inducibility of phytochemicals related to plant direct and indirect defences against herbivores is stronger at low elevation. Journal of Ecology, 104(4), 1116– 1125. Pellissier, L., Ndiribe, C., Dubuis, A., Pradervand, J.-N., Salamin, N., Guisan, A., & Rasmann, S. (2013). Turnover of plant lineages shapes herbivore phylogenetic beta diversity along ecological gradients. Ecology Letters, 16(5), 600–608. Pérez-Harguindeguy, N., Diaz, S., Garnier, E., Lavorel, S., Poorter, H., Jaureguiberry, P., … Gurvich, D. E. (2013). New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany, 61(3), 167–234.

- 138 -

Pérez‐Harguindeguy, N., Díaz, S., Vendramini, F., Cornelissen, J. H. C., Gurvich, D. E., & Cabido, M. (2003). Leaf traits and herbivore selection in the field and in cafeteria experiments. Austral Ecology, 28(6), 642–650. R Development Core Team. (2014). R: A language and environment for statistical computing. Vienna, Austria. Rasmann, S., & Agrawal, A. A. (2011). Evolution of Specialization: A Phylogenetic Study of Host Range in the Red Milkweed Beetle (Tetraopes tetraophthalmus). The American Naturalist, 177(6), 728–737. Rasmann, S., & Turlings, T. C. J. (2007). Simultaneous feeding by aboveground and belowground herbivores attenuates plant-mediated attraction of their respective natural enemies. Ecology Letters, 10(10), 926–936. Rhoades, D. F. (1979). Evolution of plant chemical defence against herbivores. In ‘Herbivores: their interaction with secondary plant metabolites’.(Eds GA Rosenthal, DH Janzen) pp. 3–54. Academic Press: London. Richards, L. A., Dyer, L. A., Forister, M. L., Smilanich, A. M., Dodson, C. D., Leonard, M. D., & Jeffrey, C. S. (2015). Phytochemical diversity drives plant–insect community diversity. Proceedings of the National Academy of Sciences, 112(35), 10973–10978. Richardson, S. J., Press, M. C., Parsons, A. N., & Hartley, S. E. (2002). How do nutrients and warming impact on plant communities and their insect herbivores? A 9-year study from a sub- Arctic heath. Journal of Ecology, 90(3), 544–556. Rønsted, N., Symonds, M. R. E., Birkholm, T., Christensen, S. B., Meerow, A. W., Molander, M., … Jäger, A. K. (2012). Can phylogeny predict chemical diversity and potential medicinal activity of plants? A case study of amaryllidaceae. BMC Evolutionary Biology, 12, 182. 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(8), 726–736. Salazar, Diego, Lokvam, J., Mesones, I., Vásquez, M. P., Ayarza, J. Z., De, P. V., & Fine, P. V. A. (2018). Origin and maintenance of chemical diversity in a species-rich tropical tree lineage. Nature Ecology & Evolution, 2, 983–990. Salazar, Diego, Jaramillo, M. A., & Marquis, R. J. (2016). Chemical similarity and local community assembly in the species rich tropical genus Piper. Ecology, 97(11), 3176–3183. Sanson, G., Read, J., Aranwela, N., Clissold, F., & Peeters, P. (2001). Measurement of leaf biomechanical properties in studies of herbivory: Opportunities, problems and procedures. Austral Ecology, 26(5), 535–546.

- 139 -

Santamaría, L., & Rodríguez-Gironés, M. A. (2007). Linkage Rules for Plant–Pollinator Networks: Trait Complementarity or Exploitation Barriers? PLOS Biology, 5(2), e31. Schädler, M., Roeder, M., Brandl, R., & Matthies, D. (2007). Interacting effects of elevated CO2, nutrient availability and plant species on a generalist invertebrate herbivore. Global Change Biology, 13(5), 1005–1015. Schoonhoven, L. M., Van Loon, B., van Loon, J. J., & Dicke, M. (2005). Insect-plant biology. Oxford University Press on Demand. Singh, A., & Agarwal, R. A. (1988). Possibility of using latex of euphorbiales for snail control. Science of The Total Environment, 77(2), 231–236. Stamp, N. E., & Yang, Y. (1996). Response of Insect Herbivores to Multiple Allelochemicals Under Different Thermal Regimes. Ecology, 77(4), 1088–1102. Tibshirani, R. (1996). Regression Shrinkage and Selection via the Lasso. Journal of the Royal Statistical Society. Series B (Methodological), 58(1), 267–288. Totland, Ø. (1999). Effects of temperature on performance and phenotypic selection on plant traits in alpine Ranunculus acris. Oecologia, 120(2), 242–251. Turlings, T. C. J., Tumlinson, J. H., & Lewis, W. J. (1990). Exploitation of Herbivore-Induced Plant Odors by Host-Seeking Parasitic Wasps. Science, 250(4985), 1251–1253. Vaieretti, M. V., Díaz, S., Vile, D., & Garnier, E. (2007). Two Measurement Methods of Leaf Dry Matter Content Produce Similar Results in a Broad Range of Species. Annals of Botany, 99(5), 955–958. Veteli, T. O., Kuokkanen, K., Julkunen‐Tiitto, R., Roininen, H., & Tahvanainen, J. (2002). Effects of elevated CO2 and temperature on plant growth and herbivore defensive chemistry. Global Change Biology, 8(12), 1240–1252. Villar, R., & Merino, J. (2001). Comparison of leaf construction costs in woody species with differing leaf life-spans in contrasting ecosystems. New Phytologist, 151(1), 213–226. Wiens, J. J., & Graham, C. H. (2005). Niche conservatism: integrating evolution, ecology, and conservation biology. Annual Review of Ecology, Evolution, and Systematics; Palo Alto, 36, 519–539. Wink, M. (2003). Evolution of secondary metabolites from an ecological and molecular phylogenetic perspective. Phytochemistry, 64(1), 3–19. Zvereva, E. L., & Kozlov, M. V. (2006). Consequences of simultaneous elevation of carbon dioxide and temperature for plant–herbivore interactions: a metaanalysis. Global Change Biology, 12(1), 27–41. .

- 140 -

Supplementary materials

Table S1. Spearman correlation between plant functional trait variables used in the natural herbivory analyses. SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, Chemical richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chemical diversity = chemical diversity based on the abundance of individual peaks per species, Chemical sum = chemical sum based on the abundance of individual peaks per species , CA Axis 1-4 = four first axes of the correspondence analyses of the untargeted metabolomics analyses, Total biomass =

biomass availability of the plant species on the field site.

SLA LDMC Toughness C:N Height Chemical richness Chemical diversity Chemical sum 1 Axis CA 2 Axis CA 3 Axis CA 4 Axis CA Silica Totalbiomass SLA 1.00 -0.45 -0.16 -0.46 -0.10 0.03 0.15 0.00 -0.01 0.01 -0.06 -0.22 -0.11 -0.39 LDMC -0.45 1.00 0.64 0.47 0.36 -0.16 -0.19 -0.16 -0.18 0.02 0.13 0.26 0.47 0.42 Toughness -0.16 0.64 1.00 0.52 0.26 -0.23 -0.27 -0.24 -0.21 -0.05 0.03 0.55 0.52 0.27 C:N -0.46 0.47 0.52 1.00 0.03 -0.31 -0.19 -0.25 0.05 0.09 -0.08 0.55 0.38 0.33 Height -0.10 0.36 0.26 0.03 1.00 -0.36 -0.10 -0.35 -0.11 0.23 0.00 0.21 0.22 0.35 Chemical richness 0.03 -0.16 -0.23 -0.31 -0.36 1.00 0.17 0.78 -0.38 -0.72 0.45 -0.17 -0.09 -0.32 Chemical diversity 0.15 -0.19 -0.27 -0.19 -0.10 0.17 1.00 -0.23 0.12 0.17 0.09 -0.42 -0.12 0.03 Chemical sum 0.00 -0.16 -0.24 -0.25 -0.35 0.78 -0.23 1.00 -0.34 -0.71 0.34 -0.15 -0.19 -0.28 CA Axis 1 -0.01 -0.18 -0.21 0.05 -0.11 -0.38 0.12 -0.34 1.00 0.64 -0.63 -0.25 -0.25 0.02 CA Axis 2 0.01 0.02 -0.05 0.09 0.23 -0.72 0.17 -0.71 0.64 1.00 -0.53 -0.06 -0.03 0.27 CA Axis 3 -0.06 0.13 0.03 -0.08 0.00 0.45 0.09 0.34 -0.63 -0.53 1.00 0.01 0.14 -0.15 CA Axis 4 -0.22 0.26 0.55 0.55 0.21 -0.17 -0.42 -0.15 -0.25 -0.06 0.01 1.00 0.56 0.26 Silica -0.11 0.47 0.52 0.38 0.22 -0.09 -0.12 -0.19 -0.25 -0.03 0.14 0.56 1.00 0.19 Total biomass -0.39 0.42 0.27 0.33 0.35 -0.32 0.03 -0.28 0.02 0.27 -0.15 0.26 0.19 1.00

- 141 -

Table S2. Spearman correlation between plant functional trait variables used in bioassay experiment analysis. SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, Chemical richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chemical diversity = chemical diversity based on the abundance of individual peaks per species, Chemical sum = chemical sum based on the abundance of individual peaks per species , CA Axis 1-4 = four

first axes of the correspondence analyses of the untargeted metabolomics analyses.

richness

SLA LDMC Toughness C:N Height Chemical diversity Chemical sum Chemical 1 Axis CA 2 Axis CA 3 Axis CA 4 Axis CA Silica SLA 1.00 -0.38 -0.10 -0.39 -0.03 -0.01 0.09 -0.05 0.02 0.10 -0.05 -0.05 -0.01 LDMC -0.38 1.00 0.62 0.34 0.24 -0.11 -0.07 -0.09 -0.10 0.05 0.09 0.07 0.33 Toughness -0.10 0.62 1.00 0.43 0.24 -0.23 -0.20 -0.21 -0.22 -0.02 -0.02 0.39 0.36 C:N -0.39 0.34 0.43 1.00 -0.03 -0.31 -0.15 -0.25 0.15 0.13 -0.12 0.41 0.25 Height -0.03 0.24 0.24 -0.03 1.00 -0.28 -0.15 -0.25 -0.16 0.14 -0.05 0.14 0.14 Chemical richness -0.01 -0.11 -0.23 -0.31 -0.28 1.00 0.11 0.78 -0.29 -0.65 0.42 -0.14 -0.02 Chemical diversity 0.09 -0.07 -0.20 -0.15 -0.15 0.11 1.00 -0.27 0.10 0.19 0.01 -0.35 -0.07 Chemical sum -0.05 -0.09 -0.21 -0.25 -0.25 0.78 -0.27 1.00 -0.28 -0.68 0.36 -0.19 -0.11 CA Axis 1 0.02 -0.10 -0.22 0.15 -0.16 -0.29 0.10 -0.28 1.00 0.61 -0.51 -0.18 -0.23 CA Axis 2 0.10 0.05 -0.02 0.13 0.14 -0.65 0.19 -0.68 0.61 1.00 -0.45 0.00 -0.04 CA Axis 3 -0.05 0.09 -0.02 -0.12 -0.05 0.42 0.01 0.36 -0.51 -0.45 1.00 -0.01 0.22 CA Axis 4 -0.05 0.07 0.39 0.41 0.14 -0.14 -0.35 -0.19 -0.18 0.00 -0.01 1.00 0.46 Silica -0.01 0.33 0.36 0.25 0.14 -0.02 -0.07 -0.11 -0.23 -0.04 0.22 0.46 1.00

- 142 -

Table S3. Spearman correlation between plant functional trait variables used in the analyses investigating the effect of warming on plant traits. LA = leaf area, SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, C:N = leaf carbon to nitrogen ratio, Chemical richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chemical diversity = chemical diversity based on the abundance of individual peaks per species, Chemical sum = chemical sum based on the abundance

of individual peaks per species.

LA SLA LDMC Toughness C:N Chemical richness Chemical diversity Chemical sum LA 1.00 0.16 -0.38 0.03 0.03 0.28 0.22 0.05 SLA 0.16 1.00 -0.45 -0.10 -0.38 0.05 0.22 -0.33 LDMC -0.38 -0.45 1.00 0.70 0.54 -0.71 -0.50 -0.34 Toughness 0.03 -0.10 0.70 1.00 0.54 -0.56 -0.44 -0.30 C:N 0.03 -0.38 0.54 0.54 1.00 -0.34 -0.19 -0.01 Chemical richness 0.28 0.05 -0.71 -0.56 -0.34 1.00 0.44 0.66 Chemical diversity 0.22 0.22 -0.50 -0.44 -0.19 0.44 1.00 -0.13 Chemical sum 0.05 -0.33 -0.34 -0.30 -0.01 0.66 -0.13 1.00

- 143 -

Table S4. Spearman correlation between plant functional trait variables used in the analyses investigating the effect of warming on plant traits and plant palatability. LA = leaf area, SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, C:N = leaf carbon to nitrogen ratio, Chemical richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chemical diversity = chemical diversity based on the abundance of individual peaks per species, Chemical sum =

chemical sum based on the abundance of individual peaks per species.

LA SLA LDMC Toughness C:N Chemical richness Chemical diversity Chemical sum LA 1.00 0.33 -0.05 0.44 0.41 -0.27 -0.19 -0.27 SLA 0.33 1.00 -0.74 -0.29 -0.19 0.32 0.52 -0.13 LDMC -0.05 -0.74 1.00 0.58 0.43 -0.36 -0.63 -0.16 Toughness 0.44 -0.29 0.58 1.00 0.62 -0.40 -0.62 -0.18 C:N 0.41 -0.19 0.43 0.62 1.00 -0.35 -0.54 0.22 Chemical richness -0.27 0.32 -0.36 -0.40 -0.35 1.00 0.75 0.21 Chemical diversity -0.19 0.52 -0.63 -0.62 -0.54 0.75 1.00 0.02 Chemical sum -0.27 -0.13 -0.16 -0.18 0.22 0.21 0.02 1.00

- 144 -

Table S5. Phylogenetic signal in plant functional traits assessed with Blomberg's K statistic. Note that for this analyses plant functional traits were averaged for each plant species across sites. Significant variables are highlighted in bold. N = number of plant species used to assess phylogenetic signal, LA = leaf area, SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, Chemical richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chemical diversity = chemical diversity based on the abundance of individual peaks per species, Chemical sum = chemical sum based on the abundance of individual peaks per species, CA Axis 1- 4 = four first axes of the correspondence analyses of the untargeted metabolomics analyses, * p- value < 0.05,** p-value < 0.01, *** p-value < 0.001.

Traits N K Z‐score P‐value LA 132 0.119 -0.938 0.127 SLA 132 0.088 -0.425 0.368 LDMC 132 0.347 -5.274 0.001 ** Toughness 132 0.433 -3.557 0.001 ** C:N 132 0.188 -2.437 0.001 ** Height 132 0.053 1.902 0.951 Silica 132 0.142 -1.068 0.059 . Chemical richness 132 0.258 -3.558 0.001 ** Chemical diversity 132 0.190 -3.029 0.001 ** Chemical sum 132 0.238 -3.365 0.001 ** CA Axis 1 132 0.565 -4.068 0.001 ** CA Axis 2 132 0.349 -4.396 0.001 ** CA Axis 3 132 0.641 -3.435 0.001 ** CA Axis 4 132 0.893 -5.381 0.001 **

- 145 -

Table S6. Changes in plant palatability between the warming treatment and the ambient control associated to plant traits and assessed using a MCMCglmm. *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1.

post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) 0.337 -0.078 0.705 19700 0.075 . Warming 0.005 -0.089 0.097 19700 0.907 LA -0.024 -0.215 0.155 19700 0.785 SLA 0.014 -0.164 0.203 19700 0.873 LDMC -0.031 -0.196 0.143 19700 0.704 Toughness -0.080 -0.286 0.126 17780 0.446 C:N -0.068 -0.189 0.053 19700 0.259 Chemical richness 0.076 -0.077 0.228 19700 0.320 Chemical diversity -0.095 -0.265 0.076 19212 0.265 Chemical sum -0.029 -0.159 0.094 19082 0.643 Warming : LA 0.022 -0.161 0.206 19700 0.801 Warming : SLA -0.061 -0.259 0.146 19700 0.534 Warming : LDMC 0.009 -0.151 0.162 19700 0.904 Warming : Toughness 0.008 -0.162 0.174 19700 0.941 Warming : C:N 0.011 -0.134 0.148 19700 0.867 Warming : Chemical richness -0.017 -0.157 0.125 19700 0.799 Warming : Chemical diversity 0.042 -0.116 0.208 19700 0.601 Warming : Chemical sum 0.017 -0.108 0.149 19700 0.781

- 146 -

Figure S1. (a) The three warming meadows, indicated by the red points, are located in the Swiss- Alps at 1800 m, 2070 m and 2270 m a.s.l. (Chablais region, State of Vaud). (b) In the warming meadows, open-top chambers (OTC) warmed eight plots directly after snowmelt and until the first snowfalls between 2014 and 2017. Photo credit: P. Descombes.

- 147 -

Figure S2. Metabolomics correspondence analysis representing the major sources of variability among plant species based on the full binarized metabolite profile. Species scores are represented along (a) axes 1 and 2, (b) axes 1 and 3, (c) and axes 1 and 4. Species belonging to the same plant family are grouped into the ellipses. The red, blue and grey colours represent plant species from the Liliopsida (monocotyledons), the Magnoliopsida (dicotyledons) and the Filicopsida clade (ferns), respectively.

- 148 -

Figure S3. Principal component analysis (PCA) representing major sources of variability among plant species based on the 12 functional traits measured on the 3 study sites. Species scores are represented along axes 1 and 2, and grouped into ellipses according to their affiliation to plant families. The lengths of the vectors in the correlation circle represent the magnitude of the correlation between the variables and the axes. SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, ChemRich = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, ChemDiv = chemical diversity based on the abundance of individual peaks per species, CA Axis 1- 4 = four first axes of the correspondence analyses of the untargeted metabolomics analyses.

- 149 -

Figure S4. Principal component analysis (PCA) representing major sources of variability among plant species based on the 12 functional traits and available biomass (i.e. Plant biomass) measured on plots of the three study sites. Species scores are represented along axes 1 and 2, and grouped into ellipses according to their affiliation to plant families. The size of the points represents the estimated dry leaf mass eaten by a natural community of herbivores in plots of the 3 study sites, and reflects the feeding preferences of a natural community of herbivores. The lengths of the vectors in the correlation circle represent the magnitude of the correlation between the variables and the axes. SLA = leaf mass per area, LDMC = leaf dry matter content, Toughness = leaf penetration force, Silica = amount of silica per mg dry leaf tissue, C:N = leaf carbon to nitrogen ratio, Height = plant height, Plant biomass = available dry biomass of the plant in plots, ChemRich = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, ChemDiv = chemical diversity based on the abundance of individual peaks per species, CA Axis 1-4 = four first axes of the correspondence analyses of the untargeted metabolomics analyses.

- 150 -

Figure S5. SLA plant trait of the 16 plant species investigated under ambient control and temperature warming. The barplot represent the mean ± sd. The four last species (i.e. Carex montana, Carex sempervirens and Sesleria caerulea) belong to the Liliopsida plant clade. Sesleria caerulea is represented two times because this species was collected on two different sites. All other plant species belong to the Magnoliopsida plant clade.

- 151 -

Figure S6. Chemical richness plant trait of the 16 plant species investigated under ambient control and temperature warming. The barplot represent the mean ± sd. The four last species (i.e. Carex montana, Carex sempervirens and Sesleria caerulea) belong to the Liliopsida plant clade. Sesleria caerulea is represented two times because this species was collected on two different sites. All other plant species belong to the Magnoliopsida plant clade.

- 152 -

Figure S7. Plant palatability of the 10 plant species investigated under ambient control and temperature warming. The barplot represent the mean ± sd. The three last species (i.e. Carex montana, Carex sempervirens and Sesleria caerulea) belong to the Liliopsida plant clade. All other plant species belong to the Magnoliopsida plant clade.

- 153 -

- 154 -

CHAPTER 4

Trophic conservatism predicts alpine plants responses to herbivore ecosystem incursion

Patrice Descombes1,2, Camille Pitteloud1,2, Gaëtan Glauser3, Emilien Jolidon3, Alan Kergunteuil3, Sergio Rasmann3, Loïc Pellissier1,2

1 Landscape Ecology, Institute of Terrestrial Ecosystems, ETH Zürich, CH-8092 Zürich, Switzerland 2 Swiss Federal Research Institute WSL, CH-8903 Birmensdorf, Switzerland 3 Laboratory of Functional Ecology, Institute of Biology, University of Neuchâtel, CH-2000 Neuchâtel, Switzerland

In preparation (target: Science)

- 155 -

Climate change promotes asynchronous range shifts of species across trophic compartments potentially reshaping ecosystem structures and functions. While, herbivore abundance and plant defences show a coupled decline along elevation gradient, the current equilibrium of ecological systems could be disrupted with animals upwardly tracking climate change faster than plants. Here, we experimentally simulated climate-driven ecosystem incursion of herbivores from lower elevation on alpine plant communities. We report that herbivore trophic conservatism predicts the response of alpine plant communities to novel herbivore incursion. By feeding preferentially on alpine plants with functional traits and metabolomics profiles matching those explaining their original trophic interactions, herbivore colonizers reduced light competition and plant biomass, but increased small stature plant species and community diversity. Contrasted with a direct warming treatment using greenhouses, we showed a stronger effect of increased herbivory than temperature on plant communities. Indirect biotic responses to climate change through climate-driven herbivore incursion might represent a stronger driver of ecosystem modification than temperature.

Herbivores are conduits of energy and matter between compartments of ecosystems, and regulate plant biomass through their trophic preferences (1). Shifts in the prevalence of herbivores along climatic clines are associated to a decrease in plant defences, underlining the strong top-down effect of herbivores on plant traits and ecosystem functioning (2, 3). Climate change is expected to modify the current equilibrium of trophic interactions between plant and herbivore along climatic gradients (4). A higher metabolic sensitivity to thermal changes and greater ability to disperse allow insect herbivores to track climate change along elevation or latitudes, faster than plants which lag behind (5–7). In this regard, a climate-driven range shift of herbivores into the alpine ecosystem may generate new opportunities to feed on poorly defended plants and could modify the structure and functions of plant communities (8, 9). So far, climate change experiments were primarily focused on abiotic effects (10–14), or on altered interactions between trophic groups within closed systems (15, 16), but did not include novel arising interactions between resident and range shifting species (but see 17, 18). Range-shifting herbivores will encounter new plants species, whose functional traits should determine the outcome of novel interaction networks (19, 20). Including direct and indirect ecological responses to climate change, via ecosystem incursion, should provide a more complete view of the effects of climate change on biodiversity.

Including the consequences of lowland insect herbivore colonizing alpine plant communities in experiments addresses the indirect effects of climate change on biodiversity. In the Alps, using niche-based models, we predict an average shift of a dominant clade of insect herbivores of 490 m by the year 2080 (ΔT= + 3.56 K; mean ± sd = 489.7 m ± 312.9 m; Table S1), an increase in

- 156 -

Figure 1. Forecasted changes in herbivore and plant communities under climate change in the study area. (A) The herbivore incursion and warming experiments were performed on species rich alpine grasslands above the treeline in Switzerland. Photo credit: P. Descombes, ETH Zürich, Switzerland. (B) The three experimental sites are located in the western Swiss Alps (black polygon on the Swiss map) at 1800 m, 2070 m and 2270 m (black circles). The black triangle corresponds to the orthopteran collection site (1400 m). The map represents the changes in species richness simulated from habitat distribution models between current time and 2080 under a business as usual scenario (RCP 8.5; mean global warming increase of 3.7 K). (C, D) Relative increase (ratio > 1) or decrease (ratio < 1) in (C) herbivore richness and (D) herbivore abundance between future (i.e. 2080 period, scenario RCP 8.5) and current climate conditions in the study region. The curves were fitted with a linear regression with polynomial terms and the grey area represents the 95% prediction interval. (E) Changes in total aboveground plant biomass (n=24), expressed as the number of plant hits per 36 pins in 50 cm x 50 cm quadrats, under herbivore incursion (Herbivore) and warming (Warming) treatments. Herbivore and warming treatments were compared to their associated controls, the herbivore- free cage (Cage control) and the ambient control, respectively. (F) Changes in total plant species richness under herbivore incursion and warming treatments (n=24). (G) Changes in number of species lost (hatched background) and gained (full background) under herbivore incursion and warming treatments (n=24). Linear mixed effect models: *** P < 0.001, ** P < 0.01, * P < 0.05.

- 157 -

herbivore richness (Fig. 1b-c) and more than a tripling in herbivore abundance at higher elevation above the treeline (i.e. 1800-1900 m; Fig. 1d). By experimentally simulating expected range shifts of herbivores, we assessed the impact of incursion on three alpine plant communities in the Swiss Alps (i.e. 1800, 2070 and 2270 m; Fig. 1 and Fig. S1). We translocated into cages of 70 x 70 cm a representative density of orthopteran herbivores from lower elevation (i.e. 1400 m) and measured the response of alpine plant communities in relation to a set of response traits, including leaf physical (SLA, LDMC, toughness) and chemical properties (leaf nitrogen content, plant metabolomics). We contrasted changes in the biotic pressure to a direct effect of temperature by modifying in situ abiotic conditions with open-top chamber (OTC) greenhouses (Fig. S2 and S3). We further assessed insect feeding preferences using two complementary approaches: (i) a cafeteria choice experiment and (ii) the reconstruction of food-webs using DNA meta-barcoding techniques applied on insect faeces. In response to reduced herbivory and abiotic factors, alpine plants are generally more palatable than low elevation plants (21, 22). We expect that herbivores colonizing alpine communities to be released from previous trophic constraints by encountering less-defended, and more palatable alpine plants.

Here, we show that trophic conservatism during herbivores’ upward range shifts predicts functional changes in alpine plant communities. The innate feeding preferences on tougher leaves in a cafeteria experiment (MCMCglmm: slope = 1.095, P = 0.004; Table S2, Fig. 2a) and Liliopsida plant species in DNA meta-barcoding (glm: slope = 4.218, P < 0.001; Table S3, Fig. 2b) was conserved when feeding on novel alpine plant species (Spearman correlation: cafeteria, r = 0.36, P < 0.001; DNA meta-barcoding, r = 0.20, P < 0.001; Fig. 2) and matches the response traits of alpine plants species in the field experiment (Fig. 3; Table S4). Herbivore trophic preferences were conserved after translocation and predicted the impact on the biomass of alpine plant species. In particular, plants with high leaf toughness (MCMCglmm: slope = -4.53, P < 0.001; Table S4) and low SLA (slope = 1.86, P = 0.018; Table S4), predominantly monocotyledon (i.e. CA axis 4: slope = -3.58, P < 0.001; Fig. S4, Table S4), decreased in biomass with incoming herbivore (e.g. Carex sempervirens, Sesleria cearulea and Festuca rubra, Fig. 3) and matches the effect of herbivore biomass removal in the communities (Spearman correlation: r = -0.38, P < 0.001). Emphasising the significance of novel interactions under herbivore range shifts, colonizing herbivores showed very different feeding behaviours (82 ± 8 % leaf attacks on tougher monocotyledon; Table S5, Fig. S5) compared to the rare high elevation herbivores on the recipient sites (74 ± 16 % leaf attacks on dicotyledons; Table S5, Fig. S5). Instead of switching to opportunistic feeding on poorly defended alpine plants, colonizing orthopteran herbivores maintained their innate trophic preference toward alpine species presenting leaf functional traits similar to their lowland habits. Hence, novel, but

- 158 - functionally conserved, trophic interactions between alpine plants and lowland colonizing herbivores mediates a distinct top-down effect on the structure of alpine plant communities.

Figure 2. Diet shifts of lowland orthopteran herbivores on alpine plant species. (A) Feeding patterns of lowland orthopteran species on a pool of low (red) or high (blue) elevation plant species assessed with a cafeteria feeding choice experiment. Pools of low and high elevation plant species were presented separately to the herbivores and correspond to 16 congeneric or closely related plant pairs typical of low or high elevation. The size of the circle is proportional to the amount of biomass eaten and correspond to the mean value across replicated feeding sessions. (B) Feeding patterns of lowland orthopteran species assessed by reconstructing the food-web using DNA meta-barcoding techniques applied on insect faeces collected on the source site and in the herbivore cages installed on the alpine plant communities (Fig. 1). The size of the circle is proportional to the number of reads detected in the faeces with DNA meta-barcoding, which is a semi-quantitative approximation of the ingested biomass by the herbivore. Blue circles correspond to novel species interactions when herbivores are translocated on high elevation grasslands. Red circles correspond to lowland interactions on the source site or similar interactions occurring on the source site and at higher elevation (same plant species). Note that values were normalized for each orthopteran species.

- 159 -

Herbivore incursion in alpine system impacted the structure of plant communities over the observed period (i.e. 2014-2017) and compared to the herbivore-free control cages by (i) increasing plant species richness by 8.3% on average (linear mixed-effects model: slope = 2.25, P = 0.004; initial species richness: mean = 27.2; Fig. 1f), (ii) increasing the compositional dissimilarity (slope = 0.11, P < 0.001) and (iii) decreasing the functional divergence of the community (slope = -0.02, P = 0.005). The increase in species richness in herbivore plots was driven by species gains (linear mixed-effects model: slope = 1.71, P = 0.007; Fig. 1g), rather than lower species losses (slope = - 0.54, P = 0.16; Fig. 1g), when compared to herbivore-free control cages. Herbivory incursion decreased the overall plant community biomass (linear mixed-effects model: slope = -139.9, P < 0.001; Fig. 1e) by feeding preferentially on the dominant plant (Spearman correlation: r = 0.64, P < 0.001) which are mainly tough monocotyledon species (Fig. 3). Increased light availability and reduced competition possibly favored a higher species richness by promoting the establishment and coexistence of ground level-growing species (15, 16) (e.g. Myosotis alpestris + 62.5%, Viola calcarata + 62.5% or Gentiana nivalis + 37.5 %). In contrast, the OTC greenhouses did not significantly impact the plant community composition and structure (linear mixed-effects model: all metrics P > 0.1; Fig. 1e-g) and showed a reduction in biomass to a significant lower degree compared to the herbivore incursion treatment (linear mixed-effects model: slope = -48.63, P = 0.02; Fig. 1e, Table S10). Considering that the herbivore incursion treatment mimicked realistic forecasted herbivore densities, our results indicate that novel trophic interactions under climate- driven range shifts might represent a stronger driver of ecosystem modification than temperature.

Because plant phenotype may change under increased temperature, thereby influencing plant- herbivore interactions (23, 24), we investigated changes in leaf chemistry and physical properties in the OTC greenhouses. On average, we found a 9.3 % increase in SLA (MCMCglmm: slope = 1.01, P = 0.03; Table S6) and a 1.1 % decrease in chemical richness (slope = -1.18, P = 0.02; Table S6) in the warming treatment compared to the ambient control, but no significant changes in other leaf traits such as leaf nutritional value or leaf toughness (Table S6). While we documented a significant change in the leaf chemistry (adonis test: R2 = 0.004, P = 0.001; Table S7), 98.4 % of the metabolomics peaks for monocotyledon and 97.3% for dicotyledon remained unchanged on average under the warming treatment (number of metabolite per species: mean ± sd = 4588 ± 127; Table S8). Changes were driven by a limited number of metabolomics peaks including flavonoids, terpenoids and phospholipide diester increasing or decreasing their expression (Table S9). Phospholipids diester are involved in membrane remodelling (i.e. glycerophosphocholine (25); Table S9), while terpenoids and flavonoids metabolites are involved in many basic functions in plant development and protection against stresses (i.e. Kaempferol (26, 27); Table S9), for UV

- 160 - protection, pigmentation and freezing tolerance (28). Shifts of the expression levels of those metabolite under the warming treatment can be associated to altered abiotic stress. Moreover, because the trophic preferences of orthopteran herbivores was predominantly driven by leaf toughness (Fig. 3, Table S4, Table S2), which does not significantly change with temperature warming (Table S6), temperature increase is likely to have an only marginal effect on the response of alpine plant communities to generalist orthopteran herbivores.

Figure 3. Treatment effect on aboveground alpine plant biomass related to plant functional traits. (A) Distribution of functional trait values across the plant phylogeny. Functional traits are classified following their effect size and association to biomass changes under the herbivore incursion treatment (Table S4), and highlighted in bold if their contribution was significant in the model. Trait values were averaged across the three field sites, log+1 transformed and normalized. The colour scale represents the size of the trait value. (B-C) Change in alpine plant biomass, expressed as the difference in total number of plant hits between 2017 and 2014 inventories in 50 x 50 cm quadrats, under (B) herbivore incursion (Herbivore) and (C) warming (Warming) treatments. Herbivore and warming

- 161 -

treatments were compared to their associated controls, the herbivore-free cage and the ambient control, respectively. The polygon area represents the distribution of the quadrats measurements around the mean (mean ± sd) in the treatment and control. Only plant species with at least 4 replicates are presented in this figure. Significant changes in plant species biomass between the treatment and the associated control are highlighted by the stars. The black line represents the net effect of the treatment on the plant biomass, and was calculated as the difference between the mean biomass change in the treatment and control. SLA = surface leaf area, LDMC = leaf dry matter content, Toughness = leaf penetration force, C:N = leaf carbon to nitrogen ratio, Height = plant height, Chem. richness = chemical richness representing number of individual compound peaks obtained from the untargeted metabolomics analyses, Chem. diversity = chemical diversity based on the abundance of individual peaks per species, CA Axis 1-4 = four first component of the correspondence analyses of the untargeted metabolomics analyses, * p-value < 0.05,** p-value < 0.01, *** p-value < 0.001.

Together, our study indicates that the incursion of novel herbivores in the alpine system can represent a faster driver of plant community changes than the direct effect of climate warming. Because herbivore abundance was higher and focused on different species than native herbivory, the treatment reduced the frequency of dominant species with tougher leaves and distinct chemistry matching their trophic preferences in their habitat of origin. As a result, increased herbivory fostered the coexistence of subordinate species by enhancing light availability at the ground level, and, resulting in a net increase in species richness, especially of plant species with low stature, likely sensitive to competition (15, 16). While the colonization to higher elevation under climate change is expected to be slower for plants than for mobile animals (5, 7), the reduction of plant biomass mediated by a first wave of novel insect herbivores might favor the establishment of novel plant species by creating gaps in the extant communities (5). Therefore, we expect that herbivore incursion in the alpine system may in turn generate a positive feedback loop that will facilitate the colonisation of lowland plant competitors through enhanced establishment rates. It is however uncertain whether the behaviour of herbivore conserved under the short time scale of the experiment, remain conserved over a decadal time scale (29). Together, our result indicate that forecasting the dynamics of plant community responses to climate change should account for the effects of ecosystem incursions.

Ecosystem incursions may represent a dominant disruptive driver of ecosystem change in a warmer future. The forecast of ecosystem responses to those indirect effects of climate change requires estimating future suitable area, dispersal rates and functional impact on ecosystems of novel interactions. Here, we showed that when shifting their range, the trophic preferences of orthopteran herbivore is conserved so that their effect on plant communities is predictable. Beyond herbivore, many organisms may exert disturbance pressures on current ecosystem functioning including pollination or pathogen organisms. The novel interactions from this variety of incursions

- 162 - may not always be predictable, in case of behavioural shifts or evolutionary effects. Future experiments combining climate change and the indirect effect of ecosystem incursions should evaluate whether the future novel biotic interactions are generally predictable across system allowing the forecast of incursions on ecosystem processes. Conserved ecological rules would allow the use of mechanistic predictive models of food-webs as changes in trophic interaction and herbivory intensity under climate change (30) and foreshadow large functional changes as climate warming continues.

Acknowledgements

We thank all the people who helped with field work and Rodolphe Muller from the refuge Giacomini for providing assistance and shelter during the field season. We thank Niklaus Zimmermann for providing help on the GCMs.

References

1. D. L. DeAngelis, Dynamics of Nutrient Cycling and Food Webs (Springer Science & Business Media, 2012).

2. O. J. Schmitz, Resolving Ecosystem Complexity (MPB-47) (Princeton University Press, 2010).

3. M. D. Hunter, The Phytochemical Landscape: Linking Trophic Interactions and Nutrient Dynamics (Princeton University Press, 2016).

4. W. H. van der Putten et al., Trophic interactions in a changing world. Basic and Applied Ecology. 5, 487–494 (2004).

5. J. M. Alexander et al., Lags in the response of mountain plant communities to climate change. Global Change Biology. 24, 563–579 (2018).

6. J. S. Bale et al., Herbivory in global climate change research: direct effects of rising temperature on insect herbivores. Global Change Biology. 8, 1–16 (2002).

7. M. P. Berg et al., Adapt or disperse: understanding species persistence in a changing world. Global Change Biology. 16, 587–598 (2010).

8. O. Schweiger, Settele Josef, Kudrna Otakar, Klotz Stefan, Kühn Ingolf, Climate change can cause spatial mismatch of trophically interacting species. Ecology. 89, 3472–3479 (2008).

- 163 -

9. R. M. Pateman, J. K. Hill, D. B. Roy, R. Fox, C. D. Thomas, Temperature-Dependent Alterations in Host Use Drive Rapid Range Expansion in a Butterfly. Science. 336, 1028– 1030 (2012).

10. A. M. Panetta, M. L. Stanton, J. Harte, Climate warming drives local extinction: Evidence from observation and experimentation. Science Advances. 4, eaaq1819 (2018).

11. A. M. Arft et al., Responses of Tundra Plants to Experimental Warming: meta-Analysis of the International Tundra Experiment. Ecological Monographs. 69, 491–511 (1999).

12. F. S. Chapin, G. R. Shaver, Physiological and Growth Responses of Arctic Plants to a Field Experiment Simulating Climatic Change. Ecology. 77, 822–840 (1996).

13. K. B. Suttle, M. A. Thomsen, M. E. Power, Species Interactions Reverse Grassland Responses to Changing Climate. Science. 315, 640–642 (2007).

14. J. Harte, R. Shaw, Shifting Dominance Within a Montane Vegetation Community: Results of a Climate-Warming Experiment. Science. 267, 876–880 (1995).

15. E. Kaarlejärvi, A. Eskelinen, J. Olofsson, Herbivores rescue diversity in warming tundra by modulating trait-dependent species losses and gains. Nature Communications. 8, 419 (2017).

16. E. T. Borer et al., Herbivores and nutrients control grassland plant diversity via light limitation. Nature. 508, 517–520 (2014).

17. J. M. Alexander, J. M. Diez, J. M. Levine, Novel competitors shape species’ responses to climate change. Nature. 525, 515–518 (2015).

18. H. Bruelheide, Translocation of a montane meadow to simulate the potential impact of climate change. Applied Vegetation Science. 6, 23–34 (2003).

19. J. X. Becerra, Insects on Plants: Macroevolutionary Chemical Trends in Host Use. Science. 276, 253–256 (1997).

20. I. S. Pearse, F. Altermatt, Predicting novel trophic interactions in a non-native world. Ecology Letters. 16, 1088–1094.

21. P. Descombes et al., Community-level plant palatability increases with elevation as insect herbivore abundance declines. Journal of Ecology. 105, 142–151 (2017).

22. L. Pellissier et al., Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecology and Evolution. 2, 1818–1825 (2012).

- 164 -

23. E. H. DeLucia, P. D. Nabity, J. A. Zavala, M. R. Berenbaum, Climate Change: Resetting Plant-Insect Interactions. Plant Physiology. 160, 1677–1685 (2012).

24. M. G. Bidart‐Bouzat, A. Imeh‐Nathaniel, Global Change Effects on Plant Chemical Defenses against Insect Herbivores. Journal of Integrative Plant Biology. 50, 1339–1354 (2008).

25. B. van der Rest, A.-M. Boisson, E. Gout, R. Bligny, R. Douce, Glycerophosphocholine Metabolism in Higher Plant Cells. Evidence of a New Glyceryl-Phosphodiester Phosphodiesterase. Plant Physiology. 130, 244–255 (2002).

26. D. Tholl, in Biotechnology of Isoprenoids (Springer, Cham, 2015), Advances in Biochemical Engineering Biotechnology, pp. 63–106.

27. I. Urquiaga, F. Leighton, Plant Polyphenol Antioxidants and Oxidative Stress. Biological Research. 33, 55–64 (2000).

28. D. C. Close, C. McArthur, Rethinking the role of many plant phenolics – protection from photodamage not herbivores? Oikos. 99, 166–172 (2002).

29. A. A. Agrawal, Host-Range Evolution: Adaptation and Trade-Offs in Fitness of Mites on Alternative Hosts. Ecology. 81, 500–508 (2000).

30. M. C. Urban et al., Improving the forecast for biodiversity under climate change. Science. 353, aad8466 (2016).

- 165 -

Methods

Study sites and plot selection

In 2014, we selected three alpine grasslands located above the treeline at 1800 m (average annual temperature = 3.09°C), 2070 m average annual temperature = 1.96°C) and 2270 m a.s.l. (average annual temperature = 1.00°C) in the Western Swiss Alps. These three grasslands were chosen because they are typical unmanaged (no cutting and no pasture grazing) species rich calcareous grasslands (82 plant species in total at 1800 m, 75 at 2070 m and 59 at 2270 m) and are similar in their dominant floristic composition. The three selected grasslands are typical high elevation dry calcareous grasslands (Seslerion community type) (1) characterized by a vegetation dominated by grasses (e.g. Carex sempervirens, Carex montana, Festuca laevigata, Festuca rubra aggr., Sesleria caerulea) and a very diverse cortege of forbs (e.g. Astrantia major, Gentiana verna, Helianthemum nummularium subsp grandiflorum, Homogyne alpina, Leontodon hispidus sstr, Polygonum viviparum, Potentilla crantzii, Scabiosa lucida, Soldanella alpina, Trifolium pratense sstr, Trollius europaeus; Table S11). We selected 96 plant communities of 50 cm x 50 cm plots distributed among the three sites. The plots at each sites (i.e. n = 32) were selected to be as similar as possible with respect to floristic composition, canopy structure and plant biomass. All plots were ground marked with 10 cm aluminium rolls embedded into the soil in order to relocate plots as precisely as possible for the subsequent years’ inventories.

Temporal survey of grassland communities

Plant inventories of the 96 grassland plots were performed in August 2014 and 2017 using two different methods. First, we used the point-intersect method to describe the spatial structure of the 96 plant communities (2, 3). All plots were sampled for pin contacts with 36 pins placed in an equally spaced grid pattern over each plot. A needle of 6 mm diameter and 50 cm of length (4) was vertically placed (following terrestrial gravity) through the plant community on each grid intersection and all species contacts (i.e. number of contacts) where recorded in five vertical height categories: 0-2 cm, 2-5 cm, 5-10 cm, 10-20 cm and 20-50 cm. Records were not performed during rainy and windy days. The total number of hits per species in each plot, expressed as the number of hits per 36 pins, was then used as an index of abundance for vascular plants (i.e. biomass). This non-destructive method is widely used to estimate plant abundance (5–8) and correlate well with true biomass (3) in grasslands. Finally, we visually estimated species cover according to a 7-level scale: : <1%, 1-5%, 5-13%, 13-25%, 25-50%, 50-75%, 75-100%. This method was used to ensure that rare species not recorded by the point-intercept method are also taken into account. We added

- 166 - the overlooked species by the point-intercept method and attributed them the minimal biomass index into the plot (1 hit per 36 pins). We used this species abundance matrix for further analyses, which is a good estimate of the plant biomass into the plot (3). Point-intercept and cover estimates inventories were performed on the exact same plots in 2014 and in 2017 using the same frames’ location.

Herbivore and warming treatments

We randomly allocated an herbivore and a warming treatment, accompanied by two controls (herbivore-free cage and ambient control) to the 32 plot on each site, leading to 8 replicates per site and treatment. The cages used for the herbivore treatment and the herbivore-free cage control correspond to 70 cm x 70 cm x 50 cm enclosures made from transparent insect-proof netting (2 mm x 3 mm mesh size, F1032 crystal clear DIATEX), which were carefully placed over the 50 cm x 50 cm plot, fixed to the ground with steel tent pegs and secured with tensioned ropes. An additional piece of net was fixed at the basis of the cage with 10cm into the ground, and regularly checked for permeability in order to prevent insects from escaping. The selected mesh size (i.e. 2 mm x 3 mm) is insect proof to most groups (e.g. orthopterans, caterpillars, butterflies and bees) and was chosen to reduce the potential effects of the cage on microclimatic conditions (9). Cages were set up each year in June and removed before snowfalls in October. The herbivore treatment consisted of a community of 10 orthopteran individuals (8 Cealifera and 2 Ensifera) randomly collected on a source site at lower elevation (i.e. 1400 m) and translocated inside the herbivore treatment cages on the three higher elevation sites. The orthopteran density per cage was fixed to 10 following a realistic density (i.e. 10 individuals in 0.5 m2) according to inventories performed in mountain grasslands (10–13). Orthopterans were collected in June during their juvenile stage with a net within the same geographic study zone at around 1400 m of altitude, and were randomly distributed among herbivore treatment cages. During the experiment, we checked every two weeks for orthopteran survival by counting the number of living individuals in each cage. We kept the density constant by replacing dead individuals by new adult individuals caught on the same collection site. The mortality during the season was globally low (personal observation). At the end of the season, all orthopterans were removed by hand from each cage and released on the collection site. Orthopteran inventories performed in the herbivore treatment plots at the end of each year (i.e. 2015, 2016 and 2017) revealed the following species mean abundances (mean ± sd individuals per species across all herbivore treatment plots; Chorthippus parallelus = 5.3 ± 2.1, Euthystira brachyptera = 1.6 ± 2.1, Metrioptera roselii = 1.8 ± 0.9, Metrioptera saussuriana = 1.2 ± 0.6, Omocestus viridulus = 1.1 ± 0.3, Stauroderus scalaris = 2.5 ± 1.5). Those abundances are

- 167 -

proportional to the natural species abundances observed on the collection site along a 10 m transect (number of observation: Chorthippus parallelus = 16, Decticus verrucivorus = 4, Euthystira brachyptera = 1, Omocestus viridulus = 2, Pholidoptera griseoaptera = 1, Mecostethus parapleurus = 1, Metrioptera roselii = 6, Metrioptera saussuriana = 1, Stauroderus scalaris = 6, Stenobothrus lineatus = 1, Tettigonia cantans = 2). Note that we did not include omnivore orthopteran species in order to avoid cannibalism (e.g. Decticus verrucivorus, Tettigonia cantans).

The warming treatment was applied by using hexagonal open-top chambers (OTC) manufactured according to ITEX (The International Tundra Experiment) standards (14, 15), which provide an effective and simple method of climate change simulation. The open-top chamber consisted of a hexagonal enclosure built of clear transparent 2 mm thick polymethylmethaacrylate material (PMMA-XT transparent clear, Angst+Pfister SA), 111 cm in ground diameter, a top opening of 60 cm in diameter and a height of 38 cm (see Fig. S1). The walls of the chambers have a 60° inclination relative to the ground. The greenhouses were carefully set up over the 50cm x 50 cm plot each spring as soon as possible after snowmelt and removed before snowfalls in October. Greenhouses were fixed to the ground with steel tent pegs and secured with tensioned ropes. The aboveground plant biomass is essentially zero at snowmelt because the previous summer's above- ground growth senesces to litter over the winter.

Effect of the experimental structures on microclimatic conditions

We assessed the effect of the experimental structures (cages and greenhouses) on the temperature by using high resolution temperature loggers (DS1922L-F5 HomeChip), parametrized at high resolution (0.0625°C) and with a sampling rate of 30 min during the July-August period 2017. The temperature loggers were placed at 20 cm above the ground and fixed under a small white cup to avoid direct solar radiation heating effects. On each of the 3 field sites, we placed a temperature logger in the middle of each OTC greenhouses (n = 8) and outside the greenhouses in the close vicinity (approx. 50 cm, n = 8). On each field sites, we also recorded temperature in herbivore-free cages (n = 4), insect herbivore cages (n = 4), ambient control (n = 4) and at 1.5 m above the ground (n = 2). We then averaged daily (0h-24h), diurnal (11h-17h) and nocturnal (23h- 5h) temperature for each logger. We assessed the effect of the structure on the temperature by using linear mixed-effects model fitted by maximum likelihood, setting site identity and plot as random factors with the “lme” function implemented in the “nlme” package (16). We compared records between cages and ambient control plots and between pairs of loggers inside/outside the greenhouses. We found no effect of the cages on the daily mean aboveground temperature (linear mixed-effects model: slope = 0.07, P = 0.64; Fig. S2). The OTC greenhouses increased the mean

- 168 - aboveground temperature (20 cm) by 1.1 K (linear mixed-effects model: slope = 1.06, P < 0.001; Fig. S2) and strengthened diurnal and nocturnal extremes (mean diurnal temperature 11h-17h = + 3.8 K, mean nocturnal temperature 23h-5h = - 0.6 K; Fig. S3) during the summer season (July- August period). While most of the heating effect inside the OTCs occurs during daytime, the overall mean daily warming effect of the chambers was reduced by occasional night time cooling below the ambient temperatures (14, 17). OTC greenhouses decreases air mixing with ambient temperature, which in turn favours heat accumulation into the chamber during daytime warming and favours night time cooling by trapping cold dense air during night time inversions (14, 17).

In addition, we quantified the effect of the experimental structures (cages and greenhouses) on photosynthetically active radiation by using a quantum meter (MQ-200 apogee instruments) measuring photosynthetic photon flux density (PPFD; μmol m-2 s-1, microEinsteins per square meter per second). We performed 4 measurements per plot (i.e. in the centre of each quarter of the plot) during a sunny day between 12h-13h30 in 2017 on the three field sites, above the plant canopy (at 25 cm above ground), and by orienting the sensor in direction of the sun. We assessed the effect of the structure on the photosynthetically active radiation by using linear mixed-effects model fitted by maximum likelihood, setting site identity and plot as random factors (16). Compared to measurements performed in ambient plots, the photosynthetically active radiation was reduced in mean by 16.8% in cages (linear mixed-effects model: slope = -351.9, P < 0.001) and by only 0.64 % in OTC greenhouses (slope = -13.3, P = 0.01).

Finally, we assessed the effect of the experimental structures (cages and greenhouses) on the soil humidity by measuring the relative water content with a moisture sensor (Field Scout Digital Moisture Sensor TDR 300). As above, we performed 4 measurements per plot (i.e. in the centre of each quarter of the plot), during a sunny day in 2014 and after a rainy period of minimum 3 days on two sites (i.e. 1800 m and 2270 m). We assessed the effect of the structure on the soil humidity by using linear mixed-effects model fitted by maximum likelihood, setting site identity and plot as random factors (16). We found no evidence of an effect of the OTC greenhouses (linear mixed- effects model: slope = -2.69, P = 0.22) or the cages (slope = -1.67, P = 0.38) on the soil humidity (mean ± sd; ambient = 39.0 ± 9.9, cages = 37.3 ± 6.7, OTC greenhouses = 36.3 ± 10.1). To account for the potential effect of the experimental structures on the microclimatic conditions, our study compares the herbivore treatment cages to the herbivore-free control cages, and compares the OTC greenhouses to the ambient control.

- 169 -

Orthopteran species distribution models

We assessed the potential distribution of 52 orthopteran species present in the study region using Swiss occurrences from the Info Fauna database (http://www.cscf.ch/). We built species distribution models using custom code in R by relating occurrence observations to six environmental variables known to have a strong influence on species distributions. The selected environmental variables included the precipitation of the growing season (June to August; mean of the 1990-2010 period), winter precipitation (December to February; mean of the 1990-2010 period), winter temperature (December to February; mean of the 1990-2010 period), solar radiation, topographic position and topographic wetness index at a spatial resolution of 25 m. We checked for multicollinearity with pairwise correlations in order to avoid spurious model calibrations (18). All selected variables showed low correlation between each other (Pearson correlations: r < |0.52|). Because models have the tendency to vary among the different statistical techniques (19, 20), we ran an Ensemble approach (21) by averaging the results of five statistical techniques: generalized linear model (22), gradient boosting model (23, 24), general additive models (25), Random Forest (26) and Maximum entropy (27, 28). Models were calibrated with Swiss occurrences and projected in the study region at a spatial resolution of 25 m. We avoided spatial autocorrelation in the presences data by using the disaggregation tool provided by the “ecospat” package (29) in R by setting a minimal distance of 125 m between presences (5 times the grid resolution). We randomly selected a set of 10000 pseudo-absences (i.e. also known as background data) (30) within Switzerland, gave equal weights to presences and pseudo-absences in the calibration of the model and ran 5 iterations of the models (31).

To evaluate the capacity of the models to correctly predict the presence and absence of the species in Switzerland, we split randomly the previously selected set of presences and pseudo- absences for each iterations into two partitions: 70% was used for model calibration, and the remaining 30% was used for model evaluation. We used the area under the ROC-plot curve (32) and the True Skill Statistics (33), which both evaluate the ability of the model to discriminate presences from absences. AUC varies between 0 (counter prediction) and 1 (perfect prediction), 0.5 meaning random predictions. TSS is scaled between -1 and 1, 0 meaning random predictions. Models are considered to have reliable prediction performances with AUC values > 0.70 (i.e. excellent AUC > 0.90; good 0.80 < AUC < 0.90; fair 0.70 < AUC < 0.80; poor AUC < 0.70) (34) and TSS values > 0.40 (i.e. excellent TSS > 0.75; good 0.40 < TSS < 0.75; poor TSS < 0.40) (35). Finally, evaluators were averaged for all models and replicates. Overall, all models showed good evaluations (Table S12).

- 170 -

We projected the models over current (1961-1990 period) and one future climate projection scenarios (RCP 8.5) averaged for five different global climate models (GCMs) of the Fifth Assessment IPCC report (AR5, 2014; CLMcom_CCLM4-8-17_CNRM_CM5, CLMcom_CCLM4- 8-17_ICHEC_EC-EARTH, CLMcom_CCLM4-8-17_MPI-ESM-LR, DMI_HIRHAM5_ICHEC_EC-EARTH, KNMI_RACMO22E_ICHEC_EC-EARTH) using one averaged time period (i.e., 2080 from the 2070-2090 period). The RCP 8.5 scenario assumes continuing rising emissions in anthropogenic greenhouses gas and predicts a mean global warming of 3.7 K (2.6 - 4.8 K) for the 2081-2100 period compared to the reference period 1986-2005 (36). The mean temperature increase for 2080 in the study area reaches 3.56 K under the RCP 8.5 scenario and the different GCMs. Because insects are long disperser and likely follow their climatic niche to higher elevation (37), we projected all models assuming unlimited dispersal of species (38). We then averaged all the model projections in a single map and binarised it using a threshold maximizing the TSS obtained with the “optimal.thresholds” function from the “PresenceAbsence” package (39) in R. Binarised maps were then combined into one consensus map representing the orthopteran species richness.

We calculated for each species and scenarios (Current and RCP 8.5) the species lowest, mid and upper range distribution (quantile 5%, 50% and 95%) in the study region. We estimated the mean upper elevation shift of all orthopteran insects between current and future climatic conditions as the mean difference between current and future upper elevation ranges (quantile 95%). Finally, we summed single species distributions and investigated forecasted changes in species richness along the elevation gradient above 1000 m between current and future climatic conditions. We calculated the ratio of “future to current herbivore richness” which indicates the relative increase (ratio > 1) or decrease (ratio < 1) in species richness in the study region.

Orthopteran community abundance model

To illustrate how herbivore abundance may shift in response to climate change along the elevation gradient, we modelled the abundance of orthopteran communities with respect to the same six environmental variables used to model single species distributions (i.e. precipitation of the growing season, winter precipitation, winter temperature, solar radiation, topographic position and topographic wetness index) at a spatial resolution of 25 m. We used the inventories of 190 orthopteran communities performed in the study region along an elevation gradient ranging from 400 m to 3200 m and which are available from a previous study (40, 41). We related orthopteran abundance to the environmental predictors by using a generalized linear model (22) including all the six predictor variables with polynomial terms and Poisson distribution. We tested the

- 171 -

performance of the abundance model by comparing the observed to the predicted abundance with a Spearman’s rank correlation test. The model showed a significant correlation between predicted and observed abundances (Spearman correlation test: r = 0.73, P < 2.2e-16). We then projected the model under current and future climate conditions (RCP 8.5 climate change scenario) in the study region. We calculated the ratio of “future to current herbivore abundance” which indicates the relative increase (ratio > 1) or decrease (ratio < 1) in herbivore abundance, a measure of herbivore pressure, in the study region.

Plant leaf traits

On each of the three field sites, we measured for each plant species a set of functional traits relating to (i) competition (i.e. plant height), (ii) the leaf economic spectrum (i.e. leaf dry matter content, specific leaf area) and (iii) physical and chemical resistance against abiotic stresses and/or herbivores (i.e. leaf toughness, carbon to nitrogen ratio, chemical richness, chemical diversity, leaf metabolomics components content).

Specific leaf area (SLA) and leaf dry matter content (LDMC) were measured according to standard protocols (42, 43). During August 2015, 10 individuals per species were randomly sampled on each of the three selected field sites, at the same phenological stage whenever possible, stored in moist bags in a cool box (10°C) and rehydrated previous to measurements (44). One well- developed and undamaged leaf per individual was then randomly chosen for trait measurement. Leaves were first scanned, weighted wet and then dried during 4 days at 50°C. The leaf area (LA; mm2) was estimated from the scanned leaf by using custom codes in R. Leaf biomass (weight; mg) corresponds to the leaf dry mass. SLA (mm2 mg-1) was calculated as the area of the leaf divided by its dry-mass and LDMC (mg g-1) was calculated as the ratio of the leaf dry mass to its saturated fresh mass. High LDMC plants have tough leaves which are more resistant to physical hazards (i.e. herbivory, wind, hail) (42). SLA is correlated with the potential relative growth rate of a plant, where high SLA plants have thin leaf blades corresponding to lower investments per unit leaf area in structural defence strategies and short leaf life span (42).

In 2016, leaf toughness was assessed using a custom made Punching test machine (Imada Inc) measuring the force required to punch a hole through the lamina of the leaf (45, 46). The device consist of a flat-ended cylindrical steel rod (2 mm diameter) mounted onto a moving head and a stationary base with a sharp-edged hole with a 0.15 mm clearance (45, 46). On each of the three selected sites, 10 individuals per species were randomly sampled, at the same phenological stage whenever possible, stored in moist bags in a cool box (10°C) and rehydrated previous to

- 172 - measurements. We avoided primary and secondary veins whenever possible and measured leaf thickness prior to punch tests with a digital caliper gauge (0.01 mm precision). We then calculated the specific punch strength (i.e. the punch strength per unit leaf thickness at the point of testing expressed in GN m-2 m-1), a measure of leaf toughness.

Plant size (mean height) was calculated for each species by using the vegetation height survey performed in 2014 on each plot and site with the point-intersect method (2, 3). We calculated the plant height for each species and sites as the mean height trough all recorded height contacts with the pins. Plant height is correlated to the plant competitive vigour (42, 47).

Leaf metabolomics analysis and nutrient content (carbon to nitrogen ratio C:N) were analysed on one sample of 10 dry mixed ground leaves per species and sites and collected in 2016. For metabolomics analysis, we extracted about 20 mg of ground tissue with 0.5 ml methanolic solution (MeOH: MilliQ water: formic acid; 80:19.5:0.5). The extraction was analysed via ultrahigh performance liquid chromatography - quadrupole time-of-flight mass spectrometry (UHPLC- QTOFMS) using an Acquity UPLCTM coupled to a Synapt G2 MS (Waters). The separation was carried out at a flow rate of 0.6 mL/min and a temperature of 40°C on an Acquity UPLCTM C18 column (50x2.1mm, 1.7μm). We applied a linear gradient from 2-100% B in 6.0 min on a mobile phases consisting in H2O+0.05% formic acid (solvent A) and acetonitrile+0.05% formic acid (solvent B). The injection volume was 2.5 μL. We performed MS detection in positive electrospray ionization over a mass range of 85-1200 Da. We used the so-called MSe mode, in which the mass spectrometer operates in data-independent analysis (DIA) and quickly alternates between low and high collision energies to yield ions of the molecular species and fragments. Samples were analysed in 3 batches over 3 consecutive days. The MS source was cleaned before each batch. We performed peak picking in Markerlynx XS (Waters) as described in Gaillard et al. (48). The complete metabolite profile was used to assess the number of individual chemical compounds per species (chemical richness) and the inverse Simpson diversity index based on the abundance of individual peaks per species (chemical diversity) using the package “Vegan” in R (49). We also retained the four first axis of a correspondence analyses (50), applied over the full binarized metabolite profile (presence and absence of metabolites), representing 5.8% of the total variability (Fig. S4). The first axis (1.8% of total variance) opposed Gentianaceae, Capriforliaceae, Orobanchaceae and Plantaginaceae species (high scores) to other plant families (low scores, Fig. S4). The second axis (1.7% of total variance) particularly distinguished Gentianaceae, Fabaceae and Primulaceae species (low scores) to Asteraceae, Geraniaceae, Ericaceae and Plantaginaceae species (high scores; Fig. S4, Table S13). We did not consider the second axis for further analyses because this axis is negatively correlated to chemical richness (Spearman correlation test: r = -0.71, P < 0.001; Fig. S4,

- 173 -

Table S13). The third axis (1.2% of total variance) opposed Orchidaceae and Gentianaceae species (low scores) to Orobanchaceae and Plantaginaceae species (high scores) and was positively correlated to chemical richness (Spearman correlation test: r = 0.44, P < 0.001; Fig. S4, Table S13). The fourth axis (1.1% of total variance) opposed the taxonomic group of Poales (Poaceae, Cyperaceae and Juncanceae; high scores) to Rosaceae, Cystaceae, Ericaceae and Fabaceae species (low scores; Fig. S4) and was negatively correlated to chemical diversity (Spearman correlation test: r = -0.40, P < 0.001; Fig. S4, Table S13). Leaf nitrogen (mg g-1) and carbon (mg g-1) contents were analysed by using an elemental analyser (NC-2500 from CE Instruments). Carbon to Nitrogen ratio (C:N) indicates plant nitrogen availability per unit of carbon to herbivores (51). We used the average trait value among all sampled individuals for each species and sites for further analyses.

Changes in richness, functional-diversity and beta-diversity

For each plant inventories performed in 2014 and in 2017 (i.e. species plant biomass derived from point-intercept and cover estimates), we calculated the inverse simpson index (diversity metric) and the total number of species (richness metric) by using the “diversity” function implemented in the “Vegan” package (49). We also calculated the functional divergence of the plant communities, using plant leaf functional traits (LDMC, toughness, SLA, C:N, plant height, chemical richness and diversity, and three axis of the metabolomics correspondence analyses) with the “dbFD” function implemented in the “FD” package (52). We then calculated the difference between 2014 and 2017 metrics. We investigated changes in species richness, diversity and functional diversity under the herbivore and warming treatments by using a linear mixed-effects model fitted by maximum likelihood, setting treatment and site elevation as fixed effects and site and plot identity as a random effect (16). We compared the herbivore treatment to herbivore-free control cages and the warming treatment to ambient control plots.

In addition, we computed abundance-based beta-diversity between pairs of 2014 and 2017 plot inventories (bray-curtis dissimilarity index) by using the “beta.pair.abund“ function implemented in the “betapart” package (53, 54). We also calculated the number of species gained (new species occurrence in the plot) and lost (species extinction) between pairs of 2014 and 2017 plot inventories. Species gains, species losses, and changes in beta-diversity (community dissimilarity) under the herbivore and warming treatments were finally investigated by using a linear mixed-effects model fitted by maximum likelihood, setting treatment and site elevation as fixed effects and site and plot identity as a random effect (16). We compared the herbivore treatment to herbivore-free control cages and the warming treatment to ambient control plots.

- 174 -

Changes in plant biomass

We investigated changes in total plot biomass (i.e. expressed as the sum of all plant hits in the plot) between 2014 and 2017 inventories with a linear mixed-effects model fitted by maximum likelihood, setting treatment and site elevation as fixed effects and site and plot identity as a random effect (16). We compared the herbivore treatment to herbivore-free control cages and the warming treatment to ambient control plots.

We investigated changes in plant species biomass (i.e. expressed as the sum of hits of the plant species in the plot) between 2014 and 2017 inventories associated to plant traits (LDMC, toughness, SLA, C:N, plant height, chemical richness and diversity, and three axis of the metabolomics correspondence analyses) with a Monte Carlo Markov Chain generalised linear mixed model with Gaussian distribution implemented in the MCMCGLMM package (55), by taking into account the interaction of the trait with the treatment, the phylogenetic relatedness of the plant species and by setting site and plot identity as random effects with uninformative priors (V=1 and nu=0.002). All selected variables were scaled around there mean and showed low correlation between each other (Spearman correlations: r < |0.71|; Table S13). Phylogenetic relationships between plants of the study sites were retrieved from a well resolved dated phylogeny of European plant species (56). We transformed the tree in an ultrametric phylogenetic tree and pruned it with the plant species pool found in all our sites. We run the model with 50000 iterations and compared separately the herbivore and warming treatments with their associated control (herbivore versus herbivore-free cage; warming versus ambient control). We checked for convergence of posteriors in the models.

Herbivory estimates

During September 2016, we counted for each plant species, plot and site the number of plant leaves with and without herbivory marks. When herbivory signs were present, the percentage of leaf eaten was visually estimated according to a 7-level scale: : <1%, 1-5%, 5-13%, 13-25%, 25-50%, 50-75%, 75-100%. We estimated herbivory damage only for chewing damage (sap sucking, leaf mining, and rasping damages were observed only few times). For each plant species in plot and sites, we estimated the eaten leaf mass, a measure of biomass removal by herbivores, by multiplying the proportion of the leaf eaten by the mean mass of ten leaves of the plant species collected on the field site. We also estimated the total leaf mass, a measure of biomass, by multiplying the total number of leaves counted in the plot by the mean mass of ten leaves. To investigate if the translocated herbivore community is feeding on dominant plant species, we related eaten leaf mass

- 175 -

to total biomass of the plant observed in the herbivory plots by using a Spearman’s rank correlation test. Finally, to investigate if changes in plant biomass are associated to biomass removal in the translocated herbivore community, we related eaten leaf mass in herbivory plots to observed changes in plant biomass between 2014 and 2017 by using a Spearman’s rank correlation test.

Cafeteria choice experiment

During August 2017, we collected adult individuals (males and females) of 4 low elevation (1000-1500 m) grasshopper species (i.e. Chorthippus parallelus, Stauroderus scalaris, Euthystira brachyptera and Omocestus viridulus) in grasslands of the municipality of Morcles in the Swiss Alps (07°02′ 55″E, 46°12′ 55′′N). The selected grasshoppers correspond to typical low elevation species, and were selected based on their known elevation ranges, their abundances and frequencies in the meadows of the study region (Table S1). All individuals were brought into cages (hereafter named “waiting cage”) containing living plant species transplanted from the study area and used in the cafeteria experiment. Grasshoppers were randomly attributed to a plant community from high or low elevation and placed in contact with the living transplanted species into the waiting cages for at least 3 days. This was performed to provide the opportunity to the grasshopper to learn and try to feed on the plant species prior to the experiment (57). The day before the experiment, fresh and undamaged leaves of 16 low elevation and 16 high elevation plant species were collected along the same elevation transect as grasshoppers and were stored in moist bags and dark conditions at 4°C. The selected plant species correspond to 16 congeneric or phylogenetically closely related plant pairs (e.g. low elevation: Knautia arvensis, high elevation: Knautia dipsascifolia) dominant or subdominant in the natural communities in the study region and were selected to cover most of the plant families and the functional range. For the cafeteria experiment, we offered whole leaves rather than standardized leaf area in order to reduce desiccation effects (58). High elevation or low elevation leaves were randomly placed in a box (40 cm × 20 cm × 4 cm) and hydrated with a 2 ml Eppendorf of water to prevent desiccation. Adult grasshoppers were collected from the high or low elevation plant waiting cage, starved during 12 hours, and introduced into the corresponding high or low elevation plant box. The box was covered with a transparent glass and a single grasshopper individual was tested at each cafeteria session, which lasted 5 hours. After each session, all leaves were checked for herbivory signs and the percentage of leaf eaten was visually estimated according to a 7-level scale: : <1%, 1-5%, 5-13%, 13-25%, 25-50%, 50-75%, 75-100%. We performed 6 sessions with 64 boxes, and leaves were replaced by fresh ones after each sessions. We estimated the eaten leaf mass by multiplying the proportion of the leaf eaten by the mean mass of ten leaves of the plant species measured before the experiment.

- 176 -

We investigated the feeding choice (eaten leaf mass) of the low elevation orthopteran species associated to three plant traits (toughness, SLA, C:N) with a Monte Carlo Markov Chain generalised linear mixed model implemented in the MCMCGLMM package (55), by taking into account the phylogenetic relatedness of the plant species and by setting orthopteran species and sessions as random effects with uninformative priors (V=1 and nu=0.002). Because the response variable (eaten leaf mass) is zero-inflated, we used a MCMCglmmm with zero-inflated hurdle Poisson distribution following the recommendation of Hadfield (55, 59) and rounded the response variable to the nearest integer. We used the average plant trait (SLA, toughness and C:N) for each plant species, which were measured by following standard protocols (42, 43) or retrieved from different plant trait databases or publications (40, 60, 61). Phylogenetic relationships between plants of the study sites were retrieved from a well resolved dated phylogeny of European plant species (56). We transformed the tree in an ultrametric phylogenetic tree and pruned it with the plant species used in the cafeteria choice experiment. We run the model with 200000 iterations and checked for convergence of posteriors in the models. We calibrated the model on the low elevation plant species and projected the model on the high elevation plant species. We then compared predicted values of feeding choice to observed values of feeding choice (i.e. eaten leaf mass) to assess diet conservatism using a Spearman’s rank correlation test.

DNA meta-barcoding

During August 2015 and 2016, we collected faeces of adult individuals (males and females) of 5 orthopteran herbivores species (Chorthippus parallelus, Euthystira brachyptera, Omocestus viridulus, Stauroderus scalaris, Metrioptera roeselii) used in the herbivore translocation experiment. Faeces were collected on the low elevation source site (collection site; 1400 m) and in the herbivore cages installed on the three alpine plant communities (i.e. 1800, 2070 and 2270 m). Individuals were separately placed into falcon tubes and faeces were carefully collected after a few hours. Faeces where then pooled into one common sample per species and site. We then reconstructed the food-web using DNA meta-barcoding techniques. The wetlab procedures and bioinformatics analyzes were conducted in collaboration with partners of Genetic Diversity Centre (GDC), ETH Zürich. DNA extractions of faeces samples were performed using the FastDNA™ SPIN Kit for Soil (MP Biomedicals, Santa Ana, USA). Metabarcoding procedure follows a two-step PCR-based approach which starts with an Amplicon PCR using published forward and reverse primers (ITS2-S2F (62); ITS4_rev (63)) that amplify the second internal transcribed spacer (ITS2) of plant nuclear ribosomal DNA (360bp), recognized as a novel plant DNA barcode to identify species at a high taxonomical resolution (62, 64). Amplicon primers were extended with 8

- 177 -

nucleotides barcodes and were compatible with Illumina Nextera XT indexing adaptors (Illumina, San Diego, USA). After purification of PCR products using 0.8x AMPure ratio (Beckman Coulter Inc., Atlanta Georgia, USA) indexing PCRs were performed using the Nextera XT kit following manufacturer instructions. After 0.8x AMPure ratio purification of indexing products and equimolar pooling of the samples, libraries were sequenced on Illumina sequencing platform (MiSeq, 300 PE). After quality control (65), quality filtering and trimming (66, 67) the reads were demultiplexed. ZOTU calling was performed using UNOISE (68) with a 97% threshold. Taxonomical assignment of OTUs (i.e. operational taxonomic units) was done using SINTAX classifier (69) and BLAST (70) against Genbank (71). OTUs were then normalized by calculating the relative read abundance per sample (RRA; 63). We ranked OTUs according to their summed RRA across samples and kept only OTUs accounting for 95% of the total RRA. This procedure allowed to reduce the number of OTUs from 466 to 70 taxa. RRA were then summed between similar OTUs genus identification (32 plant genus). The plant species identity was then retrieved from plant inventories performed on the source and the experimental sites. For plant genus showing several species representatives on a site, a weighted RRA value was attributed to each species by dividing RRA by the number of species of the corresponding genus. We then used RRA as a semi-quantitative approximation of the ingested biomass by the herbivore, which has been shown to provide a more accurate view of diet than OTUs occurrence data (72).

We investigated the feeding choice (i.e. relative read abundance in the faeces) of the low elevation orthopteran species associated to four plant traits (toughness, SLA, LDMC C:N), plant relative abundances on the field sites and Liliopsida/Magnoliodspida identity (binary variable) with a generalized linear model with Poisson distribution (22). RRA values were rounded to the nearest integer. Plant traits were measured following standard protocols (42, 43) or retrieved from different plant trait databases or publications (40, 60, 61). We visually estimated species cover over a 10 m2 surface according to a 7-level scale: <1%, 1-5%, 5-13%, 13-25%, 25-50%, 50-75%, 75-100%. We calibrated the model on the low elevation plant species (i.e. collection site) and projected the model on the high elevation plant species (i.e. three field sites). We then compared predicted values of feeding choice to observed values of feeding choice to assess diet conservatism using a Spearman’s rank correlation test.

Effect of the warming treatment on plant functional traits

We evaluated the effect of the warming treatment on the leaf chemistry of 16 frequent plant species spanning different plant families and present on the different sites (4 species at 1800 m, 5 species at 2070 m and 7 species at 2270 m). In August 2016, we randomly collected 3-5 plant

- 178 - leaves across 4 plots of the warming treatment and ambient control (i.e. 4 replicates) by collecting leaves around the central 50 x 50 cm plots whenever possible. We performed leaf metabolomics analysis on one sample of dry mixed ground leaves per species and plot by following the same protocol as mentioned above. We then used an analysis of variance to test the effect of the warming treatment on the leaf metabolomics (warming versus ambient control). We used the “adonis” function implemented in the “Vegan” package (49) in R, which preforms an analysis of variance based on Euclidean distance matrices and permutation test, by setting treatment, site identity and species as factors (with interaction) and by setting species and site identity as strata for within species randomizations (1000 permutations). Finally, we calculated for each species the number of metabolomics components significantly increasing or decreasing (two sample t-test) their expression by comparing expression profiles between the ambient control and warming treatment. For each species, we selected the 4 metabolite compounds (markers) that increased and decreased the most their expression. Markers of interest were tentatively identified on the basis of their molecular formulae (determined from accurate measurements of both mass-to-charge ratios and isotopic abundances), fragmentation patterns, and comparison with available databases such as the Dictionary of Natural Products (CRC Press), ReSpect for Phytochemicals, or Massbank.

Finally, we investigated the effect of the warming treatment on six plant functional traits measured on the selected 16 plant species: SLA, LDMC, leaf toughness, C:N, chemical richness and chemical diversity. From the aforementioned leaf metabolomics analysis, we used the full metabolomics survey to assess the number of individual chemical compounds per species (chemical richness) and the inverse Simpson diversity index based on the abundance of individual peaks per species (chemical diversity) using the package “Vegan” (49). Leaf C:N was measured on the same samples as the metabolomics analyses. Leaf LDMC, SLA and leaf toughness were measured on the collected leaves following the same protocols as described above. We calculated the mean trait value per species and plot. We then related the warming treatment (binary: ambient = 0, warming =1) to plant functional traits with a Monte Carlo Markov Chain generalised linear mixed model with binomial distribution (55) and by taking into account the phylogenetic relatedness of the plant species. Phylogenetic relationships were retrieved as described above. We set traits as fixed effect, site identity as random effect with uninformative priors (V=1 and nu=0.002) and residuals with strong priors (V=1, fix=1) following the recommendation of Hadfield (55, 59). All variable showed low correlations between each other (Spearman correlations: r < |0.71|). Leaf toughness was square root transformed and all variables were rescaled around their mean. We set thinning factor to 200 to reduce autocorrelation of consecutive samples when running 400000 iterations, and checked for convergence of posteriors in the model.

- 179 -

References

1. R. Delarze, Y. Gonseth, P. Galland, Guide des milieux naturels de Suisse: écologie, menaces, espèces caractéristiques (Delachaux et niestlé Lausanne, 1998). 2. D. Mueller-Dombois, H. Ellenberg, Aims and methods of vegetation ecology (Wiley, New York, 1974). 3. S. Jonasson, Evaluation of the Point Intercept Method for the Estimation of Plant Biomass. Oikos. 52, 101–106 (1988). 4. A. Stampfli, Accurate determination of vegetational change in meadows by successive point quadrat analysis. Vegetatio. 96, 185–194 (1991). 5. E. Post, C. Pedersen, Opposing plant community responses to warming with and without herbivores. PNAS. 105, 12353–12358 (2008). 6. A. Eskelinen, S. Harrison, M. Tuomi, Plant traits mediate consumer and nutrient control on plant community productivity and diversity. Ecology. 93, 2705–2718 (2012). 7. E. Kaarlejärvi, A. Eskelinen, J. Olofsson, Herbivores rescue diversity in warming tundra by modulating trait-dependent species losses and gains. Nature Communications. 8, 419 (2017). 8. E. Kaarlejärvi, K. S. Hoset, J. Olofsson, Mammalian herbivores confer resilience of Arctic shrub-dominated ecosystems to changing climate. Global Change Biology. 21, 3379–3388 (2015). 9. P. Evans, E. L. Davis, Z. Gedalof, C. D. Brown, Small herbivore exclosure cages alter microclimate conditions. Forest Ecology and Management. 415–416, 118–128 (2018). 10. I. Badenhausser, Estimation d’abondance des criquets (Orthoptera: Acrididae) dans les écosystèmes prairiaux. Annales de la Société entomologique de France (N.S.). 48, 397–406 (2012). 11. H. Deraison, I. Badenhausser, L. Börger, N. Gross, Herbivore effect traits and their impact on plant community biomass: an experimental test using grasshoppers. Functional Ecology. 29, 650–661 (2014). 12. J. A. Onsager, Toward an Optimum Strategy for Estimating Density of Grasshoppers (Orthoptera: Acrididae) from a Poisson Distribution. Environ Entomol. 20, 939–944 (1991). 13. H.-T. Miao, Y. Liu, L.-Y. Shan, G.-L. Wu, Linkages of plant-soil interface habitat and grasshopper occurrence of typical grassland ecosystem. Ecological Indicators. 90, 324–333 (2018). 14. G. M. Marion et al., Open-top designs for manipulating field temperature in high-latitude ecosystems. Global Change Biology. 3, 20–32 (1997).

- 180 -

15. G. H. R. Henry, U. Molau, Tundra plants and climate change: the International Tundra Experiment (ITEX). Global Change Biology. 3, 1–9 (1997). 16. J. Pinheiro, D. Bates, S. DebRoy, D. Sarkar, nlme: linear and nonlinear mixed effects models. R package version 3.1-117. Available at http://CRAN. R-project. org/package= nlme. (2014). 17. A. Dabros, J. W. Fyles, I. B. Strachan, Effects of open-top chambers on physical properties of air and soil at post-disturbance sites in northwestern Quebec. Plant Soil. 333, 203–218 (2010). 18. A. Guisan, W. Thuiller, Predicting species distribution: offering more than simple habitat models. Ecology Letters. 8, 993–1009 (2005). 19. J. Elith et al., Novel Methods Improve Prediction of Species’ Distributions from Occurrence Data. Ecography. 29, 129–151 (2006). 20. W. Thuiller, L. Brotons, M. B. Araújo, S. Lavorel, Effects of restricting environmental range of data to project current and future species distributions. Ecography. 27, 165–172 (2004). 21. M. B. Araújo, M. New, Ensemble forecasting of species distributions. Trends in Ecology & Evolution. 22, 42–47 (2007). 22. P. McCullagh, Generalized linear models. European Journal of Operational Research. 16, 285–292 (1983). 23. G. Ridgeway, The state of boosting. Computing Science and Statistics, 172–181 (1999). 24. J. H. Friedman, Greedy Function Approximation: A Gradient Boosting Machine. The Annals of Statistics. 29, 1189–1232 (2001). 25. A. Guisan, N. E. Zimmermann, Predictive habitat distribution models in ecology. Ecological Modelling. 135, 147–186 (2000). 26. L. Breiman, Random Forests. Machine Learning. 45, 5–32 (2001). 27. J. Elith et al., A statistical explanation of MaxEnt for ecologists. Diversity and Distributions. 17, 43–57 (2010). 28. S. J. Phillips, R. P. Anderson, R. E. Schapire, Maximum entropy modeling of species geographic distributions. Ecological Modelling. 190, 231–259 (2006). 29. V. D. Cola et al., ecospat: an R package to support spatial analyses and modeling of species niches and distributions. Ecography. 40, 774–787 (2017). 30. M. S. Wisz, A. Guisan, Do pseudo-absence selection strategies influence species distribution models and their predictions? An information-theoretic approach based on simulated data. BMC ecology. 9, 8 (2009). 31. M. Barbet‐Massin, F. Jiguet, C. H. Albert, W. Thuiller, Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution. 3, 327– 338 (2012).

- 181 -

32. A. H. Fielding, J. F. Bell, A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation. 24, 38–49 (1997). 33. O. Allouche, A. Tsoar, R. Kadmon, Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). Journal of Applied Ecology. 43, 1223– 1232 (2006). 34. J. A. Swets, Measuring the accuracy of diagnostic systems. Science. 240, 1285–1293 (1988). 35. J. R. Landis, G. G. Koch, The Measurement of Observer Agreement for Categorical Data. Biometrics. 33, 159–174 (1977). 36. IPCC, Fifth Assessment Report. Cambridge (Cambridge University Press, 2014). 37. M. P. Berg et al., Adapt or disperse: understanding species persistence in a changing world. Global Change Biology. 16, 587–598 (2010). 38. R. Engler, A. Guisan, MigClim: Predicting plant distribution and dispersal in a changing climate. Diversity and Distributions. 15, 590–601 (2009). 39. E. A. Freeman, G. Moisen, PresenceAbsence: An R Package for Presence Absence Analysis. Journal of Statistical Software. 23, 1–31 (2008). 40. P. Descombes et al., Community-level plant palatability increases with elevation as insect herbivore abundance declines. Journal of Ecology. 105, 142–151 (2017). 41. P. Descombes et al., Data from: Community‐level plant palatability increases with elevation as insect herbivore abundance declines. Figshare (2017), (available at http://dx.doi.org/10.6084/m9.figshare.3803955). 42. J. H. C. Cornelissen et al., A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany. 51, 335–380 (2003). 43. N. Pérez-Harguindeguy et al., New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of botany. 61, 167–234 (2013). 44. M. V. Vaieretti, S. Díaz, D. Vile, E. Garnier, Two Measurement Methods of Leaf Dry Matter Content Produce Similar Results in a Broad Range of SpeciesVaieretti et al. — Measuring Leaf Dry Matter ContentVaieretti et al. — Measuring Leaf Dry Matter Content. Ann Bot. 99, 955–958 (2007). 45. N. Aranwela, G. Sanson, J. Read, Methods of assessing leaf-fracture properties. The New Phytologist. 144, 369–383 (1999). 46. G. Sanson, J. Read, N. Aranwela, F. Clissold, P. Peeters, Measurement of leaf biomechanical properties in studies of herbivory: Opportunities, problems and procedures. Austral Ecology. 26, 535–546 (2001). 47. C. L. Gaudet, P. A. Keddy, A comparative approach to predicting competitive ability from plant traits. Nature. 334, 242–243 (1988).

- 182 -

48. M. D. P. Gaillard, G. Glauser, C. A. M. Robert, T. C. J. Turlings, Fine-tuning the ‘plant domestication-reduced defense’ hypothesis: specialist vs generalist herbivores. New Phytologist. 217, 355–366 (2018). 49. J. Oksanen et al., The vegan package. Community ecology package. 10, 631–637 (2007). 50. S. Dray, A.-B. Dufour, The ade4 Package: Implementing the Duality Diagram for Ecologists. Journal of Statistical Software. 22 (2007). 51. W. J. Mattson, Herbivory in Relation to Plant Nitrogen Content. Annual Review of Ecology and Systematics. 11, 119–161 (1980). 52. E. Laliberté, P. Legendre, B. Shipley, FD: measuring functional diversity from multiple traits, and other tools for functional ecology. R package version 1.0-12 (2014). 53. A. Baselga, The relationship between species replacement, dissimilarity derived from nestedness, and nestedness. Global Ecology and Biogeography. 21, 1223–1232 (2012). 54. A. Baselga, C. D. L. Orme, betapart: an R package for the study of beta diversity. Methods in Ecology and Evolution. 3, 808–812 (2012). 55. Hadfield, MCMC Methods for Multi-Response Generalized Linear Mixed Models: The MCMCglmm R Package. Journal of Statistical Software. 33, 1–22 (2010). 56. W. Durka, S. G. Michalski, Daphne: a dated phylogeny of a large European flora for phylogenetically informed ecological analyses. Ecology. 93, 2297–2297 (2012). 57. R. Dukas, E. A. Bernays, Learning improves growth rate in grasshoppers. PNAS. 97, 2637– 2640 (2000). 58. S. Ibanez, S. Lavorel, S. Puijalon, M. Moretti, Herbivory mediated by coupling between biomechanical traits of plants and grasshoppers. Functional Ecology. 27, 479–489 (2013). 59. Hadfield, MCMCglmm: Course notes (2015), (available at https://cran.r- project.org/web/packages/MCMCglmm/vignettes/CourseNotes.pdf). 60. J. Kattge et al., TRY – a global database of plant traits. Global Change Biology. 17, 2905– 2935 (2011). 61. C. Körner, S. Leuzinger, S. Riedl, R. T. Siegwolf, L. Streule, Carbon and nitrogen stable isotope signals for an entire alpine flora, based on herbarium samples. Alp Botany. 126, 153– 166 (2016). 62. S. Chen et al., Validation of the ITS2 Region as a Novel DNA Barcode for Identifying Medicinal Plant Species. PLOS ONE. 5, e8613 (2010). 63. T. J. White, T. Bruns, S. Lee, J. W. Taylor, in PCR Protocols: A Guide to Methods and Applications (Academic Press, New York, 1990), pp. 315–322.

- 183 -

64. C. García-Robledo, D. L. Erickson, C. L. Staines, T. L. Erwin, W. J. Kress, Tropical Plant– Herbivore Networks: Reconstructing Species Interactions Using DNA Barcodes. PLOS ONE. 8, e52967 (2013). 65. S. Andrews, FastQC: a quality control tool for high throughput sequence data (2010; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). 66. R. Schmieder, R. Edwards, Quality control and preprocessing of metagenomic datasets. Bioinformatics. 27, 863–864 (2011). 67. M. Martin, Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet. journal. 17, 10–12 (2011). 68. R. C. Edgar, UNOISE2: improved error-correction for Illumina 16S and ITS amplicon sequencing. bioRxiv, 081257 (2016). 69. R. Edgar, SINTAX: a simple non-Bayesian taxonomy classifier for 16S and ITS sequences. bioRxiv, 074161 (2016). 70. S. F. Altschul, W. Gish, W. Miller, E. W. Myers, D. J. Lipman, Basic local alignment search tool. Journal of Molecular Biology. 215, 403–410 (1990). 71. K. Clark, I. Karsch-Mizrachi, D. J. Lipman, J. Ostell, E. W. Sayers, GenBank. Nucleic Acids Res. 44, 67–72 (2016). 72. B. Deagle et al., Counting with DNA in metabarcoding studies: how should we convert sequence reads to dietary data? bioRxiv, 303461 (2018).

- 184 -

Supplementary materials

Table S1. Grasshopper elevation ranges. Lowest, mid and upper elevation range (quantile 5%, 50% and 95%) of the orthopteran species in the study region, assessed with species distribution models for the current and future (RCP 8.5) climate conditions.

Species Current RCP 8.5 Q5 Q50 Q95 Q5 Q50 Q95 Anonconotus alpinus 1374 1709 2232 2093.3 2452.4 2858.5 Antaxius pedestris 390 539.9 984.5 390.2 549.8 1034.5 Arcyptera fusca 1218.3 1583.6 2001.7 1588.5 1908.2 2647.8 Barbitistes obtusus 386.2 489.9 732.8 557.6 568.7 574.2 Barbitistes serricauda 599.6 1091.8 1541.7 828.9 1533 2074.6 Bohemanella frigida 1954 2261.6 2562.4 3052.2 3116.6 3172.1 Calliptamus italicus 386.8 506 1014.4 385 1217.3 1952.8 Chorthippus apricarius 992.7 1489.8 1946.6 1609.6 1940.6 2670 Chorthippus biguttulus 384.5 1214.5 1800.8 714.7 1457.4 2254.8 Chorthippus brunneus 377.6 711.7 1561.4 387 1310.5 2090.9 Chorthippus dorsatus 378 766.9 1470.1 388.6 1136.8 1614.5 Chorthippus eisentrauti 753.3 1408.6 1923.7 1261.1 1855.9 2608.8 Chorthippus mollis 378.3 490.1 1068.8 379.3 854.9 1575.9 Chorthippus parallelus 379.6 1223.9 1877.7 383.9 1330 2185.3 Chorthippus vagans 386.2 516.9 899.4 378.4 748.7 1414.8 Decticus verrucivorus 947.2 1446.7 1967 1474.6 1878 2612.6 Euthystira brachyptera 671.9 1389.4 1939.8 699.5 1442.2 2291.1 Gomphocerippus rufus 379.9 900 1589.2 384.4 1283.9 2092 Gomphocerus sibiricus 1555.4 1807.6 2317.3 2163.8 2618.5 2944.6 Gryllus campestris 378.7 785.7 1375.5 380 1119.5 1799.5 Leptophyes punctatissima 375.6 455.2 928.6 383.3 1148.2 1695.7 Meconema thalassinum 376.8 545.7 1187.9 387.6 1301.7 1920.1 Mecostethus parapleurus 375.4 446.1 854.8 379.9 1022.4 1606.2 Metrioptera bicolor 440.6 875.8 1477.6 380.6 1204.3 1925 Metrioptera brachyptera 1110.9 1532.7 2088.5 1792.7 2110 2783.3 Metrioptera roeselii 379.8 1214.2 1726.3 1030.4 1522.9 2291.9 Metrioptera saussuriana 1196.4 1575.6 2130.9 1669.9 1949.1 2658.7 Miramella alpina 1099 1550.3 2124.4 1273.5 1767 2563.9 Miramella formosanta 594.7 659.4 1175.1 673.4 927.2 1179.4 Nemobius sylvestris 379.4 699.7 1257.9 386.7 1272.1 1904.1 Oecanthus pellucens 374.6 409.3 778 379.9 962.9 1584.6 Oedipoda caerulescens 374.9 417.2 890 380.4 1009.8 1722.3 Oedipoda germanica 426.6 943.8 1597.6 443.8 1230 1919.7 Omocestus haemorrhoidalis 568.7 1279.1 1825.1 395.9 1142.4 1939.5 Omocestus rufipes 377.5 544.9 1214.7 379.8 935.3 1618.6 Omocestus viridulus 1047.1 1517.1 2071 1577.6 1884 2631.8 Phaneroptera falcata 375.3 433.4 893.9 380.5 1024.6 1609.5 Phaneroptera nana 374.2 399.9 611.1 379.5 885.9 1382.7

- 185 -

Pholidoptera aptera 635.3 1025.7 1610 557.3 557.3 557.3 Pholidoptera griseoaptera 379.2 898.1 1520.4 387 1340.9 2079.3 Platycleis albopunctata 380 707.2 1355.4 380 1003.3 1741.8 Podisma pedestris 1373.1 1703.5 2254.6 1961.9 2537.3 2934.5 Polysarcus denticauda 1074.4 1520 2043.5 1362.4 1707 2515.7 Psophus stridulus 1134.7 1556.2 2058.3 1560.8 1909 2667.7 Ruspolia nitidula 374.5 406.5 651.7 379.7 938.9 1432.2 Sphingonotus caerulans 374 392.9 649.6 379 834.4 1490.6 Stauroderus scalaris 996.6 1485.8 1974.1 763.4 1554.7 2400.7 Stenobothrus lineatus 658.1 1387.3 1908.7 656.1 1466.4 2160.2 Tetrix bipunctata 862.1 1473.9 2027.7 881.9 1544.2 2397.7 Tetrix tenuicornis 377.1 597.2 1406.5 380.3 1097.1 1744.3 Tettigonia cantans 676.5 1304.6 1723.6 781.3 1484.1 2287.8 Tettigonia viridissima 376.7 553.1 1279.2 379.9 1047.4 1750.7

- 186 -

Table S2. Cafeteria choice experiment. Plant choice (eaten dry leaf biomass in mg) of lowland grasshopper species on low elevation plant species, associated to plant traits and assessed using a MCMCglmm with a zero inflated hurdle-Poisson distribution. *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

post.mean l-95% CI u-95% CI eff.samp pMCMC (intercept) 1.897 -0.486 4.297 321.4 0.099 . Herbivory_hurdle 3.405 1.388 5.762 949.1 0.003 ** SLA 0.696 -0.188 1.734 207.2 0.138 Toughness 1.095 0.480 1.763 415.9 0.001 *** C:N 0.654 -0.360 1.695 221.4 0.193

- 187 -

Table S3. DNA meta-barcoding. Plant choice (relative number of reads in insect faeces) of lowland grasshopper species on low elevation plant species (reconstructed by using DNA meta- barcoding techniques applied on insect faeces) associated to plant traits, plant abundance and plant Classe identity (Liliospida = 1, Magtoliopsida = 0) and assessed using a glm. *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

Estimate Std. Error z value Pr(>|z|) (Intercept) 8.257 1.314 6.286 0.000 *** LDMC -0.033 0.003 -11.853 0.000 *** SLA -0.079 0.023 -3.401 0.001 *** Toughness 0.097 0.022 4.447 0.000 *** C:N -0.040 0.020 -1.990 0.047 * Plant abundance 0.917 0.162 5.655 0.000 *** Plant Classe 4.218 0.435 9.687 0.000 ***

- 188 -

Table S4. Changes in plant biomass associated to plant traits under herbivore translocation. Changes in plant biomass index in the herbivore translocation treatment (Herbivore) and the control cage, expressed as the difference in total number of plant hits between 2017 and 2014 inventories, associated to plant traits and assessed using a MCMCglmm. The herbivore translocation experiment changed the overall plant biomass and those changes were associated to particular plant traits. Significant variables are highlighted in bold. CA = correspondence analyses, *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) -3.357 -27.057 18.518 4700 0.767 Herbivore -4.881 -6.275 -3.486 4700 0.000 *** LDMC 1.918 -0.730 4.639 4700 0.158 Toughness -2.297 -4.565 0.013 4700 0.055 . SLA 0.956 -0.827 2.561 4700 0.276 CN -0.264 -2.183 1.544 4306 0.774 Height -0.012 -1.359 1.513 4058 0.972 Chemical richness -1.079 -2.784 0.606 4700 0.224 Chemical diversity 0.075 -1.358 1.607 4486 0.926 CA Axis 1 -1.906 -4.382 0.542 4700 0.125 CA Axis 3 -0.861 -3.001 1.282 4700 0.422 CA Axis 4 -1.043 -4.335 2.286 4700 0.536 Herbivore : LDMC -1.784 -3.958 0.571 4700 0.120 Herbivore : Toughness -4.532 -6.834 -2.198 4941 0.000 *** Herbivore : SLA 1.859 0.319 3.405 4963 0.018 * Herbivore : CN -0.178 -2.100 1.626 4481 0.867 Herbivore : Height 0.049 -1.439 1.616 4700 0.937 Herbivore : Chemical richness -1.220 -2.649 0.222 4700 0.104 Herbivore : Chemical diversity -2.084 -3.536 -0.715 4700 0.004 ** Herbivore : CA Axis 1 0.570 -0.907 1.894 4700 0.436 Herbivore : CA Axis 3 -0.355 -1.714 1.081 3921 0.614 Herbivore : CA Axis 4 -3.577 -5.518 -1.658 4700 0.000 ***

- 189 -

Table S5. Leaf attacks. Proportion of plant leaf attacks between Liliopsida (monocotyledons) and Magnoliopsida (dicotyledons) under the different treatments and controls, and estimated across all sites and replicates.

Treatment Liliopsida Magnoliopsida mean sd mean sd Ambient 0.258 0.157 0.742 0.157 Cage 0.241 0.187 0.759 0.187 Herbivore 0.824 0.087 0.176 0.087 Warming 0.312 0.187 0.688 0.187

- 190 -

Table S6. Warming effect on six plant functional traits. Warming effect on plant traits estimated with MCMCglmm under binomial distribution (ambient control = 0, warming treatment = 1). Significant variables are highlighted in bold. *** p-value < 0.001, ** p-value < 0.01, * p-value < 0.05.

post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) -0.033 -4.215 3.317 1985 0.982 SLA 1.006 0.048 1.927 2217 0.031 * LDMC -1.000 -2.222 0.206 1985 0.096 . Toughness -0.062 -0.937 0.956 1848 0.911 C:N 0.737 -0.077 1.494 2363 0.057 . Chemical richness -1.182 -2.235 -0.157 1985 0.024 * Chemical diversity 0.219 -0.356 0.849 1985 0.470

- 191 -

Table S7. Effect of the warming treatment on plant leaf metabolomics. Analysis of variance (Adonis test) of plant leaf metabolomics under the warming treatment compared to the ambient control. *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

Ambient VS Warming Df SumsOfSqs MeanSqs F.Model R2 Pr(>F) Treatment 1 66460 66460 2.701 0.00381 0.001 *** Site 2 1676550 838275 34.063 0.09604 0.027 * Species 13 12823246 986404 40.082 0.73458 0.027 * Treatment:Site 2 65575 32787 1.332 0.00376 0.101 Treatment:Species 13 413029 31771 1.291 0.02366 0.054 . Residuals 98 2411720 24609 0.13816 Total 129 17456580 1

- 192 -

Table S8. Warming effect on leaf metabolomics compounds. The table shows the total number of metabolite compounds present in the plant species, the number of metabolite compound significantly increasing or decreasing (two sample t-test between control and treatment) under the

warming treatment compared to the samples collected in the ambient control conditions.

Species Number of metabolite compounds Decreasing Increasing not Percent changing Percent decreasing Percent increasing Alchemilla_conjuncta_La_Corde 4538 81 39 97.36 1.78 0.86 Anthyllis_vulneraria_Paneirosse 4653 51 47 97.89 1.10 1.01 Aster_alpinus_Paneirosse 4563 59 24 98.18 1.29 0.53 Carex_montana_Bovonne 4482 37 39 98.30 0.83 0.87 Carex_sempervirens_La_Corde 4372 33 26 98.65 0.75 0.59 Crepis_pyrenaica_Bovonne 4727 71 34 97.78 1.50 0.72 Helianthemum_nummularium_La_Corde 4669 39 37 98.37 0.84 0.79 Homogyne_alpina_Paneirosse 4657 109 56 96.46 2.34 1.20 Leontodon_hispidus_Paneirosse 4652 133 76 95.51 2.86 1.63 Plantago_atrata_La_Corde 4742 36 72 97.72 0.76 1.52 Polygonum_viviparum_Paneirosse 4535 168 79 94.55 3.70 1.74 Potentilla_crantzii_Paneirosse 4466 93 47 96.87 2.08 1.05 Potentilla_erecta_Bovonne 4572 42 50 97.99 0.92 1.09 Prunella_grandiflora_Bovonne 4565 32 32 98.60 0.70 0.70 Sesleria_caerulea_La_Corde 4837 32 66 97.97 0.66 1.36 Sesleria_caerulea_Paneirosse 4389 21 34 98.75 0.48 0.77

- 193 -

Table S9. Leaf metabolomics peaks. Identified metabolomics compounds that (A) decrease and (B) increased significantly their expression under the warming treatment. The table shows the investigated plant species together with the site identity, the identity of the metabolite compound, the change in the expression of the metabolite compound (mean difference between control and the warming treatment peaks), the four best metabolite for each species, the tentatively identified metabolite and the associated group of metabolite. Species were collected on three sites (Bovonne = 1800 m, La Corde = 2070 m, Paneirosse = 2270 m).

A Decrease in expression

Species Identity Expressionchange BestGroup Metabolitetentatively identified Group Alchemilla conjuncta La C27H28O18: quercetin- Corde X4634_rt_0.992_mz_641.1345 -30.49 1 glucose-glucuronide flavonoid Anthyllis vulneraria C21H18O13: quercetin- Paneirosse X1326_rt_1.4732_mz_303.0511 -20.57 1 glucuronide (fragment) flavonoid C21H18O13: quercetin- Aster alpinus Paneirosse X1329_rt_1.4237_mz_303.051 -10.78 1 glucuronide (fragment) flavonoid C27H30O16: flavonoid but Carex montana Bovonne X3265_rt_1.0784_mz_611.1603 -14.55 1 no specific fragment flavonoid Carex sempervirens La C24H50NO7P: Corde X5344_rt_4.2867_mz_496.34 -42.65 1 glycerophosphocholine phospholipid Crepis pyrenaica C24H50NO7P: Bovonne X5344_rt_4.2867_mz_496.34 -43.05 1 glycerophosphocholine phospholipid Helianthemum nummularium La Corde X806_rt_2.0726_mz_430.1718 -2.85 1 C19H27NO10: alkaloid? alkaloid

C24H50NO7P: Homogyne alpina glycerophosphocholine Paneirosse X5344_rt_4.2867_mz_496.34 -42.32 1 (GroPCho) phospholipid Leontodon hispidus C23H22O14: flavonoid- Paneirosse X6966_rt_1.6359_mz_521.0937 -51.09 1 acetylhexose flavonoid Plantago atrata La C27H26O18: flavonoid- Corde X4511_rt_1.2485_mz_639.119 -19.09 1 diglucuronide flavonoid Polygonum viviparum C27H30O16: rutin or isomer Paneirosse X3263_rt_1.4266_mz_611.1604 -63.12 1 (=quercetin-glucose-glucose) flavonoid Potentilla crantzii C30H32O20: quercetin- Paneirosse X7218_rt_1.0969_mz_713.155 -16.60 1 glucose-malonylglucose flavonoid Potentilla erecta C21H18O13: quercetin- Bovonne X4194_rt_1.4657_mz_479.0828 -18.13 1 glucuronide flavonoid Prunella grandiflora C32H38O20: quercetin- Bovonne X8029_rt_1.3209_mz_743.201 -35.40 1 glucose-rhamnose-pentose flavonoid Sesleria caerulea La C26H28O15: flavonoid c- Corde X1770_rt_1.2015_mz_581.15 -123.34 1 glycoside? flavonoid Sesleria caerulea Paneirosse X4531_rt_1.0137_mz_355.1033 -2.93 1 too low peak ? Alchemilla conjuncta La Corde X2331_rt_3.2628_mz_453.3371 -17.57 2 C30H44O3: terpene? terpenoid Anthyllis vulneraria C27H30O17: quercetin- Paneirosse X3976_rt_0.9832_mz_627.1552 -16.97 2 glucose-glucose flavonoid C7H15NO3 (primary Aster alpinus Paneirosse X2106_rt_0.2476_mz_162.1139 -10.34 2 metabolite?) ?

- 194 -

Carex montana Bovonne X5284_rt_2.7005_mz_495.1293 -10.71 2 C26H22O10: phenolic? phenol Carex sempervirens La C26H50NO7P: Corde X6968_rt_4.0671_mz_520.3401 -12.65 2 glycerophosphocholine phospholipid Crepis pyrenaica C26H48NO7P: Bovonne X6839_rt_3.7829_mz_518.3244 -13.68 2 glycerophosphocholine phospholipid Helianthemum nummularium La Corde X7712_rt_1.7489_mz_269.1399 -2.78 2 C14H20O5 ? Homogyne alpina C26H50NO7P: Paneirosse X6968_rt_4.0671_mz_520.3401 -22.56 2 glycerophosphocholine phospholipid Leontodon hispidus C16H18O10: coumarin- Paneirosse X2103_rt_0.9281_mz_163.0403 -29.84 2 glucose derivative (fragment) phenylpropanoid Plantago atrata La Corde X5296_rt_0.6018_mz_369.1167 -16.14 2 C17H20O9 ? Polygonum viviparum C24H50NO7P: Paneirosse X5344_rt_4.2867_mz_496.34 -33.67 2 glycerophosphocholine phospholipid Potentilla crantzii C27H28O18: quercetin- Paneirosse X4634_rt_0.992_mz_641.1345 -11.65 2 glucose-glucuronide flavonoid Potentilla erecta Bovonne NA NA NA NA NA Prunella grandiflora C30H46O3: terpene? Bovonne X1249_rt_4.1967_mz_437.3421 -6.25 2 (fragment) terpenoid Sesleria caerulea La C27H30O15: kaempferol- Corde X2394_rt_1.349_mz_595.1657 -14.54 2 pentose-glucose flavonoid Sesleria caerulea coumaric acid derivative? Paneirosse X1547_rt_1.28_mz_147.0456 -2.64 2 (unknown precursor) phenolic Alchemilla conjuncta La C21H18O13: quercetin- Corde X1326_rt_1.4732_mz_303.0511 -10.16 3 glucuronide flavonoid Anthyllis vulneraria C26H28O16: quercetin- Paneirosse X2536_rt_1.0449_mz_597.1452 -15.49 3 glucose-pentose flavonoid Aster alpinus Paneirosse X2738_rt_1.6902_mz_457.2084 -3.58 3 ? ? C21H20O12: quercetin- Carex montana Bovonne X3202_rt_1.1227_mz_465.1039 -4.50 3 glucose flavonoid C24H50NO7P: Carex sempervirens La glycerophosphocholine Corde X5633_rt_4.2824_mz_991.6693 -4.83 3 (dimer) phospholipid Crepis pyrenaica C26H28O16: quercetin- Bovonne X2538_rt_1.153_mz_597.1448 -12.04 3 pentose-glucose flavonoid Helianthemum C10H13N5O4: adenosine or nummularium La Corde X7718_rt_0.4685_mz_268.1058 -1.90 3 another purine amino acid Homogyne alpina C12H23NO2 (unknown in Paneirosse X5127_rt_1.7397_mz_214.1814 -7.79 3 DNP) ? Leontodon hispidus C20H18O11: quercetin- Paneirosse X1067_rt_1.4517_mz_435.0926 -17.03 3 pentose (or isomer) flavonoid Plantago atrata La C24H42O7 (fragment): fatty Corde X424_rt_4.0341_mz_425.2908 -6.34 3 acid-glucose? fatty acid Polygonum viviparum C33H40O21: quercetin- Paneirosse X779_rt_1.0672_mz_773.2121 -23.96 3 glucose-glucose-rhamnose flavonoid Potentilla crantzii Paneirosse X4832_rt_2.8293_mz_487.3424 -7.64 3 C30H46O5: triterpene terpenoid Potentilla erecta C30H32O20: quercetin- Bovonne X7218_rt_1.0969_mz_713.155 -11.61 3 glucose-malonylglucose flavonoid Prunella grandiflora C26H28O16: quercetin- Bovonne X2535_rt_1.3435_mz_597.1456 -3.05 3 glucose-pentose flavonoid Sesleria caerulea La 18:3 fatty acid derivative Corde X8126_rt_3.8392_mz_277.2175 -10.14 3 (unknown precursor ion) fatty acid Sesleria caerulea Paneirosse X206_rt_1.6827_mz_755.1657 -1.88 3 ? (low peak) ? Alchemilla conjuncta La C21H18O13: quercetin- Corde X1326_rt_1.4732_mz_303.0511 -10.16 4 glucuronide flavonoid Anthyllis vulneraria C26H28O16: quercetin- Paneirosse X2536_rt_1.0449_mz_597.1452 -15.49 4 glucose-pentose flavonoid

- 195 -

Aster alpinus Paneirosse X2738_rt_1.6902_mz_457.2084 -3.58 4 ? ? C21H20O12: quercetin- Carex montana Bovonne X3202_rt_1.1227_mz_465.1039 -4.50 4 glucose flavonoid C24H50NO7P: Carex sempervirens La glycerophosphocholine Corde X5633_rt_4.2824_mz_991.6693 -4.83 4 (dimer) phospholipid Crepis pyrenaica C26H28O16: quercetin- Bovonne X2538_rt_1.153_mz_597.1448 -12.04 4 pentose-glucose flavonoid Helianthemum C10H13N5O4: adenosine or nummularium La Corde X7718_rt_0.4685_mz_268.1058 -1.90 4 another purine amino acid Homogyne alpina C12H23NO2 (unknown in Paneirosse X5127_rt_1.7397_mz_214.1814 -7.79 4 DNP) ? Leontodon hispidus C20H18O11: quercetin- Paneirosse X1067_rt_1.4517_mz_435.0926 -17.03 4 pentose (or isomer) flavonoid Plantago atrata La C24H42O7 (fragment): fatty Corde X424_rt_4.0341_mz_425.2908 -6.34 4 acid-glucose? fatty acid Polygonum viviparum C33H40O21: quercetin- Paneirosse X779_rt_1.0672_mz_773.2121 -23.96 4 glucose-glucose-rhamnose flavonoid Potentilla crantzii Paneirosse X4832_rt_2.8293_mz_487.3424 -7.64 4 C30H46O5: triterpene terpenoid Potentilla erecta C30H32O20: quercetin- Bovonne X7218_rt_1.0969_mz_713.155 -11.61 4 glucose-malonylglucose flavonoid Prunella grandiflora C26H28O16: quercetin- Bovonne X2535_rt_1.3435_mz_597.1456 -3.05 4 glucose-pentose flavonoid Sesleria caerulea La 18:3 fatty acid derivative Corde X8126_rt_3.8392_mz_277.2175 -10.14 4 (unknown precursor ion) fatty acid Sesleria caerulea Paneirosse X206_rt_1.6827_mz_755.1657 -1.88 4 ? (low peak) ?

B Increase in expression

Species Identity Expressionchange Group Matbolitecoumpounds Group/Function Alchemilla conjuncta La C21H18O13: quercetin- Corde X2321_rt_2.114_mz_595.145 14.42 1 glucuronide (fragment) flavonoid Anthyllis vulneraria C32H38O20: kaempferol- Paneirosse X8028_rt_1.0247_mz_743.2017 14.95 1 glucose-pentose-glucose flavonoid C25H24O12: dicaffeoylquinic acid Aster alpinus Paneirosse X2175_rt_1.7321_mz_163.0408 8.58 1 (fragment) polyphenol

C29H30O19: diglycosylated flavonoid (e.g. tricin Carex montana Bovonne X6245_rt_1.4424_mz_683.1445 8.86 1 diglucoside) flavonoid Carex sempervirens La C23H22O13: flavonoid- Corde X6148_rt_1.7237_mz_507.1141 7.17 1 glucuronide flavonoid Crepis pyrenaica C21H36O5: terpenoid? Bovonne X4341_rt_3.5316_mz_351.2535 5.12 1 (Fragment) terpenoid Helianthemum C39H32O15: kaempferol- nummularium La Corde X7977_rt_2.762_mz_741.1799 15.65 1 monoglycoside-dicoumaroyl flavonoid Homogyne alpina C14H18O7: methyl-benzoyl- Paneirosse X1222_rt_0.9169_mz_137.0614 60.07 1 glucose or isomer (fragment) ? Leontodon hispidus Paneirosse X6464_rt_3.5756_mz_391.2462 10.24 1 ? ? Plantago atrata La C44H52N4O16: breakdown Corde X3638_rt_2.2933_mz_893.3423 46.76 1 product of chlorophyll? ? Polygonum viviparum Paneirosse X953_rt_1.0414_mz_295.1308 2.70 1 ? ?

- 196 -

Potentilla crantzii Paneirosse X505_rt_0.2478_mz_118.0879 9.29 1 C5H11NO2: valine amino acid C15H16O8 Potentilla erecta hydroxycinnamoyl malate Bovonne X2562_rt_1.4351_mz_325.0934 5.18 1 derivative ? Prunella grandiflora Bovonne X2610_rt_4.0365_mz_455.3527 17.43 1 C30H46O3: terpene? terpenoid Sesleria caerulea La Corde NA NA NA NA NA C33H36O16: aegicin Sesleria caerulea glycoside methoxy Paneirosse X6569_rt_2.0418_mz_689.2074 5.21 1 derivative? flavonoid Alchemilla conjuncta La Corde X2983_rt_1.421_mz_177.0566 3.63 2 C10H8O3 ? Anthyllis vulneraria Paneirosse X5691_rt_1.6907_mz_667.2969 6.53 2 ? ? C13H20O3: fatty acid Aster alpinus Paneirosse X4602_rt_1.4387_mz_207.1399 2.67 2 derivative (fragment) fatty acid

Carex montana Bovonne X7075_rt_3.6766_mz_1114.3564 4.47 2 peptide peptide Carex sempervirens La Corde X673_rt_0.6216_mz_429.1367 3.65 2 too small peak ? Crepis pyrenaica Bovonne X4332_rt_0.2424_mz_353.0859 3.62 2 ? ? Helianthemum nummularium La Corde X1538_rt_1.3484_mz_147.0459 12.39 2 coumaroyl-derivative? phenolic Homogyne alpina Paneirosse X2366_rt_0.6614_mz_166.0879 31.71 2 C9H11NO2: phenylalanine amino acid Leontodon hispidus Paneirosse X2661_rt_3.6463_mz_599.4088 4.04 2 ? ? Plantago atrata La Corde X2742_rt_1.8508_mz_457.2078 34.17 2 C22H32O10 ? Polygonum viviparum Paneirosse X7576_rt_0.5787_mz_265.1294 2.65 2 C11H20O7: sugar derivative? ?

Potentilla crantzii C20H28O4: terpene or fatty terpenoid or Paneirosse X3166_rt_3.3572_mz_333.2045 8.59 2 acid derivative fatty acid Potentilla erecta Bovonne X4295_rt_0.9123_mz_935.077 4.67 2 ? ? Prunella grandiflora Bovonne X6894_rt_1.6683_mz_520.2017 5.26 2 C22H33NO13? ? C33H36O16: aegicin Sesleria caerulea La glycoside methoxy Corde X7436_rt_1.7574_mz_527.1553 4.03 2 derivative? (fragment) flavonoid Sesleria caerulea Paneirosse X7974_rt_1.616_mz_741.2014 5.18 2 ? (low peak) ? Alchemilla conjuncta La Corde NA NA NA NA NA Anthyllis vulneraria Paneirosse X4709_rt_2.4164_mz_486.1407 4.91 3 C24H23NO10 (alkaloid?) alkaloid Aster alpinus Paneirosse NA NA NA NA NA

Carex montana Bovonne X6304_rt_4.3788_mz_1054.0583 4.24 3 peptide peptide Carex sempervirens La Corde X5919_rt_0.2518_mz_231.0286 2.23 3 ? ? Crepis pyrenaica Bovonne X3346_rt_0.2544_mz_337.0903 3.40 3 ? ? Helianthemum nummularium La Corde X6439_rt_0.6002_mz_1067.1181 4.39 3 ? ? Homogyne alpina Paneirosse X793_rt_3.6891_mz_557.3113 16.76 3 C32H44O8: terpenoid? terpenoid Leontodon hispidus Paneirosse X3948_rt_1.6052_mz_195.1379 2.43 3 too small peak ?

- 197 -

Plantago atrata La Corde X4803_rt_1.2285_mz_209.1555 14.33 3 C13H20O2 ? Polygonum viviparum Paneirosse X6984_rt_1.8043_mz_253.1441 2.33 3 C14H20O4 ? Potentilla crantzii Paneirosse X8080_rt_3.357_mz_275.2017 6.43 3 fragment of X3166? ? Potentilla erecta Bovonne X6017_rt_1.6475_mz_505.2647 2.85 3 ? ? Prunella grandiflora Bovonne X7487_rt_1.2067_mz_721.1747 4.66 3 C18H16O8 (dimer) ? Sesleria caerulea La Corde X1742_rt_0.2115_mz_151.037 3.76 3 ? ? Sesleria caerulea Paneirosse X2842_rt_1.55_mz_175.1493 4.57 3 ? ? Alchemilla conjuncta La Corde NA NA NA NA NA Anthyllis vulneraria Paneirosse X7139_rt_2.3025_mz_1117.5459 4.38 4 C54H84O24: terpenoid? terpenoid C18H28O3 fatty acid Aster alpinus Paneirosse X1944_rt_4.0088_mz_315.1939 1.76 4 derivative (Na+ adduct) fatty acid C20H28O4: fatty acid Carex montana Bovonne X3166_rt_3.3572_mz_333.2045 3.85 4 derivative fatty acid Carex sempervirens La C23H22O13: flavonoid- Corde X5865_rt_1.7247_mz_1013.2178 2.20 4 glucuronide (dimer) flavonoid Crepis pyrenaica C13H20O3: fatty acid Bovonne X4607_rt_1.6051_mz_207.1394 2.16 4 derivative (fragment) fatty acid Helianthemum C21H36O4: MG nummularium La Corde X4466_rt_3.6377_mz_353.2694 3.04 4 (18:3/0:0/0:0) ? Homogyne alpina Paneirosse X877_rt_1.8168_mz_295.1042 9.92 4 C11H18O9 ? Leontodon hispidus Paneirosse X2662_rt_4.0195_mz_599.4085 2.19 4 ? ? Plantago atrata La C45H54N4O17: breakdown Corde X4096_rt_2.1682_mz_923.3527 13.33 4 product of chlorophyll? ? Polygonum viviparum Paneirosse X1569_rt_3.8011_mz_577.2775 2.28 4 ? ? Potentilla crantzii C18H28O3 (Na adduct): fatty Paneirosse X1944_rt_4.0088_mz_315.1939 2.83 4 acid derivative fatty acid Potentilla erecta Bovonne X7296_rt_1.9072_mz_525.2548 2.70 4 C23H40O13 ? Prunella grandiflora C27H46O9: MGMG Bovonne X5466_rt_3.6378_mz_497.3111 4.56 4 (18:3/0:0) ? Sesleria caerulea La Corde X6352_rt_1.2303_mz_239.0928 3.73 4 ? ? Sesleria caerulea C33H56O14: glycosylated Paneirosse X6695_rt_3.5893_mz_694.4001 4.34 4 terpene? (NH4 adduct) terpenoid

- 198 -

Table S10. Changes in plant biomass associated to plant traits under climate warming. Changes in plant biomass index under the warming treatment (Warming) and ambient control, expressed as the difference in total number of plant hits between 2017 and 2014 inventories, associated to plant traits and assessed using a MCMCglmm. The warming treatment decreased the overall plant biomass, but those changes were not associated to particular plant traits. Significant variables are highlighted in bold. CA = correspondence analyses, *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) -0.241 -9.645 8.933 4700 0.963 Warming -1.612 -2.918 -0.362 5167 0.011 * LDMC 0.565 -1.119 2.365 2768 0.514 Toughness -2.378 -4.049 -0.706 3746 0.006 ** SLA 1.728 0.579 2.866 3008 0.003 ** CN 1.683 0.246 3.043 4700 0.016 * Height 1.879 0.721 3.062 4142 0.002 ** Chemical richness -0.791 -1.900 0.356 4700 0.169 Chemical diversity -0.253 -1.330 0.839 5037 0.636 CA Axis 1 0.114 -1.081 1.382 4700 0.829 CA Axis 3 -0.266 -1.336 0.839 4700 0.620 CA Axis 4 -1.845 -3.468 -0.154 3245 0.029 * Warming : LDMC -1.014 -3.234 1.122 4700 0.371 Warming : Toughness 0.147 -1.967 2.408 4700 0.890 Warming : SLA -0.372 -1.948 1.108 4700 0.644 Warming : CN -0.558 -2.233 1.428 4700 0.541 Warming : Height -0.078 -1.511 1.391 3159 0.915 Warming : Chemical richness 0.617 -0.852 2.021 4354 0.405 Warming : Chemical diversity -1.004 -2.300 0.498 5135 0.178 Warming : CA Axis 1 -0.025 -1.423 1.470 4754 0.971 Warming : CA Axis 3 -0.169 -1.603 1.133 4700 0.791 Warming : CA Axis 4 0.583 -1.224 2.525 4700 0.556

- 199 -

Table S11. Plant inventories. Plant cover (percent of cover; mean and sd) estimated for each sites across all plots (1800m, 2070m and 2270m, n=32) with the cover estimates inventories. Values

correspond to the estimated percent of cover in the 50 x 50 cm plots.

00m_sd

Species

1800m_mean 18 2070m_mean 2070m_sd 2270m_mean 2270m_sd Acinos_alpinus 0 0 0.05 0.15 0 0 Aconitum_compactum 0 0 0 0 0.16 0.55 Agrostis_alpina 0 0 0 0 0.83 0.96 Agrostis_capillaris 0.48 0.09 0 0 0 0 Ajuga_reptans 0.06 0.17 0 0 0 0 Alchemilla_conjuncta_aggr 0 0 3.92 3.43 0 0 Alchemilla_glabra_aggr_cf 0 0 0.25 0.74 0 0 Alchemilla_hybrida_aggr 0 0 0.31 0.89 0.44 0.87 Androsace_chamaejasme 0 0 0 0 1.28 1.18 Anemone_narcissiflora 0.34 0.88 0.42 1 0 0 Anthoxanthum_odoratum 0.58 0.66 1.02 1.07 0.5 0.7 Anthyllis_vulneraria_subsp_alpestris 0 0 0 0 2.23 2.25 Anthyllis_vulneraria_subsp_valesiaca 0.02 0.09 0.38 0.88 0 0 Aposeris_foetida 1.73 1.29 0.84 1.84 0 0 Arenaria_multicaulis 0 0 0 0 0.05 0.15 Arnica_montana 0.2 0.74 0 0 0 0 Aster_alpinus 0 0 0 0 0.98 1.3 Aster_bellidiastrum 0 0 0.5 0.98 0 0 Astragalus_frigidus 0 0 0 0 1.05 2.36 Astrantia_major 8.47 4.95 0.19 0.74 0 0 Bartsia_alpina 0 0 0.17 0.55 0.09 0.53 Biscutella_laevigata 0 0 0 0 0.09 0.53 Botrychium_lunaria 0 0 0.09 0.2 0.05 0.15 Brachypodium_pinnatum 0.03 0.12 0 0 0 0 Briza_media 0.19 0.25 0.16 0.24 0 0 Bupleurum_falcatum_sstr 0.02 0.09 0 0 0 0 Calamagrostis_varia 0.16 0.55 0 0 0 0 Campanula_barbata 0.02 0.09 0.11 0.53 0 0 Campanula_scheuchzeri 0.3 0.25 0.44 0.17 1.05 1.05 Carduus_defloratus_sstr 0.31 0.89 0.52 1.1 0 0 Carex_flacca 0.16 0.55 0 0 0 0 Carex_montana 11.1 9.42 0 0 0 0 Carex_sempervirens 5.45 5.16 11.7 8.71 8.75 6.65 Carlina_acaulis_subsp_caulescens 0.12 0.54 2.11 2.65 0 0 Centaurea_montana 1.56 1.97 0 0 0 0 Centaurea_scabiosa_subsp_alpestris 0.7 1.81 0 0 0 0 Chaerophyllum_villarsii 1.75 2.66 0 0 0 0 Cirsium_acaule 0.58 1.77 0 0 0 0 Coeloglossum_viride 0 0 0.03 0.12 0.02 0.09

- 200 -

Crepis_aurea 0 0 0.09 0.2 0 0 Crepis_bocconei 0.09 0.53 0 0 0 0 Crepis_conyzifolia 0.67 1.04 0 0 0 0 Crepis_pyrenaica 1.34 1.32 0.28 0.89 0 0 Dactylis_glomerata 0.89 2.2 0 0 0 0 Danthonia_decumbens 0.11 0.53 0 0 0 0 Daphne_mezereum 0 0 0.03 0.12 0 0 Dryas_octopetala 0 0 0 0 0.09 0.53 Elyna_myosuroides 0 0 0 0 0.19 0.55 Erigeron_alpinus 0 0 0.02 0.09 0.11 0.21 Euphorbia_cyparissias 0 0 0.47 0.51 0 0 Euphrasia_minima 0 0 0.11 0.21 0.09 0.2 Festuca_laevigata 0 0 0.44 0.52 0.22 0.25 Festuca_laevigata_cf 0 0 0 0 0.19 0.25 Festuca_rubra_aggr 1.7 1.8 4.22 3.18 6.16 3.97 Galium_anisophyllon_cf 0.66 0.79 0.64 0.63 1.36 1.21 Galium_lucidum 0.05 0.15 0 0 0 0 Genista_sagittalis_tige 0.09 0.53 0 0 0 0 Gentiana_acaulis 0.03 0.12 0.75 1.22 0 0 Gentiana_campestris_sstr 0.02 0.09 0 0 0.3 0.74 Gentiana_clusii 0 0 0 0 0.81 1.29 Gentiana_nivalis 0 0 0 0 0.05 0.15 Gentiana_verna 0.02 0.09 0.28 0.55 0.55 0.68 Geranium_sylvaticum 0.86 1.17 0.75 1.12 0 0 Geum_montanum 0 0 0.34 0.88 0 0 Hedysarum_hedysaroides 0 0 0 0 0.59 1.77 Helianthemum_nummularium_subsp_grandiflorum 5.17 7.63 25 11.9 0.02 0.09 Helictotrichon_pubescens 0.17 0.24 0.11 0.21 0 0 Helictotrichon_versicolor 0 0 0 0 0.38 0.54 Hieracium_lachenalii 0.02 0.09 0 0 0 0 Hieracium_murorum_aggr 0.05 0.15 0 0 0 0 Hieracium_pilosella 0.14 0.54 0 0 0 0 Hieracium_villosum 0 0 0.02 0.09 0 0 Hippocrepis_comosa 0.11 0.21 1.09 1.93 0 0 Homogyne_alpina 0.06 0.17 3.28 2.4 1.64 2.72 Hypericum_maculatum_sstr 0 0 0.06 0.17 0 0 Knautia_dipsacifolia_sl 1.05 2.3 0 0 0 0 Koeleria_pyramidata 0.12 0.54 0 0 0 0 Laserpitium_latifolium 0.3 0.74 0 0 0 0 Laserpitium_siler 0.03 0.12 0 0 0 0 Leontodon_hispidus_sstr 0.16 0.55 0.03 0.12 2.42 2.82 Leontodon_hispidus_sstr_cf 0.12 0.22 0 0 0.09 0.53 Leontopodium_alpinum 0 0 0 0 0.02 0.09 Leucanthemum_adustum 0.11 0.21 1.52 1.34 0 0 Ligusticum_mutellina 0 0 0.09 0.53 0.61 1.07 Linum_alpinum 0 0 0.23 0.74 0 0 Linum_catharticum 0.02 0.09 0 0 0 0 Lotus_corniculatus_sl 1.17 1.17 2.17 1.75 0.16 0.55

- 201 -

Luzula_multiflora 0 0 0 0 0.33 0.55 Luzula_sylvatica_aggr 0 0 0.44 1 0 0 Myosotis_alpestris 0 0 0.25 0.25 0.31 0.74 Nardus_stricta 1.38 2.71 0.09 0.53 0 0 Nigritella_nigra_aggr 0 0 0 0 0.02 0.09 Paradisea_liliastrum 0.36 0.23 0 0 0 0 Pedicularis_ascendens 0.11 0.53 0.02 0.09 0 0 Pedicularis_verticillata 0 0 0 0 0.19 0.55 Phleum_hirsutum 0.08 0.18 0.81 1.2 0 0 Phyteuma_orbiculare 1.38 1.84 0.45 0.51 0.69 1.82 Pimpinella_major 0.97 1.21 0 0 0 0 Plantago_alpina 0 0 0.28 0.74 0 0 Plantago_atrata 0.02 0.09 1.06 1.15 0 0 Poa_alpina 0 0 0.14 0.54 0.03 0.12 Polygala_alpestris 0 0 0.28 0.55 0.2 0.55 Polygala_chamaebuxus 0.22 0.25 0 0 0 0 Polygala_vulgaris_sstr 0.09 0.2 0 0 0 0 Polygonum_viviparum 0.22 0.25 0.11 0.21 3.41 1.93 Potentilla_aurea 0.19 0.55 1.06 1.34 0.11 0.53 Potentilla_crantzii 0.61 1.07 1.05 1.26 5.05 4.65 Potentilla_erecta 3.27 3.06 0 0 0 0 Primula_auricula 0 0 0.02 0.09 0.09 0.53 Prunella_grandiflora 2.48 2.86 0 0 0 0 Pulsatilla_alpina_sstr 0 0 1.06 1.95 0 0 Pulsatilla_vernalis 0 0 0 0 0.2 0.74 Ranunculus_montanus_aggr 0 0 1.83 1.89 0.05 0.15 Ranunculus_tuberosus 0.08 0.18 0 0 0 0 Rhinanthus_alectorolophus 0.02 0.09 0 0 0 0 Rosa_pendulina 0.02 0.09 0 0 0 0 Scabiosa_lucida 0.16 0.55 0.7 1.14 0.2 0.74 Senecio_doronicum 0 0 0.28 0.74 0 0 Sesleria_caerulea 5.41 7.62 2.62 1.59 21.5 13.6 Silene_nutans_sstr 0 0 0.02 0.09 0 0 Silene_vulgaris_sstr 0 0 0.11 0.53 0 0 Soldanella_alpina 0.16 0.24 1.41 1.26 0.08 0.18 Taraxacum_officinale_aggr 0.02 0.09 0 0 0.02 0.09 Thesium_alpinum 0 0 0.22 0.25 0 0 Thymus_praecox_subsp_polytrichus 0 0 0.42 0.87 0 0 Thymus_pulegioides_sstr 0.34 0.73 0 0 0 0 Tragopogon_pratensis_subsp_orientalis 0.05 0.15 0.03 0.12 0 0 Traunsteinera_globosa 0.02 0.09 0 0 0 0 Trifolium_badium 0 0 0.11 0.21 0 0 Trifolium_medium 0.77 2.28 0 0 0 0 Trifolium_pratense_sstr 0.27 0.25 0.14 0.23 0.05 0.15 Trifolium_repens_sl 0 0 0.19 0.25 0 0 Trifolium_thalii 0 0 0 0 0.42 1.66 Trollius_europaeus 0.78 1.21 1.62 1.93 0.33 0.89 Vaccinium_gaultherioides 0 0 0.03 0.12 0 0

- 202 -

Vaccinium_myrtillus 0.17 0.55 0.44 1 0 0 Vaccinium_vitis.idaea 0 0 0 0 1.28 1.45 Veronica_aphylla 0 0 0 0 0.03 0.12 Veronica_chamaedrys 0.05 0.15 0 0 0 0 Veronica_officinalis 0.02 0.09 0 0 0 0 Viola_calcarata 0 0 0.28 0.55 0.25 0.55

- 203 -

Table S12. Habitat model evaluations. Evaluation of the species presence models under several evaluators as the average of five algorithms (GLM, GAM, GBM, RF and MAXENT) replicated five times. The table shows the mean and standard deviation of AUC and TSS evaluators.

Species AUC mean AUC sd TSS mean TSS sd Anonconotus_alpinus 0.974 0.012 0.899 0.036 Antaxius_pedestris 0.939 0.018 0.758 0.042 Arcyptera_fusca 0.944 0.013 0.766 0.034 Barbitistes_obtusus 0.960 0.013 0.835 0.035 Barbitistes_serricauda 0.863 0.016 0.558 0.033 Bohemanella_frigida 0.956 0.010 0.803 0.018 Calliptamus_italicus 0.974 0.005 0.851 0.020 Chorthippus_apricarius 0.931 0.020 0.738 0.053 Chorthippus_biguttulus 0.829 0.018 0.492 0.040 Chorthippus_brunneus 0.870 0.017 0.586 0.038 Chorthippus_dorsatus 0.820 0.034 0.485 0.060 Chorthippus_eisentrauti 0.921 0.045 0.724 0.108 Chorthippus_mollis 0.934 0.025 0.735 0.070 Chorthippus_parallelus 0.771 0.028 0.388 0.051 Chorthippus_vagans 0.971 0.010 0.853 0.029 Decticus_verrucivorus 0.841 0.025 0.544 0.046 Euthystira_brachyptera 0.840 0.027 0.551 0.046 Gomphocerippus_rufus 0.815 0.021 0.471 0.041 Gomphocerus_sibiricus 0.915 0.012 0.705 0.022 Gryllus_campestris 0.861 0.019 0.558 0.040 Leptophyes_punctatissima 0.913 0.007 0.674 0.013 Meconema_thalassinum 0.897 0.010 0.648 0.026 Mecostethus_parapleurus 0.886 0.017 0.628 0.028 Metrioptera_bicolor 0.923 0.016 0.708 0.039 Metrioptera_brachyptera 0.891 0.021 0.638 0.037 Metrioptera_roeselii 0.804 0.028 0.455 0.055 Metrioptera_saussuriana 0.935 0.010 0.754 0.019 Miramella_alpina 0.890 0.023 0.640 0.042 Miramella_formosanta 0.968 0.012 0.869 0.031 Nemobius_sylvestris 0.895 0.012 0.636 0.031 Oecanthus_pellucens 0.954 0.006 0.782 0.021 Oedipoda_caerulescens 0.945 0.014 0.761 0.043 Oedipoda_germanica 0.947 0.012 0.760 0.031 Omocestus_haemorrhoidalis 0.922 0.026 0.731 0.054 Omocestus_rufipes 0.925 0.018 0.709 0.050 Omocestus_viridulus 0.852 0.024 0.566 0.040 Phaneroptera_falcata 0.925 0.009 0.719 0.021 Phaneroptera_nana 0.978 0.002 0.903 0.015 Pholidoptera_aptera 0.901 0.026 0.650 0.059 Pholidoptera_griseoaptera 0.817 0.023 0.469 0.045 Platycleis_albopunctata 0.910 0.015 0.669 0.041 Podisma_pedestris 0.902 0.016 0.668 0.031

- 204 -

Polysarcus_denticauda 0.948 0.013 0.797 0.035 Psophus_stridulus 0.920 0.012 0.710 0.025 Ruspolia_nitidula 0.966 0.006 0.835 0.016 Sphingonotus_caerulans 0.953 0.007 0.797 0.027 Stauroderus_scalaris 0.888 0.021 0.644 0.038 Stenobothrus_lineatus 0.854 0.020 0.559 0.035 Tetrix_bipunctata 0.853 0.018 0.556 0.037 Tetrix_tenuicornis 0.863 0.011 0.558 0.026 Tettigonia_cantans 0.881 0.015 0.611 0.030 Tettigonia_viridissima 0.856 0.020 0.549 0.042

- 205 -

Table S13. Correlation between variables. Spearman correlation between functional trait

variables. CA = correspondence analyses.

SLA LDMC Toughness CN Height Chemicalrichness Chemicaldiversity Axis CA 1 Axis CA 2 Axis CA 3 Axis CA 4 SLA 1.00 -0.45 -0.14 -0.43 -0.09 0.01 0.07 0.01 0.03 -0.04 -0.18 LDMC -0.45 1.00 0.64 0.43 0.33 -0.15 -0.16 -0.16 0.06 0.09 0.24 Toughness -0.14 0.64 1.00 0.51 0.30 -0.24 -0.29 -0.26 -0.03 0.03 0.54 CN -0.43 0.43 0.51 1.00 0.02 -0.29 -0.19 0.04 0.08 -0.07 0.52 Height -0.09 0.33 0.30 0.02 1.00 -0.31 -0.17 -0.14 0.20 -0.03 0.21 Chemical richness 0.01 -0.15 -0.24 -0.29 -0.31 1.00 0.18 -0.35 -0.71 0.44 -0.17 Chemical diversity 0.07 -0.16 -0.29 -0.19 -0.17 0.18 1.00 0.12 0.14 0.06 -0.40 CA Axis 1 0.01 -0.16 -0.26 0.04 -0.14 -0.35 0.12 1.00 0.63 -0.60 -0.25 CA Axis 2 0.03 0.06 -0.03 0.08 0.20 -0.71 0.14 0.63 1.00 -0.54 -0.03 CA Axis 3 -0.04 0.09 0.03 -0.07 -0.03 0.44 0.06 -0.60 -0.54 1.00 0.01 CA Axis 4 -0.18 0.24 0.54 0.52 0.21 -0.17 -0.40 -0.25 -0.03 0.01 1.00

- 206 -

Table S14. Changes in plant biomass associated to plant traits under control cage (cage effect). Changes in plant biomass index under the control cage (Cage) and ambient control, expressed as the difference in total number of plant hits between 2017 and 2014 inventories, associated to plant traits and assessed using a MCMCglmm. Significant variables are highlighted in bold. CA = correspondence analyses, *** P < 0.001, ** P < 0.01, * P < 0.05, . P < 0.1

post.mean l-95% CI u-95% CI eff.samp pMCMC (Intercept) -0.487 -8.528 6.740 4700 0.888 Cage -0.624 -2.019 0.823 4700 0.363 LDMC 0.585 -1.023 2.377 4928 0.495 Toughness -2.747 -4.478 -1.154 5108 0.001 *** SLA 1.585 0.503 2.757 4364 0.005 ** CN 1.411 0.039 2.803 4700 0.048 * Height 1.220 0.132 2.361 4916 0.034 * Chemical richness -0.577 -1.653 0.500 4761 0.305 Chemical diversity -0.260 -1.245 0.826 4993 0.620 CA Axis 1 0.218 -0.953 1.369 5150 0.708 CA Axis 3 -0.378 -1.421 0.781 4700 0.501 CA Axis 4 -1.976 -3.642 -0.218 3704 0.025 * Cage : LDMC -0.343 -2.405 1.809 4700 0.757 Cage : Toughness 0.304 -1.778 2.273 4700 0.770 Cage : SLA 0.066 -1.407 1.484 4923 0.944 Cage : CN -0.726 -2.474 0.987 4700 0.436 Cage : Height -0.501 -1.817 0.889 4480 0.474 Cage : Chemical richness -0.023 -1.403 1.258 4700 0.977 Cage : Chemical diversity 0.062 -1.274 1.360 4700 0.920 Cage : CA Axis 1 -0.069 -1.480 1.206 4797 0.923 Cage : CA Axis 3 0.067 -1.241 1.357 5668 0.931 Cage : CA Axis 4 1.243 -0.631 3.025 4700 0.183

- 207 -

Figure S1. Pictures of the field sites and treatments. The experiment took place on three field sites in the Swiss Alps at (A) 1800 m, (B) 2070 m and (C) 2270 m. The warming treatment was applied by using (D) open-top chamber greenhouses (ground diameter: 111 cm, top opening diameter: 60 cm, height: 38 cm) and the herbivore translocation treatment by using (E) cages (ground cover: 70 cm x 70 cm, height: 50 cm) and an herbivore community collected at lower elevation (1400 m).

- 208 -

Figure S2. Mean temperature records. Mean temperature measured on three field sites (A = 1800m, B = 2070m, C = 2270m) during the July-August period in 2017 and collected using loggers installed at 20 cm above the ground surface outside the OTC greenhouses (To, n=8), inside the OTC greenhouses (Ti, n=8), in the ambient control (C, n=4), in the empty control cage (Cage, n=4), in the cage with the insect herbivores (H, n=4) and at 1.5 m above the ground surface (Air, n=2).

- 209 -

Figure S3. Diurnal and nocturnal temperatures. Mean nocturnal (23h-5h; A, C and E) and diurnal (11h-17h; B, D and F) temperature measured on three field sites (A and B = 1800m, C and D = 2070m, E and F = 2270m) during the July-August period in 2017 and collected using loggers installed at 20 cm above the ground surface outside the OTC greenhouses (To, n=8), inside the OTC greenhouses (Ti, n=8), in the ambient control (C, n=4), in the empty control cage (Cage, n=4), in the cage with the insect herbivores (H, n=4) and at 1.5 m above the ground surface (Air, n=2).

- 210 -

Figure S4. Species metabolomics correspondence analysis (CA). Representation of major sources of variability among plant species based on the full binarized metabolomics survey. Species scores are represented along (A) axes 1 and 2, (B) axes 1 and 3, (C) and axes 1 and 4. Species belonging to the same family are grouped into the ellipses. The red colour represent plant species and plant families from the Liliopsida clade (monocotyledons), the blue colour from the Magnoliopsida clade (dicotyledons) and the grey colour plant species from the Filicopsida clade (ferns).

- 211 -

Figure S5. Leaf attacks. Total number of leaf attacks under (A-C) natural ambient herbivore communities (A = ambient, B = Cage, C = Warming) and (D) translocated herbivore communities. The total number of plant leaf attacks in plots was estimated across sites and replicates (mean and standard deviation). The red and blue colour represent Liliopsida (monocotyledons) and Magnoliopsida (dicotyledons) plant species, respectively. Only species with at least 3 replicates (i.e. plot herbivory measurements) are represented on this graph.

- 212 -

CHAPTER 5

Simulated shifts in trophic niche breadth modulate range loss of alpine butterflies under climate change

Patrice Descombes1,2,3, Jean-Nicolas-Pradervand4, Joaquim Golay4, Antoine Guisan4,5, Loïc Pellissier1,2,3

1Unit of Ecology & Evolution, University of Fribourg, Fribourg, Switzerland 2Landscape Ecology Institute of Terrestrial Ecosystems, ETH Zürich, Zürich, Switzerland 3Swiss Federal Research Institute WSL Birmensdorf, Switzerland 4Department of Ecology and Evolution University of Lausanne, Lausanne, Switzerland 5Institute of Earth Surface Dynamics, University of Lausanne, Lausanne, Switzerland

Published in Ecography (2016), 39, 796-804 doi: https://doi.org/10.1111/ecog.01557 Post-print version

- 213 -

Abstract

Species currently track suitable abiotic and biotic conditions under ongoing climate change. Adjustments of trophic interactions may provide a mechanism for population persistence, an option that is rarely included in model projections. Here, we model the future distribution, of butterflies in the western Alps of Switzerland under climate change, simulating potential diet expansion resulting from adaptive behavior or new host opportunities. We projected the distribution of 60 butterfly and 298 plant species with species distribution models (SDMs) under three climate change scenarios. From known host plants, we allowed a potential diet expansion based on phylogenetic constraints. We assessed whether diet expansion could reduce the rate of expected regional species extinction under climate change. We found that the risk of species extinctions decreased with a concave upward decreasing shape when expanding the host plant range. A diet expansion to even a few phylogenetically closely related host plants would significantly decrease extinction rates. Yet, even when considering expansion toward all plant species available in the study area, the overall regional extinction risk would remain high. Ecological or evolutionary shifts to new host plants may attenuate extinction risk, but the severe decline of suitable abiotic conditions is still expected to drive many species to local extinction.

Introduction

Species currently respond to climate change by tracking favorable environments in the landscape (Parmesan 2006, Walther et al. 2010, Schweiger et al. 2012). The assumption that species are bound to particular environmental conditions fostered the development of species distribution models (SDMs). These models fit a multidimensional niche volume from correlations between occurrence and environmental conditions and use it to project species distributions (Guisan and Thuiller 2005). By changing the input climatic maps in the models, the projections of future colonization and extinction events under climate change can be derived (Thuiller et al. 2005, Pradervand et al. 2014). However, accumulating evidence indicates that evolutionary or ecological changes may modify how species respond to environmental conditions over very few generations (Thuiller et al. 2013a). An ecological or evolutionary adaptive response may thus be a possible alternative mechanism for population rescue under climate change (Pateman et al. 2012). It is not clear whether adaptive changes can operate fast enough to prevent species from extinction. Because of this uncertainty, it may be important to include adaptation potential scenarios, such as changes in biotic interactions, into climate change impact assessments. The combination of these scenarios with SDMs has been proposed (Thuiller et al. 2013a), but rarely used to assess species sensitivity to climate change (but see Brooker et al. 2007, Kearney et al. 2009, Van der Putten et al. 2010).

- 214 -

Many herbivores, such as butterfly, are shifting their distribution poleward (Devictor et al. 2012) and to higher elevation in response to climate change. Species have moved upward with an elevational shift of up to 300 m from 1967–2005 (Wilson et al. 2007, Merrill et al. 2008), and projections under expected climate change suggest further elevational shifts of 650 m by 2100 (Merrill et al. 2008). In addition, due to shrinking suitable areas, as much as 60% of the species may be lost in some mountain ranges (Thuiller et al. 2005, Engler et al. 2011). However, SDMs are criticized because they are often an oversimplified estimation of species distribution, as they may only consider abiotic factors in future forecasts (Guisan and Thuiller 2005, Wisz et al. 2013). By contrast, evidence indicates that biotic interactions, especially trophic interactions, can affect species distributions under climate change (Araújo and Luoto 2007, Pellissier et al. 2012a, 2012b, Schweiger et al. 2012, Eskildsen et al. 2015). In addition, microevolution could mitigate the effects of climate change. Process‐based models of tree phenology incorporating the divergence of phenological responses across species ranges predict less‐severe shifts in species distribution in response to climate change than niche‐based models (Morin and Chuine 2006, Morin and Thuiller 2009). While biotic interactions in general should be included when modeling the response of species to climate change (Nogués‐Bravo and Rahbek 2011, Kissling et al. 2012, Wisz et al. 2013), potential shifts in biotic interactions should also be considered.

Under climate change, evolutionary responses may especially modulate biotic interactions (Lavergne et al. 2010). For example, climate change may induce new interactions between species that never co‐occurred or evolutionary shifts in species traits modulation interactions (Reznick and Ghalambor 2001, Gilman et al. 2010, Lavergne et al. 2010). Herbivores are frequently specialized on a restricted range of host plants, as they require detoxification mechanisms to digest plant material containing secondary metabolites (Pellissier et al. 2013a, 2013b, Rasmann et al. 2014). Restricted host plant ranges are thought to reduce the ability of herbivores to colonize new geographical regions, especially if their hosts are rare or patchily distributed in the landscape (Pöyry et al. 2009). As a consequence, the possibility of partial release from biotic constraints via host‐ plant shifts may allow species to better track changing climates in the landscape and increase survival under climate change. For instance, Pateman et al. (2012) documented that the evolution of a larger larval host‐plant range of the butterfly Aricia agestis has facilitated rapid range expansion under climate change. Diet expansion can be the result of evolutionary responses (i.e. adaptation to new chemical compounds that broaden the diet to new plant clades) or the result of new, previously pre‐adapted herbivory opportunities (i.e. sharing the same chemical compounds as the host plant). The constraint of trophic interactions on species responses to climate change is well documented (Araújo and Luoto 2007, Schweiger et al. 2010, 2012, Romo et al. 2014, Eskildsen et al. 2015). Yet,

- 215 -

it remains unknown how changes in herbivores trophic regimes may modify species responses to climate change.

One possible approach is to simulate shifts in diet breadth and assess the degree of diet expansion required to rescue an herbivore species pool from climate change. Here, we used high‐ resolution maps of environmental predictors derived from weather stations and three IPCC‐based climate change scenarios to model the response of a group of specialist herbivores (butterflies) to future warming in mountain region. The goals of the study were to investigate 1) the constrain of trophic interactions on the future distribution of a regional butterfly fauna relative to abiotic conditions and 2) in which manner host plant expansion could decrease extinction risks under climate change.

Material and methods

Study area and data collection

The study area is located in the western Alps of Switzerland (Supplementary material Appendix 1, Fig. A1) and encompasses a wide elevational gradient (375 to 3200 m a.s.l.). The species data were collected using a random stratified sampling method (Hirzel and Guisan 2002) based on slope, elevation and exposure. The vegetation sampling included 912 sites surveyed exhaustively in 2 × 2 m squares across the entire gradient (Dubuis et al. 2011), whereas the butterflies were sampled in a subset of 192 of these sites from 1000 to 3200 m, employing the same sampling strategy (Pellissier et al. 2013a, 2013b). Butterflies were sampled in 50 × 50 m plots during summer 2009 and 2010 when the conditions were optimal (i.e. low wind, minimum temperature of 18°C, and between 10:00 and 17:00 during the hours of high butterfly activity; Pollard and Yates 1993).

Environmental predictor variables

To model plant and butterfly species distributions, we used predictors known to have a strong influence on species distributions and we considered only variables with low correlation (< 0.7) to avoid collinearity. We selected the climatic variables based on biological knowledge of the target groups (Boggs and Inouye 2012, Roland and Matter 2013), their use in previous studies (Engler and Guisan 2009, Pellissier et al. 2012a, 2012b) and a preliminary analysis of variable importance in the models (Supplementary material Appendix 1, Fig. A2). To model plant distribution, we used the temperature of the growing season (averaged from July to September), winter precipitation (averaged from January to March) representing snow cover, solar radiation, slope and topographic

- 216 - position as proxies for soil and wetness conditions (Engler and Guisan 2009, Randin et al. 2009). Soil variables, although important for plants (Dubuis et al. 2013), were not included as no spatially‐ explicit map is available for the study area. To model butterfly distribution, we used temperatures of the growing season, solar radiation as energy influx for larval growth and winter precipitation as snow cover indicator (Boggs and Inouye 2012). The temperature and precipitation values were obtained by relating information from weather stations to elevation as described in Zimmermann and Kienast (1999). For the current conditions, we used the 1981–2009 average that corresponds to the reference period used by the Centre for Climate System Modelling (www.c2sm.ethz.ch/). We computed the total amount of solar radiation (direct + diffuse + reflected) received by each pixel for the growing season using the spatial analyst tool in ArcGIS 10 (ESRI). We calculated the slope (degree) and the topographic position (degree of convexity/ concavity) with the spatial analyst tool in ArcGIS. Variables were computed at a resolution of 25 m to model plant species (558 452 pixels) and 50 m to model butterfly species (139 613 pixels), representing a total surface of 349 km2 (Supplementary material Appendix 1, Fig. A3).

Species distribution models

We modeled the distribution of 298 plant and 60 butterfly species presenting a minimum of 10 presences and absences in the study area. To model species distributions, we used three common statistical techniques shown to provide efficient predictions of species distributions (Elith et al. 2006): generalized linear models (GLMs; McCullagh and Nelder 1989), generalized additive models (GAMs; Hastie and Tibshirani 1990), and gradient boosting machines (GBMs; Ridgeway 1999, Friedman 2001). All the models were computed using the biomod2 R package with parameters optimized for species distribution modeling (Thuiller et al. 2013b). For model validation, we used the area under the ROC‐plot curve (AUC; Hanley and Mcneil 1982, Fielding and Bell 1997) and the true skill statistic (TSS; Allouche et al. 2006), which evaluate the ability of the model to discriminate presences from absences. Models are considered to have reliable prediction performances with AUC values > 0.7 and TSS values > 0.4 (Thuiller et al. 2009). Data were split randomly into two partitions: 80% was used for model calibration, and the remaining 20% was used for model evaluation. This procedure was replicated 10 times. We then averaged all the model projections presenting an AUC value > 0.7 to build a total consensus binary map using the sensitivity = specificity binarization criteria (Liu et al. 2005). It is expected that the accuracy of the model may vary with species traits (e.g. dispersal; Eskildsen et al. 2013, 2015). We only considered those species with at least one model with good predictive abilities (AUC > 0.7), excluding several species with long dispersal abilities potentially not associated with local habitats

- 217 -

during the monitoring (Geiger 1987). The variable importance in the models was calculated as one minus the correlation of the original model and the model with the given variable randomized (Thuiller et al. 2013b).

Climate change scenarios

We used three different climate projection scenarios (RCP3PD, A1B and A2) averaged for three climatic models (ARPEGE‐ALADIN, ECHAM5‐REMO, HadCM3Q0‐CLM). We used a unique averaged future projection time period (i.e. 2085 from the 2070–2099) developed in the Swiss Climate Change Scenario CH2011 project from the Center for Climate Systems Modeling (www.c2sm.ethz.ch/). They are based on new generations of climate models at high resolution combining global and regional models (Bosshard et al. 2011). These climatic anomalies for every weather station in Switzerland were interpolated using the same approach as the climatic data and then added to the maps of the current climate. We projected all the plant and butterfly species distributions for the three different scenarios. We assumed unlimited dispersal of species shown to be a close approximation to using dispersal kernels in mountains (Engler and Guisan 2009).

Butterfly interaction with host plants

We compiled a list of host plants for each of the butterfly species considered (see Pellissier et al. 2012a, 2012b, 2013a, 2013b for further details). Phylogenetic conservatism in plant defense traits implies that closely related species produce similar secondary compound classes and require similar detoxification mechanisms in herbivores (Becerra 1997). Herbivores are thus more likely to enlarge their diet to closely related host plant species. To support this assumption, we tested whether the phylogenetic distance between the host plants of each butterfly species was lower than expected by chance using the ‘ses.mpd’ function in the ‘picante’ package with a tip label randomization (Kembel et al. 2010). Phylogenetic clustering in host plant choice would support the use of simulating diet expansion based on phylogenetic relationships.

Diet expansion simulations

We considered two approaches to simulate diet expansion. First, we inferred the probability of trophic interaction between butterfly and plant species pairs using a statistical latent trait model (Pellissier et al. 2013a, 2013b). Probabilities of interaction between butterfly and plant species are derived from the observed matrix of trophic interactions and related to plant traits and both butterfly (Pellissier et al. 2012a, 2012b) and plant phylogenies (Ndiribe et al. 2013). The model is then

- 218 - extrapolated to the entire butterfly plant network to obtain a matrix of interaction probabilities for each butterfly and plant species pair (see Pellissier et al. 2013a, 2013b for further details). The potential trophic expansion can then be simulated by changing the interaction binary threshold in the linkage probability matrix (e.g. a threshold of 0.9 only allows interacting pairs with probabilities > 0.9). Second, we simulated diet expansion directly along a phylogenetic axis without the use of a statistical model. For each butterfly species, host plant range was broadened along a scale from zero to the most distant species pairs (i.e. 270 myr) with a 1 myr step resolution (e.g. a threshold of 10 myr only allows interacting host plants which are at a phylogenetic distance < 10 myr).

To account for the obligate trophic interaction between the butterfly and its host plants, we filtered the binary maps of the butterfly projections by those of the host plants as performed by Schweiger et al. (2008, 2012, Supplementary material Appendix 1, Fig. A4). We computed the number of species gaining or losing occupied surface under climate change and the number of species at risk of extinction along varying diet expansion thresholds. Butterfly species were considered as vulnerable to extinction when their range decreased below a threshold of 5% of the total available open habitats (i.e. total open habitats = 349 km2; 5% of total open habitats < 17.5 km2, results for thresholds of 1 and 10% available in Supplementary material Appendix 1).

Results

Model performance

Most of the plant distribution models were reliable with mean AUC values of 0.860 ± 0.050 and mean TSS values of 0.685 ± 0.099. All 298 plant species had at least one model with AUC > 0.7, except Cerastium arvense. Butterfly SDMs had a mean AUC value of 0.799 ± 0.081 and mean TSS values of 0.615 ± 0.154. Several species such as Papilio machaon showed weak predictive abilities with the three modelling techniques used. We considered a final subset of 60 butterflies with at least one good model among the three statistical techniques (Supplementary material Appendix 1, Table A1). The examination of climatic variables showed that the average winter precipitation and average summer temperature were the most important variables in the models for plant, and butterflies (Supplementary material Appendix 1, Fig. A5).

Phylogenetic structure in host plant use

We found that butterfly species with more than one host plant showed phylogenetic clustering in host plant choice (Fig. 1) with the exception of Aricia artaxerxes (2 host plants; ses.mpd = 0.25; p‐value = 0.43), Boloria napaea (2 host plants; ses.mpd = 0.26; p‐value = 0.38), Callophrys rubi (8

- 219 -

host plants; ses.mpd = −1.77; p‐value = 0.06), Erebia pluto (2 host plants; ses.mpd = −2.08; p‐value = 0.08) and Plebejus argus (6 host plants; ses.mpd = −1.04; p‐value = 0.12). In addition, the interaction matrix structure was significantly correlated with the plant and butterfly phylogenies in the trophic interaction model (phylogenetic regressions: plants latent traits 1 and 2: Pagel's‐λ = 0.8217, p‐value < 0.001; Pagel's‐λ = 0.6497, p‐value < 0.001, and for butterfly latent traits 1 and 2: Pagel's‐λ = 0.7257, p‐value < 0.001; Pagel's‐λ = 0.9230, p‐value < 0.001, Pellissier et al. 2013a, 2013b). This supports the view that herbivore diet show phylogenetic constraints, which is used to simulate diet expansion.

Figure 1. Biplot of the number of host plants in the diet of each butterfly species together with the standardized effect size of the mean phylogenetic distance (ses.mpd) between the host plants for each butterfly species. Negative values of ses.mpd indicate a phylogenetic clustering in host plant preference. Non‐significant standardized effect sizes are represented by grey points.

Projections of abiotic and biotic models

When considering the butterflies’ abiotic projections, the median loss in area between 2010 and 2085 was −21.5% (quantile 5% = −79.8%, quantile 95% = 185.1%), −64.8% (−98.9%, 163.1%) and −72.7% (−99.9%, 170.5%) under the RCP3PD, A1B and A2 scenarios, respectively. The percentage of species at risk of extinction with a predicted distribution area < 5% reached 10, 32 and 38% (Supplementary material Appendix 1, Table A2).

- 220 -

Constraining the butterfly abiotic distribution with the distribution of their known host plants, lead to more restricted butterflies current (median = −19%, quantile 5% = −0.2%, quantile 95% = −92.8%) and future projections (RCP3PD: median = −15.3%, quantile 5% = −92.3%, quantile 95% = −0.9%; A1B: −17.3%, −98.8%, −0.3%; A2: −22.6%, −99.7%, −0.3%) compared to abiotic models. The median surface loss between current and 2085 biotic projections reached −25.2% (quantile 5% = −84.6%, quantile 95% = 229.1%), −76.8% (−99.4%, 194.6%) and −87% (−99.9%, 193%) under the RCP3PD, A1B and A2 scenarios, respectively. Constraining the butterfly abiotic distribution by known current trophic interaction increased the percentage of species at risk of extinction in 2085 from 10 to 20% for RCP3PD, from 32 to 52% for A1B and from 38 to 60% for A2 (Supplementary material Appendix 1, Table A2) and modified future butterfly species richness in the area (Supplementary material Appendix 1, Fig. A6).

Response to diet expansion

Simulations of butterfly host plant diet expansion under climate change showed a concave upward decreasing function in the relationship between degree of expansion and percent of species with risk of local extinction (extinction threshold 5%: Fig. 2). Changing the selected extinction threshold to 1 or 10% of the total available open habitats did not change the shape of the curve (Supplementary material Appendix 1, Fig. A7, A8, Table A2). In contrast to a linear relationship or concave downward relationship, the decreasing concave upward shape indicates that diet expansion to even a few closely phylogenetically related host plants can significantly decrease the extinction rates. A conservative diet expansion to plants with interaction probabilities higher than 0.9 (number of species in the diet: mean = 14.9, SD = 16.5) reduced the percent of butterfly species with extinction risk under climate change from 20 to 19% for RCP3PD, from 52 to 46% for A1B and from 60 to 51% for A2 (Fig. 2a, c, e, and Supplementary material Appendix 1, Table A2), decreased the number of species presenting range loss from 66 to 58% for RCP3PD, from 69 to 66% for A1B and from 71 to 67% for A2 (Supplementary material Appendix 1, Fig. A9a), and increased the forecasted butterfly species richness in cells (Fig. 3). Similarly, enlarging the diet of the butterfly to plants with a phylogenetic distance lower than 20 myr (number of species in the diet: mean = 8.9, SD = 8.9) reduced the percent of butterfly species with high extinction risk under climate change from 20 to 18% for RCP3PD, from 52 to 43% for A1B and from 60 to 48% for A2 (Fig. 2b, d, f) and decreased the number of species presenting range loss from 66 to 60% for RCP3PD, from 69 to 65% for A1B and from 71 to 67% for A2 (Supplementary material Appendix 1, Fig. A9b). Considering an unlikely diet expansion to all plants (interaction probability thresholds = 0 and phylogenetic distance = 270) corresponds to the abiotic projections presented above.

- 221 -

Figure 2. Percent of butterfly species with high extinction risk (black line) for the 2085 RCP3PD (a and b), A1B (c and d), A2 (e and f) climate change scenarios when considering host plants diet expansion based on (a, c and e) modelled interaction probabilities between the butterfly and the plant and (b, d and f) phylogenetic distances between plants. The mean accumulated number of host plants (red line) included in the diet is shown on the right axis (number of host species) where the dashed lines represent the 5 and 95 percentiles. An interaction probability threshold of 1 and a phylogenetic distance of 0 means that only the known host plants from the literature are considered in the diet. Curves were fitted with a GAM function.

- 222 -

Figure 3. Predicted future (i.e. 2085) butterfly species richness in the study area in response to different diet expansion scenarios. The butterfly abiotic distribution was constrained with the abiotic distribution of host plants under the A2 climate change scenario when considering (a) only their known host plants, (b) a diet expansion scenario based on butterfly–plant interaction probabilities > 0.9 and (c) a diet expansion scenario based on butterfly–plant interaction probabilities > 0.5. (d) and (e) represent the changes in species richness induced by (d) a diet expansion scenario based on butterfly–plant interaction probabilities > 0.9 (i.e. b minus a) and (e) a diet expansion scenario based on butterfly–plant interaction probabilities > 0.5 (i.e. c minus a). To avoid problems of truncated response curves, projections are shown only over an elevation of 1600 m, as a conservative estimate of the thermal elevational shifts under the most extreme climate change scenario.

The scenario of diet expansion based on the interaction probability matrix was more restrictive than the one based purely on phylogenetic distance as shown by the intermediate values of the axis (Fig. 2). Enlarging the diet of the butterfly to plants with an interaction probability higher than 0.5 (number of species in the diet: mean = 21.5, SD = 20.5) reduced the percent of butterfly species with high extinction risk under climate change from 20 to 15% for RCP3PD, from 52 to 41% for A1B and from 60 to 49% for A2 (Fig. 2a, c, e, Fig. 3 and Supplementary material Appendix 1, Table A2) and decreased the number of species presenting range loss from 66 to 56% for RCP3PD, from 69 to 66% for A1B and from 71 to 66% for A2 (Supplementary material Appendix 1, Fig. A9a). In contrast, enlarging the diet of the butterfly species to plants with a

- 223 -

phylogenetic distance lower than 135 myr (number of species in the diet: mean = 22.5, SD = 15) reduced the percent of butterfly species with high extinction risk under climate change for 2085 from 20 to 15% for RCP3PD, from 52 to 35% for A1B and from 60 to 38% for A2 (Fig. 2b, d, f) and decreased the number of species presenting range loss from 66 to 42% for RCP3PD, from 69 to 58% for A1B and from 71 to 63% for A2 (Supplementary material Appendix 1, Fig. A9b).

Discussion

Currently, species are tracking suitable abiotic conditions under climate change (Parmesan et al. 1999, Hickling et al. 2006), but future persistence will also depend on the availability of trophic resources (Araújo and Luoto 2007, Van der Putten et al. 2010, Schweiger et al. 2012). The emergence of new trophic interactions may influence species response to climate change (Pateman et al. 2012, Blois et al. 2013). We show that the response of the number of species at risk of local extinction to simulated diet expansion follow a concave upward decreasing function (Fig. 2). The shape of the response curve suggests that the possibility to feed on even a few new host plants has a rapid attenuation effect – hereby the ‘rescue effect’ – on the percentage of species with high extinction risk (Fig. 3). Novel interaction may reduce species extinction risk, potentially supporting a rapid effect of adaptive responses on species persistence under climate change (Thuiller et al. 2013a).

While diet expansion in herbivores is possible within a short time period (< 75 yr; Strauss et al. 2006, Pateman et al. 2012), novel trophic interactions expected under climate change or invasions are difficult to predict (Pearse and Altermatt 2013). Statistical models of trophic interactions likelihood (Cattin et al. 2004, Pellissier et al. 2013a, 2013b, Albouy et al. 2014, Rohr and Bascompte 2014) may be used to model the effects of likely novel interactions under climate change. Using the whole range of diet expansion possibilities, we studied the shape of the response between species at risk of extinction and diet expansion. The concave upward decreasing curve found using both approaches (i.e. interactions matrix and direct phylogenetic distance) suggests that even a low degree of diet expansion may reduce the number of species at risk of extirpation from the regional fauna. Our results in an alpine region contrast with an assessment at European scale, where only few species where limited by host plant distribution under climate change (Schweiger et al. 2012). This argues for considering species adaptive potential related to biotic interactions when forecasting species responses to climate change at least at smaller scale (Van der Putten et al. 2010, Thuiller et al. 2013a).

- 224 -

We found differences in the two diet expansion approaches considered. While the phylogenetic distance approach was solely based on the host plant phylogeny, the modelled interaction probability matrix was based on information provided by host plant and butterfly phylogenies together with plant species traits (i.e. leaf nitrogen; Pellissier et al. 2013a, 2013b). This more ecologically constrained model showed a lower mitigating effect compared to the simulation based directly on phylogenetic distance (Fig. 2). While many secondary metabolites are phylogenetically conserved on plant clades (Wink 2003), other plant traits such as physical defense or leaf resource content are less conserved (Ndiribe et al. 2013) and may also modulate plant– herbivore interactions (Pellissier et al. 2012a, 2012b, Ibanez et al. 2013). Integration of trophic interactions within future projection should provide more realistic models of species interactions including the relevant ecological traits modulating species interactions (Pellissier et al. 2013a, 2013b, Ibanez et al. 2013, Albouy et al. 2014, Morales‐Castilla et al. 2015).

Even when considering the potential adaptation of butterflies to new host plants, the percent of species with high extinction risk remained high. Thus, diet expansion to new host plants will not be able to completely mitigate the loss of suitable habitats due to climate change. In mountain areas, tracking suitable climatic conditions is not a long lasting solution, since suitable surfaces decrease upward the elevation gradient (Theurillat and Guisan 2001). The most cold‐adapted species currently restricted to alpine habitats, such as Plebejus glandon, are forecasted to become extinct regionally irrespective of host plant diet expansion due to the lack of suitable cold habitats in a warmer future. Even taking into account an unlikely diet expansion toward all available plant species in the study area, the number of species with high extinction risk remained elevated, at an estimated proportion of 10% for RCP3PD, 32% for A1B and 38% for A2 (Fig. 2). Adaptive responses to new host plants may thus attenuate the risk of extinction of butterfly species, but only to a limited degree while climate change may still largely drive species range shifts and range contraction (Schweiger et al. 2012).

The current study could be improved in several aspects. For simplifications, we considered unlimited dispersal for both host plants and butterfly species. While unlimited dispersal for butterflies might be realistic given their current observed range shift (Eskildsen et al. 2013, 2015), plants dispersal is slower than that of herbivore species (Rasmann et al. 2014). Adding a dispersal kernel for plants when tracking climate change in the landscape might provide more realistic assessment of range mismatches. Furthermore, studies have shown that soil variables (Dubuis et al. 2013) or land‐use information (Randin et al. 2009, Schweiger et al. 2012) might also drive host plant distribution, so that future projections should consider refined environmental variables to provide realistic assessments. Finally, in the current study, we assumed that the host plant species

- 225 -

list obtained from the literature is exhaustive (Pellissier et al. 2013a, 2013b), but the literature may be incomplete. In such case, the ‘no expansion’ scenario might be too restrictive especially for more generalist species were omissions are more likely (Pellissier et al. 2013a, 2013b).

In this study, we forecasted butterfly species distributions under climate change considering possible adaptation and demonstrated that changes in biotic interactions may modulate species responses. Mitigation effects through the evolutionary and/or ecological capacity of butterflies to enlarge their diet to new host plants might be a possible mechanism to prevent some species from local extinction. While new interactions under climate change have been documented, it remains to be seen whether trophic changes are common enough to make a difference. The understanding and prediction of interactions between herbivores and host plants using functional traits remain at an early stage (Ibanez et al. 2013, Pellissier et al. 2013a, 2013b). Further studies of the mechanistic and chemical coupling between plant and herbivores are required to forecast future species interactions under climate change and refine the host plant diet expansion scenario developed in this study.

Acknowledgements

We thank the reviewers, especially Oliver Schweiger for highly valuable comments. We would like to thank Alexander Von Ungern, Anne Dubuis, Pascal Vittoz, Virginie Favre, Valéry Udry, Sarah Giovanettina, Vanessa Rion, Christian Purro, Saskia Godat and Aurore Gelin for field assistance. We thank Glenn Litsios for his help with the phylogenetic analysis. This study was supported by the European Commission (ECOCHANGE project, contract no. FP6 2006 GOCE 036866) and by the Swiss National Science Foundation (SNSF) grants no. 31003A‐125145 (BIOASSEMBLE project) and no. 31003A‐152866 (SESAM'ALP project) accorded to AG.

References

Albouy, C. et al. 2014. From projected species distribution to food‐web structure under climate change. – Global Change Biol. 20: 730–741. Allouche, O. et al. 2006. Assessing the accuracy of species distribution models: prevalence, kappa and the true skill statistic (TSS). – J. Appl. Ecol. 43: 1223–1232. Araújo, M. B. and Luoto, M. 2007. The importance of biotic interactions for modelling species distributions under climate change. – Global Ecol. Biogeogr. 16: 743–753. Becerra, J. X. 1997. Insects on plants: macroevolutionary chemical trends in host use. – Science 276: 253–256.

- 226 -

Blois, J. L. et al. 2013. Climate change and the past, present, and future of biotic interactions. – Science 341: 499–504. Boggs, C. L. and Inouye, D. W. 2012. A single climate driver has direct and indirect effects on insect population dynamics. – Ecol. Lett. 15: 502–508. Bosshard, T. et al. 2011. Spectral representation of the annual cycle in the climate change signal. – Hydrol. Earth Syst. Sci. 15: 2777–2788. Brooker, R. W. et al. 2007. Modelling species’ range shifts in a changing climate: the impacts of biotic interactions, dispersal distance and the rate of climate change. – J. Theor. Biol. 245: 59–65. Cattin, M. F. et al. 2004. Phylogenetic constraints and adaptation explain food‐web structure. – Nature 427: 835–839. Devictor, V. et al. 2012. Differences in the climatic debts of birds and butterflies at a continental scale. – Nat. Clim. Change 2: 121–124. Dubuis, A. et al. 2011. Predicting spatial patterns of plant species richness: a comparison of direct macroecological and species stacking modelling approaches. – Divers. Distrib. 17: 1122– 1131. Dubuis, A. et al. 2013. Improving the prediction of plant species distribution and community composition by adding edaphic to topo‐climatic variables. – J. Veg. Sci. 24: 593–606. Elith, J. et al. 2006. Novel methods improve prediction of species’ distributions from occurrence data. – Ecography 29: 129–151. Engler, R. and Guisan, A. 2009. MigClim: predicting plant distribution and dispersal in a changing climate. – Divers. Distrib. 15: 590–601. Engler, R. et al. 2011. 21st century climate change threatens mountain flora unequally across Europe. – Global Change Biol. 17: 2330–2341. Eskildsen, A. et al. 2013. Testing species distribution models across space and time: high latitude butterflies and recent warming. – Global Ecol. Biogeogr. 22: 1293–1303. Eskildsen, A. et al. 2015. Ecological specialization matters: long‐term trends in butterfly species richness and assemblage composition depend on multiple functional traits. – Divers. Distrib. 21: 792–802. Fielding, A. H. and Bell, J. F. 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. – Environ. Conserv. 24: 38–49. Friedman, J. H. 2001. Greedy function approximation: a gradient boosting machine. – Ann. Stat. 29: 1189–1232. Geiger, W. 1987. Les papillons de jour et leurs biotopes. – Ligue Suisse pour la protection de la Nature, Bâle, Switzerland.

- 227 -

Gilman, S. E. et al. 2010. A framework for community interactions under climate change. – Trends Ecol. Evol. 25: 325–331. Guisan, A. and Thuiller, W. 2005. Predicting species distribution: offering more than simple habitat models. – Ecol. Lett. 8: 993–1009. Hanley, J. A. and Mcneil, B. J. 1982. The meaning and use of the area under a receiver operating characteristic (ROC) curve. – Radiology 143: 29–36. Hastie, T. and Tibshirani, R. 1990. Generalized additive models. – Stat. Sci. 1: 297–318. Hickling, R. et al. 2006. The distributions of a wide range of taxonomic groups are expanding polewards. – Global Change Biol. 12: 450–455. Hirzel, A. and Guisan, A. 2002. Which is the optimal sampling strategy for habitat suitability modelling. – Ecol. Model. 157: 331–341. Ibanez, S. et al. 2013. Herbivory mediated by coupling between biomechanical traits of plants and grasshoppers. – Funct. Ecol. 27: 479–489. Kearney, M. et al. 2009. Integrating biophysical models and evolutionary theory to predict climatic impacts on species’ ranges: the dengue mosquito Aedes aegypti in Australia. – Funct. Ecol. 23: 528–538. Kembel, S. W. et al. 2010. Picante: R tools for integrating phylogenies and ecology. – Bioinformatics 26: 1463–1464. Kissling, W. D. et al. 2012. Towards novel approaches to modelling biotic interactions in multispecies assemblages at large spatial extents. – J. Biogeogr. 39: 2163–2178. Lavergne, S. et al. 2010. Biodiversity and climate change: integrating evolutionary and ecological responses of species and communities. – Annu. Rev. Ecol. Evol. Syst. 41: 321–350. Liu, C. et al. 2005. Selecting thresholds of occurrence in the prediction of species distributions. – Ecography 28: 385–393. McCullagh, P. and Nelder, J. A. 1989. Generalized linear models. – Chapman and Hall. Merrill, R. M. et al. 2008. Combined effects of climate and biotic interactions on the elevational range of a phytophagous insect. – J. Anim. Ecol. 77: 145–155. Morales‐Castilla, I. et al. 2015. Inferring biotic interactions from proxies. – Trends Ecol. Evol. 30: 347–356. Morin, X. and Chuine, I. 2006. Niche breadth, competitive strength and range size of tree species: a trade‐off based framework to understand species distribution. – Ecol. Lett. 9: 185–195. Morin, X. and Thuiller, W. 2009. Comparing niche‐ and process‐based models to reduce prediction uncertainty in species range shifts under climate change. – Ecology 90: 1301–1313. Ndiribe, C. et al. 2013. Phylogenetic plant community structure along elevation is lineage specific. – Ecol. Evol. 3: 4925–4939.

- 228 -

Nogués‐Bravo, D. and Rahbek, C. 2011. Communities under climate change. – Science 334: 1070– 1071. Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. – Annu. Rev. Ecol. Evol. Syst. 37: 637–669. Parmesan, C. et al. 1999. Poleward shifts in geographical ranges of butterfly species associated with regional warming. – Nature 399: 579–583. Pateman, R. M. et al. 2012. Temperature‐dependent alterations in host use drive rapid range expansion in a butterfly. – Science 336: 1028–1030. Pearse, I. S. and Altermatt, F. 2013. Predicting novel trophic interactions in a non‐native world. – Ecol. Lett. 16: 1088–1094. Pellissier, L. et al. 2012a. Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. – Ecol. Evol. 2: 1818–1825. Pellissier, L. et al. 2012b. Climate‐based empirical models show biased predictions of butterfly communities along environmental gradients. – Ecography 35: 684–692. Pellissier, L. et al. 2013a. Turnover of plant lineages shapes herbivore phylogenetic beta diversity along ecological gradients. – Ecol. Lett. 16: 600–608. Pellissier, L. et al. 2013b. Combining food web and species distribution models for improved community projections. – Ecol. Evol. 3: 4572–4583. Pollard, E. and Yates, T. J. 1993. Monitoring butterflies for ecology and conservation. – Chapman and Hall. Pöyry, J. et al. 2009. Species traits explain recent range shifts of Finnish butterflies. – Global Change Biol. 15: 732–743. Pradervand, J.‐N. et al. 2014. Functional homogenization of bumblebee communities in alpine landscapes under projected climate change. – Clim. Change Responses 1: 1–10. Randin, C. F. et al. 2009. Introduction of snow and geomorphic disturbance variables into predictive models of alpine plant distribution in the western Swiss Alps. – Arct. Antarct. Alp. Res. 41: 347–361. Rasmann, S. et al. 2014. Climate‐driven change in plant–insect interactions along elevation gradients. – Funct. Ecol. 28: 46–54. Reznick, D. N. and Ghalambor, C. K. 2001. The population ecology of contemporary adaptations: what empirical studies reveal about the conditions that promote adaptive evolution. – Genetica 112–113: 183–198. Ridgeway, G. 1999. The state of boosting. – Comput. Sci. Stat. 31: 172–181. Rohr, R. P. and Bascompte, J. 2014. Components of phylogenetic signal in antagonistic and mutualistic networks. – Am. Nat. 184: 556–564.

- 229 -

Roland, J. and Matter, S. F. 2013. Variability in winter climate and winter extremes reduces population growth of an alpine butterfly. – Ecology 94: 190–199. Romo, H. et al. 2014. Effects of climate change on the distribution of ecologically interacting species: butterflies and their main food plants in Spain. – Ecography 37: 1063–1072. Schweiger, O. et al. 2008. Climate change can cause spatial mismatch of trophically interracting species. – Ecology 89: 3472–3479. Schweiger, O. et al. 2010. Multiple stressors on biotic interactions: how climate change and alien species interact to affect pollination. – Biol. Rev. Camb. Phil. Soc. 85: 777–795. Schweiger, O. et al. 2012. Increasing range mismatching of interacting species under global change is related to their ecological characteristics. – Global Ecol. Biogeogr. 21: 88–99. Strauss, S. Y. et al. 2006. Evolutionary responses of natives to introduced species: what do introductions tell us about natural communities? – Ecol. Lett. 9: 357–374. Theurillat, J. P. and Guisan, A. 2001. Potential impact of climate change on vegetation in the European Alps: a review. – Clim. Change 50: 77–109. Thuiller, W. et al. 2005. Climate change threats to plant diversity in Europe. – Proc. Natl Acad. Sci. USA 102: 8245–8250. Thuiller, W. et al. 2009. BIOMOD – a platform for ensemble forecasting of species distributions. – Ecography 32: 369–373. Thuiller, W. et al. 2013a. A road map for integrating eco‐evolutionary processes into biodiversity models. – Ecol. Lett. 16 (Suppl. 1): 94–105. Thuiller, W. et al. 2013b. biomod2: ensemble platform for species distribution modeling. – R package ver. 3.1–18, < http://CRAN.R‐project.org >. Van der Putten, W. H. et al. 2010. Predicting species distribution and abundance responses to climate change: why it is essential to include biotic interactions across trophic levels. – Phil. Trans. R. Soc. B 365: 2025–2034. Walther, G.‐R. et al. 2010. Ecological repsonses to recent climate change. – Nature 416: 389–395. Wilson, R. J. et al. 2007. An elevational shift in butterfly species richness and composition accompanying recent climate change. – Global Change Biol. 13: 1873–1887. Wink, M. 2003. Evolution of secondary metabolites from an ecological and molecular phylogenetic perspective. – Phytochemistry 64: 3–19. Wisz, M. S. et al. 2013. The role of biotic interactions in shaping distributions and realised assemblages of species: implications for species distribution modelling. – Biol. Rev. Camb. Phil. Soc. 88: 15–30. Zimmermann, N. E. and Kienast, F. 1999. Predictive mapping of alpine grasslands in Switzerland: species versus community approach. – J. Veg. Sci. 10: 469–482.

- 230 -

Supplementary materials

Table A1. Performance of the butterfly abiotic models under two evaluators (TSS and AUC) as the average of ten replicates.

Species TSS AUC

GLM GBM GAM GLM GBM GAM Mean Sd Mean Sd Mean Sd Mean Sd Mean Sd Mean Sd

Anthocharis 0.573 0.103 0.558 0.084 0.680 0.119 0.791 0.073 0.805 0.076 0.862 0.046 cardamines Aphantopus 0.824 0.068 0.810 0.112 0.846 0.068 0.925 0.046 0.915 0.049 0.961 0.017 hyperantus Argynnis adippe 0.566 0.216 0.434 0.173 0.817 0.076 0.748 0.136 0.670 0.120 0.890 0.064 Argynnis aglaja 0.537 0.117 0.524 0.075 0.582 0.089 0.790 0.070 0.742 0.054 0.821 0.056 Argynnis niobe 0.431 0.132 0.399 0.127 0.742 0.066 0.651 0.091 0.670 0.070 0.874 0.041

Aricia artaxerxes 0.332 0.131 0.339 0.089 0.585 0.100 0.609 0.079 0.601 0.093 0.797 0.050

Aricia eumedon 0.366 0.081 0.351 0.080 0.552 0.058 0.647 0.072 0.651 0.060 0.792 0.043

Boloria euphrosyne 0.466 0.103 0.437 0.091 0.670 0.059 0.736 0.070 0.714 0.069 0.871 0.036

Boloria napaea 0.675 0.121 0.722 0.132 0.815 0.087 0.827 0.076 0.857 0.068 0.923 0.043 Boloria pales 0.778 0.068 0.742 0.103 0.822 0.060 0.917 0.037 0.895 0.053 0.942 0.023 Boloria titania 0.589 0.115 0.533 0.098 0.631 0.121 0.822 0.060 0.800 0.057 0.856 0.057 Brenthis ino 0.723 0.106 0.682 0.174 0.886 0.033 0.827 0.077 0.806 0.112 0.929 0.020 Callophrys rubi 0.433 0.073 0.585 0.143 0.786 0.073 0.620 0.048 0.776 0.052 0.891 0.051 Coenonympha 0.513 0.103 0.614 0.087 0.686 0.090 0.780 0.060 0.815 0.050 0.868 0.052 gardetta Coenonympha 0.771 0.084 0.769 0.076 0.815 0.083 0.897 0.058 0.886 0.054 0.941 0.038 pamphilus

Colias alfacariensis 0.560 0.192 0.620 0.242 0.940 0.049 0.753 0.130 0.805 0.107 0.965 0.025

Colias hyale 0.503 0.090 0.482 0.085 0.653 0.100 0.742 0.051 0.714 0.069 0.847 0.052

Colias phicomone 0.567 0.095 0.573 0.071 0.649 0.079 0.788 0.063 0.787 0.034 0.855 0.029

Cupido minimus 0.442 0.159 0.450 0.176 0.535 0.139 0.704 0.111 0.714 0.101 0.786 0.091

Erebia aethiops 0.410 0.053 0.411 0.092 0.480 0.085 0.676 0.034 0.694 0.058 0.771 0.060

Erebia cassioides 0.742 0.195 0.750 0.204 0.997 0.009 0.823 0.130 0.824 0.141 0.997 0.009

Erebia epiphron 0.735 0.073 0.684 0.087 0.906 0.063 0.842 0.052 0.838 0.046 0.959 0.032 Erebia euryale 0.528 0.141 0.449 0.168 0.613 0.102 0.735 0.071 0.688 0.072 0.812 0.054 Erebia gorge 0.872 0.092 0.884 0.096 0.962 0.034 0.957 0.028 0.932 0.076 0.984 0.014 Erebia ligea 0.613 0.117 0.615 0.078 0.724 0.092 0.810 0.082 0.805 0.050 0.866 0.045 Erebia manto 0.558 0.150 0.493 0.120 0.578 0.156 0.818 0.088 0.766 0.081 0.831 0.093

Erebia melampus 0.281 0.079 0.533 0.090 0.664 0.089 0.605 0.075 0.781 0.042 0.855 0.047

Erebia oeme 0.402 0.085 0.458 0.125 0.518 0.072 0.703 0.065 0.742 0.070 0.795 0.047

Erebia pandrose 0.686 0.115 0.586 0.146 0.746 0.097 0.854 0.060 0.799 0.064 0.887 0.049

Erebia pharte 0.590 0.137 0.583 0.105 0.659 0.116 0.795 0.085 0.803 0.078 0.878 0.051 Erebia pluto 0.884 0.154 0.899 0.108 0.980 0.024 0.941 0.082 0.955 0.062 0.994 0.007 Erebia pronoe 0.404 0.120 0.529 0.122 0.831 0.079 0.612 0.103 0.735 0.054 0.920 0.028

- 231 -

Erebia tyndarus 0.626 0.127 0.676 0.122 0.886 0.063 0.802 0.057 0.828 0.079 0.951 0.038

Euphydryas aurinia 0.569 0.125 0.565 0.086 0.744 0.092 0.792 0.060 0.775 0.051 0.883 0.034

Hamearis lucina 0.747 0.199 0.597 0.335 0.978 0.018 0.822 0.116 0.733 0.241 0.983 0.017

Lasiommata maera 0.365 0.113 0.373 0.096 0.535 0.079 0.649 0.064 0.647 0.058 0.807 0.045

Lasiommata 0.340 0.153 0.275 0.178 0.712 0.100 0.589 0.133 0.580 0.129 0.863 0.051 petropolitana

Leptidea realsin 0.659 0.089 0.647 0.073 0.742 0.087 0.846 0.043 0.832 0.035 0.884 0.037

Lycaena hippothoe 0.558 0.106 0.487 0.138 0.700 0.087 0.766 0.105 0.748 0.084 0.850 0.067

Lycaena tityrus 0.471 0.143 0.396 0.139 0.894 0.056 0.719 0.081 0.679 0.084 0.935 0.039

Maculinea arion 0.422 0.124 0.352 0.085 0.703 0.110 0.681 0.085 0.616 0.066 0.874 0.062

Maniola jurtina 0.785 0.088 0.755 0.090 0.823 0.078 0.934 0.034 0.909 0.032 0.950 0.026 Melanargia 0.744 0.185 0.704 0.173 0.906 0.042 0.864 0.116 0.854 0.112 0.972 0.018 galathea

Melitaea athalia 0.724 0.083 0.671 0.104 0.795 0.075 0.878 0.035 0.851 0.042 0.921 0.030

Melitaea diamina 0.559 0.088 0.623 0.088 0.670 0.069 0.778 0.102 0.827 0.070 0.885 0.035

Pieris brassicae 0.424 0.138 0.358 0.143 0.606 0.151 0.688 0.078 0.622 0.097 0.822 0.082 Pieris bryoniae 0.412 0.121 0.478 0.094 0.544 0.085 0.671 0.080 0.737 0.058 0.798 0.043 Pieris napi 0.508 0.133 0.535 0.166 0.572 0.125 0.793 0.079 0.769 0.085 0.820 0.081 Pieris rapae 0.314 0.097 0.286 0.147 0.524 0.138 0.629 0.063 0.592 0.098 0.766 0.069 Plebejus argus 0.279 0.142 0.359 0.091 0.829 0.110 0.521 0.108 0.636 0.082 0.909 0.056

Plebejus glandon 0.617 0.195 0.559 0.212 0.954 0.031 0.746 0.130 0.763 0.101 0.969 0.025

Plebejus orbitulus 0.755 0.176 0.683 0.225 0.992 0.014 0.847 0.110 0.808 0.093 0.994 0.010

Polyommatus 0.544 0.216 0.503 0.255 0.935 0.039 0.725 0.157 0.695 0.202 0.962 0.029 bellargus Polyommatus 0.424 0.124 0.283 0.082 0.570 0.127 0.673 0.079 0.573 0.089 0.801 0.066 coridon Polyommatus 0.256 0.129 0.324 0.196 0.544 0.075 0.514 0.126 0.553 0.169 0.759 0.073 damon

Polyommatus eros 0.420 0.109 0.358 0.157 0.792 0.113 0.631 0.080 0.620 0.095 0.908 0.040

Polyommatus 0.557 0.059 0.513 0.130 0.729 0.105 0.784 0.033 0.761 0.069 0.895 0.052 icarus Polyommatus 0.527 0.090 0.519 0.090 0.704 0.099 0.794 0.055 0.789 0.045 0.902 0.050 semiargus Polyommatus 0.617 0.162 0.600 0.175 1.000 0.000 0.771 0.089 0.739 0.133 1.000 0.000 thersites Pontia callidice 0.789 0.203 0.728 0.299 0.953 0.032 0.862 0.111 0.826 0.188 0.962 0.033

- 232 -

Table A2. Changes in percentage of butterflies’ species with high extinction risk when considering different thresholds (i.e. <1%, <5% and <10%) that define the minimal area species can occupy in the study area before being considered under risk of extinction. The values are shown for the 2010 and the future climate change scenarios (RCP3PD, A1B and A2) and different model types.

Climate change Extinction risk thresholds (percent of Model type scenario open habitats)

<1% <5% <10% 2010 0 2 12 RCP3PD 2085 0 10 18 Abiotic models A1B 2085 13 32 43 A2 2085 22 38 52 2010 2 20 30

Host plant constrained RCP3PD 2085 3 20 38 models A1B 2085 30 52 63 A2 2085 40 60 67 2010 0 12 24 Host plant constrained RCP3PD 2085 2 19 32 models with diet expansion (int. prob > 0.9) A1B 2085 24 46 58 A2 2085 34 51 64 2010 0 5 17 Host plant constrained RCP3PD 2085 2 15 27 models with diet expansion (int. prob > 0.5) A1B 2085 24 41 54 A2 2085 32 49 63

- 233 -

Figure A1. Location of the study area in the western Alps of Switzerland. The dots (black and red) represent vegetation sampling sites. The red dots correspond to the vegetation sites where butterflies were sampled. The light gray line shows the limits of study area. The dark gray line shows the 1000 m isoline (models lower limit).

- 234 -

Figure A2. Importance of seasonal climatic variables (temperature and precipitation) in the niche models for plants (a) and butterflies (b). The average winter precipitations (January to March) and summer temperatures (July to September) were the most influencing variables and are therefore good predictors for the species niche modelling. The variable importance was calculated using biomod2 and represents 1 minus the correlation of one model using the variable and one model using the variable randomized. The intermediate line is the median, and the limits of the boxes are the 1st and 3rd quartiles. The lines and the points show the minimum and maximum and the outliers.

- 235 -

Figure A3. Final area for model projections at a resolution of 50 m (139613 pixels) representing a total surface of 349 km2.

- 236 -

Figure A4. Schematic view of the modeling process from the modelling of individual butterfly and plant species distributions to their combinations considering or not diet expansion.

- 237 -

Figure A5. Importance of topo-climatic predictors in the niche models for plants (a) and butterflies (b) used in the final models. The average winter precipitations (January to March) and summer temperatures (July to September) were the most important predictors for both plants and butterflies. The variable importance was calculated using biomod2 and represents 1 minus the correlation of one model using the variable and one model using the variable randomized. The intermediate line is the median, and the limits of the boxes are the 1st and 3rd quartiles. The lines and the points show the minimum and maximum and the outliers.

- 238 -

Figure A6. Predicted current (i.e. 2010) butterfly species richness in the study area when (a) considering only butterfly abiotic distribution, and when (b) constraining butterfly abiotic distribution with the abiotic distribution of their known host plants. Constraining the butterfly abiotic distribution by known current trophic interaction decreases the butterfly species richness.

- 239 -

Figure A7. Percent of butterfly species with high extinction risk for the 2085 butterfly host plant constrained predictions under the A2 (black plain line), A1B (black long dashed) and RCP3PD (black short dashed) climate change scenarios when considering an enlargement of the host plants diet based on (a) interaction probabilities between the butterfly and the plants and (b) phylogenetic distances between plants. Species were considered in danger of extinction when their predicted area does not reach 1% of the total available open habitats (i.e., < 3.5 km2). The mean accumulated number of host plants (red line) included in the diet is shown on the right axis (number of host species) where the dashed lines represent the 5 and 95 percentiles. An interaction probability threshold of 1 and a phylogenetic distance of 0 means that only the known host plants from the literature are considered in the diet. Curves were fitted with a GAM function.

- 240 -

Figure A8. Percent of butterfly species with high extinction risk for the 2085 butterfly host plant constrained predictions under the A2 (black plain line), A1B (black long dashed) and RCP3PD (black short dashed) climate change scenarios when considering an enlargement of the host plants diet based on (a) interaction probabilities between the butterfly and the plants and (b) phylogenetic distances between plants. Species were considered in danger of extinction when their predicted area does not reach 10% of the total available open habitats (i.e., < 34.9 km2). The mean accumulated number of host plants (red line) included in the diet is shown on the right axis (number of host species) where the dashed lines represent the 5 and 95 percentiles. An interaction probability threshold of 1 and a phylogenetic distance of 0 means that only the known host plants from the literature are considered in the diet. Curves were fitted with a GAM function.

- 241 -

Figure A9. Percent of butterfly species presenting habitat loss (in term of surface) for the 2085 butterfly host plant constrained predictions under the A2 (black plain line), A1B (black long dashed) and RCP3PD (black short dashed) climate change scenarios when considering an enlargement of the host plants dietbased on (a) interaction probabilities between the butterfly and the plants and (b) phylogenetic distances between plants. The mean accumulated number of host plants (red line) included in the diet is shown on the right axis (number of host species) where the dashed lines represent the 5 and 95 percentiles. An interaction probability threshold of 1 and a phylogenetic distance of 0 means that only the known host plants from the literature are considered in the diet. Curves were fitted with a GAM function.

- 242 -

Figure A10. Percent of butterfly species with high extinction risk for the 2085 butterfly host plant constrained predictions under the A2 (black plain line), A1B (black long dashed), RCP3PD (black short dashed) climate change scenarios when considering an enlargement of the host plants diet based on (a) interaction probabilities between the butterfly and the plants and (b) phylogenetic distances between plants. Species were considered in danger of extinction when their predicted area does not reach 5% of the total available open habitats (i.e., < 17.5 km2). The extinction rate was calculated as the mean of ten replicated simulation of enlargement of the host plants diet that was limited by the number of known host plants (maintaining the distinction between generalist and specialist species in term of host plant number of species in the diet). The mean accumulated number of host plants (red line) included in the diet is shown on the right axis (number of host species) where the dashed lines represent the 5 and 95 percentiles. An interaction probability threshold of 1 and a phylogenetic distance of 0 means that only the known host plants from the literature are considered in the diet. Curves were fitted with a GAM function.

- 243 -

- 244 -

CONCLUSION

The goal of this thesis was to provide a better understanding on how plant and insect communities will respond to changes in trophic interactions under climate change, and to give new insights into the potential impact of range shifting herbivores on alpine plant communities. In particular, this thesis contributed to a better understanding of (i) the structuration of plant communities along the elevation gradient, (ii) the factors influencing plant-herbivore interactions, (iii) the diet conservatism of insect herbivores, and (iv) the response of alpine plant communities to altered trophic interactions under climate change.

Plant communities along the elevation gradient

This thesis contributed to a better understanding of the structuration of plant communities along the elevation gradient (Chapters 1 and 2). In particular, we identified a peak in plant taxonomic and phylogenetic turnover near the treeline (1900 m) (Gehrig-Fasel et al., 2007; Chapter 1), corresponding to a transitional zone that delimits low- and high-elevation plant communities (Körner, 2003). While community compositional changes have been observed along temperature and moisture gradients for many types of organisms such as plants (de Bello et al., 2012; Gentry, 1988; Pellissier et al., 2010), animals (Kergunteuil et al., 2016; Pellissier et al., 2012; Sanders, 2002) and microorganisms (Pellissier et al., 2014), only a few studies have reported how the rate of assemblage turnover varies along the elevation gradient (Mena and Vázquez-Domínguez, 2005; Bach et al., 2007; Jankowski et al., 2009, 2013). The acceleration of species turnover near the treeline indicates that this ecotone constitutes a strong barrier for some plant clades, which share analogous range limits (Chapter 1; Leibold and Mikkelson, 2002) likely modulated by variation in abiotic (Diaz and Cabido, 2009; Körner, 2007) and biotic (Alexander et al., 2015; Bruelheide and Scheidel, 1999) factors along the elevation gradient. In addition, abiotic factors may restrict plant species assemblage to a set of functionally adapted plants, thus leading to functionally distinct communities along the elevational gradient (e.g., high elevation plants with small stature and slow growth rates) (Diaz and Cabido, 2009; Körner, 2003). Hence, we found variations in plant functional traits and an overall decrease in plant defence against herbivores at higher elevation, associated to variation in abiotic conditions or a direct release of plant defences due to lower herbivore pressure on high elevation sites (Chapter 2). This result is in line with a large number of studies and recent reviews (Galmán et al., 2018; Moreira et al., 2018 and references herein). Nevertheless, contrasting patterns of herbivory and defence traits have also been reported along elevation gradients, so that broad generalizations of the factors governing the relationship between elevation and herbivory or plant traits are not yet possible (Moreira et al., 2018 and references

- 245 -

herein). Together, the variation in plant defence and herbivore pressure along the elevation gradient (Chapter 2), and the strong compositional turnover detected near the treeline, which reflect different species assemblages (Chapter 1), suggest that plant communities near and above the treeline might face very strong changes in the coming years with the upward shift of species under climate change. Manipulative experiments of plant and herbivore communities across this transitional zone (Chapter 1), and a careful understanding of the factors modulating plant-herbivore interactions might provide new insights to predict the effects of climate change on species assemblages.

Factors influencing plant-herbivore interactions

This thesis provides a better understanding of the factors (e.g., physical and chemical functional traits, environmental context) controlling herbivore performance and modulating plant- herbivore interactions (Chapters 3 and 4). By using recent developments in analytical chemistry (untargeted metabolomics) for a large range of naturally-growing plant species, we provide the evidence of a strong relationship between herbivory and plant family-specific chemical signatures (Chapters 3 and 4). The novelty compared to previous studies relies on the fact that we analyzed the entire chemical signature of a very large set of naturally-growing plant species (i.e., > 157 species) spanning different plant families (35 families). Hence, previous studies typically measured a few groups of chemical compounds, such as flavonoids, phenolics or cardenolides (e.g., Callis-Duehl et al., 2017; Pellissier et al., 2016; Rasmann and Agrawal, 2011), which are unlikely to summarize the overall complex and diverse chemical defence in plants (Mithöfer and Boland, 2012; Rhoades, 1979). Other studies used bioassays with highly polyphagous insect herbivores as a proxy of plant chemical toxicity (Pellissier et al., 2012; Pérez‐Harguindeguy et al., 2003). More recently, plant metabolomics analyses were performed over a few number of plant species, or within the same genus, mostly in tropical trees (e.g., Piper, Inga) (Coley et al., 2018; Richards et al., 2015; Salazar et al., 2018). However, to our knowledge, none investigated the chemical signature of a large set of plant species co-occurring in natural grasslands and covering a large number of plant families. Our results also suggest that family-specific chemical signatures are phylogenetically conserved and influence herbivory in a consistent way in both natural grasslands and bioassay experiments (Chapter 3). The use of untargeted metabolomics analyses in Chapters 3 and 4 is a major step forward in understanding the drivers of plant-herbivore interactions, as it allowed to identify family patterns in plant defences. Beyond chemical traits, classical physical (SLA, toughness) and nutritional (C:N) parameters were also associated to herbivore performance, but to a lower degree than chemical traits or with inconsistencies between experiments (Chapters 3 and 4). A phylogenetic conservatism in chemical traits (as observed in Chapter 3) implies that closely related

- 246 - plant species have similar secondary compounds (Becerra, 1997). Hence, phylogenetic conservatism in chemical traits might drive conservatism of plant–herbivore interactions in closely related plant species under climate change (Becerra, 1997; Futuyma and Agrawal, 2009), as herbivores evolved specific detoxification mechanisms to digest those particular chemical compounds (Becerra, 1997). Understanding the factors influencing plant-herbivore interactions is of main importance for predicting future new interactions and their consequences on communities under climate change (Chapters 4 and 5).

Diet conservatism of insect herbivores

The fourth chapter provides the evidence for a diet conservatism of insect herbivores (Chapter 4) when feeding on typical sets of low- and high-elevation plant species growing across the transitional zone that delimits lowland and alpine ecosystems (Chapter 1). The feeding preferences of lowland herbivores were found to be functionally and phylogenetically conserved on new host plants whose traits matched their trophic preferences in their habitat of origin (Chapter 4). This result is in line with previous studies showing conserved phylogenetic interaction patterns between herbivores and plants (Becerra, 1997; Ehrlich and Raven, 1964; Pearse and Altermatt, 2013; Pellissier et al., 2013). Because plant traits tend to be conserved across the tips of the phylogenies (e.g., chemical traits, toughness) (Chapter 3), trophic shifts might preferentially occur in closely related plant species (Chapter 4; Becerra, 1997; Ehrlich and Raven, 1964; Futuyma and Agrawal, 2009). Beyond theory, we provide experimental evidence of a conserved diet of herbivores when migrating to higher elevation in a context of climate change (Chapter 4). Because climate change is expected to modify trophic interactions via asynchronous range shifts of species (Berg et al., 2010; Rasmann et al., 2014; Van der Putten et al., 2010), and because herbivores can have profound and variable impact on plant communities (Belovsky and Slade, 2000; Duffy et al., 2007; Olff and Ritchie, 1998), there is an urgent need to account for novel arising interactions between resident and range-shifting species in community ecology. In this regard, and in light of our results, phylogenetic and functional information might be combined to predict future trophic interactions under climate change (Loranger et al., 2012; Pearse and Altermatt, 2013; Pellissier et al., 2013). Predicting potential, future trophic interactions might help understand the response of plant (Chapter 4) and insect herbivore communities (Chapter 5) to climate change. Hence, Chapter 5 revealed that dietary shifts might be a key mechanism for herbivore population persistence under global warming. Together, Chapters 4 and 5 provide the evidence that trophic interactions and trophic shifts between species that never co-occurred should be fully integrated in community forecasts under climate change to make reliable predictions. This study paves the way toward a better understanding of the

- 247 -

mechanisms that will shape future trophic interactions under climate change and their associated consequences on ecosystem functions and processes.

Predictability of the response of plant communities to altered trophic interactions

This thesis investigated and experimented the consequences of climate-driven herbivore incursion on high-elevation plant communities by translocating a lowland, natural herbivore community on high elevation grasslands (Chapter 4). Because the translocated herbivores tended to conserve their diet on phylogenetically and functionally similar plant species as in their native range, alpine plant communities responded in a predictable way and herbivore incursions represented a stronger driver of community compositional change than temperature warming (Chapter 4). The simulated increase in herbivory and the different diet preferences of the translocated herbivore community compared to the natural herbivores on the recipient sites induced a shift in the selective pressure on plants in the high elevation grasslands, resulting in strong compositional and structural changes. Together, the outcomes from Chapter 4 are novel and highlight the predictability of the response of plant communities to altered trophic interactions under climate change. These results have an important impact in the research area of ecological community modelling, since conserved ecological rules might allow the use of mechanistic predictive models of food-webs to simulate future plant community responses to altered trophic interactions under climate change. However, before that, future research in this field should attempt to characterize extant plant-herbivore trophic networks and collect species phylogenies and functional traits. For instance, the characterization of the whole trophic network of grasshopper (or other herbivore) species along elevation transects, together with plant and insect functional trait information, might allow to predict future interactions and impact of grasshopper herbivore species migrating in alpine ecosystems. In light of the possible strong, top-down impact of new herbivores on plants, forecasting the dynamic of plants in communities under climate change should account for the effect of new ecosystem incursions.

Limitations

Short-term experiment

Several aspects not addressed in this thesis would need careful consideration in future research. The relative short-term nature of the experiment presented in Chapter 4 (4 years) might underestimate future compositional and structural changes in the plant community. Several years of reduced primary productivity under herbivore selective grazing or increased temperature warming

- 248 - might decrease the fitness of some plant species and reshape the strength of the competition between plants (Alexander et al., 2015; Bruelheide, 2003; Panetta et al., 2018). In addition, it also remains uncertain whether the conserved feeding behavior of the herbivores will persist over a longer time period (Chapter 4), as new ecological interactions might drive the physiological and morphological adaptation of herbivores to new host plants (Agrawal, 2000). While possible behavioral and evolutionary shifts in herbivore feeding preferences might alter our capacity to predict novel interactions, the ongoing rapid climate change may also prevent species to evolve the necessary adaptations (Davis and Shaw, 2001; Etterson and Shaw, 2001). Longer-term experiments (10-20 years) might enable to sense slower ecological processes such as the colonization of more competitive, warm-adapted species, which are predicted to drastically reshape plant communities (Alexander et al., 2015), or extinction of plant species in response to decreased fitness (Bruelheide, 2003; Panetta et al., 2018).

Different dispersal rates

This thesis did not investigate the potential effect of different dispersal rates between plants and herbivores. In Chapter 5, we considered unlimited dispersal of plants and insects species in niche modelling along the elevation gradient, representing an unrealistic, best-case scenario where both plants and butterflies are able to track their abiotic niche. However, low-disperser species might not be able to pace with the current rapid climate change, which is much faster than those observed during post-glacial times (Malcolm et al., 2002). In particular, plants have been shown to have a slower response to climate change than insects, leading to asynchronous range shifts (Berg et al., 2010; de Sassi and Tylianakis, 2012). Adding dispersal kernels to plants might provide a more realistic estimation of the range mismatches between butterflies and their host plants (Engler and Guisan, 2009), and might severely increase butterfly extinction rates when no trophic adjustments are allowed (Chapter 5). As a consequence, range mismatches under future climate change could be very strong between species showing different dispersal capacities (Schweiger et al., 2008), thereby favoring range declines if species are not able to adapt to novel host plants (Pelini et al., 2010) or facilitating range expansion if species are able to adjust their diet (Pateman et al., 2012). In addition, other factors not considered in this thesis might trigger insect dispersal and their capacity to follow optimal climatic conditions, such as landscape barriers (Robinet and Roques, 2010), landscape fragmentation (Pitelka and Plant Migration Workshop Group, 1997) and species range size (Lewthwaite et al., 2018). For instance, colonization processes might be implemented via species- specific dispersal kernels based on dispersal modes and plant traits (Vittoz and Engler, 2007). In addition, the experimental settings in Chapter 4 follow recent research evidence that insects will

- 249 -

track climate change along elevation or latitude at a faster rate than plants, which lag behind (Alexander et al., 2018; Berg et al., 2010), thus leading to reshuffled communities and new ecological interactions in their new ranges (Berg et al., 2010; Rasmann et al., 2014). However, dispersal rates may also differ between insect herbivore species, so that future orthopteran communities under natural conditions on the experimental sites might be slightly different from those used in this study. As we used a natural orthopteran community of lower elevation, realistic densities and a set of species which are the most frequent and abundant in the study area, our translocated community is probably a good approximation of future orthopteran assemblages on the recipient sites. Nevertheless, future research on the processes affecting species response to climate change, such as competition or dispersal (Urban et al., 2012), might enable better forecasts of species assemblages and future community changes.

Conclusion

In light of the results, we hypothesize that insect herbivore range shifts to higher elevation might be a stronger driver of plant community turnover than temperature warming in alpine ecosystems (Chapter 4). This effect might be particularly strong in subalpine and alpine grasslands near and above the treeline (Chapter 1), where plant communities are less defended against insect herbivores (Chapter 2). Cold-tolerant alpine plants have been shown to have some resilience to abiotic changes such as temperature warming, while being very sensitive to biotic competition with lowland, warm-adapted plant competitors (Alexander et al., 2015, 2018). The upper range limit of lowland plant competitors will shift to higher elevation under climate change (Chen et al., 2011; Rumpf et al., 2018), thus allowing a progressive colonization of subalpine and alpine grasslands, where areas just above the transition zone should experience the strongest community turnover in the coming years (Chapter 1). While plant colonization to higher elevation is expected to be slower for plants than for other trophic groups (Berg et al., 2010), the reduction of plant competition mediated by a first wave of novel insect herbivores and an increase in herbivory (Rasmann et al., 2014) might favor the establishment of novel plant species at higher elevation by creating gaps in the extant communities (Alexander et al., 2018). Therefore, we expect that a reduction in plant biomass and changes in top-down selective pressure on plants within plant communities may, in turn, generate a feedback loop that will facilitate the establishment of lowland plant competitors at higher elevation (Alexander et al., 2018). In addition, the observed increase in plant species richness (Chapter 4) may only be temporary as soon as less competitive, cold-tolerant plant species are replaced by more competitive, warm-adapted species (Alexander et al., 2018). Thus, changes in trophic interactions and herbivore pressure might be key drivers of plant community turnover under

- 250 - climate change, thereby altering ecosystem processes such as nutrient cycling and primary productivity, and foreshadow large functional changes in subalpine and alpine grasslands as climate warming continues.

Consequently, this thesis contributed to a better understanding of the factors influencing plant-herbivore interactions, particularly improving the scientific knowledge about leaf physical and chemical functional traits influencing herbivore performance. While recent studies have underlined the need to account for biotic interaction when forecasting communities under future climate conditions (Davis et al., 1998; Schweiger et al., 2008), we further highlight the necessity to include scenarios of trophic interactions shifts. In particular, this thesis provides the experimental evidence for a diet conservatism during climate-driven range shifts of insect herbivore species on high- elevation plant communities, resulting in predictable responses of plants to new herbivore incursions. Together, the predictability of reshuffled trophic interactions using functional and phylogenetic approaches allows to forecast future community trends under climate change and has important implications for future research in ecological community modelling. Discarding the effects of novel trophic interactions on plant communities might lead to biased predictions under future climate change.

Perspectives

Toward a better understanding of the effect of climate change on plant communities

This thesis did not investigate the combined effect of temperature warming and altered trophic interactions on alpine plant communities (Chapter 4). Hence, climate warming might enhance the metabolism and growth of some plant species (Veteli et al., 2002), which might, in turn, alter the fitness (Panetta et al., 2018), competitive abilities and relative abundances (Lambers and Poorter, 1992) of plants, and affect trophic interactions via altered nutritional content or concentrations of defensive compounds (Coley, 1988; Evans and Burke, 2013; Gutbrodt et al., 2011). We found that temperature warming altered plant traits (Chapters 3 and 4) without associated consequences on plant palatability to herbivores (Chapter 3), thus suggesting that climate warming might not systematically alter plant palatability to herbivores or reshape plant-herbivore interactions (O’Connor, 2009). In contrast, previous studies reported species-specific responses to climate warming with decreasing or increasing herbivore performance (Bidart‐Bouzat and Imeh‐Nathaniel, 2008; Gutbrodt et al., 2011; Lemoine et al., 2013, 2014; Stamp and Yang, 1996). Because a clear understanding of the effect of climate warming on plant palatability and plant-herbivore interactions is still missing, we want to stress the necessity to investigate this question in future research by

- 251 -

focusing on plant physical and chemical traits for larger sets of plant species spanning different plant functional type levels (i.e., evergreen or deciduous shrubs, forbs, grasses and sedges) (Hudson et al., 2011).

In addition, the magnitude of the applied warming treatment (i.e., mean increase of 1.1 K; Chapter 4) also underestimates predicted future temperature increases by the end of the century under a business-as-usual scenario in the study area (rcp85 scenario; + 3.56 K; Chapter 4). In this sense, the translocation of plant communities to warmer, low-elevation conditions along elevation transects (Alexander et al., 2015; Bruelheide, 2003) or heating systems such as infrared radiators (Niu and Wan, 2008; Panetta et al., 2018) might be adequate experimental approaches and tools to control warming conditions. Nevertheless, future research on climate change should also try to account for the multidimensionality of factors affected by global warming, such as changes in temperature, precipitation, nitrogen deposition and trophic interactions (e.g. Alexander et al., 2015; Borer et al., 2014; Boutin et al., 2017; Bruelheide, 2003; Kaarlejärvi et al., 2017). Combining all these factors in experimental studies along elevation gradients might strongly improve our understanding of the ecological processes that will shape plant community changes under climate change. In this regard, we recently (2017) started an experiment along an elevation gradient in the

Figure 1. Translocation experiment along an elevation transect (Calanda, Switzerland). Alpine plant communities were translocated at lower elevation in October 2016. Among the diverse experiments performed on this field site, some translocated alpine plant communities were subjected to a natural high- or low-elevation insect community (e.g. orthopteran, butterflies, flies, bees, aphids, spiders) placed into cages. Photo credit: P. Descombes, ETH Zürich, Switzerland.

- 252 -

Swiss Alps (Calanda, Graubünden) in collaboration with other research groups by translocating coherent pieces of alpine plant communities on several sites at lower elevation (Fig. 1). Therefore, this experiment simulates changes in temperature, precipitation, herbivory intensity and trophic interactions as expected under climate change (Barry, 1992; Körner, 2003, 2007; Rasmann et al., 2014), and might provide new insights into the effect of climate change on plant traits (e.g. plant physical and chemical defence traits), growth, competition and colonization processes into alpine plant communities (Fig. 1).

Investigating different herbivore guilds

This thesis principally focused on orthopteran herbivores, which are part of the top arthropod grazers in alpine ecosystems (Blumer and Diemer, 1996). Contrary to our initial prediction that orthopteran herbivores might shift their diet on rare palatable plants (Chapter 2; Bruelheide, 2003), generalist orthopteran herbivores conserved their diet on abundant plants whose traits were matching their trophic preferences in their habitat of origin (Chapter 4). However, feeding preferences can vary across different herbivore guilds and species (as observed in Chapters 3 and 4). For instance, some specialist herbivore species, such as butterflies and slugs, might have strong impact on rare plants by selectively feeding on them (Bruelheide, 2003; Honek et al., 2017; Maze, 2009; Scheidel and Bruelheide, 2001; Schöps, 2002). As an illustrative example, Bruelheide (2003) observed that the rare alpine plant Arnica montana became extinct under increased slug herbivory when alpine plant turves where translocated under warmer conditions. Beyond insects, herbivores such as small and large mammals have been shown to considerably affect plant communities (Borer et al., 2014; Kaarlejärvi et al., 2017). Furthermore, other trophic levels, such as predators or parasitoids of herbivores, might also be affected by climate change and modulate the response of plant communities via changes in the top-down control of herbivores (Hance et al., 2007; de Sassi and Tylianakis, 2012). In light of the diverse trophic interactions occurring in natural systems, future research is needed to evaluate whether shifts in trophic interactions are predictable across other herbivore guilds and assess their consequences on plant communities by considering multitrophic level interactions.

Metabolomics analyses

Finally, recent developments in analytical chemistry and bioinformatics might allow the detection of metabolites associated to anti-herbivore defence or other functions (Johnson et al., 2014; Salazar et al., 2018). This thesis provides the evidence that untargeted metabolomics analyses are powerful for identifying family patterns in plant defences for a large range of naturally-growing

- 253 -

species (Chapter 3). While we mainly focused on the whole chemical signature of the plant species, the examination and detection of specific chemical compounds shaping family differences in herbivory rates might be a promising and fascinating research direction to investigate. For instance, recent studies are pointing the way by identifying and quantifying chemical compounds associated to herbivory (Johnson et al., 2014; Salazar et al., 2018). The identification of secondary metabolites and attempts to associate them to functions in plants, such as herbivore defence or stress responses, might profoundly improve our understanding of the mechanisms shaping plant-herbivore interactions and the evolution of plant defence. Nevertheless, future research in this field will need to cope with the high dimensionality of the plant chemical profile by using adequate tools and developing pipelines to better identify metabolites based on metabolomics peaks (Barker and Rayens, 2003; Kuhl et al., 2012; Salazar et al., 2018; Tibshirani, 1996).

References

Agrawal, A.A. (2000). Host-Range Evolution: Adaptation and Trade-Offs in Fitness of Mites on Alternative Hosts. Ecology 81, 500–508. Alexander, J.M., Diez, J.M., and Levine, J.M. (2015). Novel competitors shape species’ responses to climate change. Nature 525, 515–518. Alexander, J.M., Chalmandrier, L., Lenoir, J., Burgess, T.I., Essl, F., Haider, S., Kueffer, C., McDougall, K., Milbau, A., Nuñez, M.A., et al. (2018). Lags in the response of mountain plant communities to climate change. Global Change Biology 24, 563–579. Bach K, Kessler M, Gradstein SR (2007) A simulation approach to determine statistical significance of species turnover peaks in a species-rich tropical cloud forest. Divers Distrib 13, 863–870. Barker, M., and Rayens, W. (2003). Partial least squares for discrimination. Journal of Chemometrics 17, 166–173. Barry, R.G. (1992). Mountain Weather and Climate (Psychology Press). Becerra, J.X. (1997). Insects on Plants: Macroevolutionary Chemical Trends in Host Use. Science 276, 253–256. de Bello, F., Lavorel, S., Lavergne, S., Albert, C.H., Boulangeat, I., Mazel, F., and Thuiller, W. (2012). Hierarchical effects of environmental filters on the functional structure of plant communities: a case study in the French Alps. Ecography 36, 393–402. Belovsky, G.E., and Slade, J.B. (2000). Insect herbivory accelerates nutrient cycling and increases plant production. PNAS 97, 14412–14417.

- 254 -

Berg, M.P., Kiers, E.T., Driessen, G., Heijden, M.V.D., Kooi, B.W., Kuenen, F., Liefting, M., Verhoef, H.A., and Ellers, J. (2010). Adapt or disperse: understanding species persistence in a changing world. Global Change Biology 16, 587–598. Bidart‐Bouzat, M.G., and Imeh‐Nathaniel, A. (2008). Global Change Effects on Plant Chemical Defenses against Insect Herbivores. Journal of Integrative Plant Biology 50, 1339–1354. Blumer, P., and Diemer, M. (1996). The Occurrence and Consequences of Grasshopper Herbivory in an Alpine Grassland, Swiss Central Alps. Arctic and Alpine Research 28, 435–440. Borer, E.T., Seabloom, E.W., Gruner, D.S., Harpole, W.S., Hillebrand, H., Lind, E.M., Adler, P.B., Alberti, J., Anderson, T.M., Bakker, J.D., et al. (2014). Herbivores and nutrients control grassland plant diversity via light limitation. Nature 508, 517–520. Boutin, M., Corcket, E., Alard, D., Villar, L., Jiménez, J.-J., Blaix, C., Lemaire, C., Corriol, G., Lamaze, T., and Pornon, A. (2017). Nitrogen deposition and climate change have increased vascular plant species richness and altered the composition of grazed subalpine grasslands. Journal of Ecology 105, 1199–1209. Bruelheide H, Scheidel U (1999) Slug herbivory as a limiting factor for the geographical range of Arnica montana. J Ecol 87, 839–848. Bruelheide, H. (2003). Translocation of a montane meadow to simulate the potential impact of climate change. Applied Vegetation Science 6, 23–34. Callis-Duehl, K., Vittoz, P., Defossez, E., and Rasmann, S. (2017). Community-level relaxation of plant defenses against herbivores at high elevation. Plant Ecol 218, 291–304. Chen, I.-C., Hill, J.K., Ohlemüller, R., Roy, D.B., and Thomas, C.D. (2011). Rapid Range Shifts of Species Associated with High Levels of Climate Warming. Science 333, 1024–1026. Coley, P.D. (1988). Effects of plant growth rate and leaf lifetime on the amount and type of anti- herbivore defense. Oecologia 74, 531–536. Coley, P.D., Endara, M.-J., and Kursar, T.A. (2018). Consequences of interspecific variation in defenses and herbivore host choice for the ecology and evolution of Inga, a speciose rainforest tree. Oecologia 1–16. Davis, M.B., and Shaw, R.G. (2001). Range Shifts and Adaptive Responses to Quaternary Climate Change. Science 292, 673–679. Davis, A.J., Jenkinson, L.S., Lawton, J.H., Shorrocks, B., and Wood, S. (1998). Making mistakes when predicting shifts in species range in response to global warming. Nature 391, 783–786. Diaz, S., and Cabido, M. (2009). Plant functional types and ecosystem function in relation to global change. Journal of Vegetation Science 8, 463–474.

- 255 -

Duffy, J.E., Cardinale, B.J., France, K.E., McIntyre, P.B., Thébault, E., and Loreau, M. (2007). The functional role of biodiversity in ecosystems: incorporating trophic complexity. Ecology Letters 10, 522–538. Ehrlich, P.R., and Raven, P.H. (1964). Butterflies and Plants: A Study in Coevolution. Evolution 18, 586–608. Engler, R., and Guisan, A. (2009). MigClim: Predicting plant distribution and dispersal in a changing climate. Diversity and Distributions 15, 590–601. Etterson, J.R., and Shaw, R.G. (2001). Constraint to Adaptive Evolution in Response to Global Warming. Science 294, 151–154. Evans, S.E., and Burke, I.C. (2013). Carbon and Nitrogen Decoupling Under an 11-Year Drought in the Shortgrass Steppe. Ecosystems 16, 20–33. Futuyma, D.J., and Agrawal, A.A. (2009). Macroevolution and the biological diversity of plants and herbivores. PNAS 106, 18054–18061. Galmán, A., Abdala‐Roberts, L., Zhang, S., Teran, J.C.B.-M. y, Rasmann, S., and Moreira, X. (2018). A global analysis of elevational gradients in leaf herbivory and its underlying drivers: Effects of plant growth form, leaf habit and climatic correlates. Journal of Ecology 106, 413– 421. Gehrig-Fasel J, Guisan A, Zimmermann NE (2007) Tree line shifts in the Swiss Alps: climate change or land abandonment? J Veg Sci 18, 571–582. Gentry, A.H. (1988). Changes in Plant Community Diversity and Floristic Composition on Environmental and Geographical Gradients. Annals of the Missouri Botanical Garden 75, 1– 34. Gutbrodt, B., Mody, K., and Dorn, S. (2011). Drought changes plant chemistry and causes contrasting responses in lepidopteran herbivores. Oikos 120, 1732–1740. Hance, T., Van Baaren, J., Vernon, P., and Boivin, G. (2007). Impact of extreme temperatures on parasitoids in a climate change perspective. Annual Review of Entomology 52, 107–126. Honek, A., Martinkova, Z., Koprdova, S., and Saska, P. (2017). The role of invertebrates in seedling establishment in a heterocarpic plant, Atriplex sagittata. Annals of Applied Biology 171, 264–276. Hudson, J.M.G., Henry, G.H.R., and Cornwell, W.K. (2011). Taller and larger: shifts in Arctic tundra leaf traits after 16 years of experimental warming. Global Change Biology 17, 1013– 1021. Jankowski JE, Ciecka AL, Meyer NY, Rabenold KN (2009) Beta diversity along environmental gradients: implications of habitat specialization in tropical montane landscapes. J Anim Ecol 78, 315–327.

- 256 -

Jankowski JE, Merkord CL, Rios WF et al (2013) The relationship of tropical bird communities to tree species composition and vegetation structure along an Andean elevational gradient. J Biogeogr 40, 950–962. Johnson, M.T.J., Ives, A.R., Ahern, J., and Salminen, J.-P. (2014). Macroevolution of plant defenses against herbivores in the evening primroses. New Phytol. 203, 267–279. Kaarlejärvi, E., Eskelinen, A., and Olofsson, J. (2017). Herbivores rescue diversity in warming tundra by modulating trait-dependent species losses and gains. Nature Communications 8, 419. Kergunteuil, A., Campos-Herrera, R., Sánchez-Moreno, S., Vittoz, P., and Rasmann, S. (2016). The Abundance, Diversity, and Metabolic Footprint of Soil Nematodes Is Highest in High Elevation Alpine Grasslands. Front. Ecol. Evol. 4, 84. Körner, C. (2003). Alpine Plant Life: Functional Plant Ecology of High Mountain Ecosystems; with 47 tables (Springer Science & Business Media). Körner, C. (2007). The use of ‘altitude’ in ecological research. Trends in Ecology & Evolution 22, 569–574. Kuhl, C., Tautenhahn, R., Böttcher, C., Larson, T.R., and Neumann, S. (2012). CAMERA: An Integrated Strategy for Compound Spectra Extraction and Annotation of Liquid Chromatography/Mass Spectrometry Data Sets. Anal. Chem. 84, 283–289. Leibold MA, Mikkelson GM (2002) Coherence, species turnover, and boundary clumping: elements of meta-community structure. Oikos 97, 237–250. Lambers, H., and Poorter, H. (1992). Inherent Variation in Growth Rate Between Higher Plants: A Search for Physiological Causes and Ecological Consequences. In Advances in Ecological Research, M. Begon, and A.H. Fitter, eds. (Academic Press), pp. 187–261. Lemoine, N.P., Drews, W.A., Burkepile, D.E., and Parker, J.D. (2013). Increased temperature alters feeding behavior of a generalist herbivore. Oikos 122, 1669–1678. Lemoine, N.P., Burkepile, D.E., and Parker, J.D. (2014). Variable effects of temperature on insect herbivory. PeerJ 2, e376. Lewthwaite, J.M.M., Angert, A.L., Kembel, S.W., Goring, S.J., Davies, T.J., Mooers, A.Ø., Sperling, F. a. H., Vamosi, S.M., Vamosi, J.C., and Kerr, J.T. Canadian butterfly climate debt is significant and correlated with range size. Ecography 41, 1–11. Loranger, J., Meyer, S.T., Shipley, B., Kattge, J., Loranger, H., Roscher, C., and Weisser, W.W. (2012). Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology 93, 2674–2682. Malcolm, J.R., Markham, A., Neilson, R.P., and Garaci, M. (2002). Estimated migration rates under scenarios of global climate change. Journal of Biogeography 29, 835–849.

- 257 -

Maze, D.M. (2009). Effect of Terrestrial Mollusc Herbivory on Holocarpha macradenia (Asteraceae) Seedlings in California Coastal Prairie under Different Clipping Regimes. Madroño 56, 1–7. Mena JL, Vázquez-Domínguez E (2005) Species turnover on elevational gradients in small rodents. Glob Ecol Biogeogr 14, 539–547. Mithöfer, A., and Boland, W. (2012). Plant defense against herbivores: chemical aspects. Annual Review of Plant Biology 63, 431–450. Moreira, X., Petry, W.K., Mooney, K.A., Rasmann, S., and Abdala-Roberts, L. (2018). Elevational gradients in plant defences and insect herbivory: recent advances in the field and prospects for future research. Ecography 00, 1–12. Niu, S., and Wan, S. (2008). Warming changes plant competitive hierarchy in a temperate steppe in northern China. J Plant Ecol 1, 103–110. O’Connor, M.I. (2009). Warming strengthens an herbivore–plant interaction. Ecology 90, 388–398. Olff, H., and Ritchie, M.E. (1998). Effects of herbivores on grassland plant diversity. Trends in Ecology & Evolution 13, 261–265. Panetta, A.M., Stanton, M.L., and Harte, J. (2018). Climate warming drives local extinction: Evidence from observation and experimentation. Science Advances 4, eaaq1819. Pateman, R.M., Hill, J.K., Roy, D.B., Fox, R., and Thomas, C.D. (2012). Temperature-Dependent Alterations in Host Use Drive Rapid Range Expansion in a Butterfly. Science 336, 1028– 1030. Pearse, I.S., and Altermatt, F. (2013). Predicting novel trophic interactions in a non-native world. Ecology Letters 16, 1088–1094. Pelini, S.L., Keppel, J.A., Kelley, A.E., and Hellmann, J.J. (2010). Adaptation to host plants may prevent rapid insect responses to climate change. Global Change Biology 16, 2923–2929. Pellissier, L., Fournier, B., Guisan, A., and Vittoz, P. (2010). Plant traits co-vary with altitude in grasslands and forests in the European Alps. Plant Ecol 211, 351–365. Pellissier, L., Fiedler, K., Ndribe, C., Dubuis, A., Pradervand, J.-N., Guisan, A., and Rasmann, S. (2012). Shifts in species richness, herbivore specialization, and plant resistance along elevation gradients. Ecology and Evolution 2, 1818–1825. Pellissier, L., Rohr, R.P., Ndiribe, C., Pradervand, J.-N., Salamin, N., Guisan, A., and Wisz, M. (2013). Combining food web and species distribution models for improved community projections. Ecology and Evolution 3, 4572–4583. Pellissier, L., Niculita‐Hirzel, H., Dubuis, A., Pagni, M., Guex, N., Ndiribe, C., Salamin, N., Xenarios, I., Goudet, J., Sanders, I.R., et al. (2014). Soil fungal communities of grasslands are environmentally structured at a regional scale in the Alps. Molecular Ecology 23, 4274–4290.

- 258 -

Pellissier, L., Moreira, X., Danner, H., Serrano, M., Salamin, N., Dam, N.M. van, and Rasmann, S. (2016). The simultaneous inducibility of phytochemicals related to plant direct and indirect defences against herbivores is stronger at low elevation. Journal of Ecology 104, 1116–1125. Pérez‐Harguindeguy, N., Díaz, S., Vendramini, F., Cornelissen, J.H.C., Gurvich, D.E., and Cabido, M. (2003). Leaf traits and herbivore selection in the field and in cafeteria experiments. Austral Ecology 28, 642–650. Pitelka, L.F., and Plant Migration Workshop Group (1997). Plant migration and climate change. American Scientist; Research Triangle Park 85, 464–473. Rasmann, S., and Agrawal, A.A. (2011). Evolution of Specialization: A Phylogenetic Study of Host Range in the Red Milkweed Beetle (Tetraopes tetraophthalmus). The American Naturalist 177, 728–737. Rasmann, S., Pellissier, L., Defossez, E., Jactel, H., and Kunstler, G. (2014). Climate-driven change in plant–insect interactions along elevation gradients. Functional Ecology 28, 46–54. Rhoades, D.F. (1979). Evolution of plant chemical defence against herbivores. In Herbivores: Their Interaction with Secondary Plant Metabolites, (New York, NY, USA: Academic Press: London), pp. 3–54. Richards, L.A., Dyer, L.A., Forister, M.L., Smilanich, A.M., Dodson, C.D., Leonard, M.D., and Jeffrey, C.S. (2015). Phytochemical diversity drives plant–insect community diversity. PNAS 112, 10973–10978. Robinet, C., and Roques, A. (2010). Direct impacts of recent climate warming on insect populations. Integrative Zoology 5, 132–142. Rumpf, S.B., Hülber, K., Klonner, G., Moser, D., Schütz, M., Wessely, J., Willner, W., Zimmermann, N.E., and Dullinger, S. (2018). Range dynamics of mountain plants decrease with elevation. Proceedings of the National Academy of Sciences 115, 1848–1853. Salazar, D., Lokvam, J., Mesones, I., Vásquez, M.P., Ayarza, J.Z., De, P.V., and Fine, P.V.A. (2018). Origin and maintenance of chemical diversity in a species-rich tropical tree lineage. Nat Ecol Evol 2, 983–990. Sanders, N.J. (2002). Elevational gradients in ant species richness: area, geometry, and Rapoport’s rule. Ecography 25, 25–32. de Sassi, C., and Tylianakis, J.M. (2012). Climate Change Disproportionately Increases Herbivore over Plant or Parasitoid Biomass. PLOS ONE 7, e40557. Scheidel, U., and Bruelheide, H. (2001). Selective slug grazing on montane meadow plants. Journal of Ecology 87, 828–838. Schöps, K. (2002). Local and regional dynamics of a specialist herbivore: overexploitation of a patchily distributed host plant. Oecologia 132, 256–263.

- 259 -

Schweiger, O., Settele Josef, Kudrna Otakar, Klotz Stefan, and Kühn Ingolf (2008). Climate change can cause spatial mismatch of trophically interacting species. Ecology 89, 3472–3479. Stamp, N.E., and Yang, Y. (1996). Response of Insect Herbivores to Multiple Allelochemicals Under Different Thermal Regimes. Ecology 77, 1088–1102. Tibshirani, R. (1996). Regression Shrinkage and Selection via the Lasso. Journal of the Royal Statistical Society. Series B (Methodological) 58, 267–288. Urban, M.C., Tewksbury, J.J., and Sheldon, K.S. (2012). On a collision course: competition and dispersal differences create no-analogue communities and cause extinctions during climate change. Proceedings of the Royal Society of London B: Biological Sciences 279, 2072–2080. Van der Putten, W.H., Macel, M., and Visser, M.E. (2010). Predicting species distribution and abundance responses to climate change: why it is essential to include biotic interactions across trophic levels. Philosophical Transactions of the Royal Society B: Biological Sciences 365, 2025–2034. Veteli, T.O., Kuokkanen, K., Julkunen‐Tiitto, R., Roininen, H., and Tahvanainen, J. (2002). Effects of elevated CO2 and temperature on plant growth and herbivore defensive chemistry. Global Change Biology 8, 1240–1252. Vittoz, P., and Engler, R. (2007). Seed dispersal distances: a typology based on dispersal modes and plant traits. Bot. Helv. 117, 109–124.

- 260 -

ACKNOWLEDGEMENTS

First and foremost, I am very grateful to Prof. Dr. Loïc Pellissier who granted me complete freedom, provided me opportunities to work on various field of research, and was always available to help in solving problems and discuss. I was one of his first official PhD student when he started the research group in Fribourg (Switzerland). We have come a long way since then! Our common passion for fossil hunting and wine (from Vaud of course) lead up to fascinating field trips and discussions. What a crazy experience and good memories!

Next, I thank Prof. Dr. Sergio Rasmann, who has provided invaluable assistance and constructive criticisms on the various projects presented in this dissertation, technical support for the lab analyses in Neuchâtel and the field work during the first year of my PhD.

My work would not have been possible without the people who assisted me with field work, lab work, scientific discussions and data analysis: Camille Pitteloud, Camille Albouy, Oskar Hagen, Giulia Donati, Alan Kergunteuil, Aude Rogivue, Loïc Chalmandrier, Emilien Jolidon, Jeremy Marchon, Julia Bilat, Marco Walser, Eduardo Ottimophiore, Andrew Brown, Hélène Blauenstein, Amandine Pillonel, Andreas Zurlinden, Sabine Brodbeck, Christian Rösti, Oliver Kindler, Roland Reist, Ankara del Carmen Marjatshang-Chen, Dr. Sébastien Ibanez, Dr. Gaëtan Glauser, Prof. Dr. Nicklaus Zimmerman, Dr. Pascal Vittoz and Prof. Dr. Antoine Guisan. A special thanks to all the Landscape Ecology group for this three years at ETH Zurich, you are amazing people. We had great moment in the field, during standard days and during the many aperos we shared!

A great thanks to Rodolphe Muller and his staff from the refuge Giacomini for providing assistance and shelter during the long four consecutive summer season I spent in the beautiful region of Anzeindaz (VD, Switzerland). I greatly appreciated our discussions and your hospitality.

A great thanks to the Commune of Bex for allowing the settlement of the experiments on their territory and the State of Vaud, especially Franco Ciardo, for providing the necessary permits.

Finally, an immensurable thanks to my family who supported me all those years and who helped me to set up my experiments, especially my father Eric and my mother Margrit. Also thanks to Jean Tschopp who draw my cartoon (p. 9). Thanks to my “Didou” for the beautiful and unforgettable moments, you give me the strength to go further.

- 261 -

- 262 -

CURRICULUM VITAE

PATRICE DESCOMBES 29.05.1987, Switzerland [email protected]

EDUCATION June 2018 PhD in landscape ecology WSL Birmensdorf & ETHZ Zürich, Switzerland (2016-2018) & University of Fribourg, Switzerland (2014-2015). Supervisors: prof. Dr. Loïc Pellissier, prof. Dr. Sergio Rasmann. June 2013 Master of Sciences in Behaviour, Evolution and Conservation University of Lausanne, Switzerland. Thesis: Monitoring and high resolution species distribution modelling of invasive plant species. Supervisors: prof. Dr. Antoine Guisan, Dr. Blaise Petitpierre. June 2011 Bachelor of Sciences in Biology University of Lausanne, Switzerland.

REASEARCH EXPERIENCE At present PhD candidate at WSL and ETH, Switzerland. 2012 - 2013 MSc candidate at University of Lausanne, Switzerland. Aug – Sep 2011 Cantonal museum of Zoology, Lausanne, Switzerland.

PROFESSIONNAL EXPERIENCE Since 2013 Info Flora, Berne (BE), Switzerland. Transmission of more than 5’000 observations to the Info Flora database (> 800 species). Participation to the update of the Red List of the Swiss vascular plants. R programming for extracting spatial information on endemic plant species. Since 2010 Cercle vaudois de botanique, Lausanne (VD), Switzerland. Active participation in a floristic citizen science project in the canton of Vaud, whose goal is to update the distribution of 2’000 species and to publish an Atlas (Atlas de la Flore vaudoise; see http://www.atlasflorevd.ch). Production of distribution maps using R. Apr – Sep 2013 Biodiversity and landscape division (civilian service), St-Sulpice (VD), Switzerland. I worked on several projects of the canton of Vaud where I used my skills in data management, GIS mapping (ArcGIS) and field inventories.

- 263 -

June 2012 Ecology office, A. Maibach sàrl, Oron-la-ville (VD), Switzerland. I mapped the natural environments within an agro-ecological network and assessed the biological quality of objects according to OQE/ ÖQV criteria. Mar - July 2012 City of Lausanne, Municipality of Lausanne (VD), Switzerland. I performed field inventories of wild bees visiting insect hotels in an urban area. July - Aug 2010 Ecology office, BEB Raymond Delarze, Aigle (VD), Switzerland. Internship in applied ecology. I principally performed data management and field inventories.

TEACHING EXPERIENCE 2014 – 2018 Supervision of master and bachelor students: Emilien Jolidon (University of Neuchâtel), Jeremy Marchon, Daniel Slodowicz, Anne Häberle, Simona Colosio, Eduardo Ottimofiore (University of Fribourg). 2015 - 2016 Assistant in R spatial modelling, ETHZ, Switzerland. November 2015 Trainer for the Mendeley citation tool, University of Fribourg, Switzerland. Feb – June 2015 Assistant for ArcGIS practical’s and R, University of Fribourg, Switzerland. Feb – June 2012 Assistant for Botany 1st year students, University of Lausanne, Switzerland.

ADDITIONAL FORMATION September 2017 Introduction to the Syrphidae family and their use as bioindicators in conservation planning (Info Fauna). July 2012 Internship of alpine botany with J.-P. Theurillat (University of Geneva). May 2012 Certificate of knowledge of the Swiss amphibians (Karch). June 2011 Certificate of knowledge in field botany level 1, obtained with honors (SBS).

SKILLS R programming, ArcGIS, statistical analyses in R, species distribution models, SPLIT model (spatial diversification model of lineages through time), botany, field inventories, field experiments, traits measurements.

PUBLICATIONS In Prep Descombes, P., Pitteloud, C., Glauser G., Jolidon, E., Kergunteuil, A., Rasman, S., Pellissier, L. (2018) Trophic conservatism predicts alpine plants responses to herbivore ecosystem incursion Descombes, P., Kergunteuil, A., Glauser G., Rasman, S., Pellissier, L. (2018) Alpine plant palatability associated to physical and chemical traits in situ and under a warming treatment Häberle, A., Descombes, P., Pellissier, P. (2018) Triangular relationship between NDVI, species richness and fertilization intensity in permanent grasslands.

- 264 -

2018 Slodowicz, D., Descombes, P., Kikodze, D., Broenimann, O., Müller-Schärer, H. (2018) Areas of high conservation value at risk by plant invaders in Georgia under climate change. Ecology and Evolution. (Co-first authorship) Pellissier, L.; Descombes, P.; Hagen , O.; Chalmandrier, L.; Glauser, G.; Kergunteuil, A.; Defossez, E.; Rasmann, S. (2018) Growth-competition-herbivore resistance trade- offs and the responses of alpine plant communities to climate change. Functional ecology. Descombes, P., Gaboriau, T., Albouy, C., Heine, C., Leprieur, F., & Pellissier, L. (2018). Linking species diversification to palaeo‐environmental changes: A process‐ based modelling approach. Global Ecology and Biogeography, 27, 233-244. 2017 Descombes, P., Vittoz, P., Guisan, A., & Pellissier, L. (2017). Uneven rate of plant turnover along elevation in grasslands. Alpine Botany, 127, 53-63. Descombes, P., Marchon, J., Pradervand, J. N., Bilat, J., Guisan, A., Rasmann, S., & Pellissier, L. (2017). Community‐level plant palatability increases with elevation as insect herbivore abundance declines. Journal of Ecology, 105, 142-151. Descombes, P., Leprieur, F., Albouy, C., Heine, C., & Pellissier, L. (2017). Spatial imprints of plate tectonics on extant richness of terrestrial vertebrates. Journal of Biogeography, 44, 1185-1197. Ottimofiore, E., Albouy, C., Leprieur, F., Descombes, P., Kulbicki, M., Mouillot, D, … & Pellissier, L. (2017). Responses of coral reef fishes to past climate changes are related to life‐history traits. Ecology and Evolution, 7, 1996-2005. 2016 Descombes, P., Petitpierre, B., Morard, E., Berthoud, M., Guisan, A., & Vittoz, P. (2016). Monitoring and distribution modelling of invasive species along riverine habitats at very high resolution. Biological Invasions, 18, 3665-3679. Leprieur, F., Descombes, P., Gaboriau, T., Cowman, P. F., Parravicini, V., Kulbicki, M., ... Pellissier, L. (2016). Plate tectonics drive tropical reef biodiversity dynamics. Nature communications, 7. Pellissier, L., Eidesen, P. B., Ehrich, D., Descombes, P., Schönswetter, P., Tribsch, A., ... & Normand, S. (2016). Past climate‐driven range shifts and population genetic diversity in arctic plants. Journal of Biogeography, 43, 461-470. 2015 Descombes, P., Wisz, M. S., Leprieur, F., Parravicini, V., Heine, C., Olsen, S. M., & Pellissier, L. (2015). Forecasted coral reef decline in marine biodiversity hotspots under climate change. Global change biology, 21, 2479-2487. Descombes, P., Pradervand, J. N., Golay, J., Guisan, A., & Pellissier, L. (2015). Simulated shifts in trophic niche breadth modulate range loss of alpine butterflies under climate change. Ecography, 39, 796–804. Leprieur, F., Colosio, S., Descombes, P., Parravicini, V., Kulbicki, M., Cowman, P. F & Pellissier, L. (2015). Historical and contemporary determinants of global phylogenetic structure in tropical reef fish faunas. Ecography, 39, 825–835. Non ISI Descombes, P. (Mai 2017). Recherche scientifique à Anzeindaz. Le Patrouilleur alpin. Bulletin officiel de l’UPA10 - Union des Patrouilleurs Alpins, pages 10-12.

- 265 -

CONFERENCES Apr 2018 Descombes, P., Pitteloud C., Glauser G., Jolidon E., Kergunteuil A., Rasmann S., Pellissier L. Diet conservatism during insect’s herbivore range shifts predicts plant compositional changes under climate change. Macro2018. WSL Birmensdorf, Switzerland. Poster Feb 2018 Descombes, P., Jolidon, E., Rasmann, S., Pellissier, L. Plant responses to herbivore range shifts. Biology18. Neuchâtel, Switzerland. Poster Mar 2017 Descombes, P. Influence des orthoptères sur les communautés de plantes alpines. Conférences mensuelles du Cercle Vaudois de Botanique, Lausanne, Switzerland. Talk, invited speaker (1h) Sept 2015 Descombes P., Pellissier, L. Understanding plant community response to increased herbivore pressure under climate change. ITES Research Day. Zürich, Switzerland. Talk & Poster Feb 2015 Descombes, P., Rasmann, S., Pellissier, L. Understanding plant community response to increased herbivore pressure under climate change. Biology15. Dübendorf, Switzerland. Poster

LANGUAGES French Mothertongue (writing, speaking) English fluent (writing, speaking) German fluent (writing, speaking) Swiss-German Second mothertongue, good knowledge (speaking)

SOCIAL ACTIVITIES Sport Walking, Gymnastic Travelling National Parks: New-Zeeland, Europe, US Photography Flora & Fauna. Other Mushroom sampling, fossil hunting, astronomy

- 266 -

- 267 -