Pierre-Marc Brousseau

PIERRE-MARC BROUSSEAU

IMPACT DE LA DENSITÉ DE CERFS DE VIRGINIE SUR LES COMMUNAUTÉS D'INSECTES DE L'ÎLE D'ANTICOSTI

Mémoire présenté à la Faculté des études supérieures de l’Université Laval dans le cadre du programme de maîtrise en biologie pour l’obtention du grade de maître ès sciences (M. Sc.)

DÉPARTEMENT DE BIOLOGIE FACULTÉ DES SCIENCES ET GÉNIE UNIVERSITÉ LAVAL QUÉBEC

2011

Résumé

Les surabondances de cerfs peuvent nuire à la régénération forestière et modifier les communautés végétales et ainsi avoir un impact sur plusieurs groupes d'arthropodes. Dans cette étude, nous avons utilisé un dispositif répliqué avec trois densités contrôlées de cerfs de Virginie et une densité non contrôlée élevée sur l'île d'Anticosti. Nous y avons évalué l'impact des densités de cerfs sur les communautés de quatre groupes d'insectes représentant un gradient d'association avec les plantes, ainsi que sur les communautés d'arthropodes herbivores, pollinisateurs et prédateurs associées à trois espèces de plantes dont l'abondance varient avec la densité de cerfs. Les résultats montrent que les groupes d'arthropodes les plus directement associés aux plantes sont les plus affectés par le cerf. De plus, l'impact est plus fort si la plante à laquelle ils sont étroitement associés diminue en abondance avec la densité de cerfs. Les insectes ont également démontré une forte capacité de résilience. ii

Abstract

Deer overabundances can be detrimental to forest regeneration and can modify vegetal communities and consequently, have an indirect impact on many arthropod groups. In this study, we used a replicated exclosure system with three controlled white-tailed deer densities and an uncontrolled high deer density on Anticosti Island. The impact of deer density on the communities of four insect groups following a gradient of association with plants was studied alongside with the communities of herbivorous arthropods, pollinators and plant dwelling predators associated with three plant species whose abundance varies with deer density. The results revealed that arthropod groups more closely associated to plants were more strongly affected by deer density. Furthermore, the impact was stronger if the abundance of the plant to which they are closely associated decreased with deer density. Overall, insects appeared to be highly resilient to deer overabundance.

Avant-Propos

Ce mémoire comprend deux articles en anglais qui seront soumis pour publication dans une revue avec révision par les pairs. La récolte des donnés, l'identification des insectes, l'analyse des résultats et la rédaction de l'ensemble des textes du mémoire ont été faits par le candidat. Le directeur, Conrad Cloutier, le co-directeur Christian Hébert et le co-auteur Steeve Côté ont contribué aux manuscrits en corrigeant, révisant et en faisant des suggestions pour améliorer les textes. Cette recherche a pu être réalisée grâce à la participation conjointe de la Chaire de recherche industrielle CRSNG-Produits forestiers Anticosti de l'Université Laval et de Ressources Naturelles Canada.

Je voudrais en premier lieu remercier mon directeur et mon co-directeur de m'avoir permis d'entreprendre cette recherche et pour leur constant support et commentaires pour améliorer le travail effectué. Ils ont également su améliorer mon sens critique et mon esprit scientifique par leurs nombreux commentaires, souvent frustrants en premier, mais toujours pertinents et constructifs. J'apprécie également le fait d'avoir bénéficié d'un cadre de recherche favorable au développement personnel et un lieu de travail motivant dans le laboratoire de Christian Hébert au Centre de Foresterie des Laurentides.

Je remercie également les étudiants du laboratoire ÉcoDif avec qui il était toujours possible de parler de science ou d'autres sujets favorisant la décontraction aux moments opportuns: Ermias Azeria, Sébastien Bélanger, Richard Berthiaume, Jonathan Boucher, Éric Domaine, Jean-Philippe Légaré et Olivier Norvez. Je tiens également à remercier mes parents pour leur soutient tout au long de mes études.

J'aimerais également exprimer ma gratitude à l'équipe de la Chaire de recherche d'Anticosti, particulièrement Steeve Côté et Jean-Pierre Tremblay qui ont apporté leurs commentaires utiles à différentes étapes de la réalisation du projet, ainsi que Sonia de Bellefeuille et Caroline Hins qui ont géré le gros de la logistique de terrain.

Je souhaite également souligner la contribution de Yves Dubuc dans la préparation des travaux de terrain et son expertise au laboratoire et de George Pelletier pour son aide dans l'identification des insectes, tous deux de Ressources Naturelles Canada. Je remercie iv

également les étudiants qui m'ont aidé sur le terrain ou dans le laboratoire: Jannick Gingras, Nicolas Giasson et Yan Paiement. Finalement, je remercie Gaétan Daigle du département de mathématiques et de statistiques de l'Université Laval et Ermias Azeria pour leurs conseils pour les analyses statistiques.

Table des matières

Résumé ...... i Abstract ...... ii Avant-Propos ...... iii Table des matières ...... v Liste des tableaux ...... vii Liste des figures ...... ix Liste des annexes ...... x

INTRODUCTION ...... 1 Les surpopulations de cervidés ...... 1 L'impact des populations surabondantes sur les écosystèmes ...... 2 Effets sur les invertébrés ...... 3 L'île d'Anticosti ...... 6 Objectif de l'étude ...... 7

SHORT-TERM EFFECTS OF REDUCING WHITE-TAILED DEER DENSITY ON INSECT COMMUNITIES IN A STRONGLY DISTURBED BOREAL FOREST ECOSYSTEM ...... 9 Résumé ...... 9 Introduction ...... 9 Materials and methods ...... 12 Study area ...... 12 Experimental design ...... 12 Vegetation data ...... 13 Insect sampling ...... 14 Statistical analyses ...... 15 Results ...... 17 Diversity and abundance ...... 17 Community assemblages ...... 18 Discussion ...... 20 Acknowledgments ...... 23 References Cited ...... 24

INSECT-PLANT RELATIONSHIPS AT REDUCED DEER DENSITY IN AN OVERBROWSED BOREAL FOREST ECOSYSTEM ...... 34 Résumé ...... 34 Introduction ...... 34 Materials and methods ...... 37 Study site ...... 37 Experimental design ...... 37 Pollinators ...... 38 Herbivores and predators ...... 39 Statistical analyses ...... 40 Results ...... 41 vi

Pollinators ...... 41 Herbivores and predators ...... 42 Discussion ...... 44 Acknowledgments ...... 49 References Cited ...... 49

CONCLUSION ...... 62 Extinction et colonisation ...... 62 Subsistance des espèces ...... 64 Conservation ...... 66

Bibliographie ...... 67

Liste des tableaux

Table 1. Vegetation explanatory variables selected using the two-step forward selection procedure and used in RDA analyses on insect communities of Anticosti Island (Québec, Canada), with their mean % cover and range for harvested and forested areas. Plant species unselected, but present in the original data pool are presented in the lower part of the table...... 28 Table 2. Proportion of variance explained by the block effect based on mixed ANOVA of type 3 on the abundance of the four major insect taxa in (H)arvested and (F)orested areas (df=2/6) of the white-tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Significant variances are bolded...... 29 Table 3. Variation in the abundance and species richness of Apoidea, Syrphidae, Carabidae and macro Lepidoptera in three blocks (random factor), four experimental white tailed deer densities and two vegetation cover areas (harvested or forested) for Carabidae and macro Lepidoptera on Anticosti Island (Québec, Canada) (mixed ANOVA)...... 30 Table 4. Total species richness (S) and total abundance of Apoidea, Carabidae, macro Lepidoptera and Syrphidae along an experimental deer density gradient (0, 7.5 and 15 deer/km2 and 'U'ncontrolled) in harvested and forested areas on Anticosti Island (Québec, Canada), grouped based on relative abundance in uncontrolled deer density sites. Within each group and for each area, different letters indicate significant differences between deer densities based on mixed ANOVAs at α=0.05...... 31 Table 5. Arthropod groups (*: pollinators; +: "herbivores and predators") studied vs. plant species (Canada thistle, dwarf cornel, fireweed, and raspberry) in the different experimental units (0, 7.5, 15 deer/km2 and 'U'ncontrolled density) and blocks (A, B, C) of a deer-controlled browsing experiment on Anticosti Island (Québec, Canada)...... 52 Table 6. Mixed ANOVA on abundance of the three dominant pollinators of the flowers of three plant species on Anticosti Island (Québec, Canada) as affected by deer density (four levels, fixed effect) and blocks (three blocks, random effect)...... 53 Table 7. Proportion of variance explained by the block effect based on type 3 mixed ANOVAs on the abundance of dominant pollinators, grouped by family, on flowers of Canada thistle, dwarf cornel or fireweed in a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Significant block effect in bold...... 54 Table 8. Mixed ANOVA on abundance of the four dominant herbivorous and predator arthropods found on three plant species on Anticosti Island (Québec, Canada), as affected by deer density (four levels, fixed effect), sampling period (n=2) and blocks (three blocks, random effect)...... 55 Table 9. Proportion of variance explained by the block effect based on type 3 mixed ANOVAs on the abundance of dominant herbivores and predators, grouped by family, on Canada thistle, fireweed or raspberry in a white tailed deer controlled viii browsing experiment on Anticosti Island (Québec, Canada). Significant block effect in bold...... 56

Liste des figures

Figure 1: Distance triplots (scaling 1) of redundancy analyses (RDA) on A) the Apoidea species data (number of species (S)=35) and B) the Syrphidae (S=109) at all experimental sites (4 deer densities x 3 blocks) of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by a cross symbol (+) and a number for common species (see Annexe A). Arrows represent percent cover of each plant species retained as explanatory variables with the two-steps forward selection...... 32 Figure 2: Distance triplots (scaling 1) of redundancy analyses (RDA) on the Macro Lepidoptera species data in A) harvested area (number of species S=56) and B) forested area (S=90) at all experimental sites (4 deer densities x 3 blocks) of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by a cross symbol (+) and a number for common species (see Annexe A). Arrows represent percent cover of each plant species retained as explanatory variables with the two-steps forward selection...... 33 Figure 3: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance of pollinators associated with A) Canada thistle and B) dwarf cornel in twelve experimental units (except that Canada thistle was not sampled in density 0 deer/km2 of block A) of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (See annexe B and C)...... 57 Figure 4: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance of pollinators associated with fireweed in eight experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe D)...... 58 Figure 5: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with Canada thistle in twelve experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe B)...... 59 Figure 6: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with fireweed in eight experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe D)...... 60 Figure 7: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with raspberry in six experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe E)...... 61

Liste des annexes

Annexe A. Liste des espèces échantillonnés à l'aide de piège Luminoc® (Carabidae et macro Lepidoptera) et Malaise (Apoidea et Syrphidae) et leur abondance cumulée dans un dispositif à broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) et répété dans trois localités...... 71 Annexe B. Liste des arthropodes (N>1) échantillonnés sur les plants ou les fleurs de chardon (Cisium arvense) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) répété dans trois localités...... 79 Annexe C. Liste des insectes pollinisateurs (N>1) échantillonnés sur les fleurs de cornouiller du Canada (Cornus canadensis) avec leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) répété dans trois localités...... 84 Annexe D. Liste des arthropodes (N>1) échantillonnés sur les plants ou les fleurs d'épilobe (Epilobium angustifolium) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant trois densités de cerfs (0, 7.5, et 15 cerfs/km2) répété dans trois localités...... 86 Annexe E. Liste des arthropodes (N>1) échantillonnés sur les plants de framboisier (Rubus idaeus) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant trois densités de cerfs (0, 7.5, et 15 cerfs/km2) répété dans trois localités...... 90 1

Introduction

Dans plusieurs écosystèmes de l'hémisphère nord, les cervidés ont un grand impact sur l'environnement et les communautés qui les entourent (Danell et al. 2006, Waller et Alverson 1997). En Amérique du Nord, l'augmentation de leurs populations et de leur aire de répartition au cours du dernier siècle a entraîné des changements écologiques importants dans différents écosystèmes suite à leur broutement excessif sur plusieurs plantes (McShea et al. 1997). Plusieurs études récentes ont d'ailleurs montré que les hautes densités de cervidés pouvaient nuire à la régénération forestière (Tremblay et al. 2007) et modifier les communautés de plantes (Rooney et Waller 2003), d'oiseaux (McShea et Rappole 2000), de petits mammifères (McShea 2000) et d'invertébrés (Stewart 2001).

Les surpopulations de cervidés Selon Caughley (1981), une population animale atteint le niveau de surpopulation lorsqu'elle : 1) menace les populations humaines ou ses ressources, 2) cause une diminution de la densité d'un autre animal apprécié, 3) atteint une densité trop élevée pour son bien- être, 4) ou cause des perturbations dans l'écosystème où elle vit. Par contre, d'un point de vue strictement écologique, seule la quatrième définition correspond à de la surabondance selon Caughley. Les circonstances caractérisant les augmentations de population et pouvant mener à une irruption, i.e. une augmentation de population qui mène à la surpopulation, sont diverses et peuvent provenir d'une ou plusieurs causes agissant simultanément: l'augmentation de la capacité de support du milieu (K), la génération de nouveaux habitats par cause anthropique ou naturelle, et une diminution de la prédation ou de la chasse sont des facteurs souvent évoqués (Côté et al. 2004, McCullough 1997). L'isolement de certaines populations dans des secteurs précédemment exempts de cerfs, soit suite au franchissement d'une frontière naturelle ou suite à l'introduction par l'homme peut également entraîner des surpopulations (McCullough 1997).

Traditionnellement, on considère qu'une population connaissant une irruption voit sa densité augmenter jusqu'à un pic d'abondance pour ensuite subir un déclin rapide et se 2 stabiliser ultérieurement un peu en deçà de la capacité de support (K) du milieu (Caughley 1970); ce dernier aspect n'a par contre pas été démontré. Les populations de caribous (Rangifer tarandus L.) introduites sur les îles Saint-Paul (Scheffer 1951) et Saint-Mathieu (Klein 1968) en Alaska représentent des cas d'irruption suivi d'un déclin rapide. Dans le premier cas, 25 caribous ont été introduits en 1911 et le troupeau a atteint son pic de population en 1938 à 2046 individus. La population a ensuite chuté drastiquement à seulement 8 individus en 1950. Scheffer (1951) considère qu'à son pic d'abondance, la population de caribous était au moins trois fois plus élevée que la capacité de support du milieu, entraînant une réduction considérable de la quantité de lichens, nourriture hivernale de base du caribou, ce qui a causé sa chute. La situation sur l'île Saint-Mathieu est similaire, avec une diminution beaucoup plus rapide; le troupeau est passé de 6000 individus à 42 entre 1963 et 1966. Les causes de l'écroulement de la population seraient un hiver particulièrement rigoureux pendant lequel les individus affaiblis par la diminution de la quantité de nourriture disponible n'ont pu survivre (Klein 1968). L'évolution des populations suite à leur déclin n'est pas claire dû à un manque de suivi à long terme (>50 ans). À très long terme (~100 ans), il semble y avoir une oscillation stabilisatrice menant à l'équilibre de la densité des populations (Forsyth et Caley 2006). D'autres modèles prévoient plutôt une atteinte rapide de l'équilibre suite au déclin, ou un cycle de surabondance-déclin à amplitudes variables (McCullough 1997).

L'impact des populations surabondantes sur les écosystèmes L'influence des cervidés sur leur habitat s'explique principalement par leur broutement sélectif de certaines plantes (Augustine et McNaughton 1998) permettant à celles évitées d'augmenter fortement en densité relative (Horsley et al. 2003). La sélectivité peut également opérer à fine échelle en évitant une plante acceptable se trouvant dans le voisinage d'une plante favorisée (Bee et al. 2009). Lorsque les jeunes arbres sont fortement broutés, cela peut nuire à la régénération forestière (Watson 1983) ou modifier à long terme la succession forestière. Par exemple, les surabondances de cerfs de Virginie ont transformé des érablières (Acer saccharum Marshall) en prucheraies (Tsuga canadensis (L.) Carrière) au Michigan (Frelich et Lorimer 1985), et des forêts mixtes de feuillus en cerisaies (Prunus serotina Ehrh.) en Pennsylvanie (Horsley et al. 2003). Le broutement sélectif peut 3

également modifier les communautés de plantes de sous-bois en favorisant l'augmentation de l'abondance des fougères (Tilghman 1989) ou l'établissement d'une ou quelques espèces dominantes (Virtanen et al. 2002).

La nitrification du sol est également sujette à être modifiée par les cerfs. Dans les milieux riches en nutriments comme les prairies, l'impact peut être positif en favorisant la dégradation microbienne (Frank et al. 2000). Par contre, lorsque les plantes évitées par le cerf ont un taux peu élevé d'azote dans leur feuillage, comme l'épinette (Picea spp.), l'impact du surbroutement est négatif, nuisant ainsi au développement de certaines autres plantes (Pastor et al. 1993).

Ces modifications des communautés végétales entraînent des changements qui peuvent affecter les communautés animales de l'écosystème. La majorité des études sur l'impact des cervidés sur les oiseaux concluent que ce sont principalement ceux nichant dans les arbustes (ou la canopée intermédiaire) qui sont les plus affectés (deCalesta 1994, Martin et Daufresne 1996, Perrins et Overall 2001), soit ceux associés aux plantes ligneuses les plus accessibles aux cerfs. Également, Moser et Witmer (2000) ont observé une diminution de la richesse spécifique et de l'abondance de petits mammifères en présence de fortes densités de cerfs rouges (Cervus elaphe L.) et de vaches (Bos primigenius L.), expliquant la différence par une diminution de la quantité d'arbustes et de litière au sol.

Effets sur les invertébrés Les cervidés peuvent également modifier les communautés d'invertébrés indirectement en éliminant la ou les plantes hôtes des herbivores; en modifiant la structure végétale; ou en diminuant la biomasse et la qualité de certaines plantes (résumés dans Stewart 2001). Les invertébrés les plus étudiés dans le contexte des surabondances de cervidés peuvent être séparés en trois groupes principaux: les arthropodes épigés, les prédateurs phytophiles et les herbivores. Ces groupes représentent un gradient de relation écologique avec les plantes. Le premier groupe, qui est relativement indépendant des plantes, est généralement composé des Carabidae et/ou des araignées du sol 4

(principalement les Linyphiidae). Les principaux travaux sur les Carabidae en milieu forestier (Melis et al. 2006, Melis et al. 2007, Suominen et al. 2003) et les Linyphiidae (tous milieux confondus) (Dennis et al. 2001, Gibson et al. 1992, Takada et al. 2008) révèlent que ces groupes sont généralement plus abondants et diversifiés dans les environnements subissant une pression de broutement élevée. La diminution du taux d'humidité et l'augmentation de la luminosité causées par une simplification de la structure végétale seraient les principaux facteurs responsables de cette augmentation (Melis et al. 2007). Par contre, sur les îles Haida Gwaii en Colombie-Britannique, Allombert et al. (2005) n'ont pas observé de différence dans l'abondance et la richesse des Carabidae sur les différentes îles en fonction de l'année d'introduction du cerf. Une alternative aux hypothèses des auteurs serait qu'un historique de colonisation différent par les Carabidae sur les îles entraînerait une forte variation entre celles-ci.

Les prédateurs phytophiles, représentés par les araignées à toiles, dépendent de la structure végétale et sont généralement négativement affectés par le surbroutement (Baines et al. 1994, Miyashita et al. 2004, Suominen et al. 2008, Takada et al. 2008). Le broutement des cervidés diminue le nombre de points d'attache disponibles pour les toiles, diminuant ainsi l'espace disponible pour les araignées. La seule exception notoire est le cas des Theridiidae, une famille d'araignées qui tissent leur toile à la base des plantes, mais qui se nourrissent principalement d'arthropodes épigés. Ces araignées utilisent principalement de grosses branches non affectées par le broutement pour construire leur toile, contrairement aux autres araignées à toiles qui utilisent souvent plusieurs sous-étages de végétation, ce qui explique probablement la différence de réaction.

L'impact des cervidés et autres ongulés est généralement négatif sur les invertébrés herbivores. Différentes études ont démontré une diminution de l'abondance des gastéropodes (Suominen 1999), des lépidoptères (Baines et al. 1994, Kruess et Tscharntke 2002), des hétéroptères (Morris et Lakhani 1979) et des coléoptères, hyménoptères et diptères herbivores (Baines et al. 1994). Pour les gastéropodes, la diminution du taux d'humidité est le principal facteur responsable de la diminution de leur abondance 5

(Suominen 1999), alors que pour les autres groupes et particulièrement les spécialistes, c'est probablement la diminution de la biomasse de leurs plantes hôtes.

Quelques rares études se sont également penchées sur l'influence des cervidés sur les relations insectes-plantes. Vázquez et Simberloff (2004) ont étudié la pollinisation de plantes des Andes dans des milieux broutés par des cervidés et du bétail. Pour les quatre espèces de plantes répondant au broutement, aucune différence significative du nombre de visites de pollinisateurs n'a été observée entre les sites broutés et non broutés. Par contre, dans une étude précédente, les mêmes auteurs avaient trouvé des différences dans les communautés de pollinisateurs de sites broutés vs. non broutés (Vázquez et Simberloff 2003). Les principales différences s'expliquaient par un changement dans la fréquence de quelques relations dominantes; i.e. que les espèces pollinisatrices fortement associées à une plante lorsque celle-ci est abondante seraient moins associées à celle-ci lorsqu'elle devient rare. Den Herder et al. (2004) ont étudié l'impact du broutement du caribou sur le saule (Salix phylicifolia L.) et les insectes herbivores qui y sont associés. Les résultats montrent une diminution de 50% de la longueur des branches de saule dans les sites broutés et une diminution dans l'abondance des principaux herbivores: les chrysomèles du genre Gonioctena et les mouches-à-scie galligènes des genres Eupontania, Euura et Phyllocolpa. Finalement, dans une étude sur le broutement hivernal de l'orignal (Alces alces (L.)) sur le bouleau argenté (Betula pendula Roth), l'abondance des pucerons (Aphididae) était plus élevée sur les arbres broutés que sur les non broutés, alors qu'aucune différence significative n'a été observée pour les Curculionidae, lépidoptères et Eriophyidae (acariens) (Den Herder et al. 2009). Les résultats de ces deux études peuvent sembler contradictoires, mais ils démontrent que différents groupes d'insectes réagissent différemment aux stress subi par les plantes. Ainsi, les pucerons profitent d'une diminution de production des composés secondaires de défense de la plante en situation de stress, alors que les chrysomèles et les mouches-à-scie sont défavorisées par une diminution de la productivité de la plante causée par une diminution du taux de photosynthèse (Larsson 1989). 6

L'île d'Anticosti L'île d'Anticosti a une superficie de 7943 km2 et est située dans le Golfe du Saint- Laurent à ~35 km de la côte continentale nord et à ~72 km de la côte sud. Sa forêt fait partie de la zone boréale et est principalement composée de sapin baumier (Abies balsamea (L.) Mill.), et d'épinette blanche (Picea glauca (Moench) Voss) et noire (Picea mariana (Mill.) BSP). Le climat est de type maritime avec des étés frais et des hivers plutôt doux. Port-Menier, avec une population de 275 habitants, est le seul village de l'île. En 1896, 220 cerfs de Virginie ont été introduits sur l'île d'Anticosti dans le but d'en faire la chasse. Aucun prédateur naturel n'étant présent sur l'île, la population a rapidement augmenté pour atteindre aujourd'hui un total estimé de 125 000 individus sur l'ensemble du territoire; soit >20 cerfs/km2 (Potvin et Breton 2005).

Plusieurs espèces décidues telles que l'érable à épis (Acer spicatum Lam.), le noisetier à long bec (Corylus cornuta Marsh.) et le cornouiller stolonifère (Cornus stolonifera L.) ont pratiquement disparu de l'île. En hiver, le sapin représente 72% du régime alimentaire du cerf (Lefort 2002), nuisant ainsi à la régénération des sapinières. Les peuplements forestiers établis avant 1930 sont composés à 77-84% de sapin baumier, alors que les peuplements plus récents sont composés à 92-99% d'épinette blanche (Potvin et al. 2003). De plus, la proportion des peuplements dominés par le sapin est passée de 40% à 20% en 100 ans (Potvin et al. 2003). D'autres espèces de plantes autrefois abondantes sont également devenues rares, telles que l'épilobe (Epilobium angustifolium L.), le framboisier (Rubus idaeus L.) et la clintonie boréale (Clintonia borealis (Aiton) Raf.) (Potvin et al. 2003) au profit de plantes telle que le chardon (Cirsium arvense (L.) Scop.) (Viera 2004). Finalement, les sapinières matures de l'île d'Anticosti diffèrent de celles des îles de Mingan, un archipel sans cerf au nord d'Anticosti dont le climat et l'écologie sont semblables, par une faible représentation des semis de sapin (10-30 cm de hauteur) et l'absence d'arbustes (Viera 2004).

Après 8,5 ans d’exclusion du broutement du cerf à l’aide d’exclos dans des milieux récemment coupés, aucune différence n’a été observée dans la richesse spécifique des plantes comparativement aux endroits non protégés (Casabon et Pothier 2008). Par contre, 7 sept espèces de plantes (dont le bouleau blanc (Betula papyrifera Marsh.), l'épilobe et le sapin baumier) étaient significativement plus abondantes dans les exclos et deux l'étaient dont les zones broutées à l'extérieur des exclos (le chardon et les violettes (Viola sp.)). Finalement, en étudiant le retour des plantes à différentes densités de cerfs (0, 7,5, 15 cerfs/km2 et densité non contrôlée) dans des sapinières matures et des milieux de coupe, Tremblay et al. (2006) ont trouvé une augmentation exponentielle de la productivité de plusieurs espèces de plantes en fonction de la réduction de la densité de cerf à partir de 15 cerfs/km2.

Objectif de l'étude L'objectif général de l'étude est de déterminer l'impact de la réduction des densités de cerfs sur l'île d'Anticosti sur les communautés d'insectes en comparant trois densités de cerfs (0, 7,5 et 15 cerfs/km2) contrôlées à l'aide d'exclos avec des sites à densité non contrôlée de cerfs (>20 cerfs/km2). Ce projet entre dans le contexte d'une étude plus vaste visant à caractériser les patrons de régénération de la végétation et l'impact sur les animaux (oiseaux, petits mammifères et insectes) suite à la réduction des densités de cerfs. Un des objectifs clés est de déterminer la capacité de restauration des écosystèmes de l'île lorsque la densité de cerfs est réduite. Dans un premier temps, l'impact est étudié de façon générale sur les communautés de quatre groupes d'insectes appartenant à des guildes écologiques différentes: Carabidae (prédateurs épigés), Apoidea (pollinisateurs nicheurs), Syrphidae (pollinisateurs non-nicheurs dont les larves sont prédatrices ou saprophages) et Lepidoptera (herbivores). Dans cette section, l'hypothèse principale est que les modifications observées dans les communautés végétales suite à la réduction des densités de cerfs entraîneront des modifications dans les communautés d'insectes. Cette hypothèse permet de prédire que les taxons les plus fortement associés aux plantes (ex: phytophages) devraient réagir de façon plus prononcés que ceux qui n'y sont pas associés (ex: prédateur épigés).

En deuxième lieu, l'influence de la densité de cerfs sur les relations insectes-plantes est étudiée. Les pollinisateurs, les herbivores et les prédateurs phytophiles de quatre espèces de plantes ayant un patron de réaction différent à la réduction de la densité de cerfs sont étudiées: le chardon, le cornouiller du Canada (Cornus canadensis L.), l'épilobe et le 8 framboisier. Deux hypothèses (ce sont des prédictions) sont formulées pour ce chapitre. Premièrement, tout comme pour le premier chapitre, les insectes les plus spécifiques aux plantes devraient être les plus affectés par les diminutions de densité de cerfs. Deuxièmement, les arthropodes associés aux plantes affectées négativement par la diminution de la densité de cerf devraient être plus fortement affectés que ceux associés aux plantes absentes ou très rares à densité non contrôlée de cerfs. Celles-ci devraient principalement être utilisées par des espèces généralistes peu ou pas influencées par la densité de cerfs. 9

Short-term effects of reducing white-tailed deer density on insect communities in a strongly disturbed boreal forest ecosystem

Résumé Le broutement sélectif des cerfs peut nuire à la régénération forestière et modifier les communautés végétales. Plusieurs études ont démontré qu'il en résultait des impacts indirects sur plusieurs groupes d'arthropodes, mais ces résultats sont fragmentaires et limités à des taxons particuliers. Dans cette étude, nous avons étudié l'impact des densités de cerfs de Virginie sur les communautés de quatre groupes d'insectes suivant un gradient d'association avec les plantes sur l'île d'Anticosti. Un dispositif répliqué dans trois localités avec trois densités contrôlées de cerfs et une densité non contrôlée et élevée a été utilisé. Les résultats montrent que les taxons d'insectes les plus affectés par les densités de cerfs sont ceux dont les liens sont les plus étroits avec les plantes. L'abondance des espèces rares de Lépidoptères (herbivores) augmente et leurs communautés sont fortement affectées lorsque la densité de cerfs est diminuée, alors que les Carabidae (prédateurs épigés) ne sont pas affectés.

Introduction Since the beginning of the 20th century, the populations of cervids have been increasing in many parts of the world, mainly because of reductions in predators and hunting pressure, as well as increasing forage availability (reviewed in Côté et al. 2004). The overabundance of cervids has important ecological consequences as preferential and selective overbrowsing can be detrimental to forest regeneration (Watson 1983, Côté et al. 2004) and can modify plant communities (Rooney and Waller 2003, Côté et al. 2004). Also, the direct effect of high cervid density on vegetation is known to indirectly impact populations of invertebrates (Stewart 2001), birds (McShea and Rappole 2000), small

10 mammals (McShea 2000), and even large omnivores such as black bear (Ursus americanus Pallas) (Côté 2005).

Large herbivores can indirectly affect invertebrates by removing their host plants or by reducing plant biomass or quality, as well as by simplifying vegetation structure (reviewed in Stewart 2001). Studies involving different herbivorous insect taxa have generally shown that they were negatively affected by high ungulate density (Gibson et al. 1992b, Baines et al. 1994, Kruess and Tscharntke 2002, Martin et al. 2010), while studies on predators, which are usually less closely linked to plants, yielded variable responses. For instance, web-spiders, which highly depend on vegetation architecture (Baines et al. 1994, Miyashita et al. 2004), have been shown to be negatively affected by high ungulate density while small epigeal spiders such as Linyphiidae were favoured (Gibson et al. 1992a, Dennis et al. 2001, Takada et al. 2008). The Theridiidae, which construct webs at the base of plants, but mostly feed on ground-dwelling arthropods, seem not affected nor favoured by ungulate density (Takada et al. 2008). Most studies in forested areas have shown that Carabidae were more abundant at high grazing intensity (Suominen et al. 2003, Melis et al. 2006, Melis et al. 2007), but in at least one case, they were not affected at all (Allombert et al. 2005). However, in moorlands, a more open ecosystem, Gardner et al. (1997) reported that Carabidae were negatively affected by high cervid density. Nevertheless, Gardner et al. (1997) and Melis et al. (2007) reached the same conclusion for carabid assemblages: species associated with shady and moist habitats were more abundant at low grazing pressure, while the opposite was true for species associated with open habitats.

Some trends can be drawn from these studies based on the degree of association of organisms with plants. However, few studies have considered multiple guilds with different levels of relationship with plants in assessing the impact of high ungulate density on arthropods (but see Allombert et al. 2005). Furthermore, some groups such as pollinators which are less closely linked to plants than herbivores (flowers are available for a limited period during the life of pollinators while herbivores feed on plants during most of their life) have never been studied except in a context of observing their activity on flowers (Vázquez and Simberloff 2003, 2004).

An important difficulty arising when one wants to compare different studies is that ungulate densities are usually unknown. Therefore, because the comparison is often simply made between severely browsed and unbrowsed sites, moderate level disturbance has rarely been studied. The intermediate-disturbance hypothesis suggests that species diversity should be higher at intermediate level of disturbance (Connell 1978). Plants (Côté et al. 2004), birds (deCalesta 1994), and Carabidae and Curculionidae (Suominen et al. 2003) have been shown to be more diverse and/or abundant at intermediate deer density, which is compatible with this hypothesis. However, the results for Carabidae and Curculionidae were based on the abundance of lichen (Cladina spp.) carpets in dry forests of Lapland and are thus hardly generalizable to other systems where ungulate overbrowse saplings, shrubs and herbaceous plants.

Our goal was to evaluate the short-term effects on insect communities of reducing density of overabundant white-tailed deer (Odocoileus virginianus (Zimmermann)) in strongly disturbed balsam fir forests of Anticosti Island, Quebec, Canada in forested and harvested areas. We hypothesized that the short-term response of insects to reduced deer browsing pressure should follow their level of dependence on plants. Thus, we expected that insect responses should decrease from herbivores to pollinators and finally to epigeal predators. As it has been shown for vegetation (Tremblay et al. 2006), we hypothesized that insect responses should be stronger in harvested areas than in forested areas. From another viewpoint, dominant species of arthropods found in environments that are disturbed by overbrowsing are those that obviously have been the most favoured by conditions generated by high ungulate density. Thus, we may hypothesize that these dominant species should be the first to be negatively affected by reductions of deer density.

We predicted that 1) sensitivity to deer density should be stronger for herbivores and decrease along the guild gradient representing the degree of association with plants; 2) the species most rapidly affected should be those that dominate the disturbed ecosystem; 3) short-term responses should be stronger in harvested areas than in forested areas.

Materials and methods

Study area Study sites were located on Anticosti Island (7943 km2) in the Gulf of St-Lawrence (49º30'N 63º00'W), Quebec, Canada. The North coast of the island is ~35 km from the mainland compared to ~72 km for the South coast. White-tailed deer were introduced on Anticosti Island between 1896 and 1900. In absence of predators, the deer population increased rapidly and its density is now estimated at >20 deer/km2 across the island (Potvin and Breton 2005). The forest of Anticosti Island belongs to the boreal zone and is mostly composed of balsam fir (Abies balsamea (L.) Miller), white spruce (Picea glauca (Moench) Voss) and black spruce (Picea mariana (Miller) Britton, Sterns & Poggenburg). The dynamics of Anticosti forest ecosystems is strongly disturbed by deer overabundance. Old- growth forests originating prior to deer introduction are mostly dominated by balsam fir while recent forests are almost entirely dominated by white spruce because of browsing selectivity on balsam fir and deciduous trees (Potvin et al. 2003). The proportion of balsam fir dominated stands across the landscape has decreased from 40% to 20% over the last 100 years and many plant species once abundant are now rare, such as fireweed (Epilobium angustifolium L.), raspberry (Rubus idaeus L.) and yellow clintonia (Clintonia borealis (Aiton) Rafinesque-Schmaltz) (Potvin et al. 2003).

Experimental design A deer exclosure system where deer density was controlled was established in 2001 (Tremblay et al. 2006). The experiment formed a factorial design, where deer density was controlled at 0, 7.5 and 15 deer/km2 in fenced exclosures in three replicated blocks (A, B and C) located in different parts of the island. Each block also included an additional experimental unit with the local uncontrolled deer density (>20 deer/km2).

To control deer density in experimental units, all deer were removed from a 10 ha exclosure (0 deer/km2), whereas three deer (>11 months old) were stocked in exclosures of 40 ha (7.5 deer/km2) and 20 ha (15 deer/km2), for the other controlled densities. Deer used

13 in the controlled densities were captured in early spring, released within exclosures, and euthanized in late fall. Deer were equipped with VHF radio transmitters with mortality and activity sensors (Lotek Wireless, Newmarket, ON) to ascertain a constant deer density during the summer treatment period.

In each experimental unit (four densities x three blocks), ~70% of the forest was harvested just before the onset of the experiment in 2001, leaving ~30% of the mature balsam fir forest. The deer browsing treatment has been applied each year since 2002. In the residual forested areas, most of the vegetation biomass was represented by mature trees that were almost unaffected by the reduction of deer density, thus the main changes were on herbaceous plants density, such as dwarf cornel (Cornus canadensis L.) (Tremblay et al. 2006). However, in harvested areas, pioneering species such as fireweed and Rubus spp. rapidly established. Uncontrolled deer density in each block was estimated using line transect surveys of summer fecal pellet clusters using a distance sampling protocol (Buckland et al. 2001) and computed with DISTANCE 5.0 software (Thomas et al. 2010). For details about the estimation protocol, see Tremblay et al. (2006). Through years, uncontrolled deer densities were estimated at 26 deer/km2 for block B, and 57 deer/km2 for blocks A and C.

Vegetation data Vegetation was inventoried during summer 2007 in 20 plots of 10x10 m randomly distributed in both the harvested and forested areas of each experimental unit. The percentage cover of shrubs was visually evaluated by several observers using consensus calibration in each 10x10 m plot, while that of other vascular plant species was estimated in each of two randomly selected quadrats of 1x1 m inside the larger plot; for trees, only seedlings were considered. The technique was similar to that used by Tremblay et al. (2006). The mean percentage cover of each species across all quadrats of each experimental unit was used for the analyses.

Insect sampling Four taxonomic groups belonging to different ecological guilds along a gradient of association with plants were sampled. Nocturnal macro Lepidoptera (i.e. Bombycoidea, Drepanoidea, Geometroidea, Noctuoidea) representing mostly herbivorous insects as caterpillars; Apoidea (excluding former Sphecoidea) (Hymenoptera) representing the most important insect pollinator specialists, and are also nesting insects; Syrphidae (Diptera) were selected because they are both pollinators (polliniphage) when adults, while presenting diverse feeding behaviours at the larval stage such as predators, saprophages, herbivores or inquilines with ants, bees or wasps (Foote 1991). The absence of nesting behaviour also makes Syrphidae less affected by landscape structure (such as forest fragmentation) than Apoidea (Jauker et al. 2009). Finally, Carabidae were selected because they represent a broad diversity of mostly epigeal predators, often considered as useful bioindicators (Rainio and Niemelä 2003). These four taxa represent a gradient relative to their degree of association with plants: Carabidae (no direct relationship) → Syrphidae → Apoidea → macro Lepidoptera (close relationship with plants, and numerous cases of mono- and oligophagy).

Luminoc® traps (BIOCOM, Quebec City, QC.) (Jobin and Coulombe 1992) equipped with a blue light tube of 1.8 watt were used as pitfall traps to sample Carabidae (Hébert et al. 2000). In each experimental unit, two pitfall traps were installed in both harvested and forested areas. Traps were placed at least 100 m away from exclosure fences and whenever possible (i.e. when the forest patch was large enough), at least 50 m from the forest edge. Traps were operated during five 9-11 day periods between mid-June and mid- August 2007, using 40% ethyl alcohol as a preservative and for a total of 50 trapping days/trap in each experimental unit. At the end of each pitfall trapping period, the Luminoc® traps were raised and placed on a post at 3 m above the ground to sample macro Lepidoptera for 3-4 day periods. Vapona® strips were used to kill Lepidoptera in traps rather than ethyl alcohol. Traps at 3 m were operated when three consecutive non rainy days were forecasted by Environment Canada between 25 June and 19 August, for a total of 18 trapping days/trap in each experimental unit.

Flying adult Apoidea and Syrphidae were sampled using one Malaise trap (Gressitt and Gressitt 1962) in each experimental unit, installed in the harvested area, 50 m away from the forest edge and at least 100 m away from the exclosure fence. These traps were operated between 15 June and 19 August 2007, for a total of 64 trapping days/trap. A solution of 40% ethyl alcohol was used as preservative and samples were collected at intervals of ~14 days.

All specimens were identified at the species level whenever possible except for a few genera: i.e. Andrena (Apoidea: Andrenidae), Cheilosia and Microdon (Syrphidae), Eupithecia (Lepidoptera: Geometridae), and most females of Heringia, Platycheirus and Sphaerophoria (Syrphidae). Furthermore, specimens of the subgenus Lasioglossum (Dialictus) (Apoidea: Halictidae) were identified as morphospecies due to the lack of accurate identification keys and reference collection. Specimen identifications were cross- checked at the Canadian National Collection (CNC) of insects, arachnids and nematodes in Ottawa, Canada, and at the Insectarium René-Martineau (IRM) of the Canadian Forest Service in Québec, Canada, for families for which there was no expert at the CNC.

Statistical analyses For each major taxon (macro Lepidoptera, Apoidea, Syrphidae and Carabidae), we used mixed model ANOVAs (PROC mixed) (SAS 9.1, SAS Institute 2003) to examine variation in species richness (R) and the overall abundance of each major taxon in relation to deer density. Deer density was considered a fixed effect, and blocks a random effect. For macro Lepidoptera and Carabidae, a split-plot model with cover (forested or harvested) included as subplot was used. To meet the assumptions of ANOVA (i.e.: normality, homogeneity of variances), abundance and species richness data were square-root transformed. Furthermore, for the abundance of each major taxon, the relative contribution of blocks to the total variance among experimental units was calculated as follow:

2 2 2 σ B/(σ B+ σ ε)

2 2 where σ B is the estimated variance of the block effect and σ ε is the estimated variance of residuals. This is also a measure of the intraclass correlation related to blocks (Fleiss et al. 2003).

To test the hypothesis that dominant species should be more affected than other groups by decreases in deer density, we classified all species of each major taxon in four groups based on their ordered total abundance in uncontrolled deer densities. For Carabidae and macro Lepidoptera, this was done for both harvested and forested areas. A dominant species was defined as a species representing at least 25% of the total abundance of the taxon; common species were those that composed the first 75% of the total abundance, but were not dominant; uncommon species were present in the interval between 75-95%; and rare species were the remaining 5%. A mixed ANOVA was used to compare the abundance of each group between deer densities using the model previously described.

Variations in community composition were examined using redundancy analysis (RDA) based on Hellinger transformed species data (Legendre and Gallagher 2001). The RDAs were performed with the software R (function rda of the library vegan, version 1.15- 4) (Oksanen et al. 2009). They were conducted for each major taxon and also separately for harvested and forested areas for macro Lepidoptera and Carabidae. All RDAs included the twelve experimental units (four deer densities x three blocks), all insect species identified in the taxon and a number of explanatory variables selected using the two-steps forward selection procedure described by Blanchet et al. (2008), using the function forward.sel of library packfor in R (Dray et al. 2007), with α=0.1. Explanatory variables included deer density, % cover of each common plant species (i.e., representing ≥5% of total cover in at least one experimental unit) and the Shannon-Weiner index estimated for all vascular plants and for each of the five major plant groups (deciduous trees, coniferous trees, shrubs, herbaceous plants, and graminoids). Apoidea and Syrphidae were analysed using only vegetation data collected in harvested areas whereas data collected in forested or harvested areas were used for macro Lepidoptera and Carabidae in function of the area analysed. Shannon-Weiner indices for plants were calculated on the basis of the percentage ground cover of each species. Non-normal variables (based on the Shapiro test) were square-root or

17 fourth-root transformed; if it was not enough, they were removed from the analysis. Multicollinearity among explanation variables was assessed by calculating tolerance value (Quinn and Keough 2002); if tolerance was < 0.2 for one or more variables, the one with the lower score was removed from the pool, and the forward selection was run again until a result > 0.2 was obtained for all variables. The list of retained explanation variables for each RDA is available in Table 1.

RDA results were interpreted using distance triplots (scaling 1) where the eigenvectors are scaled to unit length (Legendre and Legendre 1998). In this scaling, projecting at right angle an experimental unit along an explanatory variable or a species response approximates the position of the unit along that variable (i.e. their degree of relation). In addition, the angle between an explanatory variable and a species represents their correlation. Finally, the distance between experimental units approximates their Euclidean distance in multidimensional space (Legendre and Legendre 1998).

Results

Diversity and abundance

A total of 1308 specimens of Apoidea representing 35 species belonging to five families (Andrenidae, Apidae, Colletidae, Halictidae and Megachilidae) were identified. For Syrphidae, 7481 specimens belonging to 109 species were caught while captures of macro Lepidoptera totalled 260 specimens in the harvested areas and 1343 in the forested areas. There were five macro Lepidoptera families (Arctiidae, Drepanidae, Geometridae, Lymantriidae and Noctuidae) and 105 species: 56 in the harvested areas and 90 in the forested areas. Finally, 1878 Carabidae belonging to 30 species were caught; 875 in the harvested areas and 1003 in the forested areas.

Although significant ANOVAs were obtained for most taxa, the effect of the block was high, explaining from 18.2% to 95.8% of the variance in the abundance of the four major insect taxa (Table 2). The block effect was not significant for Syrphidae while the

18 strongest effect was observed for Carabidae in forested areas. For both macro Lepidoptera and Carabidae, a significant block effect was found in both harvested and forested areas (Table 2).

Mixed ANOVAs did not reveal any significant effect of deer density on total abundance of any major taxa (Table 3). However, the effect of deer density was significant on the number of species of macro Lepidoptera (Table 3) for which, more species were captured at density 7.5 deer/km2 and 15 deer/km2 than in uncontrolled densities. The abundance of the dominant species of Syrphidae (Melanostoma mellinum (L.)), and Carabidae (Synuchus impunctatus (Say)), in both harvested and forested areas, decreased at reduced deer densities (Table 4). No dominant species was identified in forested areas for macro Lepidoptera, while the dominant species of harvested areas, the noctuid Leucania multilinea Walker, did not vary significantly with deer density as for the dominant species of Apoidea, Bombus borealis Kirby (Table 4). However, the dominant species of Apoidea declined in all reduced deer density sites by 25-79% while Bombus frigidus Smith became the dominant species in these sites. The abundance of rare species of macro Lepidoptera was significantly higher at reduced deer densities than in uncontrolled densities in both harvested and forested areas (Table 4).

Community assemblages Community analyses realised with RDAs showed markedly different responses of the four studied insect taxa. For Apoidea (Fig. 1A), experimental units of block A have negative values on axis 1 while those of blocks B and C are on the positive side, confirming the previously mentioned substantial block effect on insect abundance (Table 2). Furthermore, on the second axis, uncontrolled deer densities have positive values along with densities 0 and 15 deer/km2 of block B. Also, for reduced deer densities, the block effect was less strong than for uncontrolled densities. Balsam fir cover highly contributed to the formation of the second axis, and was thus mostly associated with reduced deer densities, while gold-thread (Coptis groenlandica (Oeder) Fernald) strongly contributed to the formation of the first axis, and was associated with blocks B and C.

Uncontrolled deer densities of the three blocks clearly shared a similar syrphid community, which was also shared with the 15 deer/km2 density of block B. The dominant species (M. mellinum) and one uncommon species (Platycheirus angustatus (Zetterstedt)) were strongly associated with this group of experimental units. Experimental units with reduced deer densities had different trajectories; those of block A were grouped together in the 2nd quadrant of the triplot and far from those of blocks B and C that were located in the 3rd and 4th quadrants, mostly along the 2nd axis. Contrary to Apoidea, the block effect appears to be greater for reduced densities than for uncontrolled densities, as shown by a greater separation of sites at reduced densities on the second axis. Gold-thread again opposed sites of blocks B and C to those of block A. Oak fern (Dryopteris disjuncta (Ledeb.) Morton)) was associated with high deer densities on the first axis.

For macro Lepidoptera, different RDAs were conducted for each area (harvested and forested). In harvested areas, the three uncontrolled deer density sites had negative values on axis 1 along with the 15 deer/km2 density of block B, while other experimental units had positive values (Fig. 2A). Balsam fir and dwarf red blackberry (Rubus pubescens Raf.) were associated with reduced deer densities on axis 1. Furthermore, the second axis mostly separated block A from blocks B and C. Like for Syrphidae, sites with reduced deer densities tended to differ more between themselves than uncontrolled densities. Raspberry highly contributed to separate block A from blocks B and C. In forested areas, community patterns were not as clear as in harvested areas (Fig. 2B). However, sites of block A, except that with 0 deer/km2, had negative values on the second axis along with uncontrolled density of block B, which was far from other densities of this block. Also, black spruce separated blocks B and C from block A.

Finally, for Carabidae, no significant RDA could be obtained either for the harvested or the forested areas.

Discussion As predicted, our results showed that the sensitivity of different insect taxa and feeding guilds to deer density decreased along a gradient representing their degree of association with plants. Epigeal Carabidae, who do not have any direct relation with plants, did not vary with deer density. For Syrphidae and Apoidea, we found a separation of the community between uncontrolled and reduced densities, but this separation was not as clear as for the Lepidoptera, which is the group most intimately linked to vegetation. Moreover, macro Lepidoptera was the only taxon with significantly higher abundance of rare species and higher number of species overall at reduced than at uncontrolled deer densities. In addition to the degree of association with plants, trophic level was also likely important. Indeed, predators may be slower to colonize new habitats than herbivorous species (Brown and Southwood 1983) as they depend on the recovery of herbivorous insects. Syrphidae and macro Lepidoptera also showed more differences in their communities between sites at reduced deer density than between sites at uncontrolled densities. This suggests that high deer density tends to homogenize the habitat, which results in decreased diversity of ecological niches available to insects, which are obviously closely linked with plant communities. It is also known that insect communities are composed of different species according to the plant composition (Schaffers et al. 2008), so that higher insect diversity can be expected in presence of higher diversity of ecological niches.

Because our study sites were located far from the shore and Anticosti Island itself is located at least 35 km away from the continent, we assume that most of the insects found in our study were present on the island before the 2001 forest harvesting. This implies, on one hand, that the most common species must be those that were adapted to conditions generated by deer overabundance, and were already abundant locally, and thus, were available to colonize the new habitats. On the other hand, rare species are more likely to be those inhabiting restricted areas on the island, such that their colonizing potential depends on the distance between the source populations and the new habitats (Littlewood et al. 2009). The abundance of the dominant species of Carabidae and Syrphidae significantly decreased at reduced deer densities. In Apoidea, the dominant species was nearly absent from blocks B and C and so, no significant effect of deer density was obtained. However, in

21 block A, its abundance decreased by at least 60% at two reduced densities, and by 25% in the third one. In macro Lepidoptera, no dominant species was identified in forested areas and no effect was observable on the dominant species in harvested areas. However, the abundance of rare species in this taxon was higher at reduced deer density sites. The results for all taxa except macro Lepidoptera are thus in agreement with the prediction that dominant species should be more rapidly affected by deer density reduction than other species. However, the response of rare species of macro Lepidoptera was faster than expected originally, which suggest a higher colonization capacity of this taxon.

For macro Lepidoptera, RDAs revealed that communities were more strongly structured in the harvested than in forested areas. This agrees with our hypothesis that changes are faster in an ecosystem fully exposed to sunlight, which allows rapid establishment and growth of colonizing plants. However, the abundance of rare species was higher at reduced densities than in uncontrolled densities in both harvested and forested areas. This suggests that moth communities can rapidly benefit from new vegetation growth in both open and forested areas. In temperate deciduous forests, it has been shown that rare Lepidoptera species were mostly associated with understory vegetation (Hirao et al. 2009) which could explain why only rare species seem to be affected in forested areas. Alternatively, it could also suggest that Lepidoptera in our relatively small residual forest patches benefit from the fast regeneration of the surrounding harvested areas.

The absence of response for Carabidae is intriguing because they generally show positive interactions with large herbivore density (Suominen et al. 2003, Melis et al. 2006, Melis et al. 2007). In our study, no effect was detected at the community level (RDA analysis). Similar results were also reported by Allombert et al. (2005) who worked on introduced black-tailed deer (Odocoileus columbianus (Richardson)) in the Haida Gwaii archipelago in British Colombia (Canada). Flying incapacity of many species of this family combined to random dispersion of flying species (Brouwers and Newton 2009) and landscape heterogeneity of our study sites (Halme and Niemelä 1993) can result in stochastic colonization. This is supported by the concentration of several species in only one or two experimental units, such as Calathus advena (LeConte) which was mostly

22 caught (66 out of 67 specimens) in the 7.5 deer/km2 of block A; Pterostichus melanarius (Illiger) with 57 out of 62 captures in the uncontrolled density of block A; Amara aulica (Panzer) with 26 out of 32 captures in the uncontrolled density of block A and density 7.5 deer/km2 of block B; and Pterostichus pensylvanicus LeConte with 11 out of 15 captures in density 0 deer/km2 of block C. Based on this, the absence of carabid response to deer browsing in Allombert et al. (2005) could alternatively be explained by stochastic colonisation of islands, and this could suggest that using insects with limited mobility could lead to misinterpretation when studying the impact of ecological changes on islands.

We hypothesized that insect diversity should follow the intermediate-disturbance hypothesis (Connell, 1978) and thus, be richer at mid deer densities (7.5 and 15 deer/km2). Our results are not conclusive on this aspect. For all taxa except Carabidae, the number of species was indeed higher at mid densities. However, this is significant only for macro Lepidoptera compared with uncontrolled density sites, but not with those of 0 deer/km2. Thus, this hypothesis is not confirmed on a short term in the system studied, but our results permit to hypothesize that it might be true in the long term, at least for insects closely associated with plants.

Our results support the suggestion of Hébert and Jobin (2001) that regeneration of balsam fir forests and of pioneering plant species such as raspberry are important in maintaining insect communities on Anticosti Island. These plants, once very abundant on the island are now endangered by deer overabundance and only a marked decrease of deer density could permit their restoration (Tremblay et al. 2006, 2007). Insect communities of the 15 deer/km2 of block B were different than those of other sites with the same deer densities and were more similar to uncontrolled densities for unknown reasons. A possible explanation is the presence of high level of regeneration of white spruce in this site, resulting from a hemlock looper (Lambdina fiscellaria (Guenée)) outbreak in early 1970's that may have reduced sunlight penetration and adversely affected herbaceous plant regeneration. Whatever, this indicates that deer reduction to 15 deer/km2 can be sufficient to restore insect diversity, but is unlikely to be efficient in all situations. Unfortunately, unlike plants, no historical data are available for insects of Anticosti island before deer

23 overbrowsing became noticeable in the mid 20th century (Potvin et al. 2003), and thus it is difficult to determine how reducing deer density can actually help to restore original insect natural communities.

Some very rare insect species of northeastern North-America have previously been found on Anticosti Island including Neospondylis upiformis (Mannerheim) (Coleoptera: Cerambycidae) (Hébert, unpublished data) and Papilio brevicauda Saunders (Lepidoptera: Papilionidae) (Handfield 1999), and others were found in this study: Harpalus megacephalus LeConte, Pipiza macrofemoralis Curran (first mention in the Province of Quebec), and Xylota flavitibia Bigot (first mention in eastern Canada). Considering that many insect taxa have yet to be investigated on this large island, additional rare species are expected. These are relevant arguments in favour of the long-term maintenance of the natural forests of Anticosti Island. In this context, it is important to continue studying the impact of deer on forest regeneration of Anticosti Island, but also elsewhere in North America and Europe where deer overabundance is also an issue (Côté et al. 2004). As shown here in relation to deer density, the insect fauna is directly and indirectly influenced by the surrounding vegetation, and limiting the impact of deer overabundance on plant communities is also important to maintain entomodiversity. Our results are consistent with the hypothesis that the impact of deer density on insect communities can be expected based on their degree of association with plants. Also, our results show that even in highly perturbed environment, restoration of plant communities can rapidly benefit insects and particularly herbivorous insects with high dispersal ability.

Acknowledgments Our research was financed by the Natural Sciences and Engineering Research Council of Canada (NSERC), Produits forestiers Anticosti Inc. (PFA), Natural Resources Canada (NRCan) and Université Laval. We would like to thank M. Poulin, S. Pellerin and M. Bachand for vegetation data; N. Giasson for help in field work, and Y. Paiement for help in laboratory work. We also thank Y. Dubuc for technical assistance and G. Pelletier for his taxonomic expertise, both of NRCan, and S. de Bellefeuille from the NSERC- Produits forestiers Anticosti Industrial Research Chair for logistical assistance. We are

24 grateful to Y. Bousquet from Eastern Cereal and Oilseed Research Center in Ottawa for confirming identification of Carabidae, and G. Daigle of the Département de Mathématiques et de Statistiques of Université Laval and E. Azeria from NRCan for statistical advises.

References Cited

Allombert, S., S. Stockton, and J.-L. Martin. 2005. A natural experiment on the impact of overabundant deer on forest invertebrates. Conservation Biology 19: 1917-1929. Baines, D., R. B. Sage, and M. M. Baines. 1994. The implications of red deer grazing to ground vegetation and invertebrate communities of Scottish native pinewoods. Journal of Applied Ecology 31: 776-783. Blanchet, F. G., P. Legendre, and D. Borcard. 2008. Forward selection of explanatory variables. Ecology 89: 2623-2632. Brouwers, N. C., and A. C. Newton. 2009. Movement rates of woodland invertebrates: a systematic review of empirical evidence. Insect Conservation and Diversity 2: 10- 22. Brown, V. K., and T. R. E. Southwood. 1983. Trophic diversity, niche breadth and generation times of exopterygote insects in a secondary succession. Oecologia 56: 220-225. Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas. 2001. Introduction to distance sampling: estimating abundance of biological populations. Oxford University Press, New York. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs - High diversity of trees and corals is maintained only in a nonequilibrium state. Science 199: 1302- 1310. Côté, S. D. 2005. Extirpation of a large black bear population by introduced white-tailed deer. Conservation Biology 19: 1668-1671. Côté, S. D., T. P. Rooney, J.-P. Tremblay, C. Dussault, and D. M. Waller. 2004. Ecological impacts of deer overabundance. Annual Review of Ecology, Evolution and Systematics 35: 113-147. deCalesta, D. S. 1994. Effect of white-tailed deer on songbirds within managed forests in Pennsylvania. Journal of Wildlife Management 58: 711-718. Dennis, P., M. R. Young, and C. Bentley. 2001. The effects of varied grazing management on epigeal spiders, harvestmen and pseudoscorpions of Nardus stricta grassland in upland Scotland. Agriculture, Ecosystems & Environment 86: 39-57. Dray, S., P. Legendre, and D. Borcard 2007. Package packfor: Forward selection with permutation computer program, version 0.0-7. By Dray, S., P. Legendre, and D. Borcard, Lyon. Fleiss, J. L., B. Levin, and M. C. Paik. 2003. Statistical methods for rates and proportions. John Wiley & Sons Inc., New York, NY. Foote, B. A. 1991. Syrphidae. In Immature insects. Edited by F. W. Stehr. Kendall/Hunt Publishing Company, Dubuque, IA, pp. 792-795.

Gardner, S. M., S. E. Hartley, A. Davies, and S. C. F. Palmer. 1997. Carabid communities on heather moorlands in northeast Scotland: The consequences of grazing pressure for community diversity. Biological Conservation 81: 275-286. Gibson, C. W. D., C. Hambler, and V. K. Brown. 1992a. Changes in spider (Araneae) assemblages in relation to succession and grazing management. Journal of Applied Ecology 29: 132-142. Gibson, C. W. D., V. K. Brown, L. Losito, and G. C. McGavin. 1992b. The response of invertebrate assemblies to grazing. Ecography 15: 166-176. Gressitt, J. L. and M. K. Gressitt 1962. An improved Malaise trap. Pacific Insects. 4: 87-90. Handfield, L. 1999. Le guide des papillons du Québec. Broquet, Ottawa, ON. Halme, E., and J. Niemelä. 1993. Carabid beetles in fragments of coniferous forest. Annales Zoologici Fennici 30: 17-30. Hébert, C., and L. Jobin. 2001. Impact du cerf de Virginie sur la biodiversité des forêts de l'île d'Anticosti: Les insectes comme indicateurs. Le Naturaliste Canadien 125: 96- 107. Hébert, C., L. J. Jobin, M. Fréchette, G. Pelletier, C. Coulombe, C. Germain, and M. Auger. 2000. An efficient pit-fall trap to study beetle diversity. Journal of Insect Conservation 4: 191-202. Hirao, T., M. Murakami and A. Kashizaki 2009. Importance of the understory stratum to entomofaunal diversity in a temperate deciduous forest. Ecological Research. 24: 263-272. Jauker, F., T. Diekötter, F. Schwarzbach, and V. Wolters. 2009. Pollinator dispersal in an agricultural matrix: opposing responses of wild bees and hoverflies to landscape structure and distance from main habitat. Landscape Ecology 24: 547-555. Jobin, L., and C. Coulombe. 1992. The Luminoc insect trap. Forestry Canada, Inf. Leaft. LFC-26, Québec Kruess, A., and T. Tscharntke. 2002. Grazing intensity and the diversity of grasshoppers, butterflies, and trap-nesting bees and wasps. Conservation Biology 16: 1570-1580. Legendre, P., and L. Legendre. 1998. Numerical Ecology. Elsevier, Amsterdam, NL. Legendre, P., and E. D. Gallagher. 2001. Ecologically meaningful transformations for ordination of species data. Oecologia 129: 271-280. Littlewood, N. A., R. J. Pakeman, and S. J. Woodin. 2009. Isolation of habitat patches limits colonisation by moorland Hemiptera. Journal of Insect Conservation 13: 29- 36. Martin, J.-L., S. A. Stockton, S. Allombert, and A. J. Gaston. 2010. Top-down and bottom- up consequence of unchecked ungulate browsing on plant and animal diversity in temperate forests: lesson from a deer introduction. Biological Invasions 12: 353- 371. McShea, W. J. 2000. The influence of acorn crops on annual variation in rodent and bird populations. Ecology 81: 228-238. McShea, W. J., and J. H. Rappole. 2000. Managing the abundance and diversity of breeding bird populations through manipulation of deer populations. Conservation Biology 14: 1161-1170. Melis, C., M. Sundby, R. Andersen, A. Moksnes, B. Pedersen, and E. Røskaft. 2007. The role of moose Alces alces L. in boreal forest - the effect on ground beetles (Coleoptera, Carabidae) abundance and diversity. Biodiversity and Conservation 16: 1321-1335.

Melis, C., A. Buset, P. A. Aarrestad, O. Hanssen, E. L. Meisingset, R. Andersen, A. Moksnes, and E. Røskaft. 2006. Impact of red deer Cervus elaphus grazing on bilberry Vaccinium myrtillus and composition of ground beetle (Coleoptera, Carabidae) assemblage. Biodiversity and Conservation 15: 2049-2059. Miyashita, T., M. Takada, and A. Shimazaki. 2004. Indirect effects of herbivory by deer reduce abundance and species richness of web spiders. Écoscience 11: 74-79. Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Steven, and H. Wagner 2009. The vegan library: Community ecology package computer program, version 1.15-4. By Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Steven, and H. Wagner, Oulu, FI. Potvin, F., P. Beaupré, and G. Laprise. 2003. The eradication of balsam fir stands by white- tailed deer on Anticosti Island, Québec: A 150-years process. Écoscience 10: 487- 495. Potvin, F., and L. Breton. 2005. Testing 2 aerial survey techniques on deer in fenced enclosures - visual double-counts and thermal infrared sensing. Wildlife Society Bulletin 33: 317-325. Quinn, G. P., and M. J. Keough. 2002. Experimental design and data analysis for biologists. Cambridge University Press, Cambridge, UK. Rainio, J., and J. Niemelä. 2003. Ground beetles (Coleoptera : Carabidae) as bioindicators. Biodiversity and Conservation 12: 487-506. Rooney, T. P., and D. M. Waller. 2003. Direct and indirect effects of white-tailed deer in forest ecosystems. Forest Ecology and Management 181: 165-176. Schaffers, A. P., I. P. Raemakers, V. K. Sýkora, and C. J. F. ter Braak. 2008. Arthropod assemblages are best predicted by plant species composition. Ecology 89: 782-794. Stewart, A. J. A. 2001. The impact of deer on lowland woodland invertebrates: a review of the evidence and priorities for future research. Forestry 74: 259-270. Suominen, O., J. Niemelä, P. Martikainen, P. Niemelä, and I. Kojola. 2003. Impact of reindeer grazing on ground-dwelling Carabidae and Curculionidae assemblages in Lapland. Ecography 26: 503-513. Takada, M., Y. G. Baba, Y. Yanagi, S. Terada, and T. Miyashita. 2008. Contrasting responses of web-building spiders to deer browsing among habitats and feeding guilds. Environmental Entomology 37: 938-946. Thomas, L., S. T. Buckland, E. A. Rexstad, J. L. Laake, S. Strindberg, S. L. Hedley, J. R. B. Bishop, T. A. Marques, and K. P. Burnham. 2010. Distance software: design and analysis of distance sampling surveys for estimating population size. Journal of Applied Ecology 47: 5-14. Tremblay, J.-P., J. Huot, and F. Potvin. 2006. Divergent nonlinear responses of the boreal forest field layer along an experimental gradient of deer densities. Oecologia 150: 78-88. Tremblay, J.-P., J. Huot, and F. Potvin. 2007. Density-related effects of deer browsing on the regeneration dynamics of boreal forests. Journal of Applied Ecology 44: 552- 562. Vázquez, D. P., and D. Simberloff. 2003. Changes in interaction biodiversity induced by an introduced ungulate. Ecology Letters 6: 1077-1083. Vázquez, D. P., and D. Simberloff. 2004. Indirect effects of an introduced ungulate on pollination and plant reproduction. Ecological Monographs 74: 281-308.

Watson, A. 1983. Eighteenth century deer numbers and pine regeneration near Braemar, Scotland. Biological Conservation 25: 289-305.

Plant species unselected, but present in the original data pool are presented in the lower part of the table.

Explanatory variables

Syrphidae Lepidoptera Macro (harvested) Lepidoptera Macro (forested) Carabidae (harvested) Carabidae (forested) (percentage cover) Harvested areas Forested areas Apoidea Mean ± sd Range Mean ± sd Range Abies balsamea (L.) Mill. 2.4 ± 0.5 <0.1 , 6.2 3.1 ± 0.5 0.2 , 5.4 X X Coptis groenlandica (Oeder) Fern. 3.7 ± 0.6 1.2 , 7.1 4.0 ± 0.4 1.5 , 5.7 X X X X Cornus canadensis L. 15.7 ± 1.8 4.4 , 28.0 14.5 ± 1.2 6.4 , 18.9 X Dryopteris disjuncta (Ledeb.) Morton 2.4 ± 0.7 0.2 , 8.7 1.9 ± 0.4 0.5 , 6.5 X Oxalis montana Raf. 2.5 ± 0.7 0 , 7.7 X X Picea glauca (Moench) Voss 17.5 ± 3.7 3.0 , 36.7 14.1 ± 2.4 0.8 , 30.6 X X X Picea mariana (Mill.) B.S.P. 2.0 ± 0.8 0 , 9.0 6.5 ± 1.9 0 , 22.7 X X Rubus idaeus L. 4.1 ± 1.7 0 , 18.1 X Rubus pubescens Raf. 8.3 ± 1.7 1.7 , 22.8 4.0 ± 0.6 0.1 , 7.0 X Shannon index of all vascular plants 2.7 ± 0.1 2.2 , 3.0 2.9 ± 0.1 2.4 , 3.3 X Betula papyrifera Marsh. 2.3 ± 0.6 <0.1 , 8.0 Cirsium arvense (L.) Scop. 2.8 ± 0.5 0.5 , 5.7 Dicentra canadensis (L.) Bernh. 1.8 ± 0.3 0.7 , 5.2 Epilobium angustifolium L. 7.4 ± 2.3 0 , 19.2 Fragaria virginiana Duchesne 2.2 ± 0.5 0.5 , 5.9 Gaultheria hispidula (L.) Muhl. 1.9 ± 0.6 0 , 6.2 Linnaea borealis L. 3.2 ± 0.7 0.1 , 6.8 Maianthemum canadense Desf. 4.5 ± 0.7 0.7 , 9.6 4.9 ± 0.3 3.4 , 6.5

Table 2. Proportion of variance explained by the block effect based on mixed ANOVA of type 3 on the abundance of the four major insect taxa in (H)arvested and (F)orested areas (df=2/6) of the white-tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Significant variances are bolded. Taxon % variance F P Apoidea 54.7 5.83 0.039 Carabidae (H) 63.2 7.86 0.021 Carabidae (F) 95.8 73.30 < 0.0001 macro Lepidoptera (H) 87.2 28.28 0.001 macro Lepidoptera (F) 57.3 6.36 0.033 Syrphidae 18.2 1.89 0.23

Table 3. Variation in the abundance and species richness of Apoidea, Syrphidae, Carabidae and macro Lepidoptera in three blocks (random factor), four experimental white tailed deer densities and two vegetation cover areas (harvested or forested) for Carabidae and macro Lepidoptera on Anticosti Island (Québec, Canada) (mixed ANOVA). Source of Apoidea Syrphidae variation df MS F P MS F P Abundance Density 3 77.12 1.35 0.34 96.47 2.40 0.17 Error 6 39.71 152.99 Richness Density 3 0.46 1.65 0.27 0.25 0.32 0.74 Error 6 0.28 0.78

Source of Carabidae Macro Lepidoptera variation df MS F P MS F P Abundance Density 3 6.99 2.24 0.1845 3.67 0.29 0.8296 Residual 6 3.13 1.24 Cover 1 2.17 0.85 0.3843 201.58 62.72 <0.0001 Density x Cover 3 2.16 0.85 0.5062 0.57 0.83 0.5136 Error 8 2.56 2.47 Richness Density 3 0.21 0.83 0.53 1.17 5.62 0.035 Residual 6 0.25 0.21 Cover 1 0.01 0.12 0.74 31.63 82.59 <0.0001 Density x Cover 3 0.02 0.27 0.84 0.25 0.67 0.6 Error 8 0.09 0.38

Table 4. Total species richness (S) and total abundance of Apoidea, Carabidae, macro Lepidoptera and Syrphidae along an experimental deer density gradient (0, 7.5 and 15 deer/km2 and 'U'ncontrolled) in harvested and forested areas on Anticosti Island (Québec, Canada), grouped based on relative abundance in uncontrolled deer density sites. Within each group and for each area, different letters indicate significant differences between deer densities based on mixed ANOVAs at α=0.05. Taxon Harvested Forested Group S 0 7.5 15 U S 0 7.5 15 U Apoideaa Dominant 1* 22 56 16 75 Common 3 204 272 142 87 Uncommon 8 65 88 56 56 Rare 23 38 76 43 12 Carabidae Dominant 1+ 69a 56a 67a 131b 1+ 87a 67a 72a 147b Common 2 26 41 27 142 1 60 94 45 52 Uncommon 3 35 88 56 65 3 67 80 60 55 Rare 17 10 27 9 26 17 19 78 5 15 macro Lepidoptera Dominant 1† 1 2 3 15 0 0 0 0 0 Common 4 12 5 5 15 12 114 133 136 147 Uncommon 0 0 0 0 0 17 77 104 73 42 Rare 51 43a 61a 53a 16b 61 85a 114a 85a 14b Syrphidaea Dominant 1‡ 186a 698ab 533ab 1252b Common 5 190 332 294 287 Uncommon 29 314 583 401 409 Rare 74 124 209 133 97 a This taxon was not sampled in forested areas * Bombus borealis Kirby + Synuchus impunctatus (Say) † Leucania multilinea Walker ‡ Melanostoma mellinum (L.)

Figure 2: Distance triplots (scaling 1) of redundancy analyses (RDA) on the Macro Lepidoptera species data in A) harvested area (number of species S=56) and B) forested area (S=90) at all experimental sites (4 deer densities x 3 blocks) of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by a cross symbol (+) and a number for common species (see Annexe A). Arrows represent percent cover of each plant species retained as explanatory variables with the two- steps forward selection. 34

Insect-plant relationships at reduced deer density in an overbrowsed boreal forest ecosystem

Résumé L'abondance relative de différentes espèces de plantes peut varier selon la densité de cervidés par l'action du broutement sélectif qui tend à diminuer l'abondance des plantes préférées tout en favorisant les plantes évitées. Ceci peut avoir des impacts sur les arthropodes qui sont associés à ces plantes, mais très peu d'études se sont penchées sur le sujet. Nous avons étudié les communautés d'arthropodes herbivores, pollinisateurs et prédateurs associées à quatre espèces de plantes (épilobe, chardon, cornouiller du Canada et framboisier) dont l'abondance variait différemment selon la densité de cerfs de Virginie sur l'île d'Anticosti dans un dispositif répliqué comprenant trois densités contrôlées de cerfs et une densité élevée non contrôlée. Les herbivores sont les plus affectés par la densité de cerfs et les communautés associées au chardon, une plante abondante à densité non contrôlée, mais dont l'abondance diminue avec la densité de cerfs, sont les plus modifiées.

Introduction Cervids overabundance can strongly modify plant communities (Rooney and Waller 2003). Selective browsing tends to reduce population size and density of favoured plant species (Augustine et al. 1998), thus leaving space and resources for unbrowsed or less exploited plant species (Horsley et al. 2003). High deer density can thus increase plant richness (Schütz et al. 2003), but if an unbrowsed plant benefits to the point of becoming dominant, the overall effect can also be negative (Horsley et al. 2003). These characteristics of plant communities can influence abundance and diversity of insects associated with them, and thus, deer can indirectly influence insect-plant relationships in different ways (Den Herder et al. 2004, Vázquez and Simberloff 2004).

Few studies have examined deer impact on insect-plant interactions. Vázquez and Simberloff (2003, 2004) did not find any effect of deer and cattle browsing on the number

35 of visits by pollinators to flowers of four plant species of which three were more abundant and one was less abundant in absence of browsing, although pollinator communities were affected. The main difference was a tendency for dominant relations to be less important when plant density was reduced. Browsing by reindeer (Rangifer tarandus (L.)) on tea- leaved willow (Salix phylicifolia L.) in Finland reduced the abundance of associated herbivorous leaf beetles (Chrysomelidae) and galling sawflies (Tenthredinidae) (Den Herder et al. 2004), whereas a similar study on winter browsing by moose on silver birch (Betula pendula Roth) revealed its positive impact on aphid abundance, but not on Curculionidae, Lepidoptera and Acari (Den Herder et al. 2009). These results are concordant with the observation that sap-feeding insects tend to react differently than folivorous and galling insects to plant stress (Larsson 1989).

Insect-plant relationships are diverse, including antagonism by herbivorous species, mutualism by pollinators, and more or less neutral commensalisms with predators that use plants as foraging habitat where they prey on either herbivores or pollinators (Gullan and Cranston 2005). Each of these groups could react differently to density and isolation of their host plant, as well as to plant diversity. The main factor influencing herbivorous insects seems to be plant diversity, insects showing consistently higher abundance and diversity in sites with higher plant diversity (Root 1973, Siemann 1998, Symstad et al. 2000, Borges and Brown 2001). Furthermore, positive relations were found between the abundance of a host plant and the abundance and diversity of its herbivorous insects. For example, stem borers of Canada thistle (Cirsium arvense (L.) Scopoli) were found to be more diverse and abundant in Lower Saxony, Germany when the plant was more abundant (Kruess 2003); the abundance of the thephritid fly Urophora cardui (L.) in Bavaria, Germany was also greater when it host plant C. arvense was more abundant (Eber and Brandl 2003). Similar relationships were found for pollinators.

The population size of sticky catchfly (Lychnis viscaria L.) and Wild mustard (Sinapis arvensis L.) has been shown respectively to have a positive impact on the abundance of Bumblebees (Bombus spp.) (Mustajarvi et al. 2001), and pollinators in general (Kunin 1997). Also, plant isolation (i.e., the distance between patches of the same

36 plant species) has been reported to have a negative impact on the abundance and diversity of Apoidea, whereas Syrphidae and other pollinators did not seem to be affected (Steffan- Dewenter and Tscharntke 1999). Finally, few studies addressed the impact of plant community structure on plant dwelling predators. However, it seems that web-spiders, foraging on plants are more abundant in complex plant communities (Borges and Brown 2001, Miyashita et al. 2004). Thus, the main factors affecting herbivorous arthropods and pollinators seem to be the population size and the isolation of their host plant, while predators are more influenced by the complexity of plant architecture. By inducing changes in the relative abundance of different plant species (Frelich and Lorimer 1985, Virtanen et al. 2002) and by simplifying plant architecture (Miyashita et al. 2004), deer overbrowsing can influence insect communities associated with them in different ways.

Here, we examine the impact of a reduction in white-tailed deer (Odocoileus virginianus (Zimmermann)) density, in a forest ecosystem strongly disturbed by deer overbrowsing, on the arthropod herbivores, pollinators and predators that are associated with three different plant species responding differently to deer browsing. Herbivorous and predatory arthropods were sampled on Canada thistle, fireweed (Epilobium angustifolium L.) and raspberry (Rubus idaeus L.), whereas pollinators were sampled on Canada thistle, dwarf cornel (Cornus canadensis L.) and fireweed. To assess the effects of a reduction in deer density on arthropods, replicated exclosures at three controlled deer densities (i.e.: 0, 7.5 and 15 deer/km2) were compared to nearby sites at uncontrolled deer density (>20 deer/km2) for arthropod abundance and community changes.

Our goal was to assess how reducing deer density may affect insect-plant relationships. More precisely, our study aimed to answer the following questions: 1) Do herbivores, pollinators and predators respond similarly to reduced deer density? 2) Do insect communities associated to plant species that are recovering versus those that are declining at reduced deer density react similarly to deer density reduction? First, we hypothesized that arthropod communities most directly associated with plants should be the most affected by changes in abundance and isolation of their host plant induced by reduction in deer density (chapter 1). Secondly, arthropod communities associated with

37 invasive plants at uncontrolled deer densities should be affected more rapidly than those associated with recovering plants.

Materials and methods

Study site The deer controlled-browsing experiment was carried out on Anticosti Island in the Gulf of St-Lawrence (49º30'N 63º00'W), Québec, Canada. The island of 7943 km2 is located ~35 km from the mainland to the North and ~72 km to the South. Anticosti Island is included in the boreal zone and the forest is mainly composed of balsam fir (Abies balsamea (L.) Miller), white spruce (Picea glauca (Moench) Voss), and black spruce (Picea mariana (Miller) Britton, Sterns & Poggenburg). In 1896-1897, about 220 white- tailed deer were introduced on the island. In absence of natural predators, the population increased rapidly and is now estimated at >20 deer/km2 (Potvin and Breton 2005). Deer overbrowsing resulted in many changes in the ecosystem dynamics of the island and most notably on forest regeneration (Tremblay et al. 2007). While old-growth forests are mostly dominated by balsam fir, young forests that regenerated after disturbance are generally dominated by white spruce because of deer overbrowsing on balsam fir and deciduous trees (Potvin et al. 2003). Changes in herbaceous plants and shrubs have also been observed: fireweed, raspberry and yellow clintonia (Clintonia borealis (Aiton) Rafinesque-Schmaltz) are now rare or have even disappeared locally because of deer overbrowsing, whereas invasive species such Canada thistle and Canada reed-grass (Calamagrostis canadensis (Michaux) P. de Beauvois) are now dominant (Potvin et al. 2003, Casabon and Pothier 2008).

Experimental design In 2001, enclosures were set up in three replicated blocks (A, B and C) located in different parts of the island and where deer density was controlled (Tremblay et al. 2006). Each block included three experimental units of controlled densities (0, 7.5 and 15 deer/km2) and an unfenced unit where density was uncontrolled (>20 deer/km2). In each

38 experimental unit (four densities x three blocks), ~70% of the forest was harvested, leaving ~30% of mature balsam fir forest. All deer were removed from a 10 ha area for establishing the 0 deer/km2 treatment, whereas three deer were stocked in a 40 ha and a 20 ha enclosure to establish the 7.5 and 15 deer/km2 respectively. Deer used in the experiment (mainly yearlings) were caught during spring on Anticosti Island, released in exclosures each year, and euthanized in late fall. To ascertain constant deer density within exclosures, a VHF radio transmitter equipped with a mortality and activity sensor was attached on each deer (Lotek Wireless, Newmarket, ON). Regular drives and track monitoring survey in winter were also done to prevent the presence of deer intruders in exclosures.

Uncontrolled deer density in each block was estimated with line transect surveys where summer fecal pellet groups were inventoried using distance sampling (Buckland et al. 2001) and computed with DISTANCE 5.0 software. For details, see Tremblay et al. (2006). Through years, uncontrolled densities averaged 26 deer/km2 in block B and 57 deer/km2 in blocks A and C.

Pollinators Insect pollinators foraging on three abundant plant species differently affected by deer density reductions were studied. Tremblay et al. (2006) found that dwarf cornel shows a linear increase in biomass with decreasing deer density and that fireweed, which is nearly absent at uncontrolled deer density, shows an exponential increase. By contrast, Canada thistle responds negatively to decreasing deer density. Observations were limited to experimental units where plants were sufficiently abundant, i.e. where several patches of ≥20 flowers were present (see Table 5). Observations were made during a 20 min period on the first encountered patch of at least 20 flowers. The patch was then flagged to avoid resampling. Data from 7-10 observation periods per experimental unit were obtained for Canada thistle between 8 and 22 August, for dwarf cornel between 25 June and 12 July and for fireweed between 27 July and 22 August 2008. All observations were conducted between 10:00 AM and 16:30 PM during non-rainy days with at most moderate wind and partly clouded sky. Wind strength class (null, weak or moderate), cloud cover (%) and time of observation were recorded. In each experimental unit, observations were done randomly

39 on different days and at different hours so as to minimize any possible bias. Observations were conducted in the harvested areas for fireweed and Canada thistle, and in the forested areas for dwarf cornel. All flying insects in contact with a reproductive part of a flower of the observed plant were caught with a sweep net and later identified to family level and to species level for the Apoidea, Vespidae, Syrphidae, Cerambycidae and the most abundant species of Muscidae.

Herbivores and predators Three plant species were also studied more closely to determine how changes in plant abundance induced by deer density reduction affected herbivorous and predator communities. They were fireweed and Canada thistle, for which the relation with deer density is monotonic as mentioned above, and raspberry, which has a quadratic relation to deer density, being more abundant at mid-densities (7.5 and 15 deer/km2) (Tremblay et al. 2006). However, raspberry could not be studied in block B where it was nearly absent. In each experimental unit, plants were randomly selected for sampling and individually bagged before uprooting them. Fifteen individuals of each species were collected along a transect in July and 15 more in August 2008 in each experimental unit. Upon entering an experimental unit, the first plant was collected ~100 m from the fence to avoid edge effects and subsequent ones were collected along the transect, leaving at least 20 steps between collections. Plants were frozen until they were processed in the lab, where all arthropods were collected and sorted. Leaves were carefully examined for leaf miners, stems were dissected for borers, and if present, galls were dissected to extract gallers and any associated parasitoids. For all plants, stem height, number of flowers or fruits, leaf biomass and number of stems (for raspberry) were recorded. Whenever possible, all arthropods were identified to the genus or species level. Some arthropods were identified to morphospecies when taxonomic identity could not be determined, and some taxa were recorded only at the family level (i.e. Hemiptera: Fulgoroidea, Hymenoptera: Parasitica, Diptera: Nematocera and immatures other than Syrphidae, and most Acari). Dwarf cornel was not sampled in this part of the study, because of the difficulty to standardize sampling on this very small plant. In addition, bad weather during the raspberry flowers bloom prevented us to make enough observations of pollinators foraging on this plant for statistical analysis.

Statistical analyses Community structures were examined using principal component analysis (PCA) using the software R and the function rda of the library vegan, v. 1.15-4 (Oksanen et al. 2009). A different PCA was performed on pollinators, herbivores and predators for each plant species. Species found in only one block were not included in PCA and species data were transformed into Hellinger distance, thus limiting the importance of zero abundance data (Legendre and Gallagher 2001). Only specimens identified at least to the genus level were included in PCAs. We also attempted RDA analyses using plant height, leaf biomass and number of flowers and/or fruits as explanatory variables for herbivores and predators, but none were significant, thus we do not report them.

For each plant species, mixed model ANOVAs (PROC mixed) were used to compare abundance of all species, families and orders of arthropods that were caught more than 30 times on a host-plant (SAS 9.1, SAS Institute 2003). For pollinators, the analysis was designed as a completely randomized block, with deer density as a fixed effect and block as a random effect; observation periods were considered as replicate samples. For herbivores and predators, given the two sampling periods, the analysis was designed as a split-plot with deer density and sampling period as fixed effects, and block as a random effect. When a plant species was absent in a particular density treatment of a block, it was considered as a missing value. To meet the assumptions of ANOVA (i.e.: normality, homogeneity of variance), the abundance of species were square-root transformed. To assess the influence of block effects on the groups studied here, we calculated their relative importance in the total variance of the abundance of the principal families and species caught on each plant species as follows:

2 2 2 σ B/( σ B + σ ε )

2 2 where σ B is the estimated variance of the block effect and σ ε is the estimated variance of residuals (Fleiss et al. 2003).

Results

Pollinators Overall, 1990 pollinators were captured during observations with Diptera representing 69.7%, 89.1% and 24.4% of the total catch, respectively, on Canada thistle, dwarf cornel, and fireweed, and Hymenoptera representing 29.5%, 7.1% and 75.2% on the same plants. Among Diptera, nearly 90% were Muscidae (66.7%) or Syrphidae (22.4%), while in Hymenoptera nearly 90% were Apidae (72.8%), Vespidae (11.2%) or Halictidae (5.8%). Overall, we determined 91 species of pollinators, the families with most species being the Syrphidae (47), Muscidae (15) and Apidae (7).

Pollinator communities associated with Canada thistle were strongly different between blocks along the first axis of the PCA, revealing strong local effects (Fig. 3A). The 0 deer/km2 density sites of blocks B and C were also separated from the other densities along the second axis (the 0 deer/km2 of block A was not sampled because we did not find enough patches of more than 20 flowers). In block A, 77% of all pollinators associated with Canada thistle were bumblebees, whereas they represented only 7% in blocks B and C. In these blocks, the muscid fly Thricops spiniger (Stein) was the dominant flower visiting species in all sites, representing 41-88% of total captures except at density 0 deer/km2 of block B where it represented 24% of total captures.

Communities of pollinators observed on dwarf cornel in uncontrolled deer density sites were clearly separated from reduced density sites (Fig. 3B). Sites from block A mostly stretched along the first axis whereas those from blocks B and C separated along the second axis. The muscid fly Phaonia serva (Meigen) was the dominant species in the uncontrolled density of blocks B and C representing respectively 48% and 44% of the total captures (including families not present in the PCA). In block A, no dominant species was identified in the uncontrolled density site. In reduced densities of blocks B and C, P. serva was often (4 sites out of 6) more abundant than in uncontrolled densities, but shared its dominance with the syrphid fly Meliscaeva cinctella (Zetterstedt). In block A, unlike the other blocks,

M. cinctella was nearly absent and P. serva was concentrated in the uncontrolled density site. Except for these two species, the pollinator communities of dwarf cornel were characterized by the presence of numerous species (67) from which 63 were represented by less than 10 specimens.

For pollinator communities associated with fireweed (which was nearly absent from uncontrolled densities), block A clearly separated from blocks B and C along the first axis (Fig. 4), similarly to Canada thistle. The 15 deer/km2 density of block C was widely separated from both the 0 and 7.5 deer/km2 densities of blocks B and C. Two species were apparently more abundant there than elsewhere; i.e. Megachile relative Cresson with 7 out of a total of 10 specimens and T. spiniger with 12 out of a total of 29 specimens caught on fireweed flowers. Also, 3 out of 4 species of Apidae and 4 out of 5 species of Muscidae were mostly associated with block A, whereas 4 out of 5 species of Syrphidae were more associated with blocks B and C than block A.

ANOVAs indicated no significant effect of deer density on the abundance of each pollinator species or family associated with any of the three plant species (Table 6). Contrasted effects were observed regarding the influence of the block effect depending on insect taxon and plant species. For Apidae, it was generally strong (over 80% of variance explained), except for Bombus frigidus Smith on Canada thistle (Table 7). However, this bumblebee strongly varied with blocks when associated with fireweed (88%). A contrast was also apparent for Muscidae, whose abundance varied weakly with blocks except when associated with Canada thistle (Table 7).

Herbivores and predators Overall, 11,486 arthropods were collected on the plants, from which 9197 were herbivores and 1370 were predators or parasitoids. The principal herbivorous families were the Aphididae (71%), Cecidomyiidae (11%) and Cercopidae (7%), while the principal parasitoids and predator families were the hymenopteran Platygastridae (24%), and spider families Araneidae (15%) and Clubionidae (11%).

Among the 3906 arthropods found on Canada thistle, 2477 were herbivores and 802 were predators or parasitoids with the most abundant families being the sap-feeding Aphididae (1586) and Cercopidae (502), parasitoids Platygastridae (297) and collembolan Entomobryidae (502). Herbivorous communities associated with Canada thistle were clearly separated with respect to deer density (Fig. 5A). Indeed, the three sites with 0 deer/km2 were isolated in their own quadrant. Also, the second axis separated the higher densities (uncontrolled and 15 deer/km2 density of block B) from reduced densities, except for the 7.5 deer/km2 density of block A. PCA also revealed that the most abundant mono- and oligophagous species (i.e. aphids Uroleucon cirsii (L.) and Capitophorus sp.1) had negative scores on the second axis, and thus were more associated with high deer density sites. Inversely, species associated with densities 0 deer/km2 were mainly polyphagous herbivores (cercopid Philaenus spumarius (L.), cicadellid Draeculacephala sp. and spider mites Tetranychidae sp.). For predators, most sites of block A (i.e. except the 0 deer/km2 site) were isolated from those of blocks B and C in their own quadrant (Fig. 5B). Blocks B and C, mostly separated along the second axis with respect to deer density, but it was not as clear as for herbivorous arthropods.

On fireweed, of the 5308 arthropods that were found, 4987 were herbivores and 186 were predators. The Aphididae represented 78% of total captures. The other most abundant families were Cecidomyiidae (364), Cercopidae (153) and leaf-mining Gelechiidae (98). Herbivorous arthropod communities associated with fireweed were mainly structured with respect to blocks (Fig. 6A). Again, sites of block A were isolated in their own quadrant. Sites at 0 deer/km2 of blocks B and C were relatively close to each other and distant from other sites of these blocks. For predators, the pattern was similar to that observed on Canada thistle (Fig. 6B). However, many species found on both plants showed different relative positions with respect to sites

A total of 2272 arthropods, 1733 herbivores and 382 predators and parasitoids, were found on raspberry. The most abundant families were the sap-feeding Aphididae (841), Cecidomyiidae (478), and Miridae (101). No clear community separation was observed for the herbivorous communities associated with raspberry (Fig. 7A). However, it should be

44 noted that sites with the same deer density appeared more similar at low deer density than at high deer density, thus 0 deer/km2 sites shared highly similar communities, while those at 15 deer/km2 were very different. Predators were strongly separated by blocks on the second axis (Fig. 7B). However, as for Canada thistle and fireweed, the 0 deer/km2 density of block A is remote from other sites of this block.

For all plant species, the percentage of variance explained by the first two axes of the PCAs was higher by at least 17% for herbivorous communities than for predators. Mixed ANOVAs on the abundance of herbivorous and predatory arthropods revealed a significant effect of deer density only on the collembolan Entomobrya sp. 1 on fireweed, its abundance decreasing with deer density (Table 8). The block effect was much weaker for herbivores and predators than for pollinators. For Canada thistle, the proportion of variance of the abundance of the principal families explained by the block effect was 10% on average for herbivores and predators, whereas it was 65% for pollinators (Tables 7 and 9). For fireweed, it was 25% for herbivores and predators, and 49% for pollinators (Tables 7 and 9).

Discussion

Our results show that the effect of deer density on plant communities indirectly influenced insect communities but differently on the basis of their functional relationship with plants and also with respect to the plant species itself. A significant, but variable block effect, representing among locality variation, was also observed. Thus, herbivorous communities associated with Canada thistle, a plant that clearly benefited from high deer densities and which was thus abundant and widely distributed across the island, only weakly varied with blocks and were more affected by deer density. On the opposite, communities associated with fireweed, a plant that was nearly absent at uncontrolled deer density, strongly varied with blocks. This is consistent with our prediction that insect communities associated with plants decreasing in density when deer density is reduced should be more affected. By contrast, communities associated with raspberry, which was rare and generally did not produce flowers at uncontrolled deer density sites (PMB personal

45 observation), were only slightly influenced by either blocks or deer densities. Also, the block effect was particularly strong for pollinators and predators, suggesting that local factors were more influent than vegetation changes due to deer overbrowsing in structuring these communities. The higher amount of variance represented by the first two axis of PCAs on herbivores is also consistent with the prediction that herbivorous insect communities are more strongly related to deer density (and locality) than predators, which are probably more strongly influenced by variables that were not included in our analysis.

These results also provide support for the following answers to our two original questions:

1) Arthropods that were ecologically more closely associated with plants were the most strongly influenced by changes in plant abundance resulting from reduced deer densities.

2) Herbivorous insects associated with a common plant were more influenced by changes in its abundance than those that were associated with a rare plant that suddenly became abundant following deer density reduction.

The fact that arthropods closely related to plants were the most strongly influenced by deer density is not surprising, as we previously found similar results while studying insect abundance using traps in the same system (chapter 1). Ecosystem changes resulting from deer browsing pressure on plants can modify the detectability of plants by insects associated with them, which may affect more strongly insects that are most closely associated with plants. Thus, we observed different responses for herbivores, pollinators and plant dwelling predators. The ANOVA on abundance data did not reveal any significant effect of deer density on the abundance of pollinators. However, in the sites where the flowers of dwarf cornel and Canada thistle were the least abundant (unpublished data), communities of pollinators associated to them were very different than those elsewhere. These results are in agreement with previous studies where no significant

46 difference was found in the number of pollinator visits on flowers in response to deer browsing pressure (Vázquez and Simberloff 2004). However, when comparing communities, Vázquez and Simberloff (2003) observed differences in pollinator communities between sites without and with browsing for plant species that were abundant in unbrowsed sites, but rare in browsed sites. They explained these changes by the increasing rarity of the plants affected by browsing. For spiders, the degree of relationship with plants can determine their response to deer browsing (Takada et al. 2008). However, colonization by arthropod predators is slower than for herbivorous insects (Brown and Southwood 1983), and web-spider communities may still be changing 15 years after forest harvesting (Buddle et al. 2000). Thus, the absence of response of predators as observed here could result from uncompleted colonization, because our experiment has only been running for seven years.

Herbivorous insect communities were more strongly influenced by deer densities than pollinators and predators, but they were also clearly influenced by the original density of their host plant at uncontrolled deer density (i.e. before reducing deer density). The influence of deer density reduction was particularly strong for herbivorous communities associated with Canada thistle. The community changes were principally explained by variation in abundance of the mono- and oligophagous species, which dominated at uncontrolled deer densities and thus, at higher Canada thistle density. Inversely, in absence of deer, Canada thistle density was lower and communities were dominated by polyphages. This is particularly evident for the monophagous thistle aphid U. cirsii whose average abundance decreased from 272 aphids per site at uncontrolled deer densities, to only 6 aphids at the 0 deer/km2 density sites. Such a marked decrease in abundance was already observed for the monophagous thephritid fly U. cardui in Bavaria, Germany with reduction of Canada thistle density (Eber and Brandl 2003). Thus, monophagous species seem particularly sensitive to density decrease of their host plant.

Contrary to our prediction, several oligophagous taxa such as the leaf miner Gelechiidae sp.2 and the stem borer Scythris sp. were present on fireweed in all experimental units. Even if this species of Scythris was not determined, only two species of

47 this genus are associated with Epilobium, both being specific to this genus (Landry 1991). Also, leaf miners such as Gelechiidae sp.2 are all considered at least oligophagous, and are generally monophagous (Hespenheide 1991). According to another study, the recovery of a Coleophoridae (Lepidoptera) stem borer occurred between 18 months and three years after the restoration of its host plant Sporadanthus ferrugineus de Lange when separated by approximately 800 m from the source population (Watts and Didham 2006). In our case, the distance from the source population is unknown, as fireweed is practically absent at uncontrolled deer densities across most of Anticosti Island. It is possible that the Lepidoptera came from the mainland, which would indicate greater colonization ability than expected. Alternatively, they could come from isolated fireweed populations still present on the island. However, it has been hypothesized that a minimal host plant biomass is necessary to locally sustain oligophagous species (Straw and Ludlow 1994). Our data suggest that this minimum host plant density (or minimum biomass) can be extremely low for the above fireweed herbivores of Anticosti Island.

Strong structure differences in pollinator communities on dwarf cornel separated uncontrolled from controlled deer densities (Fig. 3B). At uncontrolled deer densities, most flower patches of dwarf cornel were isolated and hardly accessible to deer. Furthermore, in forested areas (those where dwarf cornel was sampled) other flowers were even rarer (Tremblay et al. 2006). Thus, the isolation of dwarf cornel flower patches, and low flower diversity could explain the community difference observed. Indeed, other studies showed that floral diversity (Ghazoul 2006) and habitat isolation (Steffan-Dewenter and Tscharntke 1999) could be determinant for pollinators. It should be noted that B. frigidus was highly influenced by blocks on fireweed, but not on Canada thistle. When looking more closely at the raw data, we noticed that the number of caught B. frigidus was at least 50% higher on fireweed than on Canada thistle in each experimental unit where both plants were present. Thus, it seems that B. frigidus is more associated to fireweed, but that it can also use Canada thistle when the density of fireweed is too low. Another interesting observation concerning pollinators of Anticosti Island in this study is the paucity of bees in blocks B and C, with their apparent replacement by Muscidae, while bees were dominant in block A.

Thus, a decrease in competition from bee specialists seems to benefit pollinator flies, as was already observed in Arctic (Danks 1986) and Alpine regions (Arroyo et al. 1982).

Previous studies reported negative impacts of deer browsing on plant dwelling predators, particularly on web-spiders (Miyashita et al. 2004, Takada et al. 2008). In our study, even though web-spiders were the most common predators, no strong relations were found in their communities in relation to deer browsing. However, traditional sampling (i.e. net sweeping) was not used here as we focused sampling on plant species known to differently respond to deer browsing instead of all the local plant community. Vegetation architecture and complexity are often considered important predictors of web-spider abundance at the plant community scale (Borges and Brown 2001, Miyashita et al. 2004). Here, at the individual plant scale, plant architecture is not expected to vary that much and thus was probably not a local factor determining spider abundance and diversity.

Our results illustrate the complexity of insect-plant interactions at the ecosystemic level. By strongly modifying the environment and vegetation in particular, deer can influence the community of arthropods associated with a particular plant species. However, the processes involved are complex and remain very difficult to determine. Plant density and diversity, flowering and flower abundance, and patchiness all play roles in host plant selection by pollinators and herbivores, which may explain apparent contradictions in different studies trying to assess the importance of each component separately (Grez and González 1995, Kruess 2003). In our study phytophagous arthropods were affected mostly by deer browsing reduction when their abundant host plant became rare and isolated, such as observed for Canada thistle at 0 deer/km2. A similar observation occurred for pollinators associated to dwarf cornel at uncontrolled deer densities, however, in this case, we hypothesize that the plant community surrounding the visited flowers may be more important in structuring pollinator communities. Finally, plant-dwelling predators appeared mainly influenced by the surrounding plant community. Our results also revealed a fast recovery of Lepidoptera associated with population of regenerating plants that were previously almost extinct, suggesting high colonization ability for these insects.

Acknowledgments This research was financed by the Natural Sciences and Engineering Research Council of Canada (NSERC), Produits forestiers Anticosti Inc. (PFA), Natural Resources Canada (NRCan) and Université Laval. We would like to thank J. Gingras for help in the field and laboratory work and G. Pelletier from NRCan for his taxonomic expertise. We are also grateful to C. Hins from the NSERC-Produits forestiers Anticosti Industrial Research Chair for logistical assistance and G. Daigle of the mathematical and statistical department of Université Laval and E. Azeria from NRCan for statistical advice.

References Cited

Arroyo, M. T. K., R. Primack, and J. Armesto. 1982. Community studies in pollination ecology in the high temperate Andes of central Chile. I. Pollination mechanisms and altitudinal variation. American Journal of Botany 69: 82-97. Augustine, D. J., L. E. Frelich, and P. A. Jordan. 1998. Evidence for two alternate stable states in an ungulate grazing system. Ecological Applications 8: 1260-1269. Borges, P. A. V., and V. K. Brown. 2001. Phytophagous insects and web-building spiders in relation to pasture vegetation complexity. Ecography 24: 68-82. Brown, V. K., and T. R. E. Southwood. 1983. Trophic diversity, niche breadth and generation times of exopterygote insects in a secondary succession. Oecologia 56: 220-225. Buckland, S. T., D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers, and L. Thomas. 2001. Introduction to distance sampling: estimating abundance of biological populations. Oxford University Press, New York. Buddle, C. M., J. R. Spence, and D. W. Langor. 2000. Succession of boreal forest spider assemblages following wildfire and harvesting. Ecography 23: 424-436. Casabon, C., and D. Pothier. 2008. Impact of deer browsing on plant communities in cutover sites on Anticosti Island. Écoscience 15: 389-397. Danks, H. V. 1986. Insect plant interactions in Arctic regions. Revue d'Entomologie du Québec 31: 52-75. Den Herder, M., R. Virtanen, and H. Roininen. 2004. Effects of reindeer browsing on tundra willow and its associated insect herbivores. Journal of Applied Ecology 41: 870-879. Den Herder, M., R. Bergstrom, P. Niemela, K. Danell, and M. Lindgren. 2009. Effects of natural winter browsing and simulated summer browsing by moose on growth and shoot biomass of birch and its associated invertebrate fauna. Annales Zoologici Fennici 46: 63-74. Eber, S., and R. Brandl. 2003. Regional patch dynamics of Cirsium arvense and possible implications for plant-animal interactions. Journal of Vegetation Science 14: 259- 266.

Fleiss, J. L., B. Levin, and M. C. Paik. 2003. Statistical methods for rates and proportions. John Wiley & Sons Inc., New York, NY. Frelich, L. E. and C. G. Lorimer 1985. Current and predicted long-term effects of deer browsing in hemlock forests in Michigan, USA. Biological Conservation. 34: 99- 120. Ghazoul, J. 2006. Floral diversity and the facilitation of pollination. Journal of Ecology 94: 295-304. Grez, A. A., and R. H. González. 1995. Resource concentration hypothesis: Effect of host plant patch size on density of herbivorous insects. Oecologia 103: 471-474. Gullan, P. J. and P. S. Cranston 2005. The insects. An outline of entomology. Blackwell Publishing, Davis, CA. Hespenheide, H. A. 1991. Bionomics of leaf-mining insects. Annual Review of Entomology. 36: 535-560. Horsley, S. B., S. L. Stout, and D. S. DeCalesta. 2003. White-tailed deer impact on the vegetation dynamics of a northern hardwood forest. Ecological Applications 13: 98- 118. Kruess, A. 2003. Effects of landscape structure and habitat type on a plant-herbivore- parasitoid community. Ecography 26: 283-290. Kunin, W. E. 1997. Population size and density effects in pollination: Pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber. Journal of Ecology 85: 225-234. Landry, J.-F. 1991. Systematics of Nearctic Scythrididae (Lepidoptera: Gelechioidea): Phylogeny and classification of supraspecific taxa, with a review of described species. Memoirs of the Entomological Society of Canada, Ottawa, ON. Larsson, S. 1989. Stressful times for the plant stress: Insect performance hypothesis. Oikos. 56: 277-283. Legendre, P., and E. D. Gallagher. 2001. Ecologically meaningful transformations for ordination of species data. Oecologia 129: 271-280. Miyashita, T., M. Takada, and A. Shimazaki. 2004. Indirect effects of herbivory by deer reduce abundance and species richness of web spiders. Écoscience 11: 74-79. Mustajarvi, K., P. Siikamaki, S. Rytkonen, and A. Lammi. 2001. Consequences of plant population size and density for plant-pollinator interactions and plant performance. Journal of Ecology 89: 80-87. Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Steven, and H. Wagner 2009. The vegan library: Community ecology package computer program, version 1.15-4. By Oksanen, J., R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Steven, and H. Wagner, Oulu, FI. Potvin, F., and L. Breton. 2005. Testing 2 aerial survey techniques on deer in fenced enclosures - visual double-counts and thermal infrared sensing. Wildlife Society Bulletin 33: 317-325. Potvin, F., P. Beaupré, and G. Laprise. 2003. The eradication of balsam fir stands by white- tailed deer on Anticosti Island, Québec: A 150-years process. Écoscience 10: 487- 495. Rooney, T. P., and D. M. Waller. 2003. Direct and indirect effects of white-tailed deer in forest ecosystems. Forest Ecology and Management 181: 165-176. Root, R. B. 1973. Organization of a plant-arthropod association in simple and diverse habitats: Fauna of collards (Brassica oleracea). Ecological Monographs 43: 95-120.

Schütz, M., A. C. Risch, E. Leuzinger, B. O. Krüsi, and G. Achermann. 2003. Impact of herbivory by red deer (Cervus elaphus L.) on patterns and processes in subalpine grasslands in the Swiss National Park. Forest Ecology and Management 181: 177- 188. Siemann, E. 1998. Experimental tests of effects of plant productivity and diversity on grassland arthropod diversity. Ecology 79: 2057-2070. Steffan-Dewenter, I., and T. Tscharntke. 1999. Effects of habitat isolation on pollinator communities and seed set. Oecologia 121: 432-440. Straw, N. A., and A. R. Ludlow. 1994. Small-scale dynamics and insect diversity on plants. Oikos 71: 188-192. Symstad, A. J., E. Siemann and J. Haarstad 2000. An experimental test of the effect of plant functional group diversity on arthropod diversity. Oikos 89: 243-253. Takada, M., Y. G. Baba, Y. Yanagi, S. Terada, and T. Miyashita. 2008. Contrasting responses of web-building spiders to deer browsing among habitats and feeding guilds. Environmental Entomology 37: 938-946. Tremblay, J.-P., J. Huot, and F. Potvin. 2006. Divergent nonlinear responses of the boreal forest field layer along an experimental gradient of deer densities. Oecologia 150: 78-88. Tremblay, J.-P., J. Huot, and F. Potvin. 2007. Density-related effects of deer browsing on the regeneration dynamics of boreal forests. Journal of Applied Ecology 44: 552- 562. Vázquez, D. P., and D. Simberloff. 2003. Changes in interaction biodiversity induced by an introduced ungulate. Ecology Letters 6: 1077-1083. Vázquez, D. P., and D. Simberloff. 2004. Indirect effects of an introduced ungulate on pollination and plant reproduction. Ecological Monographs 74: 281-308. Virtanen, R., G. R. Edwards and M. J. Crawley 2002. Red deer management and vegetation on the Isle of Rum. Journal of Applied Ecology. 39: 572-583. Watts, C. H., and R. K. Didham. 2006. Rapid recovery of an insect-plant interaction following habitat loss and experimental wetland restoration. Oecologia 148: 61-69.

Table 5. Arthropod groups (*: pollinators; +: "herbivores and predators") studied vs. plant species (Canada thistle, dwarf cornel, fireweed, and raspberry) in the different experimental units (0, 7.5, 15 deer/km2 and 'U'ncontrolled density) and blocks (A, B, C) of a deer-controlled browsing experiment on Anticosti Island (Québec, Canada). A B C Species 0 7.5 15 U 0 7.5 15 U 0 7.5 15 U Canada thistle (C. arvense) + +* +* +* +* +* +* +* +* +* +* +* Dwarf cornel (C. canadensis) * * * * * * * * * * * * Fireweed (E. angustifolium) +* +* +* +* +* +* +* +* Raspberry (R. idaeus) + + + + + +

Table 6. Mixed ANOVA on abundance of the three dominant pollinators of the flowers of three plant species on Anticosti Island (Québec, Canada) as affected by deer density (four levels, fixed effect) and blocks (three blocks, random effect). Canada thistlea Dwarf cornelb Fireweedc 1 2 3 1 2 3 1 2 3 Source of variation df F F F df F F F df F F F Density 3 3.68 0.89 0.66 3 0.15 4.45 0.88 2 2.28 2.14 1.58 Error 5 6 3 a 1 = Thricops spiniger (Stein) (Diptera: Muscidae), 2 = Bombus frigidus Smith (Hymenoptera: Apidae), 3 = Bombus borealis Kirby (Hymenoptera: Apidae) b 1 = Phaonia serva (Meigen) (Diptera: Muscidae), 2 = Meliscaeva cinctella (Zetterstedt) (Diptera: Syrphidae), 3 = Evodinus m. monticola (Randall) (Coleoptera: Cerambycidae) c 1 = Bombus frigidus Smith (Hymenoptera: Apidae), 2 = Dolichovespula norvegicoides (Hymenoptera: Vespidae), 3 = Bombus borealis Kirby (Hymenoptera: Apidae)

Table 7. Proportion of variance explained by the block effect based on type 3 mixed ANOVAs on the abundance of dominant pollinators, grouped by family, on flowers of Canada thistle, dwarf cornel or fireweed in a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Significant block effect in bold. % Taxon variance df F Canada thistle Muscidae 80.3 2/5 15.25b Thricops spiniger (Stein) 78.0 2/5 13.35b Apidae 92.1 2/5 41.74c Bombus frigidus Smith 36.0 2/5 2.97 Bombus borealis Kirby 95.1 2/5 68.99c Syrphidae 21.0 2/5 1.93 Dwarf cornel Syrphidae 0 2/6 0.33 Meliscaeva cinctella (Zetterstedt) 38.5 2/6 3.50 Muscidae 0 2/6 0.34 Phaonia serva (Meigen) 40.4 2/6 3.71 Fireweed Apidae 81.6 2/3 12.09a Bombus frigidus Smith 88.2 2/3 20.78a Muscidae 6.2 2/3 1.17 Vespidae 59.1 2/3 4.61 Dolichovespula norvegicoides 38.6 2/3 2.57 (Sladen) a P < 0.05 b P < 0.01 c P < 0.001

Table 8. Mixed ANOVA on abundance of the four dominant herbivorous and predator arthropods found on three plant species on Anticosti Island (Québec, Canada), as affected by deer density (four levels, fixed effect), sampling period (n=2) and blocks (three blocks, random effect). Canada thistlea Fireweedb Raspberryc Sources 1 2 3 4 1 2 3 4 1 2 3 4 of variation df F F F F df F F F F df F F F F Density 3 4.31 0.65 2.62 3.40 2 0.66 0.89 3.83 12.89a 2 1.11 0.04 0.10 0.77 Error 1 6 4 2 Period 1 28.59c 4.74 84.36d 2.30 1 3.74 6.44 16.35 35.19b 1 5.49 2.17 1.86 0.14 Density x 3 3.01 0.69 2.39 0.98 2 0 1.86 0.50 0.08 2 1.03 0.29 0.02 5.05 Period Error 2 8 5 3 a 1 = Uroleucon cirsii (L.) (Hemiptera: Aphididae), 2 = Philaenus spumarius (L.) (Hemiptera: Cercopidae), 3 = Entomobrya sp. 1 (Collembola: Entomobryidae), 4 = Capitophorus sp. 1 (Hemiptera: Aphididae) b 1 = Aphis sp. (Hemiptera: Aphididae), 2 = Philaenus spumarius (L.) (Hemiptera: Cercopidae), 3 = Gelechiidae sp. 2 (Lepidoptera), 4 = Entomobrya sp. 1 (Collembola: Entomobryidae) c 1 = Amphorophora agathonica Hottes (Hemiptera: Aphididae), 2 = Entomobrya sp. 1 (Collembola: Entomobryidae), 3 = Clubiona sp. 1 (Aranaea: Clubionidae), 4 = Diastrophus rubi (Bouché) (Hymenoptera: Cynipidae) a P < 0.05 b P < 0.01 c P < 0.001 d P < 0.0001

Table 9. Proportion of variance explained by the block effect based on type 3 mixed ANOVAs on the abundance of dominant herbivores and predators, grouped by family, on Canada thistle, fireweed or raspberry in a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Significant block effect in bold. % Taxon variance df F Canada thistle Aphididae 24.6 2/17 3.61a Uroleucon cirsii (L.) 21.7 2/17 3.22 Capitophorus sp.1 10.1 2/17 1.90 Entomobryidae 0 2/17 0.27 Cercopidae 24.8 2/17 3.64a Philaenus spumarius (L.) 22.9 2/17 3.38 Platygastridae 6.4 2/17 1.54 Cecidomyiidae 0 2/17 0.26 Araneidae 6.7 2/17 1.57 Fireweed Aphididae 0 2/10 0.98 Aphis sp.1 0 2/10 0.96 Cecidomyiidae 49.4 2/10 5.88a Cercopidae 15.7 2/10 1.93 Philaenus spumarius (L.) 11.3 2/10 1.57 Gelechiidae 60.6 2/10 8.68b Entomobryidae 0 2/10 0.59 Raspberry Aphididae 0 1/7 0.58 Amphorophora agathonica Hottes 0 1/7 0.60 Cecidomyiidae 68.1 1/7 13.83b Miridae 0 1/7 0.03 Entomobryidae 0 1/7 0.76 Clubionidae 0 1/7 0.01 a P < 0.05 b P < 0.01 c P < 0.001

Figure 3: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance of pollinators associated with A) Canada thistle and B) dwarf cornel in twelve experimental units (except that Canada thistle was not sampled in density 0 deer/km2 of block A) of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (See annexe B and C). 58

Figure 4: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance of pollinators associated with fireweed in eight experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe D).

Figure 5: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with Canada thistle in twelve experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe B).

Figure 6: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with fireweed in eight experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe D).

Figure 7: Distance biplots (scaling 1) of principal component analysis (PCA) on the abundance A) of herbivorous and B) of predatory arthropods associated with raspberry in six experimental units of a white tailed deer controlled browsing experiment on Anticosti Island (Québec, Canada). Species are represented by (+) and the four first letters of the genus and species (see annexe E).

Conclusion

Les deux chapitres de ce mémoire traitent de l'impact des densités de cerfs de Virginie sur l'île d'Anticosti sur les insectes avec deux approches différentes. Dans un premier temps, nous avons regardé l'impact des modifications à l'échelle de la communauté végétale sur quatre groupes d'insectes ciblés en fonction de leur degré de relation avec les plantes. Ensuite, nous avons étudié l'impact à l'échelle d'espèces de plantes choisies selon leur abondance relative en fonction de la densité de cerfs sur les insectes qui leur sont associés. Les résultats démontrent que la densité de cerfs a une influence sur la composition des communautés d'insectes et dans certaines circonstances, peut même perturber les relations insectes-plantes. Dans les deux cas, plus la relation des groupes d'insectes étudiés avec les plantes est directe, plus ils sont affectés par les perturbations causées par les densités élevées de cerfs. Ainsi, les insectes herbivores sont les plus sensibles aux changements des communautés végétales dus aux cerfs, suivis des pollinisateurs et finalement des prédateurs. De plus, les associations insectes-plantes d'une plante hôte abondante dans un contexte ambiant (ici, la densité non contrôlée et élevée de cerfs), sont plus susceptibles d'être perturbées que pour les plantes peu abondantes.

Extinction et colonisation Le cadre de cette étude est différent des autres dans la mesure où plusieurs niveaux de densité ont été appliqués, que le cerf de Virginie est une espèce non-indigène à l'île d'Anticosti et que l'ensemble du territoire est perturbé par la surabondance de l'animal. Les changements dans la végétation ont commencé dans les années 1910-1920 et sont devenus particulièrement importants dans les années 1930 (Potvin et al. 2003). Dans un tel système, la surface des habitats occupée par les plantes les plus vulnérables au broutement diminue en taille et les parcelles restantes de ces plantes deviennent de plus en plus isolées. À l'échelle locale, ces parcelles isolées sont donc sujettes à une augmentation du taux d'extinction et une diminution du taux de colonisation des insectes et autres arthropodes qui leurs sont associés (MacArthur et Wilson 1967). Également, plus une plante ressource devient rare sous l'effet du cerf, plus la chance d'extinction à l'échelle de l'île des insectes 63 qui y sont directement associés (relation de dépendance) augmente avec le temps (Hanski et al. 1996). En 2001, lors de l'installation des exclos utilisés dans cette étude, les insectes intimement associés aux plantes fortement broutées, telles que l'épilobe (Epilobium angustifolium L.) et le framboisier (Rubus idaeus L.), étaient probablement majoritairement éteints ou présents à densité résiduelle dans quelques parcelles isolées. L'inverse devrait être observé pour les insectes associés aux plantes favorisées tel que le chardon (Cirsium arvense (L.) Scop.).

L'installation des exclos à densité réduite de cerfs en 2001 a permis de générer un habitat différent dans lequel des surfaces plus grandes et moins isolées de plantes ont pu se développer. La vitesse de colonisation d'un habitat par une espèce dépend de la taille de celui-ci et de la distance des populations sources de colonisateurs (Littlewood et al. 2009, Thomas et al. 1992). Les espèces les plus aptes à coloniser les nouveaux habitats se développant dans les exclos sont celles qui étaient abondantes dans les habitats les plus proches; soit les habitats à densité non contrôlée de cerfs. Par contre, ces espèces peuvent être défavorisées dans le nouvel environnement des exclos et donc subir une diminution d'abondance plus grande suite à la compétition avec d'autres espèces mieux adaptées et ainsi voir leur abondance diminuer (Nee et May 1992), comme nous l'avons observé pour plusieurs espèces dominantes des sites à densité non contrôlée de cerfs. Cette dynamique complexe peut modifier l'abondance relative de chaque espèce, ce qui explique probablement une partie des différences observées au niveau des communautés.

L'analyse du deuxième chapitre à l'échelle d'espèces végétales ciblées révèle des effets de la réduction de la densité de cerfs sur la structure des communautés d'insectes qui leur sont associées. Les herbivores associés au chardon, une espèce végétale très répandue en densité non contrôlée de cerfs, sont les plus affectés par la réduction de la densité de cerfs. En absence de cerf, où les surfaces de chardons sont devenues plus petites et isolées, les communautés d'insectes diffèrent des autres densités et incluent moins d'espèces monophages. À cette densité, le taux d'extinction des espèces monophages devrait être plus élevé qu'ailleurs et leur taux de colonisation plus faible (Halley et Dempster 1996). Pour

64 l'épilobe et le framboisier qui sont presque absents à densités non contrôlées de cerfs, très peu d'effets du contrôle de la densité de cerfs sont observés. Dans ces cas, la communauté végétale du site est possiblement le facteur déterminant de la structure de leur communauté (Siemann et al. 1998).

Subsistance des espèces Quelques espèces d'insectes monophages ont été retrouvées en grand nombre sur l'épilobe et le framboisier, alors que nous les considérions de prime abord comme probablement éteintes ou très rares sur l'île d'Anticosti dû à un fort impact négatif du cerf sur l'abondance de leur plante ressource. La recrudescence inattendue de ces espèces peut s'expliquer par trois processus: 1) leur subsistance localisée sur des parcelles isolées de leur plante hôte; 2) leur subsistance sur une ou des plantes hôtes alternatives non broutées; 3) la colonisation à partir de populations extra insulaires. Pour subsister sur des parcelles isolées, deux critères doivent en théorie être respectés: la "quantité minimale d'habitat propice" ("minimum amount of suitable habitat" ou MASH) et la "population minimale viable" ("minimum viable population" ou MVP) (Hanski et al. 1996). Quelques parcelles de framboisiers sont encore présentes dans les unités à densité non contrôlée de cerfs, mais nos données indiquent que les plants n'atteignent généralement pas leur pleine hauteur: leurs tiges atteignaient en moyenne 19 cm, contre 40 dans les unités à densités contrôlées. Elles peuvent probablement permettre la subsistance des espèces d'insectes qui étaient abondantes (Hanski et al. 1996) avant l'introduction du cerf, bien que des adaptations peuvent être nécessaires. Par exemple, le cynipide galligène Diastrophus rubi (Bouché) est connu pour initier ses galles sur les tiges du framboisier (Ronquist et Liljeblad 2001), mais sur les plants observés dans cette étude, 35 galles sur les 38 observées étaient à la jonction de la partie aérienne et des racines. Ce déplacement apparent des galles de D. rubi pourrait être une conséquence du broutement des tiges par les cerfs.

L'épilobe est plus problématique puisque seulement deux petites parcelles ont été trouvées durant les inventaires de végétation couvrant 117 parcelles de végétation à densité non contrôlée de cerfs. La subsistance sur les parcelles isolées d'épilobes semble donc

65 moins probable, surtout en considérant que les deux principales espèces oligophages (les micro Lépidoptères Scythris sp. (Xyloryctidae) et Gelechiidae sp. 2) sont présentes dans toutes les unités expérimentales inventoriées.

La subsistance sur des plantes hôtes alternatives est possible pour certaines espèces, dont les Aphis spp. qui ont parfois plus d'un hôte (Blackman et Eastop 2006). Par contre, les endophages comme Scythris sp. et Gelechiidae sp. 2 sont généralement spécialisées (Lewinsohn et al. 2005, Hespenheide 1991) et leur transfert sur un hôte alternatif nécessite une longue adaptation (Jobin et al. 1996). Le transfert est plus probable sur une plante biochimiquement similaire et proche phylogénétiquement (Becerra 1997), mais la circée alpine (Circaea alpina L.), seule autre plante de la même famille (Onagraceae) que l'épilobe sur l'île d'Anticosti est aussi rare.

Quant à la colonisation à partir de l'extérieur de l'île, elle dépend de plusieurs facteurs. La distance de 35 km du continent apparaît comme un problème qui n'est pas insurmontable pour des papillons tels les Gelechiidae et probablement les Xyloryctidae. Wu et al. (2006) ont démontré que le gelechiide Pectinophora gossypiella Saunders pouvait effectuer des vols sans escale d'en moyenne 41,5 km avec un maximum de 99,5 km. Cependant, une nouvelle espèce qui tente de s'installer dans un habitat déjà colonisé par une autre espèce est en principe fortement désavantagée, même si l'autre espèce est un compétiteur faible (Ward et Thornton 2000). Par contre, même si l'épilobe était déjà colonisé par d'autres insectes, aucun n'occupait leur niche particulière; soit l'intérieur des tiges pour Scythris sp. et l'intérieur des feuilles pour Gelechiidae sp. 2. La colonisation à partir de l'extérieur de l'île pour les Lépidoptères endophages semble donc être l'hypothèse la plus plausible, bien que cette colonisation semble avoir été très rapide, surtout en considérant que ces deux espèces étaient présentes dans l'ensemble des sites inventoriés. Pour les pucerons et les insectes associés au framboisier, l'hypothèse de la subsistance sur des parcelles isolées semble la plus adéquate.

Conservation Certaines espèces de plantes semblent à la base de la récupération des écosystèmes suite à une perturbation. L'identification de ces plantes et leur régénération seraient donc importantes lors de la mise en place de mesures de conservation. Dans le premier chapitre, nous avons identifié le sapin baumier et le framboisier comme étant des espèces associées aux communautés entomologiques en mutation favorisées par les réductions de la densité de cerfs et, dans le deuxième chapitre, nous avons vu que l'épilobe était favorable à l'établissement d'une faune entomologique spécialisée. Ces plantes semblent donc importantes pour le rétablissement des communautés entomologiques et sont, à tout le moins, des indicatrices de sites favorables à l'établissement d'espèces plus rares. Les résultats que nous avons observés sont à court terme (6 et 7 ans après la diminution des densités de cerfs), et il est probable que les écosystèmes soient toujours en mutation à ce stade. À plus long terme, le nombre de niches écologiques disponibles devrait augmenter dans les sites où l'abondance de cerfs a été réduite à des densités intermédiaires (7.5 ou 15 cerfs au km2) et ainsi favoriser l'augmentation de la diversité entomologique.

Nous avons identifié deux espèces de papillons, associées à des niches qui étaient probablement vacantes (l'intérieur des tissus de l'épilobe), qui pourraient provenir de l'extérieur de l'île d'Anticosti ce qui laisse supposer qu'à long terme, d'autres niches pourraient être occupées par des espèces extra-insulaires. Nos résultats démontrent donc que face à une forte perturbation, les insectes ont une forte résilience et ont une bonne capacité de recolonisation, du moins pour les espèces herbivores ailées, et que l'établissement de procédés permettant le maintien et le développement des espèces de plantes identifiées (sapin baumier, framboisier et épilobe) devraient être bénéfiques à la faune entomologique et à l'écosystème en général.

Bibliographie

Allombert, S., S. Stockton et J.-L. Martin 2005. A natural experiment on the impact of overabundant deer on forest invertebrates. Conservation Biology. 19: 1917-1929. Augustine, D. J. et S. J. McNaughton 1998. Ungulate effects on the functional species composition of plant communities: Herbivore selectivity and plant tolerance. Journal of Wildlife Management. 62: 1165-1183. Baines, D., R. B. Sage et M. M. Baines 1994. The implications of red deer grazing to ground vegetation and invertebrate communities of Scottish native pinewoods. Journal of Applied Ecology. 31: 776-783. Becerra, J. X. 1997. Insects on plants: Macroevolutionary chemical trends in host use. Science. 276: 253-256. Bee, J. N., A. J. Tanentzap, W. G. Lee, R. B. Lavers, A. F. Mark, J. A. Mills et D. A. Coomes 2009. The benefits of being in a bad neighbourhood: plant community composition influences red deer foraging decisions. Oikos. 118: 18-24. Blackman, R. L. et V. F. Eastop 2006. Aphids on the world's herbaceous plants and shrubs. Volume 2: The Aphids. John Wiley & Sons, Chichester, UK. Casabon, C. et D. Pothier 2008. Impact of deer browsing on plant communities in cutover sites on Anticosti Island. Écoscience. 15: 389-397. Caughley, G. 1970. Eruption of ungulate populations, with emphasis on Himalayan thar in New Zealand. Ecology. 51: 53-72. Caughley, G. 1981. Overpopulation, pp. 7-20. In P. A. Jewell et S. Holt (eds.), Problems in management of locally abundant wild mammals. Academic Press, New York, NY. Côté, S. D., T. P. Rooney, J.-P. Tremblay, C. Dussault et D. M. Waller 2004. Ecological impacts of deer overabundance. Annual Review of Ecology, Evolution and Systematics. 35: 113-147. Danell, K., R. Bergström, P. Duncan et J. Pastor (eds) 2006 Large herbivore ecology, ecosystem dynamics and conservation. Cambridge University Press, Cambridge, UK deCalesta, D. S. 1994. Effect of white-tailed deer on songbirds within managed forests in Pennsylvania. Journal of Wildlife Management. 58: 711-718. Den Herder, M., R. Bergstrom, P. Niemela, K. Danell et M. Lindgren 2009. Effects of natural winter browsing and simulated summer browsing by moose on growth and shoot biomass of birch and its associated invertebrate fauna. Annales Zoologici Fennici. 46: 63-74. Den Herder, M., R. Virtanen et H. Roininen 2004. Effects of reindeer browsing on tundra willow and its associated insect herbivores. Journal of Applied Ecology. 41: 870- 879. Dennis, P., M. R. Young et C. Bentley 2001. The effects of varied grazing management on epigeal spiders, harvestmen and pseudoscorpions of Nardus stricta grassland in upland Scotland. Agriculture, Ecosystems & Environment. 86: 39-57. Forsyth, D. M. et P. Caley 2006. Testing the irruptive paradigm of large-herbivore dynamics. Ecology. 87: 297-303. Frank, D. A., P. M. Groffman, R. D. Evans et B. F. Tracy 2000. Ungulate stimulation of nitrogen cycling and retention in Yellowstone Park grasslands. Oecologia. 123: 116-121. 68

Frelich, L. E. et C. G. Lorimer 1985. Current and predicted long-term effects of deer browsing in hemlock forests in Michigan, USA. Biological Conservation. 34: 99- 120. Gibson, C. W. D., C. Hambler et V. K. Brown 1992. Changes in spider (Araneae) assemblages in relation to succession and grazing management. Journal of Applied Ecology. 29: 132-142. Halley, J. M. et J. P. Dempster 1996. The spatial population dynamics of insects exploiting a patchy food resource: A model study of local persistence. Journal of Applied Ecology. 33: 439-454. Hanski, I., A. Moilanen et M. Gyllenberg 1996. Minimum viable metapopulation size. American Naturalist. 147: 527-541. Hespenheide, H. A. 1991. Bionomics of leaf-mining insects. Annual Review of Entomology. 36: 535-560. Horsley, S. B., S. L. Stout et D. S. DeCalesta 2003. White-tailed deer impact on the vegetation dynamics of a northern hardwood forest. Ecological Applications. 13: 98-118. Jobin, A., U. Schaffner et W. Nentwig 1996. The structure of the phytophagous insect fauna on the introduced weed Solidago altissima in Switzerland. Entomologia Experimentalis Et Applicata. 79: 33-42. Klein, D. R. 1968. Introduction increase and crash of reindeer on St Matthew Island. Journal of Wildlife Management. 32: 350-367. Kruess, A. et T. Tscharntke 2002. Grazing intensity and the diversity of grasshoppers, butterflies, and trap-nesting bees and wasps. Conservation Biology. 16: 1570-1580. Larsson, S. 1989. Stressful times for the plant stress: Insect performance hypothesis. Oikos. 56: 277-283. Lefort, S. 2002. Habitat hivernal du cerf de Virginie (Odocoileus virginianus) à l'île d'Anticosti. M. Sc. thesis, Université Laval, Québec, QC Lewinsohn, T. M., V. Novotny et Y. Basset 2005. Insects on plants: Diversity of herbivore assemblages revisited. Annual Review of Ecology Evolution and Systematics. 36: 597-620. Littlewood, N. A., R. J. Pakeman et S. J. Woodin 2009. Isolation of habitat patches limits colonisation by moorland Hemiptera. Journal of Insect Conservation. 13: 29-36. MacArthur, R. H. et E. O. Wilson 1967. The theory of island biogeography. Princeton University Press, Princeton, NJ. Martin, J.-L. et T. Daufresne 1996. Introduced species and their impact on the forest ecosystem of Haida Gwaii, pp. 69-85. In G. G. Wiggins (eds.), Proceedings of the cedar symposium. Canada-British Columbia South Moresby Forest Replacement Account, Victoria, BC. McCullough, D. R. 1997. Irruptive behavior in ungulates, pp. 69-98. In W. J. McShea, H. B. Underwood et J. H. Rappole (eds.), The Science of Overabundance: Deer ecology and population management. Smithsonian Books, Washington, DC. McShea, W. J. 2000. The influence of acorn crops on annual variation in rodent and bird populations. Ecology. 81: 228-238. McShea, W. J. et J. H. Rappole 2000. Managing the abundance and diversity of breeding bird populations through manipulation of deer populations. Conservation Biology. 14: 1161-1170.

McShea, W. J., H. B. Underwood et J. H. Rappole (eds) 1997 The Science of Overabundance: Deer ecology and population management. Smithsonian Books, Washington, DC Melis, C., A. Buset, P. A. Aarrestad, O. Hanssen, E. L. Meisingset, R. Andersen, A. Moksnes et E. Røskaft 2006. Impact of red deer Cervus elaphus grazing on bilberry Vaccinium myrtillus and composition of ground beetle (Coleoptera, Carabidae) assemblage. Biodiversity and Conservation. 15: 2049-2059. Melis, C., M. Sundby, R. Andersen, A. Moksnes, B. Pedersen et E. Røskaft 2007. The role of moose Alces alces L. in boreal forest - the effect on ground beetles (Coleoptera, Carabidae) abundance and diversity. Biodiversity and Conservation. 16: 1321-1335. Miyashita, T., M. Takada et A. Shimazaki 2004. Indirect effects of herbivory by deer reduce abundance and species richness of web spiders. Écoscience. 11: 74-79. Morris, M. G. et K. H. Lakhani 1979. Responses of grassland invertebrates to management by cutting. 1. Species diversity of Hemiptera. Journal of Applied Ecology. 16: 77- 98. Moser, B. W. et G. W. Witmer 2000. The effects of elk and cattle foraging on the vegetation, birds, and small mammals of the Bridge Creek Wildlife Area, Oregon. International Biodeterioration & Biodegradation. 45: 151-157. Nee, S. et R. M. May 1992. Dynamics of metapopulations: habitat destruction and competitive coexistence. Journal of Animal Ecology. 61: 37-40. Pastor, J., B. Dewey, R. J. Naiman, P. F. McInnes et Y. Cohen 1993. Moose browsing and soil fertility in the boreal forests of Isle Royale National Park. Ecology. 74: 467- 480. Perrins, C. M. et R. Overall 2001. Effect of increasing numbers of deer on bird populations in Wytham Woods, central England. Forestry. 74: 299-309. Potvin, F., P. Beaupré et G. Laprise 2003. The eradication of balsam fir stands by white- tailed deer on Anticosti Island, Québec: A 150-years process. Écoscience. 10: 487- 495. Potvin, F. et L. Breton 2005. From the field: Testing 2 aerial survey techniques on deer in fenced enclosures - visual double-counts and thermal infrared sensing. Wildlife Society Bulletin. 33: 317-325. Ronquist, F. et J. Liljeblad 2001. Evolution of the gall wasp-host plant association. Evolution. 12: 2503-2522. Rooney, T. P. et D. M. Waller 2003. Direct and indirect effects of white-tailed deer in forest ecosystems. Forest Ecology and Management. 181: 165-176. Scheffer, V. B. 1951. The Rise and Fall of a Reindeer Herd. Scientific Monthly. 73: 356- 362. Siemann, E., D. Tilman, J. Haarstad et M. Ritchie 1998. Experimental tests of the dependence of arthropod diversity on plant diversity. American Naturalist. 152: 738-750. Stewart, A. J. A. 2001. The impact of deer on lowland woodland invertebrates: a review of the evidence and priorities for future research. Forestry. 74: 259-270. Suominen, O. 1999. Impact of cervid browsing and grazing on the terrestrial gastropod fauna in the boreal forests of Fennoscandia. Ecography. 22: 651-658.

Suominen, O., J. Niemelä, P. Martikainen, P. Niemelä et I. Kojola 2003. Impact of reindeer grazing on ground-dwelling Carabidae and Curculionidae assemblages in Lapland. Ecography. 26: 503-513. Suominen, O., I. L. Persson, K. Danell, R. Bergstrom et J. Pastor 2008. Impact of simulated moose densities on abundance and richness of vegetation, herbivorous and predatory arthropods along a productivity gradient. Ecography. 31: 636-645. Takada, M., Y. G. Baba, Y. Yanagi, S. Terada et T. Miyashita 2008. Contrasting responses of web-building spiders to deer browsing among habitats and feeding guilds. Environmental Entomology. 37: 938-946. Thomas, C. D., J. A. Thomas et M. S. Warren 1992. Distributions of occupied and vacant butterfly habitats in fragmented landscapes. Oecologia. 92: 563-567. Tilghman, N. G. 1989. Impacts of white-tailed deer on forest regeneration in northwestern Pennsylvania. Journal of Wildlife Management. 53: 524-532. Tremblay, J.-P., J. Huot et F. Potvin 2006. Divergent nonlinear responses of the boreal forest field layer along an experimental gradient of deer densities. Oecologia. 150: 78-88. Tremblay, J.-P., J. Huot et F. Potvin 2007. Density-related effects of deer browsing on the regeneration dynamics of boreal forests. Journal of Applied Ecology. 44: 552-562. Vázquez, D. P. et D. Simberloff 2003. Changes in interaction biodiversity induced by an introduced ungulate. Ecology Letters. 6: 1077-1083. Vázquez, D. P. et D. Simberloff 2004. Indirect effects of an introduced ungulate on pollination and plant reproduction. Ecological Monographs. 74: 281-308. Viera, V. 2004. Effets à long terme du cerf de Virginie (Odocoileus virginianus) sur les communautés végétales de l'île d'Anticosti. M. Sc. thesis, Université Laval, Québec, QC Virtanen, R., G. R. Edwards et M. J. Crawley 2002. Red deer management and vegetation on the Isle of Rum. Journal of Applied Ecology. 39: 572-583. Waller, D. M. et W. S. Alverson 1997. The white-tailed deer: a keystone herbivore. Wildlife Society Bulletin. 25: 217-226. Ward, S. A. et I. W. B. Thornton 2000. Chance and determinism in the development of isolated communities. Global Ecology and Biogeography. 9: 7-18. Watson, A. 1983. Eighteenth century deer numbers and pine regeneration near Braemar, Scotland. Biological Conservation. 25: 289-305. Wu, H. H., K. M. Wu, D. Y. Wang et Y. Y. Guo 2006. Flight potential of pink bollworm, Pectinophora gossypiella Saunders (Lepidoptera : Gelechiidae). Environmental Entomology. 35: 887-893.

Annexe A. Liste des espèces échantillonnés à l'aide de piège Luminoc® (Carabidae et macro Lepidoptera) et Malaise (Apoidea et Syrphidae) et leur abondance cumulée dans un dispositif à broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) et répété dans trois localités. Densité Famille Taxon 0 7.5 15 N Total Apoidea Andrenidae 01 - Andrena spp. 6 15 3 7 31 Apidae 02 - Bombus ashtoni (Cresson) 0 1 0 0 1 03 - Bombus borealis Kirby 22 56 16 75 169 04 - Bombus fernaldae (Franklin) 9 15 1 4 29 05 - Bombus frigidus Smith 125 195 106 47 473 06 - Bombus insularis Smith 7 8 2 1 18 07 - Bombus ternarius Say 15 8 5 6 34 Colletidae 08 - Colletes consors Cresson 0 0 5 0 5 09 - Colletes impunctatus Nylander 12 31 24 7 74 10 - Hylaeus basalis (Smith) 2 1 0 0 3 Halictidae 11 - Halictus confusus Smith 1 6 2 0 9 12 - Halictus rubicundus (Christ) 2 3 1 0 6 13 - Lasioglossum foxii (Robertson) 2 1 3 0 6 14 - Lasioglossum quebecensis 14 4 6 13 37 Crawford 15 - Lasioglossum rufitarse 6 5 3 7 21 (Zetterstedt) 16 - Lasioglossum sp.1 0 2 0 0 2 17 - Lasioglossum sp.2 0 1 0 0 1 18 - Lasioglossum sp.3 2 2 1 4 9 19 - Lasioglossum sp.4 1 0 0 1 2 20 - Lasioglossum sp.5 1 2 0 0 3 21 - Lasioglossum sp.6 0 4 0 0 4 22 - Sphecodes solonis Graenicher 1 1 0 2 4 Megachilidae 23 - Anthophora terminalis Cresson 5 7 10 1 23 24 - Coelioxys germana Cresson 0 8 2 0 10 25 - Megachile frigida Smith 58 63 28 22 171 26 - Megachile inermis Provancher 21 14 8 18 61 27 - Megachile melanophaea Smith 1 8 13 8 30 28 - Megachile montivaga Cresson 1 1 2 0 4 29 - Megachile relativa Cresson 3 8 5 1 17 30 - Osmia atriventris Cresson 1 0 0 0 1 31 - Osmia bucephala Cresson 0 1 1 0 2 32 - Osmia inermis (Zetterstedt) 0 1 0 0 1 33 - Osmia nigriventris (Zetterstedt) 0 0 0 2 2 34 - Osmia proxima Cresson 9 16 8 3 36 35 - Osmia tersula Cockerell 2 4 2 1 9

Carabidae 01 - Agonum affine Kirby 0 2 0 0 2 72

Densité Famille Taxon 0 7.5 15 N Total 02 - Agonum gratiosum (Mannerheim) 0 1 0 0 1 03 - Amara aulica (Panzer) 1 7 1 23 32 04 - Amara lunicollis Schiodte 0 0 0 3 3 05 - Amara sinuosa (Casey) 0 0 0 1 1 06 - Bembidion grapii Gyllenhal 0 0 0 1 1 07 - Bembidion wingatei Bland 0 0 1 0 1 08 - Calathus advena (LeConte) 1 66 0 0 67 09 - Calathus ingratus Dejean 11 37 8 30 86 10 - Carabus maeander maeander 1 0 0 0 1 Fischer 11 - Carabus serratus Say 0 0 1 2 3 12 - Cymindis cribricollis Dejean 0 0 0 2 2 13 - Dicheirotrichus cognatus 1 0 0 0 1 (Gyllenhal) 14 - Harpalus fulvilabris Mannerheim 2 4 2 3 11 15 - Harpalus laticeps LeConte 0 2 1 0 3 16 - Harpalus megacephalus LeConte 0 0 1 0 1 17 - Harpalus nigritarsis Sahlberg 0 1 0 0 1 18 - Harpalus rufipes DeGeer 2 31 2 18 53 19 - Harpalus somnulentus Dejean 1 10 3 3 17 20 - Patrobus foveocollis (Eschscholtz) 0 3 0 1 4 21 - Platynus decentis (Say) 0 0 0 2 2 22 - Pterostichus adstrictus Eschscholtz 79 82 65 116 342 23 - Pterostichus coracinus (Newman) 93 145 98 79 415 24 - Pterostichus melanarius (Illiger) 3 1 1 57 62 25 - Pterostichus pensylvanicus 12 1 0 2 15 (LeConte) 26 - Pterostichus punctatissimus 6 9 14 10 39 Randall 27 - Sphaeroderus nitidicollis 4 1 4 0 9 nitidicollis Chevrolat 28 - Syntomus americanus LeConte 0 2 0 2 4 29 - Synuchus impunctatus (Say) 156 123 139 278 696 30 - Trechus apicalis Motschulsky 0 3 0 0 3 Lepidoptera Arctiidae 01 - Eilema bicolor Grote 2 1 4 0 7 02 - Platarctia parthenos Harris 0 1 1 1 3 03 - Spilosoma virginica Fabricius 1 2 0 1 4 Drepanidae 04 - Drepana bilineata Packard 0 0 0 1 1 05 - Habrosyne scripta (Gosse) 3 3 3 1 10 06 - Pseudothyatira cymatophoroides 0 0 3 0 3 (Guenée) Geometridae 07 - Cabera erythemaria Guenée 0 10 0 0 10

Densité Famille Taxon 0 7.5 15 N Total 08 - Cabera variolaria Guenée 0 0 5 0 5 09 - Caripeta divisata Walker 16 25 8 6 55 10 - Chloroclysta citrata (Linnaeus) 1 4 0 0 5 11 - Chloroclysta hersiliata Guenée 0 0 0 2 2 12 - Chloroclysta truncata traversata 4 9 6 2 21 (Kellicott) 13 - Cyclophora pendulinaria Guenée 12 7 8 2 29 14 - Ecliptopera silaceata albolineata Packard 3 5 3 0 11 15 - Ectropis crepuscularia Denis & Schiffermüller 0 0 1 0 1 16 - Epirrhoe alternata Müller 2 2 2 2 8 17 - Euchlaena muzaria (Walker) 0 2 1 0 3 18 - Eulithis destinata Möschler 0 1 0 0 1 19 - Eulithis explanata Walker 4 1 4 5 14 20 - Eulithis propulsata Walker 0 0 1 0 1 21 - Eumacaria latiferrugata brunneata 0 1 3 0 4 Packard 22 - Euphyia intermediata Guenée 0 1 1 3 5 23 - Eupithecia sp. 30 22 31 3 86 24 - Eustroma semiatrata (Hulst) 1 2 1 0 4 25 - Homochlodes fritillaria (Guenée) 0 1 1 2 4 26 - Horisme intestinata Guenée 8 4 6 10 28 27 - Hydriomena divisaria frigidata 5 6 11 3 25 Walker 28 - Hypagyrtis sp. 1 0 0 0 1 29 - Itame bitactata Walker 5 3 2 1 11 30 - Macaria marmorata (Ferguson) 1 2 4 7 14 31 - Macaria subcessaria (Walker) 0 1 0 0 1 32 - Mesoleuca ruficillata Guenée 1 2 2 1 6 33 - Nematocampa resistaria Herrich- 0 2 2 0 4 Schäffer 34 - Pero morrisonaria (Edwards) 1 2 1 0 4 35 - Probole alienaria Herrich-Schäffer 1 0 0 1 2 36 - Rheumaptera hastata gothicata 17 22 5 0 44 Guenée 37 - Scopula ancellata (Hulst) 0 0 1 0 1 38 - Scopula frigidaria Möschler 28 16 23 16 83 39 - Scopula junctaria (Walker) 5 2 3 1 11 40 - Sicya macularia Harris 0 1 0 0 1

41 - Spargania luctuata obductata 2 0 0 0 2 Möschler

Densité Famille Taxon 0 7.5 15 N Total 42 - Spargania magnoliata Guenée 3 4 0 2 9 43 - Xanthorhoe abrasaria congregata 25 22 30 55 132 Walker 44 - Xanthorhoe ferrugata Clerck 5 6 7 0 18 45 - Xanthorhoe iduata Guenée 3 3 5 4 15 46 - Xanthorhoe munitata Hübner 10 7 8 6 31 47 - Xanthotype urticaria Swett 13 5 8 2 28 48 – Geometridae sp. 1 0 1 0 0 1 Lymantriidae 49 - Dasychira plagiata (Walker) 8 1 4 8 21 50 - Leucoma salicis (Linnaeus) 0 0 1 1 2 Noctuidae 51 - Acronicta grisea Walker 1 0 0 0 1 52 - Acronicta innotata Guenée 0 1 0 0 1 53 - Acronicta oblinita (Smith) 2 1 0 0 3 54 - Actebia fennica (Tauscher) 0 0 0 1 1 55 - Amphipoea americana (Speyer) 2 4 2 0 8 56 - Anaplectoides pressus (Grote) 2 4 1 0 7 57 - Apamea contradicta (Smith) 0 0 1 0 1 58 - Apamea impulsa (Guenée) 1 0 0 0 1 59 - Apamea verbascoides (Guenée) 0 1 0 0 1 60 - Apharetra dentata (Grote) 1 0 1 1 3 61 - Aplectoides condita (Guenée) 0 0 0 1 1 62 - Autographa precationis (Guenée) 1 0 0 1 2 63 - Bleptina caradrinalis Guenée 1 0 0 0 1 64 - Caenurgina crassiscula (Haworth) 0 0 3 0 3 65 - Chrysanympha formosa (Grote) 0 0 3 0 3 66 - Diachrysia aereoides (Grote) 1 1 0 0 2 67 - Diarsia dislocata Smith 2 2 3 3 10 68 - Diarsia jucunda (Walker) 8 21 12 14 55 69 - Diarsia rosaria (Grote) 3 7 10 1 21 70 - Diarsia rubifera (Grote) 3 3 1 0 7 71 - Enargia infumata (Grote) 3 8 10 0 21 72 - Euplexia benesimilis McDunnough 3 3 1 2 9 73 - Euxoa comosa (Morrison) 0 0 0 1 1 74 - Euxoa dissona (Möschler) 5 1 0 0 6 75 - Faronta diffusa (Walker) 0 0 0 1 1 76 - Graphiphora augur (Fabricius) 2 1 0 0 3 77 - Hyppa xylienoides Guenée 1 5 3 2 11 78 - Idia aemula Hübner 0 0 1 0 1 79 - Idia americalis (Guenée) 7 9 10 11 37 80 - Lacanobia nevadae (Grote) 0 0 0 1 1 81 - Lacanobia radix (Walker) 2 7 2 0 11 82 - Lacinipolia anguina (Grote) 0 0 1 0 1 83 - Leucania insueta Guenée 0 0 2 0 2

Densité Famille Taxon 0 7.5 15 N Total 84 - Leucania multilinea Walker 9 3 3 16 31 85 - Lithacodia albidula Guenée 9 4 8 6 27 86 - Mamestra curialis (Smith) 2 0 0 0 2 87 - Mycterophora inexplicata 0 1 0 0 1 (Walker) 88 - Palthis angulalis Hübner 22 42 24 9 97 89 - Panthea acronyctoides (Walker) 0 0 0 1 1 90 - Papestra quadrata (Smith) 0 0 1 0 1 91 - Paradiarsia littoralis (Packard) 0 0 1 0 1 92 - Phalaenostola metonalis (Walker) 1 0 0 0 1 93 - Phlogophora periculosa Guenée 2 5 3 0 10 94 - Polia propodea McCabe 0 0 1 0 1 95 - Polia purpurissata (Grote) 0 1 0 0 1 96 - Pseudostrotia carneola (Guenée) 0 0 1 0 1 97 - Rivula propinqualis Hübner 0 6 12 10 28 98 - Syngrapha alias (Ottolengui) 2 0 1 0 3 99 - Syngrapha viridisigma (Grote) 3 1 1 0 5 100 - Xestia homogena (McDunnough) 1 4 4 2 11 101 - Xestia imperita (Hübner) 0 0 1 3 4 102 - Xestia mixta (Walker) 2 12 10 9 33 103 - Xestia perquiritata (Morrison) 0 13 2 2 17 104 - Xestia smithii (Snellen) 8 36 4 2 50 105 - Xestia youngii (Smith) 0 0 1 0 1

Syrphidae 01 - Blera armillata (Osten Sacken) 0 1 0 0 1 02 - Blera badia (Walker) 0 1 0 0 1 03 - Blera confusa Johnson 10 13 6 6 35 04 - Blera nigra (Williston) 5 15 9 9 38 05 - Chalcosyrphus inarmatus (Hunter) 0 0 1 0 1 06 - Chalcosyrphus libo (Walker) 6 24 1 4 35 07 - Chalcosyrphus nemorum 6 4 0 0 10 (Fabricius) 08 - Chalcosyrphus pigra (Fabricius) 1 2 0 1 4 09 - Cheilosia sp. 0 4 0 0 4 10 - Chrysogaster antitheus Walker 0 0 6 0 6 11 - Chrysotoxum derivatum Walker 30 22 26 64 142 12 - Chrysotoxum flavifrons Macquart 21 44 25 10 100 13 - Dasysyrphus venustus (Meigen) 4 3 9 4 20 14 - Doros aequalis Loew 0 0 1 0 1 15 - Epistrophe emarginata (Say) 2 0 3 0 5 16 - Epistrophe nitidicollis (Meigen) 6 2 2 0 10 17 - Epistrophe xanthostoma 0 2 1 0 3 (Williston)

Densité Famille Taxon 0 7.5 15 N Total 18 - Eriozona laxa (Osten Sacken) 8 6 10 15 39 19 - Eristalis sp.1 0 0 3 0 3 20 - Eristalis barda Say 3 1 0 0 4 21 - Eristalis dimidiatus Wiedemann 9 16 18 6 49 22 - Eristalis obscurus Loew 0 0 3 3 6 23 - Eupeodes americanus 2 1 8 0 11 (Wiedemann) 24 - Eupeodes curtus (Hine) 4 3 6 7 20 25 - Eupeodes lapponicus (Zetterstedt) 85 124 80 77 366 26 - Eupeodes latifasciatus (Macquart) 0 3 3 4 10 27 - Eupeodes luniger (Meigen) 5 29 9 6 49 28 - Eupeodes perplexus (Osborn) 32 43 60 41 176 29 - Eupeodes pomus (Curran) 1 1 1 0 3 30 - Helophilus fasciatus Walker 3 5 3 5 16 31 - Helophilus obscurus Loew 2 0 2 2 6 32 - Heringia coxalis (Curran) 7 16 6 6 35 33 - Heringia rita (Curran) 0 1 0 2 3 34 - Heringia salax (Loew) 0 0 1 0 1 35 - Heringia sp.1 9 8 6 6 29 36 - Heringia sp.2 1 3 1 2 7 37 - Heringia sp.3 1 1 0 0 2 38 - Heringia sp.4 0 0 0 2 2 39 - Lejops lineata Fabricius 0 1 0 0 1 40 - Leucozona lucorum (Linnaeus) 5 6 4 9 24 41 - Melangyna labiatarum (Verrall) 0 1 1 1 3 42 - Melangyna lasiophthalma 39 129 87 47 302 (Zetterstedt) 43 - Melangyna triangulifera 1 0 1 0 2 (Zetterstedt) 44 - Melangyna umbellatarum 0 0 1 1 2 (Fabricius) 45 - Melanostoma mellinum (Linnaeus) 186 698 533 1252 2669 46 - Meliscaeva cinctella (Zetterstedt) 10 39 20 11 80 47 - Microdon spp. 0 5 2 1 8 48 - Orthonevra pulchella Williston 4 1 1 3 9 49 - Paragus haemorrhous Meigen 1 0 0 0 1 50 - Parasyrphus genualis (Williston) 3 15 11 6 35 51 - Parasyrphus nigritarsis 0 0 1 0 1 (Zetterstedt) 52 - Parasyrphus sp.1 5 9 10 1 25 53 - Parasyrphus relictus (Zetterstedt) 0 6 1 0 7 54 - Parasyrphus tarsatus (Zetterstedt) 0 0 0 1 1 55 - Parhelophilus porcus Walker 0 1 0 1 2

Densité Famille Taxon 0 7.5 15 N Total 56 - Parhelophilus rex Curran 2 12 3 1 18 57 - Pipiza femoralis Loew 0 0 0 3 3 58 - Pipiza macrofemoralis Curran 0 0 0 1 1 59 - Pipiza quadrimaculata Panzer 1 0 4 1 6 60 - Platycheirus albimanus (Fabricius) 20 22 19 18 79 61 - Platycheirus amplus Curran 1 0 0 0 1 62 - Platycheirus angustatus 0 6 8 26 40 (Zetterstedt) 63 - Platycheirus concinnus (Snow) 0 0 2 4 6 64 - Platycheirus confusus (Curran) 11 32 17 10 70 65 - Platycheirus holarcticus 2 2 0 0 4 Vockeroth 66 - Platycheirus hyperboreus (Staeger) 10 16 15 35 76 67 - Platycheirus immarginatus 1 9 1 14 25 (Zetterstedt) 68 - Platycheirus inversus Ide 4 2 1 0 7 69 - Platycheirus jaerensis Nielsen 0 1 1 0 2 70 - Platycheirus modestus Ide 0 0 0 3 3 71 - Platycheirus nearcticus Vockeroth 47 86 46 27 206 72 - Platycheirus obscurus (Say) 21 31 49 25 126 73 - Platycheirus podagratus 0 0 0 1 1 (Zetterstedt) 74 - Platycheirus rosarum (Fabricius) 0 1 2 1 4 75 - Platycheirus scutatus (Meigen) 4 9 7 11 31 76 - Platycheirus varipes Curran 0 0 1 2 3 77 - Rhingia nasica Say 0 1 1 0 2 78 - Sericomyia chrysotoxoides 2 4 0 2 8 Macquart 79 - Sericomyia lata Coquillett 5 11 3 1 20 80 - Sericomyia militaris Walker 7 7 6 4 24 81 - Sericomyia transversa Osborn 0 1 1 0 2 82 - Sphaerophoria abbreviata 6 19 7 21 53 Zetterstedt 83 - Sphaerophoria asymetrica 3 3 6 5 17 Knutson 84 - Sphaerophoria bifurcata Knutson 7 11 4 6 28 85 - Sphaerophoria brevipilosa 1 0 0 0 1 Knutson 86 - Sphaerophoria contigua Macquart 0 2 1 0 3 87 - Sphaerophoria immarginatus 0 0 0 3 3 (Zetterstedt) 88 - Sphaerophoria novaeangliae 2 1 8 1 12 Johnson

Densité Famille Taxon 0 7.5 15 N Total 89 - Sphaerophoria philanthus 4 14 41 58 117 (Meigen) 90 - Syrphus rectus Osten Sacken 3 1 4 3 11 91 - Syrphus ribesii (Linnaeus) 40 63 35 28 166 92 - Syrphus torvus Osten Sacken 26 19 44 35 124 93 - Syrphus vitripennis Meigen 1 2 0 5 8 94 - Temnostoma vespiformis 7 3 2 3 15 (Linnaeus) 95 - Toxomerus marginatus (Say) 1 3 3 10 17 96 - Trichopsomyia apisaon (Walker) 3 5 1 3 12 97 - Trichopsomyia recedens (Walker) 1 0 0 0 1 98 - Trichopsomyia sp.1 1 0 0 0 1 99 - Tropidia quadrata (Say) 4 3 2 3 12 100 - Volucella bombylans (Linnaeus) 1 3 3 0 7 101 - Xylota annulifera Bigort 1 7 5 9 22 102 - Xylota confusa Shannon 6 6 1 8 21 103 - Xylota flavitibia Bigot 1 6 6 4 17 104 - Xylota flukei (Curran) 0 0 1 1 2 105 - Xylota hinei (Curran) 6 10 4 14 34 106 - Xylota naknek Shannon 9 9 1 2 21 107 - Xylota quadrimaculata Loew 11 24 6 15 56 108 - Xylota segnis (Linnaeus) 0 2 0 0 2 109 - Xylota subfasciata Loew 14 44 6 2 66

Annexe B. Liste des arthropodes (N>1) échantillonnés sur les plants ou les fleurs de chardon (Cisium arvense) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) répété dans trois localités. Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 N Total Acari Actinedida sp.4 In Pr 3 1 1 0 5 Actinedida sp.8 In Pr 1 1 0 0 2 Oribatida sp.3 In Pr 0 0 2 0 2 Oribatida sp.4 In Pr 0 3 0 0 3 Tetranychidae Tetranychidae sp.1 A-I Ph 21 13 18 31 83 Aranaea Araneidae Araneidae sp.2 I Pr 0 2 5 5 12 Araneidae sp.3 I Pr 0 0 11 0 11 Araneidae sp.5 I Pr 1 4 12 9 26 Araneidae sp.8 I Pr 4 5 4 0 13 Clubionidae Clubiona sp.1 A-I Pr 17 12 9 11 49 Linyphiidae Linyphiidae sp.2 A-I Pr 1 2 17 4 24 Linyphiidae sp.3 I Pr 1 0 0 1 2 Linyphiidae sp.4 A-I Pr 1 4 1 2 8 Linyphiidae sp.5 I Pr 1 1 6 2 10 Linyphiidae sp.7 A-I Pr 0 0 0 5 5 Linyphiidae sp.9 I Pr 0 0 5 0 5 Philodromidae Philodromidae sp.1 I Pr 4 2 4 5 15 Salticidae Salticidae sp.1 A-I Pr 0 1 1 13 15 Salticidae sp.2 I Pr 1 2 0 0 3 Salticidae sp.3 A-I Pr 0 1 0 1 2 Theridiidae Theridula emertoni Levi A-I Pr 5 8 6 1 20 Thomisidae Misumenops sp. 1 A-I Pr 12 10 9 7 38 Thomisidae sp.2 A-I Pr 0 2 0 0 2 ? Spider sp.9 I Pr 11 0 3 4 18 Coleoptera Latridiidae Corticaria spp. A S 0 1 1 0 2 Corticarina cavicollis A S 1 1 4 1 7 80

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 N Total (Mannerheim) Cortinicara gibosa (Herbst) A S 3 3 2 7 15 Collembola Entomobryidae Entomobrya sp.1 A S 46 85 188 182 501 Diptera Anthomyiidae I-P Ph 5 4 6 5 20 Anthomyiidae A Po 4 4 3 6 17 Bibionidae A In 0 0 2 0 2 Calliphoridae A Po 1 0 4 11 16 Cecidomyiidae A-I-P In 47 36 50 52 185 Chironomidae A In 10 8 8 5 31 Lauxaniiae A In 0 1 0 1 2 Muscidae Eudasyphora sp. A Po 0 0 1 1 2 Hydrotaea militaris (Meigen) A Po 0 1 1 2 4 Hydrotaea scambus (Zetterstedt) A Po 0 1 4 4 9 Morellia micans (Macquart) A Po 1 4 5 2 12 Muscidae spp. A-I In 8 3 2 0 13 Phaonia curvipes (Stein) A Po 0 1 1 1 3 Thricops spiniger (Stein) A Po 34 98 208 155 495 Mycetophagidae A Po 0 1 0 1 2 Phoridae A S 0 0 1 2 3 Simuliidae A Au 0 1 1 0 2 Syrphidae Chrysotoxum derivatum Walker A Po 2 0 2 1 5 Epistrophe nitidicollis (Meigen) A Po 0 2 0 0 2 Eristalis dimidiata Wiedemann A Po 0 1 0 2 3 Eupeodes lapponicus (Zetterstedt) A Po 0 1 1 1 3 Meliscaeva cinctella (Zetterstedt) A Po 0 2 0 8 10 Sericomyia militaris Walker A Po 0 0 1 1 2 Syrphus torvus Osten Sacken A Po 0 3 0 7 10 Toxomerus marginatus (Say) A Po 0 1 1 0 2 Ulidiidae A In 0 0 2 2 4

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 N Total Hemiptera Aphididae Capitophorus sp.1 A-I Ph 22 87 111 101 321 Uroleucon cirsii (Linné) A-I Ph 17 86 246 815 1164 Aphididae sp.6 A-I Ph 4 61 0 6 71 Cercopidae Aphrophora quadrinotata Say A-I Ph 1 1 0 1 3 Neophilaenus lineatus (Linné) A Ph 0 1 1 1 3 Philaenus spumarius (Linné) A-I Ph 75 115 131 97 418 Cercopidae sp.2 I Ph 3 0 0 0 3 Cicadellidae Draeculacephala sp. A Ph 5 1 0 0 6 Scaphytopius sp. A-I Ph 4 1 4 0 9 Cicadellidae sp.2 I Ph 2 1 0 1 4 Cixiidae A-I Ph 0 0 1 1 2 Delphacidae A Ph 2 0 1 1 4 Miridae Lygus borealis Kelton A-I Ph 4 3 6 19 32 Plagiognathus obscurus Uhler A-I Ph 3 0 6 9 18 Stenotus binotatus Fabricius A-I Ph 0 2 0 4 6 Miridae sp.1 I Ph 1 0 1 0 2 Pentatomidae I Ph 0 0 1 1 2 Reduviidae Reduviidae sp.1 I Pr 0 0 2 3 5 Rhopalidae I Ph 0 0 0 2 2 Hymenoptera Apidae Bombus borealis Kirby A Po 0 23 30 39 92 Bombus fernaldae (Franklin) A Po 0 2 1 0 3 Bombus frigidus Smith A Po 2 56 32 23 113 Bombus ternarius Say A Po 0 3 3 2 8 Braconidae A-I Pa 2 1 0 1 4 Colletidae Colletes impunctatus Nylander A Po 0 0 0 2 2 Crabronidae A Po 0 2 2 1 5 Eucoilidae A Pa 2 3 2 2 9 Eulophidae A-P Pa 4 5 8 21 38 Eurytomidae A Ph 0 0 12 1 13

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 N Total Formicidae Formica aserva Forel A Au 0 0 1 1 2 Formica hewetti Wheeler A Au 1 2 1 1 5 Leptothorax sp. A Au 0 0 2 5 7 Halictidae Lasioglossum foxii Robertson A Po 1 0 1 0 2 Lasioglossum quebecensis A Po 4 5 3 0 12 Crawford Lasioglossum sp.1 A Po 1 2 0 0 3 Ichneumonidae P Pa 1 0 0 1 2 Platygastridae A Pa 22 95 57 111 285 Pompilidae A Po 0 1 1 0 2 Pteromalidae A Pa 1 6 4 5 16 Scelionidae A Pa 1 2 3 1 7 Tachinidae A Po 1 0 1 4 6 Tenthedinidae Tenthredinidae sp.4 I Ph 0 0 0 2 2 Torymidae A Pa 0 0 2 0 2 Vespidae Dolichovespula norvegicoides A Po 2 3 2 3 10 (Sladen) Vespula acadica (Sladen) A Po 0 0 1 1 2 Vespula consobrina (de Saussure) A Po 2 2 0 1 5 Lepidoptera Gelechiidae Gelechiidae sp.1 I Ph 1 0 0 2 3 Gelechiidae A Po 2 1 0 0 3 Geometridae Geometridae sp.3 I Ph 0 1 1 0 2 Geometridae sp.4 I Ph 0 2 0 0 2 Hesperiidae Thymelicus lineola (Ochsenheimer) A Po 0 0 2 1 3 Noctuidae Noctuidae sp.2 I Ph 0 1 1 0 2 Tortricidae A-I Ph 5 4 2 3 14 Neuroptera Hemerobiidae A-I Pr 1 0 1 0 2 Thysanoptera Thripidae Thripidae sp.2 A-I Ph 0 1 3 8 12 Thripidae sp.3 A Ph 0 1 1 0 2

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 N Total Thripidae sp.4 I Ph 0 0 1 7 8 Thripidae sp.5 A Ph 1 1 0 1 3 a A=Adulte, I=Immature, P=Pupe, In=Inconnu/Incertain. b Au=Autre, Pa=Parasitoïde, Ph=Phytophage, Po=Pollinisateur, Pr=Prédateur, S=Saprophage, In=Inconnu/Incertain.

Annexe C. Liste des insectes pollinisateurs (N>1) échantillonnés sur les fleurs de cornouiller du Canada (Cornus canadensis) avec leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant quatre densités de cerfs (0, 7.5, 15 cerfs/km2 et 'N'on contrôlée) répété dans trois localités. Densité Ordre Famille Taxon 0 7.5 15 N Total Coleoptera Cantharidae Silis sp. 1 1 0 0 2 Cerambycidae Evodinus m. monticola (Randall) 4 2 3 7 16 Hymenoptera Crabronidae 0 2 2 5 9 Halictidae Lasioglossum quebecensis Crawford 2 1 0 3 6 Lasioglossum rufitarse (Zetterstedt) 2 0 0 0 2 Lasioglossum sp.1 4 0 0 1 5 Ichneumonidae 0 2 1 0 3 Pompilidae 0 1 0 1 2 Diptera Anthomyiidae 5 7 1 11 24 Bibionidae 3 1 0 0 4 Calliphoridae 1 0 1 0 2 Culicidae 1 0 0 1 2 Empididae 0 2 2 0 4 Muscidae Eudasyphora sp. 1 1 2 0 4 Helina sp. 3 3 0 1 7 Hydrotaea militaris (Meigen) 0 1 0 1 2 Hydrotaea scambus (Zetterstedt) 2 2 0 0 4 Morellia micans (Macquart) 7 3 4 0 14 Phaonia serva (Meigen) 29 28 23 26 106 Phaonia subfuscinervis (Zetterstedt) 5 4 0 0 9 Thricops innocuus (Zetterstedt) 3 7 1 0 11 Thricops spiniger (Stein) 4 0 0 0 4 Sciomyzidae 8 1 0 0 9 Stratiomyidae 1 1 0 1 3 Syrphidae Brachyopa ferruginea Fallen 4 3 1 1 9 Chalcosyrphus libo (Walker) 1 1 1 0 3

Densité Ordre Famille Taxon 0 7.5 15 N Total Chalcosyrphus nemorum (Fabricius) 0 2 0 0 2 Cheilosia sp. 0 2 0 1 3 Chrysotoxum flavifrons Macquart 0 1 0 1 2 Dasysyrphus venustus (Meigen) 1 1 0 0 2 Eristalis obscurus Loew 0 5 0 0 5 Melangyna umbellatarum (Fabricius) 2 0 1 0 3 Melanostoma mellinum (Linné) 2 4 0 3 9 Meliscaeva cinctella (Zetterstedt) 37 38 18 3 96 Orthonevra pulchella Williston 0 1 1 0 2 Parasyrphus genualis (Williston) 0 0 3 0 3 Parasyrphus relictus (Zetterstedt) 1 3 0 0 4 Parasyrphus sp.1 4 4 1 1 10 Sericomyia militaris Walker 1 1 0 1 3 Sphegina rufiventris Loew 3 1 2 0 6 Syrphus ribesii (Linné) 1 1 0 0 2 Syrphus torvus Osten Sacken 2 1 1 0 4 Syrphus vitripennis Meigen 0 0 1 2 3 Temnostomata vespiformis (Linné) 1 1 1 0 3 Xylota flavitibia Bigot 1 2 0 1 4 Xylota quadrimaculata Loew 1 1 0 0 2 Xylota subfasciata Loew 1 2 0 0 3 Tabanidae 0 1 0 1 2 Tachinidae 2 4 1 6 13 Tipulidae 2 0 0 0 2

Annexe D. Liste des arthropodes (N>1) échantillonnés sur les plants ou les fleurs d'épilobe (Epilobium angustifolium) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant trois densités de cerfs (0, 7.5, et 15 cerfs/km2) répété dans trois localités. Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Acari Actinedida sp. 8 In Pr 0 3 0 3 Oribatida sp.2 In Pr 0 2 0 2 Oribatida sp.5 In Pr 0 3 1 4 Tetranychidae Tetranychidae sp.1 In Ph 26 16 6 48 Aranaea Clubionidae Clubiona sp. I Pr 3 2 2 7 Linyphiidae Linyphiidae sp.2 I Pr 1 1 0 2 Linyphiidae sp.9 I Pr 2 1 5 8 Philodromidae Philodromidae sp.1 I Pr 2 2 0 4 Salticidae Salticidae sp.1 I Pr 2 3 0 5 Salticidae sp.2 I Pr 2 2 2 6 Theridiidae Theridula emertoni Levi A-I Pr 6 1 5 12 Thomisidae Misumenops sp.1 A-I Pr 7 6 1 14 ? Spider sp.11 I Pr 2 0 2 4 Coleoptera Cantharidae Malthodes parvulus (LeConte) A Pr 0 0 2 2 Latridiidae Corticaria sp. A S 0 1 1 2 Cortinicara gibbosa (Herbst) A S 2 0 0 2 Collembola Entomobryidae Entomobrya sp. A-I In 19 30 28 77 Diptera Anthomyiidae A Po 1 2 1 4 Calliphoridae A Po 1 8 0 9 Cecidomyiidae A-I In 151 111 102 364 Chironomidae A In 5 4 4 13 Muscidae Eudasyphora sp. A Po 9 11 5 25 Hydrotaea scambus (Zetterstedt) A Po 4 1 2 7 Morellia micans (Macquart) A Po 1 3 0 4 Muscidae sp. A In 2 0 2 4 Phaonia curvipes (Stein) A Po 3 2 0 5

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Thricops spiniger (Stein) A Po 6 10 13 29 Phoridae A S 2 0 3 5 Sarcophagidae A Po 1 1 0 2 Simuliidae A In 1 1 0 2 Syrphidae Melanostoma mellinum (Linné) A Po 1 1 0 2 Meliscaeva cinctella (Zetterstedt) A Po 0 3 0 3 Syrphidae sp.3 I Pr 0 4 0 4 Syrphidae sp.11 I Pr 0 1 1 2 Syrphus torvus Osten Sacken A Po 3 3 5 11 Tachinidae P Pa 1 0 1 2 Tachinidae A Po 4 3 1 8 Hemiptera Aphididae Aphis sp.1 A-I Ph 679 2739 711 4129 Aphididae sp.9 I Ph 3 4 2 9 Cercopidae Aphrophora sp. A Ph 1 0 1 2 Philaenus spumarius (Linné) A-I Ph 34 70 46 150 Cicadellidae Cicadellidae sp.4 I Ph 2 0 0 2 Cicadellidae sp.5 I Ph 0 2 0 2 Draeculacephala sp. A-I Ph 0 1 3 4 Idiocerus sp.2 A Ph 2 1 0 3 Scaphytopius sp. A-I Ph 8 7 1 16 Miridae Lygus borealis Kelton A-I Ph 4 7 5 16 Miridae sp.1 I Ph 4 0 0 4 Plagiognathus obscurus Uhler A Ph 1 2 4 7 Reduviidae Reduviidae sp.1 I Pr 4 0 0 4 Hymenoptera Apidae Bombus borealis Kirby A Po 20 5 5 30 Bombus frigidus Smith A Po 84 104 55 243 Bombus sandersoni Franklin A Po 1 1 3 5 Bombus ternarius Say A Po 8 15 3 26 Braconidae A Pa 2 0 0 2

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Colletidae Colletes impunctatus Nylander A Po 2 4 3 9 Encyrtidae A Pa 0 0 3 0 Eulophidae A-P Pa 8 8 6 22 Formicidae Formica hewetti Wheeler A Au 0 2 1 3 Halictidae Lasioglossum quebecensis A Po 1 1 2 4 Crawford Ichneumonidae A Po 0 3 1 4 Megachilidae Megachile frigida Smith A Po 7 3 1 11 Megachile inermis Provancher A Po 0 1 2 3 Megachile relativa Cresson A Po 2 0 8 10 Mymaridae A Pa 2 0 2 4 Platygastridae A Pa 5 3 13 21 Pteromalidae A Pa 0 1 2 3 Tenthredinidae Tenthredinidae sp.5 I Ph 2 4 3 9 Vespidae Dolichovespula arenaria A Po 6 7 0 13 (Fabricius) Dolichovespula norvegicoides A Po 27 17 6 50 (Sladen) Lepidoptera Gelechiidae Gelechiidae sp.2 I Ph 30 45 19 94 Geometridae Geometridae sp.4 I Ph 2 1 1 4 Geometridae sp.5 I Ph 3 0 0 3 Geometridae sp.6 I Ph 1 1 0 2 Noctuidae Noctuidae sp.4 I Ph 0 0 4 4 Pyralidae I Ph 3 2 1 6 Tortricidae I-P Ph 0 1 4 5 Xyloryctidae Scythris sp. I Ph 27 26 7 60 Neuroptera Hemerobiidae I Pr 2 0 0 2 Thysanoptera Thripidae Thripidae sp.2 A Ph 0 3 0 3 Thripidae sp.3 A Ph 3 2 0 5

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Thripidae sp.4 A Ph 0 0 2 2 a A=Adulte, I=Immature, P=Pupe, In=Inconnu/Incertain. b Au=Autre, Pa=Parasitoïde, Ph=Phytophage, Po=Pollinisateur, Pr=Prédateur, S=Saprophage, In=Inconnu/Incertain.

Annexe E. Liste des arthropodes (N>1) échantillonnés sur les plants de framboisier (Rubus idaeus) avec leur stade de développement à la capture, leur rôle trophique et leur abondance cumulée dans un dispositif de broutement contrôlé du cerf de Virginie sur l'île d'Anticosti comprenant trois densités de cerfs (0, 7.5, et 15 cerfs/km2) répété dans trois localités. Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Acari Actinedida sp.2 In Pr 1 2 2 5 Tetranychidae Tetranychidae sp.1 In Ph 4 8 6 18 Araneae Araneidae Araneidae sp.2 I Pr 11 9 4 24 Araneidae sp.4 I Pr 2 4 1 7 Araneidae sp.5 I Pr 4 2 22 28 Araneidae sp.6 I Pr 3 0 0 3 Clubionidae Clubiona sp. A-I Pr 25 33 38 96 Linyphiidae Linyphiidae sp.1 I Pr 0 0 2 2 Linyphiidae sp.2 A-I Pr 1 1 1 3 Linyphiidae sp.4 A Pr 4 0 0 4 Linyphiidae sp.7 A-I Pr 1 2 0 3 Philodromidae Philodromidae sp.1 I Pr 2 1 0 3 Salticidae Salticidae sp.1 I Pr 2 1 1 4 Salticidae sp.2 A-I Pr 1 19 2 22 Salticidae sp.4 I Pr 0 0 2 2 Theridiidae Theridula emertoni Levi A-I Pr 12 8 3 23 Thomisidae Misumenops sp.1 A-I Pr 18 6 6 30 Coleoptera Cantharidae Malthodes fragilis (LeConte) A Pr 2 1 5 8 Malthodes fuliginosus LeConte A Pr 0 0 2 2 Malthodes parvulus (LeConte) A Pr 2 1 2 5 Latridiidae Cortinicara gibbosa (Herbst) A S 1 1 0 2 Collembola Entomobryidae Entomobrya sp.1 A-I In 35 25 39 99 Sminthuridae A-I In 0 0 5 5 Diptera Agromyzidae A Ph 5 0 0 5 Cecidomyiidae A-I In 140 153 185 478 Chironomidae A In 3 1 3 7

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Heleomyzidae A In 3 1 0 4 Syrphidae Syrphidae sp.1 I Pr 1 0 2 3 Syrphidae sp.2 I Pr 4 0 0 4 Syrphidae sp.4 I Pr 2 1 0 3 Tachinidae I Pa 0 7 0 7 Hemiptera Aphididae Amphorophora agathonica Hottes A-I Ph 354 189 293 836 Aphididae sp.3 A Ph 2 1 2 5 Cercopidae Aphrophora quadrinotata Say A Ph 0 0 2 2 Philaenus spumarius (Linné) A Ph 18 14 23 55 Cicadellidae Latalus sp. A Ph 1 0 1 2 Scaphytopius sp. A-I Ph 16 16 20 52 Cixiidae A Ph 0 0 2 2 Miridae Miridae sp.1 I Ph 16 12 2 30 Miridae sp.3 I Ph 12 5 12 29 Miridae sp.4 I Ph 0 2 1 3 Plagiognathus obscurus Uhler A-I Ph 14 8 14 36 Psyllidae A-I Ph 0 5 36 41 Reduviidae Reduviidae sp.1 I Pr 7 5 2 14 Hymenoptera Braconidae A-P-I Pa 1 5 2 8 Ceraphronidae A Pa 0 3 2 5 Cynipidae Diastrophus rubi (Bouché) A-I Ph 10 21 31 62 Dryinidae A Pa 1 0 2 3 Eulophidae A Pa 3 2 6 11 Formicidae Formica hewetti Wheeler A Au 2 0 0 2 Leptothorax sp. A Au 2 0 2 4 Ichneumonidae P Pa 5 1 0 6 Mymaridae A Pa 1 1 0 2 Platygastridae A Pa 1 1 4 5 Pteromalidae A Pa 4 2 4 10

Densité Ordre Famille Taxon Stadea Rôleb 0 7.5 15 Total Tenthredinidae Tenthredinidae sp.1 I Ph 3 3 2 8 Tenthredinidae sp.2 I Ph 1 4 3 8 Torymidae A-P Pa 5 5 5 15 Lepidoptera Geometridae Geometridae sp.1 I Ph 2 3 1 6 Noctuidae Noctuidae sp.2 I Ph 1 1 2 4 Pterophoridae Pterophoridae sp.1 I Ph 2 3 0 5 Tortricidae I Ph 0 0 3 3 Pseudo- Cheliferidae A Pr 1 0 1 2 scorpionida Thysanoptera Thripidae Thripidaes sp.1 I Ph 1 1 0 2 a A=Adulte, I=Immature, P=Pupe, In=Inconnu/Incertain. b Au=Autre, Pa=Parasitoïde, Ph=Phytophage, Po=Pollinisateur, Pr=Prédateur, S=Saprophage, In=Inconnu/Incertain.