The Trophic Ecology of Wolves (Canis Lupus) and Their Predatory Role in the Ungulate Community of Two Mountainous Areas of Italy

Department of Zoology and Biological Anthropology University of Sassari

Degree of Philosophy in Environmental Biology

The trophic ecology of wolves (Canis lupus) and their predatory role in the ungulate community of two mountainous areas of Italy

Candidate: Dr. Andrea Gazzola

Supervisor: Prof. Marco Apollonio

To Delio and Ilva, my grandparents.

Ciò che apprendo, trasmetto. - Confucio -

All that I learn, I teach. - Confucio -

TABLE OF CONTENTS

Summary [in Italian] ……………………………………………………… 1

Papers list ……………………………………………………… 6

Abstract ………………………………………………………. 7

Introduction ……………………………………………………….. 10

Papers I. ……………………………………………….. 16 II. ……………………………………………….. 26 III. ……………………………………………….. 35 IV. ……………………………………………….. 63 V. ……………………………………………….. 89 VI. ……………………………………………….. 102

Conclusions ……………………………………………………….. 136

Acknowledgments ………………………………………………………… 139

SUMMARY Questa tesi si basa sui seguenti sei articoli:

I. Gazzola A., Bertelli I., Avanzinelli E., Tolosano A., Bertotto P., Apollonio M. (2005). Predation by wolves (Canis lupus) on wild and domestic ungulates of the western Alps, Italy. Journal of Zoology (London), 266: 205-213.

L’interazione preda-predatore è stata studiata in Alta Val di Susa (settore delle Alpi occidentali, provincia di Torino, Italia) dal dicembre 1999 fino al novembre 2002. L’analisi della dieta del lupo ha evidenziato un marcato utilizzo degli ungulati selvatici (% occorrenza=87), in particolar modo cervo (Cervus elaphus) e capriolo (Capreolus capreolus) (% occorrenza cumulati = 74). La preferenza verso queste due prede è particolarmente accentuata in inverno (% occorrenza cumulati = 87), mentre tende ad attenuarsi durante il periodo estivo (% occorrenza cumulati = 54.3). Il camoscio, terza specie in ordine d’importanza, è utilizzato in maniera decisamente inferiore rispetto all’elevata consistenza numerica. La presenza cospicua dei domestici in estate (maggio-ottobre) ha influenzato le abitudini alimentari dei lupi (presenza di domestico nella dieta del lupo: estate 19%; inverno 0.3%). Nel presente studio è stata evidenziata una forte relazione trofica tra lupo e cervo. Durante i tre inverni d’indagine sono state ritrovate 177 carcasse di ungulato selvatico. Escludendo la caccia, il più importante fattore di mortalità del cervo era la predazione da lupo (51%), seguita dalla malattia e denutrizione (36%) e dagli incidenti stradali e ferroviari (12%). La maggior parte dei cervi predati erano femmine adulte (59%) e giovani (29%), mentre i maschi adulti rappresentavano solo il 12%. Per contro, nel capriolo, la predazione era il fattore di mortalità di minor peso (11%), mentre gli incidenti stradali e ferroviari quello di maggiore rilevanza (78%). La preferenza alimentare del lupo sul cervo sembra essere influenzata in parte dalla stretta relazione spaziale riscontrata tra le due specie e dalla particolare attitudine del cervo a formare aggregazioni di dimensioni cospicue.

1 II. Gazzola A., Ferroglio E., Avanzinelli E., Rosso M., Kurscinski F., Apollonio M. The physical condition of roe and red deer killed by wolves in a region of the western Alps, Italy. (manuscript).

La percentuale di grasso contenuta nel midollo delle ossa lunghe delle zampe posteriori di 11 caprioli (Capreolus capreolus) predati dal lupo (Canis lupus) è stata analizzata e comparata con quella di 15 caprioli investiti da autoveicoli durante l’inverno 2003-2004. Gli stessi parametri sono stati analizzati in 14 cervi predati. Il 45 % dei caprioli uccisi dal predatore avevano un elevata percentuale di grasso midollare (75-100%). La percentuale di grasso diminuiva con il progredire della stagione invernale, ma i valori dei caprioli predati non differivano significativamente da quelli degli animali incidentati. In generale è stato riscontrato un carente stato nutrizionale dei cervi predati. Questi risultati sembrano suggerire una predazione a carico di animali in cattive condizioni fisiche nel cervo ma non nel capriolo.

III. Gazzola A., Avanzinelli E., Bertelli I., Tolosano A., Bertotto P., Musso R., Apollonio M. Wolf impact on a multi-species ungulate community of the western Alps, Italy. (manuscript).

L’impatto della predazione di due branchi di lupo sulla popolazione di prede è stata studiata in Alta Val Susa (settore delle Alpi occidentali, provincia di Torino, Italia). Durante il periodo d’indagine (maggio 2000-aprile 2003) il numero di lupi variava da minimo di 7 ad un massimo di 15, mentre le densità da 1.7 a 2.9 lupi/ 100 km2. Le principali specie preda erano il cervo (Cervus elaphus) e il capriolo (Capreolus capreolus). Il camoscio (Rupicapra rupicapra) era stato utilizzato in misura inferiore alla sua disponibilità. Per valutare l’impatto dei lupi è stata impiegata la formula di Nagy (FMR) che consente di calcolare, in base al peso corporeo di un lupo, il relativo consumo giornaliero medio di carne. Dal 2000 al 2002, i lupi annualmente prelevavano 20-34 cervi, 21-58 caprioli, e 7-14 camosci ogni 100 km2. L’impatto dei lupi equivaleva al 19-51% della mortalità annuale del cervo, al 6- 28% del capriolo e al 6-9% del camoscio. Il prelievo venatorio era responsabile del 58-94% della mortalità annuale del cervo, del 18-29% del capriolo e del 22-43 % del camoscio. Gli

2 altri fattori di mortalità (incidenti stradali, bracconaggio, malattia e denutrizione) risultavano trascurabili sia per il cervo sia per il camoscio, ma non per il capriolo (28-54% della mortalità annuale). La predazione da parte del lupo non sembrava influenzare la dinamica di popolazione degli ungulati selvatici.

IV. Gazzola A., Capitani C., Mattioli L., Avanzinelli E., Apollonio M. Livestock-damage and wolf-human conflict in the north-eastern Apennines, Tuscany, Italy. (manuscript).

Il conflitto lupo-uomo viene studiato in provincia di Arezzo analizzando i dati dei danni arrecati alla zootecnia da parte di canidi dal 1998 al 2001, e confrontando questi con la distribuzione del lupo su scala provinciale. Il lupo era presente in 22 dei 39 comuni della 2 provincia, con una densità stimata attorno 2.9 ± 0.7 lupi/100 km . Durante il periodo d’indagine la popolazione di lupo era stabile, sebbene venisse rilevata una elevata mortalità. L’area dove è stata accertata la presenza del carnivoro era caratterizzata da una maggiore copertura boscata e densità di ungulati selvatici, e da una minore densità di popolazione e di rete viaria rispetto all’area dove la specie non è stata monitorata. Il bestiame domestico era distribuito uniformemente in tutta la provincia. Per contro i danni al patrimonio zootecnico erano localizzati soprattutto nell’area di presenza del lupo. Solamente il 6% delle aziende era interessato ripetutamente dal fenomeno predatorio. Tali aziende riportavano il 38% degli attacchi e il 37% delle perdite di tutta la provincia. Il fenomeno delle uccisioni multiple coinvolgeva solamente il comparto ovi-caprino. Durante i quattro anni d’indagine, 35 attacchi causavano 536 perdite, rispettivamente il 14% e il 44% degli episodi registrati sull’intera provincia di Arezzo. Nel periodo di studio erano stati versati annualmente una media di 86863 Euro (€ 68805- € 99318) di indennizzi. Nessuna richiesta di finanziamenti, predisposti dalla Regione Toscana per la messa in opera di misure di prevenzione, erano state richieste dagli allevatori colpiti durante l’intero periodo d’indagine. Dai risultati sembra evidente che la coesistenza uomo-lupo può essere raggiunta grazie all’attenuazione dei conflitti con le attività antropiche e alla conservazione di aree idonee alla presenza del lupo.

3 V. Apollonio M., Mattioli L., Scandura M., Mauri L., Gazzola A., Avanzinelli E. (2004). Wolves in the Casentinesi Forests: insight for wolf conservation in Italy from a protected area with a rich wild prey community. Biological Conservation. 120/2: 249-260.

Le Foreste Casentinesi, localizzate in un settore dell’Appennino settentrionale, sono caratterizzate da una ricca e diversificata comunità di ungulati selvatici e dalla presenza del lupo. Tra il 1993 e il 2000, la distribuzione dei branchi era costantemente monitorata. La consistenza numerica dei lupi era monitorata attraverso la tecnica dell’ululato indotto (wolf- howling), della tracciatura delle piste d’impronte su substrato nevoso (snow-tracking), e attraverso gli avvistamenti. La presenza di 3-5 branchi è stata accertata annualmente nel periodo di studio. La dimensione media dei branchi era di 4.2 ± 0.9 lupi (massimo 7). La densità media di lupi era di 4.7 lupi/100km2 , mentre la distanza media tra branchi adiacenti era di 11.1 km. L’elevata densità di lupi riscontrata nell’area è determinata dalla ricchezza e abbondanza delle prede selvatiche. Una maggiore fedeltà ai siti di riproduzione e un più elevato successo riproduttivo era osservato nei branchi residenti totalmente in aree protette rispetto a quelli che occupavano aree marginali in parte soggette ad attività venatoria. La conservazione del lupo in Italia sembra legata alla creazione di una rete di aree protette caratterizzate dalla presenza di una ricca e diversificata comunità di ungulati.

VI. Mattioli L., Apollonio M., Gazzola A., Avanzinelli E., Bertelli I., Capitani C. Prey selection by wolves from a multi-species ungulate community in the Casentinesi Forests, Italy. (manuscript).

La selezione delle specie preda e delle classi di peso e d’età degli animali consumati da parte dei lupi è stata analizzata dal maggio 1988 all’aprile 2000. La densità media degli ungulati selvatici era di 34 capi/ km2 (924 kg/ km2). La dieta del lupo era incentrata sul cinghiale (Sus scrofa) e in misura minore sul capriolo (Capreolus capreolus). Il cinghiale risultava la specie preda preferita dai lupi, mentre il capriolo era selezionato negativamente.

4 La preferenza verso gli ungulati di età inferiore all’anno era riscontrata per tutte le specie preda. Nel cinghiale, i piccoli dell’anno erano negativamente selezionati nel primo mese di vita (marzo-aprile), ma da luglio venivano preferiti. Il peso medio delle prede consumate dal lupo era di 20.5 kg. Le abitudini alimentari dei lupi e la selezione di una particolare specie- preda sembrano determinate più dalla vulnerabilità e dalla taglia della preda che dalla abbondanza relativa.

5 This thesis is based on the following six papers, which will be referred to in the text by their Roman numerals

II. Gazzola A., Ferroglio E., Avanzinelli E., Rosso M., Kurscinski F., Apollonio M. The physical condition of roe and red deer killed by wolves in a region of the western Alps, Italy. (manuscript).

III. Gazzola A., Avanzinelli E., Bertelli I., Tolosano A., Bertotto P., Musso R., Apollonio M. Wolf impact on a multi-species ungulate community of the western Alps, Italy. (manuscript).

IV. Gazzola A., Capitani C., Mattioli L., Avanzinelli E., Apollonio M. Livestock-damage and wolf-human conflict in the north-eastern Apennines, Tuscany, Italy. (manuscript).

V. Apollonio M., Mattioli L., Scandura M., Mauri L., Gazzola A., Avanzinelli E. (2004). Wolves in the Casentinesi Forests: insight for wolf conservation in Italy from a protected area with a rich wild prey community. Biological Conservation. 120/2: 249-260.

VI. Mattioli L., Apollonio M., Gazzola A., Avanzinelli E., Bertelli I., Capitani C. Prey selection by wolves from a multi-species ungulate community in the Casentinesi Forests, Italy. (manuscript).

Papers I, V are reproduced with permission from the publisher.

ABSTRACT

This thesis examines the trophy ecology of wolves (Canis lupus) and their role in the wild ungulates community. Moreover livestock-wolf conflict were analysed and some suggestions on management issues for the conservation of wolf were discussed. The thesis contains results based on long term studies conducted in two mountainous areas of Italy. One study area is located in western Alps (Alta Valle di Susa, Turin Province, Piemonte Region). Here the wolf, was exterminated in last decade of 19th century and has recently (1994) reappeared. Domestic ungulates are present only during summer. The second study area is situated in central Italy, Northern eastern Apennines (Arezzo Province, Tuscany). Here, wolf never become extinct. Domestic ungulates are always in the pastures. Both areas are characterized by a rich and various wild ungulate community. The Alpine area is characterized by the presence of six wild ungulates: chamois (Rupicapra rupicapra), roe deer (Capreolus capreolus), red deer (Cervus elaphus), wild boar (Sus scrofa), ibex (Capra ibex) and mouflon (Ovis orientalis musimon). The wild ungulate community of Apennine area is composed of four species: wild boar, roe deer, red deer and fallow deer (Dama dama). The wolf (Canis lupus) is an opportunistic predator with a highly diversified diet. Thus, in areas characterized by poor ecological conditions, wolf populations adapt to a diversity of food resources, such as livestock, fruit, and small mammals. However, where wild ungulates are abundant, wolves feed chiefly on them. On the basis of wolf diet, in the western Alps, the most important prey of wolves was wild ungulates (87%). Cervids were the preferred prey (74%) and constituted predominant food items both in winter (84%) and summer (54%). In particular red deer was the staple food item followed by roe deer. The larger use of red deer was found especially during winter. Chamois was accessory categories during winter (10%) but its importance grew during summer (16%). On the contrary, in the northern-central Apennines, wild boar was found to be the wolf main prey. The scats analysis revealed that wild boar was the main prey, followed by roe deer, red deer, and fallow deer, with a Mean Percent Volume of 59%, 19%, 9%, and 2%, respectively.

7 With respect to their availability in the ungulate community, obtained from census data, wolves selected wild boar positively and roe deer and red deer negatively. The wolf tendency to use and select certain prey species can be influenced by several factors. At a macro scale prey community composition and human related factors (such as husbandry practices) play a decisive role. Within similar ecological contexts, prey selection can be influenced by particular local factors, such as topography, climatic conditions, relative density and structure of prey populations, spatial distribution patterns and use of space of preys and predator, social structure of prey species. In this thesis some of these factors are analysed. In general some quite clear patterns were clear i.e. prey vulnerability was influenced by physical conditions, habitat selection, aggregation pattern and age class. It was showed as red deer on the Alps that suffered of heavier under nutrition than roe deer was most preyed upon, besides this factor also a more close overlap of habitat use and a larger group size contributed to the strong selection on this prey in the Alps. Wild boar played the same role on the Apennines and this was connected to his larger group size and larger percentage of young, vulnerable animal within the population. In fact young ungulates (i.e. < 1 year) proved to be an important fraction of the wolf diet in both study areas. In Alta Val Susa, in red deer, 52% of all cases were young: this percentage was 45% in roe deer, during winter. These proportions of fawns increased in summer when they represented for red deer and roe deer 75%, 68%, respectively. In Apennine study area, young of wild boar were 89%, while in roe deer 63%.Patterns of use and selection of prey species and age classes suggest that wolf food habits are determined more by vulnerability, particularly size, than by abundance.

Our study seem to line up to other cases in which prey populations were not influenced by the presence of wolves. Wild boar and red deer were the prey species respectively more used in the two study areas by wolf, but wolf predation alone was a poor predictor of population ungulate dynamics. The larger use of livestock by wolves in the Alpine area than Apennine area is probably due to larger flocks (500-1000 sheep). Moreover, the lack of measures to prevent wolf attacks, because of the loss of any remaining tradition of coexistence with the wolf by stockmen, must be added as an explanation of these differences.

8 Wolf conservation both on the Alps and on the Apennine was clearly linked to the presence of a net of protected areas, to rich ungulate communities and to a proper husbandry practice that could help to minimize conflicts with humans

Key words: Canis lupus, conservation, wolf predation, livestock damage, population censusing.

9 INTRODUCTION

The conservation of wolf (Canis lupus) natural populations represents a priority in several European countries, where the species is endangered or was, in the recent past, severely threatened (Promberger and Schröder, 1993). The Italian wolf population suffered severe persecution till 1971 when wolf hunting was stopped and poison baits banned. This change of attitude was confirmed in 1976, when a fully protected status was given to the species. This process was stimulated by WWF International, that funded a long-term project including public educational campaigns, scientific works and management solutions in order to protect wolves. Wolf management at a national scale should not fail to take into account the knowledge of mechanisms regulating wolf population dynamics, and interactions with prey communities. Such aspects have been well studied in North America, but remain mostly unclear in several ecological contexts. In the papers III, and V information on wolf population dynamic over two Italian mountains areas are given. These are the Susa Valley (III) located in western Alps (Turin Province, Piemonte Region) where wolves have recovered during the last decade and the mountainous area of Arezzo Province, Tuscany (V) where wolf never disappeared. The wolf tendency to select certain prey species is often discussed in literature (Poulle et al., 1997, Jedrzejewski et al., 1992 and 2000, Filonov, 1989, Olsson et al., 1997, Meriggi et al., 1996, Huggard, 1993a, Mech et al., 1995, Okarma, 1995). Nevertheless, selective tendencies appear very variable at different scale levels. At a macro-level, they are related to different habitats availability and to specific prey vulnerability. At a micro-level, within similar ecological contexts, prey selection can be influenced by peculiar factors, such as topography, climate (Okarma et al., 1995, Jedrzejewski et al., 1992 and 2002, Mech and Frenzel., 1971, Peterson, 1977, Carbyn, 1983), relative density and populations structure of prey, spatial distribution patterns and use of space of preys-predator. In the papers I, and VI data on wolf food habits are presented. Particularly the selection of prey by wolves and the different impact of predation on the sex and age classes of the prey species. Finally, the preference versus certain prey species by wolves was analysed.

10 The nutritional status of preys influenced the vulnerability and determined the selection by wolves towards particular prey. Several studies reported that ungulates in poor condition were more often taken by wolves (Seal et al., 1978; Kunkel and Mech, 1994; Mech et al., 2001). The wolf is believed to catch proportionately more sick and weak animals than occur in the population, and this phenomenon is more marked in species that are usually hard to capture (Mech, 1970, Temple, 1987). Numerous researches on the physical condition of wolf kills were conducted in North America and Eastern Europe (Huggard, 1993b, Husseman et al., 2003, Okarma, 1984, Okarma, 1991) but no one in Western Europe in Italy.

In the paper II the body condition of roe and red deer killed by wolves in Alpine area is evaluated. The analysis was focused on Cervids, because they constituted the predominant food items of wolves during the winter season.

Mech and Boitani (2003) state that “Wolves form a major force in the ecosystems of which they are a part”. Wolf represent probably the most important predator on large mammals. Nevertheless, the role of wolf and his effect on prey populations dynamics has been subject of scientific debate for a long period.

The influence by wolves on ungulate populations is very difficult to evaluate especially in complex European biocenoses. Moreover, critical features of prey dynamics differ between prey population that are subject to natural mortality factors only (e.g. predation, starvation) and those that are influenced by humans impact (harvest by hunters, poaching, traffic accidents). The knowledge of the impact of predation by wolves on wild ungulate would assist wildlife managers in ensuring sustainable ungulate harvests after wolf recolonization and it is important to determine this role in the context of conservation and rational management of living natural resources (Głowaciński and Profus, 1997).

In the papers III and VI are shown prey populations dynamic, in areas where the predators is recently reappeared (Piemonte Region), and where never disappeared (Tuscany Region) respectively. In the paper III is evaluated the impact of wolf predation on wild ungulate populations, over 3 years, in the Alpine area. Moreover, the mortality induced by wolves is compared with other causes of ungulate mortality (harvests by hunters and other natural/human mortality aspects).

11 The maintenance of a viable wolf population represents a priority in most Western European countries. Land use and trophic resources availability are reported to play a leading role, but also human disturbance-associated variables seem to be important. Indeed, wolf conservation efforts can prove insufficient, when the predator has a negative impact on the economy of local communities. The high-level persistence of the human-wolf conflict is mainly due to livestock predation caused by wolves in most European countries (Zimen, 1978), and has been the motivation for eradicating the wolf from most of its former range (Boitani, 1995). Given the importance of this factor for wolf conservation a specific paper (IV) was devoted to: i) to give detailed information on livestock damage linked to wolf presence, ii) to characterize management tools that could reduce human-wolf conflict, with particular reference to damage compensation system, and iii) to illustrate how proper small-scale management can play a role in a large-scale wolf conservation policy.

12 REFERENCE

Boitani, L., 1995. Ecological and cultural diversities in the evolution of wolf-human relationships. In: Carbyn, L.N., Fritts, H.S., Seip, D.R. (Eds), Ecology and conservation of wolves in changing world: proceedings of the second North American symposium on wolves. Canadian Circumpolar Institute, Occasional Publication Number 35, pp. 3-11. Carbyn, L.N. 1983. Wolf predation on elk in Riding Mountain National Park, Manitoba. The Journal of Wildlife Management 47: 963-976. Filonov, K.P. 1989. [Ungulates and large predators in wildlife predators]. - Izdatelstvo Nauka. [In Russian] Głowaciński, Z., Profus, P. 1997. Potential impact of wolves (Canis lupus) on prey populations in eastern Poland. Biological Conservation 80: 99-106. Huggard, D.J. 1993a. Prey selectivity of wolves in Banff National Park. I. Prey species. Canadian Journal of Zoology 71: 130-139. Huggard, D.J. 1993b. Prey selectivity of wolves in Banff National Park. II. Age, sex, and condition of elk. Canadian Journal of Zoology 71: 140-147. Husseman J.S., Murray D.L., Power G., Mack C., Wenger C.R., Quiley H. 2003. Assessing differential prey selection patterns between two sympatric large carnivores. Oikos 101: 591-601. Jedrzejewski, W., Jedrzejewska, B., Okarma, H., Ruprecht, A.L. 1992. Wolf predation and snow cover as mortality factors in the ungulate community of the Bialowieza National Park, Poland. Oecologia 90: 27-36. Jedrzejewski, W., Jedrzejewska, B., Okarma, H., Schmidt, K., Zub, C., Musiani, M. 2000. Prey selection and predation by wolves in BPF, Poland. Journal of Mammalogy 81:197- 212. Jedrzejewski, W., Schmidt, K., Theuerkaf, J., Jedrzejewska, B., Selva, N., Zub, K., Szymura, L. 2002. Kill rates and predation by wolves on ungulate populations in Bialowieza Primeval Forest (Poland). Ecology 83: 1341-1356. Mech L.D. 1970. The wolf: ecology and behaviour of an endangered species. The Natural History Press, Garden City, N. Y.

13 Mech, D.L., Frenzel, L. D. Jr. 1971. An analysis of the age, sex and condition of deer killed by wolves in northeastern Minnesota. USDA Forest Service Research Paper NC-52: 35-40. Mech, L.D., Meier, T.J., Burch, J.W., Adams, L. 1995. Patterns of Prey Selection by Wolves in Denali National Park, Alaska. - In: Carbyn, L. D., Fritts, S. H. and Seip, D. R. (Eds.), Ecology and Conservation of Wolves in a Changing World. Proceedings of the Second North American Symposium on Wolves, Canadian Circumpolar Institute, 35: 231-244. Mech, L.D., Smith, D.W., Murphy, K.M., MacNulty, D.R. 2001. Winter severity and wolf predation on formerly wolf free elk herd. The Journal of Wildlife Management 64: 998- 1003. Mech, L.D., Boitani L. (Eds.), 2003. Wolves. Behavior, Ecology, and Conservation. The University of Chicago Press, Ltd., London. Meriggi, A., Brangi, A., Matteucci, C., Sacchi, O. 1996. The feeding habits of wolves in relation to large prey availability in northern Italy. Ecography 19: 287-295. Kunkel, K.E., Mech, L.D. 1994. Wolf and bear predation on white tailed deer fawns in northeastern Minnesota. Canadian Journal of Zoology 72: 1557-1565. Okarma, H. 1984. The physical condition of red deer falling prey to the wolf and lynx and harvested in the Carpathian mountains. Acta Theriologica 29, 23: 283-290. Okarma H. 1991. Marrow fat content, sex and age of red deer killed by wolves in winter in the Carpathian mountains. Holoarct, Ecol. 14: 169-172. Okarma, H. 1995. The trophic ecology of wolves and their role in ungulate communities of forest ecosystems in Europe. Acta Theriologica 40: 335-386. Okarma, H., Jedrzejewski, B., Jedrzejewska, W., Krasinski, Z., Milkowski, L. 1995. The role of predation, snow cover, acorn crop, and man-related factors on ungulate mortality in B. P. F., Poland. Acta Theriologica 40: 197-217. Olsson, O., Wirtberg, J., Andersson, M., Wirtberg, I. 1997. Wolf (Canis lupus) predation on moose (Alces alces) and roe deer (Capreolus capreolus) in south central Scandinavia. - Wildlife Biology 3: 13-23. Peterson, R.O. 1977. Wolf ecology and prey relationship on Isle Royale. - U. S. Natl. Park Serv. Sci. Monogr. Ser. 11.

14 Poulle, M.-L., Carles, L., Lequette, B. 1997. Significance of ungulates in the diet of recently settled wolves in the Mercantour mountains (southern France). Revue Ecologie (Terre vie) 52: 357-368. Promberger, C., Schröder, W. (Eds.), 1993. Wolves in Europe: Status and perspectives. Munich Wildlife Society, Ettal, Germany. Seal, U.S., Nelson, M.E., Mech, L.D., Hoskinson, R.L. 1978. Metabolic indicators of habitat differences in four Minnesota deer populations. The Journal of Wildlife Management 42: 746-754. Temple, S.A. 1987. Do predators always catch substandard individuals disproportionately from prey populations? Ecology 68: 669-674. Zimen, E., 1978. Der Wolf: Mythos und Verhalten. Meyster Verlag GmbH, Wien, München.

Predation by wolves (Canis lupus) on wild and domestic ungulates of the western Alps, Italy.

- Journal of Zoology (London) -

Gazzola A., Bertelli I., Avanzinelli E., Tolosano A., Bertotto P., Apollonio M.

The physical condition of roe and red deer killed by wolves in a region of the western Alps, Italy.

- Manuscript -

Gazzola A., Ferroglio E., Avanzinelli E., Rosso M., Kurschinski F., Apollonio M.

26 THE PHYSICAL CONDITION OF ROE AND RED DEER KILLED BY WOLVES IN A REGION OF THE WESTERN ALPS, ITALY

Gazzola A.*, Ferroglio E. **, Avanzinelli E *, Rosso M. § Kurschinski F. ° and Apollonio M.*

* Department of Zoology and Biological Anthropology, University of Sassari, Via Muroni 25, I-07100, Sassari, Italy e-mail: [email protected] (AG, EA, MA); ** Department of Animal Production, Epidemiology and Ecology, University of Torino, Via L. da Vinci 44, I-10095, Grugliasco, Torino, Italy (EF); § Gran Bosco di Salbertrand Natural Park, Via Monginevro 7, I-10050, Salbertrand, Torino, Italy (MR). ° Consorzio Forestale Alta Valle di Susa, Oulx, Torino, Italy (FK).

ABSTRACT Legbone marrow fat of 11 roe deer (Capreolus capreolus) killed by wolves (Canis lupus) were examined and compared with 15 deer killed by traffic accidents in winter (November 2003- April 2004). Moreover, a sample of 14 red deer kills was examined for the same parameter. Among individuals killed by wolf, 45% of roe deer had a high level of femur marrow fat (75- 100%). Marrow fat of leg bone showed a decrease through the winter season, but roe deer preyed by wolves did not show a significantly lower marrow fat level than those killed by cars or trains. Red deer had a generalized poor condition even if no comparison with the average condition of the population was possible.

Key words: Canis lupus, Capreolus capreolus, marrow fat, body condition.

27 INTRODUCTION The wolf is believed to catch proportionately more sick and weak animals than occur in the population, and this phenomenon is more marked in species that are hard to capture (Mech, 1970; Temple 1987). Several works have used leg bone marrow fat as an indicator of ungulate nutritional status (Ratcliffe, 1980; Fong, 1981; Fuller et al. 1986). In fact, numerous researches (Huot and Goudreault, 1985; Watkins et al. 1991; Holand, 1992) found that ungulate marrow fat value is a direct indicator of total body fat. Moreover, when the marrow fat is lower than 70-87% (depending on species), physical conditions are believed to be poor as most body fat has already been lost. Del Giudice et al. (1990) found that loss of fat stores involved also loss of protein, or muscle mass in ungulate species. Thus, marrow fat percentage should be viewed as an indicator of fat, muscle and energy depletion, and any level below a threshold of 70-85% c.a. indicates generally poor conditions (Mech et al. 1995). Numerous researches on the physical condition of wolf kills were conducted in North America and eastern Europe (Huggard, 1993; Husseman et al., 2003; Okarma, 1984; Okarma, 1991) but none in western Europe nor in Italy. The aim of the present study was to evaluate body condition of roe and red deer preyed up on by wolves and, in the case of roe deer, to compare it to the condition of traffic-caused deaths. This was done in a mountain area where these two species constituted the predominant food items in the winter season (Gazzola et al. 2005).

STUDY AREA, MATERIAL AND METHODS The study area was located in an Alpine region in the western part of Turin province (45°05’N, 7°E) (Fig. 1). It extends for 475 km2 from 800 to 2800 m a.s.l. The landscape at lower altitude is a mixture of mixed woods of beech (Fagus silvatica), maple (Acer platanoides) and birch (Betula pubescens) and fields, while from 1100 to 2300 m a.s.l., coniferous forests of pine (Pinus sylvestris), spruce (Picea abies), and larch (Larix decidua) are dominant. Over 2300 m, Alpine meadows and rocky areas replace the forest. The area is

28 characterized by the presence of six wild ungulates: chamois (Rupicapra rupicapra), roe deer (Capreolus capreolus), red deer (Cervus elaphus), wild boar (Sus scrofa), ibex (Capra ibex), and mouflon (Ovis orientalis musimon). Data on wolf kills were collected in the study area from November 2003 to April 2004. Snow- tracking surveys were made all over the study area after 1-2 days of snowfalls. This technique permits investigators to follow wolf travel routes and search for wolf-kills. Wolf signs on the snow (presence of blood, signs of a struggle) and carcass autopsy allowed us to recognize whether wolves had killed a living prey or merely scavenged a carcass. The control sample was constituted by ungulates hit by cars or trains. The bones of one hind leg were removed from each carcass. Marrow samples were taken from the middle part of the femur, metatarsus and tibia. Weighed samples were placed in Petri dishes and oven-dried at 70 °C to constant weight, and their dry weight was expressed as a percentage of fresh weight (Neiland, 1970). The Mann-Whitney test was used to estimate differences of marrow fat content among leg bones of wolf kills and road and rail victims. Relationships between leg bone marrow fat were calculated using Spearman’s correlation coefficient.

RESULTS AND DISCUSSION A sample of 11 roe deer killed by wolves and 15 hit by cars or trains was analysed. A wide variation in amount of marrow fat in all the leg bone was found. Percentage of marrow fat in roe deer killed by wolves varied in femur from 20 to 96%, in tibia from 14 to 93%, and in metatarsus from 29 to 94%. Percentage of marrow fat in roe deer hit by cars or trains varied in femur from 22 to 99%, in tibia from 31 to 99%, and in metatarsus from 19 to 99%. Considering all the leg bones analysed, marrow fat content in femur was significantly correlated with that in tibia and metatarsus, and metatarsus with tibia (Spearman’s correlation: femur-tibia r = 0.828 p<0.01; femur-metatarsus: r = 0.784; p<0.01; tibia-metatarsus: r = 0.852; p<0.01). The percentage of femur marrow fat in roe deer killed by wolves was no different from that in road and rail victims (Mann-Whitney test U = 42; p = 0.128) (Fig. 1).

29 Average percentage of marrow fat in leg bone decreased from femur to metatarsus both in roe deer killed by wolves and those hit by cars or trains (Table 1). Moreover marrow fat of leg bone shown a decrease during the cold season (Table 1). Difference of femur fat between early winter (November-January) and late winter (February-April) was significant (Mann-Whitney test 1-tailed: wolf kill U = 2; p = 0.035; traffic accident U = 7; p = 0.033). In both periods the percentage of femur marrow fat in roe deer killed by wolves and by cars and trains was similar (Mann-Whitney test: start winter U = 15; p = 1.000; late winter U = 8; p = 0.394).

Figure. 1 Distribution of percent femoral marrow fat of roe deer killed by wolves and in traffic accident. The Box-Plot represents the inter-quartile range, which contains 50% of values. The whiskers represent the highest and lowest value containing 80% of values. The circle represents outlier value (O), and the star the extreme value.

100

40 20 % marrow fat

22 20

wolf kill traffic accident

30 Table 1. Marrow fat content (%) in roe deer killed by wolves and in traffic accident in Susa Valley during winter.

Wolf Control Winter Legbone N. Mean SD N. Mean SD Femur 3 81 18 10 83 12 Nov-Jan Tibia 5 86 14 10 92 7 Metatarsus 5 92 3 10 96 5 Femur 6 44 26 4 54 26 Feb-Apr Tibia 6 44 29 4 59 25 Metatarsus 6 64 24 3 50 31

A sample of 13 femur marrow fat of red deer killed by wolves, all in the cold season, was analysed. Most red deer killed by wolves showed a low level of fat in the femur marrow (0- 75%), while most roe deer were in good physical condition (75-100%) (Fig. 2). However, the lack of a control sample of red deer did not allow us to test whether there was any selection of red deer by wolves on the basis of body condition.

Figure 2. Percentage distribution, over different intervals of fat content in femoral bone marrow, of roe deer and red deer killed by wolves.

50 45 40 Red deer 35 Roe deer

30 25 20

% of individuals 15 10 5 0 0-25% 25-50% 51-75% 75-100% % femur marrow fat

Most studies have found an increased number of pathologies or low body condition in wild ungulates killed by wolves (Peterson et al. 1984; Messier and Crete, 1985; Huggard, 1993). Nevertheless, in other studies conducted on white tailed deer, femur fat levels of wolf kills were high or did not differ from those of the population (Mech and Frenzel, 1971; Fritts and Mech, 1981). The present study found that wolves in the western Alps did not select roe deer in poor body condition. Moreover, as Okarma (1991) found in red deer killed by wolves, the condition of roe deer deteriorated considerably during winter. However this was found both in wolf kills and in the sample control. So, it would seem that body condition could not have affected wolf choice of roe deer individuals. This lack of selection could be explained by several reasons: i) the elusive and secretive life style of roe deer, and their use of habitat with scarce visibility and low group size, make it difficult for wolves to estimate their body condition; ii) its scarce mobility in deep snow and in the broken grounds of the Alpine areas makes this prey, once found, vulnerable to wolf attack even if in good body condition; iii) in this study area, roe deer is, after red deer, the second prey in order of importance in wolf diet; but it is used as available. This lack of selection by wolves for roe deer could make body condition an unimportant element of choice. Most red deer killed by wolves showed a lower level of fat in the femur marrow than did roe deer, likely due to larger body size and resistance to winter conditions. Body condition in this case may have played a role in the prey choice of wolves; however, more information on red deer body condition in this population are needed to verify this hypothesis.

ACKNOWLEDGEMENTS

We wish to thank Marco Costamagna (University of Pisa), Roberto Musso (Comparto Alpino CATO2) for their help in data collection. We are also grateful to the Servizio Tutela Flora e Fauna of Turin province, and the Corpo Forestale dello Stato for indispensable collaboration. The study was supported by the Turin Province and Piemont Regional Government.

32 REFERENCES Del Giudice, G.D., Mech, L.D., Seal U.S. 1990. Effects of winter undernutrition on body composition and physiological profiles of white-tailed deer. The Journal of Wildlife Management 54:539-550. Fong, D.W. 1981. Seasonal variation of marrow fat content from Newfoundland moose. The Journal of Wildlife Management 45: 545-548. Fritts, S.H., Mech, L.D. 1981. Dynamics, movements and feeding ecology of a newly- protected wolf population in north-western Minnesota. Wildlife Monographs No. 80. Fuller, T., Coy, P., Peterson, W. 1986. Marrow fat relationships among leg bones of white- tiled deer. Wildlife Society Bulletin 14: 73-75. Gazzola, A., Bertelli, I., Avanzinelli, E., Tolosano, A., Bertotto, P., Apollonio, M. 2005. Predation by wolves (Canis lupus) on wild and domestic ungulates of western Alps, Italy. Journal of Zoology of London 266: 205-213. Holand, O. 1992. Fat indices versus ingesta-free body fat in European roe deer. The Journal of Wildlife Management 56: 241-245. Huggard, D.J. 1993. Prey selectivity of wolves in Banff National Park. II. Age, sex, and condition of elk. Canadian Journal of Zoology 71: 140-147. Huot, J., Goudreault, F. 1985. Evaluation of several indices for predicting total body fat of caribou. Proceedings of the North American Caribou Workshop 2: 157-175. Husseman, J.S., Murray, D.L., Power, G., Mack, C., Wenge,r C.R., Quiley, H. 2003. Assessing differential prey selection patterns between two sympatric large carnivores. Oikos 101: 591-601. Mech, L.D. 1970. The wolf: ecology and behaviour of an endangered species. The Natural History Press, Garden City, N. Y. Mech, L.D., Frenzel, L.D. 1971. An analysis of the age, sex, and condition of deer killed by wolves in northeastern Minnesota. U. S. Forest Service Research Paper NC-52: 35-50. Mech, D.L., Meier, T.J., Burch, J.W., Adams, L.G. 1995. Patterns of prey selection by wolves in Denali National Park, Alaska. Pages 231-243 [In: Ecology and conservation of wolves in a changing world. Carbin L. N., Fritts S. H. and Seip D. R. eds]. Canadian Circumpolar Institute Occasional Publication No 35: 231-243.

33 Messier, F., Crete, M. 1985. Moose-wolf dynamics and the natural regulation of moose populations. Oecologia, 65: 503-512. Neiland, K.A. 1970. Weight of dried marrow as indicator of fat in caribou femurs. The Journal of Wildlife Management 34: 904-907. Okarma, H. 1984. The physical condition of red deer falling prey to the wolf and lynx and harvested in the Carpathian mountains. Acta Theriologica 29, 23: 283-290. Okarma, H. 1991. Marrow fat content, sex and age of red deer killed by wolves in winter in the Carpathian mountains. Holoarct, Ecol. 14: 169-172. Peterson, R.O., Woolington, J.D., Bailey, T.N. 1984. Wolves of the Kenai Peninsula, Alaska. Wildlife Monographs No 88. Ratcliffe, P.R. 1980. Bone marrow fat as an indicator of condition in roe deer. Acta Theriologica 25: 333-340. Temple, S. A. 1987. Do predators always catch substandard individuals disproportionately from prey populations? Ecology 68: 669-674. Watkins, B.E., Witham, J.H., Ullrey, Watkins, D.J., Jones, J.M. 1991. Body composition and condition evaluation of white-tailed deer fawns. The Journal of Wildlife Management 55: 39-51.

III

Wolf impact on a multi-species ungulate community of the western Alps, Italy.

- Manuscript -

Gazzola A., Avanzinelli E., Bertelli I., Tolosano A., Bertotto P., Musso R., Apollonio M.

35 WOLF IMPACT ON A MULTI-SPECIES UNGULATE COMMUNITY OF THE WESTERN ALPS, ITALY.

Gazzola A.*, Avanzinelli E.*, Bertelli I.*, Tolosano A.**, Bertotto P.**, Musso R.#, Apollonio M.*

* Department of Zoology and Biological Anthropology, University of Sassari, Via Muroni 25, I-07100, Sassari, Italy ** Provincial Administration of Turin, Via Valeggio 5, I-10128, Turin, Italy # Comprensorio alpino CATO2

36 ABSTRACT Predation impact by wolves (Canis lupus) on prey populations was analysed in a portion of the western Alps (Valle di Susa, Turin Province, Italy) characterized by a rich wild ungulate community. During the last decades of the 19th century, the wolf became extinct in the study area, only to reappear one hundred years later. Since 2000, the presence of two wolf packs was confirmed (Bardonecchia pack and Gran Bosco pack). The number of wolves ranged from 7 to 15, and wolf densities from 1.7 to 2.9 wolves/100 km2 during the study period. Food habits of wolves were mainly based on wild ungulate. Red deer (Cervus elaphus) and roe deer (Capreolus capreolus) were the staple preys. Chamois (Rupicapra rupicapra) was consumed less despite its high density. The potential impact of wolves on prey populations was calculated on the basis of the daily prey biomass per wolf. Two methods were adopted to evaluate it. Minimum daily food consumption per wolf was calculated using Nagy’s formula based on the field metabolic rate (FMR), which predicts that an adult wolf (32 kg) needs 2.6 kg/d of prey biomass. We hypothesized, using the second method, an upper limit of daily food intake estimated at 5.6 kg/d per adult wolf based on the kill rate of Polish wolves (PW). From 2000 to 2002, wolves annually removed 20-34 red deer, 21-58 roe deer, and 7-14 chamois per 100 km2, according to the FMR method. Those amounts were equivalent to 19- 51% of the annual mortality of red deer, 6-28% of roe deer and 6-9% of chamois. Additionally, hunting accounted for 58-94% of the annual mortality of red deer, 18-29% of roe deer and 22-43% of chamois. Other mortality factors (i.e. traffic accidents, disease, poaching) constituted a small percentage of annual mortality of red deer (5-8%) and chamois (1-5%), but not for roe deer (28-54%). During the study period, density of wolf preys was stable. Wolf predation did not seem to seriously affect ungulate populations The role of wolves on wild ungulate populations of Susa Valley seems to be compensatory, taking the place of other mortality factors.

Key words: wolf, ungulate, predation.

37 INTRODUCTION During the last century, in most countries of central and western Europe, the wolf (Canis lupus) was exterminated. Forest destruction, the lack of wild prey species and consequently higher competition with humans for resources (game species, livestock) led to a high level of persecution for the wolf on the part of rural people (Breitenmoser, 1998). Wolf populations were fragmented and survived in small and inaccessible mountainous areas of the Iberian Peninsula, the Balkans and Italy (Boitani and Ciucci, 1993; Promberger and Hofer, 1994). The Italian wolf population reached a historical minimum around 1970, when it was estimated at about 100 individuals (Boitani and Zimen, 1975). Legal protection since 1972, the setting up of protected areas, human abandoning of the countryside, expansion of woodlands, and reintroduction and restocking of wild ungulate species are important factors that have led to the recovery of the Italian wolf population and the recolonisation of its historical range (Apollonio, 1992; 2004; Apollonio et al., 2004a). Between 1970 and 1990, the wolf population grew and, in 1990, was estimated at 500-1000 individuals. During the last 20 years, wolf recolonisation started from the Apennine mountains and spread to the Maritime Alps (Italy and France) and to Piedmont (Italy), reaching the southern French Alps in 1992 (Poulle, 1995). In the Alta Valle di Susa (Turin province; Piedmont region), from the last decade of the 19th century to the last decade of the 20th (Brunetti, 1984), the wolf was extinct, but it reappeared between 1994 and 1997 (Bertotto and Luccarini, 1999). The positive trend of wild ungulate populations, followed consequently by that of their predator, has been a widespread phenomenon in Italy during the last three decades (Apollonio, 2004; Apollonio et al., 2004b). So, the natural process of recolonisation by wolves across the Italian mountainous areas has allowed a wolf-multiple prey species ecosystem to be reconstructed. The establishment of good ecological conditions had a big impact on the trophic ecology of the wolves. In fact, they responded to the recovery of wild ungulate populations by a marked shift in feeding habits that is now based on wild ungulates in most of the northern Italian and French Alps (Mattioli et al., 1995; Meriggi et al., 1996; Poulle et al., 1997; Capitani et al., 2004, Gazzola et al., 2005). However, Italy is characterized by high human population density and activity even in mountainous areas. For this reason, the Italian mountains are constituted

38 by a mosaic of small and medium-sized areas with high abundance and diversity of wild ungulates (Apollonio et al., 2004a). Italian wolf conservation is based on the maintenance of good ecological features in these areas. Information on the effects of predation by wolves on wild ungulates lends support to wildlife managers in their ensuring a sustainable ungulate hunting bag after wolf recolonisation, and it is important to determine this role in the context of conservation and rational management of living natural resources (Głowaciński and Profus, 1997). Most studies on the role of wolves in regulating or limiting wild ungulate populations have been conducted in North America (Messier and Crête, 1985; Ballard et al., 1987; Fuller, 1989; Gasaway et al., 1992; Peterson et al., 1998; White and Garrott, 2005). Instead, most European studies concern wolf diet (Salvador and Abad, 1987; Okarma, 1995; Meriggi and Lovari, 1996; Poulle et al., 1997; Capitani et al., 2004; Gazzola et al., 2005) and only a few address the influence of wolf predation on prey populations (Głowaciński and Profus, 1997; Jędrzejewski et al., 2002; Kojola et al., 2004). We evaluated the effect of wolf predation over three years in an area of the western Alpine region (the Upper Susa Valley) recently recolonised by wolves and characterised by a rich wild ungulate community. The mortality induced by wolves was compared with ungulate densities of ungulates and mortality factors, specifically harvesting by hunters and other natural/human mortality aspects.

STUDY AREA The study area is located in an Alpine region in the western part of Turin province (45°05’N, 7°E), and extends over 523 km2 from 800 to 3300 m a.s.l.. The landscape at lower altitude is a mixture of mixed woods of beech (Fagus silvatica), maple (Acer platanoides) and birch (Betula pubescens) and fields, while from 1100 to 2300 m a.s.l., coniferous forests of pine (Pinus sylvestris), spruce (Picea abies), and larch (Larix decidua) are dominant. Over 2300 m, alpine meadows and rocky areas replace the forest.

39 From 1850 to 1970, the wild ungulate community of Susa Valley was in a poor condition: cervids were rare after the end of the 19th century and became extinct during World War II. Although the chamois never disappeared, it fell to a low density. Since 1962, reintroduction and restocking of wild ungulate was the task of hunting associations and the Turin Province Administration (Luccarini and Mauri, 2000; Demeneghi et al., 1987). Thanks to these operations, rapidly growing wild ungulate populations have been restored in Susa Valley, and in 1985 hunting harvests began (Demeneghi et al., 1987). Currently, a rich wild ungulate community lives in Alta Valle di Susa constituted by six species: chamois (Rupicapra rupicapra), roe deer (Capreolus capreolus), red deer (Cervus elaphus), wild boar (Sus scrofa), alpine ibex (Capra ibex) and moufflon (Ovis orientalis mousimon). The first four ungulate specie are annually harvested by hunters. Flocks of sheep and goats and herds of cows are free-ranging on high pastures from May to October and are stabled in the valley during the rest of the year. During the study period, the presence of two stable wolf packs (Bardonecchia pack; Gran Bosco pack) was verified in the study area. The climate is continental with prolonged snow cover in winter from October- November to April-May, depending on altitude.

MATERIALS AND METHODS Wolf status To evaluate wolf status, we used snow-tracking and wolf-howling techniques. The largest number of wolves of each wolf pack was accepted as the size of the wolf pack in a given season (following Jędrzejewski et al., 2000). Two seasons were considered: May-October (summer); November-April (winter). During the winter season, wolves were tracked in presence of fresh snow (24–48 h after snowfall). When a wolf trail was found, it was followed until the number of individuals travelling along it became distinguishable. The largest number of wolves travelling together within a considered area was used as an estimate of winter pack size.

40 Wolf-howling surveys were carried out, only in summer from late June to end October, to ascertain the presence of wolf packs and their reproductive status (i.e., birth of a litter). The approach described as ‘‘saturation census’’ by Harrington and Mech (1982) was adapted to local requirements, dictated especially by the mountainous topography. Sampling sites were located in prominent places, in order to maximize the range of audibility and to minimize sound dispersion. The equipment, artificial stimuli, and session protocols have been described elsewhere (Gazzola et al., 2002). The whole census area was divided into two sectors, defined on the basis of hypothetical activities range of wolf packs, and for every working night, two adjacent sectors were covered simultaneously by different teams. Home- sites, when located, were monitored by further howling sessions once a week. Home-sites were never visited by researchers during their use by wolves, and the number of howling wolves (pup/adult) was determined when possible. Each reply by packs to wolf howling stimulus was recorded. Pack size was determined by maximum number of pups and adults heard in all replies collected during each summer.

Wolf feeding habits and energy requirements Wolf food habits were assessed by scat analysis. Prey remains (hairs and bones), fruit and grasses, found in every scat were dried at 50 °C for 24 h. Prey remains were identified on the basis of a reference collection of mammal hairs. Moreover, the age of the ungulates recovered from wolf scats was determined through the analysis of recognizable bone fragments, teeth, and macroscopic comparison of hairs (see Materials and Methods in Mattioli et al., 1995). We distinguished two age classes for cervids and chamois: juveniles (<1 year of age) and adults (> 1 year). When a component at a specific taxonomic group could not be recognised, it was considered to be undetermined. On the basis of 716 scats collected from May 2000 to April 2003, we calculated the relative biomass (Bio) of ungulate species and other mammals. Using the volume values, we applied the biomass model of Ciucci et al. (2001): Y = 0.274 + 0.011 X, where Y represents the biomass (kg) of preys for each collectable scat and X is the live weight of preys. For wolf diet analysis two seasons were considered: May-October (summer); November-April (winter).

41 Daily food consumption by wolves was calculated through the field metabolic rate (FMR) for all eutherian mammals (Głowaciński and Profus, 1997). The equation, derived from Nagy’s formula (1987), is closely correlated with body mass: FMR2 (kJ/d) = 2.58 W0.862, where W is body weight in g. This allows indirect estimates of total daily energy expenditure of a free- living animal. Since food consumption is affected by body mass, the wolf pack age structure was taken into account, but for summer only because wolves reach adult body mass in midwinter. Data from Italy give an average body weight of 32 kg for an adult wolf (> 1 year old). The average body mass of young individuals was considered to be 16 kg in summer. Calculations based on FMR yielded 2.6 kg of meat per day for an adult wolf and 1.4 kg for a young one (FMR). On the basis of the number of wolves and structure of packs (young/adults), we calculated the seasonal body mass required by wolves. Moreover, we took into account a second method to evaluate the impact of wolves on a wild ungulate community. On the basis of the kill rate by wolves in eastern Poland (Jędrzejewski et al., 2002), it was determined that a wolf over six months old needs 5.6 kg of meat each day (PW), while 2.8 kg are necessary for a young wolf (< six months) (Kojola et al., 2004).

Average body weight of prey Table 1 reports the average body mass of each wolf food item. That for staple preys (cervids and chamois) was based on local data from hunting bags. Sex and age of preys consumed by wolves were taken into account. Moreover, values of average body mass of ungulates, taken for calculations, were reduced by 10% for juveniles and 25% for adults (Fuller, 1989; Okarma, 1992; Głowaciński and Profus, 1997) on account of inedible and indigestible parts of their bodies (large bones, hair and stomach). On the basis of consumption rates for each wild ungulate, we calculated the effective usage of carcasses by wolves. Based on the mass of 112 carcass left-overs, wolves consumed 55% of their live mass in red deer, 85% in roe deer and 73% in chamois. It was assumed that young individuals were completely consumed by wolves in summer.

42 Live weights of the wild prey species quantitatively unimportant in wolf diet were taken from literature (Bassano et al., 1999; Mattioli et al., in prep.). The single values of body mass were referred to food items where the age of prey consumed by wolves was not determined by scat analysis. The mean body weight of different domestic ungulates was based on data from species bred in the study area. Moreover, sex, age and use of livestock carcass by wolves in the study area were taken into account (Dalmasso, 2003).

Table 1. Mean body mass of wolf prey items Body mass (kg) Prey items Adult or mean value Young (in winter) Young (in summer) red deer 110.4 57.8 26.5 roe deer 25.0 15.4 8.0 chamois 30.2 14.2 11.0 mouflon 29.5 _ _ wild boar 20.3 _ _ hare 3.0 _ _ marmot 4.5 _ _ rodent 0.03 _ _ sheep 66.8 _ _ goat 44.9 _ _ cattle 110.0 _ _

Surveys and mortality of the ungulate community Surveys of cervids were performed each April from 2000 to 2002. Chamois census was carried out each year in June. Data on wild ungulates were obtained by observations from vantage-points to estimate number of individuals, sex, and age structure of each species. Summer abundance of ungulates was calculated on the basis of their late winter counts, on the percentage of adult females in the population and on female fertility. The latter data were obtained by counts of foetuses found in females shot by hunters and found dead in the study

43 area. (Ferroglio, pers. comm.). Data on ungulate mortality of the study area were given by the Provincial Administration of Turin (Servizio Tutela della Flora e della Fauna). Species, age (juvenile or adult), sex, and condition of ungulate carcasses found in the whole study area were determined, and the cause of death was recorded. That for ungulates was evaluated by the veterinary staff of Turin University. Data on annual hunting harvests were provided by the Game Alpine Department (CaTO2) and by Gran Bosco Natural Park Administration.

44 RESULTS

Wolf dynamics During 2000-2003, we monitored two wolf packs: the Bardonecchia pack (2-9 wolves) and the Gran Bosco pack (3-6 wolves). The total number of wolves ranged from 7 to 15 (Table 2). During May 2000-April 2001, the Bardonecchia pack was composed of 2-3 wolves, but after summer 2001, it grew in size (7-9 wolves). Data about pack reproduction was available for summers 2001 and 2002, but not for summer 2000. The size of the Gran Bosco pack was stable from summer 2000 to winter 2001/02 (5-6 wolves) but decreased (3 individuals) during the last year. Reproduction was confirmed only in summer 2000. Direct observations and wolf howling technique provided an estimate of the number of pups in each pack. During summer 2001, two pups were heard in the Bardonecchia pack, but we hypothesised a litter size of 6 pups on the basis of snow-tracking data. In summer 2002, the wolf pack was constituted by 4 adults and 3 pups. During summer 2000, the Gran Bosco pack was composed of 2 adult wolves and 3 pups.

Table 2. Wolf population dynamics

Years 2000-2001 2001-2002 2002-2003

Season summer winter summer winter summer winter

Bardonecchia pack 2 3 3 (+6*) 9 4 (+3) 8

Gran Bosco pack 2 (+3) 6 6 6 3 3

Number of wolves 7 9 15 15 10 11 (+) litter size * number of pups hypothesised

45 Wild ungulate dynamics Roe deer and chamois were the most abundant species in the ungulate community, followed by red deer (Table 3). Wild boar was scarce and was not censused. Ibex had been recently reintroduced (1994-1996), and 75 individuals were present. Moufflon was present on the edge of the study area with about 50 individuals. Data of density and population increases of main wolf prey species were calculated for each year (Table 3).

Table 3. Population dynamics of staple wolf preys in Alta Valle Susa (2000-2002).

Years Parameters (late winter to late winter) Mean ± SD (no.individuals/100 km2) 2000-01 2001-02 2002-03

Red deer (Cervus elaphus)

Density in late winter 218 218 223 219 ± 2.7

Juveniles born in spring 83 66 78 76 ± 8.7

Density in summer 327 284 296 302 ± 22.0

Annual mortality 109 66 73 83 ± 22.8

Roe deer (Capreolus capreolus)

Density in late winter 251 234 232 239 ± 10.4

Juveniles born in spring 299 236 201 245 ± 49.4

Density in summer 613 487 435 512 ± 91.6

Annual mortality 363 253 204 273 ± 81.4

Chamois (Rupicapra rupicapra)

Density in late winter 350 355 341 349 ± 7.1

Juveniles born in spring 150 108 105 121 ± 24.9

Density in summer 568 458 461 495 ± 62.7

Annual mortality 218 102 119 147 ± 62.6

46 Wolf diet The most important prey of wolves were wild ungulates (Table 4), which constituted 51.5- 98.9% of the biomass consumed (summer: 51.5-77.7%; winter: 95.3-98.9%). Cervids represented 43.3-65.9% of total biomass eaten by wolves in summer and 87.4-92.3% in winter. Livestock was an important food item only in summer (summer: 20.7-45.6%; winter: 0.0-4.1%) while chamois was the 3rd most important prey in winter (summer: 4.4-15.2%; winter: 4.0-10.9%).

Table 4. Wolf diet (% of biomass)

2000-01 2001-02 2002-03 Food items summer winter summer winter summer winter

Red deer 39.6 64.5 22.0 71.5 36.1 41.2

Roe deer 14.4 23.0 17.4 18.7 27.4 51.1

Chamois 18.5 10.9 4.0 4.0 9.7 6.4

Moufflon 0.0 0.5 1.4 0.0 0.0 0.0

Wild Boar 0.6 0.1 1.1 0.3 0.9 0.0

Alpine Ibex 0.0 0.0 0.0 0.0 0.0 0.0

Hare 0.5 0.8 0.2 0.5 1.6 1.2

Marmot 2.9 0.2 2.0 0.0 0.0 0.0

Rodents 0.1 0.1 0.5 0.0 0.0 0.1

Sheep 9.8 0.0 43.6 4.5 24.4 0.0

Goats 4.0 0.0 7.4 0.5 0.0 0.0

Cattle 9.6 0.0 0.0 0.0 0.0 0.0

Total 100.0 100.0 100.0 100.0 100.0 100.0

47 The impact on wild ungulate population of wolves, hunting harvests and other causes of mortality

On the basis of the field metabolism rate (FMR) an adult wolf (32 kg) needs 2.56 kg of meat per day, while a young wolf (< 6 months old; weight: 16 kg) needs 1.40 kg of meat per day. Consequently, the mean daily consumed rate was 0.11 prey items per wolf per day, a food requirement amounting to 41.0 ungulates per adult wolf per year (Table 5). Considering cervids only, the average consumed rate by wolves was 0.08 prey per wolf per day. Taking into account the second method, i.e., that of 5.60 kg of meat per day per adult wolf (Jędrzejewski et al., 2002) and 2.8 kg per young wolf (Kojola et al., 2004), we obtained an average of consumed rate of 0.24 prey items per wolf per day. Food requirements per wolf yearly amounted then to 87 ungulates (Table 5). The average consumed rate by wolves was 0.17 prey item per wolf per day.

48 Table 5. Food requirement per adult wolf (32 kg) estimated by FMR formula (Nagy, 1987) and based on mean daily food intake by eastern Poland wolves (Jędrzejewski et al., 2002).

Daily consumption rate per adult wolf Annual consumption rate per adult wolf % (kg/d/wolf) (no. individuals/yr/wolf) Food items biomass (2.56 kg/d/wolf) (5.6 kg/d/wolf) (2.56 kg/d/wolf) (5.6 kg/d/wolf)

Red deer 55.3 1.42 3.10 12.4 26.8

Roe deer 22.6 0.58 1.27 20.0 43.1

Chamois 8.6 0.22 0.48 6.7 14.4

Moufflon 0.2 0.01 0.01 0.1 0.3

Wild Boar 0.4 0.01 0.02 0.4 0.8

Alpine Ibex 0.0 0.00 0.00 0.0 0.0

Hare 0.6 0.01 0.03 1.8 3.9

Marmot 0.8 0.02 0.04 2.6 5.6

Rodents 0.1 0.00 0.01 46.4 99.8

Sheep 8.6 0.22 0.48 4.5 9.8

Goats 1.4 0.04 0.08 1.0 2.1

Cattle 1.4 0.04 0.08 0.2 0.4

Total 100.0 2.56 5.60

Table 6 reports values of predation impact of wolves on staple preys (cervids and chamois) in relation to densities and annual increase due to reproduction (estimated number of young born annually). Considering wolf food expenditure calculated according to Nagy’s formula (FMR), from summer 2000 to winter 2002/03, wolves annually killed 25 ± 8.1 ungulates/100 km2. The highest predation was on cervids (41-78 individuals/100 km2), followed by chamois (7-14 individuals/100 km2). The numbers of red deer and roe deer consumed by wolves were similar during 2000-2001 (red deer: 20-34 individuals/100 km2; roe deer: 21-38 individuals/100 km2),

49 but markedly different in the last year (red deer: 20 individuals/100 km2; roe deer: 58 individuals/100 km2).

Table 6. Wolf impact on red deer, roe deer and chamois in Alta Valle di Susa in relation to population densities and increase of prey

Food requirement per wolf estimated by Mean daily food intake by FMR formula (Nagy, 1987) eastern Poland wolves (Jędrzejewski et al., 2002)

Year Annual predation on ungulates (%) Annual predation on ungulates (%) Annual predation Annual predation (no. killed/100 km2) Spring Annual Annual (no. killed/100 km2) Spring Annual Annual density increase mortality density increase mortality

Red deer (Cervus elaphus)

2000-01 20 6 25 19 44 13 53 40

2001-02 34 12 51 51 72 25 109 109

2002-03 20 7 26 27 43 14 55 59

2000-03 25 ± 8.1 8 ± 3.2 34 ± 14.7 32 ± 16.7 53 ± 16.5 17 ± 6.7 72 ± 31.8 69 ± 35.6

Roe deer (Capreolus capreolus)

2000-01 21 3 7 6 44 7 15 12

2001-02 38 8 16 15 81 17 34 32

2002-03 58 13 29 28 124 28 62 61

2000-03 39 ± 18.5 8 ± 5.0 17 ± 11.1 16 ± 11.1 83 ± 40.0 17 ± 10.5 37 ± 23.6 35 ± 24.6

Chamois (Rupicapra rupicapra)

2000-01 14 2 9 6 30 5 20 14

2001-02 7 2 7 7 16 4 15 16

2002-03 11 2 10 9 23 5 22 19

2000-03 11 ± 3.5 2 ± 0.0 9 ± 1.5 7 ± 1.5 23 ± 7.0 5 ± 0.6 19 ± 3.6 16 ± 2.5

Wolf predation on red deer, expressed as a percentage of deer consumed out of the total deer counted, constituted 8 ± 3.2% of deer censused in summer. The maximum value (12%) was

50 recorded during 2001. During the study period, the impact of wolf predation on roe deer population increased from 3 to 13% (mean value: 8 ± 5.0%), while on chamois the percentage was 2% and stayed stable among years. Wolf predation, expressed as a percentage of annual increase, constituted 34 ± 14.7% in red deer, 17 ± 1.1% in roe deer, 9 ± 1.5% in chamois; however, with respect to the total annual mortality (the difference between summer and subsequent late winter densities), it amounted for red deer to 32 ± 16.7%; roe deer to 16 ± 11.1%, and chamois to 7 ± 1.5%. On the basis of mean daily food consumption by Polish wolves (5.6 kg/day/wolf), the wolf impact clearly doubled with respect to the previous method (Table 6). The annual number of red deer consumed by wolves was 53 ± 16.5 individuals/100 km2, while for roe deer it was 83 ± 40 individuals/100 km2 and for chamois 23 ± 7 individuals/100 km2. The annual predation on ungulates did not exceed 28% of spring density in any one prey. Red deer, however, suffered much more wolf predation than did roe deer and chamois (Table 6). In the same period, hunters (Table 7) annually harvested 156-175 ungulates/100 km2 (red deer: 61 ± 0.6 head/100 km2; roe deer: 57 ± 7.6 head/100 km2; chamois: 44 ± 3.0 head/100 km2). Moreover, a supplement of ungulates dying from other mortality factors was annually found by rangers (19-41 ungulates/100 km2). Hunting harvest on red deer compared with annual mortality amounted to 77 ± 18.1%, a value much more higher than those for roe deer and chamois, respectively 22 ± 6.1% and 33 ± 10.5%. On the contrary, considering the influence of “other causes” of death, roe deer yielded the highest percentage of annual mortality (37 ± 14.5%). Low values were recorded for red deer (7 ± 1.7%) and for chamois (3 ± 2.1%). Harvest by hunters appeared a more important factor on red deer mortality than did wolf predation, and together they played a relevant role in the annual mortality (FMR: 77-145%; PW: 107-208%). On the contrary, wolf predation on roe deer played a small part in the annual mortality, while the major impact was represented by hunting pressure together with “other causes” (50-71%). Hunting harvest seemed to be the most important mortality factor for chamois (22-43%).

51 Table 7. Impact of hunting harvests and other causes of mortality on wild ungulate population in Alta Valle Susa (2000-2002)

Hunters Other causes of mortality

Annual predation on ungulates (%) Annual losses (%) Annual harvest Annual mortality (no. (no. killed/100 km2) Spring Annual Annual killed/100 km2) Spring Annual Annual density increase mortality density increase mortality Red deer (Cervus elaphus) 63 19 76 58 5 2 6 8 62 22 94 94 3 1 5 5 58 20 75 80 5 2 6 8 61 ± 2.6 20 ± 1.5 82 ± 10.7 77 ± 18.1 4 ± 1.2 2 ± 0.6 6 ± 0.6 7 ± 1.7 Roe deer (Capreolus capreolus) 64 10 21 18 34 6 12 54 49 10 21 19 15 3 6 30 54 14 29 29 16 4 8 28 57 ± 7.6 11 ± 2.3 24 ± 4.6 22 ± 6.1 22 ± 10.7 4 ± 1.5 9 ± 3.1 37 ± 14.5 Chamois (Rupicapra rupicapra) 47 8 32 22 1 0.2 1 5 44 10 41 43 1 0.1 1 1 41 9 39 34 2 0.4 2 4 44 ± 3.0 9 ± 1.0 37 ± 4.7 33 ± 10.5 1 ± 0.6 0.2 ± 0.2 1 ± 0.6 3 ± 2.1

52 DISCUSSION In the Alps, the abundance and richness of the wild ungulate community is actually much higher now than at the time of wolf eradication, 150 years ago, when chamois only survived in small numbers, while other wild ungulates became extinct. Hunting right restrictions, abandoning of the countryside, expansion of woodlands, setting up protected areas, and the reintroduction and restocking of wild ungulates caused the reestablishment of stable and rich wild ungulate communities (Apollonio, 1992; Breitenmoser, 1998). These were the major factors that led to the recovery of the Italian wolf population and to the recolonisation of its historical range (Apollonio, 1992; Apollonio et al., 2004a). Wolf restoration produced a new level of complexity in the ecosystem, but its influence on prey population is difficult to determine. Under certain circumstances, wolves can reduce, or even locally eradicate, some prey species (Mech and Karns, 1977). At other times, wolf predation may only be compensatory for other mortality causes that occur in the absence of wolves (Ballard et al., 1987). Klein (1995) reported that the black-tailed deer population of Coronation Island was wiped out by wolves after their introduction. On the contrary, on Isle Royale wolves coexist with the world’s highest density of moose (Peterson et al., 1998). The influence of wolves on ungulate populations is very difficult to evaluate especially in complex European biocenoses. In fact, in North America often only one ungulate species dominates (moose, white-tailed deer, or caribou); instead, most European countries are characterized by richer communities of ungulates (Okarma, 1985). Numerous studies on the effect of wolves on ungulates have been conducted mainly in North America (Peterson et al., 1984; Mech et al., 1987; Gasaway et al., 1992; White and Garrott, 2005), but none have yet been made in European Countries, except for Poland (Jędrzejewski et al., 2000; 2002). The average estimated daily food availability, carried out by eighteen North America studies, was 5.4 kg/wolf/day, or 0.14 kg/kg wolf/day (Peterson and Ciucci, 2003). The estimates of daily food availability ranged from a minimum of 0.06 kg/kg wolf/day (Fuller, 1989) to a maximum of 0.29 kg/kg wolf/day (Hayes, 1995). The highest values referred to wolves that killed mainly ungulates with large body mass (bison, Bison bison; moose, Alces alces), while the lowest to smaller prey (Dall sheep, Ovis dalli; white tailed deer, Odocoileus virginianus). Considering the latter studies, where body size of prey items was

53 comparable to that of European prey species, the mean food availability was <3.0 kg/wolf/day. However, these values are overestimates because the proportion of a carcass lost to scavengers must be subtracted from the consumption estimates. (Mech and Boitani, 2003). Our lower estimate of annual predation by wolves on ungulates was based on the field metabolism rate (FMR), while the upper one considered the average of meat consumption per wolf per day observed by Jędrzejewski et al. (2002) in Poland. Species, age, sex, and consumption rates (to evaluate degree of unconsumed portions) of carcasses used by wolves were considered in wolf impact calculations. Moreover, a relatively high value (25% for adult ungulates) of inedible biomass was considered in order for the food requirement value not to be an underestimate. This approach should lead to a slight discrepancy between the hypothesized and the effective impact of wolves on ungulate populations. The estimate of daily food requirement per 32 kg wolf is 2.6 kg (FMR) (3.2 kg/wolf/day taking into account that 25% of carcass biomass is inedible), while it is 5.6 kg/wolf/day for the Polish wolf. A drastic decline of red deer population should have been observed during the period of study, if the average daily consumption rate of the Polish wolves were adopted, but that was not the case. Moreover, Italian wolves are smaller than Polish ones, and thus they probably have lower energy requirements. We therefore think that the impact of wolves can be better expressed using the value obtained with the FMR method than that based on the consumption rate estimated for eastern Europe. Our study seem to be in line with other cases in which prey populations were not influenced by the presence of wolves (Mech, 1986; Adams and Dale, 1998; Mech et al., 1998; Peterson et al., 1998; Nelson and Mech, 2000). Wild ungulate densities did not change before or after arrivals of wolves in Susa Valley (census data of CaTO2). Even if, red deer was the prey species most influenced by wolf presence, wolf predation alone was a poor predictor of its population dynamics. In fact, wolf predation, as percentage of annual mortality, yielded a higher value in red deer (32%) than roe deer and chamois (respectively, 16% and 7%). A similar phenomenon was found for Białowieża Primeval Forest (Poland) where 40% of annual red deer mortality was due to wolves, against 24% for wild boar and 7% for roe deer (Jędrzejewski et al. 2002). The high susceptibility of red deer, in the Susa valley, could be a result of the close overlap of habitat and altitude use with the wolf, and to their more

54 conspicuous herding. On the contrary, the lowest wolf predation, that on chamois, was mainly due to the scarce accessibility of this prey. The high use of wide-rocky and high altitude areas by chamois and their being well adapted to snow conditions makes it difficult for wolves to catch them in Susa valley (Gazzola et al., 2005). Numerous studies (Fritts and Mech, 1981; Peterson et al. 1984; Jędrzejewski et al. 1992; Mattioli et al., 1995), indicated how young ungulates proved to be an important fraction of the wolf diet. Thus, the greatest demographic effects on ungulate population by wolves should be due to predation on young of the year (Pimlott, 1967; Mech, 1970). In our study area, wolves take up to 21% of annual production of red deer, 11% of roe deer and 5% of chamois. These low values may be partially explained by the wide use of domestic ungulates by wolf in summer. The share of wolf predation among cervids, chamois and livestock seems to guarantee the reproduction and stability of wild ungulate populations. Moreover, numerous papers report that wolf predation tends to focus primarily on the youngest and debilitated members of prey herds (Mech, 1966; Bubenik, 1972; Schwartz et al., 1992). Thus such herds, given the predation of wolves, tend to be constituted by individuals in good physical and health conditions, and therefore of high productivity. In Susa valley, most of the red deer killed by wolves were suffering from heavy undernutrition, while in roe deer, femur fat levels of wolf kills were high and did not differ from those of the population (Gazzola et al., unpublished). The viability of the red deer population was confirmed by their high reproduction rate (Meneguz et al., 2005). Some evidence for the partially compensatory role of predation was shown in central Europe, as in winter wolves selected the weakest deer with very low fat reserves (Jędrzejewski, 2005). Thus we believe that the wolf recolonisation of the western Alps should not determine the decline of the ungulate population but, quite the contrary, it will be a driving force of change in prey distribution, behaviour, and movements, and will determine a change in the population structure of prey species. Hunting harvest by humans and wolf predation together could have been important factors, limiting red deer population size in Susa Valley. The effects of these two factors were likely additive and could explain the 109% of red deer annually mortality.

55 On the contrary, wolf predation was the lowest mortality factor for roe deer, while traffic accidents were the main cause of mortality (37%) followed by hunting harvests (22%). Harvesting by hunters seemed to be the only important cause of death of chamois, but it must be emphasized that the low value of causes such as malnutrition, disease and avalanches (3%) is clearly underestimated, due to the difficulties of reaching and monitoring the winter quarters of this species and of collecting carcasses.

ACKNOWLEDGEMENTS

We wish to thank Ezio Ferroglio (Dip. di Produzione Animale, Epidemiologia ed Ecologia, Università di Torino) for his help in data collection; Alberto Dotta, Federico Kurschinski, Rinaldo Gros (Consorzio Forestale Alta Valle di Susa), Pino Arleo (Comprensorio Alpino CATO2) are thanked for their support in technical activities. We are also grateful to the Servizio Tutela Flora e Fauna of Turin Province, the Corpo Forestale dello Stato and Gran Bosco di Salbertrand Natural Park rangers for invaluable collaboration. We would also like to thank James Burge for linguistic revision. The study was supported by the Turin Province Administration and Piedmont Regional Government.

56 REFERENCES Adams, L.G., Dale, B.W. 1998. Reproductive performance of female Alaskan caribou. The Journal of Wildlife Management 62: 1184-1195. Apollonio, M., 1992. Evolution of environment and human activities in Apennines in relation to wolf (Canis lupus) presence. In: Proceedings of the Congress ‘Dalla Parte del Lupo’, 9– 10 October 1992, Parma, Italy, World Wildlife Fund, pp. 54–63 (in Italian). Apollonio, M. 2004. Gli ungulati in Italia: status, gestione e ricerca scientifica. Hystrix Italian. Journal of Mammalogy 15 (1): 21-34. Apollonio, M., Mattioli, L., Scandura, M., Mauri, L., Gazzola, A., Avanzinelli, E., 2004a. Wolves in the Casentinesi Forests: insights for wolf conservation in Italy from a protected area with a rich wild prey community. Biological Conservation 120: 249-260. Apollonio, M., Ciuti, S., Pedrotti, L., Banti, P. 2004b. Ungulate distribution, consistency, and management in Italy. International School of Ethology, 21st Course. 12-17 November 2004 Erice, Sicily, Italy. Ballard, W.B., Whiteman, J.S., Gardner, C.L. 1987. Ecology of an exploited wolf population in south-central Alaska. Wildlife Monographs no. 98. The Wildlife Society, Bethesda, MD. 54 pp. Bassano, B., Boano, G., Meneguz, P. G., Mussa, P. P., Rossi, L. 1999. I selvatici delle Alpi Piemontesi, biologia e gestione. Regione Piemonte, EDA, Torino. Pp. 1-239. Bertotto, P., Luccarini, S. 1999. Indagine sulla popolazione di lupo in Alta Val di Susa e Val Chisone. Provincia di Torino, Report. Boitani, L., Ciucci, P. 1993. Wolf in Italy: critical issue for their conservation. In: Promberger, C., Schröder, W. (Eds.), Wolves in Europe: Status and Perspectives. Wildbiologische Gesellschaft München e.V., Ettal. Boitani, L., Zimen, E. 1975. Number and distribution of wolves in Italy. Zeitschrift für Säugetierkunde 40, 102-112. Breitenmoser, U., 1998. Large predators in the Alps: the fall and rise of man’s competitors. Biological Conservation 83 (3), 279-289. Brunetti, R. 1984. Distribuzione storica del lupo in Piemonte, Valle d’Aosta e Canton Ticino Rivista Piemontese di Storia Naturale, 5: 7-22.

57 Bubenik, A.B. 1972. North American moose management in light of European experiences. Proc. N. Am. Moose Conf. 8: 279-295. Capitani, C., Bertelli, I., Varuzza, P., Scandura, M., Apollonio, M., 2004. A comparative analysis of wolf (Canis lupus) diet in three different Italian ecosystems. Mammalian Biology 69, 1-10. Ciucci P., D’Alessio S., Mattei L., Boitani L., 2001. Stima della biomassa consumata tramite analisi degli escrementi: calibrazione del modello sulle principali prede selavatiche del lupo in Italia. In: Proceedings of III Congresso Italiano di Teriologia, San Remo, Italia. Dalmasso, S. 2003. Relazione sull’attività di accertamento dei danni da canide al patrimonio zootecnico in provincia di Torino. Report, 62 pp. Demeneghi, D., Meneguz, P.G., Rossi, L. 1987. Distribuzione, status ed evoluzione delle attuali popolazioni di cervo, capriolo e camoscio in val Susa. In: Ungulati selvatici e territorio: indirizzi di gestione. Provincia di Torino - Regione Piemonte. Fritts, S.H., Mech, L.D. 1981. Dynamics, movements, and feeding ecology of a newly protected wolf population in northwestern Minnesota. Wildlife Monographs, 80: 1-79. Fuller, T.K. 1989. Population dynamics of wolves in north-central Minnesota. Wildlife Monographs, no. 105. The Wildlife Society, Bethesda, MD. 41 pp. Gasaway, W.C., Boertje R.D., Grangaard, D.V., Kelleyhouse, D.G., Stephenson, R.O., Larsen, D.G. 1992. The role of predation in limiting moose at low densities in Alaska and Yukon and implications for conservation. Wildlife Monographs, no. 120. The Wildlife Society, Bethesda, MD. 59 pp. Gazzola, A., Avanzinelli, E., Mauri, L., Scandura, M., Apollonio, M., 2002. Temporal changes of howling in south European wolf packs. Italian Journal of Zoology 69: 157-161. Gazzola A., Bertelli I., Avanzinelli E., Tolosano A., Bertotto P., Apollonio M. 2005. Predation by wolves (Canis lupus) on wild and domestic ungulates of the Western Alps, Italy. Journal of Zoology, London 266: 205-213. Gazzola A., Ferroglio E., Avanzinelli E., Rosso M., Kurschinski F., Apollonio M. 2005. The physical condition of roe and red deer killed by wolves in a region of the western Alps, Italy (manuscript).

58 Głowaciński, Z., Profus, P. 1997. Potential impact of wolves (Canis lupus) on prey populations in eastern Poland. Biological Conservation 80: 99-106. Harrington, F., Mech, L.D. 1982. An analysis of howling response parameters useful for wolf pack censusing. Wildlife Monographs 46: 686–693. Hayes, R.D. 1995. Numerical and functional responses of wolves and regulation of moose in the Yukon. Master’s thesis, Simon Fraser University, Burnaby, BC. 144 pp. Jędrzejewski, W. 2005. Impact of wolf predation, hunting harvest and food resource on red deer population dynamics. In: Proceedings of Stelvio Seventy Conference. 8-11 September 2005. Rabby Valley (Trentino), Italy. Jędrzejewski, W., Jędrzejewska, B., Okarma, H. and Ruprecht, A. L. 1992. Wolf predation and snow cover as mortality factors in the ungulate community of the Bialowieza National Park, Poland. Oecologia 90: 27-36. Jędrzejewski, W., Jędrzejewska, B., Okarma, H., Schimdt, K., Zub, K., Musiani, M. 2000. Prey selection and predation by wolves in Bialowieza Primeval Forest, Poland. Journal of Mammalogy 81 (1): 197-212. Jędrzejewski, W., Schimdt, K., Theuerkauf, J., Jędrzejewska, B., Selva, N., Zub, K., Szymura, L. 2002. Kill rates and predation by wolves on ungulate populations in Bielowieza Primeval Forest (Poland). Ecology 83 (5): 1341-1356. Klein, D.R. 1995. The introduction, increase, and demise of wolves on Coronation Island, Alaska. Pp. 275-280 in L.N. Carbyn, S.H. Frits, and D.R. Seip, eds., Ecology and conservation of wolves in a changing world. Canadian Circumpolar Institute, Edmonton, Alberta. Kojola, I., Huitu, O., Toppinen, K., Heikura, K., Heikkinen, S., Ronkainen, S. 2004. Predation on European wild forest reindeer (Rangifer tarandus) by wolves (Canis lupus) in Finland. Journal of Zoology, London 263: 229-235. Luccarini, S., Mauri, L. 2000. Il cervo in Alta Val di Susa. Comportamento spaziale e uso dell’habitat. Provincia di Torino – Servizio Tutela della Fauna e della Flora. Turin: Ananke.

59 Mattioli, L., Apollonio, M., Mazzarone, V., Centofanti, E. 1995. Wolf food habits and wild ungulate availability in the Foreste Casentinesi National Park, Italy. Acta Theriologica 40 (4): 387-402. Mattioli L., Apollonio M., Gazzola A., Avanzinelli E., Bertelli I., Capitani C. 2005. Prey selection by wolves from a multi-species ungulate community in the Casentinesi Forests, Italy. (manuscript). Mech L.D. 1966. The wolves of Isle Royale. U.S. National Park Service Fauna Series, no. 7. U.S. Govt. Printing Office. 210 pp. Mech, L.D. 1970. The wolf. The ecology and behaviour of an endangered species. New York: Natural History Press. Mech, L.D. 1986. Wolf population in the central Superior National Forest, 1967-1985. USDA Forest Service Research Paper NC-270. North Central Forest Experiment Station, St. Paul, MN. 6pp. Mech, L.D., Karns, P.D. 1977. Role of the wolf in a deer decline in the Superior National Forest. USDA Forest Service Research Paper NC-148. North Central Forest Experiment Station, St. Paul, MN. 23pp. Mech, L.D., Boitani L. 2003. Wolves. Behavior, Ecology and Conservation. The University of Chicago Press. Chicago and London. Mech, L.D., McRoberts, R.E., Peterson, R.O., Page, R.E. 1987. Relationship of deer and moose populations to previous winters’ snow. Journal of Animal Ecology 56: 615-627. Mech, L.D., Adams, L.G., Meier, T.J., Burch, J.W., Dale, B.W. 1998. The wolves of Denali. University of Minnesota Press, Minneapolis. Meneguz, P.G., Borgo, C., Ferra, M. 2005. La gestione della principale popolazione di cervo (Cervus elaphus) delle Cozie torinesi. In: Proceedings of the Stelvio Seventy Conference. 8-11 September 2005. Rabby Valley (Trentino), Italy. Meriggi, A., Lovari, S. 1996. A review of wolf predation in southern Europe: does the wolf prefer wild prey to livestock? Journal of Applied Ecology 33: 1516-1571. Meriggi, A., Brangi, C., Matteucci, C., Sacchi, O. 1996. The feeding habits of wolves in relation to large prey availability in northern Italy. Ecography 19(3):287-295.

60 Messier, F., Crête, M. 1985. Moose (Alces alces) and wolf (Canis lupus) dynamics and the natural regulation of moose populations. Oecologia 65: 503-512. Nagy, K.A. 1987. Field metabolic rate and energy requirements scaling in mammals and birds. Ecological Monographs 57: 111-128. Nelson, M.E., Mech, L.D. 2000. Proximity of white-tailed deer, Odocoileus virginianus, home ranges to wolf, Canis lupus, pack homesites. Canadian Field Naturalist 114: 503-504. Okarma, H. 1985. The trophic ecology of wolves and their role in ungulate communities of forest ecosystems in Europe. Acta Theriologica 40: 335-386. Okarma, H. 1992. The wolf – monograph of a species. [In Polish.] Bialowieza, Poland. Pp.168 Okarma, H. 1995. The trophic ecology of wolves and their predatory role in ungulate communities of forest ecosystems in Europe. Acta Theriologica 40 (4): 335-386. Peterson, R.O., Ciucci, P. 2003. The wolf as a Carnivore. In: Mech, L.D. and Boitani, L. (Eds.), Wolves. Behavior, Ecology, and Conservation. The University of Chicago Press. Chicago and London. Peterson, R.O., Woolington, J.D., Bailey, T.N. 1984. Wolves of the Kenai Peninsula, Alaska. Wildlife Monographs 88: 1-52. Peterson, R.O., Thomas, N.J., Thurber, J.M., Vucetich, J.A., Waite, T.A. 1998. Population limitation and the wolves of Isle Royale. Journal of Mammalogy 79: 828-841. Pimlott, D.H. 1967. Wolf predation and ungulate populations. American Zoologist 7: 267-278. Poulle, M.L. 1995. Le suivi des loups dans le Parc national du Mercantour. Le Bulletin mensuel de l’Office National de la Chasse 201: 36-41. Poulle, M. L., Carles, L., & Lequette, B. 1997. Significance of ungulates in the diet of recently settled wolves in the Mercantour mountains (South-eastern France). Revue Ecologie (Terre Vie) 52: 357-368. Promberger, C., Hofer, D. 1994. Der Wolf - Biologie und Management. Wildbiologische Gesellschaft München e.V., Ettal. Salvador, A., Abad, P.L. 1987. Food habits of a wolf population (Canis lupus) in Leòn province, Spain. Mammalia 51 (1): 45-52. Schwartz, C.C., Hundertmark, K.J., Spraker, T.H. 1992. An evaluation of selective bull moose harvest on the Kenai Peninsula, Alaska. Alces 28: 1-13.

61 White, P.J., Garrott, R.A. 2005. Yellowstone’s ungulates after wolves – expectations, realisations, and predictions. Biological Conservation 125: 141-152.

Livestock-damage and wolf-human conflict in the north-eastern Apennines, Tuscany, Italy.

- Manuscript -

Gazzola A., Capitani C., Mattioli L., Avanzinelli E., Apollonio M.

63 LIVESTOCK DAMAGE AND WOLF-HUMAN CONFLICT IN THE NORTHEASTERN APPENNINES, TUSCANY, ITALY.

GAZZOLA A. a, CAPITANI C. a, MATTIOLI L. b, AVANZINELLI E. a, APOLLONIO M. a a Department of Zoology , University of Sassari, Via Muroni 25, I-07100, Sassari, Italy. E-mail: [email protected] (AG, CC, EA, MA); b Provincial Administration of Arezzo, Piazza della Libertà 3, Arezzo, Italy (LM).

ABSTRACT Wolf–human conflict was investigated in the province of Arezzo by analysing data on damage to livestock by canids from 1998 to 2001, in relation to the assessed wolf distribution. Wolves (Canis lupus) were distributed over 22 municipalities (1504 km2) out of 39; with a density 2 estimated at 2.9 ± 0.7 wolves/100 km . The wolf population proved to be stable, although individual mortality was high. The area is characterised by a significantly higher percentage of forest cover and wild ungulate density, and lower human population and rail and roadway density, than in non-wolf areas. Livestock was uniformly distributed, but damage level and distribution were significantly higher in the wolf area. Only 6% of farms were considered to be chronically affected by predation, and they reported 38% of the total attacks and 37% of the total losses. Mass slaughter phenomena involved only sheep and goat farms, affecting 18% of the total. Thirty-five attacks (14% of the total attacks) reported 536 kills (44% of the total head killed in the whole province of Arezzo). During 1998-2001, compensation costs in the province averaged 86,863 Euros (range: 68,805 - 99,318 Euros). In the same period, no farmer requested prevention funding from the Tuscan Region. Coexistence of wolf and human populations can be reached by taking into account both conservation of environmental conditions and reduction of conflict, by adopting strategies to prevent damage to human activities by wolves.

Key words: Canis lupus, livestock predation, wolf-conservation, conflict with humans.

64 INTRODUCTION The maintenance of a viable wolf population represents a priority in most western European countries. Several studies have been addressed to evaluating the habitat variables influencing wolf distribution and expansion (Mech et al., 1988; Mladenoff and Sickley, 1998; Massolo and Meriggi, 1998; Jedrzejewski et al., 2004; Cayuela, 2004). Land use and trophic resources availability are reported to play a leading role, but also human disturbance-associated variables are important. In fact, wolf conservation efforts can prove insufficient, when the predator has a negative impact on the economy of local communities. Actually, the high-level of persistence of the human-wolf conflict is mainly due to livestock predation caused by wolves in most European countries (Zimen, 1978), and has been the reason for eradicating the wolf from most of its former range (Boitani, 1995). Lohr et al. (1996) believed that public acceptance, and not only ecological factors, played a fundamental role in maintenance of viable wolf populations in their original range. In fact, they stated, “public attitudes and subsequent human-caused mortality could ultimately determine the success or failure of reintroduction”. The positive trend of the Italian wolf populations and the recolonisation of areas where the predator had been absent for a long time produced a growth in the wolf-livestock conflict. The increase of livestock predation during the last decades has induced the regional government of Tuscany since 1982 to allow compensation for those farmers subjected to livestock losses. The veterinarian branch of the National Health Service (NHS) is responsible for certifying the claims. Livestock killed and injured by canid predators is refunded, but missing heads are not. Moreover, Tuscany provides financial incentives to farmers for preventive measures (fenced pastures and use of guard dogs). In the province of Arezzo, eastern Tuscany, data collected on wolf and ungulate populations since 1989 have allowed us to establish that wolves were widespread, with high density, in a vast region characterized both by a large percentage of forest cover and high wild ungulate density. Accordingly, by scat analysis, wild ungulates were found to represent the main item in wolf diet and a low use of livestock was generally observed (Mattioli et al., 1995; Capitani et al., 2004; Mattioli et al., 2004). Most of the wolf deaths found were caused by man, suggesting a high level of conflict with humans. From 1998 to 2001, we analysed the verified livestock losses caused by wolves and dogs in this area, to

65 quantify the predation impact and compensation cost. Moreover, we collected data on prevention funding provided in the last decade. The aims of the study were: i) to give detailed information on livestock damage linked to wolf presence, ii) to characterize management tools that could reduce human-wolf conflict, with particular reference to the damage-compensation system, and iii) to illustrate how proper small-scale management can play a role in a large-scale wolf conservation policy.

STUDY AREA The study area covered the whole 3226-km2 territory of the province of Arezzo (eastern Tuscany) with 323,000 inhabitants (mean density 100 inhabitants/km2) (Fig. 1). A rich community of wild ungulates is present in the province, composed of wild boar (Sus scrofa), roe deer (Capreolus capreolus), red deer (Cervus elaphus), fallow deer (Dama dama), and mouflon (Ovis orientalis musimon). Wild boar and roe deer are uniformly distributed throughout the study area, while red deer, fallow deer and mouflon occupy restricted and interspersed areas.

Fig. 1. The study area in the eastern part of Tuscany: province of Arezzo (shaded area).

Italy Tuscany

Arezzo Province

66 Livestock number is high: 16,223 head of cattle, 41,185 sheep and goats, and 2176 horses. Sheep and goat flocks vary from small ones of 2-100 head (82% of flocks) to those of 100- 2000 head (18%). Likewise, a large percentage of cattle farms (93%) are constituted by small herds (2-100 head). Depending on local topography and climatic conditions, livestock are grazed on pastures in mountainous areas from May to November and are moved to lower altitudes, or kept within enclosures, during the rest of the year. The northern portion of the province of Arezzo has been recolonised by wolves during the last 25 years, starting from the main Apennine ridge where they were always present, and is now characterised by a stable population. This portion is mainly mountainous, including part of the Apennine chain and other secondary chains. The altitude ranges between 300 and 1654 m a.s.l.. Forests, particularly deciduous ones, dominate the landscape, while urban settlements account for only 4.1% with a total resident population of over 84,000 inhabitants (ISTAT Population Census, 2001). The Casentinesi Forests National Park together with 10 provincial no-hunting zones creates a net of protected areas which includes the main ridges of the Apennine and of the other chains, covering 305 km2 in total. In the southern part, wolf presence was not verified during 1998-2001, but the temporary presence of dispersing or solitary wolves would not have been observed. This part includes the lower course of the Arno river and the Chiana Valley, the Chianti hills and some low mountains, the highest reaching 1081 m a.s.l.. Most human settlements and activities are concentrated in this part, and in 2001 the resident population consisted of around 239,000 inhabitants. The protected areas covered only 62 km2. Nowhere in the province of Arezzo has the presence of feral dogs ever been assessed; however, stray dogs and uncontrolled owned dogs are occasionally present.

67 METHODS Monitoring of wolves The presence of wolves has been regularly monitored since the beginning of the 1990’s in the Casentinesi Forests National Park, and since 1998 in a wider area including the main protected areas. Moreover, information was collected from hunters and provincial guards for the whole provincial territory. Indirect methods were mainly used, such as collection of scats, wolf howling, snow tracking, and molecular analyses. Number of packs and minimum pack size were estimated in summer by the wolf-howling method, following the “saturation” approach described by Harrington and Mech (1982) and by spectral analysis of recorded chorus responses, as described in Gazzola et al. (2002), and Apollonio et al. (2004). Packs were considered different when groups of adults and cubs were detected simultaneously in different valleys. In winter, minimum pack sizes were estimated by snow tracking (Apollonio et al., 2004; Mattioli et al., 2004). Assuming that a pack occupying an area during one year was the same as that in the same area in a subsequent year, we calculated persistence of packs as the probability of surviving for different consecutive years. Wolf diet was studied by analysis of scats collected along transects, following the standard protocol described in Mattioli et al. (1995) and Ciucci et al. (1996). All signs of wolf presence were localized on the map and geocoded using the GIS software, MAPINFO 5.0. The “wolf area” was defined as the group of municipalities whose surface was included in the wolf range, while the “non-wolf area” was all the other municipalities. Different variables were selected to verify their influence on the spatial distribution of wolves: percentage of forest cover and protected areas, and roadway and railway density (Regional Technical Map, made by the Regional Government of Tuscany), human population density (ISTAT Census of Population, 2001), wild ungulate abundance (Wildlife Service of Arezzo Provincial Government), livestock abundance, number of farms affected by predation, number of attacks and losses (Arezzo Veterinary Health Service), compensation cost (Regional Government of Tuscany).

68 Livestock damage and compensation and prevention fund The data analysed in this paper concern claims for losses attributed to canid predators from 1998 to 2001. Each claim referred to one validated attack only. Data were collected on the number, species and age class (if registered) of the domestic livestock killed or injured, and the number of farms and municipalities hit by predation. Level of predation was calculated as the number of attacks per year. Farms were considered to be chronically affected by predation if they recorded >2 attacks per year (Cozza et al., 1996). Mass slaughter was defined as those cases which involved >10 sheep killed per attack. Sheep and goats were considered as one category only, while the cattle category included cows, bulls and calves. Information on livestock production was obtained from the Arezzo Veterinary Health Service (ASL 8).

Statistical analysis The Mann-Whitney (M-W) test was used to evaluate differences in habitat variables and predation parameters between municipalities included in the wolf area and those included in the non-wolf area. The Kolmogorov-Smirnov (K-S) test was used to compare distribution of number of losses and attacks between sheep-and-goat and cattle farms, and to verify uniformity of temporal distribution of attacks. Concordance over years among distribution of predation events was tested by Kendall’s Concordance Coefficient (W). In order to verify the influence of the habitat variables on the level of damage, we developed multiple regression models including only those variables which proved not related by

Spearman’s Correlation Coefficient (rs). The final models included the dependent variable attack density (A) and the independent variables population density (P), livestock density (L), and wild ungulate density (W). Inference from models was made according to the Information-theoretic approach (Anderson et al. 2000, Anderson et al. 2001, Anderson and

Burnham 2002). Modified Akaike Information Criterion (AICc), differences with the minimum AICc (∆i), and Akaike weights (w I) were computed for each i-model to rank and scale models. Relative importance of predictor variables was measured by the sum of the Akaike weights of the models in which each variable appeared (Burnham and Anderson,

ˆ 2002). To quantify the effects of the predictors, parameters ( β ) were weighted and averaged

69 ∧ ∧ on the four models ( β ) and the unconditional sampling variance was computed (var β ) for each predictor (Burnham and Anderson, 2002). Statistical analyses were performed using SPSS 8.0 software, with significance always set at p

≤ 0.05.

RESULTS Wolf distribution and abundance During 1998-2001, we estimated that wolves were distributed over 22 municipalities (1504 km2), whereas the non-wolf area comprised 17 (1722 km2). In the period of study, this area was inhabited by a mean of 10 packs (range: 9-11). Based on census and observations, the number of wolves present each year was on average 44.7 ± 4.3. Therefore, the densities of 2 2 wolves and packs were 2.9 ± 0.7 wolves/100 km and 0.7 ± 0.1 packs/100 km , respectively. Wild ungulates were the main food items in wolf diet, whereas domestic ungulates ranged from 0.3% to 10.2% of mean percent volume (Mattioli et al., 1995; Capitani et al., 2004; Mattioli et al., 2004). The wolf area was characterized by a larger percentage of forest cover than the non-wolf area, 66% and 32%, respectively (Table 1). This difference was found significant (M-W test: U = 16.5, p = 0.001). Accordingly, wild ungulate density and biomass was significantly higher in the wolf area (M-W test: U = 100.5, p = 0.045; U = 84.0, p = 0.012, respectively). Moreover, the wolf-area contained a higher percentage of protected area than the non-wolf one, 21.2% and 3.7%, respectively. Human population density was significantly higher in the non-wolf area than in the wolf area, as indicated by the density of motorways and railways (Mann- Whitney test: U = 69, p = 0.001; and U = 88, p = 0.005, respectively). Livestock was homogeneously distributed throughout the province: density of farms was found similar in the two areas (Mann-Whitney test: U = 180.0; p = 0.843). A comparable result was observed for livestock density (Mann-Whitney test: U = 179.5; p = 0.832).

70 Table 1. Comparison of habitat variables municipalities included in the wolf area or non-wolf area in Arezzo province, Italy, in 1998-2001. Mann-Whitney p values are given for each variable.

Wolf area Non-wolf area Statistical significance Variable Mean ± SE (min-max) Mean ± SE (min-max) of difference (p)

Percentage protected area 18.3 ± 2.9 (0.1-51.9) 4.7 ± 1.3 (0.0-18.9) 0.000

Percentage forest cover 66 ± 2.8 (38.4-87.4) 32.0 ± 3.9 (0.2-55.3) 0.001

Percentage urban area 4.2 ± 0.4 (0.75-8.1) 8.2 ± 0.9 (1.7-18.3) 0.000

Wild ungulate density 38.5 ± 2.7 (13.0-62.2) 26.3 ± 3.7 (0.0-54.2) 0.014 (ind./km2) Wild ungulate biomass 1091 ± 80 (331-1838) 704 ± 103 (0-1723) 0.004 (kg/km2) Human population 66 ± 13 (10-295) 166 ± 44 (45-797) 0.001 (inhabitants/km2) Motor and railway density 3.0 ± 0.2 (2.0-5.9) 4.0 ± 0.34 (2.5-7.9) 0.004 (km/km2)

Farm (farms/100 km2) 43.3 ± 4.1 (3.9-78.4) 50.1 ± 7.7 (16.4-135.2) ns

Livestock (ind./100 km2) 1903 ± 320 (255-6925) 1804 ± 220 (528-4265) ns

Wolf-human conflict During 1988-2001, 40 wolves were found dead in Arezzo Province. Mortality causes were discovered in 67% of cases: only one death was due to a natural agent (disease), while all the others were attributed to humans. Of the latter, 78% were caused by illegal actions (shooting, poison, and snares). In all cases but four, the age of the individuals could be determined: percentage of individuals over and under 1 year of age respectively was the same (50%). Nevertheless, a high level of persistence was observed: for all the studied packs, the

71 probability of persistence was 90% for two consecutive years, and 86% and 81% for three and four years, respectively.

Livestock damages During 1998-2001, the National Health Service (NHS) received 464 claims from 179 farms attacked reporting 1462 head killed and 78 injured by canid predators. Sheep-and-goat stock proved most affected by predation (Table 2). In fact, although a similar number of attacks involved both sheep/goats and cattle, most of the losses were in sheep-and-goat stock. Only 16 attacks involved horse farms, that reported 20 losses. Among sheep and goats, adults were the age class most affected by predation (80%); 85% of cattle losses were calves; while 50% of horse predations were adults. During 1998-2001, in both sheep-and-goats and cattle farms, a high percentage of farms was subjected to a low level of predation, while a few farms were chronically affected (Table 2).

Table 2. Wolf damage to sheep-and-goat and cattle farms

Livestock damages Sheep and goats Cattle

Number of attacks 262 202

Number of animals killed 1227 235

Number of farms affected 107 72

Annual losses as percentage of stock available 0.7 0.4

Percentage of farms affected 3.6 2.8

72 Table 3. Comparison of depredation level and mass slaughter phenomena between sheep-and- goat and cattle stocks.

Percentage of attacks or losses Level of predation on farms hit by wolf

Sheep and goats Cattle

1-3 attacks 87 81

≥ 9 attacks 4 6

1-3 losses 42 79

≥ 10 losses 30 6

In fact, the number of attacks per farm, did not vary between the two categories of farms (K-S test: Z = 0.447; n1, 2 = 10; p = 0.988), and neither did the number of losses per farm (K-S test:

Z = 1.118; n1, 2 = 10; p = 0.164) (Fig. 2). Most attacks involved a small number of heads. Only in sheep-and-goat farms were a few attacks responsible for mass slaughter phenomena (Table 3). The number of losses per attack varied between sheep-and-goat and cattle farms and the difference was significant (K-S test: Z = 1.565; n1,2= 10; p= 0.015) (Fig. 2). Monthly distribution of predation events on sheep-and-goat attacks (Fig. 3) varied among years (Kendall’s Concordance Coefficient: W = 0.431; p = 0.062). On the contrary, those on cattle did not (Kendall’s Concordance Coefficient: W = 0.820; p = 0.000) (Fig. 3). Predation events were concentrated in summer and in early autumn; nevertheless, monthly attack distribution did not vary during the year either in sheep-and-goats (K-S test: Z = 0.896; n = 12; p = 0.398) or in cattle (K-S test: Z = 0.896; n = 12; p = 0.398). Overall sheep-and-goats and cattle showed the same pattern of monthly attack distribution (K-S test: Z = 1.021; n1, 2 = 12; p = 0.249) (Fig. 3).

73 Fig. 2. Frequency distributions of predation events in the province of Arezzo during 1998- 2001: number of attacks; number of losses; number of losses per attack.

Attacks per farm

Losses per farm affected

Losses per attack

74 Fig. 3. Monthly distribution of attacks on sheep-and-goats and cattle over the years, and during the whole study period (1998-2001).

Sheep and goats

Cattle

75 Mass slaughter phenomena Mass slaughter phenomena involved only sheep-and-goat farms. During 1998-2001, 19 farms were affected by these phenomena (12% of all), and 35 attacks reported 536 kills. The number of head killed per attack ranged from 10 to 54 (median = 15; mean = 17.9 SD = 9.4). Monthly distribution of these attacks was uniform (K-S test: Z = 0.866; n = 12; p = 0.441). Most farms (95%) reported low level of recurrence of attacks (1-3 attacks), while only one farm reported 8 attacks. Most farms (68%) were subjected to 10-30 losses, but one of them was subjected to 107 losses.

Factors affecting damage level Multiple regression analysis was employed to explain the variations of attack density among municipalities on the basis of the selected habitat variables. Since four models were statistically significant, we tested them with Akaike’s Information Criterion (AICc) (see Methods). The best model (Akaike’s weight ωi = 0.497) included two variables: human population density (P), livestock density (L) (Table 4). These two factors accounted for 43% of the observed variation in the log-transformed attack density: A = 4.123 − 0.769P + 0.420L, R2 = 0.426, P = 0.000, n = 39; A = Ln (attack density + 1).

Table 4. Multi-model inference based on regression models of population density (P), livestock density (L), and wild ungulate density (W) effects. R2 — coefficient of linear or multiple regression in each model, AICc — Akaike’s Information Criterion, ωi — Akaike’s weight.

2 Model k R AICc ωi

A = f(P+L) 4 0.426 24.37 0.497 A = f(P+W) 4 0.362 24.78 0.262 A = f(L+W) 4 0.325 25.00 0.225 A = f(P+L+W) 5 0.474 36.03 0.016

76 Comparison of damage in wolf and non-wolf areas Damage distribution was mainly concentrated in the municipalities where the presence of wolves was assessed (Fig. 4). The number of affected farms was higher in the wolf-area than 2 2 in the non-wolf one (10.6 ± 13.0 farms/100 km vs. 1.7 ± 3.5 farms/100 km ) and this difference was significant (Mann-Whitney test: U = 84.5; p = 0.003). Likewise, the percentage of affected farms varied significantly between the wolf- area (23%) and the non-wolf one (4%) (Mann-Whitney test: U = 81.0; p = 0.002). A total of 86% of livestock attacks (n = 397), corresponding to 1033 kills, was located in the wolf area, while the rest (n = 67) occurred in the non-wolf one, involving 429 heads.

The number of attacks varied significantly between the wolf area (26.4 ± 34.5 attacks/100 2 2 km ) and non-wolf one (4.9 ± 12.0 attacks/100 km ) (Mann-Whitney test: U = 85.0; p = 0.003). Similar results were obtained comparing the amount of livestock losses between the 2 2 wolf-area (68.7 ± 85.6 losses/100 km ) and the non-wolf one (24.9 ± 77.0 losses/100 km ) (Mann-Whitney test: U = 111; p = 0.029).

Fig. 4. Spatial distribution of wolf damage over municipalities. Shaded area represents the wolf area.

Percentage of farms affected Number of wolf attacks

77 The mean number of kills per attack was 2.7 ± 2.0 in the wolf area, and 6.3 ± 5.2 in the non- wolf one. Considering mass slaughter (M-S) phenomena. attacks represented 11.1% of total attacks in the wolf area vs. 20.3% in the non-wolf one. The number of M-S attacks did not 2 vary significantly between the wolf area (1.5 ± 2.4 head/100 km ) and the non-wolf one (0.8 2 ± 2.2 losses/100 km ) (Mann-Whitney test: U = 170.0; p = 0.514). M-S losses represented 38.3% of the total in the wolf area vs. 53.9% in the non-wolf one. The amount, 11, of M-S 2 losses did not vary significantly between the wolf area (20.4 ± 32.4 head/100 km ) and the 2 non-wolf one (13.3 ± 42.9 losses/100 km ) (Mann-Whitney test: U = 179.0; p = 0.768). Neither sheep-and-goat nor cattle farms subjected to predation, classified by municipality, were found to be correlated with farm density within the whole province (sheep and goats: rs =

0.037; n = 39; p = 0.824; cattle: rs = 0.293; n = 39; p = 0.07). On the contrary, within the wolf area, a positive correlation was found only for cattle farms (sheep and goats: rs = 0.103; n =

22; p = 0.647; cattle: rs = 0.468; n = 22; p = 0.028).

Levels of predation, compensation costs and prevention funding During 1998-2001, in the province of Arezzo, among the farms that were hit by predation, most (83%) were only slightly affected and reported an average of less than one attack per year. Only 6% of farms were considered chronically affected by predation, and these reported 38% of total attacks and 37% of total losses. The overall economic impact of predation related to livestock production was low. Annual livestock losses accounted for 0.7% of the provincial sheep-and-goat stock and for 0.4% of cattle stock (Table 3). The annual percentage of farms hit by predation was low both for farms that breed cattle (mean 2.8%; wolf-area: 5.2%, non- wolf area: 0.2%) and sheep and goats (mean 3.6%; wolf area: 6.4%, non-wolf area: 1.7%). During 1998-2001, compensation costs varied from a minimum value of 68,805 Euros to a maximum 99,318 (mean 86,863 Euros). Most of the total compensation cost was localized in the wolf area (85%). In the same period, no farmer applied for prevention funding from the Tuscany Region. During 1998-2000, the level of agriculture damage caused by other wildlife (mainly wild boar) was very high in the province of Arezzo. Compensation costs varied from a minimum value of 566,377 Euros to a maximum of 655,043 (mean 609,915 Euros). This mean value is seven times higher than the mean annual value of wolf compensation.

78 DISCUSSION In the last centuries, in western Europe and North America, wolf persecution and land exploitation led to reduction and fragmentation of the wolf’s original range, but natural resources conservation policies allowed populations to recover and expand again in some portions of these countries (Mech, 1995; Boitani, 2000). The current wolf distribution seems to be linked to a set of environmental and socio-economic variables. Jedrezjewski et al. (2004) pointed out that areas with low human disturbance, high levels of forest cover, low density of human settlements, and low density of public roads are selected by wolves in northern Poland. Also in our study, the wolf area was characterized by a larger percentage of forest cover and a higher percentage of protected area than the non-wolf one, whereas human population density and roadway and railway density were significantly lower. Moreover, in accordance with Massolo and Meriggi’s (1998) model of wolf spatial distribution, we found that wild ungulate density was a very important habitat index of stable wolf presence, while livestock abundance was not. Nevertheless, livestock losses are not necessarily a consequence of scarce availability of wild prey, and in our study area the high degree of wild ungulate density did not prevent wolf-livestock conflict. Although ecological features are important for the maintenance of viable populations of wolves in their original range, public acceptance is also decisive (Lohr et al., 1996). Wolf management is strongly conditioned by public opinion and human tolerance to wolf presence (Blanco et al., 1992; Treves and Karanth, 2003). The wolf wins support among conservationists and many urban people, but is considered a nuisance in rural areas, where people often consider this predator a pest and a threat to livestock and wildlife (Breitenmoser, 1998). Therefore, to conserve large carnivore species successfully, it is fundamental, on the one hand, to improve knowledge and acceptance of predators among rural people, mainly livestock owners and hunters (Breitenmoser, 1998; Boitani, 2000) and, on the other, to analyse the carnivore-livestock conflict in order to limit it (Linnell et al., 1996; Sagør et al., 1997). As the regional government of Tuscany refunds damage caused to livestock by both wolves and dogs, claims did not provide specific information on which species was responsible for livestock losses. Nevertheless, the distribution of livestock damage in the province of Arezzo was concentrated mainly in municipalities where the presence of wolves was assessed.

79 Depredation events were negatively influenced by population density, while the greater risk of damage to livestock in rural areas could be connected with wolf presence in the area. In several studies, a relationship was observed between wolf distribution and livestock losses (Fritts and Mech, 1981; Bjorge and Gunson, 1983; Tompa, 1983) and between wolf abundance and levels of conflict (Jedrzejewski et al., 2004). Nevertheless, considering the high density of wolves in our study area, the impact of predation was low both for farms affected (<4%) and livestock losses (<0.8%). Numerous works have reported that predation events involved only a small percentage of the total number of farms and that their economic impact related to livestock production was irrelevant (Fritts et al., 1992; Fico et al., 1993; Ciucci and Boitani, 1998; Mech, 1998). In fact, compared to compensation of damages (around 400,000 Euros in four years), the Tuscan Regional Government planned to invest in husbandry activities from 1,600,000 to 3,000,000 Euros for various measures, over five years (Tuscany Region Husbandry Development Programme, 2004). Moreover, the mean annual compensation cost represented about 0.8% of the total value of sheep-and-goat and cattle stock in Arezzo (about 10,800,000 Euros in the year 2000). But, although the damage is irrelevant to the livestock industry, it may be significant for some farmers. Several studies revealed that predation events on livestock are particularly concentrated in a few farms that reported a high level of conflict (Robel et al., 1981; O’Neill, 1988; Fritts et al., 1992). In the Abruzzo region, Cozza et al. (1996) found that 23 farmers were chronically affected and accounted for 31% of the claims. A similar level of predation was found in Tuscany, where 6% of the farms hit reported 32% of the sheep lost (Ciucci and Boitani, 1998). Accordingly, in our study, we monitored 7% of the farmers chronically affected by predation, who reported 42% of attacks and 37% of losses. In the north of Portugal, Vos (2000) found that if the losses did not exceed a certain percentage, the shepherds did not persecute the wolves. Therefore, it appears to be a critical point to act on chronicle farms in order to reduce both the overall and the individual economical losses, as well as the individual hostile reaction to predation risk pressure. A similar level of conflict is generally produced by mass slaughter phenomena. In fact, although this event is reported to occur rarely (Boitani, 1982; Gunson, 1983; Ciucci and Boitani, 1998), it is responsible for a high percentage of total losses. In a study performed in Tuscany, Ciucci and Boitani (1998) reported that mass slaughters (>20

80 head/attack) were only 2.3% of the predation events, but were responsible for 19% of the sheep losses during 1991-1995. In our study, 35 attacks alone involved 44% of total sheep and goats killed in the whole province of Arezzo during 1998-2001. Consequently, we think that another main target for the mitigation of conflict between wolf and farmer is to investigate the origins of mass slaughter phenomena, in order to reduce their number. The origin of chronicle farms could be mainly related to “site effect”, linked to the long-term habit of predators to come back to the same site, as found for the lynx in the Jura Mountains by Stahl et al. (2001). As regards mass slaughter phenomena, one important element to be taken into consideration is husbandry practices. In fact, we found that this phenomena was more frequent in the area without stable presence of wolves, suggesting a lower level of attentiveness to the predator risk factor. Compensation programs are one management tool to limit wolf-livestock conflicts and to reduce hostility in the rural community, but these goals are often unattainable. In some situations, compensation to farmers does not equal the real cost of their losses, while, in others, farmers receive the tax refunds promised by the local government more than 3 or 4 years after their applications (Vos, 2000). Moreover, compensation programmes could induce dependence on an economic incentive; this was recently targeted by the European Community, so that member states were prevented from continuing with this policy. In Arezzo, compensation to farmers is among the highest paid at the national level (Ciucci et al., 1997), and is quickly delivered by the regional government of Tuscany to farmers (3-6 months). During 1998-2001, despite the province of Arezzo paying a high mean annual compensation, no farmer applied for any prevention funding from the Tuscan Region, and wolf mortality was clearly human induced. Many studies showed that free ranging, unguarded herds were more subject to livestock predation (Blanco et al., 1990; Ciucci and Boitani 1998; Vos, 2000). In our study, repetition of attacks on a few farms, independently of farm density, suggests that the phenomenon could be linked to accessibility of domestic animals. In this context, compensation for livestock losses should not be viewed as the sole management tool to reduce the wolf-livestock conflict, but the adoption of preventive husbandry practices should also be considered, especially in areas of high level of conflict, should be encouraged.

81 CONCLUSIONS AND GUIDELINES In western European countries, a high level of natural resource exploitation has been producing vast changes in landscape and the fragmentation of habitat. In this situation, the coexistence of wolf and human populations can be achieved by decreasing the level of conflict. Italy represents a valid example of a territory where high-density population and widespread human settlement and activity limit the dimension of natural areas and increase human disturbance. In this context, the maintenance of a viable wolf population seems to be linked to the presence of a network of small areas with high productivity of natural prey and the availability of protected refuges. Our study shows that a high-density and expanding wolf population can coexist with human activities, when an adequate management system is adopted, which keeps the level of damage low. Our management system of livestock damage was successful for conservation of wolves, but failed to improve husbandry practices. In fact, even if the law required that reimbursement be conditioned by the adoption of adequate prevention measures (Tuscany Regional Law No. 72/94), this restriction was not always respected. For this reason, in the long term, the weak points in the compensation program proved to be the sustainability of its cost by the public administration and the dependency of farmers on reimbursement. In many European countries, this problem was solved by including the culling of predators as a management tool. In northern Spain, wolf harvesting is not envisaged, but hunting drives are realized when damage to livestock exceeds the threshold of what people will accept (Carranza, 2004). In France, instead, the national plan approved for 2004-2008 aimed to reduce the demographic expansion of the wolf population (55 individuals), and allowed the killing of about 10% of the population per year (Wolf Action Plan in France, 2004). In Switzerland, culling of “problematic” wolves was allowed after a great number of repeated attacks. Similar plans were adopted in Scandinavia, especially in the domestic reindeer area. Fully legal protection, instead, was guaranteed for wolves in Germany, where they were recovered after being extinct since the 19th century, and non-lethal measures are being taken to help shepherds deal with the threat. Considering the legal status of the wolf (Bern Convention, 1982, Habitats Directive 92/43/CEE), its recent reappearance in Alpine and north European regions, and the fact that

82 the mortality rate due to illegal killing (also related to conflict with hunters) or other human factors (car or train accidents) is still quite high in many southern European areas, the adoption of planned culling of this predator should be seen as an exceptional measure for managing the wolf-livestock conflict. We believe that the limitations of compensatory systems could be overcome in a different way. For instance, the Tuscany Regional Government recently proposed a new law, no longer based on damage compensation, but on the payment of insurance policies for farmers who had adopted serious prevention measures in a given time. Thus, in general, we suggest that a limitation to the wolf-human conflict could be obtained by improving or achieving the following objectives: 1. increase of wild prey stocks, where their density is low, together with the adoption of prevention measures, so that livestock is not more readily available than wild stock; 2. identification of farms suffering a high level of depredation, followed by analysis of causes and the adoption of specific damage reduction strategies; 3. intensive information campaigns and involvement of the local population both in areas where wolves are present and, in particular, in those where they have been absent for a long time and it is thought they will be recolonising in the coming future (i.e., eastern Alps); 4. substitution of compensation programs based on reimbursement of damage with a system based on insurance. According to European Union economical policy, local governments do not directly provide funds to shepherds, but pay insurance for farms that fall within the wolf distribution area. The insurance agency reimburses damage only if adequate preventive measures are adopted by shepherds, and meanwhile shepherds are induced to invest in prevention to keep the insurance premium low. In fact, the premium is covered by local Government only below a certain threshold; 5. control of free ranging dogs through specific legislation.

83 ACKNOWLEDGEMENTS

We wish to thank Bogumiła Jędrzejewska, (Mammals Research Institute, Polish Academy of Sciences, Białowieża, Poland) for her precious comments on the early draft of this paper, and Paolo Lamberti and Enrico Merli for their help with statistic analyses. Furthermore, we are grateful to the Arezzo Veterinary Health Service (ASL8) and the Department of Agriculture and Forests of the Region of Tuscany for essential cooperation, as well as to the Provincial Administration of Arezzo for logistic and financial support. We are also grateful to the ex- ASFD Administration of Pratovecchio for providing facilities during the study. Ivo Bertelli, Lorenza Mauri, Massimo Scandura, and Guido Crudele helped substantially in the collection of field data on wolf distribution. We would also like to thank James Burge for linguistic revision.

REFERENCES Anderson DR, Burnham KP, Thompson WL (2000) Null hypothesis testing: problems, prevalence, and an alternative. Journal of Wildlife Management 64: 912-923 Anderson DR, Link WA, Johnson DH, Burnham KP (2001) Suggestions for presenting the results of data analyses. Journal of Wildlife Management 65: 373-378 Anderson DR, Burnham KP (2002) Avoiding pitfalls when using Information-theoretic methods. Journal of Wildlife Management 66: 912-918

Apollonio, M., Mattioli, L., Scandura, M., Mauri, L., Gazzola, A., Avanzinelli, E. 2004. Wolves in the Casentinesi Forests: insights for wolf conservation in Italy from a protected area with a rich wild prey community. Biological Conservation 120, 249-260. Bjorge, R.R., Gunson, J.R. 1983. Wolf predation of cattle on the Simonette River pastures in west-central Alberta. In: Carbin, L.N. (Ed), Wolves in Canada and Alaska. Canadian Wildlife Service, Report Series 45, pp. 106-111. Blanco, J.C., Cuesta, L., Reig, S. 1990. El lobo en España: una vision global. In: Blanco, J.C., Reig, S., and Cuesta, L., Blanco, J.C., Reig, S., and Cuesta, L. (Eds), El lobo (Canis lupus)

84 en España. Situacion, problematica y apuntes sobre su ecologia. Ministero de Agricoltura, Pesca y Alimentacion, Icona, Madrid, Spain, pp. 45-56. Blanco, J.C., Reig, S., and Cuesta, L. 1992. Distribution, status and conservation problems of the wolf Canis lupus in Spain. Biological Conservation 60, 73-80. Boitani, L. 1982. Wolf management in intensively used areas of Italy. In: Harrington, F.H., Paquet, P.C. (Eds), Wolves of the world. Perspectives of behavior, ecology and conservation. Noyes Publications, Park Ridge, New Jersey, pp. 158-172. Boitani, L. 1995. Ecological and cultural diversities in the evolution of wolf-human relationships. In: Carbyn, L.N., Fritts, H.S., Seip, D.R. (Eds), Ecology and conservation of wolves in changing world: proceedings of the second North American symposium on wolves. Canadian Circumpolar Institute, Occasional Publication Number 35, pp. 3-11. Boitani, L. 2000. Action plan for the conservation of wolves in Europe (Canis lupus). Nature and Environment Series 113. Council of Europe, Strasbourg, France. Breitenmoser, U. 1998. Large predators in the Alps: the fall and rise of man’s competitors. Biological Conservation 83 (3), 279-289. Burnahm, K.P., Anderson, D.R. 2002. Model selection and multi-model inference. A practical information-theoretic approach. 2nd Edition. Springer-Verlag. New York. Capitani, C., Bertelli, I., Varuzza, P., Scandura, M., Apollonio, M. 2004. A comparative analysis of wolf (Canis lupus) diet in three different Italian ecosystems. Mammalian Biology 69, 1-10. Carranza, J. 2004. Wild ungulates in Spain. Proceedings of International School of Ethology. 21st Course. Ungulate management in Europe. 12-17 November 2004. Erice, Sicily, Italy. Cayuela, L. 2004. Habitat evaluation for the Iberian wolf Canis lupus in Picos de Europa National Park, Spain. Applied Geography 24, 199-215. Ciucci, P., Boitani, L., Raganella-Pelliccioni, E., Rocco, M., Guy, I. 1996. A comparison of scat-analysis methods to assess the diet of the wolf. Wilfdlife Biology 2, 267-278. Ciucci, P., Raganella-Pelliccioni, E., Boitani, L. 1997. Impatto del lupo sul patrimonio zootecnico: entità dei danni e quadro legislativo. Report to Ministero dell’Ambiente, Servizio Conservazione della Natura, Roma, Italy [in Italian].

85 Ciucci, P., Boitani, L. 1998. Wolf and dog depredation on livestock in central Italy. Wildlife Society Bulletin 26 (3), 504-514. Cozza, K., Fico, R., Battistini, L., and Rogers, E. 1996. The damage conservation interface illustrated by predation on domestic livestock in central Italy. Biological Conservation 78, 329-336. Fico, R., Morosetti, G., Giovannini, A. 1993. The impact of predators on livestock in the Abruzzo region of Italy. Revue scientifique et technique, Office International des Epizooties 12, 39-50. Fritts, S.H., Mech, L.D. 1981. Dynamics, movements, and feeding ecology of a newly protected wolf population in north-western Minnesota. Wildlife Monograph 80, 1-79. Fritts, S.H., Paul, W.J., Mech, L.D., Scott, D.P. 1992. Trends and management of wolf- livestock conflicts in Minnesota. U.S. Fish and Wildlife Service, Resource Publication 181. Gazzola, A., Avanzinelli, E., Mauri, L., Scandura, M., Apollonio, M. 2002. Temporal changes of howling in south European wolf packs. Italian Journal of Zoology 69, 157-161. Gunson, J.R. 1983. Wolf predation on livestock in Western Canada. In: Carbin, L.N. (Ed), Wolves in Canada and Alaska. Canadian Wildlife Service, Report Series 45, pp. 102-105. Harrington, F., Mech, L.D. 1982. An analysis of howling response parameters useful for wolf pack censusing. Journal of Wildlife Management 46, 686-693. ISTAT 2001. 14° censimento della popolazione e delle abitazioni G.U. Serie Generale n.81 del 7 Aprile 2003, Supplemento Ordinario n.54. Jedrzejweski W., Niedziałkowska M., Nowak S., Jedrzejweska B. 2004. Habitat variables associated with wolf (Canis lupus) distribution and abundance in northern Poland. Diversity and Distributions 10, 225-233. Linnell, J.D.C., Smith M.E., Odden, J., Kaczensky, P., Swenson, J.E. 1996. Strategies for the reduction of carnivore-livestock conflicts: a review. Norwegian Institute for Nature Research Oppdragsmelding 443, 1-118. Lohr, C., Ballard, W.B., Bath, A. 1996. Attitudes toward gray wolf reintroduction to New Brunswick. Wildlife Society Bulletin 24 (3), 414-420.

86 Massolo, A., Meriggi, A. 1998. Factors affecting habitat occupancy by wolves in northern Appenines (northern Italy): a model of habitat suitability. Ecography 21, 97-107. Mattioli, L., Apollonio, M., Mazzarone, V., Centofanti, E. 1995. Wolf food habitus and wild ungulate availability in the Foreste Casentinesi National Park, Italy. Acta Theriologica 40, 387-402. Mattioli L., Capitani C., Avanzinelli E., Bertelli I., Gazzola A., Apollonio M. 2004. Predation by wolves (Canis lupus) on roe deer (Capreolus capreolus) in north-eastern Apennines, Italy. Journal of Zoology (London) 264, 249-258. Mech, L.D., Fritts, S.H., Radde, G.L., Paul, W.J. 1988. Wolf distribution and road density in Minnesota. Wildlife Society Bulletin 16, 85-87. Mech, L.D. 1995. The challenge and opportunity of recovering wolf population. Conservation Biology 9, 270-278. Mech, L.D. 1998. Estimated costs of maintaining a recovered wolf population in agricultural regions of Minnesota. Wildlife Society Bullettin 26, 817-822. Mladenoff, D.J., Sickley, T.A. 1998. Assessing potential grey wolf restoration in the north eastern United States: a spatial prediction of favourable habitat and potential population levels. Journal of Wildlife Management 62, 1-10. O’Neill, B.B. 1988. The law of wolves. Environmental Law 18, 227-240. Robel, J.R., Dayton, A.D., Henderson, F.R., Meduna, R.L., Spacth, C.W. 1981. Relationship between husbandry methods and sheep losses to canine predators. Journal of Wildlife Management 45, 894-910. Sagør, J.T., Swenson, J.E., Røskaf, E. 1997. Compatibility of brown bear (Ursus arctos) and free-ranging sheep in Norway. Biological Conservation 81, 91-95. Stahl, P., Vandel, J.M., Herrenschmidt, V., Migot, P. 2001. The effect of removing lynx in reducing attacks on sheep in the French Jura Mountains. Biological Conservation, 101, 15- 22. Tompa, F.S. 1983. Problem of wolf management in British Columbia: conflict and program evaluation. In: Carbin, L.N. (Ed), Wolves in Canada and Alaska. Canadian Wildlife Service, Report Series 45, pp. 112-119.

87 Treves, A., Karanth, K.U. 2003. Human-carnivore conflict and perspectives on carnivore management worldwide. Conservation Biology 17 (6), 1491-1499. Vos, J. 2000. Food habits and livestock depredation of two Iberian wolf packs (Canis lupus signatus) in the north of Portugal. Journal of Zoology (London) 251, 457-462. Zimen, E. 1978. Der Wolf: Mythos

Wolves in the Casentinesi Forests: insight for wolf conservation in Italy from a protected area with a rich wild prey community.

- Biological Conservation -

Apollonio M., Mattioli L., Scandura M., Mauri L., Gazzola A., Avanzinelli E.

100

101

Prey selection by wolves from a multi-species ungulate community in the Casentinesi Forests, Italy.

- Manuscript -

Mattioli L., Apollonio M., Gazzola A., Avanzinelli E., Bertelli I., Capitani C.

102 PREY SELECTION BY WOLVES FROM A MULTISPECIES UNGULATE COMMMUNITY IN THE CASENTINESI FORESTS, ITALY

Mattioli L.*, Apollonio M.**, Gazzola A.**, Avanzinelli E.**, Bertelli I.**, Capitani C.**

*Provincial Administration of Arezzo, Arezzo, Italy; **Department of Zoology, University of Sassari, Sassari, Italy

ABSTRACT Selection of prey species and age/weight classes by a wolf pack was investigated in a multiple prey community in the Casentinesi Forest - central Apennines, Italy, from May 1988 to April 2000. Mean density of wild ungulates averaged 34 heads/km2, corresponding to a mean total biomass of 924 kg/ km2. The analysis of 1862 scats revealed that wild boar (Sus scrofa) was the main prey, followed by roe deer (Capreolus capreolus), red deer (Cervus elaphus), and fallow deer (Dama dama), with a Mean Percent Volume of 58.7%, 19.1%, 8.7%, and 1.9%, respectively. With respect to their availability in the ungulate community, obtained from census data, wolves selected wild boar positively and roe deer and red deer negatively. No functional response of wolves to density variation of the three main species was found. Wolves tended to select young (<1-year-old individuals) rather than adults in all ungulate species, but with different seasonal patterns. Wild boar piglets were negatively selected in the first months of life (March-April), whereas from July they were increasingly preferred as their weight grew. Roe deer fawns, instead, were more positively selected from birth until December than from January to April. Mean weight of wild preys was 20.5 kg for a mean pack size of 4.2 wolves. Patterns of use and selection of prey species and age classes suggest that wolf food habits are determined more by vulnerability, particularly size, than by abundance. The small size of the Italian wolf could be one of the causes of this.

Key words: Canis lupus, prey dimension, age selection

103 INTRODUCTION Wolf trophic ecology and its relationships with wild ungulate preys have been widely studied in several areas of North America (for a review: Peterson and Ciucci 2003, Mech and Peterson 2003) and in some European ones, particularly in the east-central part (for a review: Okarma 1995). However, most North American and eastern-central European forest ecosystems are markedly different from the Mediterranean ones, being characterized by lower forest productivity and, as a consequence, lower wild ungulate biomass. In many cases, the studies concerned predator-prey systems where only one or two ungulate prey species were present (Fuller et al. 2003). In this context, the Italian ecological situation is particularly interesting because of: 1) the recent secondary creation of complex zoocenosis consisting in the wolf and several ungulate species, 2) the presence of higher prey density and biomass than those in more northern latitudes, 3) the presence of a prey species (wild boar) with high productivity, 4) the wider distribution of human presence and activities. In the last 20 years, as a consequence of wolf expansion and recovery in some European areas (Promberger and Schröder 1993), interest of researchers in this predator has increased. Most of the studies concerning wolf trophic ecology agree in asserting that wild ungulates, when available, are the main preys (e.g.: Meriggi and Lovari 1996). Nevertheless, there are a few cases in which the wolf diet is analysed in relation to the actual trophic resources availability of the study area. Among these, the studies conducted in the Bialowieza Primeval Forest (BPF), Poland and Belaru,s in the last 50 years are the most important (Gavrin and Donaurov 1954, Bunevich 1988, Jedrzejewski et al. 1992, 2000 and 2002, Okarma 1995). The data collected in BPF and in the other eastern European areas show that wolves tend to select red deer, where present, as the main prey, while the use of wild boars is generally lower than its availability. From 1988 to 2000, we conducted studies on the trophic ecology of the wolf in a multispecies and highly productive ungulate community, constituted by roe deer, wild boar, red deer, and fallow deer, in the Casentinesi Forests (CF), central Apennines. Preliminary results (Mattioli et al. 1995) showed that wolves mostly preyed upon wild boar, exhibiting feeding habits different from those in eastern-central Europe, but consistent with other studies from neighbouring Italian areas. In the present study, we analysed a 12-year dataset on the feeding ecology of a CF wolf pack, to establish: 1) the existence of selective predation among species,

104 2) the presence of selection among age and weight classes of preys, 3) the relationship between the pattern of use and selection of preys and seasons.

METHODS Study area The study area (CF) is ca. 130 km2, lying between 400 and 1520 m a.s.l.. The climate is continental and the average mean annual temperature is 5 °C, snow covers the ground for 94 days a year at 1100 m a.s.l. The wild ungulate community is composed of four species: wild boar, roe deer, red deer and fallow deer. This area can be divided into two different portions. One (A), above 700 m, comprises the Casentinesi Forest, and is dominated by Fagus-Abies and Quercus-Tilia-Acer forests (89%); here, hunting is not allowed. In the other, (B), located below 700 m, forests are mainly constituted by Quercus pubescens and cover only 48% of the area, being interspersed with pastures and cultivated fields. Cervids and wild boar are hunted there. Farms are distributed exclusively in the lower part of the area: about 960 sheep and a few cattle and horses were present in 1990.

Estimating abundance, population structure and biomass of wolf and wild ungulates In CF, the presence and the number of wolves, both during the post-breeding period and in winter, were annually verified by wolf-howling and snow-tracking, respectively.

The wolf-howling census was made following the “saturation” approach described by Harrington and Mech (1982), with some adaptations to our study area. Pack size in summer was estimated by spectral analysis of chorus responses recorded during the census. Snow- tracking sessions allowed us to estimate pack size during winter and wolf movements in the study area. The protocol and instruments used during census sessions were described in detail in Gazzola et al. (2002), and Apollonio et al. (2004). The density of wild ungulates was estimated by different census methods, described in detail in Mattioli et al. (2004), for the ISA area. The roe deer population was estimated at the end of winter by drive census, performed following the protocol described in Cemagref (1984) and

105 Okarma et al. (1995). Eleven sample areas covering a total of 590 ha (6.5% of the forested area) were used, red deer being censused by counting roaring males during the rut and then extrapolating density from the percentage of stags in the population (Mazzarone et al. 1991). For fallow deer, a mean value for the whole study period was calculated, extrapolated from data obtained by both drive census and observation from vantage points in portion B in 1998 and 1999. For these species, post-parturition density taking into account the mean relative frequency of adult females in the population and a mean fertility value were calculated. These values were derived from a local sample in the case of red deer and roe deer, and taken from literature for fallow deer (Chapman and Chapman 1997). Data from drive census were also used to estimate wild boar density. For this species, we calculated post-natal density by adding the percentage of piglets, determined from monthly data collected on ungulate community structure, to that of >1-year-old individuals. From March to August, in the CF wild boar population, two cohorts of <1-year-old individuals were present at the same time: newborns of the current year (Cohort X) and those born the previous year (Cohort X-1), but direct observation allowed us to distinguish only Cohort X from the rest of the population. For all species we considered the mean annual density (MD), calculates the mean between post- parturition and late winter density for each year. The biomass of each species was calculated from density and population structure data, using weight values for different age-sex classes obtained from local data (N = 196, for roe deer; N = 173, for red deer; N = 91, for fallow deer; N = 71, for wild boar). Two biomass values were calculated for each year: one obtained in summer from post–parturition density and calculated percentage of juveniles, and one in winter using the pre-parturition density of the following year and the percentage of juveniles estimated from observations made from October to March. For roe deer and wild boar, two different weight values were adopted: one for individuals born in that year and one for adults. For red and fallow deer, instead, four classes were differentiated (young <1-year-old, yearling males, adult males, >1-year-old females).

106 Diet analysis The CF pack diet was studied through the analysis of scat contents, with samples collected between 1988 and 2000, on 13 transects of 53.6 km in total, between 700 and 1500 m a.s.l., repeated each month from 1992 onwards. Scats were prepared for the analysis following Mattioli et al. (1995) and Ciucci et al. (1996). Ungulate species were identified by analysing the morphological features of hairs (shape, colour, diameter, apex thinness) and of bones, and then comparing the samples with a reference collection. Domestic ungulates, hares (Lepus europaeus), and rodents were recognized also by microscope analyses of hair interior structures (thickness of cortex and characteristics of medulla) and comparing the samples with a specific atlas (Debrot et al. 1982).

A blind test was applied on 42 items of known species and volumes, distributed over 20 samples (26.2% wild boar, 45.2% cervids, 9.5% bovids), in order to verify the ability of operators in identifying prey species and in classifying volumes. The senior Author, who performed most of the analysis and controlled it all, recognised all the species and mistook only one class of volume in the blind test. Interpretation of data obtained by scat analysis was based on the assumption that in CF the scavenging of ungulates was negligible, so that consumption would correspond to predation. The use of all food categories was evaluated by mean percent volume (MPV) and relative frequency of occurrence (FO). For each sample, a value of volume was assigned to each type of prey using five classes: 0%, 25%, 50%, 75%, and 100%. Volume values less than 5% were not considered in calculations. For annual analyses, a period from March to the following February was considered, according to the biological cycle of wild boar, the main prey. Seasonal variations were analysed by MPV for the years from 1992 to 1999, for which the size of samples was generally greater than 100 scats. Years were divided into four seasons: spring (March-May), summer (June-August), autumn (September-November) and winter (December- February). On the basis of 1211 scats collected from November 1992 to December 1997 (period B), we calculated the relative biomass (BIO) and the relative number of preys (RNP) of ungulate species and other mammals (only for roe deer were continued these calculations till 2000). Using the volume values, we applied the biomass model of Floyd et al. (1978): Y = 0.38 +

107 0.02 X, where Y represents the biomass (kg) of preys for each collectable scat and X is the live weight of preys. This model was chosen because it was developed using preys which are comparable in size to those available in our study area. Moreover, we compared and evaluated differences among this model, the one made by Weaver (1993) (Y = 0.439 + 0.008 X) and the model 1 by Ruehe et al. (2003), weighted for European ungulates prey species (Y = 0.00554 + 0.00457 X).

Evaluation of age and weight classes of prey consumed In this study, for roe, red and fallow deer we discriminated between <1-year-old and >1-year- old individuals on the basis of the characteristics of fur coat and of the ossification and degree of bone enclosure. <1-year-old individuals were attributed to 7 classes of weight, from birth in June to the following May, according to the month of collection of each scat and a growth function curve derived, for each species, from the weight of juveniles shot in portion B of CF from the 1st of August to the 15th of March. For adults, we adopted only one weight class based on the mean weight of each class of sex and age and on their frequency in the population.

For roe and fallow deer, > or < 1-year-old individuals could be recognised throughout the year. For red deer, instead, these two age classes could be reasonably discriminated only in the period from May to October, due to different characteristics of the fur coat. Nor aged samples were attributed to adults and young classes on the basis of their relative proportion, obtained from the aged samples and calculated on two-months, six-months or one-year intervals, for roe deer, red deer and fallow deer, respectively. In calculating intraspecific selections for cervids, the annual period was considered to last from May, when most births occur, to the following April. Seasonal variations in the use of roe deer age classes were assessed by dividing each year into three seasons, summer (May- August), autumn (September-December) and winter (January-April). For wild boar, weight class of preys and scat production time are not related, because births are scattered over a wide period of time (Meriggi et al. 1988). For this reason, different

108 regression functions relating body weight and bone dimension were calculated for about 250 morphometric measurements, from nine individuals of known weight. Only those with a significance level of p<0.005 were considered. From the regression equations, extreme values for nine weight classes (each 5 kg of amplitude) were calculated. Samples containing bone remains that could be measured were attributed to one of these nine classes. Borderline samples between two classes were redistributed pro rata. Moreover, three weight classes, <10 kg, 10-35 kg, and >35 kg, could be recognised also on the basis of size and colour of hairs (Meriggi, personal comm.). In order to verify the ability to place wild boar hairs into those classes, a blind test was used comprising 61 samples of hair belonging to boars of known weight: all <10 kg (n = 5) and >35 kg individuals (n = 31) were correctly assigned, whereas two samples of the 11-35 kg class (n = 25) were attributed to the >35 kg class, representing a total error of 3.3%. On the basis of 71 wild boars whose ages and weights were known, the weight classes were converted into age classes (< or > 1-year-old individuals): individuals weighing more than 35 kg were considered > 1 year of age. The annual use of wild boar was evaluated from March to the following February according to the biological cycle of the species. For each species, we calculated the mean weight of consumed preys dividing the total ingested biomass by the number of consumed individuals, both obtained from the model.

Statistical analysis The presence of trends in the variations of both density (MD) and use (MPV) of ungulate preys from 1988 to 1999 was tested, by Regression Analysis (RA), with linear and non linear models. For the data collected from 1992 to 1999, MPV trends were also analysed by Linear Regression Analysis (MRA) with seasonal adjustment variables procedure (SEASON), which reduces the influence of seasons. Then, for the same period, seasonal variations in the use of ungulate species were evaluated by MRA of MPV with trend and seasonal dummy variables. In order to verify the relationship between the use of different species and their density, the Spearman Correlation Coefficient was calculated. Variations in the use of preys among different years, and those for each season among different years were evaluated by the

109 analysis of variance with the Kruskall-Wallis test (KW). Kendall’s Coefficient of Concordance was used to evaluate the agreement among years of seasonal patterns of the MPV for the main preys. These statistical analysis were performed using the SPSS 11.5 package. For wild ungulates, selection among species and among age classes within each species was verified comparing expected values with simultaneous Bonferroni Confidence Interval, calculated for the proportion of usage (p) (Neu et al. 1974), following the formula: CI = ±

Zα/2k√(p(1-p) /n), where n is the total number of observations, Z is the upper standard normal table value corresponding to a probability tail area of α/2k, with α = 0.05 and k = the number of categories.

RESULTS Wolf presence During the study period the continuous presence of a wolf pack was verified, mean pack size 2 being estimated at the end of summer at 4.9 ± 0.9 wolves/100 km (range: 4-7, Apollonio et al. 2004). Localizations of summer responses obtained during wolf howling sessions and spatial movements of the pack detected by snow-tracking in winter all fell within the study area; on the basis of this information, the pack was estimated to use about 81% of the study area. Reproduction took place each year, with the exception of 1996.

Wild ungulate availability Population density – In the 1989-1999 decade, the estimated mean density of all wild ungulates present in the study area was about 19.9 heads/km2 before parturition and about 34.7 heads/km2 after (Table 1). Roe deer was the most abundant species in the whole period, with the exception of 1989 (Fig. 1), and its proportion in the wild ungulate community was on average ca. 61% (range: 54-70%). The RA did not show any significant trend for roe deer density (RA, Linear model, F = 0.388, p = 0.553). Wild boar was the second most abundant species in the study area, representing ca. 21% (range: 5-40%), and nor did its density show

110 any significant trend (RA, Linear model, F = 0.505, p = 0.497). Only red deer, representing ca. 12% (range: 5-19%), significantly increased its density in the period of study (RA, Linear model, Y = 1.154+3.723*X, R2 = 0.994, F = 1235.7, p<0.0001), with a mean annual growth rate of 14.9%. For fallow deer, we calculated a constant density of 1.4 deer/km2, which corresponded to 5% of the ungulate community.

Table 1 – Mean values of density, biomass and productivity of wild boar, red deer, roe deer and fallow deer in the Casentinesi Forests area, from 1989 to 1999. Sources of data. Ungulate density: Lovari et al. 2000; Orlandi L, Gualazzi S, Bicchi F, (unpublished report for 1999- 2000); Administration Province of Arezzo (unpublished annual report for 1989-1999). Population structure: Apollonio et al., 2000.

Parameter Wild boar Roe deer Red deer Fallow deer Total

Mean body mass adult (kg) 60 24 115 60 Mean body mass Juv. (kg) in summer 3.8 3.4 13.3 7.5 in winter* 28.5 18.3 60 30.8 Juvenile/Adult ratio in summer 1.44 0.65 0.37 0.38 in winter 0.75 0.35 0.27 0.38

Biomass (kg/ km2) in summer 212 329 305 78 924 in winter 138 276 295 89 798 relative biomass 20.3 35.1 34.8 9.7 100.0

Mean density (n/ km2) in summer 9.2 20.2 3.7 1.6 34.7 in winter 3.6 12.2 2.7 1.3 19.8 relative density 23.5 59.4 11.7 5.3 100.0

111

Biomass – In the 1989-1998 period, the mean amount of ungulate biomass estimated in the study area was 924 kg/km2 in summer and 798 kg/km2 at the end of winter (Table 1). Both roe and red deer represented one third of the available biomass. However, their relative importance was inverted during the study: at the beginning, in fact, red deer biomass was about a half that of roe deer, whereas in 1998 it was more than double. Wild boar biomass was on average ca. 23% of total biomass in summer and 17% in winter, while that of fallow deer was ca. 8.5% and 11.2%, respectively.

Productivity – Roe deer productivity was estimated on the basis of a sample of 26 >1-year-old females, hunted in the study area between 1990 and 1993 and was about 1.6 embryos/female. The number of fawns/>1-year-old females, including yearlings without offspring, observed from October to March, decreased from 0.95 to 0.38 in the period of study (RA, Linear model, Y = 87.4-48.4*X, R2 = 0.786, d.f. = 9, F = 33.14, p<0.001), and averaged 0.64 fawns/>1-year-old females; fawns represented on average 39.5% of the population in summer and 26.1% in winter. In a sample of 54 >1-year-old females red deer, hunted from January 2001 to January 2002, the estimated productivity was 0.85 embryos/female. The number of calves/1-year-old female, observed from October to March averaged 0.43, with no trend among years. Calves represented 27.2% of the population in summer and 21.0% in winter. For fallow deer a productivity value of 0.9 fawns/female was assumed (see Methods). The number of fawns/female observed in autumn-winter was 0.63%, in the 1993-1997 period, and the percentage of fawns in the population was 27% both in summer and in winter. Productivity of wild boar was estimated by direct monthly observation. The ratio of piglets/adults in the May-June period, assumed as maximum annual value, substantially varied over the years from 0.79 to 3.1, with a mean value of 1.43. On average, <1-year-old individuals were 59% of the population in May and June and 42.9% from Octobe to March.

112

Figure 1 – Post parturition density estimates and annual use of wild ungulates by a wolf pack in Casentinesi Forests, from 1988 to 1999. Density is represented by bold lines and scaled on the right-hand y axes. Prey use is expressed as mean percent volume (MPV) in scats; MPV is represented by bars and scaled on the left-hand y axes. Each year runs from March to the following February. Scat sample size: 1988 = 57; 1989 = 29; 1990 = 36; 1991 = 77; 1992 = 106; 1993 = 334; 1994 = 309; 1995 = 261; 1996 = 187; 1997 = 93; 1998 = 84; 1999 = 208.

wild boar 80 20

60 15

40 10

20 5

0 0

roe deer 80 20

70 60 15

50 ha Heads/100 40 10 30

MPV MPV 20 5 10 0 0

red deer 80 20 60 15

40 10

20 5

0 0

fallow deer 80 20 70 60 15 50 40 10 30 20 5 10 0 0 88 89 90 91 92 93 94 95 96 97 98 99 113 Wolf food habits General results The diet of the wolf pack was evaluated by analysing 1862 scats collected from May 1988 to April 2000. The distribution of samples was different among seasons, and a gradual increase from summer to the following winter was usually observed (summer, n = 285, autumn, n = 351, winter, n = 544, spring, n = 622; plus n = 60 scats which were not attributable to any particular season, because of a long period of snow cover). The number of scats collected each year averaged 148.4. The diet of the pack was almost completely constituted by wild ungulates whereas domestic ungulates and other items represented only about 10% (Table 2). The main species in the diet was wild boar, while roe deer and red deer were the secondary preys. In the sample of the November 1992 – December 1997 period (n = 1211), wild boar represented about two-thirds of the total consumption of considered prey, and the MPV, BIO and RNP values were very similar. Roe deer was the most second important species in terms of consumed individuals (19.2%) and of MPV (17.5%), but in terms of BIO it had the same importance as red deer (14.3% vs 14.2%). Fallow deer represented a small portion, ca. 1.5% of RNP and 2.1% of BIO. Despite differences in slopes of regression curves among Floyd, Weaver and Ruehe models, b = 0.02, b = 0.008, and b = 0.005, respectively, that means differences in the consumed biomass per collectable scat, the results obtained with the three models were very similar, both for consumed relative biomass and relative number of preys.

114 Table 2 – Wolf diet in the Casentinesi Forests area, as relative frequency of occurrence (FO%), mean percent volume (MPV%), biomass (BIO%) and relative number of prey (RNP%). * Carnivora include fox (Vulpes vulpes), badger (Meles meles) and other Mustelidae. **Small rodents include dormouse (Glis glis) and microtine.

November 1992- Total sample December 1997

Food items n = 1862 n = 1211 FO % MPV % MPV % BIO % RNP % Wild boar 53.9 58.7 67.8 65.1 66.2 Roe deer 19.6 19.1 17.5 14.3 19.2 Red deer 9.5 8.7 8.7 14.2 6.1 Fallow deer 1.8 1.9 1.9 2.1 1.5 Unidentified Cervidae 3.4 2.5 2.6 3.5 1.9 Total Wild Ungulates 88.2 90.9 98.5 99.2 94.9 Sheep and Goats 3.4 3.3 ------Cattle or Horse 0.8 0.6 ------Pig 0.5 0.5 ------Dog 0.1 0.1 ------Total livestock 4.8 4.5 ------

Carnivora * 0.3 0.4 0.6 0.3 1.9 Hare 1.5 1.2 0.9 0.5 3.2 Small rodents** 1.9 1.0 ------Unidentified mammals 0.7 0.5 ------Aves 0.0 0.0 ------Artropoda 0.1 0.0 ------Fruit 0.7 0.2 ------Vegetable 1.2 0.5 ------Unidentified material 0.7 0.7 ------

Total 100 100 100 100 100

115 Annual variations The distribution of samples among the five volume classes was significantly different in the 1988-1999 period for the main prey species (wild boar: H = 95.92 g.l. = 11, p< 0.001; roe deer: H = 89.57 g.l. = 11, p<0.001; red deer: H = 24.37 g.l. = 11, p<0.02). Considering variations of MPV during the period 1988-2000, we observed that roe deer use declined during the first four years but this tendency was inverted after 1992 (RA, quadratic model: Y = 0.544-1.24*X + 0.97*X2, R2 = 0.68, F = 9.81, p = 0.005), while wild boar use tended to have an opposite pattern (RA, quadratic model: Y = 1.25*X - 0.98*X2, R2 = 0.35, F = 4.06, p = 0.055) (Fig. 1). From 1992 to 2000, using the seasonal adjusted series of MPV (SEASON procedure), a negative trend for wild boar and a positive one for roe deer were confirmed (wild boar: RA, linear model, Y = 0.720 - 0.034*X, R2 = 0.267, F = 12.28, p<0.001; roe deer: RA, linear model, Y = 0.1 + 0.023*X, R2 = 0.299, F = 12.79, p<0.001). Annual MPV values for red deer and fallow deer were found to be relatively stable and showed a tendency to cyclic variations with a shorter phase for red deer (three years) than for fallow deer (five years). Domestic ungulates, mainly sheep, showed two positive peaks in the years 1991 and 1995, when wild boar density was at its lowest. During the period from 1989 to 1999, no functional response was found relating MPV to density variations of the main prey species (Wild boar: Rs = -0.07, n = 11, n.s.; Roe deer: Rs = -0.06, n = 11, n.s.; Red deer: Rs = 0.21, n = 11, ns). For wild boar and roe deer, MPV values in the period from 1988 to 1999 were found inversely correlated (Rs = -0.78, n = 11, p<0.01). The same result was found for annual MPV values of wild and domestic ungulates (Rs = - 0.59, n = 11, p< 0.05).

Seasonal variations The seasonal pattern of wild boar use showed a negative peak in summer and a gradual increase in the following seasons, with the highest value in spring. This pattern was found significantly constant in the period from 1992 to 1999 (Kendall’s Coefficient of Concordance, W = 0.469, p<0.010). A significantly lower use of wild boar in summer was confirmed by MRA (Linear model, R2 = 0.489, variable summer: t = -2.704, p = 0.012). Comparing the distribution of samples among the five volume classes for each season among years, we

116 observed a very constant use in spring (H = 4.79, d.f. = 7, n.s.), a relatively constant in winter, except in 1996 (H = 16.69, d.f. = 7, p< 0.05), but a more variable use in summer (H= 23.42, d.f. = 7, p< 0.01) and particularly in autumn (H = 38.79 d.f. = 7, p<0.001) (Fig. 2 ).

Figure 2 – Mean use of wild boar, roe deer and red deer, for each season, in the wolf diet from 1992 to 1999 (n = 1530). Mean use is expressed as mean percent volume (MPV). Bars indicate standard deviation between years. Variations in the use for each season among years were verified by Kruskall-Wallis test; significance values are shown above bars: * = p<0.05; ** = p< 0.01; *** = p< 0.001; N.S. = not significant.).

Wild boar ** *** * N.S. 100 80

60 40 20 0 1234

Roe deer 60 ** ** * ** 40 MPV 20

0 1234

Red deer

*** * N.S. N.S. 60 40 20 0 SUM AUT WIN SPR

117 The use of roe deer was found on average higher in summer and winter than in autumn and spring, although without any constant pattern among years (Kendall’s Coefficient of Concordance, W = 0.094, p = 0.522) and with significant differences among years for every season (spring: H = 21.17, d.f. = 7 p< 0.01; summer: H = 22.29, d.f. = 7, p< 0.01; autumn: H = 21.69, d.f. = 7 p<0.01; winter: H = 14.51, d.f. = 7 p< 0.05). For red deer, on average we observed an opposite pattern with respect to wild boar, although it was not significantly constant among years (Kendall’s Coefficient of Concordance, W = 0.281, p = 0.80). MRA was not significant (F = 2.014, p = 0.121), although the analysis showed an almost significant higher use in summer (t = 2.029, p = 0.052). Comparing seasonal values of volume for each season among years, little variability was found for spring and winter (spring: H = 10.11 d.f. = 7, n.s.; winter: H = 7.68, d.f. = 7, n.s.), but significant variations, instead, for autumn and particularly for summer (summer: H = 29.05 d.f. = 7, p<0.001; autumn: H = 15.50, d.f.= 7, p< 0.05).

Inter-species selection The use of wild ungulates in wolf diet was compared with their relative abundance in the ungulate community, using RNP values for the period B, a significant positive selection of wild boar being observed for all the years. On the contrary, negative selection was verified for roe deer during the whole period, and for red deer, as well, with the exception of 1996. In the sample of the period B, MPV values were found closely correlated with those of RNP for wild boar (Rs = 0.90; n = 5; p<0.05), roe deer (Rs = 1.0; n = 5, p<0.01), and red deer (Rs = 0.90; n = 5, p<0.05), but not for fallow deer (Rs = 0.70; n = 5, n.s.). Therefore, we used MPV values to evaluate selection among species for the whole period; only two differences were found: the absence of selection for roe deer in 1989 and the positive selection of red deer in 1990.

Intra-species selection Roe deer - From November 1992 to April 2000, 444 scats were collected containing roe deer, and age classes were identified in 191 cases (43.0%), using both bone fragments (89.0%) and hair features (11.0%). On average, <1-year-old individuals represented the most of roe prey remains in wolf scats (Mattioli et al. 2004).

118 The annual percentage of fawns out of the total number of consumed roe deer varied between 47% and 70% using RNP, and between 41% and 61% with FO. Positive selection of <1-year- old individuals was found significant for each year with RNP in the 1988-1992 period (cumulated data) and from 1993 to 2000, with the exception of 1997 (Table 3). Considering in detail the years between 1993 and 2000, we found significant positive selection of young in summer from 1994 to 1996, but not in 1993; in autumn in 1993, 1995, 1996, and 1999, but not in 1997; and only in winter 2000, but not from 1993 to 1999. The mean weight of roe deer consumed from November 1992 to April 2000 was ca. 15.9 kg.

Table 3 – Selectivity of wolf predation for roe deer fawns from 1988 to 1999. Percent of fawns in wolf diet was calculated as RNP according to Floyd et al. 1978; n = number of aged scats. Percent of fawns in the population was obtained by averaging a calculated maximum value in summer (see Methods) and a value from direct observation during October-March; n = number of individuals seen. Significance threshold for Bonferroni Confidence Interval was

α = 0.05.

Proportion of fawns in Selection Proportion of fawns in diet Years population (Bonferroni (May x – April x+1) Confidence % n % n Intervals) 1988-1992 0.66 57 0.35 422 + 1993-94 0.68 28 0.31 516 + 1994-95 0.70 36 0.29 118 + 1995-96 0.69 23 0.30 463 + 1996-97 0.64 29 0.29 503 + 1997-98 0.47 12 0.26 578 ns 1998-99 0.59 16 0.25 451 + 1999-00 0.63 37 0.24 628 +

Mean/Total 0.63 181 0.28 3257

119 Red deer - In period B, red deer remains were found in 123 scats and, in this sample, age classes were identified in 73 cases (59.3%), both by bone remains (55.0% of cases) and by hair features (45.0%). In the reduced sample of 60 scats attributed to the May-October period, <1-year-old individuals were 73.3% of FO, 76.9% of MPV, 55.6% of BIO and 82.9% of RNP. In the years with sample size >10, from 1993 to 1996, in the May-October period, the percentage of calves varied between 57% and 91% of RNP, and between 45% and 86% of FO. Positive selection of young was always found significant, except in 1995, using RNP. Mean weight of red deer consumed from November 1992 to April 2000 was ca. 47.9 kg. Fallow deer - In the November 1992-December 1997 period, fallow deer was found in 23 scats. Age classes (< or > 1-year-old individuals) were identified in 15 cases (65.0%), mainly by bone remains (73%, n = 11). Even for fallow deer, the most represented class was that of <1-year-old individuals: 80.0% using FO, 80.7% with MPV, 60.0% with BIO and 80.0% with RNP. Considering a mean percentage of fawns in the population of 27.3%, significant positive selection for <1-year-old individuals was found. The mean weight of fallow deer consumed in the period was 28.6 kg. Wild boar - Wild boar was present in 814 scats collected in the November 1992-December 1997 period, and 331 (40.7%) of them contained identifiable bone remains that allowed the attribution of samples to weight categories. Moreover, 38 samples were assigned to piglet classes (0-5 kg and 6-10 kg, n = 5) and to those of >35-kg boars (n = 33), on the basis of hair features. The rest of the samples (n = 438) was redistributed among the11-35 kg classes. In the November 1992-December 1997 period, samples containing identified remains of <35 kg wild boars, considered <1-year-old individuals, represented 89.3% using FO, 88.2% with MPV, 82.3% with BIO and 89.3% with RNP.

120 Table 4 – Selectivity of wolf predation for wild boar age classes in 1993-1996, compared to 1988-1992 period (Mattioli et al. 1995). Wild boar with full body weight less than 35 kg were assumed to be less than 12 months old. Frequency of different wild boar age classes were calculated as RNP according to Floyd et al., 1978 and n = number of aged scats. Significance

threshold for Bonferroni Confidence Interval was α = 0.05.

Selection Years Juveniles in diet Juveniles in diet Adults in diet Juveniles in Population (Bonferroni (March x - February x+1) (Cohort X) (Cohort X-1) (> 35 Kg) (Cohort X) Confidence Interval) % % % n % n

1988-1992 0.67 0.24 0.09 77 0.59 267 ns 1993-94 0.66 0.25 0.08 98 0.47 230 + 1994-95 0.57 0.35 0.08 101 0.53 171 ns 1995-96 0.74 0.21 0.05 73 0.48 512 + 1996-97 0.33 0.58 0.09 32 0.51 513 ns Mean/Total 0.59 0.33 0.08 381 0.52 1693

121 Figure 3 – Temporal use of wild boar weight classes. Bimestrial periods are considered starting from January-February (JF). n = number of scats containing wild boar remains of known weight class.

50 JF n= 49 40 30 20 10 0

50 MA n= 74 40 30 20 10 0

50 MJ n= 51 40 30 20 10 0

50 JA n= 43 40 Relative n. of preys n. of Relative 30 py() 20 10 0

50 SO n= 45 40 30 20 10 0

50 ND n= 34 40 30 20 10 0 3 8 13 18 23 28 33 38 40 Weights (kg) COHORT X COHORT X-1 ADULTS

122 Annual use of different age classes is shown in Table 4. Cohort X was significantly preferred in 1993 and 1995. This variability was mainly due to the different use rate of the two cohorts (X and X-1) of <1-year-old individuals, rather than to different use of adults (>35 kg). The use of cohort X showed a specific pattern of use during the year. In fact, in March-April, the individuals most used were those belonging to the 11-35 classes, cohort X-1 (74.1%, Fig. 3). In May-June, the use of newborn averaged 59%, little more than their availability, while the rest (37%) was still made up of mean weight boars, belonging to cohort X-1. From July- August onwards, the use of cohort X was always over 80%. In terms of selection, we observed that cohort X was gradually selected with increasing weight. In fact, from February 1993 to August 1996, it was always negatively selected in March-April, while it was positively selected in two out of four years in May-June and in July- August, in two out of three years in September-October, always in November-December, and in two out of three years in January-February. The mean weight of wild boars used in the period from November 1992 to December 1997 was ca. 20.3 kg, the minimum value occurring in July-August (12.2 kg) and the maximum in January-February (27.6 kg).

Use of preys as a function of their size The mean weight of preys used in the November 1992-December 1997 period, obtained cumulating data of all prey species, averaged about 20.5 kg, ranging from 18.2 kg to 23.6 kg. The lowest values were found in the years when pack dimensions were the smallest (details on pack size in the “Wolf presence” section). Preys up to 30 kg amounted to 83.6%. For species whose range of weight exceeded 30 kg, as wild boar, red deer and fallow deer, a higher use of young was verified precisely in the periods when they weighed about 20-30 kg (red deer in July-August, fallow deer in September-October, wild boar in winter-spring). Analysing the variation of mean weight of prey during the year, it was observed that in May- June and July-August mean weights of preys were the lowest (14.6 kg and 13.6 kg, respectively) and the 0-10 kg weight-class predominated, as a consequence of the high use of newborns. In autumn and early winter, the mean weight of preys increased (19.5 kg in September-October and 21.4 kg in November-December). In winter and spring, finally, mean weights of preys were comprised between 20 and 30 kg (26.9 kg in January-February and 25.3

123 kg in March-April). In terms of RNP, <1-year-old individuals were 80% of ungulates on average (range: 67-90%). The mean annual percentage of young in prey species populations was ca. 33.5% (range: 32-38%), calculated as the mean between the value in May-June and that in winter. The use of young, then, was three fold their availability. As a consequence, the mean weight of prey used (20.5 kg) was lower than that of individuals in the population, ca. 34.7 kg (27.4 kg in May-June and 42.0 kg in winter).

DISCUSSION The FC wolf pack relied almost exclusively on wild ungulates. This evidence is coherent with what has been shown by other researches on wolf diet conducted in North America and in central Europe and Asia (Okarma 1995, Bibikov 1985, Peterson and Ciucci 2003). Nevertheless, in those studies the strong relationship between the wolf and its natural preys was described in regions where human impact on the ecosystem was low. The present study, as some other previous Italian ones (Meriggi et al. 1996, Ciucci et al. 1996, Capitani et al. 2004), shows that this correlation tends to be rapidly recreated also in areas where human presence is relevant, such as the Mediterranean, if the ungulate community is abundant. Under these conditions, domestic ungulates are only marginally used, confirming conclusions pointed out by Meriggi and Lovari (1996). Several studies indicated the wolf tendency to select certain prey species (Poulle et al. 1997, Jedrzejewski et al. 1992 and 2000, Filonov 1989, Olsson et al. 1997, Meriggi et al. 1996, Huggard 1993a, Mech et al. 1995, for a review Okarma 1995). Nevertheless, selective tendencies appear very variable at different scale levels. At a macro one, they are related to different available habitats and to the vulnerability and dimension range of related preys. At a micro one, within similar ecological contexts, prey selection can be influenced by particular factors, such as topography, climatic conditions (Okarma et al. 1995, Jedrzejewski et al. 1992 and 2002, Mech and Frenzel. 1971, Peterson 1977, Carbyn 1983), relative density and structure of prey populations, spatial distribution patterns and use of space of preys and predator (Kolenosky 1972, Huggard 1993a), social structure of prey species (Huggard 1993a,

124 Hebblewhite and Pletscher 2002). Therefore, we compared our results on selection mainly with those from Eurasian and Mediterranean areas characterized by highly productive deciduous forest ecosystems, and dominated by wild boar, red deer and roe deer, locally associated with other ungulate species, such as fallow deer, moose (Alces alces), chamois, and mouflon. The wolf in the northern-central Apennines tends to select wild boar positively and roe and red deer negatively. Wild boar was found to be the wolf main prey in some portions of the Apennines (Ciucci et al. 1996, Meriggi et al. 1996). In particular, positive selection of wild boar was described for two packs occupying areas next to FC (Meriggi et al. 1996 and Capitani et al. 2004). Some Authors (Huggard 1993a, Hebblewhite and Pletscher 2002) proposed the herd rather than the individual as the unit of encounter for the predator, so the selectivity of wolf predation should be calculated on the basis of the number of herds and not of individuals. Therefore, the positive selection of the wolf for wild boar in FC would be even stronger due to the larger herd size of this species (mean = 6.5; n = 566) with respect to red deer (mean = 3.1; n = 846), fallow deer (mean = 3.7; n = 366), and particularly roe deer (mean = 2.0; n = 828). The preference for wild boar can be related to several factors. Firstly, several studies reported that the encounter rate is higher for large groups of prey (Meriggi et al. 1996, Hebblewhite and Pletscher 2002, for a review), probably because better detectable. Furthermore, wild boar groups seem to have a more predictable distribution, related to the use of dense vegetation areas as resting sites, and to their habit of moving on relatively fixed trails between feeding and resting sites. Secondly, the attack success is higher on the largest groups, because the predator increases its opportunity to observe group members and find more vulnerable substandard individuals. In addition, in a wild boar population the young represent the highest percentage with respect to the other ungulates, and births are scattered over a larger period of time, so that the probability of wolves encountering vulnerable individuals is higher for this species. Finally, a more critical factor could be the defence behaviour of this prey, which tends to deal with the predator (Reig 1993, Huggard 1993a). This fact produces a high degree of vulnerability for the young within the group, especially when the attack is led by a wolf pack. This pattern of selection of wild boar is in contrast with what is generally observed in central

125 Europe and Russia where wolves positively select red deer, with rare exceptions (for a review, Okarma 1995), and anyway use wild boar less than it is available. Particularly in BPF, selection by wolves is positive for red deer and negative for roe deer and wild boar. Comparing results obtained in BPF and FC, roe deer appear to be a less vulnerable species, both at low (BPF, Jedrzejewski et al. 2000) and at high relative density, as in CF. The scarce vulnerability of roe deer is related to its social structure: in fact, it is present in small groups whose distribution is scarcely predictable. In other European areas, instead, roe deer is positively selected only where the species cohabits with larger preys, such as moose (Olsson et al. 1997), or moose and wild boar (Filonov 1989). Differences in the use and selection of wild boar and red deer, between CF and BPF, could be due to several factors. First, mean density of wild boar estimated at the end of winter was similar in the two areas (3.0 boar/km2 in BPF and 3.6 boar/km2 in FC), while red deer density estimated in BPF was about two fold that in FC (4.6 deer/ km2 vs 2.7 deer/ km2, Jedrzejewski et al. 2000; this study). Moreover, while the production of red deer calves in summer was relatively similar between the two areas (29.4% in BPF and 27.0% in CF), that of piglets was higher in CF than in BPF (59.0% vs 43.4%). Secondly, red deer body dimensions are almost the same in the two areas (Jedrzejewska et al. 1996, Jedrzejewski et al. 2002), whereas wild boar was consistently smaller in CF than in BPF (mean weight of adults: 58 kg, this study, vs 80 kg, Jedrzejewska and Jedrzejewski 1998). Additionally, in CF and in the neighbouring areas wolves are smaller than in BPF: in fact, adult males weigh on average 32. 8 ± 1.7 D.S. kg (n = 6) and females 29.3 ± 3.7 D.S. kg (n = 4), while in BPF mean weights are about 40 kg (n = 11) and 34 kg (n = 5), respectively (Jedrzejewska et al. 1996). Finally, in the Apennines wild boar is rarely subjected to mass mortality events due to harsh and snowy winters, as occurs in BPF. Besides, scarce snow cover leads to better health conditions for red deer calves, and, as a consequence, lower vulnerability. Seasonal variations in the use of different species seem to be better explained by prey dimension changes than by prey abundance. Species of the largest size, like red deer and wild boar, show more definite and constant seasonal patterns of use from the beginning to the end of each years; independently of variations in density. Decrease of red deer in diet observed from summer onwards could be related to the weight increase of calves. On the contrary, wild

126 boar is mostly used in winter-spring, when yearlings reach 15-35 kg in weight and begin to disperse from parental groups (Meriggi et al. 1996). This finding is again in contrast with what was observed in BPF, where the use of wild boar is the lowest in winter-spring and the highest in summer-autumn (Jedrzejewski et al. 1992). Roe deer, instead, is the smallest prey species among ungulates and its use does not present definite seasonal patterns and is variable among years. Wolf selectivity for young animals (i.e. < 1 year) was observed by many Authors for different species in North America, such as moose (Fuller and Keith 1980, Ballard et al. 1987, Peterson et al. 1984), elk (Carbyn 1983, Huggard 1993b, Boyd et al. 1994), white tailed deer (Odocoileus virginianus: Pimlott et al. 1969, Mech and Frenzel 1971, Mech and Karns 1977, Fritts and Mech 1981), and caribou (Rangifer tarandus caribou: Ballard et al. 1987, Gasaway et al. 1983). Analogous results emerged from European and Russian studies on wild boar and red deer (Okarma 1995, Mattioli et al. 1995, Meriggi et al. 1996, Jedrzejewski et al. 1992 and 2002). For roe deer, results of different researches are in contrast: Olsson et al. (1997) and Jedrzejewski et al. (1992 and 2002) did not find any selection for fawns contrary to what was observed in CF and in neighbouring areas by Mattioli et al. (2004). In the present study, wolves clearly selected <1-year-old individuals for all ungulate species. Selection of young was more remarkable for larger species such as red deer, wild boar and fallow deer, than for roe deer. The use of <1-year-old individuals was more important in CF than in most other European study areas, both for red deer, wild boar, and roe deer (Litvinov 1981, Brtek and Voskar 1987, Bunevich 1988, Smietana and Klimek 1993, Jedrzejewski et al. 1992 and 2000). These differences in selection of young, which are usually more vulnerable, could be related to the smaller dimensions of wolves inhabiting Mediterranean areas than those living in central Europe and Russia. This conclusion seems to be confirmed by the results obtained where wolf weight is low, such as the Indian wolf (Canis lupus pallipes) with a mean weight of 18.2 kg. In the Velavadar National Park in Gujarat-India, where the main prey is represented by blackbuck (Antelope cervicapra), wolves markedly select young (Jhala 1993). Selection of young was not constant throughout the year but showed an evolution related to the development of body growth of preys, which is different among species. Instead, very small piglets (0-10 kg) seemed not to be a profitable prey for the wolf, probably because the

127 benefits derived from their small size do not compensate the costs of predation related to their active defence by the group of adults. On the contrary, yearlings with intermediate weight appeared to be more advantageous. A very similar pattern of use of wild boar weight classes is described also for other areas (Meriggi et al. 1996, Capitani et al. 2004). In contrast, as the defence by mothers is less efficient, newborns of cervids are greatly selected beginning from the first weeks of life. Vulnerability and profitability of preys should also be evaluated in relation to the social structure and dimension of a pack during the year. In CF, in summer, when pups were present, adult wolves usually hunted alone or in pairs, because “helpers” were almost always absent from the pack (Apollonio et al. 2004). In this situation, fawns and calves likely provide a more advantageous balance between energy gain and handling time than do piglets. From autumn, when also young wolves participate to hunting, the use of young boars and of yearlings increases, whereas that of cervids diminishes, particularly red and fallow deer calves that in this period begin to weigh more than 40 kg and 30 kg, respectively. In CF, the biomass of the ungulate community in May is almost identical to BPF (924 vs 965 kg/ km2, excluding bison) but density of preys after birth in CF is two fold that of BPF (34 vs 18 heads/ km2). In fact, mean weight of wild ungulates in May in CF is about half that of BPF (26.6 kg vs 52.6 kg), due to the higher density of roe deer and wild boar with respect to red deer. The mean weight of ungulate preys observed in CF was two or three fold lower than that found in BPF in different periods of study (55.0 kg, Jedrzejewski et al. 2000; 67.2 kg, Jedrzejewski et al. 2002). Such a result was related to differences both between the main prey species in diet (red deer and wild boar in BPF and wild boar and roe deer in CF), and among the mean weights of preys for each species (red deer = 69.3-92.6 kg vs 47.9 kg; wild boar = 23.3-43.4 kg vs 20.3; roe deer = 18.1 kg vs 15.9 kg, in BPF and in CF, respectively). The differences in mean weight of preys are also partially due to methods of analysis: in the present study, only scats were analysed, whereas mainly wolf kills were examined in BPF. In fact, many Authors emphasize that the methods of kills searching tend to underestimate the importance of small preys in diet (Litvinov 1981, Olsson et al. 1997, Jedrzejewski et al. 2002), because they are usually completely consumed, and thus their discovery is rare, and also because most kills are found in winter, when mean weight of preys is higher than in summer.

128 As Jedrzejewski et al. (2002) hypothesized for BPF, the greater use of mid-weight classes (range 20 kg - 30 kg) observed in the present study for wild boar, red deer and fallow deer, suggests that these represent the most profitable preys in terms of predation costs, energy gain and losses due to scavenging, for a pack formed of about four members. The relationship between mean weight of preys and annual pack size observed in different years, although not statistically significant, seems to confirm this hypothesis.

ACKNOWLEDGEMENTS We are grateful to the Administration ex-ASFD of Pratovecchio for providing facilities during the study and to the Provincial Administration of Arezzo for logistic and financial support. We would like to thank Lilia Orlandi and Jean Claude Pucci for their contribute in collection and analysis of scats. We are also very grateful to Wlodzimierz Jedrzejewski and Alberto Meriggi, for their general suggestions, and to Alberto Meriggi for his help in improving the statistical analysis. This experiment complied with the current laws of Italy.

129 REFERENCES Apollonio, M., Lovari, C., Mazzarone, V., Gualazzi, S., Trinca, M. and Turchini, I. 2000. Population structure of Roe deer, Fallow deer and Wild boar. In: The Wild Ungulates of Casentinesi Forests: ten years of monitoring results. Regione Toscana, pp. 77-82. [In Italian]. Apollonio, M., Mattioli, L., Scandura, M., Mauri, L., Gazzola, A. and Avanzinelli, E. 2004. Wolves in the Casentinesi Forests: insight for wolf conservation in Italy from a protected area with a rich wild prey community. Biological Conservation 120: 249-260. Ballard, W. B., Whitman, J. S. and Gardner, C. L. 1987. Ecology of an exploited wolf population in south-central Alaska. Wildlife Monographs 98: 1-54.

Bibikov, D. I. 1985. [The wolf. History, systematic, morphology, ecology]. - Izdatelstvo Nauka. [In Russian]. Boyd, D. K., Ream, R. R., Pletscher, D. H. and Fairchild, M. W. 1994. Prey taken by colonizing wolves and hunters in the Glacier National Park Area. The Journal of Wildlife Management 58: 289-295. Brtek, L. and Voskar, J. 1987. [Feeding ecology of the wolf in Slovakian Carpathians] Biologia (Bratislava) 42: 985-990. [In Slovak]. Bunevich, A. N. 1988. [Dynamics of numbers and diet of wolves in Belovezha Primeval Forest]. Zapovedniki Belorussii 12: 108-113. [In Russian]. Capitani, C., Bertelli, I., Varuzza, P., Scandura, M. and Apollonio, M. 2004. A comparative analysis of wolf (Canis lupus) feeding ecology in three different Italian ecosystems. Mammalian Biology 69: 1-10. Carbyn, L. N. 1983. Wolf predation on elk in Riding Mountain National Park, Manitoba. The Journal of Wildlife Management 47: 963-976. C.E.M.A.G.R.E.F. 1984. Methodes de recensement des populations de chevreuils. C.E.M.A.G.R.E.F. Note Technique 51. Chapman, N. G. and Chapman, D. I. 1997. Fallow deer: their history, distribution and biology. 2nd edition. Coch-y-bonddu Books.

130 Ciucci, P., Boitani, L., Pelliccioni, E. R., Rocco, M. and Guy, I. 1996. A comparison of scat analysis method to assess the diet of the wolf. Wildlife Biology 2: 37-48. Debrot, S., Fivaz, G., Mermod, C. and Weber, J.-M. 1982. Atlas des poils de mammifères d’Europe. Imprimerie de l’Ouest S.A. Filonov, K. P. 1989. [Ungulates and large predators in wildlife predators]. - Izdatelstvo Nauka. [In Russian] Floyd, T. J., Mech, L. D. and Jordan, P. A. 1978. Relating wolf scats contents to prey consumed. The Journal of Wildlife Management 42: 528-532. Fritts, S. H. and Mech, L. D. 1981. Feeding ecology of a newly protected wolf population in Northwestern Minnesota. Wildlife Monographs 80: 1-50. Fuller, T. K. and Keith, L. B. 1980. Wolf population dynamics and prey relationship in northeastern Alberta. The Journal of Wildlife Management 44: 583-602. Fuller, T. K., Mech, L. D. and Cochrane, J. F. 2003. Wolf population dynamic. - In: Mech L. D., Boitani L. (eds) Wolves: behaviour, ecology and conservation. University of Chicago Press, pp. 161-191. Gasaway, W. C., Stephenson, R. O., Davis, J. L. and Sheperd, P. E. K., Burris, O. E. 1983. Interrelationships of wolves, prey, and man in interior Alaska. Wildlife Monographs 84: 1-50.

Gavrin, V. F. and Donaurov, S. S. 1954. [Wolf in Belovezha Primeval Forest.] - Zoologicheskiï Zhurnal 33: 904-924. [In Russian]. Gazzola, A., Avanzinelli, E., Mauri, L., Scandura, M. and Apollonio, M. 2002. Temporal changes of howling in south European wolf packs. Italian Journal of Zoology 69: 157- 161. Harrington, F. H., Mech, L. D. 1982. An analysis of howling response parameters useful for wolf pack censusing. The Journal of Wildlife Management 46: 686-693. Hebblewhite, M. and Pletscher, D. H. 2002. Effects of elk group size on predation by wolves. Canadian Journal of Zoology 80: 800-809. Huggard, D. J. 1993a. Prey selectivity of wolves in Banff National Park. I. Prey species. Canadian Journal of Zoology 71: 130-139.

131 Huggard, D. J. 1993b. Prey selectivity of wolves in Banff National Park. II. Age, sex, and condition of elk. Canadian Journal of Zoology 71: 140-147. Jedrzejewska, B., Okarma, H., Jedrzejewski, W. and Milkowski, L. 1994. Effects of exploitation and protection on forest structure, ungulate density and wolf predation in Bialowieza Primeval Forest, Poland. Journal Applied Ecology 31: 664-676. Jedrzejewska, B., Jedrzejewski, W., Bunevich, A. N., Milkowski and L., Okarma, H. 1996. Population dynamics and demography of wolves (Canis lupus) in Bialowieza Primeval Forest (Poland and Belarus) in relation to hunting by humans, 1847-1993. Mammal Review 26: 103-126 Jedrzejewska, B. and Jedrzejewski, W. 1998. Predation in vertebrate communities. The Bialowieża Primeval Forest as a case study. - Springer. Jedrzejewski, W., Jedrzejewska, B., Okarma, H. and Ruprecht, A. L. 1992. Wolf predation and snow cover as mortality factors in the ungulate community of the Bialowieza National Park, Poland. - Oecologia 90: 27-36. Jedrzejewski, W., Jedrzejewska, B., Okarma, H., Schmidt, K., Zub, C. and Musiani, M. 2000. Prey selection and predation by wolves in BPF, Poland. Journal of Mammalogy 81:197-212. Jedrzejewski, W., Schmidt, K., Theuerkaf, J., Jedrzejewska, B., Selva, N., Zub, K. and Szymura, L. 2002. Kill rates and predation by wolves on ungulate populations in Bialowieza Primeval Forest (Poland). Ecology 83: 1341-1356. Jhala, V. Y. 1993. Predation on Blackbuck by wolves in Velavadar National Park, Gujarat, India. Conservation Biology 7: 874-881. Kolenosky, G. B. 1972. Wolf predation on wintering deer in east-central Ontario. The Journal of Wildlife Management 36: 357-369.

Litvinov, V. P. 1981. [The wolf (Canis lupus) and wild boar (Sus scrofa) in the Kyzil-Agach State Reservation.] Zoologicheskiï Zhurnal 60: 1588-1591. [In Russian with English summary] Lovari, C., Mattioli, L., Mazzarone, V. and Siemoni, N. 2000. [Analysis of wild ungulates populations: Red deer, Roe deer, Fallow deer and Wild boar]. - In: [The Wild

132 Ungulates of Casentinesi Forests: ten years of monitoring results]. Regione Toscana, pp. 30-77. [In Italian]. Mattioli, L., Apollonio, M., Lazzarone V. and Centofanti, E. 1995. Wolf food habits and wild ungulate availability in the Foreste Casentinesi National Park; Italy. Acta Theriologica 40: 387-402. Mattioli, L., Capitani, C., Avanzinelli, E., Bertelli, I., Gazzola, A. and Apollonio, M. 2004. Wolf predation on a roe deer (Capreolus capreolus) population of the northern-eastern Apennines, Italy. Journal of Zoology (London) 264: 249-258 Mazzarone, V., Siemoni, N., Pedone, P., Lovari, C. and Mattioli, L. 1991. A method of red deer (Cervus elaphus) census during the roaring period in a forested area of northern Apennines (central Italy). - In: Csanyi S., Ernhaft J. (eds) Transactions of XXth International Union of Game Biologists Congress. Part 1. University of Agricultural Sciences, pp 140-145. Mech, D. L. and Frenzel, L. D. Jr. (1971) An analysis of the age, sex and condition of deer killed by wolves in northeastern Minnesota. USDA Forest Service Research Paper NC- 52: 35-40. Mech, L. D. and Karns, P. D. 1977. Role of the wolf in a deer decline in the Superior National Forest. USDA Forest Service Research Paper NC-143: 1-23. Mech, L. D., Meier, T. J., Burch, J. W. and Adams, L. 1995. Patterns of Prey Selection by Wolves in Denali National Park, Alaska. - In: Carbyn, L. D., Fritts, S. H. and Seip, D. R. (eds), Ecology and Conservation of Wolves in a Changing World. Proceedings of the Second North American Symposium on Wolves, Canadian Circumpolar Institute, 35: 231-244. Mech, L. D. and Peterson, R. O. 2003. Wolf-prey relations. In: Mech, L. D. and Boitani L. (eds), Wolves: behaviour, ecology and conservation. The University of Chicago Press, pp. 131-157. Meriggi, A., Stoppani, N., Cesaris, C., Zacchetti, D. and Barbieri, F. 1988. [Analisys of wild boar (Sus scrofa) harvest in north Apennine]. - In: Spagnesi, M., Toso, S. (eds), Proceedings of I National Congress of Italian Game Biologists. Suppl. Ric. Biol. Selvaggina 14: 641-642. [In Italian]

133 Meriggi, A., Brangi, A., Matteucci, C. and Sacchi, O. 1996. The feeding habits of wolves in relation to large prey availability in northern Italy. Ecography 19: 287-295. Meriggi, A., Lovari, S. 1996. A review of wolf predation in southern Europe: does the wolf prefer wild prey to livestock? Journal of Applied Ecology 33: 1561-1571. Neu, C. W., Beyers C. R. and Peek, J. M. 1974. A technique for analysis of utilization- availability data. The Journal of Wildlife Management 38: 541-545. Okarma, H. 1995. The trophic ecology of wolves and their role in ungulate communities of forest ecosystems in Europe. Acta Theriologica 40: 335-386. Okarma, H., Jedrzejewski, B., Jedrzejewska, W., Krasinski, Z. and Milkowski, L. 1995. The role of predation, snow cover, acorn crop, and man-related factors on ungulate mortality in B. P. F., Poland. Acta Theriologica 40: 197-217. Olsson, O., Wirtberg, J., Andersson, M. and Wirtberg, I. 1997. Wolf (Canis lupus) predation on moose (Alces alces) and roe deer (Capreolus capreolus) in south central Scandinavia. Wildlife Biology 3: 13-23. Peterson, R. O. 1977. Wolf ecology and prey relationship on Isle Royale. - U. S. Natl. Park Serv. Sci. Monogr. Ser. 11. Peterson, R. O., Woolington, J. D. and Bailey, T. N. 1984. Wolves of the Kenai Peninsula, Alaska. Wildlife Monographs 88: 1-52. Peterson, R. O. and Ciucci, P. 2003. The wolf as a carnivore. In: Mech L. D., Boitani L. (eds) Wolves: behaviour, ecology and conservation. University of Chicago Press, pp. 104- 130. Pimlott, D. H., Shannon, J. A., Kolenosky, G. B. 1969. The ecology of timber wolf in Algonquin Provincial Park. - Ont. Dep. Lands For. Res. Rep. Wildl. 87. Poulle, M.-L., Carles, L. and Lequette, B. 1997. Significance of ungulates in the diet of recently settled wolves in the Mercantour mountains (southern France). Revue d’Ecologie (Terre vie) 52: 357-368. Promberger, C. H. and Schröder, W. 1993. Wolves in Europe – status and perspectives. Procedings of the workshop “Wolves in Europe – current status and prospects”, April 1992, Oberammergau. Münich Wildlife Society, pp 1-136.

134 Reig, S. 1993. Non aggressive encounter between a wolf pack and a wild boar. - Mammalia 57: 451-53. Ruehe, F., Buschmann, I. and Wamelig, A. 2003. Two models for assessing the prey mass of European ungulates from wolf scats. Acta Theriologica 48: 527-537. Smietana, W. and Klimek, A. 1993. Diet of wolves in Bieszczady Mountains, Poland. - Acta Theriologica 38: 245-251. SPSS Inc. 2002. Statistical Package for Social Sciences. SPSS Inc.. Weaver, J. L. 1993. Refining the equation for interpreting prey occurrence in gray wolf scats. The Journal of Wildlife Management 43: 783-876.

135 CONCLUSIONS

The wide use of different food items by wolves is a widespread phenomenon found both in North America and on the Eurasian continent (Voigt, Kolenovsky and Pimlott, 1976; Salvador and Abad, 1987; Okarma, 1995; Spaulding et al., 1998). In the Apennines area wild boar was the main wild ungulate in the wolf diet (VI). Instead, in the Alps, in the Susa valley area (I) wolves fed on a large number of wild ungulates but in particular red deer, so the food habits of wolves are similar to those of Eastern Europe (Smietana and Klimek, 1993; Okarma, 1995; Jedrzejewski et al., 2000). The preference for red deer was high especially during winter. In literature is reported that wolf predation tend to kill primarily the youngest, and debilitated member of prey herds (Mech 1966; Bubenik 1972; Schwartz et al., 1992). Thus ungulate populations, under the influence of wolves, tend to have individuals of good condition and health, and therefore to be highly productive. In both study areas (I and VI) a larger than expected use of young ungulates (< 1 year of age) by wolves was found. Moreover, in Susa valley (II), most of red deer killed by wolves suffered of heavy under nutrition, while in roe deer, femur fat levels of wolf kills were high and did not differ from those of the population. The high susceptibility of red deer, in Susa Valley, could be the result of a combination of factors: the strong overlap of habitat and altitude with the wolf area, and their gregarious life style (I). Analogous considerations could be made for wild boar (VI) in the Apennine area. Indeed, this prey species is the most gregarious ungulate species with the more predictable distribution. Moreover, the highest percentage of young in the wild boar population and the births distributed over a larger period of time, could explain the highest probability for wolves in encountering vulnerable individuals. In spite of a wide use of wild ungulate by wolves, our study seem to line up to other cases in which prey populations were not influenced by the presence of wolves (Mech 1986; Adams and Dale 1998; Mech et al., 1998; Peterson et al., 1998; Nelson and Mech 2000). Wild ungulate densities didn’t change before and after arrivals of wolves in Susa Valley. Nevertheless, the red deer was the prey species more influenced by wolf presence, but wolf

136 predation alone was a poor predictor of deer population dynamics (III). Also the wild boar population of the Apennine study area didn’t decrease during the study period even though was the primary wolf prey (VI). Although wolves, in several ecological contexts, do help limit or retard the growth of their prey populations, this predator, on the other hand, do not necessarily hold prey numbers down. This feature seems truly our study cases, where wolf predation seems to be a compensatory mortality factor. High levels of predation on domestic ungulates were encountered, during the summer, in the Susa valley, in spite of the great richness of the wild ungulate community (I). This phenomenon is probably related to the absence for a long period of this predator, that led to the loss of any remaining tradition of coexistence with the wolf on the part of stockmen. Large flocks (500-1000 sheep) in open areas contribute to make encounters with wolves easier, and the lack of measures to prevent wolf attacks (guardian dogs, nocturnal confinement) make sheep the most vulnerable and profitable prey-species. In Tuscany, livestock damages were mainly distributed in municipalities where wolf presence was ascertained (IV). These areas were characterized by a larger percentage of forest cover and a higher percentage of protected area than the territory where wolf was absent, whereas human population density and roadway and railway density were significantly lower. Predation events on livestock are particularly concentrated in a few farms that reported a high level of conflict. Therefore, it appears to be a critical point to act on the few farms severely damaged in order to reduce both the overall and the individual economical losses and the individual hostile reaction to predation risk pressure. Nevertheless, considering the high density of wolves in the area (IV and V), the impact of predation was low both in terms of percentage of affected farms and livestock losses. Coexistence of wolf and human populations can be reached by taking into account both conservation of environmental conditions (V) and reduction of conflict, by adopting strategies to prevent damage to human activities by wolves (IV).

137 REFERENCES

Adams, L.G., Dale, B.W. 1998. Reproductive performance of female Alaskan caribou. The Journal of Wildlife Management 62: 1184-1195. Bubenik, A.B. 1972. North American moose management in light of European experiences. Proc. N. Am. Moose Conf. 8: 279-295. Jedrzejewski, W., Jedrzejewska, B., Okarma, H., Schmidt, K., Zub, C., Musiani, M. 2000. Prey selection and predation by wolves in BPF, Poland. Journal of Mammalogy 81:197-212. Mech L.D. 1966. The wolves of Isle Royale. U.S. National Park Service Fauna Series, no. 7. U.S. Govt. Printing Office. 210 pp. Mech, L.D. 1986. Wolf population in the central Superior National Forest, 1967-1985. USDA Forest ServiceResearch Paper NC-270. North Central Forest Experiment Station, St. Paul, MN. 6pp. Mech, L.D., Adams, L.G., Meier, T.J., Burch, J.W., Dale, B.W. 1998. The wolves of Denali. University of Minnesota Press, Minneapolis. Nelson, M.E., Mech, L.D. 2000. Proximity of white-tailed deer, Odocoileus virginianus, home ranges to wolf, Canis lupus, pack homesites. Canadian Field Naturalist 114: 503- 504. Okarma, H. 1995. The trophic ecology of wolves and their predatory role in ungulate communities of forest ecosystems in Europe. Acta Theriologica 40 (4): 335-386. Peterson, R.O., Thomas, N.J., Thurber, J.M., Vucetich, J.A., Waite, T.A. 1998. Population limitation and the wolves of Isle Royale. Journal of Mammalogy 79: 828-841. Salvador, A., Abad, P.L. 1987. Food habits of a wolf population (Canis lupus) in Leòn province, Spain. Mammalia 51 (1): 45-52. Schwartz, C.C., Hundertmark, K.J., Spraker, T.H. 1992. An evaluation of selective bull moose harvest on the Kenai Peninsula, Alaska. Alces 28: 1-13. Smietana, W., Klimek, A. 1993. Diet of wolves in the Bieszczady Mountains, Poland. Acta Theriologica 38 (3): 245-251. Spaulding, R.L., Krausman, P.R., Ballard, W.B. 1998. Summer diet of Gray Wolves, Canis lupus, in north-western Alaska. Canadian Field Naturalist 112 (2): 262-266. Voigt, D.R., Kolenovsky, G.B., Pimlott, D.H. 1976. Changes in summer foods of wolves in central Ontario. The Journal of Wildlife Management 40: 663-668.

138 ACKNOWLEDGEMENTS

“The strength of the wolf pack is the wolf. The strength of the wolf is the pack.” - Rudyard Kipling, 1894 -

First and foremost I would like Alessandro Gazzola and Fabrizia Bastianini (babbo e mamma lupo) for have taught me the importance of the culture and the honesty. I would like to thank my supervisor, Marco Apollonio for the numerous advises and for have believed in me. I wish to thank Bogumiła Jędrzejewska and Włodzimierz Jędrzejewski, (Mammals Research Institute, Polish Academy of Sciences, Białowieża, Poland) for their precious comments on the early drafts of this Thesis. This work has been only possible thank to the cooperation of numerous collaborators (“Alpine wolf pack and Apennine wolf pack”). I am grateful mainly to Elisa Avanzinelli (the “α – female of Alpine wolf pack”) for the efforts, delights and numerous satisfactions lived together; Ivo Bertelli (“dispersal wolf” from Apennine to Alpine area), Paola Bertotto (“cappuccetto rosso”), and Marco Costamagna for the numerous experiences. Thanks to Aldo Tolosano (the “Sheriff” of Susa Valley), Claudio Scaini, Walter Grosso, Diego Corti, and Marco Fazio (Servizio Tutela Flora e Fauna, Provincia di Torino) who have adopted me and have helped me. Federico Kurschinski (Consorzio Forestale Alta Valle di Susa) and Roberto Corti (Corpo Forestale dello Stato) fellow travellers of numerous adventures. The “α-pair of Gran Bosco wolf pack”, Massimo Rosso and Elisa Ramassa (Parco Naturale Gran Bosco di Salbertrand) for indispensable collaboration. Ezio Ferroglio, Umberto Vesco and Luca Amistadi (Università di Torino), Marco Moretti, Fabio Ghiandai for the precious contributions. Thanks to Luca Giunti and Dante Alpe (Parco Naturale Orsiera-Rocciavrè), Giorgio Favale (Corpo Forestale dello Stato), Alberto Dotta and Rinaldo Gros (Consorzio Forestale Alta Valle di Susa), Roberto Musso and Pino Arleo (Comprensorio Alpino CaTo2). I am grateful to “Apennine wolf pack”……… Thank to “the wolves of Casentinesi Forests”, Guido Crudele (ex-ASFD Amministrazione di Pratovecchio), Filippo Baldassarri (Corpo Forestale dello Stato), Ivo Bertelli, Riccardo Gambogi, Elisa Avanzinelli, Paolo Lamberti, Lorenza Mauri, Massimo Scandura, Alessia Viviani, Daniela Giustini and

139 Francesca Benvenuti (Università of Pisa, Sassari and Milano), helped substantially in the collection of field data of wolf distribution. Luca Mattioli (Provincia di Arezzo) and Claudia Capitani (Università di Sassari) for the fundamental support for realization of this work.

We would also like to thank James Burge (Università di Camerino) for linguistic revision.

I am grateful to the Arezzo Veterinary Health Service (ASL8) and the Department of Agriculture and Forests of the Region of Tuscany for essential cooperation, as well as to the Provincial Administration of Torino and Arezzo, and Piemont Regional Government for logistic and financial support. Thanks to the Journal Zoology of London, and the Biological Conservation for the permission to include the papers in my Thesis.

140