BROWNBEAR HABITATQUALITY IN GORSKIKOTAR,

JOSIP KUSAK,Biology Department, Veterinary Faculty of the Universityof Zagreb,Heinzelova 55, 10000 Zagreb,Croatia, email: [email protected] [email protected] DJUROHUBER, Biology Department, Veterinary Faculty of the Universityof Zagreb,Heinzelova 55, 10000Zagreb, Croatia, email: [email protected]

Abstract: The (Ursus arctos) population in Croatia is a possible source of bears for reintroducingand augmenting disappearing Europeanbrown bear populations in western Europe. For successful reintroduction,knowledge about bear habitat quality of both source and target areas is necessary. We developed a habitat suitability index (HSI) model to assess Europeanbrown bear habitatquality in , Croatia. Importanthabitat variables included seasonal foods, cover, roads, and fragmentation. Food sources were available year-round,whereas foraging and denning cover were more limited. Humaninfluence was manifestedthrough a relatively high density of roads (1.91 km/km2),which included forest roads. Habitatfragmentation did not occur within the study area, but a highway under constructionwas a possible threat. The overall HSI value of 0.42 for the entire area indicatedthat brown bear habitatin Gorski kotar is average. Brown bear habitatcould be improved with changes in managementpractices such as closing forest roads, seasonally avoiding logging in denning areas, and reducing reforestationof beech- (Fagus sp.-Abies sp.) areas to spruce (Picea abies).

Ursus 10:281-291

Key words: Croatia,European brown bear, habitat quality, habitat suitability index, Ursus arctos.

One of the largest remainingbrown bear populations This study was fundedby the CroatianMinistry of Sci- in Europe occurs in the , Sara, and Pindus moun- ence, Wildbiologische Geselschaft-Miinchen, John tains (approximately 60,000 km2) spanning Slovenia, Sheldon Bevins Memorial Foundation, and Stiftung Croatia, Bosnia and Herzegovina, Monte Negro, EuropaiischeNaturerbe (EURONATUR). We thank A. Macedonia (Huber and Frkovic 1993), Albania, and Frkovic, D. Majnaric,and B. Plese for logistic field sup- Greece (Servheen 1990). Within the Croatianportion of port. Special thanksare due to the reviewers and associ- this range, about 400 bears inhabit approximately9,800 ate editors for improving the style and quality of the km2(Huber and Frkovic 1993), and this numberhas been manuscript. stable since the early 1980s (Frkovicet al. 1987). Mitochondrial DNA testing has demonstrated that brown bears in the Dinara have greater ge- BROWNBEAR HABITAT SUITABILITY netic similaritiesto the small, isolated populationsin the INDEXMODEL Alps, Apennines, and Pyrenees than to other European populations(Taberlet and Bouvet 1994, Randiet al. 1994). HabitatUse Requirements Consequently,they may be the best source to reintroduce Food habits and habitatuse of brown bears in Croatia or to augment existing bear populationsin western Eu- were analyzed by Cicnjak (1991). In spring and early rope. Projectsto reintroducebears from the DinaraMoun- summer,bears were found at lower elevationswhere food tains to other countriesare in progress,with 3 (2 F, 1 M) was available. Main food sources includedbroad-leaved bears already translocatedto the AustrianAlps and an- garlic (Allium ursinum), lords and ladies (Arum other 3 (2 F, 1 M) to the FrenchPyrenees. maculatum),and carrionand corn at permanentfeeding Maintenance of suitable bear habitat is central to sites. Bears also visited meadows where they consumed the continued presence of the species in Europe grasses (Gramineaespp.), clovers (Trifolium spp.), and (Servheen 1990). Knowledge of brown bear habitat sorrels (Rumex spp.). During the summerwild angelica relationships is needed to enhance management of (Angelica silvestris) and stinking aposeris (Aposeris existing European brown bear populations and to im- foetida) were the main plant foods eaten, althoughbears prove chances for successful reintroduction efforts. occasionally fed in oat fields (Huber and Moric 1989). We developed a habitat suitability index (HSI) model By late summer,fruits ripened in meadows, abandoned for brown bears in southern Europe following the U.S. orchards,along fields and roads, and in the forest; the Fish and Wildlife Service's (1981) habitat evaluation most importantbear foods were raspberry(Rubus idaeus), procedures. We then used the HSI model to evaluate bramble (R. fruticosus), common buckthor (Rhamnus brown bear habitat in the Gorski kotar region of cathartica),and blueberry (Vacciniummyrtillus). Dur- Croatia; we recommend management changes based ing fall, bears spent the most time in forestedareas where on that analysis. beech nuts were abundant. Similar patternswere found 282 Ursus 10:1998 in Cantabrian Mountains, Spain (Clevenger et al. can increasetraffic mortality of bears. Crossing sites on 1992a,b). Beech treesstart producing nuts at 40-50 years. roadsand railroadswere recognizedas importantparts of Maximumcrop production(x = 1.38 kg/) can be ex- migratingcorridors (Huber et al. 1998). Telemetrystud- pected every 7th to 12th year; in other years production ies in Slovenia (Kaczenskyet al. 1996) showed thathigh- is 0.11 kg/tree. The cycle of beech productivityis differ- ways are a physical barrierfor movements of females, ent in differentparts of the forest (Regent 1980). Depre- whereas males occasionally cross during the mating dation on domestic animalsin Croatiawas lower than in season. other Europeancountries because of the availability of natural and supplementalfood (Cicnjak 1991). While HSIModel Applicability feeding on naturalfood, the average distance of bears to This model describeshabitat suitability in Gorskikotar, settlementsand roads was not statisticallydifferent from Croatia(Fig. 1). It can be appliedto similarmountainous randomlydistributed points (Cicnjak1991). Brown bear areas in Europe, such as the Alps, Pyrenees, Apenines, distributionand abundancehas not been associatedwith and Carpathians. It is designed to produce HSI values water availabilityin Croatia(Cicnjak 1991). for year-roundhabitat needs of the brown bear. We as- Europeanbrown bears do not rely on protectivecover sumed that carryingcapacity is restrictedby food avail- to access food; they have developedcrepuscular and noc- ability and that a weak compensatoryrelationship exists turnal activity patterns when protection cover is poor for food availability among seasons. To stress impor- (Roth and Huber 1986). The closest distance to which tance of fall food, we weighted it by a factor of 2. The man can cautiously approacha Europeanbrown bear is model can be appliedon areaswhere beech, mixed beech- about 30 m (pers. observ.). Bears that are approached fir, and spruce forests prevail. The average home range closely may abandonsome feeding places, but this may size for both sexes of bears during foraging season not be of great importance in habitat with good food was 28 km2 (range = 1-102 km2) (Huber and Roth sources elsewhere. Protective cover may be more im- 1993). Therefore, the model requires a minimum habi- portantfor selection of suitabledaybeds or dens. Resting tat area of 30 km2. sites (day beds) had the shortesthorizontal visibility (x = The model was developed based on our knowledge of 8.4 m, range = 3-46 m), followed by denning sites (x = bear biology, expert opinion (A. Frkovic, CroatianFor- 12 m, range = 1-18 m) and feeding sites (x = 36, range = ests, , Croatia,pers. commun., 1993) and a thor- 6-186 m) (Cicnjak 1991). ough comparison with other models (Aste 1993; Van Bearsin Croatiaden in forestedareas at elevationsfrom Manen and Pelton 1993, 1995; Clevenger et al. 1995); 450 m to 1370 m (x = 863 m, N = 28). The average however, it has not been tested underfield conditions. distancefrom dens to the nearestroad was 486 m (range = 39-1500 m) and from dens to the nearest settlement Components was 1435 m (range = 200-4000 m) (Huber and Roth A habitatsuitability index (HSI) is a numericalindex 1995). Feeding on energeticallyrich beech nuts, which that representsthe ability of a given habitatto supporta were the most importantfood source for bearsin Croatia specific species. An index of 0.0 representsunsuitable in autumn(Cicnjak 1991), could be of great importance habitat,whereas an index of 1.0 representsoptimal habi- for successful winteringand giving birth(Schooley et al. tat. For HSI to be between 0.00 and 1.00, an assumed 1994). linear relationshipbetween chosen habitat components In Croatia,habitat types reflect differences in eleva- and carryingcapacity must be met (U.S. Fish and Wild- tion, geology, and other ecological factors. Seasonal life Service 1981). movements of bears reflect changes of food availability We defined food, cover, andhuman impact as the main in differenthabitat types. In the contiguoushabitat found bearhabitat components (Fig. 2). Van Manenand Pelton in Gorskikotar, average home rangesize of bearsfor both (1993) defined ecological, physical, and culturalhabitat sexes did not vary between spring,summer, and fall (x = componentsfor Americanblack bear (Ursus americanus). 28 km2,range = 1-102 km2,n = 32), althoughwinter range Aste (1993) listed the following bearhabitat components of was significantlysmaller (x = 4 km2,range = 0-18 km2,n for Austria: food type, food availability,availability factors. = 5) (Huberand Roth 1993). remote and safe places, and effect of limiting in Cantabrian Forest management in the mountains of Croatia in- Clevengeret al. (1992a) rankedbear habitat volves building a dense network of forest roads. These Mountains(Spain) by quality of forest cover, elevation, roads are not closed, which increaseshuman access. In- distanceto nearestvillage, and distanceto nearestpaved creased traffic on local and regional roads and railroads road. The following sections describethe importanthabi- BROWNBEAR HABITAT IN CROATIA* Kusak and Huber 283

Fig. 1. Location of Croatia (top), the Gorski kotar study area (middle), and the study area sections (bottom). 284 Ursus 10:1998

Habitat Season Variables Habitat component

V - spring food * SPRING [--- availability V2 - spring food diversity

FOOD V3 - summerfood availability , SUMMERI ) t V4 - summer food diversity

1 FALL V5 - autumnfood availability # V6 - autumnfood productivity

- ) FORAGING V7 - protectivecover in foragingperiod HSI COVER

* DENNING V8 - protectivecover in denning period > V8a- in forest / out of forest * V8b - distance fromsettlement ) V8c - distance from road

V9 - of all road I density V10 - habitatfragmentation

Fig.2. Structureof the brownbear habitat suitability index (HSI) model for Gorskikotar, Croatia. tat variables used in the model, the suitability level of V9, density of all roads these variables,and the relationshipsamong variables. V10, habitatfragmentation Food.-We considered only plants that were present When total cover was <10% (V1 or V3 <10%), calcu- in bear diets with a frequency >5% and volume 20.5% lation for spring and summer was SIVI orV3 = 1/10 V1 or (Cicnjak1991). A total cover of 10%for 1 or for several V3. When total cover was >10% (V1 or V3 > 10%),the = plantswas takenas the thresholdof optimalindex (Fig. 3 value was SIV, orV3 1.00 (Fig. 3 A, C). Beech cover, A, C). We measuredthe following 10 variablesof suit- becauseit producedthe most importantfall food, changed ability indices (SI) and furthersubdivided variable 8 into SI linearly from 0% to 100%, (Fig. 3 E). This relation- 3 subvariables: shipwas chosenbecause productivity cycles of beech VI, springfood availability vary in time in different areas. Not all beech trees pro- V2, springfood diversity duce maximum crops in the same year (Regent 1980). V3, summerfood availability The same model was used by Aste (1993). V4, summerfood diversity Whena single significantbear plant was foundin spring V5, autumnfood availability or summer,the area was assigned an SIV2or V4 = 0.50. If V6, autumnfood productivity >2 plants were found, we assigned SIV2or 4= 1.00 (Fig. 3 V7, protectivecover in foraging period B, D). When <40% of beech trees producednuts (V6 V8, protectivecover in denning period <40%) SI was calculated as SIV6= (1/40)V6. For V6 V8a, in/out of forest >40%, SI6 = 1.00 (Fig. 3 F). V8b, distance from settlement Cover.-We evaluatedhiding cover separatelyfor the V8c, distance from road foraging and denning period. If <25% of the sites in a BROWNBEAR HABITAT IN CROATIA* Kusak and Huber 285

A V1 - spring food cover B V2 - springfood diversity SIV2=10 Sly, = 1.0 ,. -...... -. 1.0 - 1.0 S 1.....:. i-0 10

0.8 0.8

0.6 -0.6 S2 = 0 1 IV2 0.5 U~~~~~~~~~~~vo06S c 0.4 - 1 - v 1 0.4 _ -/ _z . ? 0.2 -| 0.2 - ! O0 (/')'~U ,:: :: , . c 7I I I .... 10 25 50 75 1000/o one two or more Forbs cover Forbs diversity

diversity C V3 -summer food availability D V4 - summer food Slv3 = 1 Slv4 = 1.0 C1.0 - 1.0

- 0.88

0.6 -/|o.6 s|,, = 0.5 0.4 --

0. 4 - S.3 -= V3 :: 0 ~0.2 .~ 02 . ,

10 25 50 75 1000/o one two or more Fruitcover Fruitdiversity

E V5 - autumnfood availability F V6 - autumnfood productivity

AA Slv6 = 1.0 1.0 - 1.0 -

0.8- 0.8 -

0.6- - / 0.6 /6 0.6

0.4 - / S = V5 0.4

cU 02 0- ~a .2 Slvs ~i40 V6 ? 0_ s' -/ I I I I> I I 25 50 7 000/ 75 251000/0 40 50 Beech cover % of productivebeech trees

Fig. 3. Relationships of suitability indices (SI) and 13 habitat variables used in the brown bear HSI model. (A) Spring food availability. (B) Spring food diversity. (C) Summer food availability. (D) Summer food diversity. (E) Autumnfood availability. (F) Autumnfood productivity. (G) Protective cover in foraging period. (H)Sites for denning. (I)Distance of denning areas from road. (J) Distance of denning areas from settlement. (K) Protective cover in denning period. (L) Density of all roads. (M) Habitatfragmentation. 286 Ursus 10:1998

G V7 - protectivecover in foragingperiod H V8a - sites for denning lI - = 1 n 1.0 1.0

08 0.8

v) uCO 0.6 0.6 -0 D 0.4 ._ 0.4 Slv7 = 5 V7 i 25 5 0.2 ? 0.2 co 0t Slira = 0.0 25 50 75 100% Inforest Outsideforest % of areas withhorizontal density of understory vegetationwhich obscures >60%of bear silhouetE

V8b- distance of denningareas fromroad J V8c - distance of denningareas fromsettlement ' SIvsb = 1.0 SIvs8 = 1.0 1.0 1.0

08 - 0.8

0.6 - 2 0.6 -o -0 CU 0.4 - ._ - 0.24

n 0.2 - . _ 0.2.0 O9 c, Slvsb = 0.0 Slv8c = 0.0 ., I~~~~~ 0) 500m 1500 m distancefrom road distancefrom settlement

K V8 - protectivecover in denningperiod

1.0

0.8 1 1 Slvs = V8 - ' 0.6 75 3

-o . 0.4

I 0.2 0

25 50 75 100%/ % of suitabledenning areas

Fig. 3 (Con't.). Relationships of suitability indices (SI) and 13 habitat variables used in the brown bear HSImodel. (A) Spring food availability. (B) Spring food diversity. (C) Summer food availability. (D) Summer food diversity. (E) Autumn food availability. (F) Autumn food productivity. (G) Protective cover in foraging period. (H) Sites for denning. (I) Distance of denning areas from road. (J) Distance of denning areas from settlement. (K) Protective cover in denning period. (L) Density of all roads. (M) Habitatfragmentation. BROWNBEAR HABITAT IN CROATIA* Kusak and Huber 287

L V9 - density of all roads assigned a value of 1.00 for SIV8b,whereas closer dis- tances (less suitable)were assigned SI 8b= 0.0 (Fig. 3 I). 1.0 Denning areas <1500 m from human settlements also were consideredunsuitable for denning (SIV8c= 0.0), but areas >1500 m were considered suitable bear denning 0.8 Slve = - 1 V9+19 1.5 habitat(SIV8c = 1.00) (Fig. 3 J). Areas suitable for den- ning (V8) were determinedby the factor with the lowest ) 0.6 - value, because we assumed that factors would be limit- - ing (V8 = min [V8a, V8b, V8c]). It was possible for .c 0.4 - only partof a section to be suitable for denning if it was in the forest and >500 or >1500 m away from the road or .0i, co 0.2- t settlement,respectively. The habitatsection being ana- ._ lyzed should have >25% suitable denning sites for SIV8 to be >0.0 (SIV8 = (1/75)V8 - (1/3)). The same model was used by Aste (1993) for calculating availability of remote areas. HumanImpact.-We evaluatedhuman impact by sum- ming total length (V9, km/km2)of all roads, including V10 - habitat M fragmentation forest roads that are open and accessible to the public, and for habitat We ? o V1 1 by searching possible fragmentation. 1.0 S assumed that habitatquality decreased as the density of 2490 24 9 roads increased;areas with > 1.5 km/km2roads received 0.8 SIv9 = 0.0 (Fig. 3 L). We used the same model developed by Aste (1993). 0.6- Partof the habitatwas consideredisolated if it was sur- roundedby 500 m of unforestedarea. We did not con- a/ sider habitatas isolated, even if it containedrailroads or r 0.4 - roads, if there were places for crossing (tunnel, viaduct, "greenbridge"). Areas <10 km2(VO <10km2)received 0.2 " an SIV0Oof 0.0, whereasSIVO for areasof 10 to 2,500 km2 ::3 - " were calculatedfrom SIVO= (1/2490)V10 - (1/249) (Fig. 3M). The same model was used by Aste (1993). 10 kn2 2500 km2 Size of unfragmentedpart of habitat Equations Suitabilityindexes of all habitatcomponents were com- Fig. 3 (Con't.). Relationships of suitability indices (SI) and bined to value, and 13 habitat variables used in the brown bear HSI model. according importance, compensatory (A) or nature. To the of fall Spring food availability. (B) Spring food diversity. (C) limiting emphasize importance Summer food availability. (D) Summer food diversity. (E) food, we squared average of SIV5 and SIV6. Also, for Autumnfood availability. (F) Autumnfood productivity. (G) overall HSI, SIFOoDwas multiplied by 2 because of its Protective cover in foraging period. (H)Sites for denning. (I) to other habitat Distance of areas from road. Distance of greater importance compared compo- denning (J) nents. To calculate an overall we took the SI denning areas from settlement. (K) Protective cover in SI, average denning period. (L) Density of all roads. (M) Habitat for all sections. The following equations were used to fragmentation. calculate HSI: SIFooD = (((SIVa + SIv2)/2)*((SIV3 + SI4)/2)* foraging area had adequatecover (V7 < 25%), then SIV7 (((SIv5 + SIv6)/2)2))I4 (1) = (1/25)V7. When >25%of the sites had adequatecover, SICOVER= (SIV7 + SIV8)/2 (2) SIV7was assigned the value of 1.00 (Fig. 3 G). SIUMANIMPACT = (SIV9 + SIv10)/2 (3) Suitable denning areas in forests had SIV8a= 1.00; den- HSI = (2SIFoD + SIcovER+ SIHUN ACT)/4 (4) ning areasoutside forests had SIV8a=0.0 (Fig. 3 H). Den- All statisticalanalyses were performedusing SAS soft- ning areas >500 m from roads (V8b> 500 m) were ware (SAS Inst., Inc. 1989) 288 Ursus 10:1998

Applicationof the Model ows, fields, gardens,and orchardsthat comprise 18% of This model can be appliedto areaswith plant commu- total study area (Drzavnizavod za statistiku1993). nities defined by Braun-Blanquet'smethod. Otherareas We calculated availability of plant food using would requirefield samplingto determinepresence and phytocenological data (Horvat 1938; Trinajstic 1972, amountof bear food plants. 1995;Pavletic et al. 1982;Vukelic 1985) for specific plant communities for each season except winter. The loca- Sources of OtherModels tion and size of each phytocenosis came from Delnice Existing models for evaluatingbear habitatsuitability Forest Management(1994), the agency that managesthe (Aste 1993, Van Manen and Pelton 1994, Clevenger et forest in the study area. The percentof beech trees pro- al. 1995) were not found to be appropriatefor European ducing nuts was calculatedusing informationthat classi- southeasternmixed deciduous-conifer forests, but were fied crop producers(K. Postenjak,Dipling. Jastrebarslo, used as a startingpoint for constructionof the model. Croatia,pers. commun., 1994) andthat provided the num- ber of trees/hectarein each perimeterclass for each plant Applicationof the HSIin GorskiKotar community (Cestaret al. 1986). We estimatedthe den- Gorski kotar (1,796 km2)is situatedin Croatiain the sity of understoryvegetation on random450-m2 plots (n narrowestpart of the Dinarids(Fig. 1). It comprises the = 20 for each plant community) by measuringthe per- northwest end of the Dinara Mountains, which divide centage of an averagebear silhouette(95 x 130 cm) cov- Mediterraneanfrom the continentalpart of Croatiaand ered at a 15 m distancefrom east, west, north,and south. the AdriaticSea from the drainage. This method combinedthose of Cicnjak(1991) and Van The mountainsconsist mainly of carbonaterocks with Manenand Pelton (1995). The mean of these 4 measure- oldersilicates that are not waterpermeable (Horvat 1962). ments was used as percentcover for each site. For each The highest mountains are Bjelolasica (1,533 m) and plantcommunity we determinedhow manyrandom points (1,528 m). Gorski kotar has a moderatelycold had cover values >60%. Areas suitablefor denning and climate(yearly average temperature about 8C) with a rela- fragmentationwere determinedby using 1:25,000 col- tively large amountof precipitation(up to 377 cm/year) ored topographicmaps. For total lengths of main, re- and high cover, which averages 139 days/year gional, and local roads, we used data from the State (Penzar 1959). The cover types used by brown bears in Department of Statistics (Drzavni zavod za statistiku Croatiaare defined by several plant communities. For- 1993). Forforest roads lengths, we used datafrom Delnice ests cover 66% (1,191 km2) of the total area, of which Forest Management(1994). 91% is managed. Vegetationshows clear altitudinaldif- ferentiation,and Horvat(1962) described3 main vegeta- RESULTSAND DISCUSSION tion belts. Deviations from the main zonality are varied Plants belonging to generaAllium, Arum, and Rumex by differentexposure, slope, wind, and other ecological representedspring food availabilityand provided an over- factors. Elevations from 0 m to 850 m by the Adriatic all 4.94% cover. Averagingthe SIV,values for each sec- Sea are covered with sub-Mediterraneanthermophil for- tion yielded an overall SIVI= 0.48. (Table 1). The ests belonging to Ostryo-Carpinion orientalis group. minimum % cover of spring food (1.38) was recorded Deciduous and mixed forests of Fagion illyricumgroup for habitatfacing the AdriaticSea (),whereas the cover the largestpart of Gorskikotar stretching from 900 maximumof 8.77% cover was found at lower elevations m (600 m on continentalside) to 1400 m above sea level. on the continentalside (Vrbovsko). In all sections, only Dwarf mountainpine (Pinetummugi croaticum)is found 1 plant species (lords and ladies) was present with 10% above 1400 m (Horvat 1962). coverage, which resultedin SIV2= 0.50 (Table 1) for the Gorski kotar is divided into 4 political units: Cabar, entire study area for spring food diversity. Delnice, Vrbovsko, and Rijeka (Fig. 1). These are also Summerfood covered >10% of each habitatsection (x forestmanagement units andour studyarea sections. The = 25.21%) giving an overall SIV3= 1.00 (Table 1). The area is inhabitedby 236,000 people in 308 settlements, lowest abundanceof food was again found near the sea- but the majority (74%) live in 3 towns on the Adriatic coast comparedto higher continental areas. In all habi- coast (i.e., in the 10% of land area that is marginalbear tat sections, we found >1 (max = 10) summer food habitat). The rest of the area (90%) is mountainousand plants, resulting in SIV4= 1.00 for summer food diver- is inhabitedby 24 humans/km2.A majorityof local people sity (Table 1). workin forestry(9.5%) and related industries (48%). Only Beech, the only importantproducer of fall food in the 1.4%work in agricultureon 331 km2of pastures,mead- study area,covered 55.92% of forested area,resulting in BROWNBEAR HABITAT iNIN CROATIA - Kusak and Huber 289

Table1. Suitabilityindices (SI)of brownbear habitat variables for Gorskikotar, Croatia, 1995.

Suitabilityindices (SI)

Section SIva SI2 b S Si SVV4 d e SIv6 f Si7 iv8 h S SIV9 i Delnice 0.47 0.50 1.00 1.00 0.45 0.27 0.19 0.00 0.00 0.28 Eabar 0.44 0.50 1.00 1.00 0.64 0.27 0.06 0.00 0.00 0.11 Rijeka 0.12 0.50 1.00 1.00 0.57 0.31 0.20 0.16 0.00 0.21 Vrbovsko 0.88 0.50 1.00 1.00 0.59 0.25 0.10 0.00 0.00 0.11 Overall 0.48 0.50 1.00 1.00 0.56 0.28 0.14 0.04 0.00 1.00 a spring food availability. b spring food diversity. c summer food availability. d summer food diversity. e autumn food availability. f autumn food productivity. g protective cover in foraging period. h protective cover in denning period. i density of all roads. J habitat fragmentation. a fall food availabilityof SIv5= 0.56 (Table 1). An aver- Only 2.87% of the study area had horizontalprotec- age 11.04%of all beech trees were capableof producing tive cover with a density >60%. The averageSIv7 for all nuts, for a suitabilityindex for fall food productivityof sections resultedin an overall SIV7= 0.14 (Table 1). All SIv6=0.28 (Table 1). Privateforests (20% of all forests), areas with low horizontalvisibility were found in beech which are situated mostly aroundvillages, lack mature forests on rugged terrainwith steep slopes and abundant beech trees and conifers. There were no significant dif- runningwater. Low visibility was relatedmore to terrain ferences among study area sections in food availability configurationthan to vegetation density. in differentseasons (P > 0.05). The overall area providing suitable den sites was Bear movements are related to food availability 19.56%;the averageSIV8 for all sections was 0.04 (Table (Craigheadet al. 1982, Cicnjak1991, Hameret al. 1991). 1). The numberand size of available sites varied within Home range sizes of brownbears in Croatiado not differ areas. The Delnice, Cabar, and Vrbovsko sections all among seasons except winter (Huber and Roth 1993). had SIV8= 0.0 (Table 1). The Rijeka section had an SIV8 This may be explainedby year-roundbalanced food avail- = 0.16 (Table 1); it was the only section with SIV8> 0.0. abilityin our studyarea. Springfood hadthe lowest abun- The Rijeka section had 88.76 km2 (37.09%) of suitable dance in Rijeka section, near the , whereas denning habitat, which differed significantly (P = lower continentalareas providedthe best spring habitat. 0.00046, ANOVA [analysisof variance])from othersec- Fall food availabilitywas good throughoutthe entirestudy tions. RisnjakNational Park (16.90 km2)and the section area because beech trees covered about 50% of all for- of Rijeka (31.55 km2)that borderedthe park were the 2 ested surfaces. The limiting factor seemed to be beech biggest contiguousdenning areas in the northwesternpart productivity. Management practices have resulted in of study area. 11.04% of beech trees at productive ages, decreasing Availabilityof suitabledenning sites was limitedmainly the suitability of bear habitat. Clearcuts, which are by the presence of roads, especially forest roads. The sometimes made as circular openings or clearcuts of numberand size of suitable denning areas can indicate entire slopes, may provide good summer possible habitat fragmentationin the future. Bears in foods during early succession stages. Replanting such Croatiaden in areas that we indicated as unsuitable,but areas with spruce, which suppresses most understory there is a higherpossibility of disturbance.Mild winters vegetation, will not promote good bear habitat for the and early springswith little or no snow cover are regular future. Permanent bear feeding sites, which offered periodsfor logging beech trees. At least 3 maternaldens carrion and slaughterhouse refuse, are visited mostly in 1993-94 were disturbeddue to logging and hunting. in spring (A. Frkovic, Croatian Forests, Delnice, Areas identified as suitable for denning should be ex- Croatia, pers. commun., 1993), but there is no data on cluded from logging duringwinter until late April if ma- quantities of food eaten by bears. ternaldens need to be protected. 290 Ursus 10:1998

The density of all roadsin the study areawas 1.91 km/ Table2. Suitabilityindices (SI) for habitatcomponents and km2, in 0.0 The mean habitatsuitability indices (HSI)for Gorskikotar, Croatia, resulting SIV9= (Table 1). density 1995. of all roadsin Delnice, Eabar,Rijeka, and Vrbovskowas 2.07, 1.95, 1.74, and 1.89 km/km2. respectively,resulting Section SIFOoD SICOVER SIHUMAN IMPACT HSI in SIV9= 0.0 for all sections (Table 1). This density does not brown bears from the area. In the Delnice 0.50 0.10 0.14 0.31 prevent inhabiting Eabar 0.56 0.03 0.06 0.30 CantabrianMountains of Spain,areas inhabited by brown Rijeka 0.49 0.18 0.11 0.32 bearshave an averageroads density of 0.71 km/km2while Vrbovsko 0.59 0.05 0.06 0.32 areas uninhabitedby bears have a mean road density of Overall 0.55 0.09 0.50 0.42 0.85 km/km2(Clevenger et al. 1995). Road density indi- cates possible humanaccess into bearhabitat which could ing of our model is needed to examine this assumptionas affect the bears' security (i.e., hunting, forestry opera- it relates to impacts on denning and mortality. tions, traffic caused mortality,tourism, and other activi- We conclude thatbrown bear habitat in Gorskikotar is ties). By excluding forest roads, the density of main, still in good conditionand connectedwith the remaining regional, and local roads droppedto 0.83 km/km2. The bearrange on Dinaramountain. Only moderatemanage- highest density of non-forestroads was found in Rijeka, ment changes are necessaryto maintainand improvethe (1.31 km/km2),followed by Vrbovsko (0.72 km/km2), currentstate. Proposedshort-term measures that would Delnice (0.71 km/km2),and Eabar(0.59 km/km2. Clos- have immediateeffects are closing existing forest roads, ing forest roads would increase SIV9to 0.45, SIHUMANIMPACT restrictingconstruction of new roads, and avoidingforest to 0.73, and the overall HSI to 0.48. It would also in- activities in denningareas during denning season. Long- crease the area of suitabledenning sites. term measuresinclude modifying managementto mini- Habitatfragmentation did not occur on any section or mize reforestation with spruce in beech-fir areas and on the 1,796-km2of study area. The SIVlOvalues for each ensuringproper garbage management. of the study area sections and for the entire study area were limited only by the sizes of the areas. Gorskikotar is connectedwith the rest of bearrange in Croatia(9,800 LITERATURECITED km2), which is a part of contiguous brown bear range ASTE,C. 1993. Habitatqualitatfur braunbiirenin Osterreich. Holzwirtschaft stretchingfrom Slovenian thorough Croatian,Bosnian, Diplomarbeitder Studienrichtung Forst-und verfabt am Institut fir und and Monte Negro partsof the DinaraMountains (Huber Wildbiologie Jagdwirtschaft Universitatfur Bodenkultur,Wien, Austria. 95pp. (In and Frkovic 1993). Because this range is partof a much German.) area,we 1.00 (Table 1). The main larger assigned SIVIO= CESTAR, D., V. HREN, Z. KOVAEEVIC,J. MARTINOVI,, AND Z. road stretched the middle of and the through study area, PELCER. 1986. Uputstva za izradu karte ekolosko- majorityof settlementswere located along it. The 5-km gospodarskihtipova gorskogpodrueja (I) SR Hrvatske. belt stretchingalong the main road still had 66% forest Radovi21(4). 125pp. (InCroatian.) vegetation. At least 3 bear crossing sites were identified CICNJAK,L. 1991. Foodhabits and habitat use by European along the road (Huberet al. 1998). A new highway un- brown bear in Croatia, Yugoslavia. M.S. Thesis, Univ. der constructionwill connect the AdriaticSea coast with Wisconsin, Madison. 88pp. F.J. ANDM.A. CAMPOS.1995. Brown the continental part of Croatia, through the middle of CLEVENGER,A.P., PURROY, bear habitatevaluation in the CantabrianMountains, Spain. Gorski kotar. If all measures (D. Huber,Vet- proposed Int.Conf. Bear. Res. and 9(1):165-178. are Manage. erinaryFaculty, Zagreb, Croatia, unpubl. data.) imple- , , ANDM.R. PELTON.1992a. Brown bear (Ursus cut. mented, existing bear corridorswill not be arctos L.) habitatuse in the CantabrianMountains, Spain. The final SI values for the food, cover and humanim- Mammalia56:203-214. pact habitat components were 0.55, 0.09, and 0.50, re- , AND . 1992b. Food habits of brown spectively, resultingin an overall HSI of 0.42 (Table 2). bears (Ursus arctos) in the Cantabrianmountains, Spain. J. The HSI evaluationindicated a good food base, especially . 73:415-421. 1982. A summer,below averageprotective cover, and av- CRAIGHEAD, J.J., J.S. SUMNER, AND G.B. SCAGGS. during habitatand other of humans. The overall HSI indicatedaver- definitivesystem for analysisof grizzlybear erage impact wildernessresources. Wildlife-WildlandsInst. 1, of bear habitatin Gorski kotar,Croatia. Monogr. age quality Univ. of MontanaFound., Missoula. 279pp. Our model assumes that roads have a open big impact DELNICEFOREST MANAGEMENT. 1994. Area of management units As we on brown bear habitat. noted, simply excluding by EGT. Delnice Forest Manage., Delnice, Croatia. 3pp. forest roadsraises HSI values. Field validationand test- BROWNBEAR HABITAT IN CROATIA* Kusak and Huber 291

(In Croatian.) 1994. Mitochondrial DNA sequence divergence among DRZAVNIZAVOD ZA STATISTIKU.1993. Statistieki Ijetopis 1992. some west Europeanbrown bear (Ursus arctos) populations; RepublikaHrvatska, Drzavni zavod za statistiku. Zagreb, lessons for conservation. Heredity73:480-489. Croatia. 516-531. (In Croatian.) REGENT,B. 1980. Sumsko sjemenarstvo. Dokumentacijaza FRKOVIC,A., R. RUFF,L. CICNJAK,AND D. HUBER.1987. Brown tehnikui tehnologijuu sumarstvu.Broj:79. JPSC, Beograd, bear mortalityduring 1946-85 in Gorskikotar, Yugoslavia. Yugoslavia. (In Serbian.) Int. Conf. Bear Res. and Manage. 7:87-92 ROTH,H.U., ANDD. HUBER.1986. Diel activity of brownbears HAMER,D., S. HERERRO,AND K. BRADLY.1991. Food and habitat in Plitvice Lakes NationalPark, Yugoslavia. Int. Conf. Bear used by grizzly bears, Ursus arctos, along the continental Res. and Manage. 6:177-181. divide in WatertonLakes NationalPark, Alberta. Can.Field SAS INSTITUTE,INC. 1989. SAS/STAT User's Guide. Version Nat. 105:325-329. 6. SAS Inst., Cary, N.C. 943pp. HORVAT,I. 1938. Biljnosocioloska istrazivanja Suma u SCHOOLEY,R.L., C.R. MCLAUGHLIN,G.J. MATULA, JR., ANDW.B. Hrvatskoj. Glasnik za sumske pokuse 6:127-279. (In KROHN.1994. Denning chronology of female black bears: Croatian.) effects of food, weather, and reproduction. J. Mammal. . 1962. Vegetacija planina zapadne Hrvatske. Acta 75:466-477. biologica 2. Prirodoslovnaistrazivanja JAZU 30. Zagreb, SERVHEEN,C. 1990. The statusand conservationof the bearsof Croatia. 179pp. (In Croatian.) the world. Int. Conf. BearRes. andManage. Monogr. Series HUBERD., ANDA. FRKOVIC.1993. Brown bear managementin No. 2. 32pp. Croatia. Pages 287-292 in I. Thompson,ed. Proc. 21st. Int. TABERLET,P., AND J. BOUVET. 1994. Mitochondrial DNA Union of GameBiologists Congr.,Zoetermeer, Netherlands. polymorphism,phylogeography, and conservationgenetics , J. KUSAK,AND A. FRKOVIC.1998. Traffickills of brown of the brown bear Ursus arctos in Europe. Proc. R. Soc. bears in Gorski kotar, Croatia. Ursus 10:167-171. Lond. B. 255:195-200. , ANDS. MORIC.1989. Stete od medvjedau Jugoslaviji. TRINAJSTIC,I. 1972. Fitocenoloska istrazivanja bukovih suma Zbomik treceg simpozija Savremenipravci uzgoja divljaci. Gorskogkotara. Acta Bot. Croat.31:173-180. (In Croatian.) Savez veterinara i veterinarskih tehnicara Jugoslavije, .1995. Virgin forest,natural forest and exploitedforest. Beograd:197-202. (In Croatian.) A comparisonof the floristical structure. Sauteria6:109- , ANDH.U. ROTH.1993. Movementsof Europeanbrown 132. bears in Croatia. Acta Theriologica38(2):51-159. U.S. FISH ANDWILDLIFE SERVICE. 1981. Standards for the ,AND . 1995. Denning of brown bears in Croatia. developmentof habitatsuitability index models (103 EMS). Int. Conf. Bear. Res. and Manage. 9(1):271-282. U.S. Dep. Inter. Fish and Wildl. Serv., Div. Ecol. Serv., KACZENSKY,P., F. KNAUER,T. HUBER,M. JONOSOVIC,AND M. Washington,D.C. 162pp. ADAMIC.1996. The Ljubljana- Postojnahighway-a deadly VAN MANEN,F., ANDM. PELTON.1993. Data-based modeling barrierfor brownbears in Slovenia? J. Wildl. Res. 1(4):263- of black bear habitat using GIS Pages 323-329 in I. 267. Thompson, ed. Proc. 21st. InternationalUnion of Game PAVLETIC,Z., I. TRINAJSTIC,AND I. SUGAR. 1982. Die Biologists Congr., Zoetermeer,Netherlands. warmeliebenden hopfenbuchen-buchenwalder (Ostryo- , AND . 1995. Habitat evaluation to study the FagetumWraber) in nordwestKroatien. Studia Geobotanica feasibility of reestablishing a black bear population. Int. 2:15-19. (In German.) Conf. Bear Res. and Manage. 9(1):438-457. PENZAR,B. 1959. Razdiobe godisnjih kolicina oborina u VUKELIC,J. 1985. Doprinos fotointerpretacijske analize Gorskom kotaru. Rasprave i prikazi MZH-a, 4. Zagreb, vegetacijskomistrazivanju Sumskih zajednica Nacionalnog Croatia. (In Croatian.) parka Risnjak. Glasnik za sum. pokuse 23:95-140. (In RANDI, E., L. GENTILE,G. BOSCAGLI,D. HUBER,AND H.U. ROTH. Croatian.)