Consumption of Fungal Sporocarps By

CONSUMPTIONOF FUNGALSPOROCARPS BY YELLOWSTONE GRIZZLY BEARS

DAVIDJ. MATTSON,U.S. GeologicalSurvey Forestand RangelandEcosystem Science Center,Colorado Plateau FieldStation, P.O. Box5614, Northern Arizona University, Flagstaff, AZ 86011-5614, USA, email: [email protected] SHANNONR. PODRUZNY,U.S. Geological Survey Northern Rocky Mountain Science Center, Interagency Grizzly Bear Study Team,Forestry Sciences Lab, Montana State University, Bozeman, MT 59717, USA, email: [email protected] MARKA. HAROLDSON, U.S. Geological Survey Northern Rocky Mountain Science Center, Interagency Grizzly Bear Study Team, ForestrySciences Lab, Montana State University, Bozeman, MT 59717, USA, email: [email protected]

Abstact: Sign of grizzlybears (Ursus arctos horribilis) consuming fungal sporocarps (mushrooms and truffles) was observedon 68 occasions duringa studyof radiomarkedbears in the Yellowstoneregion, 1977-96. Sporocarpsalso weredetected in 96 grizzlybear feces. Mostfungi consumedby Yellowstone'sgrizzly bears were members of theBoletaceae (Suillus spp.), Russulaceae (Russula spp. and Lactarius sp.), Morchellaceae (Morchellaelata), and Rhizopogonaceae. Consumption of false truffles(Rhizopogon spp.) was indicatedby excavationsthat were deeper, on average(1.1 dm),than excavations for mushrooms (0.6 dm). Consumptionof sporocarpswas most frequent during September (7% of all activity), althoughmedian numbers of sporocarpsexcavated at feedingsites peaked during both August and September (22-23 excavations/site).Almost all consumption(75%) occurred on edaphicallyharsh sites typically dominated by lodgepolepine (Pinus contorta). At broadscales, consumption of sporocarpswas mostlikely where these types of lodgepolepine-dominated sites wereextensive or wherehigh-elevation sites supportingmature whitebarkpine (P. albicaulis) were rare. The number of sporocarpsexcavated at a feedingsite was greatest when cone crops of whitebarkpine were smalland in standswith abundant lodgepole pine. At finescales, consumption of fungiwas positively associated with lodgepole pine basal area and negativelyassociated with total ground vegetation cover. Becauseof thestrong association of sporocarpconsumption with lodgepole pine and its disassociationat broad scales with availability of whitebarkpine seeds, consumption of mushroomsand truffles by grizzlybears will likelyincrease in theYellowstone ecosystem with global warming. Lodgepole pine is predictedto increaseand whitebark pine to declinewith global warming.

Ursus13:95-103 (2002)

Keywords: foraging behavior, fungi, grizzly bears, habitat selection, mushrooms, sporocarps, Ursus arctos, Yellowstone

Hypogeous and epigeous sporocarpsare often referred kistan (Roberts 1977), and Hokkaido (Odachi and Aoi to as truffles and mushrooms,respectively, and are com- 1987). In North America, heavy consumptionof fungi mon reproductivestructures of mycorrhizalfungi in tem- was observed among black bears in coastal Washington perate and boreal forests. These spore-producing state (Poelkerand Hartwell 1973). Among grizzly bears, structuresare eaten andefficiently digestedby omnivores morethan incidental consumption has been recordedonly such as humans and feral swine (Sus scrofa; Groot in the Yellowstone region (Mealey 1975, Mattson et al. Bruinderinket al. 1994). Moreover,fungal sporocarps 1991, Mattson 1997a). are present in the diets of numerousotherwise herbivo- Grizzly bearsin the Yellowstoneregion obtainmost of rous or frugivorous mammals including wapiti (Cervus theirenergy from seeds of whitebarkpine (Mattsonet al. Nelson and elaphus; Leege 1982, Collins and Urness 2001) or tissue of bison (Bisonbison) and wapiti(Mattson and tree 1983) squirrels(Sciurus spp., Tamiasciurusspp., 1997b),although there are marked differences in consump- and Glaucomysspp.; Hall 1981, Carey 1995, Steele 1998). tion of pine seeds andungulate tissue among bearsof dif- of animalsis Consumption fungi by consistentwith the ferentage andgender (Mattson 2000). Yellowstonegrizzly abundanceand nutritionalcharacteristics of mushrooms bears consume most pine seeds duringfall, primarilyby and truffles. Conifer forests of interiorNorth America digging whitebarkpine cones from ground-levelcaches can produce 10-17 kg of mushroomsper ha, dry weight, madeby red squirrels(Tamiasciurus hudsonicus; Mattson of 15-35% crude comprised protein(Mealey 1975, Fogel et al. 2001). Tissue is obtainedfrom ungulatesprimarily and This Trappe 1978). relatively high concentrationof during spring by scavenging winter-killed animals is protein complementedby a similarly high concentra- (Mattson 1997b). When pine seeds or carrionare scarce, tion of energy (4-6 kcal/g, dry weight), comparableto Yellowstone's grizzly bears eat diverse other foods, in- thatof most hardand soft fruits(Fogel and Trappe1978). cluding fungal sporocarps(Mattson et al. 1991). Based Some edible mushroomscontain as little as 13-18% total on field observationsof radiomarkedgrizzly bears, we fiber which dietary (Cheung 1997), equates to about 76- speculatedthat consumptionof sporocarpswas concen- 83% matter for and dry digestibility grizzly NorthAmeri- tratedin the fall and closely associated with pine trees in can black bears (U. americanus;Pritchard and Robbins forest standswith sparseground cover. 1990). In this paperwe describefeatures associated with con- The few recordsof ursids trufflesor mush- consuming sumption of fungal sporocarpsby Yellowstone grizzly rooms are limited to eitherentries in tables summarizing bears. Because virtuallynothing is known about exploi- fecal contents or to short anecdotes. Most observations tation of mushroomsand truffles by bears, the come from analysis Eurasia,including the Alps (Couturier1954), reportedhere is exploratory.We evaluatedthe effects of EuropeanRussia 1931, Novikov et al. Pa- (Ognev 1969), forest stand structure,date, type of bear,and availability 96 Ursus 13:2002 of known high-qualityfoods (i.e., pine seeds and ungu- radius forest inventoryplot at the center of grizzly bear late carrion) using data collected during a study of activity. Treeswere tallied,identified as live or dead, and radiomarkedanimals, 1977-96. the diameterof each was measuredat 1.4 m above ground. Percentground cover also was ocularly estimatedfor all graminoids,forbs, and shrubswithin about 10 m of plot STUDYAREA center. Each site was describedin terms of habitattypes The approximately23,000 km2study area, from 43?30" describedby Mattsonet al. (1999). There were 10 forest to 45? 15"Nlatitude and 109?30"to 111?30"Wlongitude, habitat types, including 2 that characterized recently correspondedto the known range of Yellowstone's griz- burnedor harvestedsites (Table 1). zly bearpopulation. Most of the areaoccupied by grizzly Field crews described all feeding signs and collected bears was >2100 m elevation and consisted of remote all feces (scats) encounteredduring 1977-96. During mountainsand plateaussurrounded by valleys and plains 1986-96 descriptionsof feeding activity includedcounts more intensivelysettled or used by humans. Annualtem- and measurementsof excavations. For excavations of peraturesaveraged about 0?C. Monthlyaverages ranged fungal sporocarps,length, width, and depth (all in dm) from -2?C to 13?Cduring April throughOctober. Tem- were measuredand multipliedto estimateexcavated vol- peraturesrarely exceeded 27?C in most areas. Precipita- ume. Excavationsof sporocarpsby bearswere identified tion variedin amountand timing with elevationand region by several criteriaincluding the presence of sporocarps, and fell mostly as snow that reached accumulationsof the absenceof otherpotential foods, the presenceof feces 20-260 cm, depending on location, before melting dur- containingsporocarp remains, the presenceof othergriz- ing March-June(Dirks and Martner1982). zly bear signs, and excavationdimensions. Comparedto About 75% of the study areawas forested. Lodgepole excavationsfor sporocarpsby otheranimals such as squir- pine was the most common dominant tree species, al- rels and wapiti, those by grizzly bears were the largest, though whitebarkpine was abundantabove 2,500 m el- averaging about 1.6 dm in width and 2.0 dm in length. evation (Despain 1990). Forest structure varied Sporocarpsjudged to be the same type as those consumed considerablyduring the study primarilyowing to mortal- by the bear were described and identified using a field ity of trees caused by fire andby epidemic populationsof key synthesizedfrom Smith(1975) and Miller (1980). In mountainpine beetle (Dendroctonusponderosae; Despain some instances, sporocarpswere collected and keyed in 1990). The largest fire occurredin 1988 and burnedap- the lab. The field key emphasizedvisible featuressuch as proximately560,000 ha. Craigheadet al. (1995) provide color, pattern,shape, size, structuresof the cap or stipe, additionaldetails aboutthe study area. site type, and whether the sporocarpwas epi- or hypogeous. The concurrentcollection of scats and informationon METHODS feeding activity maximized comparabilityof data from Grizzly bears were trapped,marked, and radiolocated these 2 sources at broad scales. We did not ascribe scat accordingto methodsdescribed by Knightand Eberhardt contents to specific feeding sites because foods found in (1985) and Blanchardand Knight (1991). A subset of scats were often not consumed at the site of collection radiolocations was visited and described according to given the 7-13 hr transittime of digesta (Pritchardand methods in Mattson(1997a, 2000). Field crews also de- Robbins 1990). Thus, we compareddata from scats and scribedsites where grizzly bearfeeding or bedding signs feeding sites at the scale of monthsbut not specific feed- were encountereden route to and from telemetry loca- ing sites. tions. At each location, field crews located a variable- Scats were collected, dried, and analyzed for percent

Table 1. Forest habitat types used to map grizzly bear habitat and describe grizzly bear activity sites in the Yellowstone region. Acronym Habitat type description

DOPEN Dry sites recently deforestedby timberharvest or fire HABLA High-elevationsubalpine fir (Abies lasiocarpa) climax sites; typified by Arnica spp. and Juniperuscommunis HPIAL High-elevationforested sites with maturewhitebark pine communis, HPSME High-elevationDouglas-fir (Pseudotsuga menziesii) climax sites; typified by Arnica cordifolia, Berberis repens,Juniperus and Symphoricarposoreophilus HVASC High-elevationsites, groundlayer dominatedby grouse whortleberry(Vaccinium scoparium) LPICO Low-elevation lodgepole pine-dominatedsites; typified by Carex geyeri, C. rossii, Calamagrostisrubescens, and Purshia tridentata MOPEN Mesic-wet sites recently deforestedby timberharvest or fire LPIEN Low-elevation Engelmannspruce (Picea engelmannii)-dominatedsites; typified by Calamagrostiscanadensis, Equisetumarvense, and Galium triflorum MABLA Mesic-wet subalpinefir climax sites; typified by Thalictrumoccidentale and Osmorhizachilensis MPSME Mesic Douglas-fir climax sites; typified by Symphoricarposalbus, Spiraea betulifolia,and Calamagrostisrubescens GRIZZLYBEARS AND SPOROCARPS* Mattson et al. 97 content accordingto methodsdescribed by Mattsonet al. bison in the ecosystem were taken from U.S. National (1991). When collecting scats we triedto minimize con- ParkService (1997). Monthlytemperatures (?C) andpre- taminationwith extraneoussoil anddebris. However,this cipitation(cm) were takenfrom regionalsummaries pub- was often impossible when collecting liquid feces typical lished by the U.S. National Oceanic and Atmospheric of those containingmushroom remains. Scats were air- Administration(1977-92). Grizzlybear activity sites were dried and later rehydrated before washing through 2 also associated with the spatialextent of habitattypes in screens. Coarsematerial was retainedby the large (0.806 surroundinglife-range-size areas on the basis of the Bear cm2) mesh screen and fine materialby the small (0.212 ManagementUnits (BMUs) where they occurred(there cm2)mesh screen. Diet items were identifiedto the finest are 18 BMUs in the Yellowstone ecosystem, each about taxonomicresolution possible and the percentvolume of the size of a female life range). Mattson(2000) provides each was visually estimated. Remains of fungal sporo- a moredetailed description of these distalfactors and their carpswere typicallyidentifiable as such, to the extentthat sources. stipes could sometimes be distinguishedfrom caps, but almost never to the extent that genera or even families could be identified. Visual estimates of percentvolume RESULTS were routinelyverified by liquiddisplacement in beakers. During 1977-96, 7,459 scats were collected and 2,134 Resultswere reported by monthand year as totalfrequency telemetrylocations from 175 radiomarkedbears were in- of occurrencein scats, percentof total fecal volume, and vestigatedto determineactivity of the associatedanimal. mean percentfor scats in which the item occurred. Feeding sign was found at 3,101 sites, including 1,940 We used logistic regressionanalysis (Demaris 1992) to sites not associated with a telemetry location. Remains specify the effects of explanatoryvariables on the log odds of fungal sporocarpswere found in 96 scats, whereassign that a bear had excavated fungal sporocarpsat a given of foraging for sporocarpswas found at 68 sites, 48 of location. Radiotelemetrylocations or feeding sites were which were associatedwith telemetrylocations of 28 dif- units of analysis. We specified 2 types of models: one ferentradiomarked bears. Feeding sign took the form of using only data from telemetrylocations, including sites shallow scrapesto more substantialpaw-sized digs in for- where no feeding signs had been found, and the otherus- est litterand soil. Scats containingsporocarps were typi- ing only data from sites with feeding signs of all types, cally grayish in color and very liquid. but including plots not at telemetry locations. We gave Of the scats containing fungal sporocarps,80% con- priorityto the first type of model and used the second to tainedadditional items. Of these mixed-compositionscats, confirm patterns. Given that excavation of sporocarps 81% containeddirt and debristhat accounted for an aver- hadoccurred, we used multiplelinear regression (Weisberg age 30% of total scat volume. Remains of grasses and 1985) to specify the effects of explanatoryvariables on sedges werenext (32%)and accounted for an average15% the total numberof excavations. We used maximumlike- of total scat volume. All otheritems occurredin <15% of lihood methods for parameterestimation and the sample mixed-compositionscats and comprised<5% of average size-adjustedversion of Akaike's InformationCriterion scat volume. (AICc;Burnham and Anderson 1998) for model selec- Field crews commentedon the type of sporocarpmost tion. We used change-in-AICc(A), obtainedby deleting likely eaten by bears at 48 of the sites where it was con- each variablein turnwhile holding the rest of the model cluded that grizzly bears ate sporocarps. A likely hypo- constant,to quantifythe relativeeffect of each variablein geous food ("truffles")was describedat 5 sites. Based on a given model (Burnhamand Anderson 1998). Because 1 positive identification,these below-groundsporocarps the value of AICCis affected by sample size, it is compa- were most likely false truffles of the genus Rhizopogon. rable within but not among models. We de-emphasized Epigeous sporocarps(i.e., "mushrooms")were described statisticalhypothesis testing (Johnson 1999) and present at 35 sites, including 9 sites with gilled mushrooms, 7 P-values solely as confirmatoryinformation. sites with boletes (Boletaceae), 4 sites with morels We considered both proximal and distal effects in the (Morchellaelata), and 1 site with puffballs(Calvatia sp.). analysis (Mattson2000). Distal effects were those opera- Three of the probableinstances where morels had been tional at broad temporaland spatial scales; proximal ef- consumed occurredduring 1989 in recently burnedfor- fects were operationalat the scale of the immediate site. ests, 1 year after the extensive wildfires of 1988. The Proximal effects were representedby measures taken at boletes were most likely of the genus Suillus, either S. telemetrylocations or other sites of grizzly bear activity. brevipesor S. tomentosus.The gilled mushroomswere in Distal effects were enumeratedfrom othersources. Num- 5 instancesmost likely of the genusRussula, in 2 instances bers of cones on whitebark pine trees were taken from of the genus Tricholoma,and in 1 instance of the genus counts at fixed transects (Mattson 2000). Numbers of Lactarius. 98 Ursus 13:2002

Excavationsfor false trufflesand mushroomsdiffered. that grizzly bears consumed sporocarpsprior to June. At the 11 sites where depths of excavations were mea- Consumptionof sporocarpswas not evenly distributed sured,those for false truffles(x = 1.1 dm, SE = 0.2, n = 4) among habitattypes. Most consumption(89%) was ob- were deeper thanthose for the bases of mushroomstipes served in just 3 forest types (the LPICO [55%], HVASC (often describedas "scrapes",x = 0.6 dm, SE = 0.1, n = 7; [20%], and HPIAL [14%; Table 1]), whereas none was F = 16.0, 1,9 df, P = 0.003). Even so, volumes of indi- observed in non-foresttypes. Consideringonly forested vidual excavationsfor trufflesand mushrooms,averaged or potentiallyforested habitat types, sporocarpconsump- by feeding site, did not differ substantially(n = 10, F = tion was distributeddifferently from all othergrizzly bear 1.8, 1,8 df, P = 0.22). feeding or bedding sign (n = 1,742, 9 df, homogeneityG2 Consumptionof fungalsporocarps by grizzly bearswas = 121.2, P < 0.0001). Consumptionwas concentrated not equallydistributed among months. Frequencyof sign more than expected in the LPICOtype and less than ex- in scats and at feeding sites peakedduring September and pected in HABLA, MOPEN, DOPEN, MPSME, and was relatively common duringAugust (Fig. 1). Median HPSME (Fig. 2; based on Bonferronisimultaneous con- numberof digs per feeding site and mean percentof indi- fidence intervals for differences between observed and vidual scats comprisedof sporocarpremains, considering expectedproportions with a = 0.05). Frequenciesof spo- only scats where such remainswere found, also peakedin rocarpconsumption in the HVASC,HPIAL, HABLA, and August and September,suggesting that meal sizes were LPIENtypes wereconsistent with whatwould be expected larger and foraging bouts longer during these months. by frequenciesof otherfeeding activitiesby bearsin these Evidence of sporocarpconsumption from scats and feed- habitattypes. ing sites differed only in that sign from sites indicated Consideringonly distal factors,consumption of sporo- comparatively common consumption during October carps was relatedto a polynomial of Juliandate for both whereassign from scats did not. With the exception of 2 telemetrylocations andfeeding sites (Tables2 and 3; Fig. scats collected duringMay 1994, there was no evidence 3a). The firstderivatives of these polynomials(univariate) suggested that probabilityof use peaked at about 0.05- 0.06 during Julian dates 264 and 266 (about 21 and 23 September)for telemetrylocations and feeding sites, respectively. Otherwise, analysis of telemetry locations suggestedthat probability of consumptionwas positively

0.6 -

( 0.5 - en- "-' 0.4 -

. 0.3 - o < - 0.2- 0.0 L<

Fig. 1. Monthlyoccurrence of sporocarpconsumption by Yellowstone as indicated scats grizzlybears, 1977-96, by (a) Habitat type and (b) feeding sites. In (a) bars (? 1 SE) indicatethe proportionof scats that containedremains of sporocarps Fig. 2. Proportionaloccurrence of feeding on fungal (left scale) and solid dots indicate the mean percent of sporocarpsby Yellowstonegrizzly bears (black bars) versus individualscats comprisedof sporocarpremains, considering all other feeding and bedding activity (hatched bars) in only scats wheresuch remainswere found (right scale). In forested habitattypes, 1977-96. Greaterthan sign (>) (b) bars indicatethe proportionof feedings sites where signifys habitattypes where consumptionof sporocarps evidence of sporocarpconsumption was found(left scale) was muchmore frequent than expected by the occurrenceof and largesolid dots the mediannumber of excavationsfor all other grizzly bear activity (Bonferroniconfidence sporocarpsat feeding sites (rightscale). The small dots intervals,a = 0.05);less thansigns (<) signifytypes where above and below the medianfor August and September consumption was much less frequent than expected. denotethe 75%and 25%quartiles, respectively. Habitattypes are describedin Table1. GRIZZLYBEARS AND SPOROCARPS* Mattson et al. 99

LPICOand HVASC, considering both telemetry locations (-0.49 and -0.46, respectively) and feeding sites (-0.44 and -0.39, respectively). By contrast,based on feeding sites, the extent of bison winterrange was positively correlatedwith the extent of LPICOand HVASC (0.56 and 0.63, respectively). Moreover,the log odds that grizzly bears had fed on bison duringAugust and September(y) was positively associatedwith the extent of bison winter range (x) in the surroundingBMU:

y = -8.6 + 0.37x

with n = 756, R2 = 0.94, 7 df, G2= 2.3, P = 0.94. Consideringdistal and proximalfactors together,both telemetry locations and feeding sites showed that consumptionof fungalsporocarps was morelikely wherethere was sparse groundcover, abundantpines, gentle slopes, and extensive LPICO habitats in the surroundingarea (Tables2 and 3; Fig 3b). Of all effects, the positive one of lodgepole pine was the greatest, especially when com- bined with the negative effect of groundcover (Fig. 3c). The effect of Juliandate also persisted. Analysis of feeding sites furthersuggested that the log odds of sporocarp consumptionwas negatively correlatedwith amountsof total forest basal area, live and dead, offset by a positive correlationwith amountsof live basalarea. In otherwords, greatertotal basal area had an inhibiting effect that was less pronouncedif comprisedof live treesrather than dead ones. Fig. 3. Univariate relations between the proportion of sites where Yellowstone grizzly bears consumed fungal Not all types of grizzly bearsappeared equally likely to sporocarps, 1977-96, and distal or proximalfactors with the consume fungal sporocarps.Considering only distal fac- greatest apparent effects on this behavior: (a) Julian date, tors, subadultfemales (autonomousbears <5 years old (b) extent of the LPICO(circles) and HPIAL(squares) habitat without were more than other of in the bear unit and and cubs) likely any type types surrounding management (BMU), sub- (c) lodgepole pine basal area and total ground cover. In (a) bearto consume sporocarps,especially in contrastto and (b) the symbols and SE bars are for proportions of adult males (Table 2). This effect persisted and was ac- quintiles of the data and illustrate goodness-of-fit centuatedwhen considering proximal effects (Table 3). The mesh in is the relation estimated (univariate). gray (c) In otherwords, even at times and when proximal from telemetry locations, the open mesh the relation optimal estimated from feeding sites. featureswere most favorable,subadult females apparently were more likely than any other type of bear to consume correlatedwith size of whitebarkpine cone crops (Table sporocarps. 2), whereas analysis of feeding sites suggested that con- Numbersof excavationsfor sporocarpsper feeding site sumptionwas positively correlatedwith numbersof bi- (y'; wherey' = In [ y + 1]) tendedto decline with increas- son in the ecosystem (Table 3). Similarly, telemetry ing size of the currentwhitebark pine cone crop (cone; locations revealed a negative correlationbetween con- cones/tree): sumptionof sporocarpsand extent of whitebarkforests (HPIAL) in the surroundingarea, whereas feeding sites y' = 3.1 - 0.00042cone2 revealed a positive correlationwith the extent of LPICO and HVASChabitat types (Fig. 3b). with n = 48, r2= 0.11, F= 5.4; 1, 46 df, P = 0.024. Some of the discrepanciesbetween resultsusing telem- Althoughthis simple model minimizedAICC (at 14.9), etry locations and results using feeding sites were likely another model containing more parameters,but with a relatedto spatialcorrelations among habitat types and other higher AICcvalue (15.4), was biologically interpretable landscapefeatures, by bearmanagement unit. The extent and helpedexplain some otherwiseobscure broader scale of HPIAL was negatively correlatedwith the extent of relations. This more complex model suggestedthat num- 100 Ursus 13:2002

Table2. Parameters(f3) of logisticregression models describing the log odds thatmushrooms were consumed at a site versus anyother type of activityfor Yellowstone grizzly bears, 1977-96. Modelsare based on telemetrylocations only, including sites whereno feedingor beddingsigns werefound. A values arechange in AICCwith piece-wise deletion of individualvariables or groupsof variables.Statistics (G2 and P) are for goodness-of-fittests; smallerG2 values and largerP values indicatebetter fit.

Distal factors only Distal and proximalfactors Independentvariables P SE A SE A Intercept - 24.2 7.3 -29.1 8.2 Distalfactors 64.6 Numberof whitebarkpine cones/tree 0.00040a 0.00019 1.9 Extentof LPICO(% of BMU) 0.61b 0.18 9.3 Extentof HPIAL(% of BMU) -0.17 0.032 36.2 Juliandate 0.17 0.062 32.9 0.22 0.071 15.1 Juliandate squared - 0.00032a 0.00013 17.2 -0.00045a 0.00015 10.9 Sex andage classc 1.9 2.7 Adultfemale - 0.18 0.24 -0.37 0.26 Adultmale 0.03 0.28 -0.33 0.31 Subadultfemale 0.87 0.32 1.13 0.36 Subadultmale - 0.72d -0.43d Proximalfactors 38.3 Lodgepolepine basal area (m2/ha) 0.049 0.0097 23.6 Whitebarkpine basal area (m2/ha) 0.00070a 0.00026 3.2 Totalground cover (%) -0.64b 0.16 12.4 Slope(B) -0.59b 0.20 6.7 Statistics G2(df) 310 (2 x 103) 281 (2 x 103) P . 1.00 RL2 0.88 0.88 N 2088 1630 a Coefficient was estimatedfor squaredvalues of the independentvariable. b Coefficient was estimatedfor values of the independentvariable (x) transformedas ln(1 + x). c Class variable. d Coefficient was calculatedby subtraction.

Table 3. Parameters(O) of logistic regression models describing the log odds that mushrooms were consumed at a site versus any other type of feeding or bedding activity, for Yellowstone grizzly bears, 1977-96. Models are based only on sites where feeding signs were found, including sites not associated with telemetry locations. A values are change in AICcwith piece-wise deletion of individualvariables or groups of variables. Statistics (G2and P) are for goodness-of-fit tests; smaller G2values and larger P values indicate better fit.

Distal factors only Distal & proximalfactors A Independentvariables P SE A P SE Intercept -54.8 10.5 -32.1 9.9 Distalfactors 56.5 Numberof bisonin studyarea 2.1a 0.59 13.4 Extentof LPICO(% of BMU) 0.50a 0.17 5.8 0.78a 0.17 19.7 Extentof HVASC(% of BMU) 0.00069b 0.00017 15.6 Juliandate 0.26 0.078 18.0 0.24 0.084 13.8 9.5 Juliandate squared -0.00048b 0.00016 12.2 -0.00046b 0.00018 Proximalfactors 71.8 28.6 Lodgepolepine basal area (m2/ha) 0.074 0.015 7.4 Whitebarkpine basal area (m2/ha) 0.0013b 0.00036 Totalbasal area (m2ha) -0.0012b 0.00034 18.6 Totallive basalarea (m2/ha) 0.059 0.021 7.3 10.9 Totalground cover (%) -0.63 0.17 0.21 5.8 Slope(B) -0.59a Statistics G2(df) 394(2x 103) 313(2x 103) P 1.00 1.00 RL2 0.89 0.90 N 2911 2207 a Coefficient was estimatedfor values of the independentvariable (x) transformedas ln(1 + x). b Coefficient was estimatedfor squaredvalues of the independentvariable. GRIZZLYBEARS AND SPOROCARPS* Mattson et al. 101 bers of excavations (y') were related to a polynomial of (Mattson2000). Even so, it is unclearif or to what extent cone crop size andotherwise increased with ambienttem- consumptionof sporocarpsby subadultfemales was part perature(temp), Julian date (jd), andlodgepole pine basal of a risk-aversestrategy and a consequenceof theirbeing area (Ipp): able to affordconsumption of a lower qualityfood (comparedto ungulatetissue or whitebarkpine seeds) in a habi- y'= -7.8 + 0.14cone - 0.0031cone2+ 1.9[ln(temp+ 1)] tat less often used by dominantbears such as adultmales. + 0.02 ljd + 0.0004lpp2 Consumptionof fungalsporocarps by Yellowstonegriz- zly bearswas positively associatedwith edaphicallyharsh with n = 44, R2= 0.45, F =6.2, 5, 38 df, P = 0.000. sites dominatedby lodgepole pine. This was manifest, distally,by an increasedproclivity to consume sporocarps where the LPICO habitattype was most extensive and, DISCUSSION proximally,by a positive association with maturelodge- Consumptionof fungalsporocarps by Yellowstonegriz- pole pine and sparse vegetation ground cover. The im- zly bears peakedin Septemberand was rareprior to July. portanceof LPICO sites is emphasized not only by the Juliandate probablyfunctioned in this analysis as a sur- concentrationof sporocarpconsumption here, but also by rogatefor seasonal availabilityof preferredfungal sporo- the apparentself-reinforcing effects of this type at broad carps in preferredhabitats. By all indications, the epi- scales;i.e., bearswere morelikely to consumesporocarps, and hypogeous sporocarpsconsumed by grizzly bears all else equal, when the LPICOtype was more common duringthis study were most abundantduring the fall, al- in the surroundingBMU. By contrast,no consumption though there may have been a secondaryspring peak in of fungal sporocarpswas observed in non-foresthabitat abundancefor some species (Smith 1975, Thiers 1975, types or in forest types that had been recently harvested. Fogel and Trappe 1978, Miller 1980, North et al. 1997, These patternsare consistentwith what is known about States and Gaud 1997). If the seasonal chronology of productionof edible sporocarpsby mycorrhizalfungi in grizzly bearconsumption tracked abundance of preferred borealor high-elevationconifer forests. For example, of sporocarps,then we predictthat peak availability occurred, the edible boletes and truffles or false truffles, most are on average, about the thirdweek of September. strongly associated with pine-dominatedforests (Thiers Consumptionof sporocarpsmay have been indirectly, 1975; Molina and Trappe 1982, 1994; Danielson 1984; positively, affected by numberof bison in the ecosystem. States 1990; States and Gaud 1997). Productionand di- The habitatswhere most sporocarpconsumption occurred versity of edible sporocarpsof all types tends to peak in (the LPICOand HVASCtypes) were positively spatially drierconifer forests (Fogel and Trappe 1978, Danielson correlatedwith bison winterrange, which was positively 1984). Otherwise,sporocarps tend to be more abundant associated with the year-round distribution of bison in thinned stands, but much less abundantin recently (Meagher 1973). Moreover,consumption of bison dur- burnedor clearcutforests of lodgepole pine and the eco- ing peak sporocarpconsumption (August and September; logically similar jack pine (P banksiana; Visser 1995, Mattson 1997b) took place nearor in bison winterranges. Kroppand Albee 1996). Even so, it is not clear how this might have led to in- Consumptionof sporocarpsappeared to be complexly creased consumptionof sporocarpsand whethersuch an relatedto availabilityof whitebarkpine trees and cones. effect, if any, might have been due to effects of bison on Interpretingthese patternsdepends on distinguishingbe- sporocarpproduction or effects of bison consumptionon tween effects of cone crops and effects of the trees them- overall grizzly bear foraging strategies,including poten- selves. Some sporocarp consumption occurred in tial avoidanceby subordinatebears of dominantbears on whitebarkpine habitatsnear whitebarkpine trees. Thus, bison carcasses (Craigheadet al. 1995). to a limited extent, the odds thatgrizzly bearswould con- Subadultfemales apparentlywere more likely thanany sume mushrooms increased where whitebarkpine was other type of grizzly bear to consume fungal sporocarps, proximally more abundant,which occurredmore often even where and when immediatesite and vegetationcon- when cone crops were large because bears concentrated ditions were otherwiseconducive to consumptionof spo- in whitebarkpine standsto eat pine seeds (Mattsonet al. rocarps by other bears. However, there is no ready 2001). On the otherhand, BMUs withextensive whitebark for this explanation behavior. It is suggestive that sub- pine habitatscontained only limited habitats,especially adultfemales wereprobably energetically the least stressed of the LPICOtype, whereconsumption of sporocarpswas of all bears (Mattson 1990), that they consumed the least most concentrated. Thus, at broad scales, the extent of tissue from ungulates (Mattson 1997b), and that, of all whitebarkpine habitats was negatively associated with of theiroverall types females, patternsof habitatuse were likelihood of sporocarpconsumption. Finally, although the most distinct from patternsof habitat use by males grizzly bearsconsumed sporocarps where whitebarkpine 102 Ursus 13:2002 was abundant,their foraging was less intense compared andinference: a practicalinformation-theoretic approach. to when they were consumingsporocarps amongst abun- Springer,New York,New York,USA. A.B. Sciuridsin Pacific Northwest dantlodgepole pine. This tendencywas probablyaccen- CAREY, 1995. managedand old forests. 5:648-661. tuatedbecause distributionsof whitebarkand lodgepole growth Ecological Applications CHEUNG,P.C-K. 1997. Dietary fiber content and composition were negativelycorrelated, probably through direct pines of some edible determinedby two methodsof analysis. between the 2 andReinhart fungi competition species (Mattson Journalof ScienceFood and Agriculture 73:255-260. 1990). COLLINS,W.B., ANDP.J. URNESS. 1983. Feeding behavior and Fungal sporocarpswill likely become a more impor- habitat selection of mule deer and elk on northernUtah tant food of Yellowstone grizzly bears with warming of summer range. Journalof Wildlife Management47:646- the global climate, especially if warming leads to drier 663. conditions. Duringthis studyconsumption of sporocarps COUTURIER,M.A.J. 1954. L'oursbrun Ursus arctos L. Grenoble, France. peaked,on average,in Septemberat only 7% of all activ- (In French.) J.S. ANDJ.A. MITCHELL.1995. The Nonetheless, under combinationsof CRAIGHEAD,J.J., SUMNER, ity. optimal sparse bears of Yellowstone: their in the cover and abundant grizzly ecology ground lodgepole pine, probabilities Yellowstone ecosystem, 1959-1992. Island Press, of reachedabout 30-50% Such consumption (see Fig. 3c). Washington,D.C., USA. the of sites that conditions were most common on types DANIELSON,R.M. 1984. Ectomycorrhizalassociations in jack areprojected to become substantiallymore widespreadin pine stands in northeasternAlberta. CanadianJournal of the Yellowstone region with climate warming (Romme Botany 62:932-939. and Turner1990, Bartlein et al. 1997). Under this sce- DEMARIS,A. 1992. Logit modeling: practical applications. USA. nario, whitebarkpine is predictedto decline and perhaps SAGEPublications, Newbury Park, California, DESPAIN,D.G. 1990. Yellowstone the virtuallydisappear. Thus, these resultsoffer a preview of vegetation: consequences of and environment in a natural setting. Robert a food and a behavior that may become more history foraging Rinehart,New York,New York,USA. to the of Yellowstone's bear important management grizzly DIRKS, R.A., ANDB.E. MARTNER.1982. The climate of population. Yellowstoneand Grand Teton National Parks. U.S. National ParkService Occasional Paper No. 6. FOGEL,R., AND J.M. TRAPPE. 1978. Fungus consumption ACKNOWLEDGMENTS (mycophagy) by small animals. Northwest Science 52:1- Many people contributedto markingbears andcollect- 31. GROOTBRUINDERINK, G.W.T.A., E. HAZEBOEK,AND H. VAN DER ing site data used for this analysis. Of those, a few de- VOOT.1994. Diet and condition of wild boar, Sus scrofa serve special recognition: S. Alt, D. Burrup,L. Cayot, D. scrofa, without supplementaryfeeding. Journalof Zoology Delsordo, D. Dunbar,G. Green,J. Henry,B. Hoskins, C. (London) 233:631-648. J. K. Kendall, M. J. D. Hunt, Jonkel, Knapp, Orpett, HALL,J.G. 1981. A field study of the Kaibabsquirrel in Grand and R. D. Reinhart,B. Schleyer,E. Shannahan, Swalley. CanyonNational Park. Wildlife Monograph 75. Stradley contributedthe most of any to radiotracking JOHNSON,D.H. 1999. The insignificance of statistical markedbears. R. Knight supervised this study during significance testing. Journal of Wildlife Management 1977-96. Data collection was funded by the U.S. Na- 63:763-772. tional ParkService, Wyoming Game and Fish, U.S. Fish KNIGHT,R.R., ANDL.L. EBERHARDT.1985. Populationdynamics bears. 66:323-334. and Wildlife Service, U.S. Geological Survey (USGS), of Yellowstone grizzly Ecology ANDS. ALBEE.1996. The effects of silvicultural and U.S. ForestService. of this analysis was KROPP,B.R., Completion treatmentson occurrence of in a the USGS Forest and mycorrhizal sporocarps supportedby RangelandEcosys- Pinus contorta forest: a Biological We preliminary study. tem and NorthernRocky MountainScience Centers. Conservation78:313-318. M. appreciatethe support,monetary and otherwise, of MATTSON,D.J. 1990. Human impacts on bear habitat use. Collopy, C. van RiperIII, and C. Schwartzof the USGS. InternationalConference on BearResearch and Management 8:33-56. 1997a. Use of lodgepole pine cover types by LITERATURECITED Yellowstonegrizzly bears. Journalof Wildlife Management 61:480-496. BARTLEIN,P.J., C. WHITLOCK,AND S. L. SHAFER.1997. Future bears climatein the YellowstoneNational Park region and its 1997b. Use of ungulatesby Yellowstonegrizzly Conservation81:161-177. potentialimpact on vegetation. ConservationBiology Ursus arctos. Biological 11:782-792. .2000. Causes and consequencesof dietarydifferences Yellowstone bears arctos). Ph.D. BLANCHARD,B.M., ANDR.R. KNIGHT. 1991. Movements of among grizzly (Ursus of Moscow, Idaho, USA. Yellowstonegrizzly bears. Biological Conservation58:41- Dissertation,University Idaho, 67. ,K. BARBER,R. MAw,AND R. RENKIN.1999. Coefficients for Yellowstone's bearhabitat. USGS BURNHAM, K.P., ANDD.R. ANDERSON.1998. Model selection of productivity grizzly GRIZZLYBEARS AND SPOROCARPS* Mattson et al. 103

Forest and RangelandEcosystem Science Center,Colorado Hokkaido,Japan. International Conference on BearResearch PlateauField Station,Flagstaff, Arizona, USA. and Management7:215-220. , B.M. BLANCHARD,ANDR.R. KNIGHT.1991. Food habits OGNEV,S.I. 1931. Mammals of eastern Europe and northern of Yellowstonegrizzly bears, 1977-1987. CanadianJournal Asia. Volume II. Carnivora (Fissipedia). Glavnauka - of Zoology 69:1619-1629. GosudarstvennoeIzdatel'stvo, Moskva. (Translatedfrom , K.C. KENDALL,AND D.P. REINHART.2001. Whitebark Russian,Israel Program for ScientificTranslation, Jerusalem, pine, grizzly bears,and red squirrels.Pages 121-136 in D.F. Israel, 1962.) Tomback, S.F. Arno, and R.E. Keane, editors. Whitebark POELKER,R.J., ANDH.D. HARTWELL.1973. Black bear of pine communities: ecology and restoration. Island Press, Washington. Washington State Game Department, Washington,D.C., USA. Biological Bulletin 14. ,ANDD.P. REINHART. 1990. Whitebarkpine on the Mount PRITCHARD,G.T., AND C.T. ROBBINS. 1990. Digestive and Washburnmassif, Yellowstone National Park. Pages 106- metabolicefficiencies of grizzly and black bears. Canadian 116 in W.C. Schmidt and K.J. McDonald, compilers. Journalof Zoology 68:1645-1651. Proceedings-Symposium on whitebark pine ecosystems: ROBERTS,T.J. 1977. The mammalsof Pakistan. ErnestBenn, ecology and managementof a high-mountainresource. U.S. London, U.K. Forest Service GeneralTechnical Report INT-270. ROMME,W.H., ANDM.G. TURNER.1991. Implicationsof global MEAGHER,M.M. 1973. The bison of YellowstoneNational Park. climate change for biogeographic patterns in the greater U.S. NationalPark Service, ScientificMonograph Series No. Yellowstone ecosystem. ConservationBiology 5:373-386. 1. SMITH,A.H. 1975. A field guide to western mushrooms. MEALEY,S.P. 1975. The naturalfood habits of free ranging University of Michigan Press, Ann Arbor,Michigan, USA. grizzlybears in YellowstoneNational Park, 1973-1974. M.S. STATES,J.S. 1990. Mushroomsand truffles of the Southwest. Thesis, MontanaState University, Bozeman, Montana, USA. University of Arizona Press, Tucson, Arizona, USA. MILLER,O.K. 1980. Mushroomsof NorthAmerica. E.P.Dutton, , ANDW.S. GAUD. 1997. Ecology of hypogeous fungi New York,New York,USA. associated with ponderosapine. I. Patternsof distribution MOLINA,R., ANDJ.M. TRAPPE. 1982. Patternsof ectomycorrhizal and sporocarp production in some Arizona forests. host specificity and potential among Pacific Northwest Mycologia 89:712-721. conifers and fungi. Forest Science 28:423-456. STEELE,M.A. 1998. Tamiasciurushudsonicus. Mammalian , AND . 1994. Biology of the ectomycorrhizal Species 586:1-9. genus, Rhizopogon. I. Host associations, host-specificity THIERS,H.D. 1975. Californiamushrooms: a field guide to the and pure culturesynthesis. New Phytologist 126:653-675. boletes. HafnerPress, New York,New York,USA. NELSON,J.R., ANDT.A. LEEGE.1982. Nutritionalrequirements U.S. NATIONALOCEANIC AND ATMOSPHERIC ADMINISTRATION. and food habits. Pages 323-367 in J.W. Thomas and D.E. 1977-92. Climatologicaldata. Annualsummary. Wyoming. Toweill, editors. Elk of North America: ecology and U.S. Departmentof Commerce,Washington, D.C., USA. management. Stackpole Books, Harrisburg,Pennsylvania, U.S. NATIONALPARK SERVICE. 1997. Yellowstone's Northern USA. Range: complexity and change in a wildland ecosystem. NORTH,M., J. TRAPPE,AND J. FRANKLIN.1997. Standingcrop U.S. National Park Service, Yellowstone National Park, and animal consumption of fungal sporocarps in Pacific Wyoming,USA. Northwestforests. Ecology 78:1543-1554. VISSER,S. 1995. Ectomycorrhizalfungal succession in jack NovIKOv,G.A., A.E. AIRAPET'YANTS,Yu.B. PUKINSKY,E.K. pine stands following wildfire. New Phytologist 129:389- TIMOFEEVA,AND I.M. FOKIN.1969. Some peculiaritiesof the 401. brown bear population in the province of Leningrad. WEISBERG,S. 1985. Applied linearregression. Second edition. Zoological Journal68:885-900. (Translatedfrom Russian, John Wiley and Sons, New York,New York,USA. Canadian Department of Secretary of State, Translation Bureau, Foreign Languages Division, Ottawa, Ontario, Received: 9 May 2001. Canada, 1970). Accepted: 18 January 2002. ODACHI, S., ANDT. AOI. 1987. Food habits of brown bear in Associate Editor: Hilderbrand.