Effects of Management of Domestic Dogs and Recreation on Carnivores in Protected Areas in Northern California

Transcription

1 Contriuted Pper Effects of Mngement of Domestic Dogs nd Recretion on Crnivores in Protected Ares in Northern Cliforni SARAH E. REED AND ADINA M. MERENLENDER Deprtment of Environmentl Science, Policy & Mngement, University of Cliforni, Berkeley, CA , U.S.A. Astrct: In developed countries dogs ( Cnis lupus fmiliris) re permitted to ccompny humn visitors to mny protected res (e.g., >96% of protected lnds in Cliforni, U.S.A.), nd protected-re mngement often focuses on regulting dogs due to concerns out predtion, competition, or trnsmission of disese nd conflicts with humn visitors. In 24 nd 25, we investigted whether crnivore species richness nd undnce were ssocited with mngement of domestic dogs nd recretionl visittion in protected res in northern Cliforni. We surveyed for mmmlin crnivores nd humn visitors in 21 recretion res in which dogs were llowed offlesh or onlesh or were excluded, nd we compred our oservtions in the recretion res with oservtions in seven reference sites tht were not open to the pulic. Crnivore undnce nd species richness did not differ mong the three types of recretion res, ut ntive crnivore species richness ws 1.7 times greter (p <.1) nd the reltive undnces of ntive coyotes ( Cnis ltrns) nd octs ( Lynx rufus) were over four times greter (p <.1) in the reference sites. Aundnces of octs nd ll crnivores declined s the numer of visitors incresed. The policy on domestic dogs did not pper to ffect species richness nd undnce of mmmlin crnivores. But the numer of dogs we oserved ws strongly ssocited with humn visittion (R 2 =.54), so the key fctors ssocited with recretionl effects on crnivores pper to e the presence nd numer of humn visitors to protected res. Keywords: crnivore, domestic dog, mngement policy, noninvsive niml survey, protected re, recretion Efectos del Mnejo de Perros Domésticos y Recreción sore Crnívoros en Áres Protegids en el Norte de Cliforni Resumen: En píses desrrolldos, se permite que perros ( Cnis lupus fmiliris) compñen visitntes humnos en muchs áres protegids (e.g., > 96% de ls áres protegids en Cliforni, E.U.A.), yelmnejo de áres protegids menudo se enfoc en l regulción de perros deido preocupciones respecto l depredción, competenci o trnsmisión de enfermeddes y conflictos con visitntes humnos. En 24 y 25 investigmos sí l riquez y undnci de especies de crnívoros se socin con el mnejo de perros domésticos y l visit recretiv en áres protegids en el norte de Cliforni. Muestremos mmíferos crnívoros y visitntes humnos en 21 áres en ls que se permitín perros con o sin corre o que fuern excluidos, y comprmos nuestrs oservciones en ls áres recretivs con oservciones en 7 sitios de referenci que no estn iertos l púlico. L riquez y undnci de crnívoros no difirió en los 3 tipos de áres recretivs, pero l riquez de especies de crnívoros fue 1.7 veces myor (p <.1) y ls undncis reltivs de coyotes ntivos ( Cnis ltrns) y linces (Lynx rufus) fueronmás de 4 veces myores (p <.1) en los sitios de referenci. L undnci de linces y de todos los crnívoros declinó medid que el increment el número de visitntes. L polític sore perros domésticos preció nofectr l riquez y undnci de mmíferos crnívoros. Pero el número de perros que oservmos est Current ddress: Deprtment of Fish, Wildlife, nd, Colordo Stte University, Fort Collins, CO , emil srh.reed@colostte.edu Pper sumitted June 2, 21; revised mnuscript ccepted Septemer 27, 21. 1, Volume **, No. **, *** *** C 211 Society for DOI: /j x

2 2 Mnging Dogs in Protected Ares fuertemente socido con l visit de humnos (R 2 =.54), sí que los fctores clve socidos con los efectos de ctividdes recretivs sore crnívoros precen ser l presenci y número de visitntes humnos ls áres protegids. Plrs Clve: áre protegid, crnívoro, muestreo no invsivo de nimles, perro doméstico, recreción Introduction Pet nd ferl domestic dogs (Cnis lupus fmiliris) occur on every continent except Antrctic (Miklosi 27; Vnk & Gompper 29). There were 77.5 million pet dogs in the United Sttes in 29 (HSUS 29), nd dogs re permitted to ccompny humn visitors to the mjority of U.S. protected res. For exmple, mong protected lnds in Cliforni tht permit pulic ccess (GIN 29) including federl, stte, nd locl prks, forests, nd privte nture reserves 78.7% permit unrestricted ccess y domestic dogs, 18.2% permit dogs only in specific res, nd only.2% exclude domestic dogs entirely. A mjority of protected-re visitors recognize tht recretion my distur ntive niml popultions (Tylor & Knight 23), nd most visitors ttriute the strongest negtive effects to recretionl ctivities with domestic dogs (Sterl et l. 28). Dogs re potentil disese vectors, predtors, nd competitors of ntive fun (Butler et l. 24). Due to concerns out their effects on nturl resources nd conflicts mong recretionl user groups (Bekoff & Meney 1997), dog ccess is regulted or restricted in some protected res (Forrest & St. Clir 26). For exmple, most U.S. ntionl prks llow dogs only on leshes, ner residences nd visitor centers, nd in cmpgrounds. Empiricl investigtions of the effectiveness of different pproches to dog mngement for the protection of ntive species re uncommon. Although the voidnce of recretionl trils nd hevily visited res y ntive irds nd mmmls is well documented (Miller et l. 1998; Firnks & Tullous 22; Tylor & Knight 23), the results from the few studies tht hve investigted the impcts of dogs in recretion res hve een mixed. Mmml ctivity levels re lower ner trils on which dogs re llowed compred with trils on which they re not (Lenth et l. 28). Similrly, ird species richness nd undnce re lower when hiker is ccompnied y dog compred with hiker wlking lone (Bnks & Brynt 27). These ptterns of sptil displcement re consistent with the results of ehviorl studies tht show elevted stress levels (McArthur et l. 1982), incresed flight distnces (Miller et l. 21), nd impired reproduction (Ylden & Ylden 199) in irds nd mmmls when dogs re present. Nevertheless, the results of other studies show few effects ttriutle to the presence or regultion of dogs. The presence of dog increses hiker s re of influence reltive to mule deer (Odocoileus hemionus), ut not reltive to two grsslnd nd one forest ird species (Miller et l. 21). In study of 22 urn prks, dog lesh lws re not ssocited with species richness or undnce of irds nd smll mmmls (Forrest & St. Clir 26). Evluting the effectiveness of mngement pproches could e confounded y the intensity of recretionl visittion to protected re ecuse visittion levels influence the mgnitude of the effects of recretion. For exmple, the undnces nd ctivity levels of mphiins (Rodriguez-Prieto & Fernndez-Juricic 25), reptiles (Grer & Burger 1995), nd irds (Vn der Znde et l. 1984) decrese s the numer of visitors increses. Among mmmlin crnivores, wolf (Cnis lupus) pcks trvel long low-use trils nd rods rther thn trils tht receive dily foot trffic or rods tht receive more thn 1, vehicles/month (Whittington et l. 25), nd in southern Cliforni, octs re detected less frequently long recretionl trils with high levels of humn ctivity (George & Crooks 26). We investigted the effects of humn visitors nd domestic dogs on the species richness nd undnce of ntive mmmlin crnivores in 28 protected res in northern Cliforni. To differentite etween the effects of dogs nd those of humns, we surveyed protected res tht represented the full rnge of dog policies (dogs offlesh, onlesh, or excluded), nd we compred the possile influences of dog mngement on ntive crnivores etween recretion res nd reference sites tht did not llow pulic ccess for recretion. We lso exmined whether humn nd dog visittion levels explin the reltive undnces of ntive crnivores. Methods Study Are Our study re ws 264-km 2 re in Cliforni s Mrin, Sonom, nd Np counties (38 18 N, W) north of Sn Frncisco By. The region hs Mediterrnen climte nd hs high concentrtions of species richness nd endemism (Myers et l. 2). The three counties support humn popultion pproching 1 million (U.S. Census Bureu 26), nd protected res in Mrin, Sonom, nd Np re hevily visited recretion destintions for locl people (Reed & Seymour 28) nd the more thn 7 million residents of the greter Sn Frncisco By Are (BAOSC 24).

3 Reed & Merenlender 3 We surveyed 21 recretion res nd seven reference sites. We defined recretion re s trct of lnd (>4 h) with nturl vegettion cover tht is mintined for the enjoyment of the pulic. The recretion res we surveyed hd three different policies on dogs: dogs permitted offlesh (n = 9); dogs permitted only onlesh (n = 7); nd dogs excluded (n = 5). The reference sites (n = 7) were protected res with no pulic ccess. To minimize vrition in vegettion chrcteristics mong sites, we surveyed only mixed-ok woodlnds etween 5 nd 5 m in elevtion long the foothills of the Costl, Mycms, nd Vc mountin rnges. The study sites hd men re of h (SD 368.3) (Tle 1). Field Surveys We visited 15 sites one to two times ech etween My nd Octoer 24 nd ll 28 sites once etween June nd Septemer 25. Effects of recretion on ntive nimls re more strongly ssocited with weekdy rther thn weekend or holidy visittion ptterns (Vn der Znde et l. 1984). Accordingly, we visited sites on weekdys for severl hours during the morning or fternoon. We conducted trnsect serches to detect scts of ntive crnivores nd domestic dogs s n index of species undnces or ctivity levels. Sct surveys re n efficient method with which to detect multiple species nd re used frequently to gther informtion on the composition nd species richness of crnivores (Long et l. 28), nd sct undnce is closely correlted with the species undnce (Wilson & Delhy 21; Hrrison et l. 24) nd popultion densities (Stnder 1998). The trget species of our surveys were ntive crnivores tht re reltively common in the study re mountin lion (Pum concolor), coyote (Cnis ltrns), oct (Lynx rufus), nd gry fox (Urocyon cinereorgenteus) nd domestic dogs. In ech site we estlished four, 5-m trnsects. Trnsects were locted long mpped recretionl trils in the recretion sites nd long closed rods or trils in the reference sites. The trils nd rods we surveyed hd grvel or nturl surfce nd were 1 5 m wide. To minimize possile effects of djcent lnd use on our inferences out crnivore distriutions, we strtified the loctions of trnsects etween the edge (<5 m from the perimeter) nd interior (>5 m from the perimeter) of ech site. We collected nd recorded the loction of ech prole mmmlin crnivore sct nd stored the sct in pper g with cly desiccnt pck (Desi-Pk, Texs Technologies, Cedr Prk, Texs). We lso recorded the coordintes of ll domestic dog scts detected with geogrphic positioning system. Becuse there were lrge numer of dog scts present, we did not collect them when we could identify them visully s dog sct. We Tle 1. Site chrcteristics nd recretionl visittion s function of domestic-dog mngement policy in 28 protected res in northern Cliforni. Vrile Rnge Offlesh Onlesh Excluded Reference F p Mens comprison Site chrcteristic re (h) (165.8) 15.6 (191.2) (76.8) 147. (96.9) men elevtion (m) (77.5) (167.9) (118.4) 22.2 (95.9) men slope ( ) (4.2) 8.5 (6.6) 14.8 (2.1) 14.9 (3.6) offlesh>onlesh herceous cover (%) (26.5) 36.2 (39.2) 14.2 (7.9) 41.7 (24.4) hrdwood cover (%) (26.7) 31.3 (32.9) 65.2 (11.2) 34.9 (18.6) Recretionl visittion tril length (km) (7.6) 7.8 (4.1) 18.3 (11.5) tril density (km/h) (1.4) 8.1 (4.5) 5.4 (3.9) onlesh>offlesh djcent properties (185.4) (41.9) (885.3) humn visittion (per hour) (4.2) 9.6 (1.7) 4. (4.9) dog visittion (per hour) (2.5) 3.1 (3.6) Men vlues (SD) for ll surveys conducted in 24 nd 25.

4 4 Mnging Dogs in Protected Ares lso recorded the coordintes nd numer of people nd dogs in ech group of recretionl visitors encountered during the trnsect surveys. Sct Identifiction We nlyzed ll scts collected in 25, ut only scts collected from suset of sites in 24 due to udget constrints. We took two susmples (pproximtely 5 mg) of ech sct within 3 d of collection nd stored them t 8 C. Between Octoer 25 nd Septemer 26, we used QigenQIAmp DNA stool extrction kits (Qigen, Vlenci, Cliforni) to extrct DNA from the scts. We performed polymerse chin rection (PCR) with the HCrn2 (Bidlck et l. 27) nd CnidL1 (Pxinos et l. 1997) primers to mplify the first 196 p of the mitochondril cytochrome gene. We then used restriction frgment length polymorphisms (RFLP) to identify the mplified DNA frgments to species. Lortory methods for DNA extrction, mplifiction, nd identifiction re descried in detil in Bidlck et l. (27) nd Reed & Merenlender (28). We repeted PCR-RFLP nlyses for scts collected in ech study site until we hd identified minimum of 75% of the smples to species. Sptil Anlyses We used sptil dtse of protected lnds in the Sn Frncisco By Are (BAOSC 29) to identify the loctions of ll recretion nd reference sites in ArcGIS 9.1 (ESRI, Redlnds, Cliforni). When two recretion res were contiguous nd mnged y the sme gency, we considered them single study site. We clculted the totl re of ech site, men elevtion, nd slope from 3-m digitl elevtion model (DEM) nd the proportion of lnd cover in dominnt vegettion types (hrdwood nd herceous) (USFS 25). We lso clculted severl vriles tht we hypothesized were proxies of intensity of visittion to the recretion res: tril length, tril density, nd djcent humn popultion density. We used tril mps provided y the mnging gencies to determine the totl length of the trils, nd we divided the tril length y site re to clculte tril density. We clculted the numer of residentil properties within 5 m of the site oundries s n index of djcent humn popultion density. Sttisticl Anlyses We performed ll sttisticl nlyses in JMP 6. (SAS Institute, Cry, North Crolin). We used one-wy nlyses of vrince (ANOVA) to compre re, elevtion, slope, nd lnd cover of study sites with different dog policies nd to compre tril length, tril density, numer of djcent residentil properties, nd oserved numers of humn nd dog visitors mong sites. We exmined vrition in the numer nd loction of visitors nd the reltive undnce of scts over time to determine whether dt from 24 nd 25 could e pooled. We used pired-smple t tests (Zr 1999) to compre the rtes of visittion (numer of people nd dogs detected per hour) in six sites etween 24 nd 25 nd to compre the totl undnce of crnivore scts collected in 15 sites etween 24 nd 25. We hypothesized tht scts could ccumulte fter the lst sustntil rin event of the yer, which typiclly occurs in April. Alterntively, scts could e trmpled on or removed from the survey trnsects. Therefore, we tested for trends in sct ccumultion with liner regression of totl sct undnce versus the numer of dys since the first trnsect survey. To test whether the reltive undnce of ll crnivore scts could serve s resonle proxy for reltive undnce of ntive crnivore scts, we exmined the correltion etween the totl numer of scts collected in ech site nd the numer of scts ttriuted to ntive crnivores in PCR-RFLP nlyses. To investigte whether dog policies influenced crnivores in protected res, we used scts to compre ntive crnivore species richness, reltive undnce of ech species, nd reltive undnce of ll ntive crnivores mong the three mngement types nd reference sites. We clculted sct undnce s the numer of crnivore scts detected divided y the length of the trnsect (Hrrison et l. 24), nd we pproximted totl crnivore undnces for ech trnsect y extrpolting the overll proportions of scts from ech species detected in the site to smples we were unle to identify in the lortory. Becuse we expected the distriutions of species scts mong sites to violte the ssumption of normlity for prmetric tests (Potvin & Roff 1993), we rnk-trnsformed the species richness nd undnce dt nd compred the rnked vlues mong the four site types. When we found differences (p <.5) mong the site types in n ANOVA, we used Tukey s honestly significnt difference (HSD) test to identify differences etween pirs of mens. We lso exmined the reltions etween the numers of humn nd dog visitors nd the reltive undnce of crnivore scts in the study sites. We clculted humn nd dog visittion rtes s the numer of people nd dogs oserved divided y the time spent surveying ech trnsect. We used model-selection pproch (Burnhm & Anderson 22) to identify the fctors tht est explined vrition in totl sct undnce nd the undnces of scts of the most common ntive crnivore species. In ddition to the humn nd dog visittion rtes, we included s covrites interctions etween the two visittion rtes; site re, elevtion, slope, nd lnd cover; nd tril distnce, tril density, nd numer of djcent residentil properties. We trnsformed the vriles to meet the ssumption of normlity nd compred univrite models for ech of the 14 covrites with Akike informtion criterion with smll smple size djustment (AIC c ). When model including humn or dog visittion

5 Reed & Merenlender 5 rte hd strong support (w i >.5), we used quntile regression (Cde & Noon 23) to further explore the reltion etween visittion rte nd reltive undnces of scts. Results Differences mong Sites nd over Time There were no significnt differences in re or elevtion mong the recretion res nd reference sites, ut sites where dogs were required to e onlesh hd shllower slopes thn where dogs could e offlesh (F = 6.66, p <.1; Tle 1). The study sites hd men of 47% hrdwood cover nd 33% herceous cover, nd sites tht excluded dogs hd the highest percentge of hrdwood cover (65%; F = 2.81, p =.6). Sites where dogs were required to e onlesh hd the highest density of recretionl trils (F = 3.57, p =.5). Humn nd dog visittion rtes were highly vrile mong study sites. We found no significnt differences in pired comprisons of the numer of humn visitors (t =.52, p =.63) or domestic dogs (t = 1.58, p =.34) etween yers. Overll undnce of scts ws greter in 25 thn 24 (t = 2.141, p =.5). Some of this difference my hve een ttriutle to the fct tht we verged sct undnce vlues for sites visited twice in 24. We did not find evidence of significnt trend in sct ccumultion or removl over time. The numer of sct detections incresed slightly over the course of the seson in 25 (.55 scts km 1 dy 1 ), ut the correltion ws very wek (R 2 =.4, p =.31). We identified n verge of 86.6% of the scts collected in ech site in PCR-RFLP nlyses. There ws strong, positive correltion (R 2 =.98, p <.1; Fig. 1) etween the totl numer of scts detected in ech site nd the numer of scts ttriuted to ntive crnivores (mountin lions, coyotes, octs, nd gry foxes). Influence of Dog Policy We detected scts from fewer ntive species in the recretion res thn in the reference sites (F = 4.8, p <.1), ut there were no significnt differences mong recretion res with different dog policies (Fig. 2). Men sct undnces for oth coyotes (F = 1.69, p <.1) nd octs (F = 5.32, p <.1) were much greter in the reference sites thn in the three types of recretion res (Figs. 2 & 2c). Smple sizes for mountin lion nd gry fox sct were too smll to detect differences in the ANOVA. Mountin lion scts, however, were detected only in the reference sites, nd undnce of gry fox sct ws 3.4 times greter in the reference sites thn in the recretion res. Pooling results for 24 nd 25, totl undnce of crnivore sct ws lso greter in the reference sites thn in the three types of recretion res Ntive crnivore sct undnce (scts/km) Totl crnivore sct undnce (scts/km) Figure 1. Totl undnce of mmmlin crnivore scts oserved (excluding domestic dog) versus the undnce of ntive crnivore (coyote, oct, gry fox, mountin lion) scts identified y PCR-RFLP nlysis in 28 protected res in northern Cliforni. (F = 1.62, p <.1; Fig. 2d). The reltive undnce of domestic dog scts vried y dog policy nd ws much greter in sites where dogs were permitted (offlesh nd onlesh) thn in sites where dogs were not permitted (excluded nd reference) (F = 25.29, p <.1; Fig. 2e). Scts from domestic cts nd red foxes were detected infrequently, nd we did not oserve ny vrition in their undnces y dog policy. Influence of Recretionl Visittion None of the models of site chrcteristics or visittion rtes explined vrition in coyote sct undnce cross the study re (Tle 2). In contrst, the totl weight of visittion models for oth oct ( w i =.994) nd totl crnivore ( w i =.983) sct undnces fr exceeded the model weights for topogrphy nd lnd cover or proxy mesures of recretionl ctivity. The dditive model of humn nd dog visittion explined the most vrition in oth oct nd totl crnivore sct undnces. None of the mesures of topogrphy or vegettion tht vried mong sites with different dog policies (Tle 1) were strongly relted to crnivore sct undnce in regression models (Tle 2). In quntile regression nlyses, the slopes of the reltions etween oct nd totl crnivore sct undnces nd dditive visittion (humns + dogs) decresed s sct undnces incresed (Fig. 3). The slope decresed from to.9 in the interqurtile rnge of oct sct undnces nd decresed from.89 to.195 in the interqurtile rnge of totl crnivore sct undnces.

6 6 Mnging Dogs in Protected Ares () Men species richness () Men sct undnce (scts/km) Men sct undnce (scts/km) (c) (d) Men sct undnce (scts/km) Men sct undnce (scts/km) (e) Offlesh Onlesh Excluded Reference Site type Offlesh Onlesh Excluded Reference Site type Offlesh Onlesh Excluded Reference Site type Offlesh Onlesh Excluded Reference Site type Offlesh Onlesh Excluded Reference Site type Figure 2. Assocition with domestic dog policy in protected res nd () species richness of ntive crnivores detected in sct surveys (25), () coyote Discussion We detected significntly more scts ttriuted to ntive crnivore species in the reference sites thn in the recretion sites (Fig. 2). Coyote nd totl crnivore sct undnces were over four times greter nd oct sct undnces were over five times greter in reference sites (Figs. 2 d). Among the three types of recretion res, ntive crnivore undnce nd species richness did not differ significntly. These results suggest the dog policies t the sites we studied do not mitigte the effects of recretion on ntive crnivores in protected res. The key fctor ssocited with effects of visitors on crnivores ws whether reserve ws open to the pulic result tht is consistent with our findings in different suite of reserch sites (Reed & Merenlender 28). We did not oserve ny dogs during our surveys in humn-only recretion res (Tle 1), ut we found domestic dog scts in ll the recretion sites we surveyed, including smll numer in sites where dogs were not llowed (Fig. 2e). Stry nd ferl dogs re not common in our study re; 9 95% of dogs cptured y county niml control progrms hd owners who were contcted successfully (K. Fennelnd nd R. Grci, personl communiction). Thus, given the strong, positive correltion etween humn nd dog visittion rtes we oserved (R 2 =.54, p <.1), we speculte tht the presence of domestic dog scts in humn-only recretion res ws primrily ttriutle to illegl dog wlking. We lso oserved dogs offlesh in sites where they were required to e onlesh. Limited complince with dog regultions my oscure the effects of dog policies on ntive niml popultions in protected res (Forrest & St. Clir 26). Despite the infrctions we inferred, the reltive undnce of domestic dog scts ws 158 times greter in sites where they were permitted thn in sites where they were excluded (Fig. 2e). In contrst to other studies (e.g., Lenth et l. 28), we did not find strong, negtive correltion etween the reltive undnces of domestic dog scts nd ntive crnivore scts (Tle 2). This difference my hve occurred ecuse we investigted the reltions etween dog nd ntive crnivore sct undnces t the site-level rther thn long individul trnsects. It could lso e ttriutle to vrile enforcement of regultions sct undnce (25), (c) oct sct undnce (25), (d) totl crnivore sct undnce (24 25), nd (e) domestic dog sct undnce (25) (offlesh, dogs llowed off lesh, n= 9sites; onlesh, dogs llowed only onlesh, n= 7; excluded, humns llowed dogs excluded, n= 5; reference, no pulic ccess, n= 7); letters ove rs, indicte mens tht re different [p <.5] ccording to Tukey s honestly significnt difference test).

7 Reed & Merenlender 7 Tle 2. Results of liner regression models of the response of coyote nd oct sct undnce nd totl crnivore sct undnce to site chrcteristics (re, elevtion, slope, nd lnd cover), proxy mesures of visittion (tril length, tril density, nd numer of djcent residentil properties), nd oserved rtes of visittion y humns nd dogs in 21 recretion res in northern Cliforni. Coyote sct undnce Boct sct undnce Totl sct undnce (scts/km; 25) (scts/km; 25) (scts/km; 24 25) Model est log(l) K AICc di wi est log( L) K AICc di wi est log( L) K AICc di wi Null Are,c Elev c,d Slope e P_he c,f P_hwd g Tril h,i Tril_dens j Adj_pr i,k Dog_sct,l Humns c,m Dogs,n Humns c,o dogs Humns dogs c,p 1 Vrile trnsformed to meet the ssumption of normlity for liner regression: x 2. Site re (h). c Vrile trnsformed to meet the ssumption of normlity for liner regression: x 1/2. d Men elevtion (m). e Men slope ( ). f Herceous lnd cover (%). g Hrdwood lnd cover (%). h Totl length of recretionl trils (km). i Vrile trnsformed to meet the ssumption of normlity for liner regression: log(x). j Tril density (km/h). k Numer of djcent residentil properties. l Aundnce of domestic dog scts (sct/km). m Humn visittion rte (humns/hour). n Dog visittion rte (dogs/hour). o Additive interction etween humn nd dog visittion rtes (humns + dogs/hour). p Multiplictive interction etween humn nd dog visittion rtes (humns dogs/hour).

8 8 Mnging Dogs in Protected Ares requiring dog owners to remove their pets scts or to n unmesured effect of the scent-mrking ehvior of ntive crnivores or domestic dogs (Lenth et l. 28). In generl, however, crnivore sct undnce ws not relted to proxy mesures of humn visittion to the recretion res (tril length, tril density, nd numer of djcent residentil properties). Insted, estimtes of humn nd dog ctivity levels otined from direct oservtion were more closely ssocited with undnce of crnivore scts (Tle 2). Mny reserchers note there re few dt on recretionl visittion to protected res (e.g., Hill & Courtney 26; Lenth et l. 28; Blmford et l. 29), nd this dt gp my limit the mesurement nd prediction of impcts of recretion on ntive species. The numer of humn visitors nd dogs ws ssocited with mny fewer oservtions of scts of octs thn coyotes (Tle 2). This finding for octs is consistent with prior oservtions in protected res in southern Cliforni nd Colordo (George & Crooks 26; Lenth et l. 28). Totl crnivore sct undnce lso decresed s visittion rtes incresed. Although coyote scts were widespred nd undnt throughout the study re, totl crnivore sct undnce more closely resemled tht of octs, presumly ecuse scts of the less common mountin lions nd gry foxes lso declined in undnce s the visittion rtes of humns nd dogs incresed. Our lortory nlyses suggested tht when it is possile to visully distinguish domestic dog scts, totl mmmlin crnivore sct undnce is relile indictor of the reltive undnce of sct of ntive species (Fig. 1). Due to the reltion etween humn nd dog visittion rtes (Tle 1), we could not seprte the effects of humn from the effects of dogs. In fct, the dditive interction etween the visittion rtes of humns nd dogs est explined vrition in oth oct nd totl crnivore sct undnces (Tle 2), which suggests crnivores my e responding to the overll level of recretion in site. Moreover, in quntile regressions, the slope of the reltion with dditive visittion decresed nd therey the mgnitude of species response to recretionl ctivity incresed for incresing quntiles of oth oct nd totl crnivore sct undnces (Fig. 3). This mens tht in plces where crnivore undnces re high, recretion could hve greter effect on crnivores. Given the uncertinty out the effectiveness of domestic dog policies nd costs ssocited with enforcing mngement regultions (Dixon & Shermn 1991), we elieve tht enforcing lesh lws my not e the est use of limited mngement resources. Prohiiting dogs in protected res, however, my ffect humn visittion rtes. Although mny fctors ffect visittion to protected res, including distnce nd ccessiility to humn popultion centers, topogrphy, lnd cover, nd site menities (Hill & Courtney 26; Reed & Seymour 28), in our study, recretion res tht llowed dogs hd 6% Boct sct undnce (scts/km) Totl sct undnce (scts/km) ; Totl visits/hour : y= Totl visits/hour Figure 3. Boct sct undnce (25 only) nd totl crnivore sct undnce (24 25) versus humn nd dog visittion rtes in 21 recretion res in northern Cliforni. Liner-regression model estimtes re shown for the.9,.75,.5,.25, nd.1 quntiles of the distriutions of sct undnces. more humn visitors thn those tht did not (Tle 1). This suggests tht people my e more ttrcted to sites where they re permitted to ring dogs. Visittion, in turn, ws ssocited with the distriution nd reltive undnces of ntive crnivore scts in our study sites (Tle 2). Becuse controlling visittion is likely to e even more difficult nd expensive thn enforcing domestic dog policies, we suggest tht designting some sites s recretion res open to the pulic nd others s nture reserves closed to the pulic my e the most efficient strtegy for mnging the effects of recretion on crnivores.

9 Reed & Merenlender 9 Acknowledgments We re grteful to the gencies, orgniztions, nd privte lndowners who grnted us permission to survey their properties. The Hoplnd Reserch nd Extension Center provided logisticl support for field surveys, nd the lortory of P.J. Plsøll provided equipment nd guidnce for genetic nlyses. S.E.R. ws supported y Budweiser Conservtion Scholrship, Ntionl Science Foundtion Grdute Reserch Fellowship, Phi Bet Kpp Doctorl Fellowship, Sigm Xi Grnt-In-Aid-of-Reserch, Switzer Environmentl Fellowship, nd the Deprtment of Environmentl Science, Policy nd Mngement. Literture Cited Blmford, A., J. Beresford, J. Green, R. Nidoo, M. Wlpole, nd A. Mnic. 29. A glol perspective on trends in nturesed tourism. Pulic Lirry of Science Biology 7:DOI:1.1371/ journl.pio Bnks, P. B., nd J. V. Brynt. 27. Four-legged friend or foe? Dog wlking displces ntive irds from nturl res. Biology Letters 3: BAOSC (By Are Open Spce Council). 24. Prks, people nd chnge: ethnic diversity nd its significnce for prks, recretion nd open spce conservtion in the Sn Frncisco By Are. BAOSC, Berkeley, Cliforni. BAOSC (By Are Open Spce Council). 29. By re protected res dtse. BAOSC, Berkeley, Cliforni. Aville from =6 (ccessed Octoer 29). Bekoff, M., nd C. A. Meney Interctions mong dogs, people, nd the environment in Boulder, Colordo: cse study. Anthrozoos 1: Bidlck, A. L., S. E. Reed, W. M. Getz, nd P. J. Plsoll. 27. Chrcteriztion of western North Americn crnivore community using PCR-RFLP of cytochrome otined from fecl smples. Conservtion Genetics 8: Burnhm, K.P., nd D.R. Anderson. 22. Model selection nd multimodel inference: prcticl informtion-theoretic pproch. Springer, New York. Butler, J. R. A., J. T. du Toit, nd J. Binghm. 24. Free-rnging domestic dogs (Cnis fmiliris) s predtors nd prey in rurl Zimwe: threts of competition nd disese to lrge wild crnivores. Biologicl Conservtion 115: Cde, B. S., nd B. R. Noon. 23. A gentle introduction to quntile regression for ecologists. Frontiers in Ecology nd the Environment 1: Dixon, J. A., nd P. B. Shermn Economics of protected res. Amio 2: Firnks, W. S., nd R. Tullous. 22. Distriution of pronghorn (Antilocrp Americn Ord) on Antelope Islnd Stte Prk, Uth, USA, efore nd fter estlishment of recretionl trils. Nturl Ares Journl 22: Forrest, A., nd C. C. St. Clir. 26. Effects of dog lesh lws nd hitt type on vin nd smll mmml communities in urn prks. Urn Ecosystems 9: Grer, S. D., nd J. Burger A 2-yr study documenting the reltionship etween turtle decline nd humn recretion. Ecologicl Applictions 5: George, S. L., nd K. R. Crooks. 26. Recretion nd lrge mmml ctivity in n urn nture reserve. Biologicl Conservtion 133: GIN (GreenInfo Network). 29. Cliforni protected res dtse. GIN, Sn Frncisco. Aville from (ccessed Octoer 29). Hrrison, R. L., P. S. Clrke, nd C. M. Clrke. 24. Indexing swift fox popultions in New Mexico using scts. Americn Midlnd Nturlist 151: Hill, G. W., nd P. R. Courtney. 26. Demnd nlysis projections for recretionl visits to countryside woodlnds in Gret Britin. Forestry 79: HSUS (Humne Society of the United Sttes). 29. U.S. pet ownership sttistics. HSUS, Wshington, D.C. Aville from /pet_ownership_sttistics.html (ccessed Octoer 29). Lenth, B. E., R. L. Knight, nd M. E. Brennn. 28. The effects of dogs on wildlife communities. Nturl Ares Journl 28: Long, R. A., P. Mcky, W. J. Zielinski, nd J. C. Ry. 28. Noninvsive survey methods for crnivores. Islnd Press, Wshington, D.C. McArthur, R. A., V. Geist, nd R. H. Johnston Crdic nd ehviorl responses of mountin sheep to humn disturnce. Journl of Wildlife Mngement 46: Miklosi, A. 27. Dog ehviour, evolution, nd cognition. Oxford University Press, Oxford, United Kingdom. Miller, S. G., R. L. Knight, nd C. K. Miller Influence of recretionl trils on reeding ird communities. Ecologicl Applictions 8: Miller, S. G., R. L. Knight, nd C. K. Miller. 21. Wildlife responses to pedestrins nd dogs. Wildlife Society Bulletin 29: Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. d Fonsec, nd J. Kent. 2. Biodiversity hotspots for conservtion priorities. Nture 43: Pxinos, E., C. McIntosh, K. Rlls, nd R. Fleischer A noninvsive method for distinguishing mong cnid species: mplifiction nd enzyme restriction of DNA from dung. Moleculr Ecology 6: Potvin, C., nd D. A. Roff Distriution-free nd roust sttisticl methods: vile lterntives to prmetric sttistics? Ecology 74: Reed, S. E., nd A. M. Merenlender. 28. Quiet, nonconsumptive recretion reduces protected re effectiveness. Conservtion Letters 1: Reed, S. E., nd K. A. Seymour. 28. Predicting the intensity of recretionl use of ok woodlnd preserves. Generl technicl report PSW-GTR-217. U.S. Deprtment of Agriculture Forest Service, Pcific Southwest Reserch Sttion, Alny, Cliforni. Rodriguez-Prieto, I., nd E. Fernndez-Juricic. 25. Effects of direct humn disturnce on the endemic Ierin frog Rnieric t individul nd popultion levels. Biologicl Conservtion 12:1 9. Stnder, P. E Spoor counts s indices of lrge crnivore popultions: the reltionship etween spoor frequency, smpling effort nd true density. Journl of Applied Ecology 35: Sterl, P., C. Brndenurg, nd A. Arnerger. 28. Visitors wreness nd ssessment of recretionl disturnce of wildlife in the Donu-Auen Ntionl Prk. Journl for Nture Conservtion 16: Tylor, A. R., nd R. L. Knight. 23. Wildlife responses to recretion nd ssocited visitor perceptions. Ecologicl Applictions 13: U.S. Census Bureu. 26. County popultion estimtes. U.S. Census Bureu, Popultion Division, Wshington, D.C. Aville from (ccessed Septemer 21). USFS (U.S. Deprtment of Agriculture Forest Service). 25. CALVEG. USFS Remote Sensing L, McClelln, Cliforni. Aville from (ccessed Ferury 21). Vn Der Znde, A. N., J. C. Berkhuizen, H. C. vn Ltesteijn, W. J. terkeurs, nd A. J. Poppelrs Impct of outdoor recretion

10 1 Mnging Dogs in Protected Ares on the density of numer of reeding ird species in woods djcent to urn residentil res. Biologicl Conservtion 3:1 39. Vnk, A. T., nd M. E. Gompper. 29. Dogs (Cnis fmiliris)scrnivores: their role nd function in intrguild competition. Mmml Review 39: Whittington, J., C. C. St. Clir, nd G. Mercer. 25. Sptil responses of wolves to rods nd trils in mountin vlleys. Ecologicl Applictions 15: Wilson, G. J. nd R. J. Delhy. 21. A review of methods to estimte the undnce of terrestril crnivores using field signs nd oservtion. Wildlife Reserch 28: Ylden, P. E., nd D. W. Ylden Recretionl disturnce of reeding golden plovers. Biologicl Conservtion 51: Zr, J. H Biosttisticl nlysis. 4th edition. Prentice Hll, Upper Sddle River, New Jersey.