R

Ds

MD

h

•••••

a

ARR1A

KABLLLS

1

beGCabe

0h

Landscape and Urban Planning 125 (2014) 17–27

Contents lists available at ScienceDirect

Landscape and Urban Planning

j our na l ho me pa g e: www.elsev ier .com/ locate / landurbplan

esearch Paper

o golf courses reduce the ecological value of headwater streams foralamanders in the southern Appalachian Mountains?

ark J. Mackey, Grant M. Connette, William E. Peterman, Raymond D. Semlitsch ∗

ivision of Biological Sciences, University of Missouri, Columbia, MO 65211, USA

i g h l i g h t s

We sampled stream salamanders on 10 golf courses in western North Carolina, USA.Stream transects running through fairways contained lower abundance and diversity of salamanders.Salamander abundance was related to leaf litter depth, coarse woody debris, and buffer width.Chemicals down-stream of fairways were rarely detected.The ecological value of golf course streams may be enhanced by protection or restoration of riparian habitat.

r t i c l e i n f o

rticle history:eceived 26 February 2013eceived in revised form1 November 2013ccepted 21 January 2014

eywords:bundanceufferand-usearvaeeaf litter

a b s t r a c t

Recent studies indicate golf courses may have a potential role in biodiversity conservation and man-agement in human dominated landscapes. To serve this ecological role, effects of current golf coursemanagement practices must first be better understood. We sampled larval, juvenile, and adult streamsalamanders in transects located upstream, through, and downstream of managed fairways of 10 golfcourses in western North Carolina, USA. We measured in-stream and riparian habitat characteristics andtested for nitrate and pesticide chemicals to explain trends in salamander abundances and diversity.Stream transects located directly on fairways contained lower abundance of larval, metamorph, juvenile,and adult salamanders than either upstream or downstream transects. The species diversity of aquaticlarval and metamorph salamanders on fairways was also reduced but only compared to the upstreamtransects, and terrestrial juvenile and adult diversity did not differ among the three transect locations.Our analysis found that leaf litter depth, CWD, soil moisture, and buffer width parameters found within

tream management several models were positive predictors of salamander abundance and diversity. Nitrate was not detectedat any of the stream reaches and two of the 16 pesticide chemicals screened were only detected in neg-ligible proportions. Our findings suggest golf courses in western North Carolina can currently provideviable habitat for stream salamanders in reaches upstream and downstream of managed areas of coursesand streams running through fairways may be enhanced through simple management practices such asretaining woody debris, leaf litter, and restoring a riparian buffer.

. Introduction

The ecological value of streams and rivers globally is influencedy increasing human land use (Allan, 2004). Currently, there arestimated to be more than 31,500 golf courses worldwide (Tanner &ange, 2005). With over 18,300 golf courses in the U.S. alone (Baris,ohen, Barnes, Lam, & Ma, 2010) encompassing over 2.7 million

cres (Colding, Lundberg, Lundberg, & Andersson, 2009), golf hasecome an appreciable portion of land use in the United States. Thecological impacts of golf courses are not always straightforward,

∗ Corresponding author. Tel.: +1 573 864 2939; fax: +1 573 864 2939.E-mail address: SemlitschR@missouri.edu (R.D. Semlitsch).

169-2046/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.landurbplan.2014.01.013

© 2014 Elsevier B.V. All rights reserved.

and popular opinion of the impacts of golf courses on the environ-ment can be in direct opposition of scientific studies (Wheeler &Nauright, 2006). Further, results from the scientific literature canbe seemingly as contradictory in their reporting (see below). Tobetter understand the ecological impacts of golf courses, it is nec-essary to move beyond the deceptive dichotomy of “good” or “bad”(Sheil & Meijaard, 2010), and to measure impacts using ecologicallymeaningful responses for target organisms.

A major focus of discussion regarding known or suspected eco-logical impacts of golf courses has been water quality, typically

focusing on chemical toxicology (Wheeler & Nauright, 2006). Golfcourses depend on agrochemicals for pest control, turf manage-ment, and esthetic purposes. Although there have been manystudies on agricultural chemicals in groundwater and surface

1 nd Urb

wasfBogWctoctsheDp

icsLspMadpltsiiduFei

mataosaNlts&pratptLiiotsai

8 M.J. Mackey et al. / Landscape a

ater, it is usually not appropriate to extrapolate results fromgricultural monitoring studies to golf course studies due to theignificantly different management practices, plant canopy, sur-ace mat, and root system of turf (Cohen, Svrjcek, Durborow, &arnes, 1999; Kenna, 1995). A study in southeastern North Car-lina reported generally greater nitrate levels in streams leavingolf courses compared to streams entering golf courses (Mallin &heeler, 2000), though concentrations varied considerably among

ourses. A study conducted on two golf courses under construc-ion and five in operation in Canada found course construction andperation had a significant impact on alkalinity, nitrogen, and baseation concentrations of streams downstream of courses comparedo forested reference streams (Winter & Dillon, 2005). A similartudy found significant differences in certain benthic algal taxa ineadwater streams downstream of golf courses compared to ref-rence streams (Winter, Dillon, Paterson, Reid, & Somers, 2003).ifferences were attributed to greater nutrient enrichment, higherH, and higher disturbance from the golf courses.

Not all studies on the effects of golf courses find significantmpacts on water chemistry. A study conducted on three golfourses in North Carolina examined the presence of chemicals inurface waters and found no chemical impact (Ryals, Genter, &eidy, 1998). Another study in the Pacific Northwest monitoredurface waters monthly following the application of fertilizers andesticides and found no significant detection of chemicals (Hindahl,iltner, Cook, & Stahnke, 2009). In fact, the most extensive meta-

nalysis to date of golf course water quality monitoring analyzedata from across 40 studies involving 80 courses over a 20-yeareriod and found relatively few pesticide detections or exceeded

imits in surface water (Baris et al., 2010). The authors attributehis finding to the combination of two factors: (1) the fact that turfystems act as a living filter, and (2) the practice of applying min-mal pesticides to the roughs, which typically surround the morentensively managed tees, greens, and fairways. As turf science haseveloped due to public scrutiny and pesticide registration eval-ations by the U.S. Environmental Protection Agency under theederal Insecticide, Fungicide, and Rodenticide Act (FIFRA; Barist al., 2010), it is possible that water chemistry and chemical runoffs no longer the foremost ecological concern at many golf courses.

The other primary impact of golf course development andaintenance is the physical alteration of the landscape. Habitat

lteration and destruction is known to be one of the biggest threatso biodiversity (Wilcove, Rothstein, Dubow, Phillips, & Losos, 1998),nd is therefore an obvious focus of ecological research. The clearingf natural vegetation, deforestation, the destruction of natural land-capes and habitat, and changes in local topography and hydrologyre all possible land use effects that result from golf (Wheeler &auright, 2006; Winter et al., 2003). A study in southeast Queens-

and, Australia found many golf courses to have negligible value aserrestrial habitat refuges which supported mostly urban-adaptedpecies compared to reference eucalypt forests (Hodgkison, Hero,

Warnken, 2007). In stream ecosystems, reach-level channel mor-hology is influenced by valley slopes, bed and bank material,iparian vegetation, and the supply of upslope water, sediments,nd wood (Montgomery & MacDonald, 2002). Human actions athe landscape scale can disrupt these factors that maintain streamrocesses and their associated biota and often result in habi-at that is both degraded and less heterogeneous (Allan, 2004).andscape level experiments have documented how such phys-cal alterations to the terrestrial landscape can have downstreammpacts on habitat and water quality of streams through alterationsf hydrology (Likens & Bormann, 1974). Headwater streams in par-

icular represent the maximal interface between aquatic-terrestrialystems (Lowe & Likens, 2005), and their sensitivity to land disturb-nce makes them both important and useful for studying land-usempacts.

an Planning 125 (2014) 17–27

Despite the potential negative impacts golf course develop-ment and maintenance can have on landscapes, a number ofstudies have found golf courses to have a general positive con-servation value on the species studied, including amphibians(Boone, Semlitsch, & Mosby, 2008; Colding, Lundberg, Lindberg,& Andersson, 2009), butterflies (Porter, Pennington, Bulluck, &Blair, 2004) pond breeding macroinvertebrates (Colding, Lundberg,Lindberg, & Andersson, 2009), reptiles (Harden, Price, & Dorcas,2009; Mifsud & Mifsud, 2008), birds (Merola-Zwartjed & DeLong,2005; Rodewald, Rodewald, & Santiago, 2004), and mammals(Eisenberg, Noss, Waterman, & Main, 2011). A review of thescientific literature studying land-use effects of golf courses onbiota found that the ecological value of golf courses increasedas the anthropogenic impact on the surrounding land increased(Colding & Folke, 2009). Therefore, the ecological impact andconversely the conservation value of a golf course will dependupon the landscape in which the golf course exists. Addition-ally, studies of golf course impacts have often focused on singlespecies or closely related taxa most likely as a result of logisti-cal constraints of field work. Many of the species studied havedramatically different life histories and therefore respond differ-ently to landscape alterations. Having a clear understanding ofspecies life history in conservation efforts offers good opportuni-ties to gain insight into the mechanisms behind species responseto land-use change (Verheyen, Honnay, Motzkin, Hermy, & Foster,2003).

Salamanders are especially prolific in headwater streamsof eastern North America where they are the most abundantvertebrate organism (Peterman, Crawford, & Semlitsch, 2008;Nowakowski & Maerz, 2009). Because they are highly philopatric,long-lived, and occur in relatively stable populations, streamsalamanders may be more appropriate and reliable indicatorsof biodiversity and habitat quality in stream ecosystems thanmany fish or macroinvertebrates (Welsh & Ollivier, 1998). Streamsalamanders may also be useful indicators of ecosystem healthbecause they are adversely affected by deforestation and phys-ical disturbance (Orser & Shure, 1972; Petranka & Smith, 2005;Willson & Dorcas, 2002), siltation (Lowe, Nislow, & Bolger, 2004;Welsh & Ollivier, 1998), and stream acidification (Kucken, Davis,Petranka, & Smith, 1994). Stream salamanders also have com-plex life cycles; reproduction and larval growth occurs in aquatichabitat followed by metamorphosis and sexual maturation interrestrial riparian habitat (Petranka, 1998). Thus, unlike someorganisms, their persistence in headwater streams is explicitlydependent upon the quality of both aquatic and terrestrial sys-tems.

The purpose of our study was to examine the influence of golfcourse management on the abundance and diversity of streamsalamanders in the southern Appalachian Mountains. We focusedon habitat changes that may have affected the abundance ofboth aquatic larvae and terrestrial juveniles and adults. Our pri-mary hypothesis was that land-use effects would be greatest instream samples that occurred directly on the golf course fairwaysbecause they would be directly impacted by habitat alteration dur-ing construction and routine maintenance of fairways comparedto upstream or downstream areas that had no direct alteration ormaintenance. We hypothesized that habitat alteration would eitherdirectly or indirectly affect aquatic larvae and terrestrial juvenilesthrough loss of canopy cover, siltation, and a reduction in leaf litter,woody debris, and soil moisture based on similar studies in refer-ence streams in the surrounding Nantahala National Forest (e.g.,Crawford & Semlitsch, 2007, 2008; Peterman & Semlitsch, 2009).

Secondarily, we tested whether chemical runoff in downstreamsamples was detectable at levels that might influence the abun-dance and diversity of salamanders on golf courses. Our objectiveswere to sample larvae, juvenile, and adult salamanders across a

nd Urb

rplgia

2

2

AwklrohctptttS

tqBlmcwacbMssmc

amvcppw2il

2

2stscbio

M.J. Mackey et al. / Landscape a

eplicate set of 10 golf courses and test our hypotheses by com-aring abundance and diversity between samples from streams

ocated upstream, downstream, and through the fairway of eacholf course. Further, we use an information theoretic approach todentify specific habitat features that are associated with either thebundance or diversity of stream salamanders.

. Materials and methods

.1. Study sites

Our study was conducted on 10 golf courses in the southernppalachian region of western North Carolina, USA. All coursesere typical for the region (Appendix A) and located within a 30-

m radius of the Highlands Biological Station, North Carolina forogistic reasons. We repeatedly sampled stream salamanders at 60eplicate transects each of which covered a 25-m long segmentf stream. All transects were selected to fall within one of threeabitat treatments which represented their location relative to golfourse fairways (downstream, through, upstream). Two replicateransects, at least 100 m apart, of each habitat treatment were sam-led on each golf course, totaling six transects per course, 60 totalransects. By necessity, three of our 60 transects were located onhe same stream but were separated by >50 m which is greater thanhe average home range of these salamanders (Barbour, Hardin,chafer, & Harvey, 1969; Ashton, 1975).

We focused sampling on both larvae and adult salamanderso fully elucidate the response of populations to changes inuality of aquatic and terrestrial environments (Price, Cecala,rowne, & Dorcas, 2010). We used a modified version of the leaf

itter bag design to sample in-stream larval and newly meta-orphosed salamanders (Pauley & Little, 1998). Leaf bags were

onstructed by filling a 70 cm × 70 cm square of 1.9-cm meshith deciduous leaf litter. Five bags were evenly distributed

long each transect for a total of 30 leaf bags deployed at eachourse and 300 bags across all 10 golf courses. We checked leafags during daylight hours three times (once each month foray, June, July 2009) during the summer. Bags were checked by

haking them over a 40 cm × 30 cm × 8 cm white tray. All capturedalamanders were identified to species, measured to the nearestm for snout-vent length (SVL), and released at their site of

apture.To sample adult salamanders, the stream bank adjacent and par-

llel to each 25-m aquatic transect was searched at night once eachonth (June, July, August 2009) during the summer. We used a

isual encounter survey (VES) during the three sample dates toapture surface-active salamanders. During each survey, two peo-le simultaneously searched the area 0–2.5 m from the stream, oneerson on each bank or side, for a total of 20 min. As salamandersere captured, they were placed in sealable plastic bags until the

0 min had expired. At the end of each survey all salamanders weredentified to species, measured to the nearest mm for snout-ventength (SVL), and released at the site of capture.

.2. Habitat measurements

We measured habitat variables once each month (May, June, July009) at each of the 60 stream transects in our study. In the tran-ect adjacent to each stream approximately 1.5 m from the bank, weook varying number of replicate measurements of soil moisture,oil temperature, leaf litter depth, ground surface temperature, per-

ent canopy cover, and width of un-mown buffer (stream edge touffer edge on each bank). In the wetted portion of the stream

n the center of each transect, we took replicate measurementsf water temperature, dissolved oxygen, percent sedimentation,

an Planning 125 (2014) 17–27 19

surface water velocity, substrate composition proportions, quali-tative estimates of cover rocks, and qualitative estimates of coursewoody debris levels. Replicate values of each habitat variable wereaveraged within each sampling date, but kept separate by date toaccount for seasonal effects.

We measured percent canopy coverage from the center of thestream transect using a spherical crown densiometer measured inthe four cardinal directions. We estimated stream width by measur-ing the wetted width of the stream at four locations approximately6 m apart to the nearest 0.01 m. Percent sedimentation was deter-mined as the surface sediment covering the streambed, and wasmeasured using a 50 cm × 50 cm quadrant that was divided into 25equal-sized square sections. We also used this method to quantifypercentages of rocks and sand in the stream bed. Stream depth wasmeasured to the nearest cm midstream at four locations approx-imately 6 m apart. We measured leaf litter depths to the nearestcm from two locations approximately 8 m apart on each side ofthe stream using a hand ruler. Stream temperatures, dissolvedoxygen, and conductivity were measured from the center of eachstream transect using a handheld YSI 85 meter. We measured pHusing an Extech ExStik® meter from the center of the stream tran-sect. Stream surface velocity was obtained using the float method(Gordon, McMahon, & Finlayson, 1992) at four locations in thestream transect approximately 6 m apart. A small fishing bobberwas dropped into the stream and the time (s) taken to float 1 mwas measured using a meter stick and a stop watch. Coarse woodydebris was visually estimated on a five point scale with 0 definedas no coarse woody debris and 4 defined as extensive woody debrisspanning the width of the stream at each transect. We measuredground surface temperature using an infrared thermometer fromtwo locations approximately 8 m apart on both sides of the stream.We measured soil temperature to the nearest 0.1 ◦C from two loca-tions approximately 8 m apart on both sides of the stream using asoil temperature probe. Buffer width was measured as the straight-line distance from the center of the stream transect to the nearestmanaged area of the golf course to the nearest m. Soil moisture wasobtained by collecting soil samples from two locations approxi-mately 8 m apart along the transect on both sides of the streamusing a hand shovel. Samples were placed and sealed into zip-locbags, returned to the laboratory and weighed to the nearest 0.01 g,dried at 35 ◦C for 24 h, and again weighed to calculate percent soilmoisture.

2.3. Chemical testing

Because of the ubiquitous presence of nitrate in water qualitystudies and its potential for adversely affect amphibian survival(Rouse, Bishop, & Struger, 1999) we tested water samples from eachstream transect for the presence of nitrate. A 25-ml water sam-ple was collected from the center of each stream transect between15 and 26 June and returned to the laboratory the same day fortesting using a LaMotte TesTab reagent test kit for nitrate/nitrogen(range 0–15 ppm; minimum detection range 1 ppm). We also testedfor the presence of pesticides in leaf litter. We chose to test leaflitter rather than water because salamanders of all age classesspend considerable time in contact with leaf litter and because leaflitter is likely to remain stationary longer than water in lotic sys-tems and accumulate any long-lived pesticides (Mississippi StateChemical Laboratory, personal communication). We collected a onequart leaf litter sample from the center of each of the 20 down-stream transects that represented worst-case scenarios and fromsix randomly chosen upstream transects that represented con-

trol/reference areas from pre-deployed leaf bags from 27 July to7 August. We assumed that differences between chemical lev-els in upstream and downstream transects would represent theeffects of runoff due to maintenance on fairways and indicate

20 M.J. Mackey et al. / Landscape and Urb

Table 1A priori models used in the prediction of abundance and diversity of lar-vae/metamorph captures and juvenile/adult captures.

Model name Model parameters

Larvae and metamorphsNull NABuffer Buffer, leaf depthSubstrate Sand, sediment, leaf depth, cover rocks,

CWDLeaf retention CWD, leaf depthHabitat complexity CWD, cover rocks, buffer, sand,

sediment, stream depth, leaf depthManagement CWD, leaf depth, buffer, ageHydrology Stream depth, stream width, flowWater quality D.O., pHGlobal All variables

Juveniles and adultsNull NAMicroclimate Soil temperature, soil moistureCover objects CWD, cover rocksLeaf retention CWD, leaf depthHabitat complexity CWD, cover rocks, buffer, sand,

sediment, stream depth, leaf depthBuffer Buffer, leaf depthManagement CWD, leaf depth, buffer, ageHydrology Stream depth, stream width, flowCrawford Leaf depth, soil moisture, soil

wapPtcpsaiS

2

miWtftdaodtdfwitamebwcwi

temperatureGlobal All variables

hether it might be an additional factor influencing salamanderbundance. Samples were collected and sealed into freezer-durablelastic bags and frozen immediately upon returning from the field.rior to sample collection, we obtained a list of the top used pes-icides (fungicides, herbicides, and insecticides) from the 10 golfourse superintendents to compile a target list of regionally appliedesticides to screen during analyses. Samples were sent to Mis-issippi State Chemical Laboratory (Mississippi State, MS, USA)nd “QuEChERS” was used for determination of pesticides residuesn the leaf litter samples (Anastassiades, Lehotay, Stajnbaher, &chenck, 2003).

.4. Data analyses

We used our spatially and temporally replicated counts of sala-anders from both leaf bag captures and VES to identify differences

n salamander communities in relation to golf course management.e fit binomial mixture models (Royle, 2004) to determine impor-

ant predictors of stream salamander abundance while accountingor temporal variation (sampling time- May, June, July) in detec-ion probability. For this analysis, we independently analyzed countata from leaf bag captures in aquatic habitat (larvae/metamorphs)nd VES captures in terrestrial habitat (juveniles/adults). Due toverdispersion in our count data, all models for salamander abun-ance were fit with a negative binomial error structure. To identifyhe overall effects of golf course management on stream salaman-er abundance, we initially compared models in which a singleactor representing golf course treatment (down, through, and up)as considered as a predictor of abundance. We then sought to

dentify the specific differences in habitat characteristics amongransects which best explained overall variation in salamanderbundance. This model comparison included a limited set of a prioriodels in which a number of specific habitat features were consid-

red as covariates for abundance (Table 1). This model set includedoth null (intercept-only) and global models (all parameters), as

ell as a number of models representing biologically-meaningful

ombinations of predictor variables. Individual habitat variablesere selected based on previous studies and pilot data suggest-

ng their potential influence on salamander abundance (Appendix

an Planning 125 (2014) 17–27

B). We calculated correlation coefficients for all pairwise combina-tions of predictor variables and combined only those which wereuncorrelated with other variables (Pearson’s r ≤ 0.7). Survey periodwas included as a covariate for detection probability in all modelsto account for variation in activity across the summer.

To calculate the effective number of species at each transect(hereafter “diversity”), we used the cumulative counts of eachspecies from across capture periods to calculate the exponen-tial of the Shannon–Weiner index [H′ = −�(pi ln pi)], where pi isthe proportional abundance of species i (Jost, 2006). These diver-sity index values were calculated for both larvae/metamorphs andjuveniles/adults and were analyzed separately. We used one-wayanalysis of variance (ANOVA) to test the hypothesis that salaman-der diversity differed among the three stream treatment locations(down, through, and up) and used Tukey’s honestly-significant-difference test to make pairwise comparisons of these streamtreatments. We then used generalized linear models to identifythe specific habitat features which best explained variation indiversity among treatments. We compared a number of a priorimodels for diversity, including biologically-relevant combinationsof non-correlated habitat variables (Pearson’s r ≤ 0.7), a globalmodel containing all predictor variables, and a null (intercept-only)model (Table 1). We based a priori model selection on publishedliterature as well as observations and pilot data (Appendix B).

To rank models for both salamander abundance and diversityin relation to habitat characteristics, we used Akaike’s Informa-tion Criterion adjusted for sample size (AICC). We then derived the�AICC for each model, which is the difference in AICC betweeneach model and the best model in the set. A �AICC less than 2 sug-gests that there is substantial support for the model (Burnham &Anderson, 2002). We also calculated Akaike weights (ωi) which rep-resent the probability that the given model is the best among theentire set of candidate models (Burnham & Anderson, 2002). Forall datasets, we determined the 95% confidence model set and usedmodel averaging and unconditional variance estimation to assessthe contribution of individual habitat variables to the model’s pre-dicting power. This can be particularly informative when ωi valuesare low and when there is no clear top model among the set of can-didate models. All statistical analyses were performed in programR (R Development Core Team, 2012). Abundance models were fitusing the unmarked package (Fiske & Chandler, 2011); model selec-tion procedures were performed using the package AICcmodavg(Mazerolle, 2013).

3. Results

3.1. Total captures and species detections

A total of 2215 salamanders were detected across all transectsthroughout the summer of 2009. Of this total, 1015 salaman-ders were caught during leaf litter bag sampling and 1200 werecaught during nighttime VES. Throughout the field season wedetected nine salamander species: seal salamander (Desmog-nathus monticola), Ocoee salamander (Desmognathus ocoee),black-bellied salamander (Desmognathus quadramaculatus), BlueRidge two-lined salamander (Eurycea wilderae), three-lined sala-mander (Eurycea guttolineata), red salamander (Pseudotriton ruber),spring salamander (Gyrinophilus porphyriticus), southern gray-cheeked salamander (Plethodon metcalfi), and red-spotted newt(Notophthalmus viridescens). Notophthalmus viridescens occurredinfrequently and was only captured during VES and was therefore

omitted from analyses. P. metcalfi, although caught frequently dur-ing VES of riparian habitat, was also omitted from analyses becauseit is strictly a terrestrial species without aquatic larvae and was notcaptured by leaf litter bags.

M.J. Mackey et al. / Landscape and Urban Planning 125 (2014) 17–27 21

F venileu verall

3

taswuTadbsd

mlitd

ig. 1. Predicted abundance and species diversity of both larvae/metamorph and jupstream of golf course fairways. Intervals represent fitted values (±95% CI) from o

.2. Comparison of stream treatments

Stream transects positioned through the fairway were predictedo have significantly lower abundance of both larvae/metamorphsnd juveniles/adults than transects either upstream or down-tream of the fairway (Fig. 1). Specifically, transects on the fairwayere predicted to have 38% fewer larvae and metamorphs thanpstream transects and 31% fewer than downstream transects.ransects through the fairway were also predicted to have 52%nd 53% fewer adults and juveniles than both upstream andownstream transects, respectively. Stream transects above andelow managed golf course fairways (up and downstream) did notignificantly differ in their predicted abundance of stream salaman-ers.

There was also a significant difference among stream treat-ents in the diversity of larvae and metamorphs captured from

eaf bags (F = 6.11, df = 2, 57, p = 0.004; Fig. 1). The transects pos-tioned through the fairway had significantly lower diversity thanhe upstream (−37.0%; Tukey’s HSD test, p = 0.004), but did notiffer significantly from the downstream (−26.1%; Tukey’s HSD

/adult salamanders on stream transects that are downstream, running through, or habitat treatment models.

test, p = 0.100) transects. Larval and metamorph diversity of theupstream and the downstream transects also did not differ sig-nificantly (Tukey’s HSD test, p = 0.358). There was no significantdifference in diversity of adults and juveniles detected during VESamong the three stream treatments (F = 0.45, df = 2, 57, p = 0.448;Fig. 1).

3.3. Habitat variables

In our study, golf course management appeared to be relatedto differences in a number of stream and riparian habitat char-acteristics (Fig. 2) which may function as the proximate causesfor changes in the abundance and diversity of stream salaman-ders among habitat treatments. As a result, we sought to identifythe specific differences in habitat characteristics which best pre-dicted the abundance and diversity of stream salamanders among

transects. Of the nine a priori models predicting larval and meta-morph abundance (Table 1), the Buffer model which included theleaf depth and buffer width variables was the most supported(ωi = 0.44; Table 2). This can be interpreted as the Buffer model

22 M.J. Mackey et al. / Landscape and Urban Planning 125 (2014) 17–27

F se stre

htopHmrpMnap

sCodawcmdn

3

tti

ig. 2. Boxplots illustrating the variation in habitat characteristics among golf cournable plotting on a common Y-axis.

aving a 44% probability of being the best predictive model amonghe set of candidate models. Leaf litter depth had a positive effectn larval and metamorph abundance and was the only significantarameter estimate identified by model averaging (Table 3). Theabitat Complexity model received the most support among theodels for adult and juvenile abundance (ωi = 0.38). This model

epresented the hypothesis that salamander abundance would beositively associated with the structural complexity of the habitat.odel averaging indicated that the abundance of adults and juve-

iles was positively associated with leaf litter depth and negativelyssociated with the amount of sand and the number of cover rocksresent (Table 3).

The a priori model best explaining larval and metamorph diver-ity was the Retention model, which consisted of the leaf depth andWD parameters (ωi = 0.62; Table 2). Following model averaging,nly CWD was found to significantly and positively affect larvaliversity (Table 3). Of the ten a priori models predicting juvenilend adult diversity, all of the top models had relatively low modeleights, with the Null model garnering the most support followed

losely by the Micoclimate model (Table 2). In the Microclimateodel, moister soil was positively associated with higher levels of

iversity, although this parameter, and all others assessed, wereot significant following model averaging (Table 3).

.4. Pesticides and nitrate

Nitrate was not detected at or above the minimum detec-ion range (1 ppm) in any of the water samples taken fromhe 60 stream transects in this study. Of the 16 chem-cals we tested, only Propiconazole and 4OH-Chlorothalonil

eam transects. The data for each habitat variable have been rescaled from 0 to 5 to

were detected at low concentrations. Propiconazole, an activeingredient of a fungicide, was detected in eight of the 20downstream transects and had an average detection propor-tion of 0.0135 ppm. 4OH-Chlorothalonil, a breakdown productof the fungicide active ingredient Chlorothalonil, was detectedin nine of the 20 downstream transects with an average detec-tion proportion of 0.0162 ppm and was also detected in one ofthe upstream transects (see Appendix C for complete pesticideresults).

4. Discussion

Stream transects located directly on fairways contained lowerabundance of larval, metamorph, juvenile, and adult salamandersthan either upstream or downstream transects. The diversity ofaquatic larval and metamorph salamanders on fairways was alsoreduced but only compared to the upstream transects, and terres-trial juvenile and adult diversity did not differ among the threetransect locations. We found that by comparing stream locationrelative to the center of golf course activity that the majority ofgolf course effects are likely the direct results of habitat alterationand maintenance on fairways. We also found little or no detectablechemical runoff in areas downstream from maintained fairways.Although all stream transects that we sampled were located ongolf course property, upstream and downstream areas appeared torepresent high-quality forested habitat and did not appear to be

compromised by proximity to fairways. Thus, the primary effectsare limited to the portion of stream actually running throughthe fairways. Although we did not include any off-site controlstreams in our study or analysis, data from previous studies using

M.J. Mackey et al. / Landscape and Urban Planning 125 (2014) 17–27 23

Table 2A priori models predicting the abundance and diversity of stream salamanders on golf courses in western North Carolina, USA.

Model Ka AICcb �AICcc ωi

d LL

Abundance modelsLarvae and metamorphs

*Buffer 7 1066.91 0 0.44 −525.38*Retention 7 1067.40 0.49 0.35 −525.62*Null 6 1069.25 2.34 0.14 −529.07*Management 9 1071.81 4.90 0.04 −525.11Substrate 10 1073.59 6.68 0.02 −524.55WaterQuality 7 1073.97 7.05 0.01 −528.91Hydrology 8 1076.79 9.88 0 −528.98HabComplexity 12 1078.44 11.53 0 −523.90Global 16 1090.72 23.81 0 −523.03

Adults and juveniles*HabComplexity 12 1014.39 0 0.38 −491.87*Buffer 7 1015.43 1.05 0.22 −499.64*CoverObjects 7 1017.03 2.64 0.10 −500.44*Retention 7 1017.13 2.74 0.10 −500.49*Crawford 8 1017.98 3.59 0.06 −499.58*MicroClim 7 1018.07 3.68 0.06 −500.96*Null 5 1018.26 3.87 0.05 −503.57Management 9 1019.79 5.40 0.03 −499.09Hydrology 8 1024.36 9.97 0 −502.77Global 17 1027.03 12.64 0 −489.23

Diversity modelsLarvae and metamorphs

*Retention 4 200.93 0 0.62 −96.10*Management 6 203.31 2.38 0.19 −94.86*Buffer 4 204.78 3.85 0.09 −98.03*Substrate 7 206.26 5.32 0.04 −95.05Null 2 207.90 6.96 0.02 −101.84WaterQuality 4 208.65 7.71 0.01 −99.96Hydrology 5 208.95 8.01 0.01 −98.92HabComplexity 9 210.21 9.27 0.01 −94.30Global 13 222.02 21.09 0 −94.05

Adults and juveniles*Null 2 138.53 0 0.33 −67.16*MicroClim 4 139.30 0.76 0.23 −65.28*CoverObjects 4 140.96 2.43 0.1 −66.12*Buffer 4 141.13 2.60 0.09 −66.20*Retention 4 141.43 2.89 0.08 −66.35*Crawford 5 141.44 2.91 0.08 −65.16*Management 6 142.29 3.76 0.05 −64.35Hydrology 5 143.02 4.49 0.04 −65.95HabComplexity 9 146.68 8.15 0.01 −62.54Global 14 154.12 15.59 0 −58.39

An asterisk preceding model names indicate that the model was included in the 95% confidence set for model averaging.a

best

sa&mvard22fibaliprid

a

Number of parameters estimated in each model.b Akaike’s Information Criterion adjusted for small sample size.c The difference between the AIC value for a given model and the AIC value of the

imilar sampling techniques in relatively pristine streams werevailable for comparison (Crawford & Semlitsch, 2007; Peterman

Semlitsch, 2009; Peterman, 2008). For example, for one of theost abundant species in the region, the average E. wilderae lar-

ae captures per sampling period in our upstream (1.04 larvae/bag)nd downstream (0.83 larvae/bag) transects fall within the rangeeported from control stream transects in a previous study con-ucted in nearby (∼50 km) Nantahala National Forest (Control 1,006: 0.67 larvae/bag; Control 1, 2007: 0.71 larvae/bag; Control, 2007: 2.22 larvae/bag; (Peterman & Semlitsch, 2009). Further,ve stream-associated species were detected at these control sitesetween 2004 and 2007 (Peterman, 2008), whereas seven stream-ssociated species were detected at each of our three streamocations (upstream, through, and downstream) on golf coursesn the summer of 2009. The two species detected on golf courseroperties and not detected at the national forest sites were P.uber and the E. guttolineata, species that utilize slower mov-

ng or higher order streams found in the region (unpublishedata).

Our analysis of in-stream and riparian habitat characteristics predictors for stream salamander abundance and diversity can

supported model for each data set.

inform stream management efforts to mitigate undesirable anthro-pogenic impacts and improve stream habitat on golf courses.Our information theoretic (AIC) approach to model selectionrevealed that leaf litter depth, CWD, soil moisture, and bufferwidth parameters were positively related to salamander abun-dance and diversity. It has long been documented that leaf litterinput is the most important basal energy source in shaded head-water streams (Vannote, Minshall, Cummins, Sedell, & Cushing,1980). This coarse particulate organic matter is colonized bymicrobes which then become a food source for aquatic macroin-vertebrates that are prey for salamanders (Johnson & Wallace,2005; Johnson, Wallace, Rosemond, & Cross, 2006). Leaf litteralso provides necessary refugia for both aquatic and terres-trial life stages of salamanders (Crawford & Semlitsch, 2008;Petranka, 1998). Reduced standing stock of organic matter is oftena consequence of human land use and can result in a reduc-tion of a stream’s capacity to intercept nutrients (Meyer, Paul,

& Taulbee, 2005). Golf courses specifically have reduced detri-tus sources due to removal of non-turf vegetation for tee boxes,fairways, and greens. Leaves and dead plant matter remaining oncourses are removed for esthetic purposes through routine course

24 M.J. Mackey et al. / Landscape and Urb

Table 3Model-averaged parameter estimates (ˇ), unconditional standard errors, and 95%confidence intervals for the analysis of stream salamander abundance and diversity.

Parameter ̌ SE Lower CI Upper CI

Abundance modelsLarvae and metamorphs

LeafDepth 0.249 0.116 0.021 0.477Buffer 0.066 0.099 −0.128 0.260Age −0.062 0.094 −0.247 0.122CWD −0.003 0.111 −0.221 0.215

Adults and juvenilesCoverRocks −0.316 0.129 −0.568 −0.064Sand −0.313 0.113 −0.535 −0.091LeafDepth 0.286 0.119 0.053 0.519Buffer −0.205 0.111 −0.422 0.013CWD 0.192 0.128 −0.058 0.442SoilMoist 0.184 0.116 −0.042 0.411StreamDepth −0.120 0.100 −0.316 0.076SoilTemp −0.018 0.105 −0.224 0.188Sediment 0.016 0.091 −0.162 0.194

Diversity modelsLarvae and metamorphs

CWD 0.402 0.202 0.006 0.797LeafDepth 0.250 0.204 −0.150 0.651Sand −0.187 0.186 −0.552 0.177Sediment 0.133 0.189 −0.237 0.503CoverRocks −0.085 0.207 −0.491 0.320Buffer −0.083 0.222 −0.519 0.352

Adults and juvenilesAge −0.190 0.102 −0.390 0.009SoilMoist 0.177 0.106 −0.030 0.384CoverRocks −0.127 0.098 −0.319 0.065LeafDepth 0.118 0.122 −0.122 0.357Buffer −0.082 0.116 −0.309 0.145CWD 0.027 0.120 −0.208 0.263SoilTemp 0.000 0.109 −0.214 0.214

P

mis3wcha1tbt2

act2a(H2ioahpMt(in

pesticide regulations since the early 1990s (Baris et al., 2010),

arameters in bold face type are significant (CI did not overlap zero).

aintenance. Average leaf litter depth at the aquatic-terrestrialnterface along streams was 97% shallower in our fairway tran-ects than in our downstream and upstream transects (upstream –.38 cm; through – 0.09 cm; downstream – 3.10 cm). Building smalloody debris dams to collect leaf litter along streams into golf

ourse management could improve food availability and provideabitat for aquatic organisms as well as improve nutrient retentionnd water quality for stream communities on fairways (Cummins,974; Aldridge, Brooks, & Ganf, 2009). We encourage researcherso test the utility of coarse woody debris and leaf retention toetter understand its restoration effects on stream communi-ies and the breadth of its application (see Hall & Fleishman,010).

We also found buffer width to be important for abundancend diversity of stream salamanders. Riparian buffer strips adja-ent to streams have been used in managed forests for more thanwo decades (Vesely & McComb, 2002; Crawford & Semlitsch,007), and they can mitigate effects of human land use suchs chemical runoff, siltation, and increased water temperaturesJones, Helfman, Harper, & Bolstad, 1999; Lowrance, Todd, Fail,endrickson, Leonard, & Asmussen, 1984; Vesely & McComb,002). Riparian vegetation, especially trees, along streams located

n actively managed areas of golf courses is often greatly reducedr completely replaced by turf grass. This reduces or eliminates

primary source for leaf litter input to streams and riparianabitat occupied by salamanders. Because of the life history andhysiology of salamanders (Southerland, Jung, Baxter, Chellman,ercurio, & Volstad, 2004) and their sensitivity to microhabi-

at alterations such as the loss of leaf litter and forest canopy

Welsh & Ollivier, 1998), we did not expect to capture as manyndividuals in streams on fairways as were captured (270 juve-iles/adults, 238 larvae). Further, results for the terrestrial juvenile

an Planning 125 (2014) 17–27

and adult diversity did not differ among the upstream, through,and downstream transects. Our results indicate that stream sala-manders are currently present in streams located on golf courses,albeit in lower abundances and diversity for the aquatic stages.We speculate that populations upstream and downstream of thefairways may provide a source to compensate losses within thestream directly on fairways, especially if reproduction throughfairways does not equal mortality. In the absence of active rein-troductions, colonization of restored habitat is entirely reliantupon the dispersal of organisms from extant populations (Bond& Lake, 2003). For populations dependent on immigration fromadjacent habitat, the degradation of either upstream or down-stream areas by additional or future land use would likely disrupteffective source-sink dynamics and lead to greater declines in abun-dance. Thus, land use in the surrounding landscape should beconsidered critical in assessing the overall effects of golf courses.Nearly all of the golf courses we studied near Highlands, NC weresurrounded by forested habitat with little or low-impact landuse.

Although golf has often received negative attention in the pastdue to chemical pesticide and fertilizer use (Baris et al., 2010),the results of pesticide and fertilizer tests from our study indi-cate that this is unlikely a major concern at our study streamsprimarily because abundance of aquatic larvae in downstreamand upstream transects were similar. For example, Propicona-zole was detected at an average concentration of more than 30times below the U.S. Environmental Protection Agency (EPA) max-imum allowable concentration (0.425 ppm; Baris et al., 2010).4OH-Chlorothalonil, a breakdown product of the fungicide activeingredient Chlorothalonil, was detected at levels more than 900times lower than toxic concentrations. Lotic systems such as ourmontane streams are characterized by flowing water and chemicalanalyses of lentic or stagnant waters may yield different results.Our chemical analyses were, however, taken from leaf litter sam-ples that are more likely to remain in place for long periods and mayabsorb chemicals unlike water samples (Mississippi State Chemi-cal Laboratory, personal communication). Although our chemicaltests were relatively extensive, it should be noted that the pres-ence or absence of a chemical reported from our analysis refersonly to the time the sample was collected, not the entire dura-tion of the study. Likewise, the timeframe of our entire study wasshort and may not accurately depict trends in nitrate and pesti-cide levels over longer timeframes that include spring snow meltthat flushes chemicals from the system or in drought years thatmight concentrate chemicals. A study examining the nutrient dis-charge of five coastal North Carolina golf courses reported coursesto have generally greater nitrate levels leaving the courses com-pared to entering the courses, but the concentrations varied greatlyamong courses (Mallin & Wheeler, 2000). Though the contamina-tion of water bodies from golf course maintenance should still beconsidered a potential concern, our findings are consistent witha large-scale review in the United States which found no signifi-cant toxicological impacts from golf courses to groundwater andsurface water (Cohen, Svrjcek, Durborow, & Barnes, 1999). A recentstudy updated and expanded this database to include all golf coursewater quality data meeting review criteria over a 20-year period(Baris et al., 2010). The study reports that pesticide levels exceed-ing surface and ground water quality criteria were rarely observed.Total phosphorous, that was added to the database, appeared to bethe analyte of greatest concern in surface waters. This result waspublished after the onset of our study. Thus, we believe intensepublic scrutiny has led to great improvements in turf science and

however continued care must be taken to ensure pesticide andfertilizer use does not compromise non-target organisms such assalamanders.

nd Urb

5

aKi&sUtottmpFigt(rmsptohtoawbh

A

toeswpBAL

A

Hga

C

C

H

H

H

M

T

M.J. Mackey et al. / Landscape a

. Conclusions

Considering ecological processes are now more widelyccounted for in golf course design and management (Jackson,elly, & Brown, 2011), courses could increasingly become an asset

n ecosystem management and biodiversity conservation (Colding Folke, 2009) and serve as models for ecological awareness andustainability. With an estimated 27.1 million golfers in 2009 in thenited States alone (NGF, 2010), integration of ecological principles

o golf courses has the capability of reaching a large audience thattherwise may not be exposed to conservation concepts and prac-ices. Golf courses can serve as opportunities for demonstration,he translation of scientific understanding into metrics of perfor-

ance and cost under real world conditions, that is key in therogression of fundamental research to applied science (Hall &leishman, 2010). From the application of our results, and build-ng on previous studies, we suggest several procedures to improveolf course management: (1) maintain or restore riparian vege-ation that includes a buffer at least 15 m from the stream edge,2) maintain or restore a tree stocking density of ∼50% within theiparian buffer to shade and provide a source of leaf litter, and (3)aintain or add small woody debris dams that retain leaf litter in

treams to provide increased refuge, maintain soil moisture, androvide food sources for salamanders. Further, these managementechniques suggested for golf courses can potentially be used inther systems with headwater streams such as parks, cemeteries,istorical sites, and a number of other human land uses that affecthe quality of stream habitat. As the golf industry becomes morepen to land stewardship, sustainability, and ecological awareness,

valuable opportunity is provided for researchers to collaborateith this group of managers and provide constructive ecologically-

ased guidelines for improvement of water quality and wildlifeabitat.

cknowledgements

We thank D. Mackey and D. Riegel for assistance in the field andhe Highlands Biological Station Director, J. Costa, and staff for usef facilities. We appreciate the comments from A. Cox and J. Earl onarly versions of this manuscript. We owe a special thanks to theuperintendents and managers of the 10 participating golf clubshose volunteered support and accommodation made this studyossible. This research was partially supported by a Highlandsiological Station Fellowship to M.J.M and a United States Golfssociation and National Fish and Wildlife Foundation’s Wildlifeinks granting program to R.D.S.

ppendix A.

Location and information for the 10 golf courses in theighlands-Cashiers area of North Carolina used in this study. Allolf courses were located within the Southern Crystalline Ridgesnd Mountains, Level III Blue Ridge Ecoregion of North Carolina.

ountry Club of Sapphire Valley, SapphirePrivate; 53 years old; elevation 966 m; 18 holes – 6859 yards, par 72

ullasaja Club, HighlandsPrivate; 20 years old; elevation 1208 m; 18 holes – 6651 yards, par 72ighlands Country Club, HighlandsPrivate; 81 years old; elevation 1114 m; 18 holes–6255 yards, par 70ighlands Cove Golf Course, Cashiers

Public; 9 years old; elevation 1190 m; 18 holes – 6587 yards, par 72ighlands Falls Country Club, HighlandsPrivate; 47 years old; elevation 1152 m; 18 holes – 6130 yards, par 70ountaintop Golf and Lake Club, CashiersPrivate; 3 years old; elevation 1168 m; 18 holes – 7100 yards, par 70

rillium Links, Cashiers

an Planning 125 (2014) 17–27 25

Private; 11 years old; elevation 1149 m; 18 holes – 6505 yards, par 71Sapphire National Golf Club, Sapphire

Private; 27 years old; elevation 1006 m; 18 holes – 6640 yards, par 72Wade Hampton Golf Club, Cashiers

Private; 22 years old; elevation 1066 m; 18 holes – 7154 yards, par 72Wildcat Cliffs Country Club, Highlands

Private; 48 years old; elevation 1262 m; 18 holes – 6493 yards, par 72

Appendix B.

Description and justification of single parameters used in a prioriregression models predicting stream salamander abundance anddiversity.

Parameter Description Justification

Sediment Percentsedimentation ofstream bottom

Can negatively impactlarval stream salamanders(Peterman & Semlitsch,2009)

Sand Percent sand ofstream bottom

Fewer species are detectedin reaches withhomogenous substrate(Mackey, Connette, &Semlitsch, 2010)

Stream depth Depth of stream May affect water flow rateand microhabitatcharacteristics (Peterman& Semlitsch, 2009)

Buffer Width of unmownor forested bufferbetween streamand managed area

Affects sediment influxesinto streams, watertemperatures,allochthonous inputs, andriparian microclimate(Crawford & Semlitsch,2007; Peterman &Semlitsch, 2009)

Soil temperature Temperature of soil Can influence streamsalamander abundance andmicrohabitat use (Crawford& Semlitsch, 2008)

Leaf depth Depth of leaf litterlocated within 1 mof stream

Primary nutrient source inallochthonous-basedheadwater streams andimportant for salamanderrefugia (Crawford &Semlitsch, 2008)

Soil moist Percent moisture ofsoil located within1 m of stream

Can influence streamsalamander abundance andmicrohabitat use (Crawford& Semlitsch, 2008)

Cover rocks Qualitativeestimate of rocksproviding suitablesalamander refuge

Are used as protectivecover and nesting sites andtherefore are stronglycorrelated withsalamander density (Davic& Orr, 1987)

Coarse woody debris Qualitativeestimate ofin-stream wood

May provide refugia aswell as nutrient source forsalamander prey(Peterman & Semlitsch,2009)

Dissolved oxygen Percent dissolvedoxygen in stream

Has been suggested to be afactor limiting streamsalamander abundances(Willson & Dorcas, 2002)

Appendix C.

Analysis of 16 chemicals by Mississippi State Chemical Labora-

tory that are used in the treatment and management of golf coursesin the Highlands-Cashiers region of North Carolina. An “X” indicatesthe chemical was not detected. Detection proportions are reportedas averages in parts per million.

2 nd Urban Planning 125 (2014) 17–27

oncy

Maximumdetection

Down stream(N = 20)

Detectionfrequency

Maximumdetection

– X – –– X – –– X – –– X – –– X – –– X – –– X – –– X – –– X – –– X – –– X – –0.054 0.0162 9 0.12– X – –– 0.0135 8 0.135– X – –– X – –

R

A

A

A

A

B

B

B

B

B

C

C

C

C

C

C

D

E

F

G

H

H

H

6 M.J. Mackey et al. / Landscape a

Chemical Lower level ofdetection

Upstream(control) (N = 6)

Detectifrequen

Tefluthrin 0.010 X –

cis-Permethrin 0.010 X –

trans-Permethrin 0.010 X –

Cyfluthrin 0.010 X –

Cyhalothrin 0.010 X –

Cypermethrin 0.010 X –

Fenvalerate 0.010 X –

Deltramethrin 0.010 X –

Chlorothalonil 0.010 X –

Fipronil 0.010 X –

Bifenthrin 0.010 X –

4OH-Chlorothalonil 0.010 0.0027 1

Azoxystrobin 0.020 X –

Propiconazole 0.010 X –

Pyraclostrobin 0.020 X –

Trifloxystrobin 0.020 X –

eferences

ldridge, K. T., Brookes, J. D., & Ganf, G. G. (2009). Rehabilitation of stream ecosys-tem functions through the reintroduction of coarse particulate organic matter.Restoration Ecology, 17, 97–106.

llan, J. D. (2004). Landscapes and rivers: The influence of land use on stream ecosys-tems. Annual Review of Ecology, Evolution, and Systematics, 35, 257–284.

nastassiades, M., Lehotay, S. J., Stajnbaher, D., & Schenck, F. J. (2003). Fast and easymultiresidue method employing acetonitrile extraction/partitioning and “dis-persive solid-phase extraction” for the determination of pesticide residues inproduce. Journal of AOAC International, 86, 412–431.

shton, R. E., Jr. (1975). A study of movement, home range, and winter behavior ofDesmognathus fuscus (Rafinesque). Journal of Herpetology, 9, 85–91.

arbour, R. W., Hardin, J. W., Schafer, J. P., & Harvey, M. J. (1969). Home range,movements, and activity of the dusky salamander, 2012. Copeia, 1969, 293–297.

aris, R. D., Cohen, S. Z., Barnes, N. L., Lam, J., & Ma, Q. (2010). Quantitative analysisof over 20 years of golf course monitoring studies. Environmental Toxicology andChemistry, 29, 1224–1236.

ond, N. R., & Lake, P. S. (2003). Local habitat restoration in streams: Constraintson the effectiveness of restoration for stream biota. Ecological Management &Restoration, 4, 193–198.

oone, M. D., Semlitsch, R. D., & Mosby, C. (2008). Suitability of golf course ponds foramphibian metamorphosis when bullfrogs are removed. Conservation Biology,22, 172–179.

urnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference: Apractical information-theoretic approach (second ed.). New York, USA: Springer-Verlag.

ohen, S., Svrjcek, A., Durborow, T., & Barnes, N. L. (1999). Water quality impacts bygolf courses. Journal of Environmental Quality, 28, 798–809.

olding, J., & Folke, C. (2009). The role of golf courses in biodiversity conservationand ecosystem management. Ecosystems, 12, 191–206.

olding, J., Lundberg, J., Lundberg, S., & Andersson, E. (2009). Golf courses and wet-land fauna. Ecological Applications, 19, 1481–1491.

rawford, J. A., & Semlitsch, R. D. (2007). Estimation of core terrestrial habitat fostream-breeding salamanders and delineation of riparian buffers for protectionof biodiversity. Conservation Biology, 21, 152–158.

rawford, J. A., & Semlitsch, R. D. (2008). Abiotic factors influencing abundance andmicrohabitat use of stream salamanders in southern Appalachian forests. ForestEcology and Management, 255, 1841–1847.

ummins, K. W. (1974). Structure and function of stream ecosystems. BioScience, 24,631–641.

avic, R. D., & Orr, L. P. (1987). The relationship between rock density and salamanderdensity in a mountain stream. Herpetologica, 43, 357–361.

isenberg, D. A., Noss, R. F., Waterman, J. M., & Main, M. B. (2011). Distribution andhabitat use of the big cypress fox squirrel (Sciurus niger avicennia). SoutheasternNaturalist, 10, 75–84.

iske, I., & Chandler, R. B. (2011). Unmarked: An R package for fitting hierarchicalmodels of wildlife occurrence and abundance. Journal of Statistical Software, 43,1–23.

ordon, N. P., McMahon, T. A., & Finlayson, B. L. (Eds.). (1992). Stream hydrology: Anintroduction for ecologists. New York: Wiley.

all, J. A., & Fleishman, E. (2010). Demonstration as a means to translate conservationscience into practice. Conservation Biology, 24, 120–127.

arden, L. A., Price, S. J., & Dorcas, M. E. (2009). Terrestrial activity and habitatselection of eastern mud turtles (Kinosternon subrubrum) in a fragmented land-

scape: Implications for habitat management of golf courses and other suburbanenvironments. Copeia, 2009, 78–84.

indahl, M. S., Miltner, E. D., Cook, T. W., & Stahnke, G. W. (2009). Surface water qual-ity impacts from golf course fertilizer and pesticide application. InternationalTurfgrass Society Research Journal, 11, 19–30.

Hodgkison, S., Hero, J.-M., & Warnken, J. (2007). The conservation value of suburbangolf courses in a rapidly urbanising region of Australia.

Jackson, D. B., Kelly, S. D., & Brown, R. D. (2011). Design guidelines for integratingamphibian habitat into golf course landscapes. Landscape and Urban Planning,103, 156–165.

Johnson, B. R., & Wallace, J. B. (2005). Bottom-up limitation of a stream salamanderin a detritus-based food web. Canadian Journal of Fisheries and Aquatic Science,62, 301–311.

Johnson, B. R., Wallace, J. B., Rosemond, A. D., & Cross, W. F. (2006). Larval sala-mander growth responds to enrichment of a nutrient poor headwater stream.Hydrobiologia, 573, 227–232.

Jones, E. B. D., Helfman, G. S., Harper, J. O., & Bolstad, P. V. (1999). Effects of riparianforest removal on fish assemblages in southern Appalachian streams. Conserva-tion Biology, 13, 1454–1465.

Jost, L. (2006). Entropy and diversity. Oikos, 113, 363–375.Kenna, M. P. (1995). What happens to pesticides applied to golf courses? USGA Green

Section Record, 33, 1–9.Kucken, D. J., Davis, J. S., Petranka, J. W., & Smith, C. K. (1994). Anakeesta stream acid-

ification and metal contamination: Effects on a salamander community. Journalof Environmental Quality, 23, 1311–1317.

Likens, G. E., & Bormann, F. H. (1974). Linkages between terrestrial and aquaticecosystems. BioScience, 24, 447–456.

Lowe, W. H., Nislow, K. H., & Bolger, D. T. (2004). Stage-specific and interactiveeffects of sedimentation and trout on a headwater stream salamander. EcologicalApplications, 14, 164–172.

Lowe, W. H., & Likens, G. E. (2005). Moving headwater streams to the head of theclass. Bioscience, 55, 196–197.

Lowrance, R., Todd, R., Fail, J., Jr., Hendrickson, O., Jr., Leonard, R., & Asmussen, L.(1984). Riparian forests as nutrient filters in agricultural watersheds. Bioscience,34, 374–377.

Mackey, M. J., Connette, G. M., & Semlitsch, R. D. (2010). Monitoring of streamsalamanders: The utility of two survey techniques and the influence of streamsubstrate complexity. Herpetological Review, 41, 163–166.

Mallin, M. A., & Wheeler, T. L. (2000). Nutrient and fecal coliform discharge fromcoastal north carolina golf courses. Journal of Environmental Quality, 29, 979–986.

Mazerolle, M. J. (2013). AICcmodavg: Model selection and multimodel inference basedon (Q)AIC(c). R package version 1.33.

Merola-Zwartjed, M., & DeLong, J. P. (2005). Southwestern golf courses provideneeded riparian habitat for birds. USGA Turfgrass and Environmental ResearchOnline, 4, 1–18.

Meyer, J. L., Paul, M. J., & Taulbee, W. K. (2005). Stream ecosystem function inurbanizing landscapes. Journal of the North American Benthological Society, 24,602–612.

Mifsud, D. A., & Mifsud, R. (2008). Golf courses as refugia for herpetofauna in an urbanriver floodplain. In J. C. Mitchell, R. E. Jung Brown, & B. Bartholomew (Eds.), Urbanherpetology (pp. 303–310). Salt Lake City: Society for the Study of Reptiles andAmphibians.

Montgomery, D. R., & MacDonald, L. H. (2002). Diagnostic approach to stream chan-nel assessment and monitoring. JAWRA Journal of the American Water ResourcesAssociation, 38, 1–16.

NGF. (2010). Summary of golf participation in the U.S. National Golf Foundation.Nowakowski, A. J., & Maerz, J. C. (2009). Estimation of larval stream salamander den-

sities in three proximate streams in the Georgia piedmont. Journal of Herpetology,43, 503–509.

Orser, P. N., & Shure, D. J. (1972). Effects of urbanization on the salamander Desmog-nathus fuscus fuscus. Ecology, 53, 1148–1154.

Pauley, T. K., & Little, M. (1998). A new technique to monitor larval and juvenile

salamanders in stream habitats. Banisteria, 12, 32–36.

Peterman, W. E. (2008). Effects of riparian buffer width on stream salamander popula-tions in the southern Appalachain Mountains (MA thesis). Columbia: Division ofBiological Sciences, University of Missouri.

nd Urb

P

P

P

P

P

P

R

R

R

R

R

M.J. Mackey et al. / Landscape a

eterman, W. E., Crawford, J. A., & Semlitsch, R. D. (2008). Productivity and sig-nificance of headwater streams: Population structure and biomass of theblack-bellied salamander (Desmognathus quadramaculatus). Freshwater Biology,53, 347–357.

eterman, W. E., & Semlitsch, R. D. (2009). Efficacy of riparian buffers in mitigatinglocal population declines and the effects of even-aged timber harvest on larvalsalamanders. Forest Ecology and Management, 257, 8–14.

etranka, J. W. (1998). Salamanders of the United States and Canada. Washington, DC:Smithsonian Institution Press.

etranka, J. W., & Smith, C. K. (2005). A functional analysis of streamside habi-tat use by southern Appalachian salamanders: Implications for riparian forestmanagement. Forest Ecology and Management, 210, 443–454.

orter, E. E., Pennington, D. N., Bulluck, J., & Blair, R. B. (2004). Assessing the con-servation value of golf courses for butterflies. USGA Turfgrass and EnvironmentalResearch Online, 3, 1–13.

rice, S. J., Cecala, K. K., Browne, R. A., & Dorcas, M. E. (2010). Effects ofurbanization on occupancy of stream salamanders. Conservation Biology, 3,547–555.

Development Core Team. (2012). R: A language and environment for statisticalcomputing. Vienna, Austria: R Foundation for Statistical Computing.

odewald, A. D., Rodewald, P. G., & Santiago, M. J. (2004). Conservation of red-headedwoodpeckers on midwestern golf courses. USGA Turfgrass and EnvironmentalResearch Online, 3.

ouse, J. D., Bishop, C. A., & Struger, J. (1999). Nitrogen pollution: An assess-ment of its threat to amphibian survival. Environmental Health Perspectives, 107,799–803.

oyle, J. A. (2004). N-Mixture models for estimating population size from spatiallyreplicated counts. Biometrics, 60, 108–115.

yals, S. C., Genter, M. B., & Leidy, R. B. (1998). Assessment of surface water qual-ity on three eastern North Carolina golf courses. Environmental Toxicology andChemistry, 17, 1934–1942.

an Planning 125 (2014) 17–27 27

Sheil, D., & Meijaard, E. (2010). Purity and prejudice: Deluding ourselves aboutbiodiversity conservation. Biotropica, 42, 566–568.

Southerland, M. T., Jung, R. E., Baxter, D. P., Chellman, I. C., Mercurio, G., & Volstad, J.H. (2004). Stream salamanders as indicators of stream quality in Maryland, USA.Applied Herpetology, 2, 23–46.

Tanner, R. A., & Gange, A. C. (2005). Effects of golf courses on local biodiversity.Landscape and Urban Planning, 71, 137–146.

Vannote, R. L., Minshall, G. W., Cummins, K. W., Sedell, J. R., & Cushing, C. E. (1980).The river continuum concept. Canadian Journal of Fisheries and Aquatic Science,37, 103–137.

Verheyen, K., Honnay, O., Motzkin, G., Hermy, M., & Foster, D. R. (2003). Responseof forest plant species to land-use change: A life-history trait-based approach.Journal of Ecology, 91, 563–577.

Welsh, H. H., Jr., & Ollivier, L. M. (1998). Stream amphibians of ecosystem stress: Acase study from California’s redwoods. Ecological Applications, 8, 1118–1132.

Vesely, D. G., & McComb, W. C. (2002). Salamander abundance and amphibianspecies richness in riparian buffer strips in the Oregon coast range. Forest Science,48, 291–297.

Wheeler, K., & Nauright, J. (2006). A global perspective on the environmental impactof golf. Sport in Society: Cultures, Commerce, Media, Politics, 9, 427–443.

Wilcove, D. S., Rothstein, D., Dubow, J., Phillips, A., & Losos, E. (1998). Quantifyingthreats to imperiled species in the United States. Bioscience, 48, 607–615.

Willson, J. D., & Dorcas, M. E. (2002). Effects of habitat disturbance on streamsalamanders: Implications for buffer zones and watershed management. Con-servation Biology, 17, 763–771.

Winter, J. G., & Dillon, P. J. (2005). Effects of golf course construction and oper-

ation on water chemistry of headwater streams on the precambrian shield.Environmental Pollution, 133, 243–253.

Winter, J. G., Dillon, P. J., Paterson, C., Reid, R. A., & Somers, K. M. (2003). Impacts ofgolf course construction and operation on headwater streams: Bioassessmentusing benthic algae. Canadian Journal of Botany, 81, 848–858.