A Comparison of Ground-Based and Tree-Based Polyvinyl Chloride Pipe Refugia for Capturing Pseudacris Regilla in Northwestern California

NORTHWESTERN NATURALIST 88:147–154 WINTER 2007

A COMPARISON OF GROUND-BASED AND TREE-BASED POLYVINYL CHLORIDE PIPE REFUGIA FOR CAPTURING PSEUDACRIS REGILLA IN NORTHWESTERN CALIFORNIA

CALE HMYERS,LISA EIGNER,JAKE AHARRIS,ROBERT HILMAN,MATTHEW DJOHNSON, RYAN KALINOWSKI,JORDAN JMUIR,MELISSA REYES, AND LESLIE ETUCCI Humboldt State University, Department of Wildlife, Arcata, California, USA 95521

ABSTRACT—Creating artificial refugia through the use of polyvinyl chloride (PVC) pipes to capture hylid treefrogs has been examined in the Eastern United States and Caribbean, but has not been evaluated in the Pacific Northwest. We compared the effectiveness of ground-based and tree-based PVC pipes for capturing the Pacific Treefrog (Pseudacris regilla) in northwestern California. We recorded a total of 464 P. regilla captures. Probability of use increased asymptotically over our 28-d sampling period and was higher (6–20%) for tree-based than ground- based pipe refugia. Tree-based pipe refugia caught 81% more P.regilla than ground-based pipe refugia but latency to first detection did not vary significantly. Probability of use was higher for pipe refugia that had been used in a previous study than for new pipes, but age did not influence capture rate or latency to detection. Our results support the hypothesis that tree-based pipe refugia are more effective than ground-based pipes for capturing P. regilla, as well as the hypothesis that P. regilla actively discriminate among refugia. Using tree-based PVC pipe refugia to monitor P.regilla may be more effective than current techniques for addressing many research questions.

Key words: Paciﬁc Treefrog, Hyla regilla, Pseudacris regilla, polyvinyl chloride, pipe refugia, northwestern California

Traditional amphibian sampling techniques, accurately monitor P. regilla populations, spe- such as the use of pitfall traps, drift fences, and cific and efficient capture techniques are need- visual encounter surveys are ineffective for ad- ed (Johnson 2005). dressing many research questions for hylid Constructing artificial refugia by means of treefrogs due to their ability to evade pitfall polyvinyl chloride (PVC) pipes has been estab- traps and drift fences, and observer biases as- lished as an effective monitoring technique for sociated with visual encounter surveys (Dodd hylids in the eastern United States and Carib- 1991; Moulton and others 1996; Lamb and oth- bean (McComb and Noble 1981; Stewart and ers 1998; Boughton and others 2000). Despite Pough 1983; Meshaka 1996; Moulton and oth- these findings, traditional techniques are still ers 1996; Boughton and others 2000), but to our employed to assess trends of the Pacific Tree- knowledge there have been no studies investi- frog (Pseudacris regilla, also referred to as Hyla gating PVC pipe refugium design for captur- regilla; family Hylidae), the most abundant am- ing hylids in the Pacific Northwest or for P. r e - phibian in western North America (Matthews gilla in particular. Our objective was to monitor and others 2001). and assess how ground-based and tree-based Urbanization, habitat destruction, and the in- positioning of PVC pipe refugia influence cap- troduction of exotic trout have reduced abun- ture success for P. regilla. dances and distributions of P. regilla in many Previous studies examining pipe refugia for parts of its range (Bull 2002; Pearl and others other hylids have demonstrated that pipe de- 2003; Knapp 2005; Pearl and others 2005; Reid sign and positioning can influence capture suc- 2005; Riley and others 2005). In addition, P. r e - cess (Townsend 1989; Moulton and others 1996; gilla populations show great natural fluctua- Boughton and others 2000; Zacharow and oth- tion, making it difficult to assess trends of the ers 2003; Bartareau 2004). Pacific Treefrogs species (Matthews and others 2001). In order to seek refuge beneath ground cover and within

147 148 NORTHWESTERN NATURALIST 88(3) trees (Nussbaum and others 1983), but it is un- were white, schedule 40-weight, 60 cm in known if tree-based PVC pipe refugia are more length, 5.08 cm in diameter, open on the top effective than ground-based pipe refugia for and capped on the bottom. To ensure a moist, capturing the species. Hylids have shown a humid environment within the confines of the preference for elevated retreat sites in other pipe, we drilled a drain-hole 15 cm from the studies examining pipe refugium design bottom of the pipe and filled it to the hole with (Townsend 1989; Greenberg and others 1996; water. To hang pipe refugia on screws, we Boughton and others 2000), and we hypothe- drilled another hole approximately 10 cm from sized that tree-based PVC pipe refugia would the top of the pipe. We mounted tree-based be more effective than ground-based PVC pipe pipe refugia vertically on tree trunks at a refugia for capturing P. regilla. We evaluated height of 2.5 m above the ground while ground- this hypothesis by testing the following predic- based pipe refugia were mounted vertically on tions: relative to ground-based pipe refugia, wooden stakes (1.27 ϫ 5.08 ϫ 122 cm) so that tree-based pipe refugia should have (1) a high- the bases of the pipes were resting on the er probability of use, (2) a higher capture rate, ground. All pipe refugia faced north to control and (3) a shorter latency to detection for P. r e - for potential differences stemming from ori- gilla. entation. We positioned tree-based pipe refugia on alder trees at approximately 5-m intervals, METHODS while ground-based pipe refugia were also po- We conducted our study on the Humboldt sitioned at 5-m intervals, 1 m from each tree- Bay National Wildlife Refuge, located on the based pipe refugium and directly in line with coast of northwestern California, approximate- the transect. We obtained 100 PVC pipes from ly 150 km south of the Oregon border. The ref- a previous study (summer 2006) and purchased uge has a variety of habitats including marshes, 100 additional pipes for our study (hereafter, seasonally flooded wetlands, riparian wet- ‘‘old’’ and ‘‘new’’ pipe refugia). We distributed lands, streams, and forests (Kwansy 2000), and old and new pipe refugia evenly among tree P. regilla are known to inhabit these types of and ground sites. During the course of our habitats at elevations ranging from sea level to study, we noticed that 10 tree-based pipe re- 2440 m above sea level (Corkran and Thoms fugia were placed onto trees used in this pre- 1996). Our study site consisted of a forested ri- vious study. To reduce possible biases associ- parian area with a flowing water source (Salm- ated with these locations, we removed these on Creek). Red Alder (Alnus rubra) and willows pipe refugia from our analyses. (Salix spp.) dominated the canopy layer while Pipe refugia were deployed on 30 September Salmonberry (Rubus spectabilis), Twinberry 2006 and we inspected pipe refugia on every (Lonicera involucrata), and California Blackberry (Rubus ursinus) comprised the under story veg- 3rd day from 2–29 October 2006, resulting in a etation. Pseudacris regilla have been observed in total of 10 surveys. To avoid predation and des- the area year-round, as the area supports a iccation, hylids seek out retreats during the day large amount of invertebrate prey (Verhey (Boughton and others 2000); therefore, we con- 1992) and wet areas for breeding (Stebbins ducted surveys between 09:00–16:00 PST. With 2003). The climate throughout the region is rel- the aid of flashlights, survey teams (consisting atively cool and moist, with annual rainfall av- of 2 field technicians at each transect) tallied the eraging 99 cm and mean annual temperatures number of frogs in each pipe. Due to budget ranging from 9.5 to 15Њ C (NOAA 2006). constraints, we were unable to mark any P. r e - gilla in this study. We designed our methods to Sampling Scheme be minimally intrusive so frogs were not sexed, We utilized a systematic sampling scheme aged, weighed, or removed from pipe refugia consisting of 2 separate 250-m transects. Each at any time during this study. We removed de- transect consisted of 50 ground-based and 50 bris only if it interfered with our ability to ac- tree-based pipe refugia. Pipe refugia were de- curately observe or count the number of frogs signed following previously successful proto- in each pipe refugium. We refilled each pipe to cols for other species of treefrogs (Boughton the 15 cm mark with water during each survey and others 2000; Staiger and others 2002): pipes when necessary. WINTER 2007 MYERS AND OTHERS: PVC PIPE REFUGIA 149

Data Analyses tory. We considered pipe refugium positioning (tree or ground) and pipe refugium age (old or We measured 3 response variables to deter- new) as factors potentially influencing␺ˆ and pˆ. mine if the positioning (ground vs. tree) and In addition, we modeled pˆ as a function of time age (old vs. new) of pipe refugia influenced (survey number 1–10) by examining a linear their effectiveness: probability of use, capture trend in pˆ over survey number, an asymptotic rate, and latency to initial detection (LTD; increase in pˆ over survey number, a unique val- Foresman and Pearson 1998). Our unit of rep- ue for pˆ at each survey, or a unique value for pˆ lication was a pipe (n ϭ 190). Because capture for surveys 1–5 and 6–10. We used Akaike’s In- rates can be difficult to interpret if the inde- formation Criterion (AIC) for model selection pendence of captures is unknown, probability and considered models with ⌬AIC values of Ͻ3 of use and LTD are useful in comparing pipe to be competitive models (Burnham and An- refugium efficacy (Gompper and others 2006). derson 2002). We presented estimates of occu- Probability of Use.—We used the occupancy pancy and pipe refugium use probabilities as x¯ modeling software program PRESENCE 2.0 Ϯ s by model-averaging the top models (⌬AIC (MacKenzie and others 2006) to determine the x¯ Ͻ3). probability of pipe refugium use. The data Capture Rate and Latency to Detection.—We de- were simplified to reflect frog presence or ab- fined capture rate as the mean number of frogs sence in a pipe for each survey, regardless of captured/pipe refugium over the 10 surveys. the number of frogs present. Occupancy mod- Latency to initial detection was defined as the eling of animal ‘‘presence-absence’’ data can be amount of time that lapsed until initial detec- conceptualized by considering 3 probabilities: tion of a frog in pipe refugia. To calculate LTD, (1) the probability a general site is occupied by we assigned a numerical value (ranging from 1 an animal; (2) the conditional probability,given to 10) to each pipe corresponding with the the animal occupies a site, that the animal used number of surveys made until initial detection the precise surveyed location when it was sur- (assigning no numerical value to pipe refugia veyed; and (3) the conditional probability, giv- where no detection was made). We used 2-way en that the animal used the surveyed location, ANOVAs to test for significant differences (␣ϭ that the animal was detected by the researcher 0.05) in capture rate and LTD between ground- (MacKenzie and others 2006). The 1st probabil- based vs. tree-based pipe refugia and old vs. ␺ ity, called occupancy probability ( ), is often new pipe refugia. Capture rate was trans- the premier parameter of interest in under- formed (log ϩ 1) for analysis; raw values are standing the spatial distribution of a species. reported for graphical purposes. The product of the 2nd and 3rd probabilities is called detection probability (p). PRESENCE al- RESULTS ␺ lows the simultaneous estimation of both and We recorded a total of 464 P. regilla captures p via a maximum-likelihood information-the- in 80 PVC pipe refugia from 2–29 October 2006, oretic model selection procedure (MacKenzie with an average of 37 pipe refugia occupied/ and others. 2006). survey. Of the 464 P. regilla captures, 299 cap- With our methods, the probability that a frog tures (64%) occurred in tree-based pipe refu- was overlooked inside a pipe was essentially gia. The total number of P.regilla counted/sur- zero. Therefore, pˆ is an estimate of the proba- vey ranged from 22–113 frogs. bility that a frog used a pipe given that a frog occupied the general site where the pipe oc- Probability of Use curred (hereafter ‘‘the probability of use’’). The best of the candidate models indicated This variable thus eliminated potentially con- that the probability of use, p, varied with pipe founding effects of variation in occupancy in refugium positioning and age, and it increased evaluating the effectiveness of a capture tech- asymptotically over the 10 surveys (Table 1). nique and represents an improvement over The top 4 models (⌬AIC Ͻ3) contained this pa- more conventional comparisons of capture suc- rameterization for p, and in each case the model cess. We developed a set of a priori candidate coefficients (␤) did not overlap zero, indicating models for pˆ and␺ˆ based on our hypotheses, strong support for the effect of these parame- experience, and knowledge of frog natural his- ters on p. Model averaging indicated that the 150 NORTHWESTERN NATURALIST 88(3)

TABLE 1. Model selection results for PRESENCE models of Pseudacris regilla occupancy (␺) and probability of use (p) at the Humbolt Bay National Wildlife Refuge, California, 2–29 October 2006; Akaike’s information criterion values (AIC), change in AIC from top model (⌬AIC), number of estimable parameters (K), and AIC ␺ ϭ Weights (wi). Subscripts give parameterization for and p:‘.’ constant over group and time variables; ‘placement’ ϭ tree- or ground-based pipe refugia; ‘age’ ϭ new pipe refugia or pipe refugia used in a previous study; ‘time࿞linear’ ϭ p increased linearly over 10 surveys; ‘time࿞asymptote’ ϭ p increased asymptotically over the surveys. Subscripts joined by a ‘ϩ’ or an ‘*’ indicate additive and factorial models, respectively. Ͻ Only the top 6 models are shown, all others had wi 0.01. ⌬ Model parameterization AIC AIC K wi ␺ (placement), p(placementϩageϩtime࿞asymptote) 1252.6 0.00 6 0.413 ␺ (.), p(placementϩageϩtime࿞asymptote) 1253.6 0.98 5 0.253 ␺ (placementϩage), p(placementϩageϩtime࿞asymptote) 1254.6 1.94 7 0.157 ␺ (age), p(placementϩageϩtime࿞asymptote) 1255.6 2.93 6 0.095 ␺ (placement*age), p(placementϩageϩtime࿞asymptote) 1256.5 3.88 8 0.059 ␺ (placementϩage), p(placementϩageϩtime࿞linear) 1259.8 7.14 7 0.017

probability of use was 6–20% higher for old asymptotically with increases in pˆ as much as pipe refugia placed on trees than for the other 10% between consecutive surveys early in our pipe refugium categories (Fig. 1). Old pipe re- study, and smaller increases (2–3%) between fugia placed on the ground and new pipe re- the later surveys. Model averaging indicates fugia placed on trees had similar probabilities that pˆ was peaking by the end of our study, as of use, while new pipe refugia placed on the models containing an asymptotic increase were ground had the lowest probability of use (Fig. over 57 times more likely to ﬁt the data than 1). For all pipe refugium categories, pˆ increased those with a linear increase. The best models

FIGURE 1. Estimated probability of use by Pseudacris regilla over 10 survey periods at the Humboldt Bay National Wildlife Refuge, California, 2–29 October 2006. Use probabilities are shown for tree- versus ground- based pipe refugia and for pipe refugia used in a previous study (old) versus new pipe refugia. Values are based on model-averaging (see Table 1) and are offset slightly on the x-axis to show error bars, which are Ϯ 1.sx¯ WINTER 2007 MYERS AND OTHERS: PVC PIPE REFUGIA 151 indicated that occupancy probability (␺) also others (2002) found ground-based pipe refugia varied by location (tree or ground), with 2 of more effective than tree-based pipe refugia for the top 3 models containing this parameter. capturing the Pine Woods Treefrog (Hyla fe- Pipe refugium age was included in the 3rd and moralis), the Green Treefrog (H. cinerea), and 4th ranked models, but its coefficient over- the Squirrel Treefrog (H. squirella). No single lapped zero, lending little support for its influ- pipe diameter, pipe length, or elevation above ␺ ␺ ϭ ence on . Model-averagedˆ was 0.46, sx¯ 0.12 ground can be considered the most effective ϭ and 0.38, sx¯ 0.05 for tree and ground loca- technique and prior to conducting detailed tions, respectively. studies, preliminary studies are needed to determine the optimal pipe refugium design for Capture Rate and Latency to Detection the hylid species under investigation (Martin Tree-based pipe refugia caught 81% more P. and others 2004; Johnson 2005). Though we regilla than ground-based pipe refugia (F1,186 used pipe diameters shown effective elsewhere ϭ 5.95, P ϭ 0.016). Pipe refugium age (old vs. (5.08 cm), it remains unknown if this is optimal ϭ new) did not influence capture rate (F1,186 for P. regilla. 0.57, P ϭ 0.45), nor was there an interaction be- Although not significantly different, laten- ϭ ϭ tween positioning and age (F1,186 0.10, P cies to detection were approximately 5.2 and 0.75). Average latency to detection was 5.17 and 4.3 surveys (16 and 13 d) for ground-based and 4.33 surveys for ground-based and tree-based tree-based pipe refugia, respectively. A possi- pipe refugia respectively, although this differ- ble explanation for this pattern is that hylids ϭ ence was not statistically significant (F1,75 have to search to find refuge after the initial 1.98, P ϭ 0.16). Pipe refugium age also did not placement of pipe refugia into a habitat (Staig- ϭ ϭ influence latency to detection (F1,75 0.39, P er and Boughton 1999; Zacharow and others 0.53), nor was there an interaction between po- 2003). ϭ ϭ sitioning and age (F1,75 0.02, P 0.89). Probability of use by P. regilla increased asymptotically with each successive survey (Fig. DISCUSSION 1). Understanding this pattern is important for Probability of use and capture rates for P. r e - managers using PVC pipe refugia to capture P. gilla were higher in tree-based than in ground- regilla, as they should be aware that pipe refu- based PVC pipe refugia, supporting the hy- gium use will likely increase as a function of pothesis that tree-based pipe refugia are more time following initial placement. In our study, effective for capturing P.regilla. These findings new tree-based pipe refugia did not reach a are consistent with studies from other regions probability of use of 50% until the 8th survey conducted on other hylids (Townsend 1989; (24 d). For managers to discern how long pipe Greenberg and others 1996; Boughton and oth- refugia should be deployed, it is necessary to ers 2000; Bartareau 2004) and support the hy- conduct studies to determine an asymptote for pothesis that hylids actively discriminate total P.regilla captures as a function of time fol- among artificial refuge sites (McComb and No- lowing initial placement. In our system, that ble 1981; Boughton and others 2000; Bartareau time was about a month, but may fluctuate if a 2004). study is conducted over a longer period of time, It remains unknown why some hylids prefer during a different season, or in a different hab- elevated retreat sites. Boughton and others itat. (2000) suggested that elevated retreat sites offer Pipe refugium age did not influence capture favorable microclimates. Although our study rate or latency to initial detection, but old pipe was conducted during the non-breeding sea- refugia had somewhat lower probabilities of son, other studies (Greer and Wells 1980; Na- use than did pipe refugia constructed from rins and Hurley 1982; Mitchell 1991) have hy- newly purchased PVC pipe. Though all pipe re- pothesized that elevated sites offer better fugia were cleaned before being deployed and acoustical properties than ground sites for call- the purchased PVC was housed outdoors at the ing male hylids. supplier (and thus somewhat weathered), it is No single pipe refugium design is appropri- possible that the 2 groups differed in odors or ate for all hylid species in all locations (Martin off-gassing. Researchers considering PVC pipe and others 2004; Johnson 2005). Staiger and refugia may seek to standardize odors across 152 NORTHWESTERN NATURALIST 88(3) all pipe refugia in a study, either by making all Johnson 2005). Nevertheless, our study sup- purchases simultaneously, or ‘‘weathering’’ ports the use of tree-based PVC pipe refugia new pipe refugia in the field before deploy- over ground-based PVC pipe refugia for cap- ment. turing P. regilla. Other studies (Moulton and Several confounding factors may have influ- others 1996; Lamb and others 1998; Boughton enced our results. First, we tried to eliminate and others 2000) have suggested that sampling biases associated with a preliminary experi- techniques currently utilized to monitor P. r e - ment in the same area, but P. regilla may have gilla are ineffective for addressing many re- grown accustomed to occupying pipe refugia search questions for the species. Tree-based during the preliminary work, and future stud- PVC pipe refugia may offer a valuable tool for ies involving naive frogs may not achieve cap- monitoring P. regilla in the Pacific Northwest. ture rates as high as ours. Second, our study was only conducted over a 4-wk period in mid- ACKNOWLEDGMENTS fall, and longer studies have demonstrated a re- We thank members of Humboldt State University’s duction in the effectiveness of tree-based pipe Fall 2006 Wildlife Techniques Class for contributions refugia during the hylid breeding season to our study design and collection of data. We rec- (Staiger and Boughton 1999). During the P. r e - ognize A Desch for assistance in pipe refugium de- gilla breeding season, frogs are known to in- sign. The HSU Wildlife and Forestry Stockroom pro- habit permanent and semi-permanent breed- vided necessary equipment. We thank the Humboldt ing ponds that can be found in open areas (Cal- Bay National Wildlife Refuge, operated by the United hoon and Jameson 1970). We did not measure States Fish and Wildlife Service, for granting per- mission to conduct this study on their property. T the effectiveness of pipe refugia under these Bartareau and an anonymous reviewer provided conditions. Future studies should be designed helpful comments on an earlier version of this man- to place refugia out over the breeding season as uscript. Our research was approved by the Hum- well as the non-breeding season. Third, budget boldt State University Animal Care and Use Com- constraints prevented us from conducting a mittee (Permit No. 05/06.W.06.E). mark and recapture analysis to differentiate among individual P. regilla, and we may have LITERATURE CITED recaptured many of the same frogs. Boughton BARTAREAU T. 2004. PVC pipe diameter influences and others (2000) demonstrated high levels of the species and sizes of treefrogs captured in a retreat fidelity in a mark and recapture analysis Florida coastal oak scrub community. Herpeto- among 6 different hylid species. Zacharow and logical Review 35:150–152. others (2003) recognized that hylids take up BOUGHTON RG, STAIGER J, F RANZ R. 2000. Use of PVC residency in the same PVC pipe for various pipe refugia as a sampling technique for hylid amounts of time. Future studies should inves- treefrogs. The American Midland Naturalist 144: tigate retreat site fidelity for P. regilla by incor- 168–177. porating alpha numeric tag marking tech- BUCHAN A, SUN L, WAGNER RS. 2005. Using alpha numeric fluorescent tags for individual identifi- niques (Buchan and others 2005). In addition, cation of amphibians. Herpetological Review 36: future studies should analyze P. regilla demo- 43–44. graphics such as sex, age, body size, and color BULL EL. 2002. Influence of fish and habitat on am- with respect to pipe refugium design. It is pos- phibian communities in high elevation lakes in sible that certain pipe refugium designs may be northeastern Oregon. Northwest Science 76:240– more effective for targeting a certain demo- 248. graphic group of P. regilla BURNHAM KP, ANDERSON DR. 2002. Model selection To our knowledge, this is the 1st study to ex- and multimodel inference: a practical informa- amine the effectiveness of PVC pipe refugium tion-theoretic approach. 2nd Edition. New York, design for capturing P. regilla and investigate NY: Springer-Verlag. 488 p. CALHOON RE, JAMESON DL. 1970. Canonical corre- PVC pipe refugium design for capturing hylids lation between variation in weather and variation in the Pacific Northwest. Additional studies ex- in size in the Pacific tree frog, Hyla regilla in amining other aspects of pipe refugium design, Southern California. Copeia 1970:124–134. such as pipe diameter, pipe length, and pipe CORKRAN CC, THOMS C. 1996. Amphibians of height above ground for capturing P. regilla Oregon, Washington, and British Columbia. Red- would be beneficial (Martin and others 2004; mond, WA: Lone Pine Press. 175 p. WINTER 2007 MYERS AND OTHERS: PVC PIPE REFUGIA 153

DODD CK JR. 1991. Drift fence-associated sampling ing characteristics in a southern Mississippi cho- bias of amphibians at a Florida sandhill tempo- rus of Hyla cinera. Copeia 2:521–524. rary pond. Journal of Herpetology 25:296–301. MOULTON CA, FLEMING JW, NERNEY BR. 1996. The FORESMAN KR, PEARSON DE. 1998. Comparison of use of PVC pipes to capture hylid frogs. Herpe- proposed survey procedures for detection of for- tological Review 27:186–187. est carnivores. Journal of Wildlife Management. NARINS PM, HURLEY DD. 1982. The relationship be- 62:1217–1226. tween call intensity and function in the Puerto Ri- GOMPPER ME, HACKETT HM, JACHOWSKI DS, MO- can coqui (Anura: Leptodactylidae). Herpetolog- NELLO RJ, OLIFIERS N. 2006. Large carnivores and ical 38:287–295 the conservation of biodiversity. Animal Conser- [NOAA] NATIONAL ATMOSPHERIC AND OCEANIC vation 9:355. ADMINISTRATION. 2006. National Climatic Data GREENBERG CH, NEARY DG, HARRIS LD. 1996. A Center. http://www.ncdc.noaa.gov/oa/climate/ comparison of herpetofaunal sampling effective- regionalclimatecenters.html#WRCC. Accessed 21 ness of pitfall, single-ended, and double-ended December 2006. funnel traps used with drift fences. Journal of NUSSBAUM RA, BRODIE ED, STORM RM. 1983. Am- Herpetology 28:319–324. phibians and reptiles of the Pacific Northwest. GREER J, W ELLS KD. 1980. Territorial and reproduc- Moscow, ID: University Press of Idaho. p 137–138. tive behavior of the tropical American frog, Cen- PEARL CA, ADAMS MJ, SCHUTYTEMA GS, NEBEKER AV. trolenella fleischmani. Hereptologica 36:318–326. 2003. Behavioral response of anuran larvae to JOHNSON JR. 2005. A novel arboreal pipe-trap de- chemical cues of native and introduced predators signed to capture the Gray Treefrog (Hyla versi- in the Pacific northwestern United States. Journal color). Herpetological Review 36:274–277. of Herpetology 37:572–576. KWANSY, DC 2000. Reproductive response of Savan- PEARL CA, ADAMS MJ, NIELS L, BURY RB. 2005. Am- nah Sparrows (Passerculus sandwichensis) to live- phibian occurrence and aquatic invaders in a stock grazing at Humboldt Bay National Wildlife changing landscape: implications for wetland Refuge [thesis]. Arcata, CA: Humboldt State Uni- mitigation in the Willamette Valley, Oregon, USA. versity. 32 p. Wetlands 25:76–88. KNAPP RA. 2005. Effects of nonnative fish and habi- REID IS. 2005. Amphibian, fish stocking, and habitat tat characteristics on lentic herpetofauna in Yo- relationships in Siskiyou Mountain Wilderness semite National Park, USA. Biological Conserva- Lakes, California and Oregon. Northwestern Nat- tion 121:265–279. uralist 86:25–33. LAMB T, G AUL RW JR,TRIPP ML, HORTON JM, GRANT RILEY S, BUSTEED GT, KATS LB, VANDERGON TL, LEE BW. 1998. A herpetofaunal inventory of the lower ES, DAGIT RG, KERBY JL, FISHER RN, SAUVAJOT Roanoke River floodplain. Journal of the Elisha RM. 2005. Effects of urbanization on the distri- Mitchell Scientific Society 114:43–55. bution and abundance of amphibians and inva- MACKENZIE DI, NICHOLS JD, ROYLE JA, P OLLOCK K, sive species in southern California streams. Con- BAILEY L, HINES JE. 2006. Occupancy Estimation servation Biology 19:1894–1907. and Modeling—Inferring Patterns and Dynamics STAIGER JS, BOUGHTON RG. 1999. Factors affecting of Species Occurrence. Amsterdam, Elsevier Pub- hylid treefrog use of PVC pipe refugia in north cen- lishing. 324 p. tral Florida. Gainesville, FL: United States Geo- MARTIN D, WIKE LD, PALLER MH. 2004. PVC pipe logical Service, Center for Aquatic Resource Studies. samplers for hylid frogs: a cautionary note. Her- http://cars.er.usgs.gov/posters/Herpetology/ petological Natural History. US Department of PVC࿞Piperefugia/pvc࿞piperefugia.html. Accessed Energy Environmental Services and Technology 22 September 2006. Department. General Technical Report DE-AC09- STAIGER JS, SMITH LL, BARICHIVICH WJ, DODD CK JR. 96SR18500. 6 p. 2002. Using artificial pipe refugia to sample Hylid MATTHEWS KR, POPE KL, HAIGANOUSH K, PREISLER frogs in Okefenokee National Wildlife Refuge. K, KNAPP RA. 2001. Effects of nonnative trout on Gainesville, FL: United States Geological Service, Pacific Treefrogs (Hyla regilla) in the Sierra Ne- Center for Aquatic Resource Studies. http://cars. vada. Copeia 4:1130–1137. er.usgs.gov/posters/Herpetology/Artificial࿞ MCCOMB WC, NOBLE RE. 1981. Herpetofaunal use of Piperefugia/artificial࿞piperefugia.html. Accessed natural tree cavities and nest boxes. Wildlife So- 22 September 2006. ciety Bulletin 9:261–267. STEWART MM, POUGH FH. 1983. Population density MESHAKA WE JR. 1996. Vagility and the Florida dis- of tropical forest frogs: relation to retreat sites. tribution of the Cuban Treefrog (Osteopilus sep- Science 221:570–572. tentrionalis). Herpetological Review 27:37–40. TOWNSEND DS. 1989. The consequences of microhab- MITCHELL SL. 1991. Intermale spacing and site call- itat choice for male reproductive success in a 154 NORTHWESTERN NATURALIST 88(3)

tropical frog (Eleutherodactylus coqui). Herpetolo- Using ground-placed PVC pipes to monitor hylid gica 45:451–458. treefrogs: capture biases. Southeastern Naturalist VERHEY CL. 1992. Bird and invertebrate communities 2:575–590. in grazed and ungrazed ﬁelds at Humboldt Bay National Wildlife Refuge [thesis]. Arcata, CA: Humboldt State University. 69 p. Submitted 30 March 2007, accepted 12 July 2007. ZACHAROW M, BARICHIVICH WJ, DODD CK JR. 2003. Corresponding Editor: Ryan O’Donnell.