Forest Ecology and Management 368 (2016) 45–54

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Anuran occupancy and breeding site use of aquatic systems in a managed pine landscape Bethany A. Johnson a,⇑, Kyle Barrett a, Jessica A. Homyack b, Robert F. Baldwin a a b

Department of Forestry and Environmental Conservation, Clemson University, 132 Lehotsky Hall, Clemson, SC 19634, USA Weyerhaeuser Company, 1785B Weyerhaeuser Road, Vanceboro, NC 28562, USA

a r t i c l e

i n f o

Article history: Received 24 November 2015 Received in revised form 1 March 2016 Accepted 1 March 2016

Keywords: Amphibian Anuran Aquatic system Breeding Forest management Occupancy

a b s t r a c t Although the southeastern United States has >12 million ha of intensively managed pine forest, we have a poor understanding of how aquatic systems embedded in managed landscapes contribute to biodiversity. Further, the influence of local- and landscape-scale environmental factors on occupancy of aquatic habitat types by wetland-breeding species in managed forests is unclear. Thus, we investigated anuran occupancy across three ephemeral aquatic system types (altered sites, unaltered sites, and roadside ditches) embedded in an intensively managed pine landscape in eastern North Carolina, USA. These aquatic systems varied in management, disturbance intensity, and landscape context. Altered sites are actively managed as part of the surrounding plantation, unaltered sites are avoided by silvicultural activities (set aside), and ditches receive maintenance as part of routine forest management. We examined occupancy of anuran species at 53 aquatic sites surrounded by early-, mid-, or late-rotation aged stands. During January–July 2013–2014, we conducted repeated call surveys for anurans at aquatic sites and detected 14 species. We used single-species, multi-season occupancy models to examine associations between species occupancy and site- and landscape-scale habitat characteristics for 9 commonly encountered anurans as a function of aquatic system type and stand age class while accounting for imperfect detection. Detection probabilities by species ranged from 0.27 to 0.53 and increased seasonally through the year for most anurans. Species occupancy ranged in 2013 from 0.28 to 0.81 and in 2014 from 0.33 to 0.82. Species occupancy varied by aquatic system type, but the stand age surrounding an aquatic site had little effect on occupancy for most anurans we modeled. Our results indicate that local-scale factors commonly had a larger influence than landscape context on anuran occupancy. We detected evidence of breeding across all aquatic systems and stand age classes, suggesting at least a subset of species are calling and reproductively active. Our study highlights how novel landscape structure and reconfigured ephemeral aquatic systems embedded in intensively managed forests can support anuran occupancy across a range of disturbance intensities. Ó 2016 Elsevier B.V. All rights reserved.

1. Introduction Pine plantations in the southern United States are among the most intensively managed in the world (Schultz, 1997). Since 1952, land area covered by plantations has increased by 1672% to cover >12.9 million ha, with management intensity also increasing (Jokela et al., 2004; Fox et al., 2007). Typical silvicultural operations include mechanical and chemical site preparation, fertilization, thinning, clear-cut harvesting, and declining rotation lengths (Jokela et al., 2010) that produce a landscape mosaic of forest patches varying in age and structural conditions and benefit ⇑ Corresponding author at: Collins Pine Company, 500 Main Street, Chester, CA 96929, USA. E-mail address: [email protected] (B.A. Johnson). http://dx.doi.org/10.1016/j.foreco.2016.03.004 0378-1127/Ó 2016 Elsevier B.V. All rights reserved.

many species (Wigley et al., 2000; Hartley, 2002; Brockerhoff et al., 2008). Similarly, expansive areas of wet pine flats and pocosins wetlands were reconfigured in many areas of the Atlantic Coastal Plain by ditching and draining to remove excess water for supporting forestry, agriculture, and development (Cashin et al., 1992). However, forest landowners also set aside areas of ecological importance, including aquatic habitats, rare ecotypes, and riparian zones, which can still be abundant on reconfigured landscapes (Jones et al., 2010; Leonard et al., 2012). Thus, naturally derived aquatic habitat types and networks of ditches occur through large areas of managed forest, yet their contributions to amphibian diversity and occupancy are only beginning to be understood (Homyack et al., 2014, 2016).

46

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

Many amphibians depend on aquatic sites surrounded by terrestrial habitat to complete their lifecycle (Semlitsch and Bodie, 2003). Aquatic sites provide habitat for breeding, larval development, and over-wintering, while adjacent terrestrial habitat provides adults and dispersing juveniles with foraging opportunities, escape cover, and hibernation or estivation sites. Among aquatic habitat types, small ephemeral pools that are often difficult to detect play critical ecological roles by supporting high species diversity and abundance of amphibians (Russell et al., 2002; Semlitsch and Bodie, 2003; Gibbons et al., 2006). Although natural forests typically support greater amphibian richness than altered areas (Guerry and Hunter, 2002; Denton and Richter, 2013; Walls et al., 2014), there is abundant evidence that pine plantations contribute to wildlife diversity and population maintenance (Wigley et al., 2000; Hartley, 2002; Brockerhoff et al., 2008; Homyack et al., 2014). Prior research on the contributions of altered aquatic habitat types suggests that some anuran species can exhibit behavioral plasticity and use a variety of habitat types. For example, Hyla and Pseudacris species were more abundant in harvested gaps relative to undisturbed bottomland wetlands (Cromer et al., 2002) and other anurans used ditch systems in reconfigured landscapes, even where natural wetlands were present (Mazerolle, 2004; Homyack et al., 2014). Further, other altered wetlands and incidentally created aquatic habitat types, such as machinery ruts and blocked drainages, provide oviposition sites, but these may sometimes serve as population sinks (Adam and Lacki, 1993; DiMauro and Hunter, 2002). At the global scale, herpetofauna are impacted negatively by habitat alteration, road mortality, disease, and other factors (Gibbons et al., 2000; Blaustein et al., 2011, 2012), but managed forests may provide a refuge from some salient threats (O’Bryan et al., 2016). Although numerous studies have examined effects of forest management in the Southern United States on amphibians (Cushman, 2006; Semlitsch et al., 2009), less is known about species-specific responses to different management strategies applied to wetlands within the forested matrix. In many forested systems of the Atlantic Coastal Plain, three kinds of aquatic systems occur on pine landscapes: unaltered sites that are avoided by silvicultural activities, altered sites that are actively managed as part of the surrounding plantation and not set aside, and roadside ditches. We investigated anuran occupancy and breeding effort in these aquatic system types embedded in an intensively managed pine landscape in the Atlantic Coastal Plain of North Carolina. We used occupancy models to examine the influence of forest structure and habitat characteristics at the site and landscape scale while accounting for imperfect detection. We hypothesized that roadside ditches would support occupancy of generalist species, and that altered and unaltered sites also would support occupancy of smaller frogs (e.g., Hyla and Pseudacris spp.) that have more specific habitat requirements, such as increased aquatic vegetation and a lack of predatory fish.

exists a complex ditch network used to lower the water table to promote pine growth and improve operability. In addition, embedded in plantations are several types of ephemeral aquatic systems (Leonard et al., 2012; O’Bryan et al., 2016). The surrounding landscape was a mixture of forest, agriculture lands, and low-density residential housing. From this landscape we selected and sampled the three dominant aquatic system types, which vary in the intensity of disturbance received during forest management activities: (1) unaltered sites, (2) altered sites, and (3) roadside ditches. Unaltered sites were excluded from harvesting, site preparation, and planting activities, and thus remained relatively undisturbed, often with a hardwood canopy (aquatic set-asides). Altered sites were typically small temporary depressions that received the same silvicultural regime as the surrounding stand and thus had been subject to harvesting, site preparation, and planting. Ditch sites were maintained approximately 5 years prior to this study and were adjacent to a single plantation. Because ditches were continuous linear systems, we randomly selected a 150-m transect/site by generating a random center point and sampled 75 m in either direction. We selected aquatic sites embedded in (altered and unaltered sites) or adjacent to (roadside ditches) plantations across a range of stand ages. We stratified aquatic study sites by structural condition of the surrounding plantation (early, mid-, or late rotation age) and by landscape context (sites surrounded by forest to a mixture of forest, agriculture, and rural housing). Early rotation stands were open-canopy, regenerating clearcuts planted from 2008 to 2013 (mean = 2.4 years old, SE = 0.3). Mid-rotation stands were commercially thinned 2008–2013 to approximately 210 trees/ha and were 12–24 years old (mean = 15.5 years old, SE = 0.9). Late rotation stands were commercially thinned 1991–2003 (with the exception of one site thinned in 2008, but maintained structural conditions similar to other late aged stands) and were 21–40 years old (mean = 28.1 years old, SE = 1.4). We identified potential altered aquatic sites using GIS data from previous remote sensing research that predicted the geographic location of small depressions in the study area (see Leonard et al., 2012 for details). We identified potential unaltered sites with GIS data and imagery, and identified potential roadside ditch sites using local forestry records. We cross-referenced potential sites with harvest plans and excluded those with a harvest planned during the project duration. From this pool, we visited sites within each category in a random order. Due to high variability in hydroperiod (i.e., sites may not have had water during site visits), we selected sites based on presence of aquatic vegetation, visible ground depressions, and/or standing water, and selected those P500 m away from the nearest site. We continued selecting sites until we reached our target sample size. We selected 51 of the 110 sites we visited in 2013 and added 2 additional sites in 2014 (one site in 2013 was not used in 2014). These included 16 unaltered aquatic sites (size range = 0.05–2.25 ha), 18 altered aquatic sites (size range = 0.02–0.86 ha), and 19 roadside ditches.

2. Materials and methods

2.2. Aquatic site characteristics

2.1. Study area and aquatic system types

At each site, we quantified habitat characteristics before (March) and after leaf-out (May) in 2013 and 2014. We measured the size of altered and unaltered sites by walking the perimeter of each site with a Garmin e-trex10 GPS unit (Garmin, Chicago, IL) with a WAAS accuracy of <3 m. We delineated perimeters of altered and unaltered sites by identifying the high water line, ground depressions, or abrupt changes in vegetation. We estimated the size of a ditch by measuring the width of the ditch at its widest point within the 150-m transect, then calculated 150 m ⁄ ditch width. We estimated water depth (cm) at the deepest point and canopy openness using a

We conducted our study in an intensively managed loblolly pine (Pinus taeda) landscape in the Coastal Plain of North Carolina (Fig. 1). Plantation silviculture involved clearcutting mature stands (25–35 years old), followed by mechanical (V-shearing and bedding) and chemical (banded or broadcast herbicide prescribed at the stand-level) site preparation, loblolly pine seedlings planted at 1100 trees/ha, fertilization, and a commercial thinning entry (Homyack et al., 2014). Embedded in and adjacent to these stands

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

47

Fig. 1. We examined anuran occupancy across different aquatic system types in an intensively managed pine landscape in Jones, Craven, and Beaufort Counties, North Carolina, USA.

spherical densiometer (Forestry Suppliers, Jackson, Mississippi). At ditch sites, we measured canopy openness at the center and near both ends of the 150-m transect and for altered and unaltered sites, we measured canopy openness near the center of the site, and at 2–7 locations along the perimeter. We characterized vegetation cover by categorizing the percent of site area comprised of reeds, aquatic grass, other aquatic vegetation, cattail (Typha spp.), switch cane (Arundinaria gigantea), and upland grasses (1 = <5%, 2 = P5 to <25%, 3 = P25 to <50%, 4 = P50 to <75%, 5 = P75%) (Weyrauch and Grubb, 2004). Fish presence was a binary variable based on 1– 4 dipnet surveys in 2013 and 1–6 dipnet surveys in 2014 at a site. The number of dipnet surveys at a site depended on water presence. During habitat surveys and all site visits, we noted the presence or absence of water. Sites were visited every 1–2 weeks through the duration of the study both years. We estimated hydroperiod as the percent of time a site was holding water by summing the number of times water was present divided by the total number of unique visits, which accounted for temporary drying and filling periods. We evaluated differences in habitat features among aquatic system types and stand age classes by averaging habitat data across both years within a site and using a two-way ANOVA, followed by a pairwise comparison using Tukey’s HSD for significant results (R Core Team, 2014). 2.3. Anuran surveys We documented presence of calling male anurans at aquatic sites with automated recording devices (Song Meters model SM2 +, Wildlife Acoustics, Maynard, MA). We equipped Song Meters with two microphones and recorded at a sample rate of 22,050 Hz. We deployed Song Meters January–June 2013 and 2014 to capture the wide breeding phenology of anurans in the

study area (Homyack et al., 2014). We recorded anurans across 3 seasons approximately following the North Carolina Calling Amphibian Survey Program (CASP) survey dates; season 1: January 15–February 28, 2: March 15–April 30, and 3: May 15–June 30 (North Carolina Partners in Amphibian and Reptile Conservation, 2014). We attached Song Meters on trees or polyvinyl chloride piping at approximately 1.2 m high at the site perimeter and oriented them facing the site. At roadside ditches, we mounted Song Meters at the center point of the sampling segment. We deployed one Song Meter/site for five consecutive nights (i.e., secondary sampling period) and recorded 5 min/h for 8 h beginning 30 min after sunset. We recorded two secondary sampling periods/season/year at each site. This sampling schedule allowed us to examine temporal variation in calling male anurans both seasonally and within a night. Across the study we had 6 secondary sampling occasions within one primary sampling period (year) for use in our occupancy models. We manually analyzed recordings using SongScope software (Wildlife Acoustics, Maynard, MA). A single user (BJ) listened to recordings and visually scanned spectrograms to identify species. A site was considered occupied if the species was detected (heard in the recording or visually identified in the spectrogram) in any of the recordings during a single year. We collected site-specific air temperatures from Song Meters for each 5-min recording and obtained daily total rainfall estimates from the nearest National Oceanic and Atmospheric Administration weather station from January–April 2013 and from an onsite weather station from April 2013 to June 2014 (National Oceanic and Atmospheric Administration Climate Center, 2014). Thus, air temperatures were specific to a site and survey, and rainfall measurements did not vary across sites for a given sampling date. We averaged temperatures and combined rainfall within each secondary sampling occasion.

48

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

Additionally, we investigated reproductive effort among aquatic systems with dipnet surveys of larval anurans. We performed four dipnet surveys April–July 2013 and six dipnet surveys March–June 2014, spaced approximately monthly. For altered and unaltered sites, we used dip nets with a 40.6 cm
40.6 cm frame and a net that was 40.6 cm deep with a 3/1600 mesh. At ditch sites, due to steep banks, we used a telescoping pool pole with a 40.6 cm
45.1 cm frame and a fine mesh. Each site received 5– 40 sweeps based on a visual estimate of the percentage of water available at a site (<0.1 ha  10 sweeps, P0.1 to <0.5 ha  20 sweeps, P0.5 to <1 ha  30 sweeps, P1 ha  40 sweeps) (Gunzb urger, 2007). We dipnetted approximately every 4–6 m along the perimeter of a site using a 1 m scoop while jabbing the net into the substrate toward the shoreline (Heyer et al., 1994). We also sampled unique micro-habitats that were otherwise missed, including patches of aquatic vegetation, and we counted and identified larval amphibians in the field to species when possible (Richter-Boix et al., 2007). We conducted a contingency analysis using the Fisher’s exact test to test for independence between aquatic system type and tadpole presence on ten species using combined tadpole data from 2013 to 2014. 2.4. Occupancy analyses We used anuran call data to develop single-species multiseason occupancy models for commonly observed anurans to investigate relationships among species occupancy and environmental covariates (Mackenzie et al., 2003). Multi-season models incorporate four parameters: wi, the probability a species initially occupies site i; pij, the probability a species is detected at site i on survey j given that is it present at the site; ct, the probability than an unoccupied site in season t is occupied by the species in season t + 1; and et, the probability that a site occupied in season t is unoccupied by the species in season t + 1. We used maximum likelihood techniques to estimate these parameters based on MacKenzie et al. (2003). We modeled dynamic changes in occupancy as a first-order Markov process, so that occupancy at a site in season t depends on the occupancy state at a site in the previous season, t  1, or wt+1 = wt(1  et) + (1  wt)ct. Under this model, it is expected that a site occupied in year t is more likely to be occupied the following year than one currently unoccupied. Multi-season occupancy models assume the population is closed to changes in occupancy within, but are open to changes between primary seasons. Movements of species in or out of aquatic sites within a primary period could violate this assumption. We conducted surveys within the core breeding time of each species to decrease the likelihood that species were not available for detection within a primary period. However, random moving among ponds was unavoidable and likely random, so that we did not expect this to result in significant bias with the closure assumption. Prior to evaluating factors that influenced occupancy, we identified survey-specific covariates that may have influenced detection, including Julian day, temperature, rain, and presence of water. We modeled Julian day (middle day during a 5 night survey) using a linear and quadratic effect since species core breeding times varied throughout the year. We estimated temperature as the average and maximum air temperature recorded by the Song Meter across a secondary sampling occasion. We estimated rain as (1) the total amount of rainfall that occurred one day prior to and during the five days of recordings and (2) the number of nights it rained during a sampling period and the day prior to recordings. We modeled presence of water as a binary covariate during a 5 night survey. For southern leopard frog (Lithobates sphenocephalus), American bullfrog (Lithobates catesbeiana), and green frog (Lithobates clamitans), we excluded water as a detection covariate

because they are typically found near water (Oseen and Wassersug, 2002; Saenz et al., 2006). We standardized all variables except water presence and total rainfall sensu Homyack et al. (2014). We also incorporated covariates to account for differences in detection between years and among surveys and a null model. We modeled detection covariates individually and as all combinations for each species to develop the best fitting models while keeping occupancy, extinction, and colonization constant. After identifying species-specific detection covariates from the most supported model (DAIC = 0) for each species, we incorporated these detection covariates into models that evaluated the influence of site- and landscape-scale covariates on species-specific occupancy probabilities. We did not incorporate covariates to address local extinction and colonization. These parameters were left as constants in our models because we felt this was appropriate given the short-term nature of the study and because our primary objective was to understand patterns of occupancy across aquatic sites, not examine hypotheses related to colonization and extinction rates. We modeled occupancy as a function of 12 site-scale and 8 landscape-scale covariates, using a logit link function (Table 1). These site- and landscape-scale factors initially were evaluated separately, and then the variables from top competitive models in each separate analysis were combined into multi-scale candidate models. At the site-scale, occupancy models were created using single covariates and additional multi-covariate models were created to test the effect of hydroperiod, stand age, or aquatic system type in combination with the other site-level covariates mentioned. Our approach produced estimates of initial occupancy for year 1 and derived estimated for year 2. Occupancy covariates reflect conditions in year one (hydroperiod and fish presence) or covariates that we did not expect to change between years (all other covariates). We averaged vegetation percentages across years and then simplified variables into classes to examine aquatic vegetation (reeds, aquatic grass, and other aquatic vegetation combined), hard-structured vegetation (cane and cattail), grass, and the combination of all three vegetation variables in a combined model (aquatic vegetation + grass + hard structured vegetation). We felt this was appropriate as we did not anticipate site vegetation conditions to significantly differ between years. Additionally, we quantified the amount of wetland and mature forest within a 50 m radius of a site as local continuous covariates to address terrestrial habitat immediately surrounding sites. We estimated wetland area as a combination of known habitat set-asides (other unmanaged aquatic areas), predicted unmanaged sites (Leonard et al., 2012), and other hand-delineated aquatic sites using colorinfrared (CIR) images. We compared these identified areas to the National Wetland Inventory (NWI) in GIS to identify additional aquatic features. We characterized habitat features at broader scales (50 m and 400 m) in a geographic information system (GIS) using ArcMap 10.1 (ESRI Inc., Redlands, CA). We incorporated the 50-m scale to examine local features and the 400-m radius because it represents the average terrestrial area frogs use (Semlitsch and Bodie, 2003). We bounded covariates representing the percent of area surrounding a Song Meter, the center point for anuran occupancy, between 0 and 1. We also modeled wetland area (50 m) and total ditch length within a 50 m radius in a combined model. We assessed the weight of evidence for a model with Akaike’s Information Criterion (AIC) (Burnham and Anderson, 2002). All occupancy analyses were performed in Program Presence 6.4 (Hines, 2006). To further understand how species occupancy varied among aquatic system types and stand age classes, we did a posthoc comparison of mean occupancy across categories for species where aquatic system type or stand age class covariates appeared in most competitive models (DAIC < 2).

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

49

Table 1 Site and landscape scale occupancy covariates used in the single-species multi-season occupancy models for anurans across a managed forest landscape in the Atlantic Coastal Plain of North Carolina. Occupancy covariate

Scale

Type

Description

Size Stand.age Can.open Aq.veg Grass Hard.struct Vegetation Aq.syst Hydro Fishless %Wet50, 400 %MatFor50, 400 %YoungFor400 %NonFor400

Site Site Site Site Site Site Site Site Site Site Site/landscape Site/landscape Landscape Landscape

Continuous Continuous Continuous Categorical Categorical Categorical Categorical Categorical Continuous Binary Continuous Continuous Continuous Continuous

%NLCDFor400

Landscape

Continuous

HYW400 WEYrd400 Ditch50

Landscape Landscape Landscape

Binary Continuous Continuous

Size of site (Ha) Age of pine stand surrounding a site (Years) Average site canopy openness Average vegetation percent across both years at a site. Combined reeds, aquatic grass, and other aquatic vegetation Average vegetation percent across both years at a site Average vegetation percent across both years at a site. Combined cattail and cane Combined model assessing Aq.Veg + Grass + Hard.Struct variables Altered pond, unaltered pond, or ditch Percentage of time a site held water (0–100) Presence or absence of fish at a site Percent of wetland surrounding a 50 and 400 m radius of a Song Meter Percent of mature forest >20 years old surrounding a 50 and 400 m radius of a Song Meter Amount of young forest <4 years old surrounding a 400 m radius of a Song Meter Percent of National Land Cover Database 2011 designation of non-forest area (barren, scrubland, herbaceous, and planted/cultivated combined) within a 400 m radius of a Song Meter (NLCD, 2011) Percent of National Land Cover Database 2011 designation of forest area within a 400 m radius of a Song Meter (NLCD, 2011) Presence or absence of a public road within a 400 m radius of a Song Meter Total length of forest roads within a 400 m radius of a Song Meter Total length of non-roadside (internal) ditches within a 50 m radius of a Song Meter

3. Results 3.1. Environmental variation Environmental characteristics of sites varied by both aquatic system type and stand age class. In both years, many sites dried by May–June and never re-filled, with 19 (37%) drying in 2013 and 28 (53%) drying in 2014. We detected a significant interaction between the effects of aquatic system type and stand age class on hydroperiod (F4,50 = 4.91, P = 0.002). Overall, hydroperiods were longer in roadside ditches and retained water through 98% of the field season with little differences among stand age classes. In contrast, altered sites and unaltered sites had a mean hydroperiod of 64% and 77% of the study period, respectively. In altered sites, hydroperiod increased in mid-rotation stands and decreased in late-rotation stands, while unaltered sites had an opposite trend (decreased hydroperiod in mid-rotation stands and increased hydroperiod in late-rotation stands). Most (90%) roadside ditches had fish present both years, whereas we detected fish in 17% of altered and 56% of unaltered sites. Aquatic systems also varied by depth (F2,50 = 5.61, P = 0.006). Altered sites were shallower than unaltered sites (Tukey’s HSD, P = 0.009) by 27 cm and roadside ditches (Tukey’s HSD, P = 0.04) by 21 cm. We detected a significant difference in canopy openness among aquatic systems (F2,50 = 10.21, P < 0.001) and among stand age class (F2,50 = 54.60, P < 0.001), but there was not a significant interaction (F4,50 = 1.92, P = 0.12). Unaltered sites had significantly greater canopy cover (63%) than roadside ditches (41%; Tukey’s HSD, P > 0.001) and altered sites (44%; Tukey’s HSD, P = 0.002). The canopy of earlyaged sites was more open (79%) than middle-aged (41%) (Tukey’s HSD, P > 0.001) and late-aged sites (29%) (Tukey’s HSD, P > 0.001). Vegetation characteristics also differed among aquatic sites. We found a significant interaction between the effect of aquatic system type and stand age class on aquatic vegetation (F4,50 = 3.98, P = 0.008). In altered and unaltered sites, as stand age class increased, the amount of aquatic vegetation cover decreased. For roadside ditches, aquatic vegetation cover increased in midrotation stands and then decreased in late-rotation stands. Further, we detected significant differences among aquatic systems for grass (F2,50 = 15.04, P = <0.001) and hard-structured vegetation cover (F2,50 = 7.94, P = 0.01). Altered sites had greater grass cover than either unaltered sites (Tukey’s HSD, P = 0.002) or roadside

ditches (Tukey’s HSD, P < 0.001) and roadside ditches had less hard-structured vegetation cover (cattail and cane) than altered (Tukey’s HSD, P < 0.001) or unaltered sites (Tukey’s HSD, P = 0.01). 3.2. Anuran occupancy We recorded calling anurans for 1454 sampling nights during 6 periods in 2013 and 1569 sampling nights during 6 periods in 2014, which resulted in 1064 detections of 14 anuran species. One species, oak toad (Anaxyrus quericicus), which was detected at 10 sites, is North Carolina priority species (North Carolina Wildlife Resources Commission, 2015). For variable detection for the 9 most commonly observed species, we developed singlespecies, multi-season occupancy models assessing site- and landscape-scale habitat features while accounting for imperfect detection. We examined a large number of models related to detection and occupancy probabilities at multiple scales to describe multi-scale relationships, not predict occupancy in other systems. Our approach produced a small number (64) of competitive models (DAIC < 2) for seven of nine species (Table 2), but American bullfrog and southern cricket frog (Acris gryllus) had 9 and 14 competitive models, respectively. Estimated probabilities of detection varied across species (Table 3). Except for southern leopard frog and green frog, detection probabilities were influenced positively by Julian day, number of days with rain, or total rainfall. Air temperature positively influenced detection probability only of pine woods treefrog (Hyla femoralis). Estimates of initial occupancy also varied across species and sites. After combining most-supported site and landscape covariates, all nine species’ occupancies had at least one competitive model (DAIC < 2) associated with a combination of site- and landscape-scale habitat features. Two species [Spring peeper (Pseudacris crucifer) and Brimley’s chorus frog (Pseudacris brimleyi)] had additional competitive models only associated with site-scale habitat features while squirrel treefrog (Hyla squirella) was the only species to have a landscape-scale only model also appear as a competitive model (Table 2). Our results indicate that aquatic system type influenced occupancy for three species. Spring peeper and Brimley’s chorus frog had reduced occupancy in roadside ditches (Fig. 2). Spring peeper also had lower occupancy in altered sites while Brimley’s chorus frog had increased occupancy in altered sites relative to other sys-

50

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

Table 2 Most supported (DAIC < 2, up to four models) single-species multi-season occupancy models by species for combined site- and landscape-scale habitat covariates with the top species-specific detection covariate. Local colonization and extinction were kept constant in all models. Top site and landscape occupancy models <2 DAIC Model, by species

DAIC

wi

k

Spring peeper w(aq.syst + aq.veg + ditch50), c(.), e(.), p(julian2 + rain.nights) w(aq.syst + aq.veg + %Wet50), c(.), e(.), p(julian2 + rain.nights)

0 0.54

0.36 0.28

11 11

Brimley’s chorus frog w(aq.syst + aq.veg + ditch50), c(.), e(.), p(julian2) w(aq.syst + aq.veg + ditch50 + %Wet50), c(.), e(.), p(julian2)

0 0.88

0.28 0.18

10 11

Little grass frog w(size + HWY400), c(.), e(.), p(julian + rain.total) w(size), c(.), e(.), p(julian + rain.total) w(size + stand.age + HWY400), c(.), e(.), p(julian + rain.total) w(size + %YoungFor400), c(.), e(.), p(julian + rain.total)

0 0.42 0.96 1.36

0.29 0.23 0.18 0.15

8 7 9 8

Southern cricket frog (Top 4 of 14 models DAIC < 2) w(size + hydro13), c(.), e(.), p(julian2 + water) w(size), c(.), e(.), p(julian2 + water) w(size + hydro13 + WEYrd400), c(.), e(.), p(julian2 + water) w(size + hydro13 + %MatFor400), c(.), e(.), p(julian2 + water)

0 0.45 0.85 1.32

0.10 0.08 0.07 0.06

9 8 10 10

Pine woods treefrog w(Stand.age + size + WEYrd400), c(.), e(.), p(julian + rain.total + air.max + water)

0

0.52

11

Squirrel treefrog w(%Wet400), c(.), e(.), p(julian2 + rain.nights + water) w(%Wet400 + size), c(.), e(.), p(julian2 + rain.nights + water) w(%Wet400 + hydro), c(.), e(.), p(julian2 + rain.nights + water) w(%Wet400 + hard.struct), c(.), e(.), p(julian2 + rain.nights + water)

0 0.3 0.78 1.09

0.19 0.16 0.13 0.11

10 11 9 9

Bullfrog (Top 4 of 9 models DAIC < 2) w(hydro13 + hard.struct + %YoungFor400), c(.), e(.), p(julian + rain.nights) w(hydro13), c(.), e(.), p(julian + rain.nights) w(hydro13 + can.open + %YoungFor400), c(.), e(.), p(julian + rain.nights) w(hydro13 + aq.syst), c(.), e(.), p(julian + rain.nights)

0 0.51 0.67 0.68

0.10 0.08 0.07 0.07

9 7 9 9

Green frog w(hydro13 + aq.veg + %Wet50 + ditch50), c(.), e(.), p(t) w(hydro13 + vegetation), c(.), e(.), p(t) w(hydro13 + aq.veg), c(.), e(.), p(t) w(hydro13 + aq.veg + %Wet400), c(.), e(.), p(t)

0 0.63 0.73 1.85

0.26 0.19 0.18 0.10

15 15 13 14

Southern leopard frog w(hydro13 + %MatFor50 + WEYrd400), c(.), e(.), p(t) w(hydro13 + WEYrd400), c(.), e(.), p(t)

0 1.88

0.27 0.11

18 17

Table 3 Species-specific probabilities of occurrence (w estimated for year 1 and derived in year 2), detection (p), colonization (c), and extinction (e) rates based on the top model (DAIC = 0). Species

Year 1 w (SE)

Year 2 w (SE)

p (SE)

Spring peeper Brimley’s chorus frog Southern leopard frog American bullfrog Green frog Little grass frog Squirrel treefrog Pine woods treefrog Southern cricket frog

0.574 0.369 0.809 0.421 0.281 0.511 0.568 0.627 0.406

0.720 0.450 0.818 0.326 0.407 0.568 0.621 0.700 0.500

0.528 0.501 0.505 0.419 0.475 0.335 0.369 0.268 0.312

(0.120) (0.108) (0.057) (0.110) (0.089) (0.099) (0.096) (0.134) (0.105)

(0.094) (0.112) (0.075) (0.100) (0.095) (0.092) (0.089) (0.134) (0.114)

tem types. Additionally, American bullfrog had 2 of 9 competitive models that included aquatic system type, showing higher occupancy in roadside ditches compared to altered and unaltered sites (Fig. 2). Similarly, stand age class influenced occupancy for little grass frog (Pseudacris ocularis) and pine woods treefrog. Little grass frog showed a positive relationship with stand age for one of four competitive models and pine woods treefrog had a negative relationship with stand age (1 of 1 competitive models) (Fig. 3). Though few species had relationships with stand age at the site level, two additional species had occupancy probabilities related to stand

c (SE) (0.047) (0.054) (0.075) (0.068) (0.109) (0.039) (0.057) (0.054) (0.060)

0.493 0.178 0.334 0.085 0.331 0.317 0.399 0.429 0.485

e (SE) (0.113) (0.077) (0.140) (0.057) (0.082) (0.099) (0.125) (0.161) (0.127)

0.113 0.098 0.045 0.321 0.389 0.191 0.230 0.138 0.476

(0.062) (0.078) (0.037) (0.127) (0.143) (0.098) (0.105) (0.125) (0.157)

structure at the landscape scale. A positive relationship with the amount of young forest within 400 m of a site appeared in five of nine competitive models for American bullfrog. Forest cover within 400 m of a site was positively related to occupancy for southern cricket frog in 5 of 14 competitive models. In contrast to our predictions, occupancy probabilities of other species were more influenced by site characteristics than forest management. Site size was positively related to occupancy for southern cricket frog (14 of 14 competitive models), little grass frog (4 of 4 competitive models) and American bullfrog (2 of 9 competitive models), and negatively related for pine woods tree-

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

51

Fig. 3. Species-specific mean probabilities of occurrence (± standard error) by stand age class. These data are only plotted for species with stand age occurring in competitive occupancy models (delta AIC < 2) including (a) little grass frog and (b) pinewoods treefrog. Gray circles represent estimates of occupancy for 2013 and black circles represent derived estimates of occupancy for 2014. Early rotation stands were regenerating clearcuts planted from 2008 to 2013, mid-rotation stands were established 1989–2001 and commercially thinned 2008–2013 and late rotation stands were previously commercially thinned 1991–2003 and were 21– 40 years old.

Fig. 2. Species-specific mean estimated probabilities of occurrence (± standard error) by aquatic system type. These data are only plotted for species with aquatic system occurring in competitive occupancy models (DAIC < 2), including (a) spring peeper, (b) Brimley’s chorus frog, and (c) American bullfrog. Gray circles represent estimated occupancy for 2013 and black circles represent derived estimates of occupancy for 2014. Altered sites and ditches are managed with the surrounding stand and are considered high disturbance systems, whereas unaltered sites are avoided during forest management activities and are a low disturbance system, relative to altered sites and ditches.

frog (1 of 1 competitive models) and squirrel treefrog (1 of 4 competitive models). Hydroperiod helped explain patterns of occupancy for five species. Hydroperiod was positively related to occupancy for southern leopard frog (2 of 2 competitive models), American bullfrog (9 of 9 competitive models), green frog (4 of 4 competitive models), and southern cricket frog (9 of 14 competitive models) and negatively related for squirrel treefrog (1 of 4 competitive models). Lastly, vegetation characteristics explained occupancy patterns for five species. Aquatic vegetation cover was positively related to occupancy for spring peeper (2 of 2 competitive models) and Brimley’s chorus frog (2 of 2 competitive models).

Aquatic vegetation cover was negatively related to green frog (4 of 4 competitive models) occupancy while hard structured vegetation cover was positively related to occupancy (1 of 4 competitive models). Hard structured vegetation was also positively related to occupancy for American bullfrog (2 of 9 competitive models) and squirrel treefrog (1 of 4 competitive models). 3.3. Observed breeding effort We detected amphibian larvae in all aquatic system types during both years. In 2013, we observed larvae in 44% of roadside ditches, 53% of altered sites, and 56% of unaltered sites. In 2014, we observed larvae in 33% of roadside ditches, 61% of altered sites, and 69% of unaltered sites. During dipnet surveys, we captured 3,568 individuals of 10 anuran species in 2013 and 1,736 individuals of 7 anuran species in 2014. In 2013, the majority were southern leopard frog (52%) and southern toad (Anaxyrus terrestris) tadpoles (30%) and southern leopard frog tadpoles were most common in 2014 (80%). Across both years, we detected 8 tadpole species in both altered and unaltered sites and 4 species in roadside ditches (Table 4). Of those in altered and unaltered sites, 4 [southern cricket frog, Eastern narrowmouth toad (Gastrophryne carolinensis), Brimley’s chorus frog, and little grass frog] had >90% of individuals detected in altered sites. Based on the contingency table analysis, breeding was dependent on aquatic system type

52

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

Table 4 Tadpole captures during dipnet sampling of three different aquatic system types on a managed pine plantation in the Coastal Plain of North Carolina, U.S.A. Sampling occurred April–July 2013 (4 surveys/site) and March–June 2014 (6 surveys/site) and results are expressed as Catch Per Unit Effort. Mean CPUE ± SE

Southern cricket frog Spring peeper Brimley’s chorus frog Little grass frog Southern toad Eastern narrowmouth toad Pine woods treefrog American bullfrog Green frog Southern leopard frog

Percent of sites occupied

Unaltered sites (number of sites with detections)

2013

2014

2013

2014

2013

5.8 7.7 1.9 5.8 15.4 5.8 21.2 5.8 9.6 48.1

– 26.4 17.0 1.9 3.8 – – 1.9 1.9 43.4

– 0.05 ± 0 (1) – – 27.24 ± 24.4 (2) 0.05 ± 0 (1) 2.75 ± 1.8 (4) 0.50 ± 0.3 (2) 0.82 ± 0.3 (3) 4.96 ± 3.4 (9)

– 2.09 ± 3.1 (5) 0.04 ± 0 (2) – – – – – 0.05 ± 0 (1) 1.65 ± 1.4 (9)

0.29 ± 0.2 1.71 ± 0.3 6 ± 0 (1) 0.52 ± 0.3 8.49 ± 4.5 0.77 ± 0.6 2.67 ± 1.0 – 0.27 ± 0.1 8.39 ± 4.7

for spring peeper (P = 0.002), pine woods treefrog (P = 0.004), green frog (P = 0.03), and southern leopard frog (P = 0.04). For all of these, breeding was absent (spring peeper, pine woods treefrog, green frog) or reduced (southern leopard frog) in roadside ditches. 4. Discussion Within the Atlantic Coastal plain, managed pine plantations contain aquatic systems that are subjected to a range of disturbance intensity and reconfiguration due to silvicultural practices. Our results suggest the aquatic systems we studied supported numerous frog and toad species that were generally affected more by site- and landscape-scale habitat characteristics than disturbance from forest management. However, aquatic system type was an important driver for two small-sized hylids (spring peeper and Brimley’s chorus frog) and American bullfrog and stand age emerged as an important determinant for little grass frog and pinewoods treefrog. Though we primarily detected species of little conservation concern, common species represent a large portion of biomass and can contribute disproportionately to ecosystem function relative to uncommon species (Gaston and Fuller, 2007). Further, common species are affected by many of the same stressors as rare species. Therefore, understanding how common species respond to these stressors is important for managing population persistence across the landscape (Rustigian et al., 2003). Species responses to aquatic system types were often a function of habitat characteristics at the site. For example, the importance of hydroperiod to anurans is well documented and influenced some amphibian species occurring in our study area (Babbitt, 2005; Baldwin et al., 2006; Denton and Richter, 2013). Southern leopard frog, American bullfrog, and green frog had positive relationships with hydroperiod and had higher occupancy rates in roadside ditches and some unaltered sites which contained water through most of the study. In contrast, these species had lower occupancy at altered sites, some of which had very short hydroperiods. Permanent sites are likely important breeding habitat for these Lithobates species, particularly American bullfrog, which require permanent water for extended larval development. Other site characteristics related to hydroperiod also likely affected occupancy. Many of the species we modeled generally use ephemeral systems. In our landscape, ditches were interconnected aquatic systems that exhibited short-term flow after rain events. This combined with long hydroperiods and the presence of predatory fish may have limited occupancy at some sites, yet our overall species pool was similar to those detected solely in ditches (Homyack et al., 2014). Ditches may serve as refugia when upland aquatic sites dry and may assist dispersal across the landscape (Mazerolle, 2004; O’Bryan et al., 2016). The study area is highly hydrologically con-

Altered sites (number of sites with detections)

(2) (2) (3) (3) (2) (7) (1) (6)

Roadside ditch (number of sites with detections)

2014

2013

2014

– 0.75 ± 0.5 (9) 1.16 ± 1.3 (7) 0.1 ± 0 (1) – – – – – 3.58 ± 4.0 (10)

0.20 ± 0 (1) – – – 12.17 ± 10.9 (3) – – – – 0.43 ± 0.2 (8)

– – – – 1.80 ± 0.7 (2) – – 0.1 ± 0 (1) – 0.47 ± 0.3 (4)

nected and overland flow potentially contributes to temporally and spatially dynamic fish presence even in altered and unaltered sites. This may partially explain why fish presence was not associated with decreased occupancy across any site type. It is also possible that our sampling methods for fish presence were insufficient. Future work aimed toward assessing fish influence would likely benefit from additional techniques that would yield higher detection probabilities. Vegetation differences at the site-level among aquatic systems also influenced occupancy for many species. Of the nine species that we modeled, five (2 small hylids, 1 treefrog, and 2 Lithobates) had relationships with vegetation covariates. Altered sites had high structural complexity with greater cover from hard-structured vegetation than roadside ditches, greater grass cover than both unaltered and roadside ditches, and abundant aquatic vegetation. Vegetation at wetlands provides forage and cover for tadpoles and adults, calling sites for males, and egg attachment sites (Dorcas and Gibbons, 2008; Gorman and Haas, 2011). Vegetation differences among aquatic systems are due, in part, to forest management. Although altered sites received significant disturbances from harvesting, site preparation and planting, this management regime may promote abundant herbaceous vegetation that supports occupancy. Further, management activities such as clearcut harvesting and commercial thinning decrease pond shading and can increase water temperature, enhancing larval development for some pond-breeding amphibians (Skelly et al., 2014). Structural condition surrounding an aquatic site did not influence occupancy for most of the species we examined. It is possible we observed this trend because the study area was dominated by forest (Baldwin et al., 2006). Heavily shaded ponds are often species poor relative to open canopy ponds, suggesting that encroachment of overstory vegetation is associated with species loss (Skelly et al., 1999; Werner et al., 2007). We did not observe this in our study area, potentially because historically this area was wetland habitat and despite current forest management, there is still high hydrological connection. Although forest management affects habitat conditions, the heterogeneity of structural conditions, aquatic system types, and connectivity likely allows amphibians in the area to choose sites based on site conditions, rather than explicit management regime. There are several advantages and disadvantages to using automated recording devices such as the Song Meter. Vegetation can influence how far away from the recording device a species may be detected. We believe this error was minimized at our sites since recorders were placed at the edge of a pond and ponds tended to be relatively free of vegetation across the pond surface and our interest was in males calling within the immediate vicinity of a site. Further, this interference from vegetation would have varied

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

greatly among sites. We additionally collected a large number of samples in both primary and secondary sampling periods that allowed us to assess anuran presence across a range of conditions. We feel confident in our methods as our species pool was similar to Homyack et al. (2014) and Homyack et al. (2016), who used different survey methods, suggesting that our technique was appropriate for the species we modeled. Other research has observed adult frogs at ditches in peat bogs, but found no sign of reproductive activity (Mazerolle, 2004) and some research suggested altered sites may act as ecological traps (Dimauro and Hunter, 2002). In contrast, managed forests supported 15 species of calling male anurans in ditches (Homyack et al., 2014) and forest management, such as clearcutting, can indirectly benefit many larval amphibians through warmer water temperatures leading to increased survival and faster development (Semlitsch et al., 2009). We detected amphibian larvae across all three aquatic system types and stand ages classes, suggesting all habitat types can provide breeding opportunities for at least a portion of the regional assemblage. Lower captures of individuals and total species in 2014 was likely related to drier conditions and a shorter sampling period in 2014. Most species for which breeding was compared across sites showed no preference among aquatic system types. This may be an artifact of the low number of total sites where individual species were detected breeding. Where breeding was dependent on aquatic system type, we found reduced (southern leopard frogs) or no breeding effort (spring peeper, pine woods treefrog, and green frog) in roadside ditches. The majority of our species may not have produced tadpoles in roadside ditches, but these species were also only detected at one or two altered or unaltered sites. Though we did not follow the fate of observed tadpoles, we observed dispersing metamorphs at multiple sites, including southern leopard frog, spring peeper, Brimley’s chorus frog, pine woods treefrog, and southern toad. Future work that examines reproductive success and juvenile dispersal in different aquatic systems and stand ages would provide insight into how different aquatic systems in managed forests contribute to persistence of these species.

5. Conclusion Voluntary implementation of forestry best management practices (BMPs) and third-party sustainability certification requirements by forest industry have led to increased water quality and protection of unique habitat types, including aquatic systems (Jackson, 2014; Cristan et al., 2016). Consequently, the contributions of managed pine forests to biodiversity conservation may be greater than sometimes appreciated, particularly for aquatic and semi-aquatic species. Our study area consisted of a wide range of structural conditions and aquatic habitats, likely benefiting many species that require different types of habitat at the site scale. Our analyses show that species occupancy varied by aquatic system type, with different site- and landscape-scale habitat characteristics influencing occupancy, while stand age surrounding an aquatic site had little effect on occupancy for the majority of the species we modeled. Given different life histories of the many species occurring in this area, it is expected that species would respond differently to aquatic sites that vary in their habitat characteristics. Roadside ditches provided well connected habitat with long hydroperiods that are critical for large-bodied anurans with long developmental stages and may also provide routes for anuran dispersal into forest stands or to other aquatic features. Altered and unaltered sites also provide suitable habitat for a number of species, both as adults and for tadpole development. We detected breeding effort at over half of these aquatic sites, suggesting

53

species were calling and reproductively active. However, we only modeled occupancy of nine common species occurring in this managed forest. Information gaps still exist on survivorship, movement patterns, and species that were less common in the landscape. Despite this, forest management in this area is likely contributing to regional diversity of commonly occurring anurans by creating a wide range of aquatic systems and structural conditions, and these diverse habitats appear to support the many different life histories of anurans that occur across the South Atlantic Coastal Plain. Acknowledgements Funding was provided by Weyerhaeuser Company, the North Carolina Wildlife Resources Commission, and the National Council for Air and Stream Improvement. We thank L. Paden and A. Temple for field assistance and Jeff Hall for providing knowledge of herpetofauna within the study area. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foreco.2016.03. 004. References Adam, M.D., Lacki, M.J., 1993. Factors affecting amphibian use of road-run ponds in Daniel Boone National Forest. Trans. Kentucky Acad. Sci. 54, 13–16. Babbitt, K.J., 2005. The relative importance of wetland size and hydroperiod for amphibians in southern New Hampshire, USA. Wetlands Ecol. Manage. 13, 269– 279. Baldwin, R.F., Calhoun, A.J.K., deMaynadier, P.G., 2006. The significance of hydroperiod and stand maturity of pool-breeding amphibians in forested landscapes. Can. J. Zool. 84, 1604–1615. Blaustein, A.R., Han, B.A., Relyea, R.A., Johnson, P.T.J., Buck, J.C., Gervasi, S.S., Kats, L. B., 2011. The complexity of amphibian population declines: understanding the role of cofactors in driving amphibian losses. Ann. N. Y. Acad. Sci. 1223, 109– 119. Blaustein, A.R., Gervasi, S.S., Johnson, P.T.J., Hoverman, J.T., Belden, L.K., Bradley, P. W., Xie, G.Y., 2012. Ecophysiology meets conservation: understanding the role of disease in amphibian population declines. Phil. Trans. R. Soc. B 367, 1688– 1707. Brockerhoff, E.G., Jactel, H., Parrotta, J.A., Quine, C.P., Sayer, J., 2008. Plantation forests and biodiversity: oxymoron or opportunity? Biodivers. Conserv. 17, 925–951. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference. Springer, New York, NY. Cashin, G.E., Dorney, J.R., Richardson, C.J., 1992. Wetland alteration trends on the North Carolina Coastal Plain. Wetlands 12, 63–71. Cristan, R., Aust, W.M., Bolding, M.C., Barrett, S.M., Munsell, J.F., Schilling, E., 2016. Effectiveness of forestry best management practices in the United States: literature review. For. Ecol. Manage. 360, 133–151. Cromer, R.B., Lanham, J.D., Hanlin, H.H., 2002. Herpetofaunal response to gap and skidder-rut wetland creation in a Southern bottomland hardwood forest. For. Sci. 48, 407–413. Cushman, S.A., 2006. Effects of habitat loss and fragmentation in amphibians: a review and prospectus. Biol. Conserv. 128, 231–240. Denton, R.D., Richter, S.C., 2013. Amphibian communities in natural and constructed ridge top wetlands with implications for wetland construction. J. Wildl. Manage. 77, 886–896. DiMauro, D., Hunter, M.L., 2002. Reproduction of amphibians in natural and anthropogenic temporary pools in managed forests. For. Sci. 48, 397–406. Dorcas, M., Gibbons, W., 2008. Frogs and Toads on the Southeast. The University of Georgia Press, Athens, GA. Fox, T.R., Jokela, E.J., Allen, H.L., 2007. The development of pine plantation silviculture in the southern United States. J. Forest. 105, 337–347. Gaston, K.J., Fuller, R.A., 2007. Commonness, population depletion and conservation biology. Trends Ecol. Evol. 23, 14–19. Gibbons, J.W., Winne, C.T., Scott, D.E., Willson, J.D., Glaudas, X., Andrews, M.M., Todd, B.D., Fedewa, L.A., Wilkinson, L., Tsaliagos, R.N., Harper, S.J., Greene, J.L., Tuberville, T.D., Metts, B.S., Dorcas, M.E., Nestor, J.P., Young, C.A., Akre, T., Reed, R.N., Buhlmann, K.A., Normal, J., Croshaw, D.A., Hagen, C., Rothemel, B.B., 2006. Remarkable amphibian biomass and abundance in an isolated wetland: implications for wetland conservation. Conserv. Biol. 20, 1457–1465. Gibbons, J.W., Scott, D.E., Ryan, T.J., Buhlmann, K.A., Tuberville, T.D., Metts, B.A., Greene, J.L., Milles, T., Leden, Y., Poppy, S., Winne, C.T., 2000. The global decline of reptiles, déjà vu amphibians. Biosphere 50, 653–666.

54

B.A. Johnson et al. / Forest Ecology and Management 368 (2016) 45–54

Gorman, T.A., Haas, C.A., 2011. Seasonal microhabitat selection and use of syntopic populations of Lithobates okaloosae and Lithobates clamitans clamitans. J. Herpetol. 45, 313–318. Guerry, A.D., Hunter, M.L., 2002. Amphibian distributions in a landscape or forests and agriculture: an examination of landscape composition and configuration. Conserv. Biol. 16, 745–754. Gunzburger, M.S., 2007. Evaluation of seven aquatic sampling methods for amphibians and other aquatic fauna. Appl. Herpetol. 4, 47–63. Hartley, M.J., 2002. Rationale and methods for conserving biodiversity in plantation forests. For. Ecol. Manage. 155, 81–95. Heyer, W.R., Donnelly, M.A., McDiarmid, R.W., Hayek, L.C., 1994. Measuring and Monitoring Biological Diversity: Standard Methods for Amphibians. Smithsonian Institution Press, Washington, D.C.. Hines, J.E., 2006. PRESENCE – Software to estimate patch occupancy and related parameters. USGS-PWRC. . Homyack, J.A., O’Bryan, C.J., Thornton, J.E., Baldwin, R.F., 2014. Anuran assemblages associated with roadside ditches in a managed pine landscape. For. Ecol. Manage. 334, 217–231. Homyack, J.A., O’Bryan, C.J., Thorton, J.E., Baldwin, R.F., 2016. Community occupancy of herpetofauna in roadside ditches in a managed pine landscape. For. Ecol. Manage. 361, 346–357. Jackson, C.R., 2014. Forestry best management practices: a mitigated water pollution success story. J. Forest. 112, 47–49. Jokela, E.J., Dougherty, P.M., Martin, T.A., 2004. Production dynamics of intensively managed loblolly pine stands in the southern United States: a synthesis of seven long-term experiments. For. Ecol. Manage. 192, 117–130. Jokela, E.J., Martin, T.A., Vogel, J.G., 2010. Twenty-five years of intensive forest management with southern pines: important lessons learned. J. Forest. 108, 338–347. Jones, P.D., Hanberry, B.B., Demarais, S., 2010. Managing the Southern pine forest – retained wetland interface for wildlife diversity: research priorities. Wetlands 30, 381–391. Leonard, P.B., Baldwin, R.F., Homyack, J.A., Wigley, T.B., 2012. Remote detection of small wetlands in the Atlantic coastal plain of North America: local relief models, ground validation, and high-throughput computing. For. Ecol. Manage. 284, 107–115. MacKenzie, D.I., Nichols, J.D., Hines, J.E., Knutson, M.G., Royle, A.D., 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84, 2200–2207. Mazerolle, M.J., 2004. Drainage ditches facilitate frog movements in a hostile landscape. Landscape Ecol. 20, 579–590. National Oceanic and Atmospheric Administration, 2014. National Climatic Data Center. NLCD 2011. Jin, S., Yang, L., Danielson, P., Homer, C., Fry, J., Xian, G., 2013. A comprehensive change detection method for updating the National Land Cover Database to circa 2011. Remote Sens. Environ. 132, 159–175. North Carolina Partners in Amphibian and Reptile Conservation. CASP: NC Calling Amphibian Survey Program, 2014.

North Carolina Wildlife Resources Commission. North Carolina Wildlife Action Plan. Raleigh, North Carolina, 2015. O’Bryan, C., Homyack, J., Harrison, A., Kanno, Y., Baldwin, R., 2016. Novel habitat use supports population maintenance in a reconfigured landscape. Ecosphere, in press. . Oseen, K.L., Wassersug, R.J., 2002. Environmental factors influencing calling in sympatric anurans. Oecologia 133, 616–625. R Core Team, 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. . Richter-Boix, A., Llorente, G.A., Montori, A., 2007. A comparative study of predatorinduced phenotype in tadpoles across a pond permanency gradient. Hydrobiologia 583, 43–56. Russell, K.R., Guynn Jr., D.C., Hanlin, H.G., 2002. Importance of small isolated wetlands for herpetofaunal diversity in managed, young growth forests in the Coastal Plain of South Carolina. For. Ecol. Manage. 163, 43–59. Rustigian, H.L., Santelmann, M.V., Schumaker, N.H., 2003. Assessing the potential impacts of alternative landscape designs on amphibian population dynamics. Landscape Ecol. 18, 65–81. Saenz, D., Fitzgerald, L.A., Baum, K.A., Conner, R.N., 2006. Abiotic correlated of anuran calling phenology: the importance of rain, temperature, and season. Herpetol. Monogr. 20, 64–82. Schultz, R.P., 1997. Loblolly pine. The ecology and culture of loblolly pine (Pinus taeda L.). Agricultural Handbook 713, U.S. Forest Service, Washington, D.C. Semlitsch, R.D., Bodie, J.R., 2003. Biological criteria for buffer zones around wetlands and riparian habitats for amphibians and reptile. Conserv. Biol. 17, 1219–1228. Semlitsch, R.D., Todd, B.D., Blomquist, S.M., Calhoun, A.J.K., Gibbons, J.W., Gibbs, J.P., Graeter, G.J., Harper, E.B., Hocking, D.J., Hunter, M.L., Patrick, D.A., Rittenhouse, T.A.G., Rothermel, B.B., 2009. Effects of timber harvest on amphibian populations: understanding mechanisms from forest experiments. Bioscience 59, 853–862. Skelly, D.K., Werner, E.E., Cortwright, S.A., 1999. Long-term distributional dynamics of a Michigan amphibian assemblage. Ecology 80, 2326–2337. Skelly, D.K., Bolden, S.R., Freidenburg, L.K., 2014. Experimental canopy removal enhances diversity of vernal pond amphibians. Ecol. Appl. 14, 340–345. Walls, S.C., Waddle, J.H., Faulkner, S.P., 2014. Wetland reserve program enhances site occupancy and species richness in assemblages of anuran amphibians in the Mississippi alluvial valley, USA. Wetlands 34, 197–207. Werner, E.E., Skelly, D.K., Relyea, R.A., Yurewicz, K.L., 2007. Amphibian species richness across environmental gradients. Oikos 116, 1697–1712. Weyrauch, S.L., Grubb, T.C., 2004. Patch and landscape characteristics associated with the distribution of woodland amphibians in an agricultural fragmented landscape: an information-theoretic approach. Biol. Conserv. 115, 443–450. Wigley, T.B., Baughman, W.M., Dorcas, M.E., Gerwin, J.A., Gibbons, J.W., Guynn, D.C., Lancia, R.A., Leiden, Y.A., Mitchell, M.S., Russell, K.R., 2000. Contributions of intensively managed forests to the sustainability of wildlife communities in the south. In: Sustaining Southern Forests: the Science of Forest Assessment. Southern Forest Resource Assessment.