Responses of a Federally Endangered Songbird to Understory Thinning in Oak-Juniper Woodlands
Wildlife conservation and management on military lands must be accomplished in the context of military readiness, which often includes ground-based training that is perceived to conflict with wildlife needs and environmental regulations. From 2008‒2012, we examined territory density, pairing success, and fledging success of the federally endangered golden-cheeked warbler (Setophaga chrysoparia; hereafter warbler) in relation to removal of small-diameter trees from the understory of mature oak-juniper (Quercus-Juniperus) woodland at the 87,890 ha Fort Hood Military Reservation in central Texas. Understory thinning created troop maneuver lanes, but left canopy vegetation intact. Warbler density, pairing success, and fledging success were similar across thinned and control sites. We found that warbler pairing and fledging success were best predicted by Ecological site (hereafter Ecosite), an indicator of hardwood tree species composition. Warbler pairing and fledging success were about 1.5 and 1.6 times higher, respectively, for territories dominated by the Low Stony Hill Ecosite than territories dominated by the Redlands Ecosite. Our results indicate that understory thinning for military training purposes did not have a negative effect on warblers at Fort Hood in the manner tested, and suggest that removal of smaller trees from the understory in a way that replicates historic conditions may elicit neutral responses from this forest-dependent songbird. Quantifying wildlife responses to military activities provides the Department of Defense and US Fish and Wildlife Service with data to guide conservation of threatened and endangered species on Department of Defense facilities while maintaining the military mission, and supports wildlife management efforts on other public and private lands.
KeywordsEcological site description Golden-cheeked warbler Military training Oak-juniper woodland Setophaga chrysoparia Understory thinning
As the largest landholder in the United States, the federal government owns and manages about 260 million ha of land in the country (Gorte et al. 2012). This includes about 7.9 million ha of land under the jurisdiction and management of the US Department of Defense (DoD) (Gorte et al. 2012). The 4127 DoD parcels range from <0.5 ha to 1.25 million ha and represent a wide variety of vegetation types that support high levels of biological diversity (Stein 2008; Gorte et al. 2012). The importance of DoD lands to sustain wildlife populations is well established, especially for species of conservation concern; 458 federally listed threatened and endangered species and 555 at-risk species occur across military defense sites, and over 40 of these species are found exclusively under DoD stewardship (Stein 2008; Stein et al. 2008; NatureServe 2015). Wildlife conservation and management activities on DoD lands are extensive, but must be accomplished in the context of military readiness, which often includes ground-based training that is perceived to conflict with wildlife needs and federal environmental regulations. Quantifying wildlife responses to military activities can provide the DoD and US Fish and Wildlife Service (USFWS) with reliable scientific data to guide conservation efforts for threatened and endangered species on DoD facilities while maintaining the military mission, and provides information to help guide wildlife management efforts on other public and private lands.
Fort Hood Military Reservation (hereafter Fort Hood) is a 87,890 ha Army installation located in central Texas that maintains infrastructure and training lands for ~ 50,000 military personnel from combat units and support teams. Fort Hood is host to a wide variety of anthropogenic disturbances, including artillery fire, dismounted and mechanized maneuvers, and aircraft gunnery, and provides an opportunity for military personnel to train in ecological conditions similar to those experienced by military personnel in the Middle East (personal communication with Integrated Training and Management Department at Fort Hood 2012). Fort Hood also supports breeding habitat for the federally endangered golden-cheeked warbler (Setophaga chrysoparia; hereafter warbler). This small insectivorous songbird breeds exclusively in oak-juniper woodland (Quercus spp.-Juniperus) in central Texas (Pulich 1976; Ladd and Gass 1999). Warbler specificity to this vegetation type reflects the species’ use of mature Ashe juniper (Juniperus ashei) bark for nesting material; warblers also use a combination of Ashe juniper, oaks, and other hardwoods as nesting and foraging substrates (Pulich 1976; Ladd and Gass 1999). Though early observers suggested the species requires highly closed canopy oak-juniper woodlands, more recent research found that warblers breed under a wider variety of canopy cover characteristics across their breeding range (e.g., Farrell et al. 2012; Klassen et al. 2012; Long 2014).
Like many military instillations, Fort Hood must balance wildlife conservation and Endangered Species Act compliance with competing military requirements (Tazik and Martin 2002; Boice 2006). One component of military training with the potential to impact warblers at Fort Hood is understory thinning of oak-juniper woodland to create infantry maneuver lanes. In coordination with Fort Hood, the Texas A&M Institute of Renewable Natural Resources conducted Ashe juniper removal from the understory of mature oak-juniper woodland and initiated a study to quantify the potential impacts of understory thinning activities on warblers. From previous research, we know that avian responses to understory thinning are species-specific. In Arkansas, for example, Rodewald and Smith (1998) found that understory thinning resulted in higher abundance of canopy nesting species and lower abundance of understory-nesting species. Artman (2003) also found that understory thinning in western hemlock (Tsuga heterophylla) enhanced habitat conditions for some birds, while having a minimal effect on other species. Though relatively few studies have examined the influence of understory thinning activities on reproductive parameters, results similarly suggest that avian responses are species-specific. For example, Eggers and Low (2014) found that understory thinning differentially influenced survival and reproduction of sympatric Parid species; when the forest understory was reduced, bird species associated with canopy vegetation had higher nest and adult survival than species associated with understory vegetation.
At Fort Hood, the goal of understory thinning activities was to create maneuver lanes by selectively removing small Ashe juniper from the understory of oak-juniper woodland with a combination of hydraulic shears and hand-cutting while leaving the overstory and hardwood components intact. Our goal was to examine warbler responses to this vegetation management practice. Specifically, we used data collected from 2008–2012 to compare warbler territory density, warbler pairing success, and warbler fledging success in thinned woodland and ecologically similar controls that received no understory thinning treatment. Because warblers are a canopy-dwelling species, we predicted that understory thinning would have no effect on warblers. However, we also hypothesized that warblers would elicit a positive response to understory thinning, as clearing of small Ashe juniper from the understory had the potential to functionally replicate natural disturbance events that remove vegetation from the understory of mature woodland (e.g., wildfire, windstorms, seasonal flooding), and other canopy dwelling bird species have shown positive responses to removal or thinning of the under and mid-story (USFWS 2003; Hagar and Friesen 2009). We knew from previous research on Fort Hood that warbler pairing and fledging success is higher in areas dominated by Texas oak (Q. buckleyi) when compared to post oak (Q. stellata) (Marshall et al. 2013). As such, we also hypothesized that warblers may respond differently given the dominant tree species within the site or territory. Our research contributes to DoD’s understanding of how vegetative management practices associated with military training activities influence warblers at Fort Hood, and provides information to help predict warbler responses to natural or anthropogenic disturbance on other public and private lands in Texas.
Materials and Methods
Study Area and Experimental Design
We conducted our research at five mature oak-juniper woodland study sites inhabited by warblers on the eastern portion of Fort Hood (Fig. 1). Each study site consisted of treated areas (i.e., sampled woodland that received understory thinning) and ecologically similar controls (i.e., sampled woodland with no understory thinning). In Training Area 11 (TA 11), this included two treated areas (34–78 ha) per year of our study and 4–6 control areas (45–146 ha) per year of our study. In Land Group 2 (LG 2), this included four treated areas per year of our study (30–102 ha) and four control areas per year of our study (88–123 ha). The initial locations of our study sites were determined by military activity and understory thinning efforts. However, a study conducted on Fort Hood by Marshall et al. (2013) indicated that warbler reproductive success varied by dominant hardwood species, which may be coarsely identified by Ecosite (i.e., Ecological Site Descriptions that identify distinctive land types with physical characteristics, such as soil and geologic conditions, which can influence vegetation assemblages; Creque et al. 1999). Specifically, Marshall et al. (2013) found that warblers had higher pairing and fledging success in the Low Stony Hill Ecosite, which is generally dominated by Texas oak, when compared to the Redlands Ecosite, which is dominated by post oak. Therefore, we also categorized our study sites and territories by Ecosite to account for the potential effects of vegetation composition on warbler territory density, pairing success, and fledging success in relation to understory thinning. We assigned each study site to an Ecosite classification based on the proportion of Low Stony Hill or Redlands within the sampled area (i.e., areas with >50% Low Stony Hill or >50% Redlands were classified as such). We similarly classified territories as Low Stony Hill or Redlands territories based on the proportion of each Ecosite within the territory boundaries. We also included a Steep Adobe Ecosite classification for analyses conducted at the territory scale, as this was another Ecosite occupied by warblers within treatments and controls at both of our study sites. Within the Steep Adobe Ecosite, slope and geologic structure have a strong influence on the composition of the plant community, with rocky slopes producing woody vegetation consisting mainly of trees such as Texas oak, Live oak, Texas redbud (Cercis canadensis), and Ashe juniper (USDA-NRCS 2011c).
In 2008 and 2009, we sampled pre- and post-treatment vegetation across each of our study sites scheduled for thinning treatments to quantify the change in vegetation structure. Within each treated area, we established grid points at 125-m spacing. From each grid point, we randomly established a 20-m line transect. At 1-m increments along the 20-m transects, we estimated the number of times a woody plant stem intersected with a 6 m retractable pole (i.e., hits) within each vegetation height class (0–1.5 m, >1.5–3.0 m, >3.0–4.5 m, and >4.5–6.0 m). We marked the beginning and end of each line transect with rebar so that we could relocate our vegetation sampling points following the understory thinning treatment. We calculated the mean number of pre- and post-treatment hits per vegetation height class and their associated standard deviations.
Warbler Territory Mapping and Monitoring
During the first 3 weeks of March 2008–2012, we conducted transect surveys every 7–10 days across each of the study sites to determine initial locations of warblers. Transects were spaced ≥75 m apart and transect length varied between 400 and 1700 m, depending on the size of the study site. Every 50 m along transects, we stopped for about 2 min to listen for singing male warblers. We recorded the location of detected males with Global Positioning System (GPS) handheld devices and recorded the direction and distance of all neighboring warblers. We did not include these initial detection points in our analyses because initial locations may shift somewhat during the earliest stages of territory establishment. Rather, we used these detection points to provide a basis for relocating male warblers to conduct territory mapping.
After identifying the general locations of warblers at our study sites, we returned to the sites once per week for the entire warbler breeding season (March–June) to map each male’s territory. This resulted in ≥10 total visits for each warbler territory. During territory mapping visits, we relocated males and recorded sequential GPS points on the focal male until the bird was no longer visible or was mapped for ≤1 h. From 2008–2010, we recorded GPS locations of male warblers if the focal male moved >20 m, and in 2011 and 2012, we recorded GPS locations every 2 min. We used data collected using both methods to create minimum convex polygons, which represent the outermost points in a location dataset, for each territorial male (i.e., male relocated for >4 weeks) in each year when >15 points were recorded over the course of the breeding season. We then used the territory delineations to estimate territory density (territories/ha) within the sampled areas at each study site. Given the large number of warblers within LG2, we were unable to monitor all males within the control areas. Therefore, we restricted our comparisons of mean territory density to TA11, where we were able to map and monitor all known warbler territories.
During territory mapping, we conducted behavioral surveys to determine warbler pairing and fledging success. We used a modified version of the Vickery index to assess warbler productivity (Vickery et al. 1992). The Vickery index is a method of estimating reproductive success that avoids potential biases associated with non-randomly collected nest data, limits disruption of nesting pairs, provides an opportunity to sample a larger geographic area in less time, and allows observers to predict territory outcomes (e.g. unpaired male, paired, fledged) using behavioral observations when nests are difficult to locate and monitor. We assume our productivity estimates are conservative and any error associated with sampling or observers was similar for all sites and in all years. We considered a male successfully paired if we detected a female warbler within his territory for ≥4 weeks. We considered a territory reproductively successful if ≥1 fledgling was located within a territory. Because we covered large study sites and the treatment application was not under our control, we could not color band a large proportion of breeding pairs. Thus, we took extreme care to properly link breeding outcome with a specific territory by repeatedly visiting each territory once fledglings were detected. If we could not determine the breeding status of a territory (e.g. inability to link fledglings with a specific territory), we excluded that territory from our productivity data set. Our sample unit for questions regarding warbler productivity was the territory and not the study site as described above for territory density. Therefore, we included all sampled territories in our statistical analyses for warbler pairing and fledging success.
We used paired t-tests to compare the mean number of hits recorded in each vegetation height category during pre- and post-thinning vegetation surveys. We used two-way analyses of variance (ANOVA) to test the interactive effects of year, treatment, and Ecosite on mean warbler territory density. Though our research question was specific to the influence of treatment and Ecosite on our warbler responses, we included year in all analyses because inter-annual variation in weather conditions can influence avian habitat selection and reproductive success (e.g., Morrison and Bolger 2002, Marshall et al. 2013). If there was no significant interaction between our variables, we examined the main effects separately, using an ANOVA for our comparison of mean warbler territory density across years and a paired t-test for our comparison of mean warbler territory density between treatments and Ecosites. When analyses of our warbler metrics indicated significant differences, we used Tukey’s HSD (honest significant difference) test for pairwise comparisons of means. We used α = 0.05 to demarcate statistical significance for our statistical tests, but acknowledge that α is arbitrary (Johnson 1999) and present follow-up analyses and discussion when α < 0.10.
Comparisons of the mean number of woody stem hits per vegetation height class before (Pre) and after (Post) vegetation thinning of treated study sites within Training Area 11 (TA11) and Land Group 2 (LG2) at the Fort Hood Military Reservation in Coryell and Bell counties, TX (2008‒2012). Means are presented with their associated standard deviations (SD)
TA 11 (n = 80)
LG 2 (n = 58)
We found statistically significant differences in the mean number of woody stem hits recorded in the understory (0.0–1.5 and 1.5–3.0 m height categories) between pre- and post-treatment vegetation in TA11 and LG2 (Table 1). Specifically, we found a 29–57% reduction in the 0.0–1.5 m category and a 14–32% reduction in the 1.5–3.0 m height category after the vegetation thinning treatment. Conversely, we found no statistically significant difference in the number of hits recorded in the mid-story (3.0–4.5 m) or canopy (4.5–6.0 m) at TA11 and LG2 (Table 1) after the vegetation thinning treatment, where there was only an 8–11% reduction in the mid-story and canopy at both study sites.
Model results for golden-cheeked warbler pairing and territory success at the Fort Hood Military Reservation located in Coryell and Bell counties, TX (2008–2012)
We found no evidence that understory thinning of mature Ashe juniper woodland conducted for military training purposes at Fort Hood Military Installation in central Texas had a negative effect on the federally endangered golden-cheeked warbler. Understory thinning treatments conducted in warbler habitat created troop maneuver lanes as intended, but left canopy vegetation intact. Warbler density, pairing success, and fledging success were similar across thinned and control sites. We found that warbler pairing and fledging success were best predicted by Ecosite, an indicator of hardwood tree species composition on Fort Hood, and not by understory thinning treatment; the predicted probably of warbler fledging success was about 1.6 times higher for territories dominated by the Low Stony Hill Ecosite when compared to territories dominated by the Redlands Ecosite.
It is possible, though unlikely, that the understory thinning treatment had negative effects on warbler demographic or population-level responses we did not measure, such as fledgling survival, breeding site fidelity, and rates of cowbird parasitism. Within primarily forested landscapes, a variety of silvicultural treatments have shown little or no effect on nest success or brood parasitism rates of forest birds (Annand and Thompson 1997; King and DeGraaf 2000; Powell et al. 2000) and often landscape context appears to have more influence on demographic parameters than does forest stand structure (Rodewald and Yahner 2001). However, understory thinning could have had a negative impact on other avian species that we did not study. For example, dense understories tend to benefit shrub-nesting forest birds, but may be unsuitable for ground-nesting species.
Because warblers are a canopy-dwelling species, our results are not unexpected. The thinning treatment at Fort Hood only involved removal of understory woody vegetation, thus leaving the canopy where warblers most often nest and forage intact (Pulich 1976; Ladd and Gass 1999). Studies indicate that silvicultural practices can modify forest structure and affect avian demographics (Thompson et al. 1995; Annand and Thompson 1997). However, other studies suggest that understory thinning may replicate natural disturbance events (e.g., windstorms, wildfires, and seasonal flooding), which can achieve many forestry objectives such as tree regeneration, forest restoration, fuel reduction, enhancement of biodiversity, and disease control (Franklin et al. 1997; Moore et al. 1999; Palik et al. 2003; Halpern et al. 2005; Atwell et al. 2008) while eliciting neutral or positive responses from birds associated with canopy vegetation (Rodewald and Smith 1998; Artman 2003; Eggers and Low 2014). Oak-juniper woodland in central Texas has undergone decades of fire exclusion and the understories have become dense, particularly with Ashe juniper (Fuhlendorf et al. 1996; Lyons 2005; Sparks et al. 2012). Thus, warbler habitat before Euro-American settlement likely had substantially less woody species in the understory compared to present day conditions. It is possible that understory thinning replicated historic conditions, thus resulting in limited variation in warbler responses across the treated and control sites.
Although our estimates for warbler territory density, pairing success, and fledging success were similar to other study sites across the warbler’s range (Groce et al. 2010), it is interesting to note that these metrics were consistently higher in the Low Stony Hill Ecosite compared to the Redlands Ecosite. This pattern is similar to results found by Farrell et al. (2012), who reported that warbler pairing and fledging success increased with increasing territory density. In addition, Marshall et al. (2013) noted a relationship between Ecosite and habitat quality, whereby warbler productivity was significantly higher in Low Stony Hill sites compared to Redlands sites, and the authors linked differences in productivity to arthropod density across tree species. The main difference in vegetation between the two Ecosites lies in the deciduous component (i.e., Low Stony Hill is predominated by Texas oak and Redlands is predominated by post oak). However, there also appears to be a major historical difference between the Ecosites. Low Stony Hill and Steep Adobe sites are typically associated with hillsides and slopes, making them less prone to frequent fires, and thus more likely to contain a significant Ashe juniper component (USDA-NRCS 2011a). Conversely, Redlands sites are typically found at lower elevations and are not typically associated with slopes, which means they were more prone to frequent fires, and thus historically contained little to no Ashe juniper (USDA-NRCS 2011b). In the absence of fire, woody species have encroached into post oak dominated woodland that was historically open and park-like (Diggs et al. 2006), again, likely creating warbler habitat conditions that were not present prior to Euro-American settlement. While we were unable to examine the interaction between treatment and Ecosite for warbler pairing and fledging success, our analyses indicated that mean warbler territory density was lowest in Redlands controls and we found no differences in mean territory density when comparing the Redlands treatments to the Low Stony Hill treatments or controls. This supports the hypothesis that understory vegetation modification that mimics historic disturbance may benefit warblers inhabiting certain types of oak-juniper woodland. However, it is also possible that unmeasured site-specific factors influenced our results. More research is needed to examine this potential relationship across a larger number of sites, and studies should concurrently examine other demographic parameters important to warbler population dynamics.
Quantifying potential impacts of human disturbance on wildlife provides data to guide conservation efforts for threatened and endangered species. This is especially important on DoD land where species management must be conducted while maintaining the military mission. We used an impact assessment to investigate the effects of understory thinning on federally endangered warblers at the Fort Hood Military Reservation in central Texas. We found no evidence to suggest that understory thinning of mature oak-juniper woodland to create troop maneuver lanes had a negative impact on this species, which has implications for on-post military training capabilities.These results are also encouraging in the context of species recovery, ecosystem sustainability, multi-purpose land use objectives, and wildfire safety concerns, among others, as removal of trees from the understory in a way that replicates historic conditions may elicit neutral or, in some cases positive, responses from this forest-dependent songbird. We encourage more quantitative studies addressing the potential impacts of understory thinning practices on warblers and other species when vegetation manipulation is required, the results of which will help guide more effective management efforts on public and private lands.
We thank the US Army Environmental Command (Fort Hood Military Reservation) for funding and the research team at the Texas A&M Institute of Renewable Natural Resources for logistical support. We also thank Justin Tatum and Reynaldo Navarro for assisting with study site access and guidance applicable to our research objectives. Andy James provided comments on earlier drafts and many field technicians contributed to data collection.
Compliance with Ethical Standards
Conflict of Interest
The authors declare no Conflict of Interest.