Journal list menu
Impacts of umbrella species management on non-target species
Abstract
- Restoration of anthropogenically altered habitats has often focused on management for umbrella species—vulnerable species whose conservation is thought to benefit co-occurring species. Woody plant encroachment is a form of habitat alteration occurring in grasslands and shrublands around the globe, driven by anthropogenic shifts in disturbance regimes. Conifer encroachment is a pervasive threat to historically widespread sagebrush communities, as trees outcompete sagebrush and can negatively affect sagebrush-obligate animal species. Degradation and loss of sagebrush plant communities in western North America have been associated with drastic declines in wildlife populations. The imperilled Greater Sage-Grouse is assumed to be an umbrella species for the sagebrush community, so habitat restoration, including removal of encroaching conifers, is commonly targeted towards sage-grouse. How this conservation action affects the demography of species other than sage-grouse is largely unknown.
- We quantified the demographic effects of landscape-level restoration of sagebrush communities through conifer removal on an assemblage of sagebrush-obligate, shrubland generalist and woodland-associated songbirds. We compared songbird density and reproduction between adjacent restored and uncut conifer-encroached sagebrush plots in southwest Montana. We found and monitored nests to record nest fate and number of offspring produced.
- We found demographic benefits for sagebrush-obligate species in restored areas. Sage Thrashers colonized restored areas. Brewer's Sparrow density was 39% higher and nest success was 63% higher in removal treatments, resulting in 119% higher fledgling production compared with uncut areas. The density of Vesper Sparrows, a shrubland generalist, was 308% higher and fledgling production was 660% higher in restored areas.
- Another shrubland generalist, the White-crowned Sparrow, experienced 55% lower density and 37% lower fledgling production in conifer removal areas. Two woodland-associated species, Chipping Sparrows and Dark-eyed Juncos, were nearly extirpated following conifer removal. A third woodland associate, the Green-tailed Towhee, experienced 57% lower density and 69% lower fledgling production in removal than non-removal areas.
- Synthesis and applications. Our study demonstrates the benefits of conifer removal for sagebrush-obligate species, while highlighting species that may be sensitive. Umbrella species management can benefit co-occurring species with similar habitat associations, but demographic analyses for all impacted species are essential for effective conservation.
1 INTRODUCTION
Humans are increasingly driving habitat alteration through actions such as clearcutting, burning, urban development and introduction of invasive species (Haddad et al., 2015; Molnar et al., 2008). Habitat alteration and loss have driven dramatic biodiversity declines, loss of ecosystem services and decreases in many wildlife populations (Bender et al., 1998; Ceballos et al., 2017). As a result, conservationists have attempted to restore ecosystems and reverse species declines. Due to the complexity of ecosystem restoration, efforts are often focused on conservation shortcuts, such as management for ‘umbrella species’, which are particularly vulnerable and thought to reflect the benefits of conservation action for co-occurring species (Caro & O'Doherty, 1999; Roberge & Angelstam, 2004; Wilcox, 1984). Species with the greatest range-wide spatial overlap with the umbrella species are often assumed to benefit from umbrella-based conservation action, and the impacts are typically measured through metrics of occupancy or abundance (Branton & Richardson, 2011; Crosby et al., 2015; Rich et al., 2005; Rowland et al., 2006). However, this approach may overlook impacts on species with less spatial overlap and may obscure unintended demographic consequences, especially at the edges of umbrella species' ranges (Carlisle et al., 2018; Caro, 2003; Seddon & Leech, 2007). Thus, an in-depth assessment of the impact of umbrella species management requires an investigation into the demography of both umbrella and non-target species. Detailed studies into both abundance and reproductive impacts of umbrella species management for all co-occurring species are key to ensuring the success of restoration.
One broad-scale example of habitat alteration is woody plant encroachment into grasslands, tundra and shrublands across the globe (Myers-Smith et al., 2011; Nackley et al., 2017; Stevens et al., 2017). Native tree and shrub ranges are expanding due to anthropogenic fire suppression, overgrazing and climate change (Archer et al., 2017; Donohue et al., 2013; Miller & Rose, 1999). In western North America, active fire suppression over many decades has led to native conifers expanding into almost 150,000 km2 of rangelands in the past three decades (Miller et al., 2005, 2008; Morford et al., 2022; Figure 1a). Conifer expansion, along with the proliferation of invasive annual grasses and continuing sagebrush removal for ranching, agriculture, and urban development have led to historically widespread sagebrush shrublands becoming one of the most threatened ecosystems in North America (Doherty et al., 2022; Noss et al., 1995). Many wildlife species associated with sagebrush communities have experienced steep population declines due to habitat loss, especially Greater Sage-Grouse (Centrocercus urophasianus; Connelly et al., 2011; Davies et al., 2011; Figure 1c). Consequently, sage-grouse are considered an umbrella species for the sagebrush biome and they have received significant management and research focus (Connelly et al., 2011; Rowland et al., 2006; Runge et al., 2019; Timmer et al., 2019).
Conifer encroachment is one of the causes thought to underlie sage-grouse population declines, so management has focused on conifer removal as a strategy to benefit sage-grouse populations (Baruch-Mordo et al., 2013; Sandford et al., 2017; Severson, Hagen, Tack, et al., 2017). Conifer removal results in increased nest success, survival and population growth rates for sage-grouse (Olsen, Severson, Maestas, et al., 2021; Severson, Hagen, Tack, et al., 2017), and is assumed to provide similar benefits for other sagebrush-associated species (Donnelly et al., 2017; Reinhardt et al., 2020). However, only abundance responses to conifer removal have been quantified in many cases (Bombaci & Pejchar, 2016; but see Bergman et al., 2014). Sagebrush-obligate species have demonstrated increases in abundance following restoration through conifer removal, while some woodland-associated species have decreased (Donnelly et al., 2017; Holmes et al., 2017; Howard et al., 1987; Knick et al., 2014, 2017; Reinkensmeyer et al., 2007; Van Lanen et al., 2023a; Young & Johnson, 2024). While abundance often appears to align with species' habitat associations (e.g., woodland vs. shrubland), it is not always a reliable indicator of habitat quality (Johnson, 2007; Van Horne, 1983; but see Bock & Jones, 2004). Demographic rates are the most reliable way to assess the impacts of restoration efforts (Grant et al., 2017; Kramer et al., 2019), so we quantified species reproductive success relative to conifer removal. Reproduction may positively correlate with abundance among habitat types and reflect adaptive habitat preferences (Martin, 1998). However, reproductive success has been observed to negatively correlate with abundance in some habitats, including sagebrush (Chalfoun & Martin, 2007; Chalfoun & Schmidt, 2012; Misenhelter & Rotenberry, 2000). Thus, the study of the demographic consequences of conifer removal is necessary for ensuring the effective restoration of sagebrush habitats.
Associations of abundance and reproductive success with habitat variation due to conifer removal are particularly important to understand because of the transitional nature of conifer encroachment. Conifer encroachment occurs at the border between sagebrush and forested habitats and will transition to forest without conifer removal or natural disturbance in many cases (Miller et al., 2005). Complicated and shifting edge dynamics can result in increased uncertainty about patterns of abundance and reproductive success (Kristan et al., 2003; LaManna et al., 2015). For example, abundance patterns of shrubland-generalist or edge-associated species are less clear across the ecotone than those of sagebrush-obligate or woodland-associated species (Hamilton et al., 2019; Magee et al., 2019; Noson et al., 2006). Most importantly, the potential for abundance to negatively correlate with reproductive success can be especially problematic in habitat transition zones (Chalfoun & Schmidt, 2012; Gates & Gysel, 1978; Ries et al., 2017). An unexamined possibility is that conifer removal may encourage shrubland-associated wildlife to settle in removal areas, but suffer poor reproduction due to high predator densities in adjacent conifer forests (Coates et al., 2017; Tewksbury et al., 1998). Thus, examination of both abundance and reproductive success is imperative to understand the effectiveness of umbrella species conservation approaches.
Songbirds provide an important test case of umbrella species conservation strategies because they use a wide range of habitats, span a range of habitat specialists to generalists and often vary widely in their spatial overlap with potential umbrella species (Crosby et al., 2015; Knick et al., 2017; Micheletti et al., 2023). In the case of conifer removal for sage-grouse habitat restoration, songbirds ranging from sagebrush obligates to woodland associates may fall under the sage-grouse umbrella based on overlapping habitat use. However, despite the overlap, species with varying habitat preferences may differ in their response to conifer removal (Rowland et al., 2006; Zeller et al., 2021).
Like sage-grouse, sagebrush-obligate songbirds such as Brewer's Sparrows (Spizella breweri) and Sage Thrashers (Oreoscoptes montanus) have experienced broad population declines associated with habitat loss and have been listed as species of conservation concern in many western states (Davies et al., 2011; MTNHP, 2021; Rich et al., 2005; Sauer et al., 2019; Figure 1d). Other generalist and woodland-associated songbirds that use sagebrush, such as Green-tailed Towhees (Pipilo chlorurus) and Chipping Sparrows (Spizella passerina), are also declining in some areas of their ranges (Rich et al., 2005; Sauer et al., 2019; Figure 1d). The demographic consequences of conifer removal for the songbird community are unknown but understanding them is essential for making targeted conservation decisions. To understand the effect of conifer removal on sagebrush songbirds, we used a large-scale experiment to compare species density, nest success and reproductive productivity between removal and non-removal areas. We calculated the number of fledglings produced per hectare to quantify the impact of habitat restoration on declining species. We compared eight species that occur in conifer-encroached sagebrush and exhibit varying amounts of range overlap with Greater Sage-Grouse (Rowland et al., 2006). Given the overlap, these species should fall under the sage-grouse umbrella. However, we predicted that species with similar habitat associations to Greater Sage-Grouse would benefit more from conifer removal than woodland-associated species.
2 MATERIALS AND METHODS
2.1 Study site
We conducted this study in the Medicine Lodge Valley in southwestern Montana, USA (Figure 1a) in May–August of 2019–2022. Breeding occurred almost exclusively in June and July. The elevation at the site ranged from 2000 to 2300 m a.s.l. The vegetation communities in this area consisted of mountain big sagebrush (Artemisia tridentata vaseyana) encroached primarily by Douglas fir (Pseudotsuga menziesii) and juniper (Juniperus spp.) with additional woody vegetation present along riparian corridors. Some areas in the early stages of conifer encroachment containing relatively small, scattered conifers were removed in 2017–2018. Other areas, ranging from early to middle stages of encroachment, were left untreated in anticipation of this project. In removal areas, conifers were hand-felled, leaving sagebrush cover intact and allowing a more accurate comparison between treatment and control than prescribed burns. We selected six plots in treatment areas and six in control, paired as closely as possible by elevation, slope, aspect, and similarity of sagebrush height and vigour. Plots were largely adjacent to one another and averaged 55 ha (Figure 1b). The total area monitored was 674 ha.
All (eight) of the reasonably abundant species that occurred in removal and non-removal treatments and often nested in sagebrush habitat were selected as study species. Riparian-associated species were excluded from the analysis. Study species included sagebrush obligates, shrubland generalists and woodland-associated species. Sagebrush obligates were Brewer's Sparrows and Sage Thrashers. Shrubland generalists included Vesper Sparrows (Pooecetes gramineus) and White-crowned Sparrows (Zonotrichia leucophrys) and woodland-associated species included Green-tailed Towhees, Chipping Sparrows, American Robins (Turdus migratorius) and Dark-eyed Juncos (Junco hyemalis). Green-tailed Towhees were of special interest because they are often termed a near-obligate of sagebrush habitats (Dobbs et al., 2020; Noson et al., 2006). Green-tailed Towhees are associated with open forest habitats with a shrubby understorey, similar to conifer-encroached sagebrush areas (Dobbs et al., 2020; Knick et al., 2003; Martin, 1998; Pavlacky & Anderson, 2004), and are a species of management concern in Montana due to population declines (MTNHP, 2021). Other conifer-dependent species of concern, such as Pinyon Jays (Gymnorhinus cyanocephalus), were not present at the study site (Tack et al., 2023).
All field surveys and bird handling were conducted under Montana Fish Wildlife and Parks permits 2019-010-W, 2020-008-W, 2021-021-W and 2022-033-W, US Fish and Wildlife Service permit MB791101-0, US Geological Survey federal bird banding permit 21,635, and University of Montana Institutional Animal Care and Use Committee Animal Use Proposals 061-18 and 046-20.
2.2 Vegetation cover
We analysed vegetation cover data to confirm differences in tree cover and to determine whether other vegetation types varied significantly between treatment and control plots. We downloaded per cent cover data of trees, shrubs, litter, forbs and grass, and bare ground from the Rangeland Analysis Platform (RAP; Jones et al., 2018) across the study site. RAP calculates fractional cover data annually within 30 m pixels based on Landsat and other remotely sensed datasets (Allred et al., 2021). We overlaid a 50 m grid of points across all plots and downloaded vegetation cover data each year from 2019 to 2021 in a 25 m buffer around each grid point. We then summarized mean per cent cover for each plot in each year. We combined annual and perennial forbs and grasses into one category. Invasive grasses (i.e., Bromus tectorum) were not present at our study site.
2.3 Density
To understand how treatment impacted breeding pair density, we used spot-mapping to create territory maps in each year of the study (Chalfoun & Martin, 2007; Ralph et al., 1993). We surveyed all plots three to five times throughout the breeding season and mapped every bird seen and heard. The 50 m grid of points was used to assist with the accurate identification of bird locations using a GPS unit and visual distance estimation. We paid attention to males who were counter-singing to determine territory boundaries. Using the spot-mapping observations and nest locations, we created a territory map for each species each year of the study. We assumed that every territory held a pair of birds and if a territory overlapped the plot boundary; then, it was considered half of a territory. We used the final territory maps to estimate the number of pairs of each species on each plot and divided by plot area to calculate density.
2.4 Nest success and offspring production
To estimate reproductive success and offspring production between treatments, we located nests by observing parental behaviour and following individuals to their nests, as well as systematic searching (Martin & Geupel, 1993). Once nests were found, we monitored them every 1–3 days, recording nest contents at each visit. Nests were considered failed if the nest was empty two or more days before the end of the mean nestling period and no evidence of fledglings was detected. Nests were considered successful if at least one nestling fledged. We recorded egg number when the final clutch size was confirmed on two or more consecutive visits. We recorded the number of eggs hatched only if all young were accounted for and none disappeared at hatch. We recorded the number of fledglings based on nestling count in the nest within 2 days of average fledging age and evidence of fledging was observed (i.e., fledglings seen or heard, adults carrying food in the vicinity or nest edge flattened).
To determine whether more nesting attempts were possible in one of the treatment types due to a longer active breeding period, we calculated mean season length for each species. We took the average of the earliest and the latest three Julian initiation dates for each species in each year to account for outliers and subtracted to calculate season length. Finally, to understand how removal treatments may impact population dynamics and impact declining species, we calculated the number of fledglings produced per hectare.
2.5 Statistical analysis
We conducted all analyses in Program R (version 4.3.0), using packages nlme and lme4 for modelling (Bates et al., 2015; Pinheiro et al., 2022; R Core Team, 2023). To assess the treatment impact on tree cover, we ran a linear model of tree cover with treatment as a fixed effect and plot and year as random effects. Then, to determine whether vegetation other than tree cover differed between treatments, we checked normality and collinearity of the cover variables and ran a MANOVA with shrub, litter, bare ground, and grass and forb cover as dependent variables and treatment as the independent variable. Finally, to determine which vegetation types differed between treatments, we ran a linear model for each cover type with treatment as a fixed effect and year and plot as random effects.
We used species-specific mixed effects models to assess the effect of treatment on various dependent variables, including density, nest success, season length, number of eggs laid, number of eggs hatched, nestlings fledged and fledglings produced per hectare. Each model included treatment as the only fixed effect. We used linear models to test for a treatment effect, so we could account for temporal and spatial variation using random effects of year and plot. Species abundance varied between years (Table 1) and was measured at the plot level, so models of species density included only year as a random effect. For analysing nest success, we used a logistic-exposure analysis with nest fate as the dependent variable (1 = success, 0 = fail) and exposure days accounted for in the logistic-exposure link function (Shaffer, 2004). We expected that varying species abundance between years and spatial autocorrelation may impact nest success, so we tested random effects of year and plot. We used the Akaike information criterion to select the top model and calculated daily nest success by taking the inverse logit of beta estimates from the models. For the remaining models of offspring number, we assumed similar spatial and temporal impacts and used the same random effects structure as used in nest success models. For all mixed models, we calculated degrees of freedom using the Satterthwaite method (Satterthwaite, 1941) in the jtools package (Long, 2022).
| Habitat association | Common name | Species code | Treatment | Nests found | ||||
|---|---|---|---|---|---|---|---|---|
| 2019 | 2020 | 2021 | 2022 | Total | ||||
| Sagebrush-obligate | Brewer's Sparrow | BRSP | Removal | 27 | 124 | 120 | 30 | 301 |
| Non-removal | 5 | 69 | 59 | 15 | 148 | |||
| Sage Thrasher | SATH | Removal | 2 | 8 | 4 | 5 | 19 | |
| Non-removal | 0 | 0 | 0 | 0 | 0 | |||
| Shrubland generalist | Vesper Sparrow | VESP | Removal | 16 | 50 | 12 | 14 | 92 |
| Non-removal | 12 | 9 | 1 | 1 | 23 | |||
| White-crowned Sparrow | WCSP | Removal | 9 | 34 | 16 | 23 | 82 | |
| Non-removal | 19 | 50 | 49 | 11 | 129 | |||
| Woodland associated | American Robin | AMRO | Removal | 0 | 6 | 6 | 5 | 17 |
| Non-removal | 4 | 8 | 11 | 5 | 28 | |||
| Green-tailed Towhee | GTTO | Removal | 1 | 4 | 4 | 7 | 16 | |
| Non-removal | 3 | 5 | 6 | 11 | 25 | |||
| Dark-eyed Junco | DEJU | Removal | 1 | 1 | 0 | 0 | 2 | |
| Non-removal | 3 | 16 | 1 | 5 | 25 | |||
| Chipping Sparrow | CHSP | Removal | 1 | 1 | 0 | 2 | 4 | |
| Non-removal | 22 | 28 | 8 | 21 | 79 | |||
Finally, to estimate the number of fledglings produced per hectare, we first calculated entire-period nest success by raising daily nest success to the power of the average length of the entire period of each nest (Mayfield, 1961) for each species in both treatments. To capture the variability of fledglings per hectare, we bootstrapped 1000 iterations of the density dataset in each treatment for each species and took their means. Then, we multiplied density, entire-period nest success and number of fledglings per successful nest to calculate the number of fledglings produced per hectare. Finally, we ran linear models with treatment as a fixed effect and the same random effects as nest success models.
3 RESULTS
3.1 Vegetation cover
Tree cover was 6.16 percentage points lower in removal versus non-removal plots (95% CI [−8.23, −4.09], p < 0.01, Figure 2). Vegetation cover excluding trees still differed significantly between plots (F(1,50) = 9.34, p < 0.001), but some values were strongly correlated (Pearson correlation coefficient litter: bare = 0.74, Litter: grass and forb = −0.58). Additional linear models accounting for year and plot showed that forb and grass cover was 7.01 percentage points higher in removal than non-removal plots (95% CI [5.05, 8.96], p < 0.01), but litter cover was 1.07 percentage points lower (95% CI [−1.63, −0.52], p < 0.01). Shrub and bare ground cover did not differ between treatments (shrub model: βrem = 0.58, 95% CI [−1.17, 2.32], p = 0.52; bare model: βrem = 0.04, 95% CI [−0.56, 0.63], p = 0.91).
3.2 Density
Sagebrush-obligate species were significantly denser in restored, conifer removal areas than in non-removal, encroached areas despite fluctuating density between years (Tables 1 and 2). Brewer's Sparrows abundance was almost 40% greater when trees were removed, and Sage Thrashers were absent from areas with trees present (Table 3; Figure 3). Vesper Sparrows demonstrated the same pattern as the sagebrush obligates and were 308% more dense where conifers were removed (Table 3; Figure 3). The other shrubland generalist, the White-crowned Sparrow, was 55% less dense where trees were removed (Table 3; Figure 3). Finally, all the woodland-associated species except American Robins were more abundant where trees were still present (Table 3). Green-tailed Towhees were 57% denser, Chipping Sparrows were 94% denser, and Dark-eyed Juncos were 97% denser where trees remained.
| Species | Model | K | AICc | ΔAICc | Weight | Loglik |
|---|---|---|---|---|---|---|
| BRSP | Trt + (1|Year) | 3 | 781.03 | 0.00 | 0.54 | −387.49 |
| Trt + (1|Year) + (1|Plot) | 4 | 782.06 | 1.03 | 0.32 | −386.98 | |
| Trt | 2 | 784.53 | 3.50 | 0.09 | −390.26 | |
| Trt + (1|Plot) | 3 | 786.33 | 5.30 | 0.04 | −390.14 | |
| VESP | Trt | 2 | 231.92 | 0.00 | 0.56 | −113.96 |
| Trt + (1|Year) | 3 | 234.06 | 2.14 | 0.19 | −113.92 | |
| Trt + (1|Plot) | 3 | 234.13 | 2.21 | 0.18 | −113.96 | |
| Trt + (1|Year) + (1|Plot) | 4 | 236.21 | 4.29 | 0.07 | −113.92 | |
| WCSP | Trt | 2 | 417.63 | 0.00 | 0.74 | −206.81 |
| Trt + (1|Year) | 3 | 419.70 | 2.07 | 0.26 | −206.79 | |
| Trt + (1|Plot) | 3 | Fails to converge | ||||
| Trt + (1|Year) + (1|Plot) | 4 | Fails to converge | ||||
| AMRO | Trt + (1|Plot) | 3 | 112.59 | 0.00 | 0.45 | −53.00 |
| Trt | 2 | 113.25 | 0.66 | 0.32 | −54.62 | |
| Trt + (1|Year) + (1|Plot) | 4 | 115.01 | 2.41 | 0.14 | −53.00 | |
| Trt + (1|Year) | 3 | 115.84 | 3.25 | 0.09 | −54.63 | |
| GTTO | Trt | 2 | 62.57 | 0.00 | 0.62 | −29.28 |
| Trt + (1|Year) | 3 | 65.22 | 2.65 | 0.17 | −29.28 | |
| Trt + (1|Plot) | 3 | 65.22 | 2.65 | 0.17 | −29.28 | |
| Trt + (1|Year) + (1|Plot) | 4 | 67.68 | 5.11 | 0.05 | −29.28 | |
| CHSP | Trt | 2 | 164.10 | 0.00 | 0.57 | −80.05 |
| Trt + (1|Year) | 3 | 166.40 | 2.30 | 0.18 | −80.05 | |
| Trt + (1|Plot) | 3 | 166.40 | 2.30 | 0.18 | −80.05 | |
| Trt + (1|Year) + (1|Plot) | 4 | 168.61 | 4.51 | 0.06 | −80.05 | |
- Abbreviation: AIC, Akaike information criterion.
| Species | Model | Random effects | Conifer removal | |||
|---|---|---|---|---|---|---|
| β | SE | t-Value | p | |||
| BRSP | Density | Year | 0.160 | 0.060 | 2.682 | 0.01 |
| Nest success | Year | 0.573 | 0.165 | 3.473* | <0.01 | |
| Eggs laid | Year | −0.031 | 0.070 | −0.45 | 0.65 | |
| # eggs hatched | Year | 0.010 | 0.122 | 0.083 | 0.93 | |
| # fledged | Year | 0.191 | 0.138 | 1.381 | 0.17 | |
| Fledglings per ha | Year | 0.418 | 0.002 | 253.9 | <0.01 | |
| SATH | Density | None | 0.023 | 0.005 | 5.136 | <0.01 |
| VESP | Density | Year | 0.152 | 0.029 | 5.196 | <0.01 |
| Nest success | None | 0.406 | 0.383 | 1.062* | 0.29 | |
| Eggs laid | None | −0.097 | 0.170 | −0.567 | 0.57 | |
| # eggs hatched | None | 0.158 | 0.277 | 0.569 | 0.57 | |
| # fledged | None | −0.286 | 0.391 | −0.730 | 0.47 | |
| Fledglings per ha | None | 0.153 | 0.002 | 94.33 | <0.01 | |
| WCSP | Density | Year | −0.173 | 0.050 | −3.485 | <0.01 |
| Nest success | None | 0.248 | 0.248 | 1.0* | 0.32 | |
| Eggs laid | None | −0.025 | 0.111 | −0.227 | 0.82 | |
| # eggs hatched | None | 0.023 | 0.159 | 0.146 | 0.88 | |
| # fledged | None | 0.398 | 0.211 | 1.885 | 0.06 | |
| Fledglings per ha | None | −0.139 | 0.002 | −68.93 | <0.01 | |
| AMRO | Density | None | −0.014 | 0.011 | −1.354 | 0.18 |
| Nest success | Plot | 0.214 | 0.756 | 0.283* | 0.78 | |
| Eggs laid | None | 0.233 | 0.200 | 1.17 | 0.25 | |
| # eggs hatched | None | 0.125 | 0.307 | 0.408 | 0.69 | |
| # fledged | None | −0.286 | 0.430 | −0.664 | 0.52 | |
| Fledglings per ha | None | −0.009 | 0.0001 | −72.29 | <0.01 | |
| GTTO | Density | Year | −0.034 | 0.009 | −3.825 | <0.01 |
| Nest success | None | −0.476 | 0.598 | -0.797* | 0.43 | |
| Eggs laid | None | 0.193 | 0.302 | 0.639 | 0.53 | |
| # eggs hatched | None | 0.500 | 0.545 | 0.918 | 0.37 | |
| # fledged | None | −0.194 | 0.514 | −0.378 | 0.71 | |
| Fledglings per ha | None | −0.057 | 0.0004 | −126.5 | <0.01 | |
| CHSP | Density | Year | −0.221 | 0.031 | −7.071 | <0.01 |
| Nest success | None | −0.400 | 0.860 | -0.465* | 0.64 | |
| Eggs laid | None | −0.027 | 0.457 | −0.06 | 0.95 | |
| # eggs hatched | None | −0.027 | 0.940 | −0.029 | 0.98 | |
| # fledged | None | 0.114 | 1.007 | 0.113 | 0.91 | |
| Fledglings per ha | None | −0.247 | 0.002 | −117.0 | <0.01 | |
| DEJU | Density | Year | −0.109 | 0.015 | −7.159 | <0.01 |
- * z-values.
3.3 Nest success
Only Brewer's Sparrows experienced a significant difference in daily nest success between removal and non-removal areas (Table 3; Figure 4). Daily nest success was 2.3% higher in removal areas compared with non-removal areas, which accumulates to a 63% higher likelihood of nest success across the nesting period (Table 3; Figure 4). Sage Thrashers avoided areas where conifers remained, so we were not able to find any nests in non-removal areas. Dark-eyed Juncos and Chipping Sparrows occurred at such low densities in removal areas that we were unable to calculate nest success for the former and our estimate was based on very few nests for the latter (Table 1).
We found no differences in nesting season length between treatments for any of the species that we had sufficient sample sizes for estimation (AMRO: βrem = 1.81 ± 8.90, p = 0.85; BRSP: βrem = 9.45 ± 7.51, p = 0.25; VESP: βrem = −4.73 ± 5.92, p = 0.47; WCSP: βrem = 0.47 ± 5.13, p = 0.93). Thus, we assumed that each species had relatively equal numbers of nesting attempts in the two treatments throughout the season.
3.4 Offspring production
The number of eggs laid or hatched did not differ between treatments for any species we considered. Most species also did not differ in the number of fledglings produced from successful nests, but White-crowned Sparrows fledged slightly more offspring in removal areas than in non-removal areas (Table 3; Figure 5).
While nest success differences were difficult to test for species lacking pairs and nests in one treatment type, the reproductive consequences of such differences were clearer when based on fledgling production. Indeed, we found that mean number of fledglings produced per hectare differed between treatments for all species and was strongly impacted by species density (Table 3; Figure 6). Brewer's Sparrows produced 119% more fledglings in restored areas compared with encroached areas, while Vesper Sparrows produced 660% more. The remaining species produced fewer fledglings in restored areas. White-crowned Sparrows produced 37% fewer and American Robins produced 36% fewer fledglings where trees were removed. Green-tailed Towhees produced 69% fewer, and Chipping Sparrows produced 96% fewer fledglings in removal areas compared with non-removal areas (Table 3; Figure 6).
4 DISCUSSION
Under a conifer removal restoration plan, the Greater Sage-Grouse appears to be an appropriate umbrella for species with similar habitat associations but provides minimal benefit for species with increasing reliance on woodland habitats. Sagebrush-obligate and grassland-associated species benefitted from conifer removal, while species associated with open forest habitat had fewer, if any benefits. Unsurprisingly, species most closely associated with woodland habitats were negatively impacted following conifer removal (Figure 7). Our demographic analyses confirm and quantify beneficial management outcomes for declining sagebrush-obligate species. These differences between woodland and shrubland species emphasize the importance of assessing the varying impacts of umbrella species management.
Conifer removal resulted in similar benefits for sagebrush-obligate songbirds as have been seen for sage-grouse. Conifer removal created additional nesting habitat for Sage Thrashers, which were completely absent from encroached areas. Similarly, sage-grouse are unlikely to lek or nest in areas with as little as 4% tree cover (Baruch-Mordo et al., 2013; Severson, Hagen, Maestas, et al., 2017a), so conifer removal increases suitable habitat (Olsen, Severson, Allred, et al., 2021; Severson, Hagen, Maestas, et al., 2017b). Higher nest success for Brewer's Sparrows in removal areas mirrors increases seen in sage-grouse nest success (Olsen, Severson, Maestas, et al., 2021; Severson, Hagen, Tack, et al., 2017). Similar responses to conifer removal between umbrella and non-target species may be a result of high predation risk where trees are present. Trees in open sagebrush are thought to provide perch sites for predators and nearby conifer forests may provide a source of additional predators (LaManna et al., 2015; Tewksbury et al., 1998; Young et al., 2023). However, the landscape context around removal treatments may also affect species responses. For example, whether conifer removal creates an abrupt or gradual edge with conifer forest can alter predator behaviour and abundance (Angelstam, 1986; Batáry & Báldi, 2004; Söderström et al., 1998; Thompson & Burhans, 2003), as well as songbird settling patterns (Gates & Gysel, 1978; With & King, 2001). Removals at our study site were done such that a relatively gradual edge remained (Figure 1b), so further study into abrupt edges is warranted.
Predictably, shrubland generalist density and fledgling production differed between treatments in directions that reflected their habitat associations. Vesper Sparrows typically occur in open grasslands and shrublands and were denser where trees were removed. White-crowned Sparrows use a wide variety of shrubby habitats and forest edges and were denser in non-removal areas. Interestingly, despite lower density, White-crowed Sparrows fledged more young from successful nests in removal areas. Given that the number of eggs laid and nestlings hatched did not differ, White-crowned Sparrows must experience higher rates of partial brood loss in non-removal areas. Partial brood loss can occur from partial predation or from starvation of the smallest nestling (Mock, 1994; Ricklefs, 1965; Thompson et al., 1999). Thus, White-crowned Sparrows are likely experiencing higher partial predation or lower food availability in non-removal areas. However, despite fledging more young per nest in removal areas, density was more important for determining fledgling production per hectare. Despite potential localized negative impacts to White-crowned Sparrow populations where conifers are removed, more widespread negative effects are unlikely due to their abundance across North America (Chilton et al., 2020).
Often considered a near-obligate of sagebrush habitats, Green-tailed Towhee range closely overlaps the sagebrush biome (Dobbs et al., 2020; Knick et al., 2003; Pavlacky & Anderson, 2004; Rich et al., 2005). Green-tailed Towhees are typically associated with the ecotone between open sagebrush and forested habitats, and they exemplify some of the complexity associated with a transitional edge habitat. Our results demonstrate lower abundance and fledgling production in removal areas, highlighting the importance of scattered trees for them (Figures 3 and 5). To complicate matters, towhees have mixed population trajectories across their range, with decreases in some areas and increases in others (Sauer et al., 2019; Van Lanen et al., 2023b). Recent large-scale mapping shows that conifer encroachment has likely increased towhee habitat across broad swaths of their range while infill has dramatically decreased the suitability of historically open conifer forests (Dobbs et al., 2020; Filippelli et al., 2020). Complete conifer removal from encroached sagebrush may decrease reproductive output, but thinning of dense conifer stands adjacent to sagebrush may benefit Green-tailed Towhee populations. Additional study is warranted to understand towhee response to future thinning of their historic, infilled conifer forests.
Sagebrush restoration did not benefit woodland-associated species following tree removal. Dark-eyed Juncos and Chipping Sparrows were nearly absent from removal areas (Figure 3), leading to difficulty in estimating nest success (Figure 4). Both species are more closely associated with forested habitats, so their avoidance of open sagebrush is unsurprising (Middleton, 2020; Nolan et al., 2020; Reinkensmeyer et al., 2007). While conservationists should be aware of the potential negative impacts of conifer removal on these species, population-level consequences will likely be minimal given their limited use of this habitat type across their wide distributions. Widespread forest loss and degradation, introduction of invasive species, and increased nest parasitism are more important for their population trajectories (Middleton, 2020; Nolan et al., 2020; Ortega et al., 2006). Thus, the proportion of a species' distribution that coincides with areas of umbrella species management may be a useful indicator of population-level impact.
Conifer removal is a restoration practice that has the potential to provide significant benefits for declining sagebrush-associated species with limited harm to woodland-associated species (Tack et al., 2023; Van Lanen et al., 2023b). Our demographic analysis provides confirmation of the assumed benefits for sagebrush songbirds and allows decision-makers to quantify the impact of removal treatments on populations. Additionally, most of the woodland-associated species we considered are widespread across North America and are not species of conservation concern. However, species that are more reliant on the sagebrush-forest ecotone, such as Green-tailed Towhees and Pinyon Jays could be harmed (Van Lanen et al., 2023a). Planning conifer removal in areas where these species are absent can reduce negative outcomes (Reinhardt et al., 2023; Zeller et al., 2021). Finally, given that widespread conifer encroachment should have increased habitat for these species, the causes of population declines are unclear and further study is necessary for effective conservation.
Ultimately, our results indicate that habitat restoration for an umbrella species will likely benefit co-occurring species with similar habitat associations. Our study found no evidence of unintended demographic consequences from habitat restoration. However, restoration can create ecological traps or population sinks (Hale & Swearer, 2017), so the study of the demographic consequences of umbrella species management remains essential for effective restoration.
AUTHOR CONTRIBUTIONS
Elise C. Zarri, David E. Naugle and Thomas E. Martin conceived the ideas and designed the methodology; Elise C. Zarri collected and analysed the data and led manuscript writing. All authors contributed critically to the drafts and gave final approval.
ACKNOWLEDGEMENTS
We thank two anonymous reviewers, Mark Hebblewhite, Angela Luis, John Maron and the Martin Lab for their helpful comments on earlier drafts of this manuscript. We thank the many field assistants who made this work possible and the Hansen family for providing access to the field site. We acknowledge that this work was conducted on the traditional lands of the Shoshone-Bannock and other native peoples. We thank them for their stewardship of the land. Funding was provided by the Bureau of Land Management and the US Fish and Wildlife Service.
CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest to declare.
Open Research
DATA AVAILABILITY STATEMENT
Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.zgmsbcckx (Zarri et al., 2024).
