Number of Papers and Responses Found

Using the search criteria described above, we found 719 potentially relevant papers describing vegetation-environmental responses to pinyon-juniper treatments. Upon further review, we determined that only 48 were directly relevant, 10 of which were included in the analysis by Jones (2019). Most of the initial papers were excluded because either they had no relevance to the review (for example, medicinal use of Juniperus species), PJCX-permitted treatments were not conducted (for example, modeling or review papers), relevant treatments were combined with those not permitted by the PJCX (for example, cutting followed by prescribed fire), treatments were not conducted on pinyon-juniper species addressed in this review, or significance of treatment effects was not reported. We found papers that tested for effects relevant to all PJCX-permitted treatments except yarding and piling of cut trees (which was never specifically mentioned).

The 48 papers reported 1,709 relevant responses to PJCX-permitted treatment types (fig. 5; app. 1). All studies included cutting (37 papers) or mastication (23 papers) as a primary treatment (fig. 5). Three studies combined cutting and mastication, and we assigned these to the cutting category for our assessment because only a few masticated plots were combined with a larger number of cut plots (Roundy and others, 2018, 2020) or because large trees were cut and only small trees were masticated (Ashcroft and others, 2017). When cutting was the primary treatment, 73 percent of variables were measured following cutting alone, with the remaining responses measured when secondary treatments of pile burning, off-site tree removal, or seeding native species followed cutting. Additionally, there were instances in which seeding was conducted following pile burning or off-site tree removal, as well as instances in which those treatments were conducted with no seeding. We highlight differences at this “tertiary level” (fig. 5) only when they vary from those reported at the secondary level. For mastication, only one paper (Redmond and others, 2014) included a secondary treatment of seeding native species. We report these seeding responses only when they differ from overall mastication results.

Geographical, Ecological, and Treatment Context of Studies Included

For purposes of this summary, a “study site” was defined as a location in which one or more plots were measured. Among the 48 papers that addressed vegetation-environmental responses, most included more than one study site, with a total of >200 unique locations reported. In some instances, authors did not provide precise study site locations and (or) treatment responses were summarized using numerous sites. As such, study site locations displayed in fig. 6 are a combination of provided (that is, authors reported geographic coordinates) or estimated (that is, authors provided a map that was used to approximate site location). Many sites located in the Central Basin and Range, Northern Basin and Range, Eastern Cascade Slopes, Southern Rockies, Wasatch and Uinta Mountains, and Colorado Plateau could not be mapped with confidence due to lack of or vague location information provided by authors, including data from more than 100 treated sites in Utah assessed in a retrospective study by Monaco and Gunnell (2020). Furthermore, some locations represent more than one site in very close proximity, and other locations represent individual sites that were utilized in more than one paper.

Reported study sites were located across seven western states, with the majority (56 percent) of sampling conducted in either Utah (34 percent) or Oregon (22 percent). Across 8 ecoregions, most sampling occurred in the Central Basin and Range (52 percent), followed by the Blue Mountains (16 percent), and Northern Basin and Range (15 percent). Five additional ecoregions accounted for <6 percent of sampling each. One additional paper (Monaco and Gunnell, 2020) that analyzed 129 previously sampled study sites in Utah (primarily in the Central Basin and Range, Colorado Plateau, Wasatch and Uinta Mountain ecoregions) lacked information necessary to adequately determine the exact locations of sites that qualified for our review. Among research papers, many study sites were located near the margins of rangeland-dominated ecoregions, likely reflecting areas with higher potential for tree expansion along ecotones. Study sites were also located across a broad range of climatic conditions, with reported mean annual precipitation (represented by 30-year norms and [or] conditions around time of treatment) ranging from 229 to 573 millimeters (mm; 40 papers) and elevation ranging from 800 to 3,900 meters (m; 47 papers).

Among all papers, pinyon and juniper species were mostly studied when occurring together, with 37 papers reporting at least one site that was a pinyon-juniper woodland. Eleven papers reported exclusively on J. occidentalis and no papers reported on pinyon species without juniper present. The most studied species in our review were J. osteosperma (27 papers), J. occidentalis (22 papers), and P. edulis (25 papers), while papers examining treatment of P. monophylla (17 papers), J. monosperma (6 papers), or J. scopulorum (3 papers) were less common (fig. 7A). Five papers did not specify the pinyon and (or) juniper species treated. Successional development of pinyon-juniper woodlands being studied ranged from Phase I to Phase III (see definitions in table 1), with 12 papers comparing responses to treatments at different developmental phases. However, roughly 80 percent of reported treatments were applied in Phase II or III, and authors sometimes used different criteria to designate development phases (for example, tree cover versus tree dominance index).

Study sites and treatment sizes represented varying spatial extents. The total area spanned by study sites for any given paper ranged from local scale (<1,000 ha, 22 papers), to large-landscape scale (about 1,000–100,000 ha within a single ecoregion, 8 papers), to multi-ecoregional (11 papers). Of the 11 multi-ecoregional papers, 10 were sampled as part of the Sagebrush Steppe Treatment Evaluation Project (SageSTEP), which is a long-term experimental project designed to evaluate sagebrush steppe restoration methods in the Great Basin (McIver and Brunson, 2014). The total pinyon-juniper treated area also varied widely by study, ranging from 0.2 to 2,800 hectares when reported (40 papers). Twenty-eight papers examined effects where <200 hectares of pinyon-juniper were treated in total, 11 papers evaluated effects where 200–1,000 hectares were treated, and one paper examined effects where 2,800 hectares of pinyon-juniper were treated. When the sizes of individual treatments were reported (38 papers), most treatments were implemented in small (<8 ha; 17 papers) or medium-sized (8–30 ha; 17 papers) patches, with only 5 papers measuring responses in large, contiguous treatment patches (50–166 ha; including one paper that also included a medium-sized patch).

The temporal sampling design and duration of studies also varied greatly; for example, many papers analyzed measurements taken before and after experimental treatments, while others were retrospective studies of previous land management treatments of various ages. When the maximum time since treatment was reported (46 papers), the number of papers that measured at less than or equal to (≤) 3, 4–8, and 9–13 years after treatment was similar, accounting for approximately a third of all papers each (fig. 8A). Only one paper included responses that were measured >14 years after treatment (at 25 years; Bates and others, 2017b). However, even more germane to our assessment of results among papers, when considering responses that we could confidently place into one of the aforementioned time periods (n = 1,619), 49 percent of responses were measured ≤3 years posttreatment, while only 3 percent of responses were measured greater than or equal to (≥) 14 years posttreatment (fig. 8B). The earliest treatment considered in our review occurred in 1991 and the most recent in 2018.

Native Herbaceous Plant Abundance Responses to Treatments

Pinyon-juniper removal treatments generally did not benefit native herbaceous plant abundance, with most responses reported as non-significant across and among treatment types (fig. 9). However, when significant effects were observed, the relative dominance of positive versus negative responses varied among individual treatment types (appendix 1). When all native herbaceous species were considered (All Native Grass/Forb Abundance, n = 199 [16]), the majority of responses to all treatments combined were non-significant (57 percent), and there were slightly more positive (24 percent) than negative (20 percent) responses (fig. 9). Notably, following mastication there were >8 times more positive (42 percent) than negative (5 percent) All Native Grass/Forb Abundance responses (n = 19 [6]). Though a less pronounced difference, cutting only (n = 87 [10]) also had more positive (25 percent) than negative (11 percent) responses. When secondary treatments occurred after cutting, off-site tree removal (n = 51 [2]) had the highest percentage (73 percent) of responses that were non-significant, while pile burning (n = 42 [2]) had the highest percentage (50 percent) of negative responses (and higher than either positive or non-significant responses). Seeding following cutting (n = 62 [2]) had a higher proportion of non-significant (44 percent) than either positive (26 percent) or negative (31 percent) responses for All Native Grass/Forb Abundance.

When considering native plants based on life-cycle duration (Native Annual Grass/Forb Abundance, n = 25 [4]), responses to all treatments combined were mostly non-significant (60 percent), and positive responses (28 percent) were more frequent than negative (12 percent; fig. 9). Cutting treatments accounted for most of these responses (n = 19 [3]) and resulted in similar trends (fig. 9). Among all treatments combined, Native Perennial Grass/Forb Abundance (n = 166 [13]) also had mostly non-significant responses (56 percent), with positive responses (23 percent) slightly more common than negative (21 percent). Cutting accounted for most Native Perennial Grass/Forb Abundance responses (n = 158 [11]). At the tertiary level of treatment effects, seeding made an apparent difference in Native Perennial Grass/Forb Abundance. Specifically, when pile burning (n = 12 [2]) and off-site tree removal (n = 14 [2]) followed cutting with no subsequent seeding, there were no positive responses for either treatment. However, when seeding did occur, 42 percent of responses were positive for pile burning (n = 24 [2]) and 11 percent for tree removal (n = 28 [2]); although, negative responses were similar between seeded and non-seeded sites for both cutting and pile burning. Following mastication treatments, both Native Annual Grass/Forb Abundance (n = 6 [1]) and Native Perennial Grass/Forb Abundance (n = 8 [4]) had 50 percent positive and 50 percent non-significant responses.

Exotic Herbaceous Plant Abundance Responses to Treatments

Unlike the relatively variable response of native herbaceous species abundance, exotic herbaceous species abundance was nearly always more positive than negative following pinyon-juniper removal (fig. 9). There were 6 times more positive (42 percent) than negative (7 percent) All Exotic Grass/Forb Abundance (n = 103 [18]) responses to all treatment types combined. Following both cutting only (n = 29 [7]) and mastication treatments (n = 31 [9]), more than half of all responses were positive (52 percent and 58 percent, respectively), with no negative responses for either treatment. When pile burning followed cutting (n = 23 [4]), there were >2 times more positive (30 percent) than negative (13 percent) responses. Seeding following cutting (n = 29 [5]) also resulted in more positive than negative responses (21 percent and 17 percent, respectively), although the difference was less substantial. Off-site tree removal (n = 18 [2]) was the only treatment to result in more negative (22 percent) than positive (17 percent) responses in the All Exotic Grass/Forb Abundance category, but data were from just two related studies in the same location (Kerns and Day, 2014; Kerns and others, 2020). When comparing exotic plant responses based on life-cycle duration, Exotic Annual Grass/Forb Abundance (n = 70 [12]) responses to all treatments combined were >4 times more positive (44 percent) than negative (10 percent), with 46 percent non-significant responses. Additionally, both cutting only (n = 18 [5]) and all mastication (n = 23 [6]) treatments had 61 percent positive responses, with the remaining non-significant. There were no response data available for exotic perennial grass/forb abundance among studies we considered.

General Understory Vegetation Abundance Responses to Treatments

When all herbaceous plants were considered irrespective of life-cycle duration or native status (All Understory Abundance, n = 786 [36]), 56 percent of responses were non-significant, and there were >5 times more positive (37 percent) than negative (7 percent) responses. Among nearly all individual treatment types, either the majority or the largest proportion of All Understory Abundance responses were non-significant, and positive responses were substantially more common than negative. The exception was for mastication followed by seeding, which had mostly (57 percent) positive All Understory Abundance responses, but data were from a single study (n = 7 [1]). Among all treatments combined, Annual Grass/Forb Abundance (n = 180 [20]) had mostly non-significant responses (52 percent), and 7 times more positive (42 percent) than negative (6 percent) responses. Perennial Grass/Forb Abundance responses among all treatments combined (n = 376 [22]) were also mostly non-significant (58 percent), and there were >3 times more positive responses (32 percent) than negative responses (10 percent, fig. 9).

Native Plant Species Richness/Diversity Response to Treatments

Among all treatments combined, Native Plant Species Richness/Diversity (n = 54 [7]) responses were mostly non-significant (65 percent) and with slightly lower proportions of positive (15 percent) than negative (20 percent) responses (fig. 10). Following cutting treatments (n = 40 [3]), negative responses (23 percent) were >1.5 times more frequent than positive (13 percent). In contrast, following mastication treatments (n = 14 [4]), positive responses (21 percent) were 1.5 times more numerous than negative (14 percent). Among secondary treatments following cutting, pile burning (n = 18 [2]) had 39 percent negative and no positive responses, while seeding after cutting (n = 28 [3]) had a slightly lower proportion of positive (14 percent) than negative (18 percent) responses. At the tertiary treatment level, when seeding occurred after off-site tree removal (n = 12 [2]), positive responses (17 percent) were >2 times more frequent than negative (8 percent); however, when there was no seeding after off-site tree removal (n = 6 [2]), positive and negative responses were each 17 percent.

Shrub Abundance Responses to Treatments

Treatments had mostly non-significant effects on measures of shrub abundance (fig. 10); however, of those with significant effects, positive responses were generally more common than negative. Among all treatment types combined, General Shrub Abundance (n = 165 [27]) had mostly non-significant responses (69 percent), but positive responses (26 percent) were >5 times more frequent than negative (5 percent). Notably, following cutting treatments there were >14 times more positive (29 percent) than negative (2 percent) General Shrub Abundance responses (n = 112 [19]), with the remaining responses non-significant. Concerning secondary treatments conducted after cutting, off-site tree removal (n = 4 [2]) had only non-significant responses, seeding (n = 9 [4]) had 78 percent non-significant responses, and pile burning (n = 9 [4]) had 67 percent non-significant responses. Following mastication, responses (n = 53 [15]) were also largely non-significant (70 percent), and positive responses were >1.5 times more frequent than negative (19 percent and 11 percent, respectively). There were relatively few Sagebrush Abundance responses recorded among studies (n = 37 [7]), but 46 percent of responses to all treatments combined were positive, with the remaining non-significant (54 percent; fig. 10). Most Sagebrush Abundance responses were measured after cutting only (n = 34 [6]), and the few responses following mastication followed a similar pattern (one positive, two non-significant).

Tree Abundance Responses to Treatments

Among all treatments combined, Tree Abundance responses (n = 86 [17]) were largely negative (63 percent), with 8 percent positive and 29 percent non-significant responses (fig. 10). Tree Abundance responded similarly to cutting (n = 56 [12]) and mastication (n = 30 [8], with 64 percent and 60 percent of all responses negative after treatment, respectively. When secondary treatments were conducted following cutting, tree removal (n = 7 [2]) and pile burning (n = 6 [3]) both had only negative responses, while seeding (n = 12 [4]) had 75 percent negative and 25 percent non-significant responses.

Ground Cover Responses to Treatments

Among all treatments combined, Downed Woody Debris (n = 77 [9]) had >3 times more positive (48 percent) than negative (14 percent) responses (fig. 11). Importantly, most of the Downed Woody Debris responses were reported by fuel size class (ranging from 1-hour [hr] to 1000-hr; n = 57). Downed Woody Debris positive responses were most pronounced following cutting treatments (n = 44 [6]), with >12 times more positive (64 percent) than negative (5 percent) responses. When secondary treatments followed cutting, seeding (n = 10 [1]) had 80 percent non-significant responses, while tree removal and pile burning had only one response each (positive and negative, respectively). Following mastication, 45 percent of Downed Woody Debris (n = 33 [5]) responses were non-significant, with the remaining responses split evenly between positive and negative. While this may seem counterintuitive, the Downed Woody Debris category includes dissimilar fuel types (for example, measurements of mulch and 1,000-hr woody debris) and a range of years since treatment, resulting in variable responses. When all treatments were combined, Litter (n = 79 [15]) had mostly non-significant responses (68 percent) and >2 times more positive (22 percent) than negative (10 percent) responses (fig. 11). This trend was similar for both cutting (n = 61 [12]) and mastication (n = 18 [7]). Concerning secondary treatments after cutting, there were only four Litter responses (one negative and three non-significant) for seeding, only one positive response for pile burning, and five responses for off-site tree removal (four non-significant and one positive). Bare Ground Cover was generally unchanged after treatment, with mostly non-significant responses (66 percent) among all treatment types combined (n = 100 [16]), but with nearly 6 times more negative (29 percent) than positive (5 percent) responses (fig. 11). Mastication (n = 33 [9]) resulted in 13 times more negative (39 percent) than positive (3 percent) Bare Ground Cover responses, and cutting only (n = 58 [9]) resulted in 14 times more negative (28 percent) than positive (2 percent) responses. However, secondary treatments following cutting suggest potential for increasing Bare Ground Cover. Two of three Bare Ground Cover responses for pile burning treatments were positive, and one of six responses after off-site tree removal were positive (the remaining were non-significant). There was little evidence that treatments benefited Biocrust Cover/Microbial Activity (n = 62 [8]), with 71 percent of responses non-significant among all treatment types combined, and >3 times more negative (23 percent) than positive (6 percent) responses (fig. 11). Biocrust Cover/Microbial Activity responses to cutting only (n = 27 [6]) were predominantly negative (52 percent), with the remaining responses non-significant (48 percent). Following secondary treatments after cutting, pile burning had 92 percent non-significant and 8 percent positive responses (n = 13 [2]), while tree removal had only one non-significant response, and seeding had no responses. Few studies reported on Biocrust Cover/Microbial Activity following mastication (n = 21 [4]), with 86 percent non-significant and 14 percent positive responses.

Abiotic-Water Responses to Treatments

When all treatments were combined, Soil Water (n = 47 [6]) had a slight majority of positive responses (51 percent), with most (47 percent) of the remaining responses non-significant (fig. 12). Notably, most of the Soil Water responses (n = 38 [3]) were included in the three papers (Ashcroft and others, 2017; Roundy and others, 2018, 2020) that were assigned to cutting treatments, but in which some mastication was also conducted. Following cutting only (n = 41 [4], including 38 responses with some degree of mastication), responses were 56 percent positive, 42 percent non-significant, and 2 percent negative (fig. 12). When off-site tree removal followed cutting (n = 6 [2], including 4 responses with some degree of mastication), there were 83 percent non-significant responses, with the remaining positive (17 percent; fig. 12). Hydrologic responses (n = 157 [6]), which include measures of water runoff/flow and sediment yield, were mostly non-significant (83 percent), but there were more negative (11 percent) than positive (6 percent) responses among all treatment types combined (fig. 12). Hydrologic responses to cutting and mastication treatments (n = 103 [5] and n = 54 [3], respectively) were also mostly non-significant (82 percent and 85 percent, respectively). However, the remaining responses to cutting were split nearly evenly between positive and negative, while the remaining mastication responses were only negative. Regarding secondary treatments after cutting, off-site tree removal had just twelve Hydrologic responses from a single study (Ashcroft and others, 2017), and these were evenly split between negative and non-significant. No included studies evaluated Hydrologic response variables for pile burning or seeding after cutting treatments.

Carbon Storage Responses to Treatments

The overall effects of treatments on Carbon Storage (n = 59 [5]) were ambiguous, as 80 percent of responses from the reviewed studies were non-significant, and the remaining responses were split evenly (10 percent each) between positive and negative (fig. 12). No clear patterns emerged for belowground (that is, soil and root) Carbon Storage (n = 52), with 88 percent of responses non-significant, 10 percent positive, and 2 percent negative. There were only 6 total responses (n = 5 negative, 1 positive) for aboveground Carbon Storage, and one additional non-significant response measured both total above and belowground Carbon Storage. The majority of Carbon Storage responses were for cutting only (n = 32 [3]), with 66 percent of responses non-significant, 16 percent positive, and 19 percent negative. Only one paper reported on pile burning (n = 9) following cutting, with all responses non-significant. Mastication treatment effects on Carbon Storage (n = 18 [3]) were overwhelmingly non-significant (94 percent), with the remaining responses positive.

Number of Papers and Responses Found

We identified 238 papers that met our search criteria for wildlife responses to PJCX-permitted treatments in the Web of Science database. We reviewed an additional 100 of the most relevant papers returned by the Google Scholar database, 44 of which were already included in our Web of Science search. Of the 294 papers assessed, 11 of them were relevant based on our selection criteria. Most of the excluded papers either did not measure wildlife responses, did not include PJCX-permitted treatments, or did not include pinyon-juniper species addressed in this review. One of the included papers (McIver and Macke, 2014) was also included in the review by Bombaci and Pejchar (2016).

Most of the 11 papers that met our selection criteria considered more than one type of treatment. In total, eight considered cutting, five included mastication, five included pile burning, one included seeding, and one included tree removal (table 6; fig. 13). In one of the papers that included mastication (Bristow and others, 2020), 9 percent of plots had been burned by wildfire before treatment; these plots were combined with unburned plots in the reported responses and, thus, were included in our results. Two papers included both cutting and mastication treatments. One of these reported wildlife responses separately for each treatment (McIver and Macke, 2014), whereas the other paper reported most responses (18 of 28) for both treatments combined (Magee and others, 2019; fig. 13). Thus, for this second paper, we report the 18 combined cutting and mastication responses in a separate category. Six papers included either secondary or tertiary treatments (four papers had cutting then pile burning; one had cutting then tree removal; and one had cutting followed by pile burning and seeding). We did not find any papers that specifically mentioned wildlife responses to yarding and piling treatments.

From the 11 papers, we identified 132 responses to treatments, of which 48 percent were non-significant, 22 percent were positive, and 30 percent were negative. However, the prevailing direction (positive or negative) of significant responses varied by treatment type, response type, taxonomic group, and habitat functional group (figs. 14–16). In general, regardless of the treatment type, Habitat-Related responses were most frequently reported (n = 74 [5]; table 6), with Community-Level (n = 31 [4]) and Demographic (other than reproductive; n = 24 [3]) responses also common. Reproductive (n = 3 [3]) responses were infrequently reported (table 6).]

Geographical, Ecological, and Treatment Context of Studies Included

There were 52 reported study sites among our 11 papers that addressed wildlife responses to PJCX-permitted treatments (maximum = 29 sites; Magee and others, 2019). Study sites were distributed across seven western states, with the majority (73 percent) of sampling conducted in Colorado (60 percent) and New Mexico (13 percent). Nine woodland sites within the SageSTEP network were sampled in one paper, which reported butterfly responses (McIver and Macke, 2014). One location was sampled in two papers, both of which reported sage-grouse responses (Severson and others, 2017c, 2017d). Most papers did not provide coordinates of study sites and thus we do not specifically organize studies by ecoregion; however, approximate locations are mapped (fig. 6) and most sites are in or near the Southern Rockies and the Northern Basin and Range ecoregions. Studies occurred across a broad range of elevations (800–2,832 meters [m]; 10 papers) and annual precipitation (242–462 millimeters [mm]; 7 papers).

All 11 papers specified the pinyon and (or) juniper species treated. The most commonly treated tree species was Pinus edulis (seven papers), while only one paper examined treatment of P. monophylla. No papers treated pinyon species alone, and most included at least one pinyon and one juniper species (eight papers; fig. 7B).

Study areas ranged from local to regional scales, with the smallest study encompassing approximately 120 ha and the largest spanning multiple ecoregions and four states. The paper with regional scale extent included study sites within the SageSTEP network and reported butterfly responses (McIver and Macke, 2014). Total PJCX treatment area ranged from 7 ha to approximately 6,500 ha when reported (9 papers); however, the actual area studied did not always encompass the entire treatment area. When the sizes of individual treatments were reported (5 papers), treatments ranged from 1 ha to a mean of 144 ha. The two papers that studied the same location included a total area of 6,488 ha that was treated from 2007 to 2014 and reported sage-grouse responses (Severson and others, 2017c, 2017d).

Time since treatment ranged from <1 to 11 years. Four papers measured responses ≤3 years and 4–8 years after treatment, and three papers measured responses 9–11 years after treatment (fig. 8B). When considering responses that we could confidently place into one of the aforementioned time periods reported (n = 59), 39 percent of responses were measured ≤3 years, and 6 percent were measured 4–8 years; the remaining responses were reported across temporal scales. The earliest treatment application occurred in 2002 and the most recent was in 2018.

Habitat-Related Responses

Papers that measured Habitat-Related wildlife responses relative to PJCX-permitted treatments focused on mule deer (n = 19 [2]), greater sage-grouse (n = 5 [1]), and songbirds (n = 50 [2]; fig. 15). For mule deer, habitat selection was inconsistent by season, treatment age, and primary land cover type. In one paper, selection by mule deer was generally for pinyon-juniper savanna more than pinyon-juniper woodlands, and the deer selected for recently treated (that is, 2 years previously) areas in summer but for older treatment (that is, 4 years previously) areas in winter (Sorensen and others, 2020). In the second paper, mule deer selected for lower vegetation heights, but treatment age was not associated with habitat selection (Bristow and others, 2020). For sage-grouse, Habitat-Related responses included both habitat use and selection. In a Before-After-Control-Impact (BACI) study, extent of nesting habitat was predicted to have increased each year after three consecutive years of treatment (Severson and others, 2017c). When habitat selection was measured, sage-grouse selected for older-treated areas (22 percent increase in probability of use each year after treatment) and sites closer to treatments (5.5 percent decrease in probability of use for every 100 m away from treatments; Severson and others, 2017c).

Finally, responses for songbirds were studied exclusively for habitat use. Most pinyon-juniper specialists and other woodland species used treated areas less or no differently than they used control areas, whereas one shrubland-grassland species (lark sparrow [Chondestes grammacus]) used treated areas more than it used control areas (Magee and others, 2019). Bird species associated with both dense and open woodlands used treated areas less than they used control areas (Bombaci and others, 2017).

Community-Level Responses

Papers that measured response of Community-Level parameters to treatments focused on butterflies (n = 18 [1]), birds (n = 6 [1]), and small mammals (n = 7 [2]; fig. 15). For butterflies, abundance and richness of local species did not differ between treatment and control areas (McIver and Macke, 2014). Additionally, transient whites, legume-feeding sulfurs, and Melissa blues (Plebejus melissa) were more abundant in treated areas relative to control areas, whereas juniper hairstreaks (Callophrys gryneus) were less abundant (McIver and Macke, 2014). For birds, richness and overall abundance for a community of songbirds and raptors declined faster in treated than in untreated sites, but species diversity did not change (Fair and others, 2018). Finally, for small mammals, species richness, evenness, and biomass (that is, the sum of the mean masses of each captured individual) did not change after treatment (Hamilton and others, 2019). In another study, abundances of deer mice (Peromyscus maniculatus) and least chipmunks (Tamias minimus) did not differ between treated and control areas (Bombaci and others, 2017).

Reproductive Responses

Papers that measured Reproductive responses to treatments focused on greater sage-grouse (n = 2 [2]) and pinyon jays (Gymnorhinus cyanocephalus; n = 1 [1]; fig. 15). For sage-grouse, nest success was higher (19 percent increase) within treated areas relative to the control area (Severson and others, 2017d). Nest abundance was predicted to have increased by 9.5 percent during each of the 3 years after treatment (Severson and others, 2017c). Pinyon jays abandoned nesting in treated sites, although they did often nest near treatments (Johnson and others, 2018).

Other Demographic Responses

Three papers measured Other Demographic responses to treatments and focused on greater sage-grouse (n = 2 [1]), small mammals (n = 7 [1]), and songbirds (n = 15 [1]; fig. 15). Annual survival and population growth rate of female sage-grouse did not differ significantly between treatment and control areas (Severson and others, 2017d). Density of small mammals, both overall and for most species, did not change after treatment (Hamilton and others, 2019). The one exception to this pattern was for the piñon mouse (Peromyscus truei), which declined in density following treatment. Finally, some sagebrush, grassland, and shrubland songbird species had higher density, and one shrubland-grassland species (rock wren; Salpinctes obsoletus) and one woodland-shrubland species (gray flycatcher) had lower density, in treated areas relative to control areas (Holmes and others, 2017).

Need for Continuous Long-term Study Across Multiple Spatial Scales

The value of long-term data collection and analysis has been widely recognized in ecological research (Likens, 1989; Lindenmayer and others, 2012). The studies included in our review varied widely in terms of the period of measurement for post-treatment effects, with measured responses ranging from <1 year to 25 years after treatment. Although we found relatively even distribution among studies that sampled a maximum of 3 years, 4–8 years, or 9–13 years after treatment (fig. 8A), studies with responses measured beyond 13 years after treatment were nearly nonexistent (only 1 study for vegetation-environmental responses, and none for wildlife). Some of the longer-term studies were retrospective assessments of past management treatments of various ages (for example, Coop and others, 2017; Monaco and Gunnell, 2020), while others measured responses over time following experimental treatments (for example, Chambers and others, 2021). Moreover, when considered as a proportion of responses, short-term information was clearly more abundant than long-term information. For example, for vegetation-environmental responses that could be assigned a time since treatment (n = 1,619), 49 percent of responses were recorded ≤3 years post-treatment versus 3 percent for responses measured ≥14 years post-treatment (fig. 8B).

The relatively short duration of recently published studies may not be adequate to address complex community dynamics operating on timescales dictated by slow-growing perennial plants in the understory and long-lived pinyon-juniper tree species, a concern that has been expressed by others. For instance, Hartsell and others (2020) reviewed the geographic and temporal context of 131 pinyon-juniper treatment-based studies that occurred over a longer publication period and wider variety of treatment types than considered here (that is, types not permitted by the PJCX). They found most studies only measured management effects within one year of treatment and concluded this represented “…consistently short timescales for evaluating post-treatment effects and were not representative of the possible spatial or temporal extent of responses to management actions.”

Understanding differences between short-term and long-term vegetation-environmental and wildlife responses to pinyon-juniper removal treatments will be essential for land managers to estimate the potential for restoration success. For instance, although shrub response was 83 percent non-significant among short-term studies assessed here, positive responses of shrub abundance became the majority (69 percent) among studies that measured responses ≥10 years post-treatment. If shrub cover or abundance drives habitat use or demography for wildlife species, which is likely at minimum for the sagebrush-obligate birds during the breeding season, then wildlife responses may also need longer time scales to be fully realized. Similarly, there were more positive than negative responses from exotic annual plant species over time after treatment; however, because availability of responses over time decrease substantially and environmental settings vary, firm conclusions about temporal trends cannot be drawn based on these data alone. Other research has shown that the abundance of both native and exotic species can vary broadly over time after disturbance or treatment in pinyon-juniper and sagebrush ecosystems (for example, Shinneman and Baker, 2009; Chambers and others, 2014, 2021), as can hydrologic and other abiotic responses (for example, Williams and others, 2020). Although our wildlife dataset is also insufficient to assess temporal trends, well-established temporal trends in wildlife response to disturbance found elsewhere (for example, Sawyer and others, 2017; Poessel and others, 2020) suggest such trends are likely. For these reasons, research projects that regularly measure responses to treatments in semiarid rangelands over long time frames will become increasingly valuable to the land management decision-making process, including for treatments permitted under the PJCX. Several studies reviewed here are part of the multidisciplinary SageSTEP project (McIver and Brunson, 2014) that has investigated both short and long-term (>10 years) treatment effects for a variety of variables (for example, vegetation, wildlife, soils) using consistent and repeatable methods at sagebrush steppe and woodland sites throughout the Great Basin. Treatment monitoring protocols developed by SageSTEP can also be integrated into agency objectives for long-term monitoring programs (for example, Assessment, Inventory, and Monitoring; Kachergis and others, 2022).

The size of land management agency treatments often far surpasses those of the experimental design treatments that are established primarily for research purposes. Thus, long-term agency monitoring and assessment of plant, animal, and abiotic resources across a range of typical agency treatment sizes will also be important, especially when replicating large treatments for controlled scientific study is not practical or feasible. Only three papers assessed here measured vegetation-environmental responses where pinyon-juniper removal had individual treatments of greater than 100 contiguous hectares (the largest individual treatment size reported was 166 ha); a size far less than the maximum sizes of treatments in the past (table 2) and the 4,047-ha treatment size permitted under the PJCX. This gap between practice and research is important because many ecological outcomes may be influenced by the size and placement of treatments. For instance, sagebrush shrub re-establishment into the interiors of disturbed areas may decrease with distance to seed sources (Young and Evans, 1989), necessitating a greater focus on seeding after treatment in larger treated areas (Davies and Bates, 2019). Moreover, animal movement and demographic responses to treatments may also vary with treatment size, as well as with the relative placement of multiple treatment patches on the landscape (for example, treatment proximity or environmental setting), and the size of the study area relative to the size of the overall treatment (Doherty and others, 2010; Doherty and Driscoll, 2018). For animals, effects of pinyon-juniper may not be limited to the footprint of remaining, uncut stands. For example, if the composition or activity of predator communities is affected by pinyon-juniper cover, there could be effects of predators within a limited distance of remaining stands. Therefore, proximity to untreated areas may influence animal responses to treatments as well.

Under-Studied Pinyon-Juniper Ecosystems

The geographic distribution of research sites (fig. 6) roughly corresponds to the geographic distribution of treatment history in pinyon-juniper woodlands (fig. 4), with heaviest concentrations of both activities in the ecoregions comprising the central to northern Great Basin and adjacent areas, and secondarily scattered within or near the Colorado Plateau and Arizona-New Mexico Plateau ecoregions. However, as woody species expand into various grassland and shrubland ecosystems in the western U.S., there is increasing interest to remove trees and restore non-wooded habitat across a range of environmental settings (Maestas and others, 2015; Falkowski and others, 2017; Reinhardt and others, 2020). For instance, although there are ongoing juniper removal projects occurring in ecoregions such as the Wyoming Basin and Middle Rocky Mountains, we found no peer-reviewed research addressing tree removal treatment effects in these regions. Thus, there is a need for commensurate increase in research for determining the efficacy of tree removal among various grassland and shrubland environments throughout the ecoregions of the western U.S. This need for geographic expansion is also reflected in the pinyon-juniper tree species being studied. Although some species are well-studied relative to their range-wide distribution, particularly P. edulis and J. occidentalis, other species lack regional representation in treatment research, for instance a lack of studies for J. osteosperma across the entire northeastern portion of its range. Moreover, there are disproportionately few papers addressing wildlife effects after treatment in P. monophylla ecosystems, and relatively few vegetation-environmental research papers have addressed treatment effects in J. monosperma and J. scopulorum ecosystems relative to their overall distribution (figs. 1, 6). Using a broader set of criteria for study inclusion, Hartsell and others (2020) similarly found that treatment effect studies were lacking for individual pinyon-juniper species among various ecoregions, and especially for J. scopulorum.

Under-Studied Processes, Including Interactions with Land Use and Climate Change

Even among the vegetation-environmental response categories reported here, there were often relatively few or no responses among specific treatment-response category combinations (for example, response of native herbaceous species to mastication and seeding). Specifically, of the 171 possible combinations for vegetation-environmental variables (appendix 1), 33 percent had fewer than 10 responses, and 15 percent had no responses. Of the 40 possible combinations for wildlife variables (appendix 3), 70 percent had fewer than 10 responses, and 45 percent had no responses. Moreover, we were unable to assess the response of exotic perennial forbs and grasses to any treatment because that variable was not specifically assessed in any paper. Furthermore, information on 39 combinations of treatments and responses for the vegetation-environmental variables were from just 1 or 2 independent studies. Similarly, some categories reported here largely reflect the influence of a disproportionately large number of observations from a single study. Specifically, the soil water, hydrologic, and carbon storage response categories each had a single study comprise over half of the responses assessed in this review. Additionally, other post-treatment responses were poorly represented in qualified studies, and we did not analyze them in this review. For instance, soil chemistry responses were too few among specific nutrients to summarize effectively. Changes in soil chemistry such as soil nitrogen (N) availability can affect plant growth after treatment, especially invasive species such as cheatgrass. For instance, mulch left on the ground after mastication can increase N beneficial to invasive species (Young and others, 2013b), but can also enhance soil carbon (C), simulating the growth of soil microbes that in turn utilize and reduce inorganic N (Rhoades and others, 2012).

Only 35 percent of vegetation-environmental papers included in our review addressed tree responses after treatment (for example, Havrilla and others, 2017) and only 6 percent included measures of post-treatment tree reestablishment (for example, tree seedling density). Variability among response types (that is, different measures of abundance, density, or biomass) and imprecise or inconsistent reporting among studies made it difficult to assign a specific post-treatment year to each tree abundance response; thus, we did not attempt to collectively quantify a temporal pattern or trend in tree regrowth after treatment. However, this is critical information, as potential tree reestablishment and growth may impact long-term success rates of treatments, whether for habitat restoration or hydrologic function, and can inform the need for strategic retreatment schedules.

The effects of past and continuing land use were not specifically quantified or addressed in most studies. Perhaps the most glaring omission is that only two research papers (Jacobs, 2015; Dittel and others, 2018) reviewed here experimentally incorporated the effects of livestock grazing with tree removal treatments, and only 46 percent of studies mentioned whether treatment plots were grazed by cattle or not (for example, within a fenced livestock exclosure). Treatment interactions with livestock grazing can enhance tree seedling or exotic species survival over native herbaceous species, damage remnant soil crust or impede biocrust redevelopment, alter soil chemistry, and contribute to increased erosion (Neff and others, 2005; Shinneman and others, 2008; Shinneman and Baker, 2009; Condon and Pyke, 2018). Hartsell and others (2020) reported that only 31 percent of the studies they examined considered grazing impacts relative to treatment effects, which they suggested was inadequate given that grazing is a dominant land use in pinyon-juniper landscapes. Jones (2019) also found that most reviewed studies did not control for grazing following treatments. Moreover, few studies mentioned the previous treatment or disturbance history of the sites they studied (for example, past clearings, seedings of nonnative wheatgrasses), a process that is now facilitated by creation of databases on historical treatment records (using the LTDL; Pilliod and others, 2019) and historical wildfire perimeter data (for example, Welty and Jeffries, 2021). Given the availability of historical datasets, as future research is designed and implemented, it will be critical to ensure these types of past disturbances and their interactions are considered.

Only four of the vegetation-environmental papers reviewed here specifically examined the influence of ambient climate variability on responses to treatment, and none utilized experimental techniques to mimic potential climate change impacts on the efficacy of tree removal treatments. Post-treatment climate variability can have dramatic effects on regeneration success and species composition (for example, Chambers and others, 2014; Brabec and others, 2015). Climate-warming induced changes in the intensity and patterns of pinyon-juniper mortality events (for example, drought dieback, insect outbreaks, wildfire), declining recruitment, and shifts in species distribution also need to be considered more fully in future treatment research. For instance, a recent study (Shriver and others, 2022) suggests that P. edulis, P. monophylla, J. monosperma, and J. scopulorum populations are all declining in the western United States, especially in warmer and drier portions of their ranges, due to both enhanced rates of mortality and inadequate rates of recruitment. P. edulis experienced the largest proportion (24 percent) of its populations in decline (Shriver and others, 2022). The reasons for these changes are complex but include extensive drought-induced dieback events in some regions (Shaw and others, 2005; Brewer and others, 2017; Filippelli and others, 2020; Stanke and others, 2021). Intensified disturbance events may be followed by warmer and drier post-disturbance climates and interact with grazing or other dominant land uses that create inhospitable conditions for tree regeneration, especially for lower elevation forest and woodland ecotones (Stevens-Rumann and others, 2018; Urza and others, 2020). Such changes will have consequences for pinyon-juniper woodland distribution, plant species composition, hydrologic processes that determine availability of water resources, and wildlife habitat and food sources. For instance, substantial declines in pinyon seed production via periodic “masting” events could have serious ramifications for regeneration of pinyon trees as well as a host of wildlife species that depend on abundant seed as food (Redmond and others, 2012).