Conservation Evidence strives to be as useful to conservationists as possible. Please take our survey to help the team improve our resource.

Providing evidence to improve practice

Action: Remove vegetation by hand/machine Terrestrial Mammal Conservation

Key messages

Read our guidance on Key messages before continuing


  • Richness/diversity (1 study): A site comparison study in the USA found that mechanically clearing trees within woodland reduced small mammal diversity.


  • Abundance (11 studies): Eight of 11 site comparison or controlled studies (nine of which were replicated), in the USA, Israel, Portugal, Spain and the Netherlands, found that clearing woody vegetation or herbaceous and grassland vegetation benefitted target mammals. Population or density increases were recorded for small mammals, European rabbits and Stephens’ kangaroo rat while black-tailed prairie dog and California ground squirrel colonies were larger or denser and Utah prairie dog colonies established better than in uncleared areas. Two studies found mixed results of clearing woody vegetation, with hazel dormouse abundance declining, then increasing and small mammal abundance increasing, then declining in both cleared and uncleared plots alike. One study found no effect of scrub clearance from sand dunes on habitat specialist small mammals.
  • Survival (1 study): A replicated, site comparison study in the USA found that mechanical disturbance of woody vegetation within forest (combined with reseeding, follow-up herbicide application and further seeding) increased overwinter survival of mule deer fawns.


  • Use (8 studies): Four of seven studies (of which six were site comparisons or controlled), in the USA, Canada, Norway, France and Thailand, found that areas cleared of woody vegetation or herbaceous and grassland vegetation were utilized more by mule deer, reindeer, mouflon and gaur. One study found that clearing woody vegetation promoted increased use by white-tailed deer in some but not all plots, one found that it did not increase use by mule deer and one found that carrying out a second clearance on previously cleared plots did not increase use by white-tailed deer. A before-and-after study in South Africa found that clearing woody vegetation from shrubland increased wildebeest and zebra abundance following subsequent burning but not when carried out without burning whilst other mammals did not show consistent responses.

Supporting evidence from individual studies


A replicated, controlled, before-and-after study in 1966–1970, of pinyon-juniper woodland and grassland at six sites in Utah, USA (Baker & Frischknecht 1973) found that, after clearance of pinyon-juniper and seeding with grassland species, small mammal abundances in both cleared and uncleared plots followed similar patterns. Comparisons between treatments were not tested for statistical significance. Two years after clearance and seeding, more deer mice Peromyscus maniculatus were caught in cleared plots (107–118 from 180 trap nights) and in uncleared plots (89 from 180 trap nights) than were caught before clearance and seeding (19 from 270 trap nights). However, after three to four years, abundance in cleared plots declined (16–37 mice from 180 trap nights) and abundances in uncleared plots also declined (27–30 from 180 trap nights). Trees were cleared by dragging a heavy chain or were bulldozed. Aerial seeding followed. Felled wood was gathered into lines and left in place or burned, or was dispersed during a second pass of the chain. In 1966–1970, small mammals were sampled using snap-traps over a range of dates in August–November.


A replicated, controlled study in 1975–1977 on grassland in British Columbia, Canada (Willms et al. 1980) found that in mown areas, bluebunch wheatgrass Agropyron spicatum was consumed more by foraging mule deer Odocoileus hemionus than in unmown areas. Deer took a higher average number of bites/observation of bluebunch wheatgrass in mown plots (12 bites) than in unmown plots (two bites). Plots were studied at two sites in sagebrush and two in Douglas fir Pseudotsuga menziesii forest. At each site, plots (1.25 × 5 m) were established in a block. In each block, in October 1975, three plots were clipped using a lawnmower and electric-powered sickle and three were uncut. In April 1976, three deer were fenced onto the block and their selection between plots was assessed through direct observations at intervals through the day. The same three deer were used on all blocks and observed twice/block for one day each time. In April 1977, four deer were observed, on two blocks combined, over four days.


A site comparison study in 1977 of five areas within a pinyon-juniper woodland in Colorado, USA (O'Meara et al. 1981) found that mechanically clearing trees increased small mammal abundance but reduced diversity. More small mammals were caught in area cleared areas (175–295 individuals) than in the uncleared area (102 individuals). However, diversity was lower in cleared areas than in the uncleared area (results reported as Shannon-Weaver diversity index). Small mammals were sampled in four study areas (≤28 km apart). One area was mature pinyon-juniper woodland whilst other areas comprised woodland that had been cleared by chaining (a heavy anchor chain was dragged between two bulldozers) 1, 8, and 15 years previously. Small mammals were live-trapped on three grids in each area (32 trap stations/grid). Trapping was conducted concurrently on all areas, during two trapping sessions of eight days each, in mid-July and mid-August 1977.


A controlled study in 1978–1981 of grassland at four sites in a national park in Utah, USA (Player & Urness 1982) found that mechanical disturbance of vegetation promoted establishment of translocated Utah prairie dogs Cynomys parvidens. In the first year of translocation, more prairie dogs (8–16) were counted on sites where vegetation was disturbed than on sites where vegetation was not disturbed (0.3). The same pattern held over the second year (disturbed: 9–14; undisturbed: 0 prairie dogs) and third year (disturbed: 15–16; undisturbed: 0 prairie dogs) after translocation. In August 1978, vegetation in one site was disturbed using a rotobeater. In another site, four railroad rails were dragged twice over the site. Vegetation was not disturbed at a third site. Sites were 5 ha each. On each site, 200 artificial burrows were created. In early-summer 1979, a total of 200 prairie-dogs were translocated and released across four sites (these three sites and a fourth site, not detailed here). Counts were conducted through summer and autumn of 1979 and in summer 1980–1981.


A replicated, controlled study in 1981–1983 of a pinyon-juniper woodland in New Mexico, USA (Severson 1986) found that 13–18 years after treatment, felled or thinned stands had more small mammals than did undisturbed stands. The number of animals caught in stands that were thinned (432) or bulldozed (433) did not differ from each other but both were greater than the number in undisturbed stands (246). Species composition differed, with more grassland species in bulldozed stands (bulldozed: 95–175; thinned: 35; undisturbed: 46) and more woodland mice in thinned stands (thinned: 58; bulldozed: 6–11; undisturbed: 26). Plots, approximately 120 ha each, were established in each of two woodland blocks, one in 1965, one in 1970. In each block, one plot was thinned (trees ≥6.1 m apart), one was bulldozed (trees pushed over and left) and one was undisturbed. Small mammals were trapped in the second and third week of September, each year, in 1981–1983. Each plot was sampled for four days each year.


A randomized, controlled, before-and-after study in 1981–1983 of forest and grassland on a ranch in Texas, USA (Rollins et al. 1988) found that after partial clearing of woody vegetation, there was a mixed response in white-tailed deer Odocoileus virginianus use of these areas. Changes in use of partially cleared areas were not tested for statistical significance. In two of four plots that were partially cleared, average deer numbers increased (after: 22–24 deer/100 ha; before: 3–13 deer/100 ha). In the other two plots that were partially cleared, average deer number declined (after: 11–15 deer/100 ha; before: 13–15 deer/100 ha). In the plot that was not cleared, deer numbers declined (after: 20 deer/100 ha; before: 27 deer/100 ha). On a 20,000 ha ranch, five plots (120 ha each, ≥4 km apart) were studied. Two tractors dragged a heavy-duty chain in a U-shape to partly clear four plots of woody vegetation in May–June 1981. Plots had 30, 50, 70, and 80% of woody vegetation cleared. Uprooted woody material was removed by burning in July 1981. A fifth plot remained uncleared. Treatments were assigned randomly to plots. Deer were counted from helicopter transects, every three months, from March 1981 to March 1983.


A replicated, controlled study in 1995–1996 of a coastal sand dune in Israel (Kutiel et al. 2000) found that removing scrub did not increase abundances of habitat specialist sand-living small mammals. The total number of Anderson’s gerbils Gerbillus allenbyi in cleared plots (124) did not significantly differ from that in uncleared plots (107). The same applied for Tristram’s jird Meriones tristrami, (cleared: 3; uncleared: 8). However, scrub clearance reduced numbers of invasive house mice Mus musculus (cleared: 6; uncleared: 109). All aboveground woody vegetation was removed from two 50 × 50-m plots, in September 1995. Plots were >200 m apart. Uncleared plots were located 50–200 m from each cleared plot. Small mammals were surveyed using 36 Sherman live traps in each plot, over four nights, each month, from December 1995 to September 1996.


A before-and-after study in 2001–2002 of a shrubland site in Texas, USA (Rogers et al. 2004) found that carrying out a second mechanical vegetation clearance of plots already subject to an earlier mechanical clearance did not increase their utilization by white-tailed deer Odocoileus virginianus. There was no significant difference in deer track counts between plots before (37 track crossings/km) or after (47 track crossings/km) the second mechanical clearance. Plots (3–9 ha), were established in a 6,154-ha study area. In March–April 1999, five plots were cleared of woody vegetation using a mechanical aerator pulled by a tractor. Plots were mechanically cleared again in September 2000. Deer utilization was assessed by counting tracks along prepared track lanes, over three days on four occasions. Surveys were conducted once before clearance, in late-May to July 2000, and three times after clearance, in December 2000 to January 2001, May 2001 and June–July 2001.


A replicated, controlled, before-and-after study in 1996–2000 of a grassland area in California, USA (Kelt et al. 2005) found that after vegetation mowing commenced, Stephens’ kangaroo rat Dipodomys stephensi abundance increased. More animals were estimated to be in mown plots two years after mowing began (mown: 21; before mowing 18) and in plots that were both mown and grazed (mown: 15; before mowing: 8). Plots that were neither grazed nor mown contained more animals than mown or mown and grazed plots, although the number after management of other plots commenced did not differ from that before management (28 vs 28 kangaroo rats). Seven plots (80 × 80 m) were surveyed. Two were mown in 1998 and 1999, three were mown in 1998 and grazed by sheep in 1999 and two were not grazed or mowed. Mowing cut vegetation as short as the mower allowed. Cut vegetation was left on site. Grazing removed all available forage. Kangaroo rats were surveyed using grids of Sherman live traps, over three consecutive nights, bimonthly, from November 1996 to October 2000.


A replicated, controlled, before-and-after study in 2002–2003 in a national park in Dakota, USA (Milne-Laux & Sweitzer 2006) found greater areas occupied by black-tailed prairie dog Cynomys ludovicianus colonies and more prairie dog burrows, in plots that were burned and mechanically cleared of woody vegetation than in plots that were not cleared or burned. The study does not distinguish between the effects of mechanical vegetation clearance and burning. At the end of the second summer after vegetation clearance, prairie dog colonies had expanded more (into 18–70% of available habitat) in burned and cleared plots compared to unmanaged plots (0–5%). In burned and cleared plots, there were more new burrows (191–458) after two summers than in unmanaged plots (41–116). At each of three prairie dog colonies, a 2-ha treatment plot, just beyond the colony boundary, underwent prescribed burning in May 2002 and mechanical removal of woody vegetation in June 2002. Similarly, selected 2-ha plots were left unmanaged. Colonies boundaries were mapped in May–September 2002 and May–August 2003. New burrows were mapped monthly during these periods.


A replicated, controlled study in 2003–2005 of pasture at a site in northern Norway (Colman et al. 2009) found that mown pasture was selected by feeding reindeer Rangifer tarandus more than was unmown pasture. Reindeer spent more time feeding in mown plots (25% of all feeding observations) than in unmown plots (17%). Sixteen plots were established in each of two 0.3-ha fields. Each field contained four replicate plots of high-intensity sheep grazing, low-intensity sheep grazing, mowing and unmanaged. Sheep grazing treatments are not reported on in the paper. Mown plots were cut in July, to 5 cm height, with cuttings removed. Four 2-year-old male reindeer grazed in each field for two weeks in autumn 2003, spring and autumn 2004 and spring 2005. Reindeer feeding patch choice was determined during timed observations.


A replicated, controlled, before-and-after study in 2000–2002 on scrubland in a nature reserve in southwest Portugal (Ferreira & Alves 2009) found that clearing scrub (through establishing firebreaks) increased densities of European rabbits Oryctolagus cuniculus. In areas where firebreaks were established average annual rabbit pellet densities (1.1–3.6/m2) were higher than prior to establishment of firebreaks (0.5–1.5/m2). Pellet densities were also higher than in areas where no firebreaks were established (firebreaks: 1.1–3.6/m2; no firebreaks: 0.4–2.2/m2). Four 300-ha sites, ≥3 km apart, were studied. In February 2001, areas of grassland were restored by cutting 5-m-wide firebreak strips through scrub. The other two sites remained unmanaged. Rabbit pellets were counted, monthly, at fixed points along transects, from May 2001 to October 2002.


A controlled study in 2004–2008 of heather moorland at a site in southern France (Cazau et al. 2011) found that cutting heather (Calluna vulgaris and Erica tetralix) resulted in greater use of it by mouflon Ovis gmelini musimon × Ovis sp. Average density of feeding mouflon was higher on cut plots (27/ha) than on uncut plots (5/ha). Prior to the study, each 360 × 80-m plot had not been modified for >40 years. Two plots were cut in spring 2004, to an average height of 5 cm, and two were left uncut. Mouflon use of plots was determined by counting feeding animals in each plot, at 20 minute intervals, for two hours up to sunset. In total, 668 such counts were made in 2004–2008.


A replicated, site comparison study in 2008–2012 in grassland and scrubland along a mountain chain in Andalusia, Spain (Guerrero-Casado et al. 2013) found that removing scrubland vegetation to create pasture increased abundances of translocated European rabbits Oryctolagus cuniculus in areas of high scrub coverage but not of medium- or low-scrub coverage. In high scrub cover areas, there were more rabbits around plots where scrub was cleared (5.9 latrines/km) than where scrub was not cleared (2.6 latrines/km). There was no significant difference in rabbit abundance in areas of medium cover scrub (scrub clearance: 7.1 pellets/km; no scrub clearance: 5.0 pellets/km) or low scrub cover (scrub clearance: 1.6 pellets/km; no scrub clearance: 2.1 pellets/km). In autumn and winter of 2008–2009, between 75 and 90 rabbits/ha were released into fenced plots (0.5–7.7 ha). Wooden branches and artificial warrens were added within a 500-m radius outside plots and, at some, scrubland was cleared to create pasture (number of plots/treatment and pasture sizes not reported). At the end of each breeding season in 2009–2011, small gates allowed rabbits to disperse through fences into adjacent areas. Rabbit abundance was estimated by latrine counts in four 500-m-long transects around each plot, in summer 2012.


A before-and-after study in 2009–2010 on savannah in South Africa (Isaacs et al. 2013) found that in areas cleared of woody vegetation, wildebeest Connochaetes taurinus and zebra Equus burchelli abundance was higher than in uncleared areas after areas were burned, but not before burning, whilst other mammals did not show consistent responses. Wildebeest faecal pellet prevalence was higher in cleared than in uncleared plots after burning (cleared: in 4–7% of plots; uncleared: 1%) but not before (cleared: 0%; uncleared: 2%). Similarly, zebra pellet prevalence were higher in cleared than in uncleared plots after burning (cleared: in 18–30% of plots; uncleared: 7%) but not before (cleared: 16–19%; uncleared: 20%). Impala Aepyceros melampus, kudu Tragelaphus strepsiceros and giraffe Giraffa camelopardalis did not show consistent differences between responses in cleared versus uncleared land. Herbivore abundance was determined by establishing presence or absence of faecal pellets for each species in plots along transects through areas on sandy soils subject to mechanical clearance of woody vegetation by barko crawler, bosvreter and chainsaw (date of clearance not stated) and uncleared areas. Pellets were counted in April–May 2009, prescribed burns were carried out in June–November 2009 and plots were resampled in June 2010.


A replicated, site comparison study in 2005–2008 of a pine-juniper forest in Colorado, USA (Bergman et al. 2014) found that mechanical disturbance of vegetation (combined with reseeding, follow-up herbicide application and further seeding – referred to as advanced management) increased overwinter survival of mule deer Odocoileus hemionus fawns. Management actions were not carried out individually, so their relative effects cannot be determined. Average overwinter survival was highest under advanced management (77%), intermediate under mechanical disturbance and seeding without follow-up actions (69%) and lowest with no habitat management (67%). Mechanical management, commencing in 1998–2004, involved removing and mulching trees to create open areas. These were seeded with grasses and flowering plants. Follow-up actions in advanced management plots, two to four years later, involved controlling weeds with herbicide and further seeding with deer browse species. Fawns were radio-collared on eight study plots; two advanced management plots, four mechanical management plots and two unmanaged plots. Survival was assessed by monitoring fawns from capture (1 December to 1 January) until 15 June, in winters of 2004–2005 through to 2007–2008, three to six years after mechanical treatments.


A replicated, before-and-after, site comparison study in 2009–2013 at six forest sites in the Netherlands (Ramakers et al. 2014) found that after clearance of most mature trees, hazel dormouse Muscardinus avellanarius nest abundance declined briefly but then increased relative to areas where no trees were cleared. Dormouse nest numbers in cleared plots fell in the year after clearing to 32% of pre-clearance levels. Two to four years after clearance, nest numbers were higher, at 374–803% of pre-clearance levels. Data were presented as standardised indices. In uncleared plots, there was a declining trend throughout with, at the end of the study, nest numbers 21% of the count made at the start of the study. Dormouse nests were counted along transects in September and November each year in 2009–2013. In 10 arbitrarily chosen ‘managed’ segments along transects (average 92 m long), 75–100% of mature trees were cut in winter 2009–2010. Ten unmanaged transect sections (average 181 m long) were monitored as controls.


A replicated, site comparison study in 2006–2009 of pine and juniper forests interspersed with meadows on a plateau in Colorado, USA (Bergman et al. 2015) found that mule deer Odocoileus hemionus densities did not differ between plots where trees were cleared and those where trees were not cleared. Average deer density was 6–37 deer/km2 on plots where trees were cleared and 5–85 deer/km2 on plots where no trees were cleared. Tree clearance was carried out on four plots, two to eight years prior to deer surveys. This comprised uprooting trees with a bulldozer, followed by mechanical roller chopping to break vegetation into smaller pieces, or hydro-axing, whereby individual trees were mulched to ground level. In two plots, no trees were cleared. Deer numbers were estimated by resighting marked individuals, in late winter each year in 2006–2009, from aerial surveys. Surveys were conducted over 15–94 km2/plot.


A replicated, controlled, paired sites study in 2011–2014 of two areas of grassland and scrubland in southern California, USA (McCullough-Hennessy et al. 2016) found that in mown areas, California ground squirrel Otospermophilus beecheyi burrow densities were higher compared to in unmown areas. Three years after management commenced, there were more squirrel burrows in mown (11–122/subplot) compared to in unmown (12–54/subplot) areas. Each of six plots comprised a circle covering 0.8 ha, divided into three equal wedge-shaped subplots. One subplot in each plot was mown in May, for two years, at 7.5–15 cm height, with cut material removed and one was unmown. (Management details for the third subplot are not relevant to this intervention). Management commenced in 2011 (two plots) and 2012 (four plots). Squirrels were translocated into plots at a target rate of 30–50/plot. Squirrel abundance was determined by counting squirrel burrows.


A site comparison study in 2010–2012 in two secondary forest plots in Nakhon Ratchasima Province, Thailand (Prayong & Srikosamatara 2017) found that clearing vegetation using chainsaws increased the density of gaur Bos gaurus using these areas. Average gaur density was higher in a plot where pioneer trees were felled (8.6 individuals/km2/day) than in a plot where the vegetation was left unmanaged (4.0 individuals/km2/day). The study was conducted within an 8-km2 area, reforested since 1994. In May–September 2010, a total of 407 pioneer Macaranga siamensis trees were felled with chainsaws to open up 28% of a 5.7-ha plot. Trees were not felled in a nearby 4.7-ha plot. The ground within the felled and unfelled plots was cleared, using a tractor, in June and December 2011. Gaur dung piles were counted monthly, between February 2011 and March 2012, with the exception of June and December 2011. Dung piles were counted by 9–10 volunteers along 50-m-long transects (number not stated) with counts used to estimate guar usage of plots.

Referenced papers

Please cite as:

Littlewood, N.A., Rocha, R., Smith, R.K., Martin, P.A., Lockhart, S.L., Schoonover, R.F., Wilman, E., Bladon, A.J., Sainsbury, K.A., Pimm S. and Sutherland, W.J. (2020) Terrestrial Mammal Conservation: Global Evidence for the Effects of Interventions for terrestrial mammals excluding bats and primates. Synopses of Conservation Evidence Series. University of Cambridge, Cambridge, UK.