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: Use group-selection harvesting Forest Conservation

Key messages

Read our guidance on Key messages before continuing

  • Four of eight studies (including one replicated, controlled study) in Australia, Canada, Costa Rica and the USA found that group-selection harvesting increased cover and diversity of understory plants and the density of young trees. Two studies found it decreased understory species richness2 and biomass.Two studies found no effect on understory species richness and diversity and two found no effect of group-selection harvest on tree density and growth-rate.


Supporting evidence from individual studies


A replicated, controlled study in 1977-1997 in temperate mixed coniferous forest in California USA (Battles et al. 2013) found no effect of group- or single-selection harvesting on understory plant species richness. Numbers of species/1.13 ha in group (58) and single-tree selection harvest (52) was similar to unharvested plots (48). The study area was divided in sections of 8–80 ha that were assigned to the following treatments: group and single tree selection (approximately 11% of the section was harvested every 10 years in groups smaller than 0.6 ha and smaller than 0.1 ha respectively) and unharvested (since early 1990s). Understory vegetation was monitored in 30 m radius plots within each treatment annually from 1977.



A site comparison study in 2000 in a Mediterranean jarrah forest in Western Australia (Burrows, Ward & Cranfield 2002) found that group selection harvesting decreased plant species richness and abundance. The number of native plant individuals/m2 (group selection: 31; uncut: 38) and individuals/30 m2 (group-selection: 943; uncut: 1,138), as well as the number of native plant species/m2 (group-selection: 10.1; uncut: 13.3) were lower in group-selection than uncut plots. The number of species/30 m2 was similar between treatments (group-selection: 53; uncut: 57). Data were collected in five lines of 30 quadrats (1 × 1 m) in group-selection (retaining gaps of 4–7 ha, applied in 1995) and uncut treatments located in an 11,000 ha study area.



A replicated, controlled study in oak–pine Quercus–Pinus forest in Maine, USA (Schumann, White & Witham 2003) found no effect of group selection harvesting on species richness and diversity of understory vegetation. Numbers of species/1 m2 (group-selection: 18-34; uncut: 18-25) and diversity (Shannon's index group-selection: 1.7-2.2; uncut: 1.9-2.1) were similar between treatments. Data were collected in 1998 in 40 pairs of group-selection (36-3,393 m2 gaps harvested in 1987-1988) and uncut sites inside a 40 ha study area. Equal number (proportional to the gap size) of 1 m2 plots were monitored in each pair.



A replicated, controlled study in 1994-2000 in mixed hardwood forest in North Carolina USA (Elliott & Knoepp 2005) found that group-selection harvesting increased the diversity of shrubs and herbaceous plants, but not the density of shrubs and trees. Numbers of shrub species/plot (group-selection: 10; uncut: 4) and diversity (Shannon index) of herbaceous plants (group-selection: 2.2; uncut: 1.8) were higher in group-selection than uncut plots. The density (individuals/ha) of shrubs (group-selection: 28,347; uncut: 21,789) and of trees (group-selection: 742; uncut: 771) was similar between treatments. Three group-selection (0.1–0.2 ha openings, 25% tree-cover removed) and two uncut sites (4.0-6.6 ha) were established in 1994. Monitoring was in 2000 in four plots (20 × 40 m) in each treatment site.



A replicated, controlled study in 1997-1999 in tropical forest in Costa Rica (Dupuy & Chazdon 2008) found that group-selection harvesting increased the density of new tree seedlings. The density of new tree seedlings was 2.5/m2 in group-selection, and <0.5/ m2 in uncut plots. In 1997, large gaps (320–540 m2) were created inside five 40 × 40 m plots (group-selection) by cutting and removing all stems ≥5 cm diameter at breast height. Five other similar size plots (uncut) were unmanipulated with respect to canopy cover. Data were recorded every two months for one year after treatment.



A replicated, controlled, before-and-after trial in 2004-2005 in temperate broadleaf forest in Ontario Canada (Falk et al. 2006) found that group-selection harvesting increased the diversity of early spring herbaceous species and decreased the percent of plant species lost. The increase in the diversity (Shannon's index) of early spring herbaceous species was higher in group-selection (0.15 to 0.25) than in unharvested plots (0.32 to 0.34). Overall, the percentage of plant species lost was higher in unharvested (15%) compared to the group-selection treatment (8%). The percentage of plant species gained was similar (unharvested: 29%; group-selection:  35%). Two replicates (average 33 ha) of each group-selection harvest (creating five 400 m2, four 700 m2 and three 1,400 m2 gaps) and unharvested plots, were established between November 2004 and April 2005. Sampling was in April 2004 (pre-harvesting) and in April-May 2005 (post-harvesting) in 4 m2 regeneration growth plots (45×2 in control and 112×2 in group-selection).



A replicated, controlled, before-and-after study in 1995-2001 in temperate coniferous forest in Oregon, USA (Davis & Puettmann 2009) found that group-selection harvesting increased the change over time in herbaceous and shrub cover. The increase in herbaceous cover (group-selection: 3; uncut: -2%) and in low shrub cover (group-selection: 20%, uncut: -4%) was higher in group-selection than in uncut plots. The increase in bryophyte cover (group-selection: 14%; uncut: 5%) and in tall shrub cover (group-selection: 6%; uncut: 0%) was similar between treatments. Gaps (0.2 ha circular gaps, retaining 250 trees/ha) and uncut treatment units (15-53 ha) were established in each of four sites in 1995-1997. In uncut units about 7.5% of the area was covered using 0.1 ha circular plots. In group-selection units, one 0.1 ha plot was placed in each of ten gaps, ten gap-edges, and ten areas between the gaps. Data were collected before treatments and again in 2001 in 16 subplots of 0.1 m2 in each plot.


A replicated, randomized, controlled study in 2004-2008 in temperate broadleaf forest in Wisconsin, USA (Dyer et al. 2010) found that group-selection harvesting decreased the above ground biomass, but not the annual biomass increase. Above ground biomass was lower in group-selection (242,000 kg/ha) than in uncut plots (260,000 kg/ha), while the annual biomass increase was similar between treatments (11,000 kg/ha). Biomass of all plants <1.4 m tall was higher in large (700 kg/ha) than in medium (620 kg/ha) and small (480 kg/ha) gaps, and was higher in all gap-sizes compared with the transition zones (250-300 kg/ha). In 2007, all trees >5 cm diameter at breast height were cut in one small, one medium and one large circular subplots (gaps) of 4, 8 and 11 m radius in each of 15 plots of 80 × 80 m group-selection. In other 20 similar plots, subplots remained uncut. Each subplot was surrounded by an untreated transition zone 4, 8, and 11 m wide respectively. Total above ground biomass was determined for the entire plot, biomass of plants <1.4 m tall was measured in four 2 × 2 m quadrat at each gap and transition zones.



A replicated, controlled, before-and-after study in 1995-2007 in mixed conifer and broadleaf temperate forest in Maine, USA (Arseneault et al. 2011) found that two group-selection harvesting treatments affected tree annual growth rates differently, but neither differed from the uncut control. Average basal area annual growth was higher in the large group (0.27 m2/ha) than small group treatment (−0.05 m2/ha). There was no difference in average basal area annual growth between any of the group-selection treatments and the uncut treatment (−0.09 m2/ha). Three treatments were replicated at three different sites: large-group (trees removed from 20% of the area creating 1,000-2,000 m2 gaps); small-group (trees removed from 10% of the area creating 500-1,000 m2 gaps); and uncut. Treatments were applied in 1995-1997. Monitoring was in 2005-2007 in 20 plots (0.05 ha) randomly selected in each treatment.


Referenced papers

Please cite as:

Agra H., Schowanek S., Carmel Y., Smith R.K. & Ne’eman G. (2019) Forest Conservation. Pages 331-347 in: W.J. Sutherland, L.V. Dicks, N. Ockendon, S.O. Petrovan & R.K. Smith (eds) What Works in Conservation 2019. Open Book Publishers, Cambridge, UK.