1. Introduction
Conservation of mature and old-growth forests has reached a crucial point in time because there are unpredictable and large-scale losses of existing timber due to clearcut harvesting, natural disturbances from wildfire, insect epidemics, windthrow, and potentially widespread drought owing to climate change [
1,
2,
3,
4]. Naturally disturbed forests may also have large-scale salvage harvesting that creates very large (>1000 ha) contiguous openings [
5]. Clearcutting continues to be common and may reduce the abundances of some mammalian species because of a loss of food, cover, and forest stand structure [
6,
7,
8,
9,
10].
Forest-floor small mammals are excellent ecological indicators of significant change in forest structure and function [
11,
12,
13,
14]. These functions include prey for many predators [
15,
16], consumers of seeds [
17,
18], seedlings [
19,
20], other plant products [
21], and invertebrates [
22,
23], and dispersal of fungal (including mycorrhizae) spores [
24,
25]. Many studies have used small mammals as a model to evaluate improvements in forestry practices and sustainable management for conservation of forest biodiversity across landscapes [
14,
26].
Responses of forest-floor small mammals to clearcutting in North America are species-specific with generalists such as the deer mouse (
Peromyscus maniculatus) and chipmunks (
Neotamias spp.) occupying a range of habitats, whereas
Microtus voles and
Sorex shrews may persist on clearcuts for variable periods. Similarly, common species of small mammals usually increased in abundance, or were not affected, by clearcutting of temperate and boreal forests in Europe [
10]. Small mammals take advantage of increased vegetation cover and food on clearcuts. Foods include herbaceous forbs and grasses, seeds, fruits, and mast of various tree and plant species, invertebrates, and fungi [
17,
24,
27,
28].
Specialists such as the southern red-backed vole (
Clethrionomys gapperi) require closed-canopy forest and disappear on clearcuts, often within a year after harvest, at least in coniferous stands in North America [
29,
30]. Thus,
C. gapperi is an important indicator species of closed-canopy forest conditions in managed landscapes [
31]. This microtine does not return to old forest-level abundance for several decades [
32,
33]. Recent reviews by [
9,
26] have corroborated this pattern of response in abundance of small mammal species to clearcutting in North America.
A major limitation of determining the explanation for population changes of forest-floor small mammals is that most studies are short-term (e.g., 3–5 years). However, the crucial importance of identifying and understanding how species and ecosystems respond to environmental change requires long-term studies of many decades [
34]. Research undertakings in forest ecology are long-term endeavors that may be termed “the long now” [
35]. The value of long-term monitoring of ecological sites has been known for some time [
36,
37,
38,
39], Long-term monitoring programs provide managers with input on the effectiveness of past actions (e.g., silvicultural systems) and/or environmental change (e.g., habitat alteration and climate change) that help provide input into management and policy decisions [
34,
39]. This feature of “the long now” allows us to look forward in a way that is informed by the past [
35]. For example, long-term datasets (46 years) from the Yukon Territory have recorded major changes in northern red-backed vole (
C. rutilus) and deer mouse numbers over several years, possibly related to climate change [
40,
41].
In many forest regions, the cumulative removal of forest cover by clearcutting and salvage harvesting has generated large expanses of early successional habitat on an unprecedented scale [
4]. We ask what are the long-term consequences of this dramatic change in habitat for abundance and species composition of forest-floor small mammal communities? We have a unique 42-year window of a landscape from the first clearcut-harvest event in lodgepole pine (
Pinus contorta var.
latifolia) forest through four independent periods of cumulative clearcutting (1979 to 2020) in southern British Columbia (BC), Canada. Cover of early successional habitat progressed from 0% of the landscape in 1977 (pre-clearcutting) to ≥70% in 2020. Conversely, cover of standing mature and old-growth coniferous forest has declined from near 100% to ≤30%. In addition, a second anthropogenic disturbance is that virtually all clearcuts are grazed by cattle (
Bos taurus) which are ubiquitous throughout much of the inland Pacific Northwest, at least where summer forage is relatively abundant [
42]. A further disturbance is climate change, which is particularly severe in arctic regions, but is affecting all ecosystems including temperate forests [
43].
Firstly, we investigated the responses in abundance and species composition of small mammal communities to cumulative clearcutting of coniferous forests over a landscape that covered four independent harvest events (Periods 1 to 4) over a 42-year interval from 1979 to 2020. In particular, we ask if the small mammal communities have changed significantly over these decades owing to removal of old-growth forest by clearcut harvesting. Secondly, we report on changes in abundance, species richness, and diversity of small mammal communities in newly clearcut harvested sites and uncut old-growth forest sites over these same four Periods. Thirdly, we compared the responses in the abundance of early seral vegetation (herbs and shrubs) post-harvest and abundance, species richness, and diversity of small mammal communities in cattle-grazed and ungrazed clearcuts over three comparable periods of cumulative clearcutting at our Summerland and Golden study areas, respectively.
To better understand these long-term changes, we tested three hypotheses (H): the small mammal community would (H1) increase in abundance, species richness, and diversity on new clearcuts owing to the availability of early seral post-harvest habitats from cumulative clearcutting; and (H2) have higher mean abundance, species richness, and species diversity in clearcut than uncut forest sites, owing to availability of vegetative food and cover. A third hypothesis (H3) predicted that abundance of (i) early seral vegetation (herbs and shrubs) and (ii) small mammal populations, will be greater in ungrazed than grazed clearcut sites
2. Methods
2.1. Study Areas
Studies were conducted in BC, Canada from 1979 to 2020. The long-term cumulative clearcutting (LTCC), clearcut-forest (CC-FOR), and ungrazed vs. grazed clearcut (CC-GR) studies were located at two study areas in south-central BC: (i) Summerland (LTCC + CC-FOR + CC-GR) 25 km west of Summerland and (ii) Golden (CC-GR) 25 km and 35 km east and northwest, respectively, of Golden. The Summerland area is primarily in the Montane Spruce (MS
dm; dry, mild) biogeoclimatic subzone with a small part in the upper Interior Douglas fir (
Pseudotsuga menziesii var.
glauca) (IDF
dk; dry, cool) subzone [
44] (
Table 1). Hybrid interior spruce (
Picea glauca ×
P. engelmannii) and subalpine fir (
Abies lasiocarpa) are the dominant shade-tolerant climax trees. Trembling aspen (
Populus tremuloides) and black cottonwood (
Populus trichocarpa) occur on some moist sites [
44].
The Golden area is in the MS
dk (dry, cool) and Interior Cedar-Hemlock (ICH
mk; moist, cool) biogeoclimatic subzones with topography ranging from hilly to steep terrain at 1060–1350 m elevation in the lower ranges of the Rocky Mountains. Western red cedar (
Thuja plicata) and western hemlock (
Tsuga heterophylla) dominate mature climax forests with Douglas-fir, lodgepole pine, white spruce (
Picea glauca), Engelmann spruce (
Picea engelmannii), and subalpine fir common in these stands [
44].
The Summerland studies were in a commercial forest landscape with clearcut harvesting of lodgepole pine beginning in 1977 and continuing periodically throughout the 1980s and 1990s in response to an outbreak of mountain pine beetle (MPB) (Dendroctonus ponderosae). Approximately 30% of uncut old-growth forest remained in this area, and hence approximately 70% of the original standing forest of lodgepole pine had been harvested. Thus, the clearcut harvests that initiated the four Periods of the LTCC and CC-FOR and three Periods of the CC-GR studies (1979–1982, 1997–2002, 2007–2011, and 2017–2020) included harvest of lodgepole pine and other available coniferous species.
The Golden study area was located in a similar commercial forest landscape but with some larger expanses of unbroken forest (100 s to 1000 s of ha) than at Summerland. Clearcutting was initially dominated by salvage harvesting of lodgepole pine from stands of MPB-killed and susceptible trees and then conventional harvest focussed on Douglas-fir and interior spruce. The three Periods (2, 3, and 4) of the CC-GR study at Golden (2004–2009, 2012–2016, and 2016–2019) were approximate matches for those at Summerland (1997–2002, 2007–2011, and 2017–2020) with respect to number of post-harvest years after each new clearcutting.
2.2. Study and Sampling Designs
For the LTCC study at Summerland, there were 1, 3, 3, and 4 replicate sites of each treatment in
Periods: (1) 1979–1982, (2) 1997–2002, (3) 2007–2011, and (4) 2017–2020, respectively. Timing of clearcut harvesting and mean area of clearcut sites are listed in
Table 1. Pinegrass (
Calamogrostis rubescens), Arctic lupine (
Lupinus arcticus), fireweed (
Epilobium angustifolium), and heart-leaved arnica (
Arnica cordifolia) were the major herbaceous species on these 1- to 6-year-old clearcut sites. All clearcut units were aerially seeded with an agronomic grass-legume mix in the first year after harvest to enhance forage production for cattle in
Period 1 at Summerland [
42]. However, this practice was dramatically reduced by 83% by the 1990s [
45] and was not evident in
Periods 2,
3, or
4 in our study areas. Seasonal grazing by cattle continues to be common and relatively consistent on harvested sites, particularly clearcuts at Summerland, and may last for at least 10
–15 years post-harvest [
42].
The CC-FOR study at Summerland had a completely randomized design with two treatments: (a) clearcut harvest and (b) uncut old-growth forest in each of the four independent
Periods. Forest sites were composed of a mixture of lodgepole pine and Douglas-fir with scattered interior spruce and subalpine fir in wetter sites. Mean ages of lodgepole pine ranged from 80 to 120 years and Douglas-fir and other conifers ranged from 120 to 220 years. Area of forest sites ranged from 10 to 100+ ha. Canopy closure ranged from 82% to 88%. The CC-GR study (
Periods 2,
3, and
4) included clearcut sites at Summerland which had summer grazing by cattle and at Golden which had no history of grazing (
Table 1). However, seeding of landings, road-sides, and skid-trails with an agronomic-legume mix for slope stabilization and erosion control was conducted, as an operational practice, on some harvested sites at Golden.
All clearcut sites were planted with lodgepole pine, Douglas-fir, and interior spruce seedlings at 1- or 2-years post-harvest. All sites at Summerland and Golden were spatially segregated to enhance biological and statistical independence [
46] (
Table 1). Forest sites in the CC-FOR study were separated by a mean (±SE) of 1.79 ± 0.49 km (range 0.67–2.96 km). For the major species, very few or no voles or deer mice were captured on more than one grid or line, and hence our sites were considered independent. Sites were not considered independent for the northwestern chipmunk (
Neotamias amoenus).
2.3. Forest-Floor Small Mammal Populations
There were nine species of forest-floor small mammals: five major species that included the deer mouse, southern red-backed vole, long-tailed vole (Microtus longicaudus), meadow vole (M. pennsylvanicus), and northwestern chipmunk; and four less common species: heather vole (Phenacomys intermedius), montane shrew (Sorex monticolus), common shrew (S. cinereus), and western jumping mouse (Zapus princeps). Two small mustelids: the short-tailed weasel (Mustela erminea) and long-tailed weasel (M. frenata) were also captured occasionally. At Summerland, populations were sampled at 3–4-week intervals on grids in clearcut and forest sites in Period 1: May to September or October 1979–1982; Period 2: May to October 1997–2002; Period 3: May or June to September or October 2007–2011; and on index-lines in Period 4: May or June to October 2017–2020. At Golden, populations were sampled on clearcut sites at 4-week intervals with grids in Period 1: May to September 2004–2008; at 4- to 8-week intervals with index-lines in Period 2: May to September 2012–2016; and Period 3: May to September 2016–2019.
One live-trapping grid (1 ha) or index-line was in each site. Grids had 49 (7 × 7) trap stations at 14.3-m intervals with 1 Longworth live-trap at each station. An index-line had 7 stations at 14.3-m intervals with four Longworth live-traps at each station [
47]. Traps were baited with whole oats, a slice of carrot, and cotton as bedding. Each trap had a 30-cm × 30-cm plywood cover for protection from sunlight (heat) and precipitation. Traps were set on the afternoon of day 1, checked on the morning and afternoon of day 2 and the morning of day 3, and then locked open between trapping periods. All animals captured were ear-tagged with serially numbered tags and point of capture recorded [
48]. Animals were released immediately after processing. Unfortunately, the overnight trapping technique resulted in a high mortality rate for shrews. Therefore, shrews were collected, frozen, and later identified according to tooth patterns [
49]. All handling of animals followed guidelines approved by the American Society of Mammalogists [
50] and the Animal Care Committee, University of British Columbia.
2.4. Population Data Analyses
Abundance estimates of the major species (numerically dominant) were derived from the Jolly–Seber (J-S) stochastic model for open populations with small sample size corrections [
51,
52]. Minimum number alive was used to estimate populations of the heather vole; number of individuals was used for the montane shrew, common shrew, and western jumping mouse. We calculated the effective trapped area (ETA) for the major species on each grid based on mean maximum distance moved (MMDM) as a boundary strip method [
53]. Estimates of population size were converted to a density estimate by dividing population estimates for each trapping period by the ETA. At Summerland, mean ETAs (±SE) (ha) in
Periods 1,
2, and
3 for deer mice in clearcut sites was 1.44 ± 0.04 and forest sites was 1.54 ± 0.09; for red-backed voles in forest sites was 1.28 ± 0.04; for long-tailed and meadow voles in clearcut sites was 1.19 ± 0.06; and for chipmunks, where sample size was sufficient, was 2.05 ± 0.02 in clearcut sites and 1.80 ± 0.19 in forest sites. Mean ETAs (±SE) (ha) for
Period 2 in clearcut sites at Golden were 1.40 ± 0.03 for deer mice, 1.27 ± 0.00 for red-backed voles, 1.09 ± 0.05 for long-tailed voles, 1.08 ± 0.00 for meadow voles, and 1.82 ± 0.07 for chipmunks.
Regression relationships of numbers of animals captured on index-lines to numbers on an ETA-adjusted 1-ha grid system were conducted for each of the major species: deer mice (
y = 8.45ln(
x) + 0.69;
R2 = 0.47,
p < 0.01) [
54]; long-tailed vole (
y = 0.57
x0.80;
R2 = 0.62,
p < 0.01) [
55]; and northwestern chipmunk: (
y = −0.14
x2 + 1.57
x;
R2 = 0.51,
p < 0.01). We consider each of these estimates to be a “density index” [
53]. Jolly trappability was calculated according to [
56]. Species richness was the total number of species sampled for the small mammal communities in each site [
52]. Species diversity was based on the Shannon-Wiener index [
52,
57].
2.5. Early Seral Vegetation
Early seral vascular plants were sampled following the method of [
58]. Vegetation responses were coordinated for 3 and 5 years post-harvest at Summerland (
Period 2; 1999 and 2001) and Golden (
Period 2; 2006 and 2008) and again for 1, 2, and 3 years post-harvest at Summerland (
Period 3; 2007–2009) and Golden (
Period 3; 2012–2014). All sampling was conducted in July–August and plant species were identified in accordance with [
59]. No vegetation sampling was done in
Periods 1 or
4 at Summerland nor
Period 4 at Golden.
2.6. Statistical Analyses
In the LTCC study at Summerland, a repeated-measures analysis of variance (RM-ANOVA) [
60] was conducted to determine the effect of clearcutting
treatment:
Period (
1, 2,
3, and
4), and
time: years 1 to 4 post-harvest, on mean annual total abundance, species richness, and species diversity of small mammals in the four years immediately post-clearcutting in each
Period. Where significant treatment effects were detected that also had significant treatment × time interactions over the four
Periods, additional univariate ANOVAs were conducted within individual
Periods.
In the CC-FOR study at Summerland, a RM-ANOVA was also conducted to determine the effect of cumulative clearcutting on mean annual total abundance, species richness, and species diversity of small mammals, as well as time and treatment × time interactions, between clearcut and forest sites over the three Periods (2, 3, and 4). Mean values were calculated for these attributes in Period 1 (one replicate only) for comparison with mean results for Periods 2, 3, and 4 in the RM-ANOVA. A univariate ANOVA was conducted to determine the effect of the two treatments on overall mean trappability for P. maniculatus and N. amoenus in clearcut and forest sites.
In the CC-GR study, a RM-ANOVA was conducted to determine the effect of cattle grazing on mean ground cover and abundance (crown volume index) of herb and shrub layers between grazed sites (Summerland) and ungrazed sites (Golden) for the two comparable Periods where vegetation data were available. This analysis was also done for mean annual total abundance, species richness, and species diversity of small mammals, as well as time and treatment × time interactions, between grazed and ungrazed sites for the three comparable Periods. Where significant treatment effects were detected that also had significant treatment × time interactions over the three Periods, additional univariate ANOVAs were conducted within individual Periods.
For all analyses, homogeneity of variance was measured by the Levene statistic. Mauchly’s
W-test statistic was used to test for sphericity (independence of data among repeated measures) [
61,
62]. For data found to be correlated among years, the Huynh-Feldt (H-F) correction was used to adjust the degrees of freedom of the within-subjects
F-ratio [
63]. Duncan’s multiple range test (DMRT), adjusted for multiple contrasts, was used to compare mean values based on RM-ANOVA results for the LTCC analysis [
64]. In all analyses, the level of significance was at least
p = 0.05 [
65].
5. Conclusions
Over the 42-year period, cumulative clearcutting of the original forest resulted in creation of early successional habitat that ranged from 0% of the landscape in 1977 (pre-clearcutting) to ≥70% in 2020. Overall mean annual abundance of forest-floor small mammals declined despite the availability of early post-seral habitats and much of this decline was owing to loss of M. longicaudus and M. pennsylvanicus, thereby not supporting H1. The deer mouse and northwestern chipmunk contributed to high mean species richness and diversity in Periods 2 and 3 before these metrics declined in Period 4, and hence partly supported H1. Except for Period 1, numbers of small mammals were often similar in clearcut and forest sites, thereby not supporting H2. However, species richness and diversity remained relatively high on clearcut sites in Periods 2 and 3, particularly in the first two years after cutting, before declining to forest levels in Period 4, and hence partly supported H2.
Cattle grazing seemed to have a significant negative effect on overall mean abundance of small mammals in all three comparableyes Periods, and hence supported the small mammal part of H3. However, this difference was apparently not related to cattle consumption of herbaceous vegetation but was possibly related to the lower abundance of shrubs in grazed than ungrazed sites. Thus, the vegetation part of H3 was supported for shrubs but not herbs. The decline and near disappearance of both species of Microtus was possibly related to the reduction in plant community abundance and structure from grazing (at least for shrubs) and potentially from drought effects associated with climate change. Loss of microtines from these early seral ecosystems may have profound negative effects on various ecological functions and predator communities.