Edge-related responses of understory plants to aggregated retention harvest in the Pacific Northwest

(click here for full paper)

Introduction

Aggregated retention of overstory trees is now a standard component of timber harvest prescriptions on federal lands in the Pacific Northwest. Patches of remnant forest (hereafter, “forest aggregates”) retained during harvest are thought to enhance the structural and biological diversity of managed forests, but the extent to which they maintain components of the original understory or promote recovery in adjacent harvest areas has not been tested. Although small forest aggregates can have high conservation value, their ecological functions may be diminished as a result of fragmentation effects or edge influences. We examined short-term (1- and 2-yr) responses of understory plants to disturbance and creation of edges in structural retention harvest units at two of the DEMO experimental blocks in western Washington. Our design utilizes pre- and post-treatment measurements of permanent plots allowing us to reliably quantify the spatial pattern, magnitude, and time course of vegetation response. We pose the following questions:

• Do species richness and community composition remain stable in forest aggregates?
• Do aggregates retain disturbance-sensitive herbs that decline in, or are lost from, adjacent areas of harvest?
• Within forest aggregates, are there edge-related gradients in vegetation response (changes in species richness, community composition, or abundance of individual species) and, if so, do these gradients correlate with changes in light availability or disturbance?

Methods

Field methods

Studies were conducted in the 40% aggregated retention treatments at Butte and Paradise Hills (see Study Areas). Pre-treatment sampling was conducted in 1996 and post-treatment sampling in 1998 and 1999 (years 1 and 2). At each site, two of the five 1-ha circular aggregates marked for retention were randomly selected. In each of these, we established four perpendicular transects, 81 m in length, which extended in cardinal directions from the aggregate center and ended 25 m into the surrounding area to be harvested (Figure 1). Twelve bands of permanent plots were established along each transect, eight in the area marked for retention (at distances of 0, 5, 10, 15, 20, 30, 40, and 50 m from the edge) and four in the area marked for harvest (at distances of 5, 10, 15, and 25 m from the edge). Each band consisted of five, 1-m² subplots, within which we estimated the cover of all vascular plant species.

To explore possible correlates of vegetation change, we quantified cover of logging slash and disturbed soil (year 1) and light availability (year 2). Cover of logging slash and disturbed soil was estimated along the interior edge of each band (Figure 1), using the line-intercept method. Light availability was estimated with a CI-110 digital canopy imager with a 150-degree lens. Digital photographs were taken from the end points of each band (Figure 1) at a height of 1 m from the ground surface. Digital images were analyzed using Scanopy 2.0b software to calculate percent open sky.

Figure 1. Position of forest aggregates within a 13-ha harvest unit (a). Data were collected in two, undisturbed forest aggregates (b) at each site, along transects (c) originating at the center of each aggregate and ending 25 m into the surrounding harvested area. Twelve bands (d) were established at 5-10 m intervals along each transect; each band consisted of five, 1-m2 subplots (e) within which we estimated cover of all vascular plant species. Cover of logging slash and disturbed soil were sampled along the interior edge of each band (f) and light availability at the two interior endpoints (g).

Analyses

We considered three types of response variables: richness, community composition, and species abundance (percent cover). To compare the contributions of the original forest flora and open-site species, separate richness calculations were made for species classified as “forest understory” and “early-seral.” Changes in community composition were expressed as the percent dissimilarity (PD) between pre- and post-treatment measurements, using the quantitative form of Sørensen’s community coefficient. To standardize for spatial variation in species richness and abundance prior to treatment, a “change value” was computed for each variable as the arithmetic difference between pre- and post-treatment values.

Responses within forest aggregates and adjacent areas of harvest. -- We assessed the comparative responses of vegetation in forest aggregates and adjacent areas of harvest by conducting a series of two-sample t-tests using mean “change values” (or mean PD for community composition) as the measure of response. Tests of individual species’ responses were limited to 29 of the more common taxa.

Edge-related gradients in vegetation response and physical environment within aggregates.-- Edge-related gradients in vegetation response were assessed by calculating Spearman rank correlation coefficients between mean values of vegetation variables (changes in richness and cover, and PD) and distance from the aggregate edge, with separate analyses for each post-treatment year. Species-level analyses were limited to the 29 common taxa noted above. Environmental variables (open sky, logging slash, and disturbed soil) were also correlated with distance from edge and with vegetation responses.

Results

Differences between forest aggregates and adjacent areas of harvest

• Percent open sky and cover of logging slash were significantly lower (P<0.002) in forest aggregates than in adjacent harvested area; cover of disturbed soil did not differ significantly between environments.
• Two years after creation, aggregates remained stable with respect to forest species richness and were largely resistant to colonization by early-seral species (Figure 2).

Figure 2. Species area curves for (a) forest understory species (before and 2 yr after treatment) and (b) early-seral species (1 and 2 yr after treatment), in forest aggregates (squares) and harvest areas (circles). Points along curves (from left to right) represent means for bands, transects, individual aggregates or surrounding harvest areas, sites, and both sites combined.

• Changes in community composition (percent dissimilarity, PD) were significantly greater in harvest areas than in aggregates, with levels of significance increasing from year 1 to 2 (P=0.003 and <0.001, respectively).
• Eight of 29 common species showed significant differences in response between environments: seven of these showed greater magnitudes of decline in harvest areas than in aggregates; only one showed a significantly greater increase in abundance in harvest areas than in aggregates (Figure 3).

Figure 3. Relative change in cover of individual species in forest aggregates and harvest areas, 2 yr after treatment. Note different scales of x and y axes. Closed circles and bold labels indicate species with significant differences between environments (based on t-tests), and open circles non-significant relationships. Species codes are: Ace_cir = Acer circinatum, Ach_tri = Achlys triphylla, Ane_del = Anemone deltoidea, Ber_ner = Berberis nervosa, Bro_vul = Bromus vulgaris, Chi_men = Chimaphila menziesii, Chi_umb = C. umbellata, Cli_uni = Clintonia uniflora, Cor_can = Cornus canadensis, Gau_ova = Gaultheria ovatifolia, Goo_obl = Goodyera oblongifolia, Hie_alb = Hieracium albiflorum, Lin_bor = Linnaea borealis, Lis_cau = Listera caurina, Pte_aqu = Pteridium aquilinum, Pyr_asa = Pyrola asarifolia, Pyr_pic = P. picta, Pyr_sec = P. secunda, Ros_gym = Rosa gymnocarpa, Rub_las = Rubus lasiococcus, Rub_urs = R. ursinus, Smi_rac = Smilacina racemosa, Smi_ste = S. stellata, Tri_ova = Trillium ovatum, Vac_mem = Vaccinium membranaceum, Vac_ova = V. ovalifolium, Vac_par = V. parvifolium, Vio_sem = Viola sempervirens, and Xer_ten = Xerophyllum tenax. Only those species present prior to treatment in ≥ 3 aggregate/harvest area pairs and 10% of all sample bands are illustrated.

Gradients in environment and vegetation response within forest aggregates

• Percent open sky and cover of logging slash increased significantly with proximity to forest edge (P<0.001); however, these increases were largely restricted to a distance of ca. 10-15 m from the edge (Figures 4a and b). Cover of disturbed soil did not show a significant correlation with proximity to edge, although there were several relatively high values within 5 m of the forest margin (Figure 4c).

Figure 4. (a) Open sky, (b) cover of logging slash and (c) cover of disturbed soil with distance from the edges of forest aggregates. Lines represent mean values (n=16) at each sampled distance. Points in the shaded region represent bands within the aggregates.

• Proximity to edge explained little of the variation in the change in richness of forest species; however, the strength of this relationship increased with time (year 1, r=-0.32, P=0.073; year 2, r=-0.44, P=0.011; Figure 5).

Figure 5. Changes in richness of forest understory (circles) and early-seral (squares) species with distance from the edges of forest aggregates. Values represent mean differences (n=16) in band-level richness between pre- and post-treatment measurements (closed symbols = year 1, open symbols = year 2). Points in the shaded region represent bands within the aggregates.

• Changes in community composition (PD) increased significantly with proximity to edge (year 1, r=0.41, P=0.022; year 2, r=0.73, P<0.001), due in large part to changes at the forest margin (0-5 m) (Figure 6).

Figure 6. Changes in community composition with distance from the edges of forest aggregates. Values represent mean percent dissimilarity (n=16) between pre- and post-treatment composition at each distance (closed circles = year 1, open circles = year 2). Points in the shaded region represent bands within the aggregates.

• Herbaceous species were more affected by proximity to edge than were shrubs, and the number and strength of significant relationships increased with time: declines were significant for three herbaceous species in year 1 and eight in year 2, but not for any shrub in either year (Figure 7a). Of the eight species that showed significant negative correlations between change in cover and edge proximity (Figure 7a), five also showed significant negative correlations with light availability (Figure 7b) and six with cover of logging slash (Figure 7c).

Figure 7. Spearman rank correlation coefficients (r, n=32) between change in species abundance within forest aggregates and (a) edge proximity, (b) percent open sky, or (c) cover of logging slash, one and two years after treatment. For edge analyses, negative correlations indicate a decline with proximity to forest edges. Closed circles indicate significant first-year relationships, and open circles significant second-year relationships. Absence of circles indicates non-significant relationships. Arrows illustrate the direction and magnitude of change in response from year 1 to 2. Only those species present prior to treatment in >3 aggregate/harvest area pairs and 10% of all bands were tested, and only those species with significant relationships (P=0.05) are illustrated.

Discussion

Forest aggregates vs. adjacent harvest areas

Compared to harvested areas, forest aggregates showed minimal change in species richness and composition two years after treatment. Aggregates were largely resistant to colonization by early-seral species, and changes in composition were small compared to those in adjacent harvest areas. In the short-term, forest aggregates provide refugia for shade-tolerant herbs that are extirpated from, or decline in, adjacent areas of harvest: one quarter of the species tested showed significantly greater declines in harvest areas than in aggregates, and two formerly common species, Chimaphila menziesii and Listera caurina, disappeared from harvest areas at Paradise Hills. Because most forest species do not maintain a viable seed bank, local persistence in and subsequent dispersal of seeds from aggregates may greatly facilitate reestablishment of populations in harvested areas.

Gradients in response within forest aggregates

Within forest aggregates, increased light availability and harvest-related disturbance were limited to a 10- to 15-m-wide band, leaving approximately 50% of the forest aggregate unchanged for these attributes. However, this large outer band was notably altered, with logging slash covering 38% of the ground surface and open sky roughly double that at the center of the aggregate.

Spatial gradients in community composition, species richness, and the abundance of individual forest species correlated to varying degrees with proximity to forest edge. Changes in community composition were most apparent at the forest border (0-5 m). We also found slightly reduced richness at the edge, reflecting declines of some forest species and minimal establishment of early-seral species. We anticipate gradual increases in richness near aggregate edges with time, as early-seral species become more abundant in adjacent harvested areas.

Although none of the common shrub species showed significant response, eight of 23 common herbs declined near edges. A marked increase from year 1 to 2 in both the number of species showing significant declines and the magnitude of decline suggests that edge effects will become more apparent with time. Most forest herbs are clonal and, during short periods of unfavorable resource conditions, many are capable of drawing upon nutrient reserves or of physiological integration among ramets. With time, however, reserves may be depleted and rhizome connections may decay.

Management considerations

Identifying minimum sizes for protected areas is an important issue in conservation biology. Although large reserves are clearly necessary for many ecosystem processes and components (e.g., interior-forest microclimate and wide-ranging carnivores), smaller forest remnants also may have high conservation value, especially in landscapes that are intensively managed for timber production. Our results suggest that, over short timeframes, aggregates of at least 1 ha in size may play an important role in protecting late-seral plant species through retention harvest of mature, Douglas-fir forest. However, temporal trends suggest that edge effects judged to be small in the short term may become more prominent with time. Additional research at these and other sites in the Pacific Northwest is necessary to identify the temporal and spatial scales over which forest aggregates serve their intended ecological functions.