Oak Woodland Restoration: Understory Response to Removal of Encroaching Conifers Warren D

Woodland Research Report

Oak Woodland Restoration: Understory Response to Removal of Encroaching Conifers Warren D. Devine, Constance A. Harrington, and David H. Peter

Abstract Oregon white oak (or Garry oak, Quercus garryana) woodlands and savannas of the coastal Pacific Northwest are legacies of an anthropogenic fire regime that ended with European settlement in the mid-1800s. Historically, these oak stands had a sparse overstory and an understory dominated by fire-tolerant grasses and forbs. Post-settlement fire suppression resulted in widespread invasion and subsequent overstory dominance by conifers, causing mortality of shade-intolerant oak trees and shifting understory plant communities to shade-tolerant species. In a study on four southwestern Wash- ington sites, our objective was to determine the effects of overstory conifer removal, primarily Douglas-fir (Pseudotsuga menziesii), on microclimate, native and non-native understory cover, and sapling growth. Overstory conifer removal created a warmer, drier understory microclimate during summer months. Conifer removal had little effect on native understory cover during five years post-treatment; however, cover of non-native plants, primarily grasses and woody understory species, increased significantly during the same period. Height growth of Oregon white oak and Douglas-fir saplings exhibited a delayed, but positive, response to overstory conifer removal, although the treatment response of Douglas-fir was 133% greater than that of oak. Increases in non-native understory cover and the rapid growth of young Douglas-fir indicate the importance of pre- and post-treatment understory management to control undesirable plants and promote native species such as Oregon white oak.

Keywords: Douglas-fir (Pseudotsuga menziesii ), microclimate, Oregon white oak (Quercus garryana), restoration, savanna, understory

Introduction Harrington 2006). Conifers overtop centuries to replace due to their slow and suppress the shade-intolerant oak growth (Stein 1990). Where former he Oregon white oak (or Garry trees, leading to crown die-back and canopy oak trees are already dying as oak; Quercus garryana) wood- T eventual mortality of the oak. a result of suppression, promoting the lands, savannas, and associated prairies Restoration of Oregon white oak growth and survival of oak seedlings of the coastal Pacific Northwest are woodlands and savannas to presettle- and saplings becomes critical. part of an anthropogenic ecosystem, ment conditions requires removal of This encroachment of conifers is historically maintained by frequent, the conifer overstory to “release” the widespread, and substantial losses low-intensity fires set by Native Amer- suppressed oak trees from competition of prairie and oak habitats also have icans (Agee 1993). After European and to restore a more open stand struc- resulted from conversion of land to settlement in the mid-1800s, the lack ture (Harrington and Devine 2006). agricultural and urban uses (Crawford of regular burning allowed fire-intol- Results from a companion study have and Hall 1997). In British Columbia, erant conifers, primarily Douglas-fir shown that large oak trees (diameter less than five percent of pre-settlement (Pseudotsuga menziesii), to regenerate at breast height over 20 cm) respond Oregon white oak habitat remains (Lea in prairies and oak stands where they to release from overtopping Douglas- 2006) (Figure 1). Recognizing the loss were previously excluded (Thysell and fir by increasing their rate of growth, of this ecosystem, private landowners, Carey 2001). Growth and survival of expanding leaf area through formation conservation organizations, and state Oregon white oak trees are substan- of new (i.e., epicormic) branches, and and federal land managers in Oregon’s tially reduced where these conifers increasing acorn production (Devine Willamette Valley, the Puget Sound have invaded (Stein 1990, Devine and and Harrington 2006). Preservation region of Washington, and Vancouver of existing oak trees is a priority when Island, British Columbia, have become Ecological Restoration Vol. 25, No. 4, 2007 ISSN 1522-4740 E-ISSN 1543-4079 restoring these stands because the trees involved in various aspects of restoring ©2007 by the Board of Regents of the provide an important structural com- Oregon white oak communities. Two University of Wisconsin System. ponent and would take decades or non-profit organizations, the Oregon

December 2007 Ecological Restoration 25:4 • 247 colonize these former oak savanna and woodland sites. The Douglas-fir had been lightly thinned two to three times before this study at 10- to 15-year intervals, most recently immediately prior to the study (2000–2001). Stand basal area averaged 34.3 ± 8.2 m2/ha for Douglas-fir and 4.7 ± 2.4 m2/ha for oak. Approximate age range for the suppressed oak was 90 to 150 years, while Douglas-fir averaged 65 to 90 years. A detailed description of these oak trees appears in Devine and Harrington (2006). The understory of the oak woodland and savanna stands in the Puget Sound region was historically dominated by grasses and forbs in a plant community likely similar to the Oregon white Figure 1. An Oregon white oak (Quercus garryana) woodland (dormant season) with encroaching oak, long-stolon sedge (Carex inops), Douglas-fir Pseudotsuga( menziesii ) in the background. In the foreground is our most problematic invasive shrub, Scotch broom (Cytisus scoparius). Photo by Warren Devin common camas (Camassia quamash) community (Chappell and Craw- ford 1997). A lack of regular fire and Oak Communities Working Group information on understory response increasing conifer dominance results and the Garry Oak Ecosystems Recov- will facilitate selection of appropriate in a more shade-tolerant, shrubby ery Team (British Columbia), have restoration techniques in similar plant understory community such as the focused on preserving and restoring communities. Oregon white oak, Douglas-fir/snow- Oregon white oak ecosystems since the berry (Symphoricarpos albus), sword- late 1990s. When Oregon white oak Study Area fern (Polystichum munitum) commu- is released from overtopping conifers, nity (Chappell and Crawford 1997) the intensity of release ranges from The study took place in four stands of which was present at our study sites. individual-tree treatment (removing Oregon white oak and Douglas-fir on The most common understory spe- only the conifers adjacent to selected the Fort Lewis Military Reservation cies present across all treatments and oak trees) to total removal of conifers in southwestern Washington, east of years are listed in Table 1. Control at the stand level. These practices are the city of Tacoma. The four sites, of the invasive shrub Scotch broom influenced by the fact that on many located 10 to 15 km apart, are on gla- (Cytisus scoparius) is part of nearly all private lands the invading conifers cial outwash terraces and moraines, at oak savanna and prairie restoration represent a source of timber revenue elevations from 85 to 135 m. Soils are efforts in the region; we performed while the oak trees may have little gravelly to very gravelly sandy loams, a one-time removal of Scotch broom monetary value. Thus, depending on and loamy fine sands, and are moder- from the study by cutting (at ground- the merchantability of the conifers, ately to somewhat excessively drained. line to prevent sprouting) or uprooting a release treatment may require an Annual precipitation in Tacoma is 995 all plants within one tree-height radius expenditure, allow the owner to break mm, but total precipitation from 1 of the study trees at the beginning of even, or even yield a profit. May through 30 September averages the study in June 2001. The number The goal of this study was to only 158 mm (WRCC 2005). Mean of Scotch broom plants per study tree evaluate the influence of oak release temperatures in January and July are was variable, averaging 43 ± 80 and treatments on the understory, as 5 and 19°C respectively. ranging from zero to 350. the effects of release are currently The four stands were similar in spe- not well understood. We compared cies composition, with an overstory Study Design and primarily composed of Douglas-fir overtopped and released conditions Sampling to determine how removal of coni- averaging 167 ± 46 trees/ha (mean ± fers affects microclimate, native and one standard deviation) and a mid- At each of four sites, we selected 18 non-native understory cover, and story of suppressed Oregon white oak oak trees for the study that were each growth of Oregon white oak and averaging 115 ± 66 trees/ha. This is overtopped by at least two Douglas-fir Douglas-fir saplings. Our hope is that the first generation of Douglas-fir to trees. Average height of these 72 oak

248 • December 2007 Ecological Restoration 25:4 Table 1. Shade tolerance and percent coverage in years 1 and 5 by basal area. The full-release treatment treatment for species occurring on at least 5% of microplots. Shade tolerance represents a near-total removal of classification is tolerant (T), intermediate (M), or intolerant (I). The symbol * overstory competition, while the half- indicates a significant change in cover p( < 0.05) between year 1 and year 5. release represents an individual-tree Cover (%) release that has been applied by some Control Full Release landowners to prolong the life of oak Shade Group / Species Yr 1 Yr 5 Yr 1 Yr 5 trees while removing relatively few Tolerance overstory conifers. During thinning Native forbs and treatment, trees were cut with Polystichum munitum (swordfern) T 16.6 20.1 7.2 12.1 chainsaws and moved to the landings Galium aparine (stickywilly) M 7.0 1.6* 6.0 3.4 with skidders. Clinopodium douglasii (yerba buena) T 2.8 2.0 1.1 2.2 We measured air temperature and Nemophila parviflora M 2.4 0.3 2.8 1.2 relative humidity every two hours near (smallflower nemophila) 18 randomly selected full-release and Fragaria vesca (woodland strawberry) M 0.5 1.1 1.1 1.6 control treatment trees using HOBO® Pro dataloggers (Onset Computer Native grasses Carex inops (long-stolon sedge) M 1.9 2.1 2.2 3.6 Corp., Bourne, MA). Dataloggers Bromus vulgaris (Columbia brome) M 1.1 0.5 2.0 1.3 were mounted 25 cm above the forest floor, 1 m south of the study oak tree. Native woody species Vapor pressure deficit (VPD), which Symphoricarpos albus (snowberry) M 32.7 42.4* 25.5 37.2* influences plant-water relations and Rubus ursinus (California blackberry) M 32.6 36.9 24.1 41.6* photosynthesis rate (Elliott and Vose Corylus cornuta var. californica 1994, Singsaas et al. 2000), was cal- T 13.8 16.3 14.5 20.2* (California hazelnut) culated from relative humidity and Mahonia aquifolium (tall Oregongrape) M 2.5 3.7 2.3 6.7* air temperature (Lee 1978). Data Lonicera ciliosa (orange honeysuckle) M 4.9 7.4 3.7 5.8 presented are from 2002 and 2003 and are representative of the collec- Non-native forbs tion period (July 2001 through May Hypochaeris radicata (hairy catsear) M 0.7 1.5 0.6 2.2 2005). Hypericum perforatum M 0.3 0.5 0.4 1.5* We conducted vegetation surveys (common St. Johnswort) near 24 study trees in the full-release Mycelis muralis (wall-lettuce) M 0.4 0.6 1.8 0.6* and control treatments in June of 2001, 2003, and 2005 (study years Non-native grasses one, three, and five). We established a Agrostis capillaris (colonial bentgrass) I 2.8 2.2 3.5 8.9* permanent transect from east to west, Holcus lanatus (common velvetgrass) M 0.8 2.2 3.1 8.9* bisected by the study tree, with 20- × Dactylis glomerata (orchardgrass) M 0.8 3.5 1.3 0.8 50-cm microplot frames every two Poa pratensis (Kentucky bluegrass) M 0.0 1.6 0.0 1.2 meters. Transect length was equal to twice the study tree height, and the Non-native woody species number of microplots per transect Cytisus scoparius (Scotch broom) M 0.3 4.2 0.6 14.9* averaged 16 ± 3. Within each microplot, experienced surveyors measured “study trees” was 16.0 ± 3.1 m, and release” treatment was removal of all duff layer thickness and visually esti- crown diameter averaged 7.5 ± 2.3 m. Douglas-fir trees within a distance of mated percent total understory cover, Height of the two largest overtopping one-half tree-height (an average of 8 percent exposed soil, and percent cover Douglas-fir per oak averaged 40.6 ± m) of the study oak tree, an average by species for vascular plants. 6.9 m. of 6 ± 3 Douglas-fir trees per oak. In In spring 2001, we tagged, mapped, We applied three levels of treat- the control treatment, no Douglas-fir and measured the height of naturally ment, replicated six times per study trees were removed. An average of one established oak and Douglas-fir sap- site, in April and May of 2001. The to two Douglas-fir trees within one lings (height = 0.5 to 3.0 m; n = 288) “full-release” treatment was removal of tree-height radius of each study tree within one tree-height of the study all Douglas-fir trees within a distance was removed across all treatments in trees. We remeasured the height of of one tree-height (an average of 16 m) a light commercial thinning that took these saplings after each growing of the study tree, an average of 15 ± 8 place prior to the study in 2000–2001, season from 2001 to 2005. For oak Douglas-fir trees per oak. The “half- removing approximately 15% of stand saplings, we classified the level of

December 2007 Ecological Restoration 25:4 • 249 understory competition in 2002 and Daily maximum 6 crown shape in 2003. Crown shape Daily mean A was classified as 1) a vertically oriented Daily minimum crown with one dominant leader; 2) a 4 vertically oriented crown with multiple competing leaders; or 3) a flat-topped, horizontally oriented crown. 2 The study followed a randomized complete-block design (site = block). Data were analyzed by analysis of vari- 0 ance and correlation analysis (details of analysis available upon request). Air temp. treatment difference (°C) difference treatment temp. Air Significance was judged at the 95% -2 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec confidence level. Month (2003) Results and Discussion 2.5 Daily maximum Understory microclimate 2.0 Daily mean B The full-release treatment increased maximum daily air temperature, par- 1.5 ticularly from June through August when the maximum temperature 1.0 averaged 5.2°C higher than in the control (Figure 2a). Conversely, 0.5 daily minimum air temperature was slightly but consistently lower in the 0.0 full-release treatment than in the VPD treatment difference (kPa) difference treatment VPD control throughout the year. Mean -0.5 daily air temperature did not differ Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec between the full-release and control Month (2003) treatments. Maximum daily VPD was 5 35 significantly higher in the full-release Full release treatment than in the control during Control C 30 summer months, particularly from 4 Air temp. June through August (Figure 2b). Mean daytime VPD did not differ by 25 3 treatment. Air temperature and VPD during a typical clear summer day (12 20 VPD 2

July 2002) are shown in Figure 2c. (kPa) VPD Both air temperature and VPD peaked 15 slightly earlier in the day in the full- 1 (°C) temperature Air release treatment than in the control. 10 We assessed direct sunlight reach- ing the study trees and found that 0 5 the half-release and full-release treat- 06:00 08:00 10:00 12:00 02:00 04:00 06:00 08:00 ments approximately doubledFigure and 2 Time of day quadrupled, respectively, the duration Figure 2. The effect of full release, relative to the control treatment on A) average daily maxi- of direct sunlight that the oak trees mum, mean, and minimum air temperatures at 25 cm (positive values indicate full release received each day in summer (Devine was warmer than the control and negative values indicate full release was cooler); B) daytime maximum and mean vapor pressure deficit (VPD) at 25 cm. C) mean VPD (solid line) and air and Harrington 2006). The over- temperature (dashed line) for control and full-release treatments on a typical clear summer day topped oak trees in the control treat- (12 July 2002). ment received, on average, only 13% of the photosynthetically active radiation that would have been available

250 • December 2007 Ecological Restoration 25:4 Figure 3 ) 50 30 140 Treatment 25 Year * 120 40 Treat. x year 100 20 30 80 15 60 20 10 Full release 40 Control 10 Treatment Treatment 5 20 Year *** Native forbs total cover (%) cover total forbs Native Year Treat. x year Treat. x year *** 0 graminoids (%) Native total cover 0 0 Native woody species total cover (% cover total woody species Native 1 3 5 1 3 5 1 3 5 Year Year Year

50 30 140 Treatment * Treatment * Treatment Year 25 Year *** 120 Year *** 40 Treat. x year Treat. x year Treat. x year * 100 20 30 80 15 60 20 10 40 10 5 20 Non-native forbs total cover (%) cover total forbs Non-native 0 0 0 Non-native graminoids total cover (%) graminoids total cover Non-native

1 3 5 1 3 5 woody (%) Non-native totalspecies cover 1 3 5 Year Year Year Figure 3. Total cover (± SE) of native and non-native forbs, grasses, and woody understory species for two treatments and three sampling periods. Significance of fixed effects atp -values of 0.05 and 0.001 is denoted by * and ***, respectively.

under unobstructed sunlight (Devine Native and non-native resulted in a temporary but significant and Harrington 2006). understory cover reduction in understory cover and an Our microclimate findings are in increase in exposed soil (Figure 4). Cover of native and non-native veg- agreement with other studies that Soil seed banks in 40- to 60-year-old etation followed different trends after found greater diurnal fluctuation in conifer forests in the Pacific Northwest treatment. Cover of native forbs and air temperature and greater VPD in have been found to contain a signifi- grasses was not significantly changed openings relative to a closed-canopy cant fraction of non-native grasses by the full-release treatment compared forest (Carlson and Groot 1997, Gray and forbs, and disturbance may favor to the control, but cover of non-native et al. 2002, Ritter et al. 2005). The establishment of these non-native grasses, and, to a lesser extent, non- increases in near-ground maximum species in the understory (Halpern native forbs, was increased by the full- air temperature, VPD, and solar radia- et al. 1999). Because the understory release treatment (Figure 3, Table 1). tion, resulting from removal of the disturbance and the removal of the Non-native grass cover also increased conifer overstory, created an under- Douglas-fir overstory in our study substantially over time from year one story microclimate that is likely more appear to have created an environ- to year five. Cover of native woody similar to that of historical conditions. ment favorable to non-native, rud- species was reduced in the full-release The return of a warmer, drier under- eral species (i.e., those common to treatment compared to the control story with greater solar radiation is disturbed areas), a strategy to control in year one, likely due to disturbance important not only for reestablish- undesirable plant species may be most during the release operation. Cover of ing native plant species that formerly effective if initiated prior to the release non-native woody species (predomi- existed in that environment, but also operation (Hanna and Dunn 1997). nantly Scotch broom) increased over for reducing the moisture content of On particularly sensitive sites, aerial time, and in year five was significantly fuels and thus creating conditions that harvesting systems may reduce forest greater in the full-release treatment will facilitate a prescribed burning floor disturbance, although costs are than in the control. regime similar to that which existed much greater than those associated The release operation used ground- historically (Agee 1993, Tveten and with ground-based systems (Stokes based harvesting equipment that inev- Fonda 1999). and Schilling 1997). itably disturbed the forest floor. This

December 2007 Ecological Restoration 25:4 • 251 110 Full release the release area as well as litter from five. Scotch broom is an invasive spe- Control the understory. The current accumula- cies of particular concern in restoration 100 tion of duff is certainly much greater of the region’s prairies and savannas as than the organic layer that was present it establishes and spreads rapidly and 90 prior to settlement, under a regime dramatically changes the understory of frequent, low-intensity fire. Thus, structure (Parker et al. 1997, Parker 80 this duff layer presents different seed- 2000). Repeated control treatments

70 Treatment ** bed and fuel conditions (Tveten and (e.g., burning or pulling) offer the Understory cover (%) cover Understory Year *** Fonda 1999). best chance to reduce Scotch broom Treat. x year *** 60 A site may become susceptible to populations, although its seed bank 1 3 5 invasion by non-native species when may endure for decades (Alexander Year unused resources increase (Tilman and D’Antonio 2003). In our study, 25 1985 Davis et al. 2000). In this study, proximity to seed source influenced Treatment ** removal of overstory Douglas-fir caused the spread of Scotch broom after Year *** 20 Treat. x year *** a rapid increase of several resources in Douglas-fir removal. For example, the understory: direct sunlight, grow- where Scotch broom plants occurred 15 ing space, soil water, and possibly soil along forest roads in the vicinity of N (Devine and Harrington 2006, released trees, there appeared to be 10 2007). Among the three understory a greater likelihood of the species’ growth forms, the non-native grasses encroachment upon those trees. Exposed soil (%) soil Exposed 5 appeared to increase most rapidly in Oak and Douglas-fir response to the increased resource 0 availability (Figure 3). Coverage of regeneration 135both colonial bentgrass (Agrostis capil- During the five years following treat- Year laris) and common velvetgrass (Holcus ment, growth of oak saplings increased 8 lanatus) increased significantly over in the full-release treatment relative to time in the full-release treatment, the control, while growth in the half-

6 but not in the control (Table 1). The release treatment remained intermedi- moderate increase in non-native grass ate (Figure 5). Growth of oak across all cover in the control treatment may treatments was significantly affected 4 have been associated with the com- by the level of understory competition mercial thinning that removed scat- (Figure 6). Overtopped saplings grew 2 tered Douglas-fir trees prior to the significantly less than those that had Treatment ** Duff thickness (cm) Year *** study. In southwestern British Colum- more direct sunlight. Establishment Treat. x year bia, disturbance of the understory in and vigor of oak can be increased by 0 removing understory and midstory 135Oregon white oak savannas led to vegetation to provide greater access Figure 4 Year overall increases in invasive species and shifts in abundance of native and non- to sunlight and soil water (Janzen Figure 4. Total percent cover (± SE) of under- native grasses as different competitive and Hodges 1985, Davis et al. 1998, story, exposed soil, and duff layer thickness for two treatments and three sampling peri- strategies were favored (MacDougall Rey-Benayas et al. 2005). On similar ods. Significance of fixed effects atp -values 2002, MacDougall and Turkington sites, height and diameter growth of of 0.01 and 0.001 is denoted by ** and ***, 2004). Following removal of Douglas- planted Oregon white oak seedlings respectively. fir from an Oregon white oak savanna was increased by controlling compet- in Oregon’s Willamette Valley, there ing vegetation using plastic mulch Mean duff layer thickness increased was a rapid increase in abundance within a 60-cm radius of seedlings over time in the full-release and con- of non-native forb and grass species, (Devine et al. 2007). trol treatments, but was consistently although the fraction of non-native Post-treatment growth of oak sap- greater in the control treatment species had begun to decline slightly lings was significantly affected by (Figure 4). These patterns in duff by the fourth year after treatment crown shape. Across all three treat- layer thickness were clearly influ- (Vance et al. 2006). ments, annual growth of flat-topped enced by needle litter from overstory Scotch broom, although removed oak saplings was significantly less Douglas-fir. The increase over time in from around the study trees soon after (5.9 ± 1.0 cm) than saplings with duff layer thickness in the full-release release in year one, comprised 95% of one leader (12.1 ± 1.3 cm) or sap- treatment was apparently due to litter the cover of non-native woody species lings with multiple leaders (12.9 ± from Douglas-fir trees at the edge of in the full-release treatment in year 1.0 cm). Similarly, oak saplings with

252 • December 2007 Ecological Restoration 25:4 Figure 6 Figure 5

45 20 40 Oregon white oak Treatment A 35 Year * 15 A Full release A 30 Treat. x year ** AB Half release 25 Control 10 B 20

15 5 10 2002 height growth (cm) growth height 2002 5

0 0 12345 40 Douglas-fir Treatment Understory competition class 35 Year *** Treat. x year *** Figure 6. Height growth increment (± SE) in 2002 for Oregon white oak 30 (Quercus garryana) saplings in five understory competition classes: 1) no Annual height growth (cm) competition near sapling crown, 2) sapling has three or more years of 25 growth above competition, 3) sapling has one or two years of growth 20 above competition, 4) sapling overtopped; some direct sunlight, and 5) sapling overtopped; little direct sunlight. Same letter denotes no 15 difference at p = 0.05. 10 5 time relative to growth of the conifers 0 in the control and half-release treat- 12345 ments. By year five, Douglas-fir in Year the full-release treatment was growing at more than twice the rate of Figure 5. Annual height growth (± SE) for Oregon white oak and Douglas-fir saplings in three treatments. Significance of fixed effects oak in the same treatment. This dif- at p-values of 0.05, 0.01, and 0.001 is denoted by *, **, and ***, ference in growth rate is typical for respectively. these species when they are exposed to full sunlight, and, assuming it per- flat-topped crowns (white oak (Q. The increase in the rate of oak sists, the Douglas-fir will eventually alba), northern red oak (Q. rubra) and growth in response to release, although overtop and shade the released oak chestnut oak (Q. montana)) in West statistically significant, was relatively trees. A restoration effort that includes Virginia responded slowly after partial small in magnitude, especially during removal of undesired overstory trees canopy removal, although half had the first few years after treatment. The also must include control of young formed a dominant leader four years same was true for Douglas-fir. These trees, and if a seed source remains on post-treatment (Carvell 1967). In our trends may be indications of low root or near the site, repeated treatments study, the growth rate of saplings with to shoot ratios of these saplings prior will be necessary. flat-topped crowns had not improved to release. Although both of these spe- by year five in any of the treatments. cies grow best in full sunlight, as sap- Implications for Practice Saplings that exhibited apical domi- lings they may survive for many years nance (i.e., a dominant terminal shoot beneath a Douglas-fir canopy. Growth Removal of invading conifers is a nec- suppressing other shoots via hormone under such conditions generally favors essary early step in restoring native oak production) responded to release with stem and foliar development, rather woodlands and savannas because of greater height growth throughout the than roots, as a response to the limited the conifers’ substantial effect on the five-year period. In a related study, we light availability (Hodges and Gardiner understory microclimate and species improved growth rates of flat-crowned 1992). Thus, during the initial years composition. However, removal of oak saplings and oak saplings damaged after release, stem growth may have overstory trees will abruptly change during a release operation by cutting been negatively affected as resources the microclimate and forest floor saplings near the base and allowing were instead allocated to root develop- condition, making the site suscep- the stump to sprout. Three-year sprout ment in response to the increased light tible to colonization by invasive spe- growth averaged 86 ± 34 cm, while (Beon and Bartsch 2003). cies. Control of particularly undesired growth of untreated saplings averaged Annual growth of Douglas-fir in the species may be best initiated prior 25 ± 15 cm. full-release treatment increased over to the overstory treatment because

December 2007 Ecological Restoration 25:4 • 253 heavy machinery will likely expose Jim Rohde (both retired) for their work in supply and demand. Journal of Ecology soil, creating a seedbed prone to colo- installing the study. We thank members of 86:652–661. nization. Dispersal of seed via log- the Silviculture and Forest Models Team for Devine, W.D. and C.A. Harrington. 2006. their assistance with field and office work Changes in Oregon white oak (Quercus ging machinery and vehicular traffic and Christel Kern for her work in the estab- garryana Dougl. ex Hook.) following may be reduced by cleaning vehicles lishment of this study. We thank Jeffrey release from overtopping conifers. Trees and equipment before entering a site Foster, Roberta Davenport, and Peter Gould 20:747–756. (Dalsimer 2002). for reviewing a draft manuscript. _____. 2007. Release of Oregon white oak Changes in the plant understory from overtopping Douglas-fir: Effects on soil water and microclimate. Northwest five years after release from Douglas-fir References indicate that post-release treatments Science 81:112–124. Agee, J.K. 1993. Fire Ecology of Pacific will be necessary to control undesir- Devine, W.D., C.A. Harrington and Northwest Forests. Washington, DC: L.P. Leonard. 2007. Post-planting able plant species and favor the native Island Press. treatments increase growth of Oregon plant community. While prescribed Alexander, J.M. and C.M. D’Antonio. 2003. white oak (Quercus garryana Dougl. ex burning is the traditional method of Control methods for the removal of Hook.) seedlings. Restoration Ecology maintaining savannas and prairies in French and Scotch broom tested in 15:212–222. the Pacific Northwest, it is a logistical coastal California. Ecological Restoration Elliott, K.J. and J.M. Vose. 1994. challenge to implement with regular- 21:191–198. Photosynthesis, water relations, Beon, M.S. and N. Bartsch. 2003. Early ity or on a large scale. Alternative tools and growth of planted Pinus strobus seedling growth of pine (Pinus L. on burned sites in the southern for restoring native vegetation include densiflora) and oaks (Quercus serrata, Q. Appalachians. Tree Physiology mowing, tilling, and applying herbi- mongolica, Q. variabilis) in response to 14:439–454. cide to undesirable species (Schuller light intensity and soil moisture. Plant Ewing, K. 2002. Effects of initial site 1997, Tveten and Fonda 1999, Ewing Ecology 167:97–105. treatments on early growth and three- 2002, MacDougall et al. 2004). Four- Carlson, D.W. and A. Groot. 1997. year survival of Idaho fescue. Restoration year results from a restoration study Microclimate of clear-cut, forest interior, Ecology 10:282–288. show that native grasses and forbs may and small openings in trembling aspen Gray, A.N., T.A. Spies and M.J. Easter. forest. Agricultural and Forest Meteorology 2002. Microclimate and soil moisture be established by sowing and plant- 87:313–329. responses to gap formation in coastal ing, respectively, following overstory Carvell, K.L. 1967. The response of Douglas-fir forests. Canadian Journal conifer removal from an oak savanna understory oak seedlings to release after of Forest Research 32:332–343. (Vance et al. 2006). partial cutting. Agricultural Experiment Halpern, C.B., S.A. Evans and S. Nielson. The growth of oak saplings after Station Bulletin 553. Morgantown: 1999. Soil seed banks in young, closed- overstory Douglas-fir removal was University of West Virginia. canopy forests of the Olympia Peninsula, affected by competition from the Chappell, C.B. and R.C. Crawford. Washington: Potential contributions 1997. Native vegetation of the South understory and by sapling crown to understory reinitiation. Canadian Puget Sound prairie landscape. Pages Journal of Botany 77:922–935. shape. Oak saplings would prob- 107–122 in P. Dunn and K. Ewing Hanna, I. and P. Dunn. 1997. Restoration ably benefit from control of adjacent (eds.), Ecology and Conservation of the goals for Oregon white oak habitats in understory competitors until the oak South Puget Sound Prairie Landscape. the south Puget Sound region. Pages grows above the understory. Oak sap- Seattle, WA: The Nature Conservancy 231–246 in P. Dunn and K. Ewing lings with dominant leaders are likely of Washington. (eds.), Ecology and Conservation of the to grow above understory competi- Crawford, R.C. and H. Hall. 1997. South Puget Sound Prairie Landscape. Changes in the South Puget Sound tors more quickly than saplings with Seattle, WA: The Nature Conservancy prairie landscape. Pages 11–15 in P. of Washington. flat-topped crowns. Because Douglas- Dunn and K. Ewing (eds.), Ecology and Harrington, C.A. and W.D. Devine. 2006. fir grows rapidly following overstory Conservation of the South Puget Sound A practical guide to oak release. USDA removal, continuing efforts to control Prairie Landscape. Seattle, WA: The Forest Service General Technical Report this species must take place to pre- Nature Conservancy of Washington. PNW-666. vent it from regaining dominance of Dalsimer, A. 2002. Why DOD is interested Hodges, J.D. and E.S. Gardiner. 1992. the site. Control of this regeneration in invasive species. Federal Facilities Ecology and physiology of oak is most easily implemented and least Environmental Journal 13:41–54. regeneration. Pages 54–65 in D. Davis, M.A., J.P. Grime and K. Thompson. Loftis and C. McGee (eds.), Oak expensive when these trees are still 2000. Fluctuating resources in plant regeneration: Serious problems, practical small. communities: A general theory recommendations. USDA Forest Service of invasibility. Journal of Ecology General Technical Report SE-84. Acknowledgments 88:528–534. Janzen, G.C. and J.D. Hodges. 1985. Davis, M.A., K.J. Wrage and P.B. Reich. Influence of midstory and understory We thank the Fort Lewis Forestry Program 1998. Competition between tree vegetation removal on the establishment for financial and logistical support for this seedlings and herbaceous vegetation: and development of oak regeneration. project, particularly Gary McCausland and Support for a theory of resource Pages 273–278 in E. Shoulders (ed.),

254 • December 2007 Ecological Restoration 25:4 Proceedings of the 3rd Biennial Southern and Conservation of the South Puget Sound Tilman, D. 1985. The resource-ratio Silvicultural Research Conference. Prairie Landscape. Seattle, WA: The hypothesis of plant succession. American USDA Forest Service General Technical Nature Conservancy of Washington. Naturalist 125:827–852. Report SO-54. Rey-Benayas, J.M., J. Navarro, T. Espigares, Tveten, R.K. and R.W. Fonda. 1999. Fire Lea, T. 2006. Historical Garry Oak J.M. Nicolau and M.A. Zavala. 2005. effects on prairies and oak woodlands ecosystems of Vancouver Island, British Effects of artificial shading and weed on Fort Lewis, Washington. Northwest Columbia, pre-European contact to the mowing in reforestation of Mediterranean Science 73:145–158. present. Davidsonia 17:34–50. abandoned cropland with contrasting Vance, N., A. Neill and F. Morton. 2006. Lee, R. 1978. Forest Microclimatology. New Quercus species. Forest Ecology and Native grass seeding and forb planting York: Columbia University Press. Management 212:302–314. establishment in a degraded oak savanna MacDougall, A.S. 2002. Invasive perennial Ritter, E., L. Dalsgaaard and K.S. Einhorn. plant community in the Coast Range grasses in Quercus garryana meadows 2005. Light, temperature and soil foothills of western Oregon. Native of southwestern British Columbia: moisture regimes following gap Plants Journal 7:35–46. Prospects for restoration. Pages 159–168 formation in a semi-natural beech- Western Regional Climate Center (WRCC). in R.B. Standiford, D. McCreary and dominated forest in Denmark. Forest 2005. Washington climate summaries. K.L. Purcell (tech. coords.), Proceedings Ecology and Management 206:15–33. www.wrcc.dri.edu/summary/climsmwa. of the 5th symposium on oak woodlands: Schuller, R. 1997. Vegetation response html. Oaks in California’s changing landscape. to fall prescribed burning within USDA Forest Service General Technical Festuca idahoensis-dominated prairie, Report PSW-184. Mima Mounds Natural Area Preserve, MacDougall, A.S., B.R. Beckwith and C.Y. Washington 1985–1992. Pages 207–216 Maslova. 2004. Defining conservation in P. Dunn and K. Ewing (eds.), Ecology Warren D. Devine, Research Forester, strategies with historical perspectives: and Conservation of the South Puget USDA Forest Service, Pacific Northwest A case study from a degraded oak Sound Prairie Landscape. Seattle: The Research Station, Forestry Sciences grassland ecosystem. Conservation Nature Conservancy of Washington. Laboratory, 3625 93rd Ave SW, Olympia, Biology 18:455–465. Singsaas, E.L., D.R. Ort and E.H. Washington 98512, 360/753-7675, MacDougall, A.S. and R. Turkington. 2004. DeLucia. 2000. Diurnal regulation of email [email protected] Relative importance of suppression- photosynthesis in understory saplings. based and tolerance-based competition New Phytologist 145:39–49. Constance A. Harrington, Research in an invaded oak savanna. Journal of Stein, W.I. 1990. Oregon white oak. Pages Forester, USDA Forest Service, Pacific Ecology 92:422–434. 650–660 in R.M. Burns and B.H. Northwest Research Station, Forestry Parker, I.M. 2000. Invasion dynamics Honkala (tech. coords.), Silvics of Sciences Laboratory, 3625 93rd Ave SW, of Cytisus scoparius: A matrix model North America: Hardwoods. USDA Olympia, Washington 98512 approach. Ecological Applications Agricultural Handbook Number 654. 10:726–743. Stokes, B.J. and A. Schilling. 1997. Improved David H. Peter, Ecologist, USDA Forest Parker, I.M., W. Harpole and D. Dionne. harvesting systems for wet sites. Forest Service, Pacific Northwest Research 1997. Plant community diversity and Ecology and Management 90:155–160. invasion of the exotic shrub Cytisus Thysell, D.R. and A.B. Carey. 2001.Quercus Station, Forestry Sciences Laboratory, scoparius: Testing hypotheses of garryana communities in the Puget 3625 93rd Ave SW, Olympia, Washington invasibility and impact. Pages 149–162 Trough, Washington. Northwest Science 98512 in P. Dunn and K. Ewing (eds.), Ecology 75:219–235.

December 2007 Ecological Restoration 25:4 • 255