Experimental evaluation of herbicide use on biodiversity, ecosystem services, and timber production tradeoffs in forest plantations
Stokely, Thomas et al. (2021), Experimental evaluation of herbicide use on biodiversity, ecosystem services, and timber production tradeoffs in forest plantations, Dryad, Dataset, https://doi.org/10.5061/dryad.bzkh1898j
The value of non-commodity ecosystem services provided by forests is widely recognized, but intensive forest management practices are increasing, with uncertain consequences for a multitude of these services. Quantitative relationships among biodiversity conservation, timber production, and other ecosystem services remain poorly understood, especially during the early successional period of intensively managed forestlands.
We manipulated management intensity in regenerating forest plantations to test the prediction that treatments aimed at maximizing timber production decrease biodiversity conservation and non-timber services. We measured species richness of three taxonomic groups and thirteen proxies for provisioning, cultural, and regulating services within stands randomly-assigned to one of three herbicide application intensities or an untreated control.
Herbicides increased allocation of net primary production to crop trees, increasing projected timber volume and revenues at 40- and 60-year harvest ages. Commonly-used herbicide prescriptions reduced culturally-valued plants by 71%, wild-ungulate forage by 41%, avian richness by 20%, and pollinator floral resources by 42%, the latter being associated with 38% fewer pollinator species. However, agriculturally-valued bumblebees, pollination of blueberries, avian-mediated arthropod control, ungulate observations, and regulation services tied to forest productivity appeared unaffected by increasing management intensity and timber production.
Species richness and flora-provided services in young forest plantations exhibited strong tradeoffs with projected timber production, whereas post-treatment vegetation regeneration and site-level variation likely maintained a range of other services. Although vegetation recovery is important for supporting wildlife and some ecosystem services on industrial forestlands, it is unlikely that any single prescription can optimize both timber and non-timber benefits to society across managed forest landscapes. Instead, producing different services in discrete portions of the landscape may be necessary.
Synthesis and applications. We tested the effects of intensive forest management via herbicides on ecosystem services and found that biodiversity responses and services from early-successional vegetation trade off with crop-tree production. A number of services appeared to be compatible with timber production, although no single prescription optimized the full range of services. Stand-level biodiversity conservation and a variety of services may be provided by treatment skips and less-intensive management on productive sites, although it is unlikely that all services can be optimized without landscape-level planning.
Experimental design: The experiment was conducted in the Oregon Coast Range, in stands harvested during 2009/2010. Four stands per block (8 blocks total) were selected based on the criteria of being in proximity to one another (no greater than 5 km) and harvested during the same time period. In 2010, each stand was randomly assigned to one of three herbicide treatment intensities (light, moderate, intensive) or an untreated control. The light treatment received an aerial applications of spring-herbaceous specific (in 2011) and fall woody-broadleaf specific treatments (in 2012). The moderate treatment was treated with aerial applications of a broad-spectrum, site preparation spray (in 2010), aerially-applied spring herbaceous specific spray (in 2011) and spot treatments of coppice-sprouting maple where present (in 2012; 3/8 stands received this treatment). The intensive treatment was sprayed with an aerial site preparation (in 2010), aerial spring herbaceous sprays (2011, 2012, 2013), and fall woody-broadleaf sprays (2012, 2014).
Vegetation (Native plant richness, floral resources, culturally-valued plants, wild ungulate forage): Within each stand, we measured plant species cover from 12, 1 m2 quadrats in a randomly located 15 x 15 m permanent vegetation plot for six years, during the peak in vegetation production (i.e., July-August, 2011-2016). Due to the logistics of a companion wildlife exclusion study, we restricted vegetation measurements to 7 of the 8 blocks. We used information from Von Hagen et al. (1996) to classify plants as having cultural value for food, traditional medicine and as economically valuable non-timber forest products. We also used the USDA Plant Database to classify each species as native or non-native in the state of Oregon. For cultural plants, we summed the cover of all species within each group across the six years, 2011-2016. For native plant species richness, we tallied the number of observed species from 2011-2016 and averaged years for an average-yearly richness estimate. Using information from Cook et al. (2016), Ulappa (2015), the USDA Plant Database and the US Forest Service Fire Effects Information System (FEIS), we assigned plant species to forage and non-forage groups. Further, we used Cook et al. (2016) and Ulappa (2015) to assign plants to selected and accepted forage groups to develop biomass regression equations for each specific forage group. In 2015, we randomly assigned 8 locations throughout each stand to harvest above-ground current annual growth of each species (not including woody biomass) up to 2 m from the ground (i.e., maximum browsing heights) and simultaneously recorded ocular cover estimates using the above methods. We dried plant samples at 55°C and pooled forage group biomass samples for each quadrat and summed cover estimates among species for each group per quadrat, allowing cover values to exceed 100%. We defined forage groups as: selected broadleaf, neutral broadleaf, selected herbaceous (forbs and graminoids), neutral herbaceous, non-forage shrubs, non-forage herbaceous and non-forage conifers. We also directly collected and dried biomass per species from 4, 50x50 cm quadrats, located within excluded and open plots. We used biomass-cover samples from 2015 to develop allometric equations predict forage biomass production for each year from 2011-2016. We fit models for each forage group, including the log-log relationship between biomass and cover estimates, with log-transformed average plant height as a covariate. We used Akaike Information Criterion (AIC) to compare full models with models only including cover as a predictor variable, generally opting for the full model. We then used the equations to predict biomass production at the plot-level from 2011-2016 using the equations below. For all models, we used the Consistent I estimator bias correction factor (Hayes & Shonkwiler 2007), calculated as biomass+exp(Mean Square Error/2). We then summed the biomass of all forage groups across years to obtain an estimate of net forage production. To estimate the frequency of animal-pollinated flowering plants, we recorded the presence or absence of flowers for each species in each quadrat. We then tallied the number of species-quadrat level observations from 2012 (when flower surveys began) until 2016. We used information from Niell and Puettmann (2013) to classify each flowering species as to whether they provide pollen sources for animals.
Blueberry pollination: To test the effects of herbicide application on insect pollination efficiency, we conducted a pollination experiment using potted sentinel Northern Highbush Blueberries (Vaccinium corymbosum, variety 'Duke', Alpha Nurseries, Salem, Oregon, USA), an economically important crop plant in Oregon. Northern Highbush blueberries are self-fertile, but cross-pollination by pollinators increases seed set and berry mass, which is positively correlated with pollinator abundance (Isaacs & Kirk 2010). Pollinators of blueberries in the study region include bumblebees and honeybees, but bumblebees are thought to be more efficient pollinators (Rao, Stephen 2009). Blueberry plants were planted in 1-gallon pots, and we added hydrogel to minimize desiccation. Before bringing the plants to the field, we kept them in a cooling chamber at 10oC for 10 days to synchronize flowering. We brought out a group of 4 blueberry plants that received no prior pollinator visits to each stand in early May and placed them at least 50 m from the stand edge. All plants per block were simultaneously in the field. Plants were protected from ungulate herbivory with a fence and watered every 5-6 days. Before placing the plants on the stands, we removed all open flowers, individually marked 5 branches per plant and counted the number of flowers on those branches. After 17-19 days in the field, we brought the plants back to a mesh-covered exclosure to prevent further pollinator access. We counted all flowers that had opened during the field period (>95% of the total) and removed flowers that had not yet opened. For all marked branches, we determined fruit set and total mass of ripe berries per branch.
Wild ungulate observations: During four seasons (2012-2015), we assessed stand use by wild ungulates (black-tailed deer [Odocoileus hemionus colombianus] and Roosevelt elk [Cervus canadensis roosevelti]) using camera traps (Bushnell Trophy Camera, model 119436). Cameras were placed in the corner of each vegetation plot. We considered the number of photos taken of individuals within 10 m of the camera per day per sampling period (May-October from 2012 to 2015) for each species to be estimates of detection. We included the number of active camera days as an offset in the further analysis of ungulate detections, as the total observation time per stand ranged from between 397 to 736 days, due to occasional camera malfunction. To address detection issues regarding vegetation density, we deployed distance markers and only included photos taken of animals within 10-m of each camera. We then determined that treatments did not differ in detection probability for pictures taken in 1-m bins from 1-10 meters, using Weibull and lognormal distributions.
Pollinator richness: To quantify pollinator richness, we randomly assigned 8 pollinator plots per stand in 7 research blocks. At every pollinator richness plot, an observer determined four 1 x 2 m subplots, situated 25 m away from the pollinator richness plot, one in every cardinal direction (N,E,S,W, n = 28*3*4 = 336 subplots). Every subplot was observed for 10 minutes. All invertebrate pollinators touching reproductive plant organs were caught using netting or an aspirator and dispatched in ethanol for identification in the lab. After the 10-minute period, observers walked slowly back to the plot center, collecting all pollinators within the 1 m wide transect. For each block and round, two stands were surveyed in the morning and 2 stands in the afternoon of the first visit. This order was switched for the second visit. Observations only occurred during favorable weather conditions (<50% cloud cover, no rain, no strong winds). Pollinators were air-dried and pinned for identification to the species level. Samples will be curated at the Oregon State Arthropod Collection (https://osac.oregonstate.edu).
Bumblebee counts (Bombus vosnesenskii): We quantified the abundance of a native key pollinator species, the yellow-faced bumble bee Bombus vosnesenskii, on the experimental stands, as a proxy of regenerating forest stands to serve as source habitats for agriculturally relevant key pollinators. The yellow-faced bumble bee is critical for cranberry, blueberry and clover seed production in the region (Rao and Stephen 2009, 2010; Stephen et al. 2009; Broussard et al. 2011), and this species is widespread and abundant in early-successional forests (Rivers and Betts, in revision). Bumblebees were sampled during three sampling rounds in 2015 at three locations (corresponding to the avian point count stations) per stand. The rounds corresponded to late May to early June, mid to late July and early to mid-August. The locations were placed according to a stratified random design, in order to cover representative portions of the stands while avoiding stand edges and maximizing the distance between locations. At every location, we placed a total of four blue vane traps, at 25m distance away from the location center in each cardinal direction. Two opposing traps had propylene glycol in the bottom of the trap (1inch of fluid), while the two other traps were dry. The cardinal directions were chosen at random. Blue vane traps were collected and emptied after 24 hours. To preserve bumblebees, we washed them of propylene glycol using hot water, soap, and ethanol, and then air dried and pinned each bumblebee for identification. We then tallied the total number of B. vosnesenski collected in each stand. Samples will be curated at the Oregon State Arthropod Collection (https://osac.oregonstate.edu).
Avian richness: We surveyed birds at three randomly placed point count plots during 6 years (2011-2016) using standard avian point count protocols (Ralph 1995). The three bird census points were located with at least 130 m separation between adjacent points, and at least 75 meters from the treated stand edge. Survey points were sampled four times during the breeding season (25 May to 4 July) in each survey year. Survey order and observers were randomized throughout the season to avoid associated biases. Observers recorded every bird seen or heard with an associated first detection distance from the census point, including aerial insectivores but not fly-over individuals. Each point was sampled four times for a ten-minute interval during the breeding season (May 28–July 3rd) and were conducted between sunrise and 10 am. For all analyses, we aggregated data over all point count stations within a stand to obtain one response per harvest unit per visit. Using a 3-m radius subplots, we quantified broadleaf vegetation cover yearly around each point count location, a measure we then integrated into the detection sub-model of the avian statistical models. The spatially and temporally replicated sampling allowed us to use Bayesian N-mixture models that could account for differences in detection probabilities between stands (i.e. multispecies site occupancy and multispecies N-mixture abundance models). In brief, the occupancy process models included terms for block, treatment, year, and the treatment × year interaction. The species-specific detection model included terms for year, percent cover of woody vegetation, and quadratic ordinal date (Table S1). We used the estimated latent occupancy and abundance state variables to compute species richness per stand and associated credibility intervals. Bayesian models were fit using JAGS 46 called from R version 3.5.1 41 using the ‘jags’ function in package R2jags version 0.05-07 47, with three Markov chains of length 100,000 with a burn-in period of 20,000 and 1/80 thinning.
Avian-herbivore control: In the second year of stand regeneration following harvest (2012), we constructed two adjacent 15 m x 15 m exclosures: one fenced and one fenced and netted. Numerous terrestrial taxa prey on arthropods, yet our focus was on birds. Our netted exclosure excluded both birds and bats. However, bats in our study region have been shown to utilize old-growth forest and water as nesting and feeding sites at much higher rates than young-growth forest (up to 10.3 times, Thomas 1988, Clare et al. 2011). The Myotis bats that comprise the majority of the bat species in our study area are also more likely to feed on flies and aquatic insects (Waldien et al. 2000), and not the arthropods for which we found strong exclosure effects. Therefore, birds likely comprise the majority of vertebrate predators in this experiment. Each exclosure consisted of a 2.4 m high fence with a mesh size of 10 cm x 8 cm. The netted exclosure also included a net with mesh size 18 mm x 18 mm that was stretched over the top of the fence and draped over the sides, extended to the ground. Exclosures were randomly placed at least 50 m from stand edges, occurred on a wide range of slopes, covered all aspects, and avoided riparian areas, logging trails, and logging debris piles. Nets were removed in the winter to avoid snow loading and reinstalled in the spring. We sampled arthropods in each exclosure, once in July and once in August in 2012, 2013, 2014, and 2015 (3 to 6 years following harvest) using methods intended to sample a wide range of substrate and arthropod feeding guilds: sweep net, pitfall trap, and a restricted leaf-turning method. For the sweep net, an observer walked three evenly spaced parallel transect lines across the length of the exclosure, sweeping a net at ankle- to waist-height with sufficient force to dislodge arthropods while not damaging the vegetation. Three pitfall traps, using a 9 cm diameter plastic cup with rain cover, were evenly spaced across the middle of the exclosure and remained in place for 24 hours. For the leaf-turning method, one representative specimen of the five most abundant tree and shrub species in the plots, one representative of bracken fern (Pteridium aquilinum), one representative of sword fern (Polystichum munitum), and all Douglas-fir (Pseudotsuga menziesii), were sampled. Each sample consisted of examining a fixed number of leaves per plant for deciduous species and five 30 cm branch segments for Douglas-fir. Sampled deciduous tree and shrub species included California hazelnut (Corylus cornuta), oceanspray (Holodiscus discolor), vine maple (Acer circinatum), big-leaf maple (Acer macrophyllum), cas¬cara (Rhamnus purshiana), red alder (Alnus rubra), Rubus spp., Vaccinium spp., salal (Gaultheria shallon), and Oregon grape (Mahonia nervosa), depending on site. We detected no difference in the number of Douglas-fir trees in the fenced (Mean = 19.5, SE = 4.6) and netted (Mean = 20.7, SE = 5.1) exclosures (paired t-test, t27 = -1.48, P = 0.15). Arthropods were identified to the family level in most cases and assigned to a feeding guild (predator, herbivore, detritivore, mixed). For arthropods that were difficult to identify, we assigned a higher taxonomic rank: Chilopoda (class), Acari and Collembola (sub-class), Isopoda, Microcorphyia, and Opiliones (orders); microwasp and midges (sub-orders). The length of each arthropod was measured prior to live release in the center of each plot. No vegetation was removed to sample arthropods. The composite count from the sweep net, pitfall trap, and leaf-turning methods from the two surveys (July and August) represents the relative arthropod abundance for that year. For the analyses, we included only herbivorous arthropods to quantify the potential effect birds have in controlling herbivorous arthropod abundance. To estimate an effect size for trophic strength, we calculated the log-response ratio (LRR) with pairwise comparisons of the treatment factors using the emmeans package in R (Russell 2018). Because our response variable was log-transformed, the pairwise comparison between the responses in the presence versus absence of birds results in LRR: ln[NB+] – ln[NB-] = ln[NB+ / NB-] = LRR where NB+ and NB- are the mean responses in the presence (fenced exclosure) and absence (netted exclosure) of birds, respectively. The log-response ratio is frequently used in trophic studies to estimate cascade strength (Hedges et al. 1999, Mooney et al. 2010).
Timber projection: We conducted tree growth projections from a list of planted crop trees (Pseudotsuga menziesii) and naturally regenerating conifer and hardwood species, measured from 18-20 systematically located plots within each stand. All plots were measured at the end of the growing season in 2015 (i.e., October-December) and consisted of a 5-m radius plot, in which planted-crop trees were tagged and measured. We also established a nested 3-m radius plot in which naturally regenerating conifer and hardwood species were tagged and measured. Measurements included bole diameter at 15 cm from root collars, diameter at breast height (dbh, i.e., 1.37 m) and height to apical leader. To avoid overestimates of hardwood regeneration, stems from coppice-sprouting hardwood clumps were only included if the diameter for each stem was at least 80% the size of the largest stem within each sprouting clump. From the 3 m radius plots, we tallied the number of naturally regenerating conifer seedlings less than 20 cm in height and assigned them all a height of 15 cm. During the growing season of 2015 (i.e., July-September), we ocularly estimated the percent cover of competing vegetation, grouped by forb, graminoid, fern and broadleaf shrub and non-commercial broadleaf woody (i.e., excluding Alnus rubra, Acer macrophyllum and Prunus emarginata) life-form functional groups. Life-form cover estimates were fit into projection models as the average sum of competing life-form percentages per plot within each stand. We combined two growth models to project development of the 5-year-old tree lists for 100 years, thereby using the most appropriate equations to account for both the competing vegetation in the young plantations and hardwood growth in older stands. We projected tree lists from age 5 to 20 with Center for Intensive Planted-forest Silviculture (CIPS) annualized growth equations. These are a set of proprietary equations for diameter and height growth, and mortality of Douglas-fir and western hemlock plantations, and account for the negative effect of competing vegetation cover. Although these equations can be used to estimate growth of trees ranging from newly planted seedlings to mature stands, additional equations needed to be added to grow other species. Trees of the twenty-year-old projected tree lists were then further grown with SMC-ORGANON, a regional growth model constructed from plantation data in the Pacific Northwest (26). Because the CIPS equations cannot estimate growth of species other than Douglas-fir and western hemlock, other means were used to model growth for other tree species present on the plots. Although there were few other conifers sampled, grand fir (Abies grandis) and noble fir (Abies procera) were modeled as western hemlock (Tsuga heterophylla). No regional equations were available for bitter cherry (Prunus emarginata), so they were grown as bigleaf maple (Acer macrophyllum). Red alder (Alnus rubra) diameter growth was estimated with annualized ORGANON equations (26), with the red alder site index assumed to be 50% of the Douglas-fir site index. Bigleaf maple diameter growth was estimated with a linear interpolation of the ORGANON 5-year diameter growth equation (27) for maple. Estimated heights of both species were based on ORGANON height:diameter equations for alder (26) and maple (28), and ORGANON equations were also used to estimate crown base (alder: (26); maple: (29)) and mortality (alder: (26); maple: (27)). Assignment of diameter to hardwood stems surpassing 1.37 m was based on an equation constructed from data collected from this study. Douglas-fir 50-year site index (30) used within both models was based on landowner estimates. Western hemlock site index was based on a published conversion equation from the Douglas-fir site indices (31). We then calculated expected yield (board feet * ha-1) and expected yield gains for each herbicide treatment at three different rotation ages typical for the region (40 and 60 years). For this, we fitted a linear mixed effects models in R for each rotation age (9), with the expected yield per stand as the response, herbicide treatment as explanatory factor, and block random intercept. We chose the best fitting variance structure per model (Gaussian, VarExp, VarPower, varIdent(form=~1|Treatment)) using corrected Akaike Information Criteria (AICc) and residual plots, to account for residual heteroscedasticity. We fit final models using restricted maximum likelihood.
Revenue projection: To compare the economic performance of the different herbicide treatments under a broad spectrum of realistic investment decisions (discount rates ranging from 4 – 10%), we performed a series of cash-flow analyses (plantations costs, herbicide costs, harvest costs, harvest income) and transformed those into land expectation values (LEV, also called soil expectation value). LEV is a standard discounted cash-flow technique used to compare timber management strategies for plantations with periodic harvest. LEV calculates the value of a forest plantation under perpetual production (a proxy for the willingness to pay for forest land). Unlike net present value, LEV values are not associated only with current timber rotations and can be compared between forest lands with different rotation ages. Importantly, future cashflows are discounted at a rate (hereafter “discount rate”) which can be interpreted as a measure of opportunity cost of capital (i.e., the rate of return from the best alternative available investment project). We calculated land expectation values using estimated costs of planting ($650 ha-1), logging ($32/m3), and hauling ($14/m3) that were based on communications with local professional foresters. Herbicide costs were based on the current costs of applying the specific products used (site prep ($200 ha-1); release 1 ($175 ha-1); release 2 ($150 ha-1)). Economic tabulations also included an annual 2% inflation rate and an annual 0.5% timber price increase. All calculations were performed for nominal discount rates of 6% representing the spectrum of applied interest rates in the region based on communications with local timber companies. We then assessed whether herbicides caused differences in expected LEV. Using the maximum LEV across rotation ages for each stand and for each discount rate (e.g. using the rotation age that maximized the LEV for the discount rate under consideration), we compared the expected LEV for each herbicide treatment under 4 and 6% nominal discount rate, rates typical of state and private-industrial private lands, respectively. We used the same statistical procedure as for the yield analysis.
Standing carbon projections: Tree-level aboveground biomass was predicted for Douglas-fir using six-component biomass equations constructed from data collected in intensively managed stands (Mainwaring et al. in prep). Tree-level aboveground biomass of other species was estimated using previously published equations (Ung et al. 2008). Carbon content of aboveground tree tissues was calculated assuming a carbon content of 50% (Matthews 1993).
Soils (nitrogen concentration, soil carbon stocks, litter decomposition): We collected organic and mineral soils from four subplots within all 225 m2 plots between June and November, 2015. Subplots were randomly located 2.35 m right or left of the center point of each side of the plots and 1.5 m from plot edges. These locations were selected to avoid interference with permanent vegetation plots. In each subplot, we clipped vegetation using a 50 cm x 50 cm quadrat as a guide, and then removed the organic horizon (O-horizon) directly beneath by cutting around the perimeter of a 25 cm x 25 cm frame with a knife. O-horizons were then carefully separated from mineral soils. We included harvest residuals in our samples, but we only collected branches that were 1 cm in diameter. We collected mineral soils to a depth of 30 cm in all four subplots after organic layers were removed using a bulk density corer containing plastic sleeves with internal diameters of 4.8 cm. After removal, we split mineral soil samples into increments of 0-15 cm and 15-30 cm. For both O-horizons and mineral soils, we composited the four samples collected from each subplot. O-horizon and mineral soil samples were dried at 50ºC for 72 hours or until a constant mass was achieved. For mineral soils, we also dried subsamples at 105ºC for 48 hours to determine moisture content. After samples were dried, we noted that there was wide variability in the quantity of decay class IV and V buried wood among O-horizon samples. We elected to remove buried wood from the samples to reduce plot-level variation. We separated O-horizon samples into three components: 1) Oi material that was too large to pass through a 2 mm sieve; 2) a combination of Oe and Oa material, that did pass through a 2 mm sieve; and 3) buried wood. Buried wood was removed from both Oi and Oe/Oa fractions, although some fine buried wood could not be removed from the Oe/Oa fraction. Rocks were removed and discarded during the sieving process. Once samples were separated into three fractions, we ground Oi material using an electric wood chipper to homogenize woody and herbaceous components. Oi subsamples were then ground to a fine powder using an IKA grinder. Subsamples of both Oe/Oa material and buried wood were pulverized using a Kleco ball mill. We analyzed ground samples of all three fractions for C and N concentrations using a ThermoFlash 1112 Elemental Analyzer. We sieved dried mineral soils into coarse (>2 mm) and fine (<2 mm) fractions, after which fine fractions were ground using a roller grinder and analyzed for C and N concentrations using a ThermoFlash 1112 Elemental Analyzer. We calculated C stocks of all five soil components (Oi, Oe/Oa, buried wood, 0-15 cm mineral soils, and 15-30 cm mineral soils) separately. For O-horizon components, we used the C concentrations from elemental analysis, dry masses, and known areas of our frames to calculate stocks on a Mg/ha basis. For mineral soils, we corrected fine fraction masses for moisture content to calculate bulk density and then used C concentrations and the moisture corrected bulk density values to determine C stocks for 0-15 cm and 15-30 cm layers. Total C stocks were calculated by summing organic and mineral stocks. We excluded buried wood C stocks from the total C stocks reported in the analysis. In spring of 2016, we collected needles from current year’s growth on three-year-old Douglas-fir branches from three locations in the Oregon Coast Range, all separate from our study sites. After collection, we dried needles at 50ºC for 72 hours. Ten grams of needles were then placed in litterbags, which we constructed using 40 um polypropylene mesh bottoms, and 3.2 mm polypropylene mesh covers. To enable access by macroarthropods, we cut two 7 mm diameter holes in the top of each bag. We then attached five bags together using nylon contractor’s string. Needles from the three collection sites were not homogenized prior to filling litterbags, but with the exception of one plot, needles from a given collection site were consistently used in all plots in a block. Between late March and early April 2016, we placed one set of bags in randomly selected corners of all plots. We then collected bags after approximately 3, 6, 12, 18, and 26 months. Here, we only present data from the first year, as contamination from litterfall from surrounding vegetation likely influenced our measurements of mass remaining after the one-year mark. Upon collection, we dried litterbags at 50ºC for 72 hours, removed any contaminants that had fallen into the bags (rocks and detritus that was visually distinct from Douglas-fir needles), and then weighed the needles. We calculated daily decomposition constants over the course of one year by regressing the natural log of the percent mass remaining against time and fixing the y-intercept to ln(100) at time 0, using the following equation: ln(x_t/x_0 *100)=ln(100)-k(t), where xt is the mass (g) of needles at time t (days), x0 is the original needle mass, and k is the decomposition constant (Matkins, 2009; Olsen, 1963). We calculated and present daily k constants rather than monthly or annual constants, as not all litterbags were in the field for the same number of days. However, all litterbags within individual blocks were placed and collected within one day of each other. We set intercepts to ln(100) given that we knew that 100% of the initial mass was remaining when bags were placed in the field.
Broussard, M., S. Rao, W. P. Stephen, and L. White. 2011. Native bees, honeybees, and pollination in Oregon cranberries. HortScience 46:885-888.
Clare, E. L., B. R. Barber, B. W. Sweeney, P. D. N. Hebert, and M. B. Fenton. 2011. Eating local: influences of habitat on the diet of little brown bats (Myotis lucifugus). Molecular Ecology 20:1772–1780.
Cook, J.G., Cook, R.C., Davis, R.W. & Irwin, L.L. (2016) Nutritional Ecology of Elk During Summer and Autumn in the Pacific Northwest. Wildlife Monographs, 195, 1-81.
Hedges, L.V., J. Gurevitch, P.S. Curtis. 1999. The meta-analysis of response rations in experimental ecology. Ecology 80, 1150–1156.
Isaacs, R. & Kirk, A.K. (2010) Pollination services provided to small and large highbush blueberry fields by wild and managed bees. Journal of Applied Ecology, 47, 841-849.
Matkins, J. 2009. Decomposition and nitrogen dynamics of red alder and Douglas-fir leaf litter in Oregon Coast Range riparian forests, Thesis, Oregon State University, Corvallis, Oregon, USA.
Matthews G. The carbon content of trees. Forestry Commission Technical Paper 4. Forestry commission, Edinburgh, Scotland, 1993.
Mooney, K. A., D. S. Gruner, N. A. Barber, S. A. Van Bael, S. M. Philpott, and R. Greenberg. 2010. Interactions among predators and the cascading effects of vertebrate insectivores on arthropod communities and plants. Proceedings of the National Academy of Sciences 107:7335–7340.
Neill, A.R. & Puettmann, K.J. (2013) Managing for adaptive capacity: thinning improves food availability for wildlife and insect pollinators under climate change conditions. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere, 43, 428-440.
Ralph, C. J., S. Droege, and J. S. Sauer. 1995. Monitoring Bird Populations by Point Counts. Page 191. General Technical Report, U.S. Department of Agriculture, Forest Service, Albany, CA, USA.
Rao, S., and W. P. Stephen. 2010. Abundance and diversity of native bumble bees associated with agricultural crops: The Willamette Valley Experience. Psyche: A Journal of Entomology 2010:354072.
Rivers, J.W., Galbraith, S.M., Cane, J.H., Schultz, C.B., Ulyshen, M.D. & Kormann, U.G. (2018) A Review of Research Needs for Pollinators in Managed Conifer Forests. Journal of Forestry, 116, 563-572.
Russell, L. 2018. emmeans: estimated marginal means, aka least-squares means. R package version 1.2.3.
Stephen, W. P., S. Rao, and L. White. 2009. Abundance, diversity and foraging contribution of bumble bees to blueberry production in Western Oregon. Acta Horticulturae 810:557–562.
Thomas, D. W. 1988. The distribution of bats in different ages of Douglas-fir forests. The Journal of Wildlife Management 52:619.
Ulappa, A. C. (2015). Doctoral Dissertation: Using foraging dynamics to answer landscape management questions: The nutritional ecology of blacktailed deer. Pullman, WA: Washington State University, Research Exchange.
Ung, C.H., Bernier, P. and Guo, X.J. 2008. Canadian national biomass equations: new parameter estimates that include British Columbia data. Can. J. For. Res. 38: 1123-1132.
Von Hagen, B., Weigand, J.F., McLain, R., Fight, R., Christensen, H.H., 1996. Conservation and development of nontimber forest products in the Pacific Northwest; an annotated bibliography. General Technical Report. Portland, OR. PNW-GTR- 375:1-246. https://doi.org/10.2737/PNW-GTR-375.
Waldien, D. L., J. P. Hayes, and E. B. Arnett. 2000. Day-roosts of female long-eared Myotis in Western Oregon. The Journal of Wildlife Management 64:785–796.
See the readme file for metadata descriptions.
U.S. Department of Agriculture, Award: AFRI-2009-04457
Agricultural Research Service, Award: AFRI-2015-67019-23178
Oregon Forest Industries Council
Oregon State University
National Council for Air and Stream Improvement
Oregon Forest Industries Council