Net Primary Production and Biomass Distribution in the Blue Oak Savanna 1 John J. Battles, 2 Randall D. Jackson, 3 Ayn Shlisky, 4 Barbara Allen-Diaz,2 and James W. Bartolome2 Abstract The complexity of savanna ecosystems with the shared dominance between woody and herbaceous primary producers poses challenges to measuring such fundamental ecosystem characteristics as net primary productivity (NPP). We address these challenges in the blue oak savanna in California by constructing comprehensive estimates of NPP for three adjacent watersheds. We directly assessed annual biomass increment of all plant components in 12 randomly stratified plots (380 m2) per watershed. Annual estimates of mean NPP for the three experimental watersheds during 2001 to 2002 ranged from 4.35 to 5.69 Mg ha-1 yr -1 of dry biomass. On average, belowground NPP accounted for 22 percent of total NPP. In general, there was much greater uncertainty in the belowground estimates. Trees accounted for approximately 50 percent of aboveground NPP in all three watersheds. Across these watersheds, tree productivity increased in a linear fashion with canopy closure. In contrast, herb productivity was nearly constant for relatively open sites (canopy closure < 40 percent) and then monotonically declined as canopy closure increased. The result is that total NPP increased gradually from the most open sites in the watersheds to a maximum around 55 percent canopy closure. Keywords: Biomass distribution, carbon budget, Mediterranean ecosystem, net primary production, temperate savanna. Introduction A fundamental characteristic of savanna ecosystems is the co-dominance of tree and grass life forms. Savannas and synonymous designations (e.g., woodland, rangeland, shrubland) constitute a gradient of ecosystems that fall between grasslands and forests. Together they account for more than an eighth of the terrestrial biosphere (Scholes and Archer 1997). They are economically and ecologically important in tropical and temperate regions throughout the world. Despite their extent and value, basic ecosystem functions of savannas are poorly understood relative to grasslands and forests. In a recent review, House and others (2003) noted that challenges to developing a robust understanding of savanna ecosystems included a preponderance of studies that focused on either the grass or tree component in isolation and a lack of studies that addressed belowground productivity and biomass. Yet it is this sort of integrated ecosystem-level information that is crucial to understanding savanna dynamics and to 1 An abbreviated version of the paper was presented at the Sixth California Oak Symposium: Today’s Challenges, Tomorrow’s Opportunities, October 9-12, 2006, Rohnert Park, California. 2 University of California, Berkeley, Department of Environmental Science Policy and Management, 137 Mulford Hall, Berkeley, CA 94720-3114. 3 University of Wisconsin-Madison, Department of Agronomy, 1575 Linden Drive, Madison, WI 53706. 4 The Nature Conservancy, 2424 Spruce St., Boulder, CO 80302. 511 GENERAL TECHNICAL REPORT PSW-GTR-217 managing them appropriately. In this paper, we summarize our efforts to confront these challenges for the blue oak savanna in California. A critical management objective in California oak savanna is to sustain livestock productivity while maintaining long-term ecosystem health. Despite a wealth of information regarding the productivity of certain components of this ecosystem including forage yield (Holland 1980, Kay 1987, Connor and Willoughby 1997), leaf and acorn production (Dahlgren and Singer 1994, Knops and others 1996, Koenig and others 1996, Koenig and others 1999), fine root growth (Jackson and others 1990, Millikin and Bledsoe 1999), and tree growth (Kertis and others 1993), there exists no estimate of total net primary productivity (NPP). Without NPP estimates, we cannot address basic ecological questions about NPP allocation under various biotic and abiotic scenarios–information that is useful for predicting the effects of land-use modifications or climate changes on C cycling and plant community distribution. Moreover, in savannas where NPP has been estimated, there exists what House and others (2003) describe as the “NPP conundrum”–competing hypotheses about the nature of NPP allocation between trees and herbs as tree dominance varies. The general relationship noted by House and others (2003) is for declining grass NPP with increasing tree dominance where dominance is measured as a function of tree density, basal area, or canopy cover. Robust tests of this relationship remain rare, though Reich and others (2001) found that aboveground NPP decreased with increasing tree dominance in oak savanna of the upper Midwest supporting this relationship. Alternatively, Mitchell and others (1999) reported that in the pinewiregrass savanna of the southeastern United States, the highest grass productivity occurred at sites with the greatest tree density. In the blue oak savanna, several studies have concluded that the presence of trees enhances understory productivity (Callaway and others 1991, Dahlgren and others 1997) primarily due to processes associated with nutrient cycling. However, results from tree removal studies conflict (Connor and Willoughby 1997). For example, experiments conducted at the same research station in the blue oak savanna have drawn opposing conclusions (Holland 1980, Kay 1987). In 2001, we initiated a study to explore the patterns in productivity and the distribution of biomass in the blue oak savanna. We will use this information to develop a measure of ecosystem health, based on a robust and integrated estimate of NPP for an entire management unit (i.e., the watershed). We measured NPP across the grass-tree mosaic using techniques from landscape and community ecology to efficiently allocate our sampling effort and to properly extrapolate our plot-level results. Specifically, our objectives were to produce watershed-level productivity and biomass budgets for blue oak savannas and to examine the internal relationships between productivity and tree abundance in the blue oak savanna. Methods Study Site In 2001, we identified three adjacent watersheds and instrumented them to measure NPP. The experimental watersheds are located in the foothills of the northern Sierra Nevada at the University of California Sierra Foothill Research and Extension Center (SFREC) near Marysville, CA. (39º 15' N, 121º 17' W). These three adjacent watersheds, collectively known as the Lewis watersheds, contain hilly, rolling terrain. 512 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles They range in size from 35 ha (WS1) to 116 ha (WS3) and encompass an elevation gradient from 189 to 584 m. The climate is Mediterranean with cool, wet winters and hot, dry summers. Based on the onsite weather records for the past 15 years (California Irrigation Management Information System, Browns Valley Station #84, http://wwwcimis.water.ca.gov/cimis), total annual precipitation ranges from 49 to 133 cm with a mean of 77.5 cm. Most of the rain (98 percent) falls between October and May. There is a prolonged summer drought (June-September) where relative humidity averages less than 45 percent and mean daily air temperature is 24°C. Annual estimates of productivity coincide with the water year, which starts on October 1, the typical beginning of winter rains, and ends the following September 30. Thus, the estimates reported here for 2002 include all the plant production that occurred during the 2002 growing season, even though the year began on October 1, 2001. Soils within the watersheds formed in basic metavolcanic (greenstone) bedrock. They are classified as fine, mixed, thermic Typic Haploxeralfs (Dahlgren and others 1997). These soils can extend to a depth of 100 to 150 cm and overlie relatively massive bedrock. The vegetation of the foothills consists of an overstory dominated by the winterdeciduous blue oak (Quercus douglasii). Interior live oak (Quercus wislizenii) and foothill pine (Pinus sabiniana) are present at lower densities (Shlisky 2001). As is typical for savanna ecosystems, the trees are patchily distributed across the landscape. Canopy cover varies from less than 4 percent to more than 80 percent with a watershed-level mean of 56 percent. Based on the 2002 inventory, the three most common plants in the grass-dominated understory (42 percent cover) are the introduced annual grasses: Bromus hordeaceus, Bromus madritensis and Cynosurus echinatus. The next most common species (8 percent cover each) are the introduced annual forbs, Trifolium hirdum and Torilis nodosa. Common shrubs include poison oak (Toxicodendron diversiloba) and coffeeberry (Rhamnus californica). The prevailing management regime in these watersheds includes fire exclusion and grazing at moderate intensity by a cow-calf herd in the green season (December through May). In 2002, herbivore consumption (domestic and native) on the grass layer was approximately 0.37 Mg yr -1 (22 percent of grass productivity). Data Collection and Analysis Productivity sampling was based on the stratified random design described in Shlisky (2001). Using color aerial photographs, watersheds were divided into four tree-cover classes (<15 percent, 15 to 30 percent, 30 to 60 percent, and >60 percent). The minimum mapping unit was 2 ha. Sampling intensity in each watershed was allocated to each cover class proportional to its abundance. Plot locations were randomly assigned in each cover class. From this set of 64 plots, we randomly choose a subset of 36 locations to establish NPP plots (3 per strata, 12 per watershed). Note that all watershed-level estimates (means and variances) were weighted by the proportional abundance of the cover class in the watershed (Cochran 1977). NPP plots consist of an 11-m radius circular area (380 m2). Shlisky (2001) demonstrated that plant composition and canopy structure is relatively homogeneous at this scale. In 2001, all live trees ≥ 5 cm diameter at breast height (1.37 m, dbh) 513 GENERAL TECHNICAL REPORT PSW-GTR-217 within the plot were identified, measured, and tagged. Dendrometer bands were installed on a size-stratified subset (15 percent of total, 94 trees) of these trees in order to measure annual stem increment. Within each plot, three “cattle-resistant” litterfall traps were systematically placed at 3, 5, and 7 m from the center at 120° angle from each other. These traps were approximately 1.5 tall to avoid disturbance from cows. The tops were outfitted with burlap sacks. The maximum opening was 0.23 m2; however, the actual opening was often smaller because the cages were often deformed by cattle impact. Thus the opening size was measured at each collection. Litter is defined as leaves, seeds, and twigs ≤1 cm in diameter. The cages were maintained year-round and litter was collected three times per year with the last collection timed soon after maximum leaf fall. Shrub biomass estimates for the interior 5-m radius of each NPP plot were based on light interception methods (sensu Reich and others 2001). A calibration curve relating shrub biomass to shrub leaf area index (LAI) was developed by harvesting all the biomass in the shrub layer (0.5 m to 2.5 m in height) in nine plots adjacent to the experimental watersheds. Before harvest, we measured shrub LAI using the techniques described above. We found a strong correlation between the two parameters (shrub mass in Mg ha-1 = 17.53 × shrub LAI, r2 = 0.95, p <0.0001). Herbaceous layer aboveground net primary productivity (ANPP) was estimated by harvesting and weighing all herbaceous material from 0.0625 m2 quadrats from within three randomly located, 1 m2 livestock exclusion cages at peak standing crop (May 15 to June 15). Exclusion cages were randomly relocated each year postharvest to avoid resampling or potential cumulative cage effects. Fine root NPP (roots ≤ 2 mm diameter) was estimated using sequential coring (Vogt and others 1998, Fahey and others 1999). This method consists of estimating fine root production as the difference between live root biomass at minimum and maximum periods of plant growth. The method is most useful in systems that undergo distinct growth pulses such as annual grasslands (Fahey and others 1999). Hence, we collected three randomly located cores (15 cm depth, 5 cm inside diameter, AMS Core Sampler, American Falls, ID, USA) during the winter slow growth phase (December) for minimum and at peak standing crop (May/June) for maximum fine root biomass estimation. Soil cores were stored at 5ºC until being washed over a 1-mm mesh screen, which allowed us to collect all root and organic matter fragments trapped by the screen. We used a variant of the stand increment approach (sensu Clark and others 2001) to estimate wood biomass and production. First, we developed allometric equations to predict aboveground woody biomass as a function of dbh. We combined the results from two separate studies of blue oak trees at SFREC (Millikin and others 1997, Dahlgren and Singer 1994) to obtain dbh and biomass estimates on eight trees (dbh range: 7.6 – 48.5 cm). For coarse root biomass (roots > 2 mm in diameter), we used the data from six trees in Millikin and others (1997). We applied the same allometric equations to calculate interior live oak biomass. There are no specific allometric equations for foothill pine, so we used the general equations that predict the aboveground woody biomass of pines from dbh (Jenkins and others 2003, table 3, species id = 100) and the equation from Omdal and others (2001) for Pinus ponderosa to estimate large-root biomass from dbh. 514 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles As noted above, we inventoried all the live trees (dbh ≥ 5 cm) in our plots in 2001. We censused these plots again in late September 2002 to document tree mortality and ingrowth. We also recorded the annual (October 2001 to September 2002) diameter growth from the subset of trees with dendrometer bands. Statistical Framework We evaluated the importance of differences in productivity among ecosystem components and watersheds by comparing means and their respective 95 percent confidence intervals. We explored the correlation of plot-level NPP with a suite of potential indicators, including tree basal area, canopy closure, herbaceous productivity, height of the herbaceous canopy, species richness, elevation, slope, and aspect. To visualize changes in tree, herb, and total NPP across the vegetation mosaic in the blue oak savanna, we smoothed the plot-level responses using locally weighted regressions (Cleveland and Devlin 1988). We used percent tree canopy closure (determined by light interception techniques; sensu Reich and others 2001) as the independent variable measuring proportional tree dominance. Results Productivity and Biomass Annual estimates of mean NPP for the three experimental watersheds during 2001 to 2002 ranged from 4.35 to 5.69 Mg ha-1 yr-1 of dry biomass (table 1). Confidence intervals (95 percent) around these means averaged ± 33 percent of the mean. Given this variation, there were no significant differences in NPP between watersheds. Belowground production accounted for a low 18 percent of the NPP in WS1 and a high of 27 percent in WS2 (table 1). In general, there was much greater uncertainty in the belowground estimates. Trees accounted for approximately 50 percent of ANPP in all three watersheds. In contrast, herb contributions were more varied. They produced 47 percent of ANPP in WS1, 39 percent in WS2, and only 35 percent in WS3 (table 1). Live tree biomass was greatest in WS2 at 129 Mg ha-1 (table 2) with smaller pools in both WS1 (113 Mg ha-1) and WS2 (111 Mg ha-1). However, like NPP, these differences were not significant. Shoot-to-root ratios in live tree mass ranged from a low of 2.5 in WS1 to 3.3 in both WS2 and WS3 (table 2). Shrub mass was a consistently small component (1.4 -2.0 Mg ha-1) of the live biomass pool in these watersheds, but they fixed proportionally more biomass (9 to 13 percent of ANPP, table 1). 515 GENERAL TECHNICAL REPORT PSW-GTR-217 Table 1—Results from the Lewis Watersheds at Sierra Foothill Research and Extension Center, Browns Valley, CA. Fluxes reported in Mg ha-1 yr-1 of oven-dried biomass. xr refers to the mean of 1000 randomizations of the estimated watershed value using stratified random sampling. Component (Mg ha-1 yr-1) xr WS1 95%CI xr WS2 95%CI xr WS3 95%CI Herbs 1.91 1.59 – 2.26 1.61 1.23 – 2.07 1.20 0.80 – 1.72 Shrubs 0.39 0.24 – 0.52 0.43 0.25 – 0.59 0.45 33 – 0.57 Bole wood 0.88 0.71 – 1.04 1.15 0.81 – 1.54 0.92 76 – 1.10 Tree litter 1.41 1.03 – 1.81 0.97 0.71 – 1.25 0.89 0.66 – 1.12 3.97 – 5.28 4.13 3.51 – 4.80 3.47 2.87 – 4.06 ANPP 4.59 Fine roots 0.80 0.20 – 1.36 1.27 0.71 – 1.81 0.65 -0.01 – 1.32 Coarse roots 0.25 0.19 – 0.30 0.30 0.19 – 0.41 0.20 0.16 – 0.23 BNPP 1.05 0.37 – 1.63 1.55 0.96 – 2.10 0.88 0.25 – 1.56 NPP 5.63 4.73 – 6.52 5.69 4.69 – 6.67 4.35 3.74 – 5.09 Table 2—Results from the Lewis Watersheds at Sierra Foothill Research and Extension Center, Browns Valley, CA. Pools reported in Mg ha-1 of oven-dried biomass. xr refers to the mean of 1,000 randomizations of the estimated watershed value using stratified random sampling. CWD = coarse woody debris. Component WS1 WS2 WS3 xr 95%CI xr 95%CI xr 95%CI Bole mass 82 70 – 96 99 76 – 122 85 69 – 102 Root mass 31 25 – 37 30 24 - 36 26 21 – 31 Shrub mass 1.4 0.72 – 2.0 1.7 1.0 – 2.4 2.0 1.4 – 2.6 CWD 3.5 0.78 – 6.3 2.8 0.99 – 4.4 1.9 0.66 – 3.2 Dead trees 14 -12 – 38 0.60 -19 – 20 9.2 -14 – 35 -1 (Mg ha ) Live mass Dead mass 516 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles There were important differences in the tree composition among the watersheds (table 3). The mean tree basal area was 14.7 Mg ha-1 (95 percent CI: 12.8 to 16.7), but tree basal area was notably but not significantly lower on WS3 (11.7 m2 ha-1) compared to the other two watersheds (table 3). The pine fraction was much higher in WS2 (20 percent of basal area), whereas interior live oak was more abundant in WS3 (28 percent of basal area). Paralleling the differences in the tree stratum, understory plant composition varied among the watersheds (Multiple Response Permutation Procedure, p < 0.001, McCune and Grace 2002). For example, Rhamnus californica (coffeeberry) was more abundant and frequent in WS3, while Bromus madritensis (red brome), an introduced non-native annual grass, was a major component of the understory flora in WS1. Table 3—Results based on 2001 inventory from the Lewis Watersheds at Sierra Foothill Research and Extension Center, Browns Valley, CA. Watershed-level means (standard error of the mean) of tree basal area and density calculated from a stratified random sample of 12 plots per watershed. Species WS1 WS2 WS3 density (%) basal area (%) density (%) basal area (%) density (%) basal area (%) Blue oak 95.2 89.9 56.7 52.4 60.1 58.6 Foothill pine 2.8 8.1 20.5 20.2 6.0 7.8 Interior live oak 1.9 1.9 19.2 19.2 33.0 28.1 Black oak1 0.2 0.1 3.6 8.1 -- -- Ponderosa pine1 -- -- -- -- 1.0 5.5 stems ha-1 m2 ha-1 550 (241) 16.6 (2.3) Totals 1 stems ha1 m2 ha-1 438 (75) 15.7 (1.9) stems ha1 m2 ha-1 317 (36) 11.7 (0.6) Black oak (Quercus kelloggii) and ponderosa pine (Pinus ponderosa). Patterns in Productivity From the suite of potential indicators, tree basal area was the most strongly correlated with plot-level NPP (r = 0.63, p < 0.001. fig.1A). The next best was a positive relationship with herb productivity (r = 0.49, p = 0.003, fig. 1B). No other correlations with NPP had a coefficient better than ± 0.30. 517 GENERAL TECHNICAL REPORT PSW-GTR-217 Figure 1—Correlation between plot-level estimates of total NPP and tree basal area (A) and herb productivity (B) in a Californian blue oak savanna. Tree productivity increased in a linear fashion with canopy closure (fig. 2). In contrast, herb productivity was nearly constant for relatively open sites (canopy closure < 40 percent) and then monotonically declined as canopy closure increased. The result is that total NPP increased gradually from the most open sites in the watersheds to a maximum near 55 percent canopy closure. At higher canopy cover there was little change in total NPP (fig. 2) though the ratio of tree biomass to herb biomass continued to increase. 518 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles 7.25 Total 6.25 NPP (Mg ha-1 yr-1) 5.25 4.25 Tree 3.25 Herb 2.25 1.25 0.25 0 20 40 60 80 Proportional Tree Dominance (% canopy closure) Figure 2—Regression models of productivity versus tree dominance in a Californian blue oak savanna. The final regressions were based on local linear models with the smoothing parameter set to 0.75 (Cleveland and Devlin 1988). Error bars for total NPP are the standard errors of the fit. The coefficient of variation (CV = standard error of the fit over the fit) for NPP = 10 percent; for tree productivity CV = 16 percent; for herb productivity, CV = 15 percent. Discussion Our watershed-level estimates of biomass and NPP for the blue oak savanna (table 1, table 2) fell within the wide range of values reported for this diverse biome. While the live biomass pool was near the global mean for tropical savannas, productivity was closer to the low end of the range (Chen and others 2003, House and Hall 2001). As a consequence, the biomass-to-NPP ratio for the blue oak savanna (23 years) was much higher than the global mean for savannas (3.4 years, Chen and others 2003). Clearly carbon in the blue oak savanna is cycled much more slowly than expected for a savanna ecosystem. In general, the low productivity of the blue oak savanna is typical of more arid tropical sites that support tree-grass ecosystems. For temperate savannas, our ANPP results (table 1, watershed mean = 4.06 Mg ha-1 yr -1) closely matched two recent findings. Reich and others (2001) measured biomass and productivity in 20 oak savanna stands in eastern Minnesota. Extrapolating from the reported regression relationship, ANPP in comparable stands in Minnesota (i.e., stands with a canopy openness around 44 percent) was around 3.75 Mg ha-1 yr-1. Mitchell and others (1999) documented an ANPP gradient from a low of 3.91 Mg ha-1 yr-1 in xeric sites to a high of 7.43 Mg ha-1 yr-1 in mesic sites for pine-wiregrass ecosystems in Georgia. The productivity of our research watersheds is broadly representative of the blue oak savanna in California. Long-term forage (i.e., herb productivity) records indicate that Sierra Foothill Research and Extension Center (SFREC) has yields comparable to other oak savannas in California (George and others 2001). In regard to temporal 519 GENERAL TECHNICAL REPORT PSW-GTR-217 patterns, 2002 was an average year in terms of herb productivity. From monitoring sites across SFREC between 1980 and 1999, annual average productivity ranged from a minimum of 1.2 Mg ha-1 yr-1 to a maximum of 5.3 Mg ha-1 yr-1 (George and others 2001). However just as we have found (fig. 2), Connor and Willoughby (1997) showed that herb productivity in the blue oak savanna varies relative to canopy cover. The long-term average forage yield at SFREC for sites with 50 percent canopy cover (a canopy level close to the 56 percent mean canopy measured in our research watersheds) was 1.58 Mg ha-1 yr-1, an estimate commensurate with the mean of our three watersheds (1.57 Mg ha-1 yr-1). Acorn production is a spatially and temporally variable element of the productivity budgets of blue oak savannas. For California oaks, annual acorn output varies greatly both interannually and geographically (Koenig and others 1996, Koenig and others 1999). At SFREC, Dahlgren and Singer (1994) reported an average acorn production under blue oak canopies of 2.69 Mg ha-1 yr-1 but during the three years of observation (1990 to 1992) yields varied from 0.3 to 5.8 Mg ha-1 yr-1. In 2002, acorns made a negligible contribution to the NPP of our research watersheds but clearly acorn production is a major component of NPP. It will be interesting to examine the relationship between tree growth and reproductive output as our longterm record develops. In 2002, watershed-level means of leaf litter production varied from 1.41 Mg hayr in WS1 to 0.89 Mg ha-1 yr-1 in WS3 (table 1). These values fall toward the low end of the range reported for blue oak savanna (1.4 to 2.1 Mg ha-1 yr-1, Dahlgren and Singer 1994, Knops and others 1996). 1 -1 Root productivity is a notoriously difficult parameter to measure (Vogt and others 1998) and we acknowledge the limitations of our min-max approach by reporting the large uncertainties associated with the means (table 1). Our watershedlevel estimate of 0.9 Mg ha-1 yr-1 of fine root production (fig.1) is half the production reported by Cheng and Bledsoe (2002). They calculated their estimate from sequential harvesting of ingrowth cores in a nearby blue oak watershed at SFREC. However, despite their physical proximity, direct comparisons between the studies are difficult. Ingrowth cores tend to overestimate root biomass (Steingrobe and others 2000). Moreover, the studies were done in different years and conducted at different spatial scales (topographic sites versus whole watersheds). Nevertheless, the range of our values from individual plots overlapped with the range of value reported in Cheng and Bledsoe 2002. Living woody biomass dominated the carbon pool in these watersheds (table 2). Dead biomass (dead trees and coarse woody debris) accounted for a characteristically small portion of the budget. As Tietje and others (2002) documented, dead biomass does not accumulate in the blue oak savanna. Given the compositional dominance of the annual plants in the herb layer, it is reasonable to infer that dead herbaceous tissue made an even smaller contribution to the biomass pool. We did not measure soil organic matter in the research watersheds, but Dahlgren and others (2003) reported that soil organic matter (surface 15 cm) averages approximately 110 Mg ha-1 under oak canopies and 50 Mg ha-1 in open grasslands for comparable sites at SFREC. 520 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles Tree-Grass NPP Ratio The empirical pattern we described for the relationship between NPP and the treegrass ratio in the blue oak savanna (fig. 2) does not precisely match any of the conceptual models described in House and others (2003). Across these research watersheds, total NPP increased linearly with increasing canopy cover until it saturated at approximately 50 percent cover. Herb productivity did not decline until this saturation point was reached. In contrast, tree NPP increased monotonically with canopy cover across the observed range. Thus, it appears that the interaction between the trees and grasses is contingent upon where they are on the landscape. In the more open areas, tree and grass productivity do not seem to be related, as predicted by McClaran and Bartolome (1989). However, as the proportional dominance of trees increased beyond ~50 percent canopy cover, grass productivity was negatively impacted. These results generally correspond to the findings of Connor and Willoughby (1997) that canopy cover levels of 40 percent to 60 percent do not suppress forage production in the blue oak savanna. The factors that determine the abundance of trees and grasses in savannas continue to be debated (Sankaran and others 2004). Resolution of this debate will require a comprehensive approach to savanna ecology. Our results concerning the productivity of this Mediterranean savanna explicitly incorporated the spatial complexity of these ecosystems. We contend that the uncertainty associated with these NPP estimates provide a crucial and underappreciated dimension to our understanding of this ecosystem. Future work will address the temporal variability in tree-grass productivity relationships and the impact of land-use changes. Given the documented sensitivity of savannas to changes in land use (House and others 2003), we need methods to evaluate the impact of these changes on ecosystem function at relevant spatial scales. The biometric approach described here provides an initial estimate of the resolution of any changes in NPP that we can expect to detect in a small watershed. Acknowledgments This work was supported by the Integrated Hardwood Range Management Program (Project # 00-1) and the California Agricultural Research Station. This project is part of the research program of the Sierra Foothill Research and Extension Center. We appreciate the hard work of our many research technicians, especially Jennifer York, Angela Kong, and Rebecca Wenk. References Callaway, R.M.; Nadkarni, N.M.; Mahall, B.E. 1991. Facilitation and interference of Quercus douglasii on understory productivity in Central California. Ecology 72:1484-1499. Chen, X.; Hutley, L.B.; Eamus, D. 2003. Carbon balance of a tropical savanna of northern Australia. Oecologia 137:405-416. Cheng, X.M.; Bledsoe, C.S. 2002. Contrasting seasonal patterns of fine root production for blue oaks (Quercus douglasii) and annual grasses in California oak woodland. Plant and Soil 240:263-274. 521 GENERAL TECHNICAL REPORT PSW-GTR-217 Clark D.A.; Brown, S.; Kicklighter, D.W.; Chambers, J.Q.; Thomlinson, J.R.; Ni, J. 2001. Measuring net primary production in forests: concepts and field methods. Ecological Applications 11:356-370. Cleveland, W.S.; Devlin, S.J. 1988. Locally-weighted regression: an approach to regression analysis by local fitting. Journal of the American Statistical Association 83:596-610. Cochran, W.G. 1977. Sampling techniques. 3rd edition. Santa Barbara: John Wiley & Sons. Connor, J.M.; Willoughby, B.L. 1997. Effects of blue oak canopy on annual forage production. In: Pillsbury, N.H.; Verner, J.; Tietje, W.D., technical coordinators. Proceedings of a symposium on oak woodlands: ecology, management, and urban interface issues. 1995 19-22 March; San Luis Obispo, CA. Gen. Tech. Rep. PSW-GTR160. Albany, CA: Pacific Southwest Research Station, Forest Service, U.S. Department of Agriculture; 321-326. Dahlgren, R.A.; Horwath, W.R.; Tate, K.W.; Camping, T.J. 2003. Blue oak enhances soil quality in California oak woodlands. California Agriculture 57:42-47. Dahlgren, R.A.; Singer, M.J. 1994. Nutrient cycling in managed and non-managed oak woodland-grass ecosystems. Final Report, Integrated Hardwood Range Management Program. Land, Air, and Water Resources Paper – 100028. University of California, Davis, CA. Dahlgren, R.A.; Singer, M.J.; Huang, X. 1997. Oak tree and grazing impacts on soil properties and nutrients in a California oak woodland. Biogeochemistry 39:45-64. Fahey, T.J.; Bledsoe, C.S.; Day, F.P.; Ruess, R.W.; Smucker, A.J.M. 1999. Fine root production and demography. In: Robertson, G.P.; Coleman, D.C.; Bledsoe, C.S.; Sollins, P., editors. Standard soil methods for long-term ecological research. New York: Oxford University Press; 437-455. George, M.; Bartolome, J.; McDougald, N.; Connor, M.; Vaughn, C.; Markegard, G. 2001. Annual range forage production. University of California Division of Agriculture and Natural Resources: Report 8018. Holland, V.L. 1980. Effect of blue oak removal on rangeland forage production in Central California. In: Plumb T.R., technical coordinator. Proceedings of the symposium on the ecology, management, and utilization of California oaks; 1979 June 26-28; Claremont, CA. Gen. Tech. Rep. PSW-GTR-44. Berkeley, CA: Pacific Southwest Forest and Range Experiment Station, Forest Service, U.S. Department of Agriculture; 314-318. House, J.I.; Archer, S.; Breshears, D.D.; Scholes, R.J., Participants NT-GI. 2003. Conundrums in mixed woody-herbaceous plant systems. Journal of Biogeography 30:1763-1777. House, J.I.; Hall, D.O. 2001. Productivity of tropical savannas and grasslands. In: Roy, J.; Saugier, B.; Mooney, H.A., editors. Terrestrial global productivity. San Diego:Academic Press; 363-400. Jackson, L.E.; Strauss, R.B.; Firestone, M.K.; Bartolome, J.W. 1990. Influence of tree canopies on grassland productivity and nitrogen dynamics in deciduous oak savanna. Agriculture, Ecosystems and Environment 32:89-105. Jenkins, J.C.; Chojnacky, D.C.; Heath, L.S.; Birdsey, R.A. 2003. National-scale biomass estimators for United States tree species. Forest Science 49:12-35. Kay, B.L. 1987. Long-term effects of blue oak removal on forage production, forage quality, soil, and oak regeneration. In: Plum, T.R.; Pillsbury, N.H., technical coordinators. Proceedings of the symposium on multiple-use management of California’s hardwood resources; 1986 November 12-14; San Luis Obispo, CA. Gen. 522 Net Primary Production and Biomass Distribution in the Blue Oak Savanna—Battles Tech. Rep. PSW-GTR-100. Berkeley, CA: Pacific Southwest Forest and Range Experiment Station, Forest Service, U.S. Department of Agriculture; 351-357. Kertis, J.A.; Gross, R.; Peterson, D.L.; Arbaugh, M.J.; Standiford, R.B.; McCreary. D.D. 1993. Growth Trends of Blue Oak (Quercus douglasii) in California. Canadian Journal of Forest Research 23:1720-1724. Knops, J.M.H.; Nashi, T.H.I.; Schlesinger, W.H. 1996. The influence of epiphytic lichens on the nutrient cycling of an oak woodland. Ecological Monographs 66:159-179. Koenig, W.D.; Knops, J.M.H.; Carmen, W.J.; Stanback, M.T.; Mumme, R.L. 1996. Acorn production by oaks in central coastal California: influence of weather at three levels. Canadian Journal of Forest Research 26:1677-1683. Koenig, W.D.; McCullough, D.R.; Vaughn, C.E.; Knops, J.M.H.; Carmen, W.J. 1999. Synchrony and asynchrony of acorn production at two coastal California sites. Madrono 46:20-24. McClaran, M.P.; Bartolome, J.W. 1989. Effect of Quercus douglasii (Fagaceae) on herbaceous understory along a rainfall gradient. Madrono 36:141-153. McCune, B.; Grace, J.B. 2002. Analysis of ecological communities. Oregon: MjM Software Design. Millikin, C.S.; Bledsoe, C.S. 1999. Biomass and distribution of fine and coarse roots from blue oak (Quercus douglasii) trees in the northern Sierra Nevada foothills of California. Plant and Soil 214:27-38. Millikin, C.S.; Bledsoe, C.S.; Tecklin, J. 1997. Woody root biomass of 40-to-90 year-old blue oak (Quercus douglasii) in western Sierra Nevada foothills. In: Pillsbury, N.H.; Verner, J.; Tietje, W.D., technical coordinators. Proceedings of a symposium on oak woodlands: ecology, management, and urban interface issues. 1995 19-22 March; San Luis Obispo, CA. Gen. Tech. Rep. PSW-GTR-160. Albany, CA: Pacific Southwest Research Station, Forest Service, U.S. Department of Agriculture; 83-89. Mitchell, R.J.; Kirkman, L.K.; Pecot, S.D.; Wilson, C.A.; Palik, B.J.; Boring, L.R. 1999. Patterns and controls of ecosystem function in longleaf pine – wiregrass savannas. I. aboveground net primary productivity. Canadian Journal of Forest Research 29:743-751. Omdal, D.W.; William, R.J.; Shaw, C.G.I. 2001. Estimating large-root biomass from breast-height diameters for ponderosa pine in northern New Mexico. Western Journal of Applied Forestry 16:18-21. Reich, P.B.; Peterson, D.W.; Wedin, D.A.; Wrage, K. 2001. Fire and vegetation effects on productivity and nitrogen cycling across a forest-grassland continuum. Ecology 82:1703-1719. Sankaran, M.; Ratnam, J.; Hanan, NP. 2004. Tree-grass coexistence in savannas revisited – insights from an examination of assumptions and mechanisms invoked in existing models. Ecol Letters 7:480-490. Scholes, R.J.; Archer, S.R. 1997. Tree-grass interactions in savannas. Annual Review of Ecology and Systematics 28:517-544. Shlisky, A.J. 2001. Hierarchical relationships between plant species communities and their ecological constraints at multiple scales in an oak woodland/annual grassland system of the Sierra Nevada foothills. Berkeley, CA: University of California, Berkeley. Ph.D. Dissertation. Steingrobe, B.; Schmid, H.; Claassen, N. 2000. The use of the ingrowth core method for measuring root production of arable crops – influence of soil conditions inside the 523 GENERAL TECHNICAL REPORT PSW-GTR-217 ingrowth core on root growth. Journal of Plant Nutrition and Soil Science 163:617622. Tietje, W.D.; Waddell, K.L.; Vreeland, J.K.; Bolsinger, C.L. 2002. Coarse woody debris in oak woodlands of California. Western Journal of Applied Forestry 17:139-146. Vogt, K.A.; Vogt, D.J.; Bloomfield, J. 1998. Analysis of some direct and indirect methods for estimating root biomass and production of forests at an ecosystem level. Plant and Soil 200:71-89. Continue 524