Dr. Barbara Bond

Vegetation water use and stream flow at the H.J. Andrews Experimental Forest

GOAL: The long-term of this project is to better understand how vegetation age, structure, and species composition affects hydrological patterns in small watersheds at the H.J. Andrews Experimental Forest.

BACKGROUND: The vegetation cover within watersheds has an important influence on stream flow, but the specific details of this influence are not well understood. In the Oregon Cascade Range, up to 25% of national forest land has been altered by logging activities since the 1930's, resulting in patches of 20 to 40 ha with vegetation of varying ages and species composition (Jones and Grant 1996). It is important to understand how these changes in vegetation affect the hydrologic cycle in order to manage watersheds for multiple uses that include water supply and slope stability.

Following a vegetation-removing disturbance to a watershed, such as fire or harvesting, peak flows as well as total flows of water generally increase for a period of a few to several years (Amthor 1998, Watson et al. in press). Jones and Grant (1996) determined that forest harvesting increased peak discharges in experimental basins in the H.J. Andrews by as much as 50% in small basins and 100% in large basins. After this initial increases in flow, water discharge decreases over a period of time in many watersheds. In at least three long-term datasets (Hubbard Brook, Coweeta, and Eucalyptus regnans forests of southeast Australia), total discharge from watersheds with the vigorous, recovering forests 10-30 years after a major disturbance was actually less than the initial condition with old-growth vegetation.

Some of these changes have good explanations. The initial increase in flow following disturbance is generally attributed to reduced transpiration due to the low leaf area after the disturbance, resulting in a greater amount of surface flow. In a few cases where flow decreased after harvest, it appears that the finely articulated conifer needles of the original forest intercepted significant fog and cloud water, so water input was reduced when trees were removed (Harr 1982, 1986). In most systems, as vegetation recovers following disturbance, the use of water by vegetation for transpiration reduces outflow. However, some of the interactions between vegetation and watershed hydrology are more elusive. Why is there less discharge from watersheds with recovering vegetation than in the original condition? Amthor (1998) proposed that change in atmospheric CO2 levels are affecting forest water use at Hubbard Brook; in Coweeta, it is possible that a shift in species composition in the recovering vegetation has resulted in increased transpiration. In the E. regnans forests of Australia, the overstory trees of the young, recovering forests had a higher leaf area compared with the original old forest, resulting in higher total transpiration (Watson et al. in press). Also in the old forest there was a better-developed understory which used less water per unit leaf area than the overstory (Watson et al. in press). Still, the transpiration per unit leaf area was much higher for young E. regnans trees compared with older trees. This could result from changes in the hydraulic resistance in the trees themselves as they age (Ryan and Yoder 1997). The changes in water discharge in all of these systems apparently result from combination of changes to vegetation structure, composition and age and possibly climate interactions.

H.J. Andrews (HJA) Experimental Forest has maintained records of water discharge from small basins (60-101 ha) for over 35 years. These are part of a paired-basin experiment. Experimental harvests were conducted in the 1960s in half of the basins, with clear pre-treatment and post-treatment measurement periods. The HJA is also fortunate to have an outstanding team of hydrology researchers who are on the forefront of new concepts and analysis techniques. For example, Jones and Grant (1996) introduced a new level of statistical rigor to watershed research and revealed surprisingly strong influences of vegetation succession and roads on peak discharges. Tracer studies by Wondzell (unpublished) have revealed strong diurnal trends in the flow from Watershed 1; flow increases over night and decreases during the day. Newer, high-precision weirs are now able to measure this change in flow directly; these instruments indicate that some of the small watersheds in the H.J. Andrews show strong diurnal variation and others do not (Post and Jones 2001). It is likely that the diurnal variation is influenced by water use by streamside vegetation, but there are no data to support this hypothesis. Missing from the current research effort at the H.J. Andrews are studies that relate vegetation processes to change in hydrological cycle.

RESEARCH OBJECTIVES:

Objective I. Evaluate and quantify the impact of three components of vegetation structure and composition and structure on vegetation water use. These are:

1. Species composition (especially hardwoods vs. softwoods, and Douglas-fir vs. hemlock). Within a decade after harvesting, hardwoods and young Douglas-fir dominate young stands. As forests age, hardwoods become less prevalent and in very old forests, hemlock replaces Douglas-fir.
2. Sapwood basal area. Although total basal area increases with stand age after harvesting, preliminary evidence indicates that sapwood basal area is lower in very old forests than in young, mature forests.
3. Tree size/age. Preliminary evidence from Wind River and other locations indicates that tree water use per unit sapwood area is significantly lower in old growth Douglas-fir than in

Objective II. Compare and contrast measurements of vegetation water use with stream flow measurements on time scales ranging from hours to decades.

EXPERIMENTAL APPROACH:

The study focuses on two small watersheds in the Experimental Forest. WS1 was cut in the mid 1960's and is used to evaluate vegetation structure and function of "young/mature" forests. WS2 is an uncut control watershed that is immediately adjacent to WS1. The last major disturbance in WS2 was about 450 ybp. The study includes these key measurements:
1. continuous measurements of transpiration in "young hardwoods" (we are using red alder as a surrogate for all hardwoods), "young Douglas-fir", "old Douglas-fir" and "old western hemlock" (additional species and age classes may be added over time).
2. estimates of vegetation composition in the two watersheds, accomplished with vegetation surveys.
3. continuous measurements of streamflow (part of the core LTER measurements).

Transpiration is measured at 20 minute intervals with "Granier-type" sapflow sensors (Granier 1996) installed in 5 to 7 trees of each species/age-class, with at least two sensors per tree (more details on sensor positions is provided below). These trees were cored with an increment borer to determine sapwood depth. It is critical to note that the power requirements for sapflow measurements preclude a good random sample of trees throughout the watersheds. Instead, the sample trees lie in a cluster near the base of the watersheds and thus the data to date are unavoidably biased due to the sampling design. In WS1 trees lie along two "transects", one each of alder and Doug-fir, just above the weir. These run normal to the stream through a pocket of each vegetation type up the southern (north facing) slope. The transects are 50m long, and 7 trees were selected along the transect at roughly equal intervals. Due to limitations of power and equipment, we measured red alder only in 1999; in subsequent years we estimated sap flux in red alder based on relationships between red alder and Douglas-fir in 1999. We began measurements in WS2 in 2000. In WS 2 we selected 5 Douglas-fir and 4 western hemlock (all overstory trees) in a transect on the N side of the stream about 50m below the weir.

Vegetation surveys were conducted in 1999 in WS1 and 2000 in WS2 to quantify the species composition and basal sapwood area of all woody vegetation >1 cm diameter in the riparian zones (arbitrarily defined as 100m swath centered on the stream bed) of the two watersheds. In each WS, we established transects normal to the stream every 200m upstream from the weir. The transects alternated from one side of the stream to the other. Along each transect we established contiguous square plots - in WS1 there were five 10m-square plots and in WS2 there were three 20m-square plots on each transect (plot dimensions were determined for the horizontal plane - i.e., they were slope-corrected). Within each plot we measured the diameter and species of every tree greater than 1 cm diameter as well as height and sapwood depth of 5 trees of each species in the plot, systematically selected to represent the size distribution in that plot. From the sample of trees used for measurements of height and sapwood depth, we developed species-specific regression equations to predict sapwood area from DBH outside the bark. Cover (by percent area) estimates of shrubs and herbaceous species were made using the line intercept technique from a diagonal transect running from the SW to the NE corners of the plots, with species identified when possible.

Scaling procedures: Sap flux density for each individual sensor over each 20 minute period was determined from temperature differentials using equations in Granier (1987). These measurements were scaled to the whole-tree and species level generally using the procedures described in Phillips et al. 2002. In red alder we installed sensors at three depths (0-20 cm, 20-40 cm and 40-60 cm) in five trees and we determined the average gradient in sapflow from the outer to inner sapwood. Using this gradient we "scaled" outer flux measurements in the other trees to a whole tree basis, and then divided by the total sapwood area of that tree to come up with the average sap flux density. In Douglas-fir and western hemlock we installed most sensors at a depth of 0-20 cm, but we also installed sensors at 20-40 cm in 4 of the old-growth trees. We combined information from these four trees with sap flux measurements from Douglas-fir at Wind River to analyze how radial gradients in sap flow are affected by site, tree age and seasonal variation. From this analysis we developed a predictive relationship to estimate radial variation based on measurements in the outer 2 cm of the sapwood, and we then used these relationships to estimate whole-tree sap flow over 20 min intervals for each measurement tree. For hemlock we took advantage of radial measurements of sapflow by F.R. Meinzer at Wind River. Meinzer's data show that sapflow declines linearly from the outer edge of sapwood to the sapwood/heartwood boundary. We used this relationship to estimate whole tree sapflow from measurements in the outer 2 cm in hemlock. We found no difference in whole-tree sap flux density for any species or size/age class as a function of distance from the stream, so we averaged the data (for each time increment) over the total number of sample trees to develop the mean sapflux density for measurement period for each species/size class. We multiplied this value for red alder by the sapwood basal area of hardwoods in WS1 to estimate hardwood transpiration. We multiplied this value by the sapwood basal area of all conifers (which is >95% Douglas-fir) to estimate conifer transpiration in WS1. We multiplied the valued for old hemlock and Douglas-fir, respectively, by the sapwood basal areas of these species in WS2 (which account for >95% of the sapwood basal area of all trees in this watershed) to estimate transpiration in WS2. The sap fluxes over 20 min intervals were summed to obtain daily sap fluxes.

In many cases, individual sensors were not functional over periods of several days. Because of the small sample sizes, dropping these individuals from the overall mean could result in large artifacts in the time-series data. Therefore, we interpolated to fill "missing" data based on relationships among the sensors when all functioned properly.

Streamflow is monitored continuously as part of the core LTER program. During the summer months, 90-degree v-notch weirs are used to precisely measure stream flow variations at 15 min intervals.

References:

Amthor, J.S. 1998. Searching for a relationship between forest water use and increasing atmospheric CO2 concentration with long-term hydrologic data from the Hubbard Brook Experimental Forest. Environ. Sci. Div. Publ. 4833, Oak Ridge Nat. Lab., Oak Ridge, TN.

Bond, B.J. and K.L. Kavanagh. 1999. Stomatal conductance of four woody species in relation to leaf-specific hydraulic conductance and threshold water potential. Tree Physiology 19:503-510.

Granier, A. 1987. Evaluation of transpiration in a Douglas-fir stand by means of sap flow measurements. Tree Physiology. 3:309-320.

Harr, R.D. 1982. Fog drip in the Bull Run municipal watershed, Oregon. Water Resour. Bull. 18(5):785-789.

Harr, R.D. 1986. Effects of clearcutting on rain-on-snow runoff in western Oregon: A new look at old studies. Water Resour. Res. 22:1095-1100.

Jones, J.J. and G.E. Grant. 1996. Peak flow responses to clear-cutting and roads in small and large basins, western Cascades, Oregon. Water Res. Res. 32:959-974.

Phillips, N., B.J. Bond, N.G. McDowell and M.G. Ryan. 2002. Canopy and hydraulic conductance in young, mature, and old Douglas-fir trees. Tree Physiology 22(2/3):205-212.

Ryan, M.G. and B.J. Yoder. 1997. Hydraulic limits to tree height and tree growth. BioScience 47(4):235-242.

Watson, F.G.R., Vertessy, R.A., Grayson, R.G. In press. Large scale modeling of forest hydrological processes and their long term effect on water yield. Hydrological Processes, Special Issue on Process Interactions in the Natural Environment.