Rain shelter for the drought plot, Savilleta LTER site, NM. The shelter will collect all rainfall so that grasses and shrubs in this plot will experience a multiyear drought.

E. Small, J. Elliott (NMT)

Semi-arid shrubland and grassland ecosystems should respond differently to precipitation variability due to contrasting plant traits, nutrient cycling, and surface hydrology. During drought, shrublands may respond more slowly than grasslands because soil resources are concentrated in "islands of fertility" and the vegetation is more deeply rooted and drought-tolerant. If this is the case, then drought may be an important force behind the transition from grassland to shrubland that has been observed in semi-arid ecosystems worldwide. Using rainfall manipulation at twenty-one extensively instrumented plots (grassland, shrubland, and mixed grass-shrub) we are addressing three questions: 1) How do grasslands and shrublands respond to multiyear drought? 2) On what timescales do different ecosystem components respond? 3) Is the plant response simply a direct result of changes in rainfall amount or do changes in surface/vadose zone water cycling lead to feedbacks?

Activities and Results

Our activities during the past year fall into two main categories:
1) Establishment of additional rainfall manipulation plots: At the request of the NSF ecosystems panel, we expanded our experimental design to include 6 pure grassland and 6 pure shrubland plots in addition to the 9 mixed grass-shrub plots already in place. (See web site http://www.ees.nmt.edu/Hydro/faculty/Small/research_2000/drought_folder/drought_main.html for photos and details). To ensure that all plots were subjected to the same conditions, we delayed the onset of water limitation/addition treatment until July 2002. Drought plots receive 50% less rainfall than control plots, yielding a drought similar to that observed in the 1950's. Our monitoring includes the primary components of surface/vadose zone water cycling, nutrient and carbon cycling, and plant productivity and physiology.
2) Monitoring and analysis of components of water cycling and plant-water interactions: We continued our intensive monitoring of soil moisture, ET, and others aspects of the surface water and energy budgets. In addition, we completed 9 'reference storms' at the ecotone plots to assess the coupled plant-water response to summertime rainfall. A 15 mm artificial rainfall event was added to each plot, followed by ~10 days of intensive measurements. Finally, we began a detailed spatial analysis of the patterns of canopy and interspace in grassland and shrubland environments.

We have measured and analyzed two summers of data from grassland and shrubland. Key results include:
a. ET dynamics are nearly identical in grassland and shrubland, controlled primarily by near-surface soil moisture. However, this does not mean that evaporation and transpiration are identical. Instead, evaporation is expected to be much higher in shrubland due to extensive bare soil (~70%).
b. There is intense temporal variability of ET and evaporative fraction (EF) following rainfall events. Drydown is faster in shrubland, however both locations return to low values (ET ~ 0.5 mm/day; EF ~ 0.1) within only a few days.
c. We observed a linear relationship between EF and surface soil moisture (Figure 1), however EF increases more rapidly with soil moisture at the shrubland site.

d. Following rain events, the soil tends to be wetter beneath grass than shrub canopies. Compared to both canopies, interspaces are relatively dry. In general however, infiltration is shallow; soil at depths greater than 45 cm is usually dry.
e. Grass response to a 15 mm rainfall event is more dramatic than for shrubs. Grass plant water potential increased substantially following a rainfall event whereas shrub plant water potential increased only slightly. The transpiration responses, normalized to pre-storm values, were similar: grass transpiration increased by a factor of five whereas shrub transpiration did not even double. This results in greater carbon assimilation by grass.

The implications are as follows: First, shallow soil moisture (0-5 cm) is the key control of ET in these environments, suggesting a large bare soil evaporation component. Second, temporal variability of ET is substantial and must be represented by models of land-atmosphere interactions (e.g., NOAH). Third, vegetation type does not noticeably influence the total flux of water back to the atmosphere, although it influences the partitioning between evaporation and transpiration. Fourth, the primary control on the amount of carbon fixed during the observed storm was the pattern of infiltration: the soil beneath grass was wetter, so this plant type fixed more carbon and lost more water.

Plans

During years 4 and 5, we will continue rainfall manipulation treatment and monitoring at the drought plots at the Sevilleta. This will yield drought and control experiments >2.5 years in duration, including three consecutive summer monsoon seasons (2002-2004). At that point, it will be sensible to reevaluate whether to continue drought treatment or to end the treatment and observe the subsequent recovery. Our measurements and analysis will encompass three main areas of ecosystem function: 1) water cycling: spatial and temporal patterns of soil moisture and soil water potential, surface water redistribution and infiltration, runoff, and evapotranspiration; 2) nitrogen and carbon cycling: mineralized N in soil, soil organic carbon, and carbon assimilation; and 3) plant physiology and productivity: above- and below-ground production, percent cover of canopy and interspace, leaf gas exchange, and water relations. The plant and soil components of this work are greatly enhanced via leveraged support from the Sevilleta LTER.

Our results thus far and those we plan to get from monitoring will better constrain surface and vadose zone water cycling in the extensive valley floor environments of semi-arid regions, in particular with respect to the role of vegetation. Results and data from this process study will be integrated into SAHRA-wide efforts in two ways. First, we will derive effective parameters (e.g., Ks) for LANL model grid cells at the hillslope scale. Second, the changes in these effective parameters caused by our drought experiments will be used as landscape boundary conditions in drought scenarios. This information is critical to understand water balance at the basin scale, and to predict how it changes in response to drought and human-induced land surface change.

Publications and Presentations

Shirley A. Kurc, Eric E. Small, The Influence of Vegetation Type on the Surface Water and Energy Balance in Semiarid Ecosystems.

View poster