1999-2002. NSF Ecology Program DEB-9815607 NSF-Ecology. Facilitation, competition and woody patch dynamics in a savanna parkland

Project Abstract

(Project Proposal)

Savanna research has emphasized tree effects on soils, microclimate and grasses. Little is known of tree-tree interactions and how these might influence savanna function and dynamics. Furthermore, savanna research has largely been conducted where individual trees are scattered throughout a grassy matrix. However, the formation of tree or shrub 'islands' is a common, but little-studied phenomenon in many dry lands.

In this project, we seek to quantify mechanisms of facilitative and competitive interactions between an overstory tree, honey mesquite (Prosopis glandulosa), and associated understory shrubs in a subtropical savanna ecosystem. Our goal is to elucidate mechanisms of patch dynamics associated with the physiognomic transformation of grassland and open savanna to closed-canopy thorn woodland. Our overall objective is to quantify woody plant interactions in a subtropical savanna to: (a) elucidate mechanisms associated with the initiation and development of tree-shrub islands; and (b) predict how species interactions in these islands change across topoedaphically diverse landscapes with time.

Our recent studies suggest (a) Prosopis plants facilitate the establishment of other woody species beneath its canopy to create distinctive tree-shrub clusters; (b) woodlands develop as new clusters form and existing clusters grow and coalesce; (c) facilitation gives way to asymmetric competition whereby (d) mortality of the founding Prosopis plant is hastened by root competition with understory shrubs on some soils; on others (e) Prosopis suppresses development of understory shrubs and a negative feedback develops whereby plants co-exist in dynamic equilibrium. This conceptual model is based on space-for-time field sampling. Experimental confirmation of inferences our studies of pattern and explicit identification of mechanisms controlling patch dynamics are therefore our specific goals.

Specifically, we will test the hypotheses that: 1) Prosopis facilitation is passive: the plant is merely a perch site for birds dispersing seeds of other woody species; 2) Prosopis facilitation is active and the result of its modification of microclimate (1 , early) and soil fertility (2 , later); or 3) The deep-rooted Prosopis nurse plant, benefits seedlings and shallow-rooted understory shrubs by the nocturnal transport and redistribution of deep soil moisture to drier surface soils (hydraulic lift), but, (4) The efficacy of this process is mediated by topoedaphic properties (presence or absence of an argillic [clay-rich] horizon) known to influence soil hydrology and rooting patterns; 5) At latter stages of woody patch development, overstory-understory interactions become highly asymmetrical such that (a) on sites with a well-developed argillic horizon (=shallow soils) understory shrubs exerting a strong competitive influence over Prosopis, whereas (b) enhanced growth of Prosopis on lowland sites receiving run-on or on upland sites lacking the argillic layer (= deeper soils) results in a suppression of understory shrub growth.

Our approach will involve a combination of selective removal and reciprocal transplant experiments that will be coupled with manipulations of resources (light, nutrients and soil moisture) and measurements of plant growth, foliar chemistry, leaf gas exchange, plant/soil water relations, and 2H/ 18O natural abundance. Results will increase our ability to generalize the roles of facilitation and competition in succession and mechanisms of woody plant increase in grasslands and savannas, enhance our ability to represent and parameterize spatially explicit simulation models for tree-shrub-grass ecosystems, and contribute directly to the development of a sound basis for monitoring, managing, and manipulating grasslands and savannas ecosystems.

2001-2004. NSF-Ecosystem Studies DEB-9981723 Scaling Soil C and N Storage in a Changing Savanna Parkland Landscape - Spatial Structure, Prediction and Uncertainty Assessment

It is widely known that woody plants significantly alter soils subsequent to their establishment. This leads to the formation of ‘fertile islands’ in shrublands, woodlands and savannas. Although the biophysical characteristics of ‘islands of fertility’ are well-known, we know little of their rates of formation, the extent of within-island variability or how to develop an integrated landscape-scale assessment of nutrient pools in relation to the spatial distribution of plant life forms or growth forms and their change with time. 

The proposed study will be conducted at a savanna parkland landscape where woody plant encroachment over the past century has been well-documented. We will quantify spatial variation in soil organic carbon (SOC), total nitrogen (TN) and root mass (0-20 cm) in relation to herbaceous and woody patch age-states across a topographically heterogeneous landscape, assess within-patch spatial structure of these variables, and generate landscape-scale estimates of SOC, TN, and root mass pools. Our goal is to understand the spatial structure of landscape-scale SOC, TN, and root mass and quantify changes which have accompanied the shift from grass to woody plant domination. The overall objectives are to: (a) compare and contrast patterns of SOC, TN, and root mass distribution within different woody plant and herbaceous patch types; (b) quantify the spatial structure of SOC, TN, and root mass distribution at the landscape scale using geostatistical tools; (c) utilize information on spatial structure and uncertainty patterns to estimate the intensity and distribution of ‘point samples’ needed to quantify landscape-scale SOC, TN, and root mass; (d) determine the geostatistical relationships between aboveground patch features (which can be readily quantified on aerial photos or satellite imagery) and SOC, TN, and root mass; and (e) quantify changes in SOC and TN pools accompanying woody plant expansion using information derived from historical (1930-2000) aerial photography; and (f) compare these estimates to those generated from linked biogeochemistry-succession models.

Specifically, we will test hypotheses that (1) there are significant spatial structures in near-surface SOC, TN, and root mass distributions across savanna parkland landscapes; (2) patterns of SOC, TN, and root mass vary within patches as influenced by individual plants, patch type (e.g., discrete cluster,  groves, closed-canopy woodlands) and patch age-state; (3) SOC, TN, and root mass distributions in savannas have nested spatial structures, with local structures induced by vegetation pattern and large-scale structure determined by geomorphology and surface hydrology; (4) efficient point sample distributions across landscapes can be designed based on knowledge of spatial structure and uncertainty distribution generated from stochastic simulation and univariate statistics; (5) accuracy of landscape-scale SOC, TN, and root mass inventories based on remotely sensed vegetation data can be substantially enhanced using co-kriging and stochastic simulation at different scales; and (6) an enhanced deterministic extrapolation model, based on an understanding of the nested spatial structure and uncertainty, will generate acceptable estimates of SOC, TN, and root mass compared against the estimates using kriging and co-kriging. This approach will then be used to estimate historical changes in landscape SOC, TN, and root mass distribution associated with woody plant expansion as quantified on historical aerial photos.

1996-2001. USDA-NRICGP Ecosystems Program, 96-35101-3210 Historic Changes in Tree/Grass Abundance: Implications for C and N Storage 

The abundance of woody plants has increased in grasslands and savannas throughout the world in recent history. In addition to direct and immediate effects on animal production systems, these dramatic changes in ecosystem structure have the potential to profoundly influence hydrology, biogeochemistry, biodiversity, and future land use options in the affected areas.

Our recent work has shown that rates and patterns of vegetation change in southern Texas have been temporally dynamic, spatially variable, and constrained by edaphic/geomorphic controls. The dominant woody plants in these systems are both highly productive and capable of symbiotic nitrogen fixation. As a result, their displacement of grasses has multiple consequences for fundamental biogeochemical processes. Notable among these are the cycling and sequestration of two elements critical to the climate system: carbon and nitrogen. Unfortunately, we have little knowledge of how shifts in plant physiognomy have affected C and N dynamics across topoedaphically variable landscapes. A recently completed project has linked a process-based ecosystem model (CENTURY) to a grassland-to-woodland succession model to evaluate the magnitude of changes in plant and soil carbon and nitrogen that might have resulted from alteration of grazing and fire regimes since European settlement.

This modeling experiment has generated explicit predictions regarding C and N storage in plants and soils across the landscape. We propose to test these predictions. If our representation of success ional processes and mechanisms are adequate, if grassland and forest versions of CENTURY have been appropriately modified for use in savannas and woodlands, and if the CENTURY-successional model linkages have been handled properly, then we should be able to reasonably simulate changes in soil C and N associated with changes in gross vegetation composition and structure. The data generated in conducting these "validations" will enhance our ability to correctly represent and parameterize simulation models for tree-shrub-grass ecosystems, and will also provide a wealth of information on the structure and function of a subtropical savanna woodland with functional counterparts in Central and South America, Africa, Southeast Asia, and Australia.