In LTER V (2000-2006, we propose to expand on the Jornada desertification model in order to understand and predict variation in ecosystem properties and dynamics across multiple scales, rather than focusing only on average conditions through time or across space. This shift in focus to connections across scales was necessitated by our inability to predict temporal and spatial variation in vegetation dynamics (i.e., shrub invasion success and grass persistence and recovery) in the previous model. In particular, we have observed that fine-scale events associated with woody plant invasion can amplify or cascade to result in desertification over increasingly larger areas through time (Peters et al. 2004). However, important questions remain unresolved, including:
(1) How do we integrate diverse observations about flora, fauna, soils, hydrology, climate, and human populations across spatial and temporal scales to improve our ability to understand current and historic patterns and dynamics?
(2) How do processes interact across a range of scales and under different conditions to drive desertification dynamics and constrain the conservation of biological resources? More specifically, under what conditions do fine-scale processes cascade to affect larger spatial scales, and under what conditions do broad-scale drivers constrain or overwhelm fine-scale processes to influence system dynamics?
(3) How do we disentangle interactions driving landscape dynamics such that we can predict spatial and temporal variation in ecosystem properties related to desertification? How can we use knowledge of desertification dynamics to more effectively promote the conservation of biological resources and the recovery of grasslands?
In addressing these questions, JRN LTER V will focus on quantifying interactions between ecosystem processes and patch structure (i.e., area or size, composition, spatial arrangement of bare and vegetated patches at multiple scales) as a means of improving our mechanistic understanding and ability to integrate, predict, and extrapolate across spatial and temporal scales up to and including those relevant to land management and policy.
Our overall hypothesis is that spatial and temporal variation in ecosystem dynamics is the result of patch structure interacting with transport vectors (wind, water, animals) and environmental drivers (e.g., precipitation, temperature, human activities) to influence cross-scale resource redistribution. These interactions feed back to patch structure and dynamics to cause cascading events (Fig. 3) with affects on ecosystem goods and services (Fig. 4). Historic legacies and geomorphic templates are important modifiers of this relationship.
Here we describe specific hypotheses to be tested by integrating long-term data with a strategic suite of new multi-scale experiments designed to elucidate how interactions and feedbacks play out across scales to determine pattern-process relationships and spatial and temporal variation in system dynamics. We also describe how we will use our spatially explicit, multi-scale, multi-transport vector ENSEMBLE simulation model to synthesize and integrate this information in order to generate new testable hypotheses and to predict future system dynamics under alternative environmental conditions and management regimes (Figs. 5, 6).
LTER V seeks to elaborate on the landscape linkages framework that emerged in the latter stages of LTER IV by: (1) testing specific elements of our framework using existing long-term studies, (2) conducting a suite of new integrated, cross-scale experiments for three geomorphic units and a nearby suburban interface with the Jornada, and (3) forecasting alternative future landscapes under a changing environment that includes socioeconomic processes and explicit interactions with ecosystems. We will also expand our modeling efforts by integrating a fine-scale model of vegetation and soil water dynamics (ECOTONE, Peters 2000) with existing transport models of wind (SWEMO, Okin et al. 2006), water (Műller 2004), and animal dynamics (under development with ARS funding). This multi-scale, spatially-explicit ENSEMBLE model complements our field studies, and will be used to improve our understanding of complex interactions among system components and to make predictions about future states and dynamics.
Our conceptual framework builds on previous frameworks that seek to quantify the redistribution of resources within and among a hierarchy of spatial units (Fig. 7) (Peters et al. in review). We hypothesize that interactions among five key elements connect spatial units across scales to generate complex dynamics: (1) historical legacies that include climate and past disturbances, (2) environmental drivers, such as weather, current disturbance regimes, and human activities, (3) a geomorphic template that includes both local properties, such as soil texture and geomorphology, and the context and arrangement of spatial units, and (4) multiple horizontal and vertical transport vectors (fluvial, aeolian, animal) that interact to (5) redistribute resources within and among spatial units (Fig. 3) and (Fig. 4). Interactions and feedbacks among these elements within and across spatial scales generate threshold changes in patch structure and dynamics that result in cascading events and associated broad-scale conversion from perennial grasslands to shrub dominance. Feedback mechanisms operate at multiple scales: via plants, animals, and soils to influence transport vectors and resource redistribution within each spatial scale, and via the spatial arrangement of vegetation patches, transport vectors, and spatial connectivity in resource redistribution among scales. Complexity, contingency, and the interdependence of system components are major obstacles to prediction in ecosystem science. We believe that the integration of the above elements is a powerful approach for advancing our understanding and forecasting ecosystem dynamics. We propose to test the proposition that an explicit accounting of processes in the context of patch and geomorphic structure, spatial context, and cross-scale interactions will improve our predictive capabilities by resolving what heretofore has been a large pool of unexplained variance.
Our landscape linkages framework addresses three aspects largely missing from previous desertification frameworks. First, dominant processes and dominant vectors interact and change through time and across space. These cross-scale interactions often generate unexpected dynamics (Peters et al. 2004a). For example, both wind and water can operate as broad-scale drivers for soil, nutrient, litter, and seed redistribution. In arid systems, wind is often the dominant broad-scale driver on sandy soils with little topographic relief, and water is the dominant broad-scale driver on fine-textured soils along topographic gradients. At fine scales, the importance of each driver depends on factors such as surface soil texture, patch structure of the vegetation, and topographic position. In addition, large and small animals redistribute resources and seeds within and among spatial units across a range of scales. Interactions among these multiple transport vectors operating across spatial and temporal scales determine the relative importance of within versus among spatial unit processes to ecosystem dynamics. Second, spatial context (i.e., location or adjacency of spatial units) and patch structure influence resource redistribution both within and among spatial scales. In arid systems, a key characteristic is the influence of bare patches on horizontal resource redistribution by wind and water. Landscapes with highly connected bare soil patches are expected to promote the rapid movement of materials and disturbances over greater distances, whereas landscapes with low connectivity may have barriers or spatial configurations that restrict horizontal movement of materials (e.g., Ludwig et al. 2005). Highly connected landscapes for one vector may have low connectivity for another vector. For example, grasslands with many small bare patches have low potential for wind and water erosion, yet can be highly connected for grass seed dispersal. By contrast, shrub-dominated systems with large bare patches are highly connected for wind and water erosion and shrub seed dispersal, yet have limited connectivity for grass seed dispersal. Third, soil-geomorphic organization is a primary determinant of the importance of particular vectors and spatial context that controls resource redistribution. For example, sites on sandy soils with high infiltration rates are expected to experience relatively short distances of horizontal redistribution of nutrients by water compared with sites located on slopes with silty soils and physical soil crusts that limit infiltration rates and promote overland flow. Further, the spatial context of geomorphic units in arid and semiarid systems is predictable within physiographic regions, and these units have predictable relationships with climate and soil development (Monger and Bestelmeyer in press).