95th ESA Annual Meeting (August 1 -- 6, 2010)

COS 91-7 - Nitrogen Dynamics: A Graph-Theoretic Approach

Thursday, August 5, 2010: 3:40 PM
406, David L Lawrence Convention Center
Kusum Naithani and Erica Smithwick, Department of Geography, The Pennsylvania State University, University Park, PA
Background/Question/Methods

 Elevated nitrogen deposition alters soil biogeochemistry and associated ecosystem processes that can lead to plant mortality and decline in stand productivity.  Understanding the effects of elevated nitrogen deposition as a result of different land use practices is critical for predictive understanding of ecosystem productivity and resilience.  Nitrogen cycling in terrestrial ecosystems is a nonlinear, complex and dynamic process which can be characterized by coupling of physical, chemical and biological processes and their feedback at different spatial and temporal scales.  This paper presents a graph theoretic approach to characterize the behavior of nitrogen dynamics in terrestrial ecosystems.  Different pools and observations are presented as nodes and their direct and indirect effects are displayed as edges connecting them.  The directed network allows the percolation of perturbations among couplings in a desired direction.  Additionally, we used a Probabilistic Boolean Network (PBN) approach to test the effect of elevated nitrogen deposition on ecosystem productivity by comparing the network topology of networks with and without (control) elevated nitrogen input.  Prior information about the processes involved in nitrogen cycle and its influence on plant growth is used to determine a set of rules that encodes causal relationships and assign associated uncertainty to the strength of each relationship.  Change in one component leads to the change in another component of the network which is used as an input to a Boolean form of the process network to study the effects of elevated nitrogen deposition on ecosystem productivity.

Results/Conclusions

We found that network is robust against simulated perturbations via manipulation of edge weights and random attacks (node or edge removal) and showed great vulnerability under targeted attacks. Introduction of fertilizer nitrogen input to network increased closeness centrality by ~45% and decreased betweenness centrality by ~10%.  PBN predicts decline in ecosystem productivity under chronic nitrogen deposition in different soil moisture regimes via :1) increased acidity in soil and consequently leaching of available nitrogen to ground water, and 2) higher uptake of available nitrogen leading to fine root stress and nutrient toxicity. Introduction of drought under chronic nitrogen deposition increases plant mortality by 60% than control network.  Simulation of ecosystem behavior from our model indicates the potential vulnerability of key nitrogen cycling process and provides basis for future experimental and field research to test the outcomes of our model