Background/Question/Methods Background: Land management strategies to address the impacts of rapid climate change on a variety of ecological systems include creating habitat corridors, replacement of invasive species, re-establishment of disturbance regimes, restoration of communities and species translocation. All of the proposed strategies imply that land management in the future will largely be involved in explicitly designing and establishing ecological systems to meet management objectives. Whether we are designing intensive agro-ecosystems or biodiversity reserves, robust ecological systems design is improved by insights gathered from the study of patterns and processes in unintentional systems.
Approach: In three landscapes across an elevation gradient in the Laupâhoehoe Experimental Tropical Forest on the Island of Hawai`i (3200 to 4700 feet elevation) we determined the species-specific distribution patterns of a Hawaiian montane rainforest plant community in response to underlying environmental and successional gradients.  Principle components of environmental and plant community composition were used as explanatory factors in a spatially explicit generalized linear mixed effects model (glmmPQL) for each plant species. The aim is to provide specific ideas about what promotes and constrains particular species in different contexts as well as think about the general implications for the design of similar ecological systems.

Results/Conclusions

Results: Analysis suggested four major environmental gradients in all three landscapes: 1) soil surface type, 2) surface water flow and microtopography and 3) canopy gap disturbance and 4) an environmental complex associated with fallen large diameter tree boles. Community composition trends varied by landscape but all seemed to be associated with soil organic matter content or successional status. Individual species showed diverse responses to potential environmental drivers. The lowest elevation landscape was dominated by species models that emphasized spatial autocorrelation alone (44%). The middle elevation landscape had higher portions of species models involving plant community composition as a significant co-variate (52%). The highest elevation landscape demonstrated species models featuring environmental gradients as significant co-variates (57%). A large number of species had differing model types in the three landscapes implying a large degree of context-dependence over small geographic distances.

Synthesis: Species responses to underlying environmental gradients and community composition patterns were individualistic and context-dependent. Ecological systems design in new contexts must therefore be experimental, adaptive, inspired and unrelentingly localized. The observed diversity of distribution patterns in different contexts, even within species, is a call to the describers of pattern to favor active and iterative participation in localized management projects.