What Are the Limitations of LANDIS-II?

LANDIS-II has broad capabilities and can be used at many different scales for many different questions. However, LANDIS-II is limited in both reality and potential. LANDIS-II, like similar models, is also fundamentally limited by our current knowledge.

LANDIS-II is limited by its intended scale. Processes that are important for understanding fine spatiotemporal scales are not included, stomate opening and closing for example. Such processes may inform the inputs to LANDIS-II - such as the probability of establishment - but are not simulated within the model. Nor is LANDIS-II intended for continental or global scales. The emphasis on spatially dynamic processes precludes efficient computation at these scales.

LANDIS-II is designed for landscapes dominated by woody vegetation. Annual species would not be well represented. LANDIS-II assumes that species can be differentiated by their life history attributes. If many or most of these attributes are unknown, LANDIS-II would produce tenuous results at best. LANDIS-II is not appropriate if succession is driven by hydrology or other un-represented processes.

LANDIS-II is also limited by the extensions written to date. Significant gaps include browsing, land use change, and other important processes - both known and unknown.

Finally, no forest simulation model, including LANDIS-II, should be used for making predictions. We simply do not understand all the processes necessary to produce a prediction similar to what the Weather Service can provide. In addition, LANDIS-II is highly stochastic and will produce highly stochastic results as it includes random selection from various distributions (e.g., normal, uniform, log-normal, etc.) throughout.