Quantifying the deviation of landscape diversity from potential natural vegetation with Shannon's entropy

Abstract

In the past 20 years, several metrics have been developed to quantify various aspects of landscape structure and diversity in space and time, and most have been tested on grid- based thematic maps. Once landscape patterns have been quantified, their effects on ecological functions can be explained if the expected pattern in the absence of specific processes is known. This type of expected pattern has been termed a neutral landscape model. In the landscape-ecological literature, researchers traditionally adopt random and fractal computer-generated neutral landscape models to verify the expected relationship between a given ecological process and landscape spatial heterogeneity. Conversely, little attention has been devoted to distribution patterns of potential natural vegetation (PNV) as an ecological baseline for the evaluation of pattern-process interactions at the landscape scale. As an application for demonstration, we propose a neutral model based on PNV as a possible reference for a quantitative comparison with actual vegetation (ARV) distribution. Within this context, we introduce an evenness-like index termed "actual-to-potential entropy ratio’ (HA/P=HARV/HPNV, where H is Shannon’s entropy). Results show that, despite the hypothetical character of most PNV maps, the use of PNV distribution as a baseline for a quantitative comparison with ARV distribution may represent a first step towards г general model for the evaluation of the effects of disturbance on vegetation patterns and diversity.