43 
total diversity of a macro system such 
as an ecosystem, blome or continent 
(Suffling et al 1988). The diversity 
literature is dominated primarily by 
studies of alpha diversity (Peet 1974, 
Pielou 1975) in part because it has 
been easier to measure than beta or 
gamma diversity. 
Regardless of the level of diversity 
being examined, two components of 
diversity are recognized. The first is 
richness, or the number of species or 
number of habitats present, and the 
second is evenness, or the 
distribution of individuals or habitats 
among species or landscapes (Peet 
1974, Pielou 1975). Thus, estimates 
of alpha diversity for a given region 
requires two variables, the number of 
species present (also called species 
richness) and the relative 
abundance of each species. Beta 
diversity requires knowing the 
number of habitats in a given 
landscape (also called landscape 
richness) and the relative 
abundance of each habitat across a 
given landscape. 
METHODOLOGY 
Sources of Data. Preprocessing, and 
the Study Area 
The data used to calculate diversity 
are the number of pixels for each 
spectral class in all or a portion of a 
SPOT multispectral image. I have 
personally visited more than half of 
the 423 islands in the study and 
conclude that the spectral classes of 
the preprocessed data correlate well 
to specific habitats or landscapes. I 
assume therefore, that the diversity 
of spectral classes for a given island 
is a measure of habitat or landscape 
diversity. Consequently, I use the 
terms spectral diversity, landscape 
diversity and habitat diversity 
interchangeably. 
Landscape diversity and richness 
was measured directly from a full 
SPOT multispectral scene centered 
on Penobscot Bay, Maine (Figure 1.). 
The image was acquired October 27, 
1988 and totaled 64.25 km 2 of which 
423 islands comprised17.27 km 2 . 
The remaining 46.98 km 2 was 
Fig. 1. SPOT Image scene of Penobscot Bay, Maine showing the larger islands. © [1988] ONES, 
provided by SPOT Image Corporation, Reston, VA. 
comprised of water and mainland. 
Data on mammal species richness 
on 18 islands was provided by 
Crowell 1986. 
Data were extracted from the image 
using GAIA software. GAIA allows 
the display, manipulation and 
analysis of a variety of digital earth 
images as well as the integration of 
vector maps on any 8 bit Macintosh II 
(Podolsky and Morehouse 1990 and 
Podolsky, et. al. 1990). File 
compatibility and data complexity 
dictate that the raw SPOT images be 
preprocessed. Preprocessing is a 
two step procedure. Step one 
converts SPOT data to Macintosh 
format and step two reduces the 16.7 
million reflectance values (256