Vegetation Map of the Watersheds Between Kawela and Kamalö Gulches, Island of Moloka‘i, Hawai‘i

Mapping plant communities in the Hawaiian Islands
presents many challenges. Primary complicating factors are
extreme topographic variation in many areas, availability of
suitable imagery, and a large number of invasive plant species
that have become established in an otherwise relatively simple
set of native vegetation units. Additionally, the vegetation in
dry habitats may appear to be very different during the wet
season versus the dry season. Although the land area of the
main Hawaiian Islands is relatively small, two of the islands
(Hawai‘i and Maui) rise to over 3,000 meters (m) elevation, and
many parts of these and the other islands are highly eroded—
resulting in steep terrain. This highly variable topography
leads to problems interpreting the data because of shadows on
the imagery, and to dramatic transitions in plant community
composition and structure over short distances. Cloud cover
and shadows also compromise the imagery, making it difficult
to perform standard classification analysis using digital image
analysis or photo-interpretation. Finally, while the native
ecosystems generally have only a few canopy species or
species associations that dominate the landscape, since human
colonization of this archipelago thousands of species of plants
and animals have been either intentionally or accidentally
introduced and have become established in the native
ecosystems (Smith, 1985; Staples and Cowie, 2001). Invasive
species have led to major alteration of some of the original
plant communities, particularly in lowland habitats, as well as
the incorporation of alien plants into many of the remaining
native vegetation units (Pratt and Gon, 1998; Warshauer,
1998; Pratt and Jacobi, 2009). As a result, a majority of the
lowland communities are now (2012) dominated by introduced
species at various stages of community succession, yielding an
extremely high number of potential map units.
Many approaches have been used to map the vegetation
of the Hawaiian Islands. These include extremely generalized
(1:500,000 and smaller scale) maps depicting broad vegetation
zones (for example, Robyns and Lamb, 1939; Ripperton and
Hosaka, 1942; Krajina, 1963; Pratt and Gon, 1998); midscale maps (1:100,000–1:500,000) displaying generalized
plant communities, such as the Hawai‘i GAP land cover
map (Gon and others, 2006) or the recent LANDFIRE EVT
map by the USGS (U.S. Geological Survey, 2009); as well
as very large scale maps (1:12,000–1:100,000) with units
that display very detailed plant communities with boundaries
that could be identified relatively easily in the field (such as
Mueller-Dombois, 1966; Jacobi, 1990; Shaw and Castillo,
1997; Cogan and others, 2011). This document describes the
methods and results of a project to produce a large-scale map
of the dominant plant communities for a 5,118.5-hectare (ha)
area encompassing the Kawela and Kamalö watersheds on
the Island of Moloka‘i using digital image analysis of multispectral satellite imagery. Besides providing a base map of
the area for land managers to use, this vegetation map serves
as spatial background for the USGS Moloka‘i Ridge-to-Reef
project, which is an interdisciplinary study of erosion and
sediment transport within these watersheds (Stock and others,
2011). Research components of the Ridge-to-Reef project
include an assessment of current vegetation and projected
changes with ungulate control, collecting data on local
climate conditions and surface water flow, modeling sediment
mobilization and transport, and sediment impacts on the nearshore reef ecosystem.