Abstract:
This study aimed to evaluate the potential use
of normalized difference vegetation index (NDVI) from
satellite-
derived remote sensing data for monitoring
rodent abundance in semi-arid areas of Tanzania. We
hypothesized that NDVI could potentially complement
rainfall in predicting rodent abundance spatially and tem-
porally. NDVI were determined across habitats with differ-
ent vegetation types in Isimani landscape, Iringa Region,
in the southern highlands of Tanzania. Normalized differ-
ences in reflectance between the red (R) (0.636–0.673 mm)
and near-infrared (NIR) (0.851–0.879 mm) channels of
the electromagnetic spectrum from the Landsat 8 [Opera-
tional Land Imager (OLI)] sensor were obtained. Rodents
were trapped in a total of 144 randomly selected grids
each measuring 100 × 100 m 2 , for which the corresponding
values of NDVI were recorded during the corresponding
rodent trapping period. Raster analysis was performed by
transformation to establish NDVI in study grids over the
entire study area. The relationship between NDVI, rodent
distribution and abundance both spatially and tempo-
rally during the start, mid and end of the dry and wet sea-
sons was established. Linear regression model was used
to evaluate the relationships between NDVI and rodent
abundance across seasons. The Pearson correlation
coefficient (r) at p ≤ 0.05 was carried out to describe thedegree of association between actual and NDVI-predicted
rodent abundances. The results demonstrated a strong
linear relationship between NDVI and actual rodent
abundance within grids (R 2 = 0.71). NDVI-predicted rodent
abundance showed a strong positive correlation (r = 0.99)
with estimated rodent abundance. These results support
the hypothesis that NDVI has the potential for predicting
rodent population abundance under smallholder farming
agro-ecosystems. Hence, NDVI could be used to forecast
rodent abundance within a reasonable short period of
time when compared with sparse and not widely available
rainfall data.h
