Mhosisi Masocha

Integrating conventional classifiers with a GIS expert system to increase the accuracy of invasive species mapping

Mapping the cover of invasive species using remotely sensed data alone is challenging, because many invaders occur as mid-level canopy species or as subtle understorey species and therefore contribute little to the spectral signatures captured by passive remote sensing devices. In this study, two common non-parametric classifiers namely, the neural network and support vector machine were used to map four cover classes of the invasive shrub Lantana camara in a protected game reserve and the adjacent area under communal land management in Zimbabwe. These classifiers were each combined with a geographic information system (GIS) expert system, in order to test whether the new hybrid classifiers yielded significantly more accurate invasive species cover maps than the single classifiers. The neural network, when used on its own, mapped the cover of L. camara with an overall accuracy of 71% and a Kappa index of agreement of 0.61. When the neural network was combined with an expert system, the overall accuracy and Kappa index of agreement significantly increased to 83% and 0.77, respectively. Similarly, the support vector machine achieved an overall accuracy of 64% with a Kappa index of agreement of 0.52, whereas the hybrid support vector machine and expert system classifier achieved a significantly higher overall accuracy of 76% and a Kappa index of agreement of 0.67. These results suggest that integrating conventional image classifiers with an expert system increases the accuracy of invasive species mapping.

Frequent burning promotes invasions of alien plants into a mesic African savanna

Fire is both inevitable and necessary for maintaining the structure and functioning of mesic savannas. Without disturbances such as fire and herbivory, tree cover can increase at the expense of grass cover and over time dominate mesic savannas. Consequently, repeated burning is widely used to suppress tree recruitment and control bush encroachment. However, the effect of regular burning on invasion by alien plant species is little understood. Here, vegetation data from a long-term fire experiment, which began in 1953 in a mesic Zimbabwean savanna, were used to test whether the frequency of burning promoted alien plant invasion. The fire treatments consisted of late season fires, lit at 1-, 2-, 3-, and 4-year intervals, and these regularly burnt plots were compared with unburnt plots. Results show that over half a century of frequent burning promoted the invasion by alien plants relative to areas where fire was excluded. More alien plant species became established in plots that had a higher frequency of burning. The proportion of alien species in the species assemblage was highest in the annually burnt plots followed by plots burnt biennially. Alien plant invasion was lowest in plots protected from fire but did not differ significantly between plots burnt triennially and quadrennially. Further, the abundance of five alien forbs increased significantly as the interval (in years) between fires became shorter. On average, the density of these alien forbs in annually burnt plots was at least ten times as high as the density of unburnt plots. Plant diversity was also altered by long-term burning. Total plant species richness was significantly lower in the unburnt plots compared to regularly burnt plots. These findings suggest that frequent burning of mesic savannas enhances invasion by alien plants, with short intervals between fires favouring alien forbs. Therefore, reducing the frequency of burning may be a key to minimising the risk of alien plant spread into mesic savannas, which is important because invasive plants pose a threat to native biodiversity and may alter savanna functioning.

Predicting maize yield in Zimbabwe using dry dekads derived from remotely sensed Vegetation Condition Index

Abstract Maize is a key crop contributing to food security in Southern Africa yet accurate estimates of maize yield prior to harvesting are scarce. Timely and accurate estimates of maize production are essential for ensuring food security by enabling actionable mitigation strategies and policies for prevention of food shortages. In this study, we regressed the number of dry dekads derived from VCI against official ground-based maize yield estimates to generate simple linear regression models for predicting maize yield throughout Zimbabwe over four seasons (2009–10, 2010–11, 2011–12, and 2012–13). The VCI was computed using Normalized Difference Vegetation Index (NDVI) time series dataset from the SPOT VEGETATION sensor for the period 1998–2013. A significant negative linear relationship between number of dry dekads and maize yield was observed in each season. The variation in yield explained by the models ranged from 75% to 90%. The models were evaluated with official ground-based yield data that was not used to generate the models. There is a close match between the predicted yield and the official yield statistics with an error of 33%. The observed consistency in the negative relationship between number of dry dekads and ground-based estimates of maize yield as well as the high explanatory power of the regression models suggest that VCI-derived dry dekads could be used to predict maize yield before the end of the season thereby making it possible to plan strategies for dealing with food deficits or surpluses on time.

Estimating forest standing biomass in savanna woodlands as an indicator of forest productivity using the new generation WorldView-2 sensor

Abstract Accurate and up-to-date information on forest dendrometric traits, such as above ground biomass is important in understanding the contribution of terrestrial ecosystems to the regulation of atmsopheric carbon, especially in the face of global environmental change. Besides, dendrometric traits information is critical in assessing the healthy and the spatial planning of fragile ecosystems, such as the savanna dry forests. The aim of this work was to test whether red-edge spectral data derived from WorldView-2 multispectral imagery improve biomass estimation in savanna dry forests. The results of this study have shown that biomass estimation using all Worldview-2 raw spectral bands without the red-edge band yielded low estimation accuracies (R2 of 0.67 and a RMSE-CV of 2.2 t ha−1) when compared to when the red-edge band was included as a co-variate (R2 of 0.73 and a RMSE-CV of 2.04 t ha−1). Also, similar results were obseved when all WorldView-2 vegetation indices (without the red-edge computed ones), producing slightly low accuracies (R2 of about 0.67 and a RMSE-CV of 2.20 t ha−1), when compared to those obtained using all indices and RE-computed indices(R2 of 0.76 and a RMSE-CV of 1.88 t ha−1). Overall, the findings of this work have demontrated the potential and importance of strategically positioned bands, such as the red-edge band in the optimal estimation of indigeonus forest biomass. These results underscores the need to shift towards embracing sensors with unique and strategeically positioned bands, such as the forthcoming Sentinel 2 MSI and HysPIRI which have a global footprint.

The relationship between satellite-derived indices and species diversity across African savanna ecosystems

Abstract The ability to use remotely sensed diversity is important for the management of ecosystems at large spatial extents. However, to achieve this, there is still need to develop robust methods and approaches that enable large-scale mapping of species diversity. In this study, we tested the relationship between species diversity measured in situ with the Normalized Difference Vegetation Index (NDVI) and the Coefficient of Variation in the NDVI (CVNDVI) derived from high and medium spatial resolution satellite data at dry, wet and coastal savanna woodlands. We further tested the effect of logging on NDVI along the transects and between transects as disturbance may be a mechanism driving the patterns observed. Overall, the results of this study suggest that high tree species diversity is associated with low and high NDVI and at intermediate levels is associated with low tree species diversity and NDVI. High tree species diversity is associated with high CVNDVI and vice versa and at intermediate levels is associated with high tree species diversity and CVNDVI.

Testing utility of Landsat 8 for remote assessment of water quality in two subtropical African reservoirs with contrasting trophic states

Abstract Water quality problems continue on a global scale and this creates the need for regular monitoring using cheaper technologies to inform management. The objective of this study was to test for significant relationships between the field-measured and Landsat 8 OLI sensor-retrieved water quality parameters. The study was carried out in two reservoirs with contrasting trophic states in Zimbabwe. Results show that the Blue/Red ratio had strong predictive relationships with Secchi disc transparency (R2 > 0.70) and turbidity (R2 ≥ 0.65). The Near-infrared/Red ratio was a strong predictor of chlorophyll-a in Mazvikadei (R2 > 0.84) whereas in Lake Chivero, which is more polluted, the red band was the most useful predictor (R2 = 0.69). Overall, our work demonstrates the utility of using Landsat 8 band ratios for remote assessment of water quality in African reservoirs as a value-addition to the traditional field-based methods, which are expensive resulting in data scarcity.

Remote sensing of nutrients in a subtropical African reservoir: testing utility of Landsat 8

Abstract Remote sensing is useful for water quality assessments but current remote sensing applications favour parameters that are easy to detect such as chlorophyll-a. An assessment of the utility of Landsat 8 for detecting nutrients was conducted in Mazvikadei reservoir in Zimbabwe. The main objective was to determine whether nutrients often overlooked by remote sensing and yet are the main determinants of water quality can be remotely sensed. Sampling targeted ammonia, nitrates and reactive phosphorus from May to October 2015. In situ nutrient concentrations were regressed against reflectance derived from Landsat 8 imagery. Strong negative relationships were found between ammonia and the near-infrared band in July (R2 = 0.80, p < 0.05) as well as between nitrates and the blue band (R2 = 0.67, p < 0.05) in June. Overall, the results suggest that the cool dry season is the optimum time to use Landsat 8 for monitoring nutrients in tropical lakes.

Long-term changes in the spatial distribution of lumpy skin disease hotspots in Zimbabwe

Outbreaks of lumpy skin disease (LSD) are reported almost every year in Zimbabwe but not much is known regarding whether the pattern of the disease is changing in response to major socio-economic programmes such as the land reform launched in 2000. In this paper, geo-referenced data of LSD cases was used to detect and map significant LSD hotspots over a 20-year period (1995–2014). The hotspots were then overlaid on top of a land tenure map to explore whether hotspots have spread or persist in some land tenure types. The main results are that LSD outbreaks are on the rise and the disease is spreading throughout the country with areas formerly large-scale commercial farms now experiencing more outbreaks. These results suggest that regular vaccination should be now recommended in most districts in the country.

Modelling Opuntia fulgida invasion in Zimbabwe

The invasion of agro-ecosystems, as well as natural ecosystems by Opuntia fulgida, compromises the ability of these systems to provide goods and services to the society and this has a direct bearing on human livelihoods. Yet, not much is known about the current, as well as potential distribution of this invasive alien species in invaded landscapes, such as the southwestern districts of Zimbabwe. In this study, we apply maximum entropy species distribution modelling to georeferenced occurrence data and a set of predictor variables (that is, climatic, ecological and demographic variables) in order to generate the first distribution maps of O. fulgida in Zimbabwe. Our results suggest that a total of 17 districts in Zimbabwe are vulnerable to invasion by O. fulgida and communal lands are most susceptible to invasion relative to other land tenure systems. We discuss the possible mechanisms of spread and highlight threats to livestock production systems posed by this invasive alien species.

Spatiotemporal patterns of clinical bovine dermatophilosis in Zimbabwe 1995–2014

A retrospective study of clinical bovine dermatophilosis outbreaks and cases for the period 1995–2014 was conducted, using data obtained from the Division of Veterinary Services (DVS). A total of 3856 outbreaks and 26 659 cases of dermatophilosis were reported countrywide during this period. The post rainy season accounted for 37.9% of the outbreaks followed by the rainy season (26.7%), cold dry season (22.1%) and the hot dry season (13.2%). A retrospective space–time scan statistic in SaTScan™ was used to detect clusters. From this study, it was evident that dermatophilosis was spreading from the north-west of Zimbabwe through the central to the north-east during the period 2010–2014. Five clusters were identified mainly in the central and north-western regions of Zimbabwe. The primary cluster was centred at Ungwe, Gokwe district in Midlands; the second, third, fourth and fifth likely clusters were centred at Bonga (Mashonaland Central), ARDA (Mashonaland West), Nsenga (Matabeleland North) and Zanda in Gokwe, respectively. The findings of this study suggest the continued spread of dermatophilosis across the country; as such the Department of Livestock and Veterinary Services are advised to develop measures aimed at managing this spread such as dipping, quarantine, movement control and raising farmer awareness.