Quantification of spatial variation caused by Messor galla ants using remote sensing in the Koutiala region, Mali
Besides assisting the field team during their daily activities using the eBee drone, the goals of this internship were to 1) quantify the arable land lost due to harvester ants, 2) find a relation between the location of the ant colonies and crop and/or soil type and 3) to improve the Beer-Lambert light absorption formula to correct for spatial variation caused by Messor galla ants in Maize fields of smallholder farms.
Spatial variation in smallholder farms in the Sahelian region makes it difficult to estimate yields using models. One of the reasons for spatial variation is the Messor Galla ant. Messor galla belongs to the guild of harvester ants. These ants collect seeds as this forms the main staple of their diet. The seeds are stored in underground granaries where the biggest workers will mill the seeds with their powerful jaws (Figure 1).
Figure 1: Close-up photo of the head of a major worker ant from the Messor galla species (Ericson 2013).
For the smallholder farmers in the Sahelian zone these harvester do prove to be an annoyance since they create barren spots that can have an area of several square meters (Figure 2). These spots are created at the start of the growing season when the young vegetation is easy to cut. Due to the removal of vegetation the surrounding vegetation receives more light and can therefore benefit. The changed light conditions brake the assumptions of the Beer-Lambert light attenuation formula, namely that the light-extinction coefficient (k) and the leaf-area index (LAI) have to be constant. Therefore an attempt was made to develop parameters that could improve the formula to correct for the spatial variation caused by harvester ants in Maize fields of smallholder farms.
Figure 2: One of the bigger barren spots created by Messor galla ants in a field of cotton.
Quantification of the arable land lost and the relation to soil and crop type
The first step was to mark all possible ant colony locations using the high resolution mosaics based on the eBee-imagery taken from the seven field clusters near Sougoumba (Figure 3). A total of 1251 barren spots were delineated using GIS-programs. To make sure that the marked spots were indeed made by Messor galla ants I visited 1201 barren spots. Of which 1071 were effectively caused by harvester ants. The other barren spots were caused by other factors such as stagnant water, water wells or termites. An analysis of the locations of the colonies with regard to the underlying soil type showed that the highest ant colony densities were situated in deep soils with a sandy to sandy-loamy structure. These soils probably provide the best substrate to build the nest structure. Although most ant colonies were situated in fields of cotton, there is probably no relation between the locations of the ant colonies and the crop type due to the yearly crop rotation system the local farmers apply. The arable land lost due to Messor galla was in general very low. The average size of a barren spot made by a Messor galla colony was 6 m2, in almost all cases no more than 1% of the total field area was lost due to Messor galla ants. This made me conclude that it is not worth using pesticides on Messor galla colonies but rather apply alternative techniques such as the transplantation of better developed sorghum or millet plants from other fields. Since Messor galla ants lack the capacity to cut more mature leaves, transplantation is probably the best solution to regain the arable land that was lost.
Figure 3: The productive field (Cotton) with the highest relative arable land loss of 1.01%. The background image is an orthomosaic made from the UAV flight of Cluster 1 on 20/08/2015.
Parameter development for the Beer-Lambert formula
In this last part I made an attempt to improve the Beer-Lambert light absorption formula by adding 2 parameters (a and b) that would correct for the spatial variation caused by Messor galla ants (Equation 1).
Equation 1: 𝐼=a 𝐼0(1−𝑒−𝑘 𝐿𝐴𝐼 𝒃)
The first parameter (a) would correct for the arable land that was lost. The value of this parameter was calculated using the average arable land lost due to Messor galla ants in maize fields, and equals 0,9988. The calculation of the second parameter proved to be more difficult. The first step was to find a relation between the spatial variation measured in the field and the light extinction coefficient (k). Since k is unknown it was represented by an alternative parameter: the fraction of light absorbed. This value was calculated using data collected by the field-team using a light interception measuring device (ACCUPAR LP-80). The spatial variation was represented by 2 parameters (Pearsons skewness index and Coefficient of variance) that were calculated from perpendicular vegetation index (PVI) – orthomosaics. However no relation between the spatial variation and the light extinction coefficient was found. It is believed that two main reasons contributed to this. First a lot of noise and missing data were found in the data collected with the light interception measuring device. And secondly, the impact of the Messor galla ants is only minor in Maize fields, which means that the distribution of PVI values is barely affected. Although the parameter development to correct the Beer-Lambert light absorption formula was not a success, the line of thoughts used to estimate the parameters could be used in future research.