SA-SOURCED SOFTWARE USED TO IMPROVE DESERT LOCUST MONITORING AND FORCASTING IN MAURITANIA

Computing

The International Association for Applied Acridology (the study of grasshoppers and locusts) is using SAS software - sourced in SA - to determine which plant habitats favour locust communities in the desert region of Mauritania in North Africa.

According to Jeffrey Lockwood, director at the Association, the ability of any single nation to sustain a critical mass of expertise in grasshopper and locust pest management has diminished, yet the quality of geographically dispersed experience and knowledge is extremely high.

The association aims to bring together the world’s leading Acridology experts to develop and provide unbiased analyses, along with culturally, technologically, economically, and environmentally appropriate methods, for managing locust and grasshopper outbreaks.

Grasshopper and locust outbreaks reduce standards of living, displace human populations, induce famine, and erode environmental quality. To combat this, the highest level of practical expertise and experience is essential to build the capacity of agricultural communities to implement safe and effective prevention and control programmes.

The Association has developed a database with 241 variables and over 20,000 records, which contains detailed information on locust occurrence, insecticide applications, plant habitat, meteorology, landscape and soil parameters. "We selected SAS software to maintain the database because of its high quality and flexibility in analysing complex datasets, its relative simplicity and the wide availability of SAS programmers," says Lockwood. "What sets SAS apart from other software packages is that it can debug data very easily - it provides a detailed log of all analytical steps so that errors in programming can be detected - and it has the capacity to unravel complex relationships with relatively simple programming," he adds.

The SAS licence was obtained through the SAS regional headquarters in SA in 2001. Initial steps focused on debugging the existing database - a number of erroneously entered records that had slipped through previous proof-reading rounds were easily identified using summary statistics.

A spatial component was then added to the analysis by using SAS code to create a grid of cells, each measuring approximately 50km by 50km. This grid is used to assign each geo-referenced observation to a particular cell. In a preliminary analysis, counts of gregarious locusts were totalled by grid, month, and year, and regressed against presence counts of 68 different plant species.

It was found that some species of plants do not appear to have any impact on locust abundance and these may be ignored during surveys. However, analysis is continuing, and the next step will be to relate locust and host plant abundance to soil and meteorological parameters, such as rainfall, relative humidity, soil moisture and soil type in areas where these measurements were taken.

"Understanding where and when locusts are most likely to occur will lead to a better understanding of locust ecology and an improvement in locust management. Focusing survey and control operations to those areas will lead to a reduction of survey costs and potential reduced pesticide use through more precise control operations," Lockwood concludes.

ICT World