The study in this article showed that using rats to evaluate mine risk is a very promising mine-detection method. When three rats were used to evaluate a contaminated area, the success rate was 95 percent, showing that rats can be a speedy and cost-effective means of mine detection.
Trainer treats his rat with a banana, following an indication of a buried mine.
Photo courtesy of APOPO
After an acclimatisation and training period, Cricetomys rats started to evaluate five demarcated boxes in Limpopo, Mozambique. The boxes, 5 metres by 20 metres (5 by 22 yards), were constructed by Menschen gegen Minen (MgM—German for "People Against Landmines") in a strategic minefield along the Limpopo Railway. Vegetation in the area can be characterised as low, relatively dense bush with scattered trees of 5 to 12 metres (6 to 13 yards) high.
MgM staff separated the five boxes with safe lanes that were prepared with a bush-cutter; those safe lanes were manually demined. High vegetation within the boxes was removed as much as possible but was still dense in some parts.
Weather during the test period was variable with rain on the first day (13 Nov. 2003) and sun during the other days (14–18 Nov. 2003).
Each box was evaluated by three rats according to the method as described in "Preliminary Results on the Use of Cricetomys Rats as Indicators of Buried Explosives in Field Conditions."1 Testing was done early in the morning (between 5 and 8 a.m.). After that, temperatures became too high for the rats to operate. Within each box, rats walked parallel lanes 0.5 metre (2 feet) wide and all relevant behaviour was recorded on a test sheet.
Graphic by APOPO/MAIC
The five boxes had a total area of 427.5 square metres (511 square yards) and were divided in subunits of 0.5 square metre (5 square feet). We used letters to distinguish between strong marking behaviour ("S" for scratching the soil or "B" for biting the soil for a long time) and weak indications ("s" for a short scratch or "b" for a short bite) of the rats. Using these recorded indications, a risk value was calculated for each subunit ranging from 0 (i.e., no indications in that and surrounding subunits by the three different rats) up to 6 (i.e., all rats indicated that particular subunit). "S" and "B" indications were scored as 2 while "s" and "b" indications were given a score of 1. An example of how values were calculated for each subunit is given in Figure 1. The risk value of the central subunit is equal to the score of the central subunit divided by two plus the scores of the surrounding subunits divided by 16.
Using this method, each subunit had a risk score and they were divided into the five classes shown in Table 1.
|1 - Score Subunit = 0||no risk|
|2 - Score Subunit = < 0.1||no risk, indications (s or b) close to this subunit|
|3 - Score Subunit = 0.1- 0.5||indications within this subunit or close by (s or b)|
|4 - Score Subunit = 0.5 -1.0||indications within this subunit or close by (S or B)|
|5 - Score Subunit = >1.0||strong indications by one or more rats within subunit|
According to this ranking, subunits were given different colours and maps were constructed for each box (results of each box are given in Figure 2). After the rats tested the boxes, each box was inspected and cleared by an MgM deminer using a metal detector and manual prodding. All objects found by the deminer (mines, bullets, fragments, etc.) were mapped in the same way as shown in Figure 2.
|Figure 2: Grids showing rat indications for each test box and risk maps based on these indications. (Click on an image to enlarge.)
Graphics by APOPO
Table 2 summarizes the results of the evaluation by the rats done on the five boxes in comparison with what was found by the manual deminer. All mines present in the boxes were indicated by the rats and were located in the subunits that were categorized as risk class 4 (one mine) and 5 (19 mines). Of the 20 mines present, 12 were visible due to erosion of the soil. Although clearly visible, the rats did not indicate the mines directly, but rather detected them all within a distance of 1 metre (3 feet). When the rats found a covered mine, they marked the exact spot by scratching directly over the mine.
|Box A||Box B||Box C||Box D||Box E||Total||%||Mines||% sub-units
|Risk class||Number of subunits in each risk class|
In total, seven rats were used to evaluate the five boxes. Three were experienced rats (Johan, Jullie and Josse), and the others were young trained rats (Gilgamesh, Lothar, Respect and Sargon). At least two experienced rats tested each box. Table 3 gives the success scores and number of false positive indications of the individual rats. Indications within 1.25 metres (4 feet) of bullets, mine fragments or detonator pins were not considered as false positives as it is not clear for the moment if these particles were really indicated because of explosive residue on them (they will be tested to determine this). As can be seen from Table 2, there is an obvious marking behaviour by the rats for those items. Of the 70 metal objects found by the deminer, 39 were indicated by the rats (56 percent). Nearly all detonator pins were indicated by the rats (87 percent), while other items scored less frequently (fragments = 53 percent, bullets = 33 percent).
|Rat||Boxes tested (area)||Success score||False positives
S+B/100 sq. m
S+B +s+b/100 sq. m
|Johan||A – C – E (265 sq. m)||9/16 = 56.3%||0.75%||0.75%|
|Jullie||B – C – D – E (362.5 sq. m)||15/20 = 75.0%||0.28%||3.00%|
|Josse||B – C – E (262.5 sq. m)||10/16 = 62.5%||1.14%||1.53%|
|Gilgamesh||A (65 sq. m)||No mines||0.00%||1.54%|
|Lothar||A – D (165 sq. m)||4/4 = 100%||2.42%||3.64%|
|Respect||B (67.5 sq. m)||No mines||0.00%||0.00%|
|Sargon||D (100 sq. m)||0/4 = 0.0%||1.00%||1.00%|
With the exception of Sargon, all rats scored relatively well (mean = 63.3 percent) with very few false positive indications (mean < 0.8 indications per 100 square metres [120 square yards] for the major markings S+B and 1.6 for all markings S+B+s+b). It should be noted that many of the false positive indications given by different rats were clustered, which might indicate an explosives-contaminated spot.
Figure 3: Mean success score of the sequence of three rats that tested the five boxes.
Graphic by APOPO
Although the individual success score might seem low, the overall score on the C, D and E boxes (those containing mines) was 100 percent after three rats evaluated a box (see Figure 3).
The mean time for a rat to inspect a box was 32 minutes/100 square metres (120 square yards), so when a box was inspected by three rats, this was done in 96 minutes. When we include handling and exchanging animals, the total average time to evaluate one 100-square-metre box (120 square yards) was about 116 minutes.
The mine-detection rat is harnessed and linked to the search line. Two leashes are connected to the glider to allow manipulation of the animal's position in the box from the safe lanes.
Photo courtesy of APOPO
The test area was a very dense minefield with 20 mines within an area of less than 30 square metres (36 square yards). Besides the mines, the area was highly contaminated with all kinds of war materials (bullets, detonator pins, mine fragments, etc.), which were also often indicated by the animals, especially the detonator pins. After three rats evaluated a box, all mines present in that box were scored.
The construction of risk maps based on the indications of the animals seems to be a very useful tool as 95 percent of the mines were found in the highest calculated risk area and the other mine in the second highest risk area. Using this method, more than 80 percent of the total area evaluated by the rats could be declared free of mines.
Professor Ron Verhagen, APOPO’s deputy director and chief scientist,
analyses and interprets the animals’ performance in relation to
environmental and other influences. Ron is a biologist and leading
expert in rodent biology. He is a professor at the department of
biology of the University of Antwerp (Belgium).
Frank Weetjens is APOPO’s resident representative in Mozambique and
manages APOPO’s operational field work and country office. Frank is
a paramedic and has extensive experience working in post-conflict
countries in Africa.
Christophe Cox works on the development of the Remote Explosive Scent Tracing concept and manages the Morogoro Centre. Christophe is a product engineer with a lot of experience in eastern Africa.
Bart Weetjens initiated the idea of using rats for landmine detection. His focus is the development of the direct detection concept and the coordination of the Tuberculosis Project.2 Bart is a product engineer and a practising Zen monk.
From the Belgian office, APOPO’s chairman, Professor Mic Billet, handles logistic, financial, administrative and legal matters. Mic is a sociologist and communications expert, and has been a driving force of the project since its earliest stage.
- Verhagen R, Cox C., Machang’u R., Weetjens B. and M. Billet. 2003. "Preliminary Results on the Use of Cricetomys Rats as Indicators of Buried Explosives in Field Conditions." In: Mine Detection Dogs: Training Operations and Odour Detection. Geneva International Centre for Humanitarian Demining. Geneva (ISBN 2-88487-005-5). pp. 175–193.
- The Tuberculosis Project is a study hoping to change the way Tuberculosis is diagnosed using the exceptional sniffing abilities of rats. For more information, please see http://news.bbc.co.uk/1/hi/health/3486559.stm. Accessed 11 Nov. 2005.
2020 Antwerpen - Belgium
Fax: + 32 3 265 3474
University of Antwerp
Tel: + 32 3 474 362 799
Fax: + 32 3 265 3474
Christophe Cox and Bart Weetjens
Sokoine University of Agriculture
PO Box 3078
Tel: +255 23 2600 635
Fax: +255 23 2600 636
University of Antwerp
2020 Antwerpen - Belgium
Tel: +32 3 265 3457
Fax: +32 3 265 3474
Mine Detection Rats
PO Box 649
Tel/Fax: +258 51 23 977