Cassava in flower

men planting trees

Principal Investigator: Prof. Roy Ellen
Project dates: 2009 – 2011
Funding: British Academy (ASEASUK Research Committee) £4000

A comparative study of the sociocultural concomitants of cassava diversity in four eastern Indonesian populations in relation to environmental security

Aims: The objectives are to examine local knowledge and practices regarding cassava (Manihot esculenta) diversity in three local populations (Debut on Kei Kecil, Rouhua in the Nuaulu area of south Seram and Buano, to the west of Seram), in relation to agricultural histories, ecology and food culture. It is hypothesized that the number of landraces (locally recognized varieties) identified and the proportion of ‘bitter’ to ‘sweet’ types would correlate with the extent of cassava dependency.


1. The work in Debut (August 2009) was conducted by Ellen (the grant-holder) and Dr. Hermien Soselisa of Pattimura University, that in Rouhua (August 2009) by Ellen, and that on Buano (October 2009) by Soselisa. The research was conducted under the terms of reference of a Memorandum of Understanding between the University of Kent and Pattimura University, Ambon.

2. Fieldwork has involved interviews with farmers, food processors, traders and consumers of cassava; mapping of agricultural plots; preference ranking of landraces for planting and eating; photography of cassava landraces and management techniques; measurements of plants in situ, collection of leaf specimens for subsequent analysis (morphology, DNA analysis) and tuber specimens for morphological and chemical analysis. In particular, we wished to test the extent to which distribution of cyanogenic glucoside toxicity was reflected in how subjects allocated landraces to categories, and what the consequences of this might be.

3. Debut is situated in a relatively dry, deforested shrubby savanna area with some relict forest. Cassava is grown in a mixture of swiddens (kebun ladang), dryfields near the village and in a government-sponsored development project zone. We recorded approximately 39 named landraces, 19 of which were described as kasbi (i.e. sweet) and 20 as enbal (i.e. bitter), and collected data on the characteristics of each, and on management techniques. The number, type and proportions of landraces are in a continual state of flux, with much movement of germplasm between villages and between Kei and other parts of Indonesia, particularly Seram. By comparison Nuaulu in Rouhua  recognize about 17 named, and predominantly ‘sweet’, landraces, while on Buano farmers recognized 12 landraces, two of which were regarded as ‘bitter’.

4. In Debut enbal is reportedly better on drier and poorer soils, harvesting can be delayed and is more resistant to pig depradations. By comparison, kasbi was less flexible but produced better tubers for cooking. Subjects insisted on separate preference rankings for kasbi and enbal: management factors proved to be more important than taste. All landraces frequently flower and fruit, though virtually all planting is through clonal selection. F1 hybrids dispersed via cockatoos (enbal kanar tean) do not produce good tubers, though may in the long-term be a means of introducing further genetic variability into local stock. Farmers are aware that toxicity is variable genetically, but also heavily microenvironmental. All enbal is processed and deaths or long-term ill-effects are seldom reported.

5. Using ‘Bradbury’ picrate tests on portions of cassava tuber, we found that for 23 landraces collected in Debut the mean toxicity was 359 ppm (parts per milliliter). There was a considerable amount of overlap between landraces designated kasbi and those designated enbal, though kasbi landraces showed a mean of 182 and enbal 406. Interestingly, tests conducted on 9 landraces from Rouhua in South Seram also showed extensive variation in toxicity but a lower average of 121, though the one landrace marked out as ‘bitter’, kasipii paru (from Buton) was 200. Using leaf measurements, we have also been able to show a strong local correlation between leaf shape and perceived toxicity, and evidence for the possible selection for perceptual distinctiveness e.g. enbal lislis. The tests conducted in Buano proved inconclusive.

6. We conclude that folk classification provides a framework for making sense of toxicity, though high micro-environmental variation means that it is by no means a perfect guide. Local differences in number and type of landraces recorded reflect ecology, environmental change, previous food histories and human population movement and contact. DNA analysis of 32 leaf specimens from Rouhua and Debut currently being undertaken in Bandung is beginning to shed interesting light on the genetic relatedness of existing landraces, patterns of historical movement and factors influencing toxicity. Our main conclusions from the DNA work at this stage indicate a weak correlation between genetic relatedness and toxicity and perceptions of toxicity, suggesting that toxicity is largely phenotypic and micro-environmental, and as a genetic trait is repeatedly re-selected for by farmers in new and different habitats. Similarly, farmer evidence for the geographic provenance of individual landraces does not correlate well with genetic relatedness, suggesting that naming follows phenotypic resemblance rather than firm evidence based on traceable germplasm. On the other hand, the DNA evidence shows close genetic relatedness between most of the larger number of Debut landraces and a distant genetic relatedness between all of the smaller number of Rouhua landraces, strongly suggesting that farmers in Debut are much more active in selecting propagative material than the Nuaulu.

7. While wet micro-environments, low population density, extant lowland forest and complex vegecultural farming and agroforestry (e.g. Rouhua and Buano) favour low diversity with fewer toxic landraces, in Debut a dry micro-environment, higher population densities, deforestation, denudation of sago swamp, favour high diversity with a significantly larger majority of toxic landraces. High levels of toxicity in cassava is generally understood as a response by the plant to poor soils and low rainfall. Agricultural change in Debut (in particular the externally-funded project ‘Pola Usaha Tani Berbasis Enbal’), with its more rigid management techniques has potential to raise overall productivity, but is more selective in planting of cassava landraces. Landrace diversity is, rather, maintained in traditional swiddens under the personal supervision of individual farmers.

8. While we did not undertake a systematic survey of the health effects of dependency on cassava landraces with high toxicity levels, it is clear that local processing techniques are effective in reducing toxicity levels in consumed food, while the general adequacy of diet provides a buffer against toxicity in cassava tubers processed in ways which less effectively remove cyanogens. This was found in Debut, Rouhua and in Buano. There may be lessons to be learned in other parts of Indonesia (e.g. West Java) and in east Africa, where cassava toxicity is less well managed, and where it has resulted in major health problems.

9. Consumption of cassava is likely in increase in Maluku, as soils and water availability deteriorate through global warming effects, deforestation and more intensive farming, replacing more traditional starch tubers and sago. In Kei, this process began about 100 years ago and enbal has progressively replaced sago in the diet, as more recently rice is replacing cassava. The role of cassava (particularly enbal) in Kei culture is significant partly because it can be processed in ways (hard biscuits) that resemble traditional sagu lemping in this part of Maluku. Indeed, enbal features as an item in traditional feasting, gift-giving and rituals in parts of the Kei archipelago in ways that suggest that in these respects also it is replacing the earlier social role of sago. Enbal has become a cultural food for contemporary Kei islanders and part of their identity. This is reflected in recent attempts to market it.


Ellen, R and H. L. Soselisa 2012. Cassava diversity and toxicity in relation to environmental degradation: a feature of food security in the Moluccas, Indonesia. In Environmetal uncertainty and local knowledge: Southeast Asia as a laboratory of global ecological change, eds. A.-K. Hornidge and C. Antweiler. Bielefeld: Transcript Verlag, pp. 215-242.

Ellen, R., H. L. Soselisa and A. P. Wulandari 2012. The biocultural history of Manihot esculenta in the Moluccan islands of eastern Indonesia: assessing the evidence for the movement and selection of cassava germplasm.  Journal of Ethnobiology 32(2): 157–184.

Ellen, RF and HL Soselisa 2012. A comparative study of the socio-ecological concomitants of cassava (Manihot esculenta) diversity, local knowledge and management in Eastern Indonesia. Ethnobotany Research and Applications 10, 15-35.

Soselisa, Hermien L. and Ellen, Roy (2013) The management of cassava toxicity and its changing sociocultural context in the Kei Islands, Eastern Indonesia. Ecology of Food and Nutrition, 52 (5). pp. 427-450. ISSN 0367-0244.