Effects of socio-economic household characteristics on traditional knowledge and usage of wild yams and medicinal plants in the Mahafaly region of south-western Madagascar

The study area is situated in the northern part of the Mahafaly region. The studied villages are located on the adjacent coast (littoral) and on the west side (plateau) of the Tsimanampetsotsa National Park (24°03′-24°12′S, 43°46′-43°50′E; Figure 1). The area is characterized by a dry and spiny forest vegetation with the highest level of endemism in plant species registered in Madagascar (48% of genera and 95% of species; [17]). The natural vegetation consists of a deciduous forest characterized by drought tolerant woody species of Didieraceae and Euphorbiaceae, xerophytic bushland and savannah. In the littoral zone dry forests on sandy soil dominate while on the plateau dry and spiny forests on tertiary limestone or ferruginous soil occur [18]. The semi-arid climate is characterized by an annual mean temperature of 24°C and a highly variable annual rainfall ranging between 300–350 mm in the littoral and 400-450 mm on the plateau [19]. The dry season lasts nine to ten months and the rainy season five months from November to April. The unreliability and unpredictability of rainfall is one of the major factors limiting agricultural production by the predominantly small holder farmers and herders, which partly rely on forest products to fulfil their daily needs throughout the year. During the past 40 years forest cover declined by 45% due to slash and burn agriculture and uncontrolled bushfires [20, 21]. In addition, the region has the lowest education rate of Madagascar and the majority of the households were classified as poor [22] in combination with a lack of basic health services and infrastructure. Altogether, 41% of the local population on the Mahafaly region is affected by food insecurity and famine [23]. Rapid population growth and the recent expansion of the Tsimanampetsotsa National Park (from 42,200 to 203,000 ha in 2007) have increased the pressure on the forests resources in and outside the park area [21, 24, 25]. Combined with the effects of climate change this leads to an increasing over-use of the natural resources in the Mahafaly region.

In the Mahafaly region wild yams are used to supplement cassava (Manihot esculanta Krantz) and maize (Zea mays L.), especially during hunger periods (‘Kere’). Local reports indicate that during the past years the amount of harvested wild yam tubers has strongly increased given a rising insufficiency of crop production.

The field work was conducted from June to December 2012 in five villages that were part of a larger village and household survey [21, 26]: (1) Efoetse in the littoral (S 24°4′42,41″- E 43°41′54,78″), (2) Ampotake (S 23°52′27,78″- E 43°58′36,55″), (3) Andremba (S 23°58′17,60″- E 44°12′17,05″), (4) Itomboina (S 23°51′59,15″- E 44°5′10,9″) and (5) Miarintsoa (S 23°50′14,21″- E 44°6′17,68″) on the plateau. Village selection was based on (1) market accessibility, (2) distance to the national park, (3) intensity of forest product collection of village inhabitants and (4) diversity of household activities. For each village, 50 households (HH) were randomly selected based on a complete household list (total N = 250). Pre-testing interviews and field observations were performed with key informants selected by snowball sampling [27]. Semi-structured interviews [28] were conducted with the household head after we received his consent. The Code of Ethics of the International Society of Ethnobiology was followed. If household head disagreed to take part in an interview, an alternative household was chosen based on an existing household list of the village. The questionnaire was divided in three thematic sections: (1) Information on socio-cultural and economic characteristic (family size, source of income, agricultural harvest, origin of the head and spouse, land area available for cultivation, livestock owned, harvest satisfaction, education level, ethnic group, religion, gender affiliation and age of respondents); (2) Household consumption, collection and use of wild yam species; (3) Medicinal plants and the knowledge about their uses. Respondents were also asked about specific plant parts used and the habitat from which they collected the plant material. All interviews were supplemented with field observations and forest walks. Since informants were only able to mention the local species name, plant specimen were collected in the field to establish a digital herbarium of inventoried specimens for botanical identification [29] in the Herbarium of the Botanical and Zoological Parc of Tsimbazaza (PBZT) in Antananarivo (Madagascar), following the nomenclature of the Tropicos database of the Missouri Botanical Gardens [30].

In the absence of any formal ethics committee the concept, content and questions related to this study conducted within the participatory SuLaMa (sustainable Land Management in South-Western Madagascar) project (http://​www.​sulama.​de) were discussed and approved at the governmental and the village level in several meetings as were the outcomes of the interviews.

The consumption, collection intensity and usage of wild yams were analysed using the following interview data: number of species collected, frequency of collection per month, period of collection per year, average number of tubers collected per collection event (estimated by the number of harvest holes), number of collectors per households, type of consumption (staple or additional food) and sale of tubers. The types of medicinal usage were categorized in different medicinal categories according to Cook [31]. To estimate the informant knowledge on the use of medicinal plants, the diversity of medicinal plant uses [32, 33] was calculated for each informant. The species (UV_S) and the family use values (FUV) were computed (Table 1) [34, 35] to compare the importance of plant species and families.

All statistical analyses were carried out using SPSS 17.0. A two-step cluster analysis was used to identify household groups based on socio-economic characteristics and plant use patterns. The existence of collinearity was tested based on correlation coefficients and suspicious data was removed from the dataset resulting in the following parameters used for cluster analysis: Education level, agricultural harvest, household activities, family size, tropical livestock units, agricultural area, medicinal plants used, number of medicinal uses and diversity of medicinal plant use (D), wild yam species collected, amount of tubers harvested (number of holes harvested for each collection), frequency of collection, sale, collection period and use of wild yams.

To evaluate the contribution of each variable in separating the resulting households groups, a Discriminant Analysis (DA) was conducted using the standardized canonical coefficients, canonical correlation coefficients, Eigen value and Wilk’s Lambda. A structure coefficient matrix was established which allowed to assess the importance of each variable in relation to the discriminant function.

A One Way ANOVA (Analysis of variance) was performed to compare the differences of knowledge and use between communities in relation to their location (villages). Additionally, we used Jaccard’s similarity index, which was based on species usage data to determine the similarity of species usage among villages [36].

To determine which cultural and socio-economic variables influence the use intensity and knowledge on medicinal plants and wild yams (response variables), we used a Generalized Linear Model (GLM) based on a Poisson distribution. The GLM consisted of two models with eight response variables, which explain the relationship between predictors and the knowledge on medicinal plants (number of medicinal plants used) and the use of wild yams (frequency of yam collection per month). The performance and the fit of the models were assessed using the Akaike Information Criterion (AIC; [37]). In each model, we only included main effects and choose the Type III analyses and Wald chi-square as statistical tests. The 0.05 significance level was used to assess if an independent variable related significantly to a dependent variable.

Average household size varied between 6.3 persons in Itomboina and 7.2 persons in Miarintsoa (Table 2) whereby big households typically comprised a polygamous household head. Thus, each sub-family might live separately, but all family members eat together and share the same income. The education level of the households was highly variable across the villages, but in general, 30% of interviewed households did not receive formal education and only half visited at least the first year of primary school. The village with the highest rate of illiteracy, Ampotake, had no school. However, in Efoetse, where public and even private schools are available, literacy was high. The majority of the households comprise small holder farmers, which conduct different off-farm activities for cash income generation, such as salaried work, artisanal activities, trading, fishing, charcoal production or the collection of wood and other forest resources. The average household’s agricultural area was 2.2 ha of which some was partly left uncultivated due to heavy weed encroachment or a perceived decline in soil productivity. For the majority of households, periods of food insecurity due to unpredictable and insufficient rainfall are frequent and people heavily depend on supplementary off-farm income. Most of the household heads were born in the village where they live, only 26% are immigrants. The majority of households (60%) has traditional religious beliefs (ancestor reverence) and conduct ritual practices, while 30% are Christian (Catholic, Protestant or Anglicans).

Based on their socio-economic characteristics and the use intensity of forest products, the cluster analysis revealed two groups of households (Table 5). The well-off farmers represent households with a high number of livestock, off-farm activities and a higher education level. They use yam as a supplementary food, practice a more sustainable harvest technique and collect less wild yam tubers compared with the poorer farmers. The latter are characterized by lower household assets and off-farm activities. Farmers of this group collect more yam species and use their tubers as staple food.

Most of the socio-economic variables used for the cluster analysis were effective in discriminating the two defined household groups except for the education level and the diversity of medicinal plant use. Together the predictors accounted for 51% of the between-group variability. Based on the conclusions of Rach et al. that structure coefficients ≥ 0.30 indicate a strong discriminating power [39], households cluster groups were determined by the amount of agricultural harvest, livestock owned by household, and the frequency of wild yams collection. In contrast, the number of medicinal plants used and the use intensity of medicinal plants differed only slightly among the two groups.

Collection and use of forest plants differed between the littoral (Efoetse) and the plateau (the other three villages) which may be mainly explained by the lack of forest resources and wild yams in the coastal area. The number of medicinal plants and wild yam species used were higher on the plateau (Ampotake, Andremba, Itomboina, Miarintsoa), and the number of species collected was highest in Itomboina and Miarintsoa (Table 6). However, the collection frequency, period, and the amount of harvested wild yam were higher in Ampotake. This may be mainly due to the proximity of community based forests, where collection of forest products is not restricted. Itomboina and Miarintsoa are situated in the middle of the plateau, where different soil types (ferralitic, red sandy and calcareous soils) and forest habitats prevail, which may explain the high diversity in species collection by the informants. Knowledge, traditional uses and the number of species used differ significantly (P < 0.01) among villages. Overall, the knowledge and the uses of plants are higher in Ampotake than in the other villages. In Ampotake, Miarintsoa and Itomboina, similar medicinal plant species are used as indicated by the Jaccard similarity indices ranging between 0.68-0.7 (Table 7).

The number of livestock owned (TLU), education level, family size and agricultural harvest were significant predictors for the number of medicinal plants used and the frequency of yam collection. The TLU and the age of respondents significantly affected the collection of wild yams (P < 0.001; Table 8). In the study region, a high number of livestock owned is a sign of wealth. Households with a low TLU are characterized by higher yam collection intensities. For the number of medicinal plants used, the only significant predictor variables were family size and healer consultancy. The latter indicates how often a household asked a traditional healer for advice on appropriate medicinal plants. The higher the diversity of different household activities (salaried work, trading, artisanal), the more cash income is produced. Consequently, the households have the possibility to buy food during difficult seasons, and depend less on wild food collection. In addition, female respondents use and know more plants than men. Age did not affect the use and knowledge on medicinal plants, which is maybe due to the direct knowledge transfer within one household. In this study, 79% of the households did not report to consult a traditional healer in case of illness.

The basic characteristics of the interviewed households correspond to the results of INSTAT [22] for SW Madagascar even though our survey indicated a higher education level. In Ampotake, the majority of the households heads (52%) are illiterate, which reflects the percentage of the non-educated people in the rural area in this region. The average land size per household (2.2 ha) corresponds to the respective value in Mozambique [40]. In this study, we used off-farm activities to determine the different cash income sources and diversification level of households based on the assumption that higher diversification leads to higher income [41, 42].

Among the six species of wild yam recorded, only D. alatipes and D. bemandry were frequently harvested by local people to substitute for staple food. This is comparable to the collection of wild yam species in the dry forest of NW-Madagascar [43]. Mavengahama et al. [44] recorded a similar importance of wild yam collection for rural livelihoods in South Africa, where wild vegetable are of high importance in supplementing staple food diets based on maize, sorghum (Sorghum bicolor Moench.), and millet (Pennisetum glaucum L.).

In our study, the collection intensity of wild yams depended not only on the availability of the species, but also on the taste of the yam tubers. For Malagasy yams, the preference in taste was analysed by Jeannoda et al. [14] who observed a significant correlation (P < 0.001) between the preference and the sensitivity to saccharose. Polycarp et al. [45] stated that the high level of carbohydrate and energy with appreciable levels of minerals makes yam a very nutritious source of food. Bhandari et al. [46] found that the nutritional composition of selected wild yams in Nepal was similar to those reported for cultivated species of yam. When analyzing the nutritional value of Malagasy yam germplasm, including those of wild species, Jeannoda et al. [14] determined high contents of calcium in Dioscorea ovinala, which makes some wild yams physiologically important.

However, a decline in the availability of wild yams was already reported by the respondents of our study who are forced to increase the search radius for tuber harvests. One main reason for the decline in this essential resource securing local livelihood strategies against drought related hunger risks may be the exploitative harvesting methods used by the majority of the collectors in the Mahafaly region, which hampers the regeneration of the species. In contrast, Ackermann [43], who conducted a study in the NW-Madagascar reported that traditional people try to harvest the tubers carefully to guarantee the survival of the plant stand. In our study only 15% of the household took care of the regeneration of the lianas. While the sale of wild yam tubers provides valuable cash income for many households it may also be one of the causes for its overexploitation and increasingly threatened existence [47]. About 20% of the harvested tubers per households are sold on local markets.

The majority of the medicinal plants used by the local people belong to the Fabaceae, Apocynaceae and Euphorbiaceae. In contrast to yams, none of the interviewed households was selling medicinal plants. Local people complained that some species are nowadays hard to find, which was confirmed by our field observation. Hamilton [48] stated that globally 4,160 to 10,000 medicinal plants are endangered by habitat losses or overexploitation in areas where rural families traditionally collected them. The present study shows that the most popular plants with high use values, such as Aloe divaricata, Erythroxylum retusum, Cedrelopsis grevei, Neobeguea mahafaliensis, Salvadora angustifolia and Croton geayi are native species collected from forest habitats. This shows that the wild habitats are important for local communities in terms of basic needs. Beltrán-Rodríguez et al. [49] also pointed to the importance of wild habitats for peoples’ livelihood in a rural community of Mexico and found a greater diversity of plant uses in wild habitats than in managed environments.

Some plants are less frequently used, which does not decrease their importance since most of them are needed for very specific therapeutic purposes. The increasing scarcity of such plants may also enhance the loss of traditional knowledge about the medicinal uses [50, 51]. On the other hand there are cultivated species such as Tamarindus indica and Sclerocarya birrea subsp. caffra, Citrullus lanatus and Ziziphus spina-christi, which are nowadays used more intensively for medicinal purposes.

Different parts of the same plant are used for different purposes or by different population groups. Sometimes, a specific plant part is used for children and another part of the same plant for adults to treat a disease such as in the case of Aloe divaricate. The use of plant roots as traditional remedies is often problematic as it prevents plant regeneration [52]. Muthu et al. [53] reported that the choice of plant species most used by people depended largely on the type of diseases treated. In our study, digestive disorders, post-delivery care, body injuries and wounds were the most frequently mentioned diseases. This is comparable to similar studies conducted in Africa [54, 55] China [56] and in Colombia [57], where digestive disorders were most frequently treated by medicinal plants. Compared to other developing countries, where sexually transmitted infections are most commonly treated with herbal medicines [58] this category was rarely cited in our study. Except for venereal diseases which are treated using a combination of different species [59, 60] the majority of plant species utilized had a single therapeutic use.

Some of the recorded medicinal plants in Madagascar are already pharmaceutically analysed and the active ingredients confirm traditional therapeutic uses. For example, Koehneria madagascariensis has a large and strong antimicrobial activity [61]. Hernandia voyronii[62] is known for its antimalarial active substances, Neobeguea mahafaliensis and Cedrelopsis greveii for effectiveness against cardiovascular diseases [63]. Although the World Health Organization (WHO) reported that 60-70% of Madagascar inhabitants have ready access to primary health care [64], accessibility of effective modern medicines is still a challenge for the local population in the Mahafaly region and they thus make use of native plants for alternative treatment.

Our study revealed that the collected quantities and qualities of plants vary greatly between households. Very poor and poor farmers consume and sale more yams and have higher knowledge on traditional usages of medicinal plants than well-off or “rich” individuals. Households with lacking off-farm income collect and consume more frequently wild yams than households with regular off-farm income. In addition, the regression results revealed, that households with more cropland and higher crop harvest collect less forest products. This was also confirmed by Reddy and Chakravarty [65] in India. Variables showing the collection and consumption of wild yams (P < 0.01) were important discriminators for household groups in contrast to the variables on the use of medicinal plants (P < 0.05).

The use of forest products was significantly higher in villages near forests, where wild yams and medicinal plants are more readily available. This confirms findings of Banana and Turiho-Habwe [66] in Uganda and Kerapeletswe and Lovett [67] in Botswana, where the dependency on the forests for food supply decreased rapidly with an increasing distance of the respondent’s home from the forests. Furthermore, poor market access may increase the importance of forest products to sustain people’s livelihood [68].

The number of livestock owned by the household, education level, agricultural harvest and family size affected the collection of wild yams and the usage of medicinal plants. Livestock and off farm activities determine the wealth condition of the household in this region and were negatively correlated with the use of wild yams and medicinal plants. However, we cannot generalize these findings as with time and location the direction of the relationship may change [69]. Socio-cultural factors are of higher importance for the use of medicinal plants than for the collection of wild yams. In contrast to other findings [49] female respondents use more plant species than males. The use of medicinal plants is the basic health care for the majority of the households and the knowledge about their use was maybe shared over generations, which might explain, that there is no significant influence of informant age on the collection intensity of medicinal plants. In the study of Kirstin [70] on the usage of Budongo’s forest products, the use of wild food such as Dioscorea spp. increased with age, whereas young village people focused on the use of fruits and wild game because of their higher income potential. This might also be true for our study region, were younger farmers predominate in collecting wild yams for sale.

Overall, this study indicates that a household’s wealth status affects the traditional knowledge and use intensity of forest products, which confirms previous studies [49, 71, 72]. The World Resources Institute [4] reported that families facing poverty, sickness, drought, wars and economic crisis depend to a higher degree on the collection of wild resources. Although, our study focused only on medicinal plants and wild yams as forest products, the rate of change in social and economic attributes of rural households is likely proportional to the rate of change in resource use [73]. Therefore, whatsoever the products extracted, a household’s socio-economic dynamics ultimately drives its ability to use village forest resources.