Study area

The study was carried out in Paterasi Village Development Committee (VDC; the lowest administrative unit; called Paterasi rural municipality since 2017) in the Chaudabise valley of Jumla District (29.34°N, 82.50°E) in the mountains of midwestern Nepal (Figure 1). Altitudes range from 2700 to 6500 m and the area of 44,010 ha consists mainly of forest and rangelands (Larsen 2002; Jumla Local Development Office 2010). All villages are located in the valley bottom at around 2700 m. The area is relatively isolated, with a dry-season dirt road, constructed in 2012, and livelihoods are made up of the 3 traditional pillars of limited crop production, raising livestock, and engaging in limited trade. Maize, millet, wheat, barley, potatoes, and beans are the most common crops. In 2010, the total population of the VDC was 3791, distributed among 625 households, with 74% being high-caste Chhetri, 19% lower-caste Dalits (such as Kami and Sarki), and 7% others (Jumla Local Development Office 2010). All villages have similar access to commercial medicinal plants that are harvested in communal areas regulated through forest user groups or in government-owned and controlled habitats outside the VDC. Collection in the latter areas formally requires a permit, but this is not enforced.

FIGURE 1

Map of Jumla District and Paterasi VDC. (Map by Dipesh Pyakurel)

Data collection

Data were collected in March and April 2015 using a structured household questionnaire supplemented with interviews with key informants (5 local district officers, eg from the District Forest Office, to inform the content and structure of the household questionnaire) and a mix of participatory rural appraisal methods: a seasonal calendar to temporally map agriculture, livestock, and medicinal plant collection activities; a preference ranking to understand main livelihood activities (eg showing that medicinal plant collection is done for cash and considered risky, on par with agricultural crop growing); and a historical timeline, showing, for example, that O. sinensis harvesting began on a large scale in 1998 and that the first tractor drove to the area in 2012.

The questionnaire followed the Poverty Environment Network format and guidelines (PEN 2007; Angelsen et al 2011), and was developed to allow structured quantification of total household income, including environmental income. The questionnaire also recorded household characteristics, such as caste, household size, and education of the household head. The questionnaire was tested in a neighboring village outside the sampling frame.

Households were randomly sampled (n = 72) from the 3 main villages in Paterasi VDC, using forest user group membership lists as sampling frame (all households must be members in order to gain access to daily products such as firewood), with a proportional distribution reflecting the number of households in each village.

Interviews lasted 20–40 minutes and all were undertaken in Nepalese with the head of the household. They were conducted inside the house to minimize disturbances. Confidentiality was stressed and informed consent obtained. The structured survey included 3 reference years: 2006–2007, 2009–2010, and 2014–2015 (Nepalese years 2063, 2066, and 2071), selected to aid recall: 2006–2007 marked the end of the civil war, 2009–2010 was prominent as a suspension bridge was built in Paterasi VDC, while 2014–2015 was the ongoing year (where medicinal plant collection had been completed in the previous fall).

Household income and asset estimation

Annual total household income is the sum of all outcomes of household economic activities, thus including both subsistence and cash income (not subtracting the value of household's own labor). It is measured per adult equivalent unit (AEU) using the equivalence adjustments proposed by Cavendish (2002; adult [15–60 years] male given a value of 1.00, adult female 0.88, children [0–14 years] 0.67), enabling comparison of empirical results across households differing in composition and size (using the common scale of the number of adult male equivalents).

Total income is divided into 3 major components: farm, environmental, and nonfarm incomes. “Farm income” refers to subsistence and cash income from households’ own farming activities, including crop production and livestock products. External inputs were not used in agricultural production, as households used their own seeds and livestock manure as fertilizer, and net and gross incomes are assumed to be similar. Likewise, environmental income relies on household labor and involves negligible capital costs. Environmental income is subdivided into forest and medicinal plant incomes. The former includes incomes from collection of firewood and forest litter used for forage. Other forest income sources were marginal and are not included. Medicinal plant income includes cash income from wild-harvested products, excluding the value of medicinal plants for own-subsistence use. Nonfarm income includes household businesses (including shopkeeping and income from buying and selling medicinal plants), gifts, and loans. All registered values are farm-gate prices as recommended by Wunder et al (2011), except for firewood and forest litter, which were estimated using contingent valuation (willingness to pay) because of lack of markets.

Total asset value was estimated as the sum of savings, debt, and standardized livestock and land values. These were estimated using the average of the own-reported unit values (eg per cow, horse, chicken) obtained during the household survey. There was no trading of houses in Paterasi and these are not included in asset estimation. Furthermore, human and social capital is not included because human relations cannot be monetized (Cavendish 2002). All prices were converted to 2014–2015 prices using the Nepal Rastra Bank consumer price index.

Wealth groups

Households were divided into 3 wealth groups: poorest, poor, and better off. To include the dynamic aspects of poverty and to better predict a household's ability to adapt to income changes, the wealth groups were ranked based on total annual income/AEU and total asset value/AEU (Nielsen et al 2012). The “poorest” wealth group comprises households from the lowest tercile in income and assets (1 + 1) or medium tercile in either income or assets (2 + 1 or 1 + 2). The “poor” group includes households with terciles 1 + 3, 2 + 2, and 3 + 1, and the “better-off group” are households with scores 2 + 3, 3 + 2, and 3 + 3.

Data analysis

All local units were converted to the International System of Units and analyzed using the SPSS statistics package. Frequency and percentage histograms were used to understand how data were clustered or grouped. Differences between wealth groups were tested using t-tests, and analysis of variance (ANOVA) was used to compare the means of the 3 observation points for key variables such as the number of medicinal plant collection days per household (post hoc corrections were not applied).

We applied descriptive statistics and multiple ordinary least squares (OLS) linear regression models to determine which variables influenced total annual income and medicinal plant income. Predictive variables were household characteristics (age, sex, and education of household head; size; caste; location), assets (cropland, pastureland, livestock units), and medicinal plant collection (number of days, years of collection experience). The variance inflation factor for each of the explanatory variables was low (ranging from 0.2 to 1.7), indicating that multicollinearity is not problematic.

Foster–Greer–Thorbecke decomposable poverty measures (Foster et al 1984) were calculated to estimate the prevalence (the proportion of households with per AEU income below the poverty line), depth (the average income shortfall as a proportion of the poverty line), and severity (the variation in income distribution among households below the poverty line) of poverty.

The national poverty line was defined in 2010–2011 as NPR 19,261/person/y (NPR 102 = US$1; DFID 2013). Adjusted for inflation, the 2014–2015 national poverty line was estimated as NPR 25,039/person/y, equivalent to US$ Purchasing Power Parity 1.63/person/d.

Data checking

At the end of each working day in the field, data were checked for errors and inconsistencies. This allowed revisits to households the following day to clarify issues. Mean, mode, and median values of households' own-reported value estimates, as well as the mean standard deviation, were compared (Wunder et al 2011) and showed the pattern reported by Rayamajhi and Olsen (2008)—similar mean, mode, and median values showing little skewness, standard deviations lower than the means—thus indicating that households provided valid and reliable estimations of product values. Medicinal plant values for 2014–2015 were also checked against the central wholesaler purchasing prices published online by the Asia Network for Sustainable Agriculture and Bioresources and consistently found to be lower, as expected; prices obtained by collectors were lower than those offered by central wholesalers further up the value chain (eg Olsen and Helles 1997, 2009).

Household characteristics

The households were in many ways similar (Table 1), with a few notable exceptions: (1) The better-off households were significantly smaller (as measured in AEU), with fewer children and fewer adults (the latter significant only in relationship to the poorest households). (2) The poorest households had significantly lower landholdings and were more food insecure compared to the better-off poor households. The wealthier households also tended to own more businesses, have higher savings, and be less food insufficient. Education levels were low across the board, with high illiteracy: 65% of respondents never went to school. Wage work, recorded in 28 households (40%), was mainly as masons, while 10 households (14%) had their own business, a small shop or trade in medicinal plants.

TABLE 1

Characteristics of households across 3 wealth groups, Jumla District, 2014–2015.^a)

Household incomes

Again, the households were similar in many ways (Table 2). In relative terms, the distribution of income was similar across the 3 terciles, with environmental income being the dominant source (average of 65% of annual total household income, ranging from 62–67%) of income, followed by nonfarm income (19%, 18–21%) and farm income (16%, 15–17%). The largest variation was in wage income, which was relatively less important for the better-off households (8% versus 15% for the other groups), who instead generated income from their own businesses (13% versus 2% and 4%). In absolute terms, the total annual household income per AEU of the better-off households was significantly higher, and more than double the income of the poorest, including significantly higher incomes from environmental resources, agricultural products, and own businesses. There was also a remarkably similar distribution of productive assets, with the largest difference being significantly higher total asset value in the better-off group. Around 40% of the households in the sample fell below the national poverty line, with a poverty depth of 13% and 6% of households living in severe poverty.

TABLE 2

Nested total absolute (Abs) (NPR/AEU) and relative (Rel) (%) annual household income and productive assets by wealth groups, Jumla District, 2014–2015.^a)

The single most important and dominant source of income was from wild harvest and sale of medicinal plants. There were no significant differences between the wealth groups, with an average share in annual total household income of 58% (ie of the sum of household cash and subsistence income), ranging from 55–60% across the groups. In absolute terms, however, medicinal plant income for the poorest households was significantly lower than for the other households, with the better-off households earning an average of almost 3 times more. Medicinal plant income made up an average of 78% of all cash income, with 46% of households having medicinal plant sales as their sole source of cash.

Products harvested and traded were bulbs of Fritillaria cirrhosa D. Don, morels Morchella spp, rhizomes of Nardostachys jatamansi (D. Don) DC. and Neopicrorhiza scrophulariiflora (Pennell) D.Y. Hong, roots of Rheum australe D. Don, tubers of Dactylorhiza hatagirea (D. Don) Soó, rhizomes of Paris polyphylla Sm., pseudobulbs of Eulophia spp., and O. sinensis (Berk.) G.H. Sung, J.M. Sung, Hywel-Jones & Spatafora (the high unit price for the latter meant that 29% of households collected only this product).

All species were harvested, air-dried, and then sold to local traders who feed production networks leading to India and China, with local and domestic consumption being minimal. There was no local knowledge of end uses besides the anecdotal, for example that O. sinensis is used as an aphrodisiac in China.

Determinants of medicinal plant income

Because of the low variation in the sampled households, as noted above, the OLS regressions (Table 3) did not produce many significant results. Only the number of medicinal plant collection days significantly and positively influenced the total annual medicinal plant income. There was no significant difference in the number of medicinal plant collection days between any wealth groups. This variable was also significant and positive for “total annual household income,” as was “collection experience” (expressed as number of medicinal plant collection years). The number of household members had a negative effect, probably because of more unproductive members: while our data do not allow for a fine-grained analysis, the poorest households had significantly more children, which could prevent the allocation of adult labor to medicinal plant collection through the need to provide childcare. Likewise, dominant caste membership had a negative and significant influence on total annual household income (see Discussion).

TABLE 3

Household (n = 71) socioeconomic determinants of total annual household and medicinal plant incomes, Jumla District, 2014–2015.^a)

Changes in medicinal plant collection and income from 2006–2007 to 2014–2015

The number of medicinal plant-collecting households increased from 78% in 2006–2007 to 100% in 2014–2015. Men were the main collectors, but women's participation increased by 27% in the period. Plants were usually collected during 2 to 3 dedicated trips per year.

In the observation period, the average number of collection days per household increased significantly, from 25 days in 2006–2007 to 62 in 2014–2015 (Table 4), as did income, from NPR 11,000/AEU to NPR 21,000/AEU. The income per collection day, however, decreased significantly, from NPR 577/d to NPR 375/d. The total income in the studied population increased from NPR 800,000 in 2006–2007 to NPR 1.5 million in 2014–2015.

TABLE 4

Medicinal plant collection: collection days and income in 3 reference years, Jumla District.

Some households had specialized in medicinal plant collection to the extent that they had abandoned other livelihood activities. In particular, O. sinensis collectors had left crop production; as the collection season overlaps with the agricultural season, cultivation of rice had completely stopped, and the cultivation of barley and wheat had decreased. Respondents reported that everybody had enough money to purchase food, limiting the need to cultivate it. In the words of a respondent: “Nowadays we don't wait for food, we just buy it.” Decreasing crop production had reduced agricultural labor demand, allowing more children to go to school. Two new private schools opened in 2014–2015, indicating that parents could afford school fees and wanted to invest in their children's future.

Use of medicinal plant income

Most households (54%) indicated that they were spending medicinal plant income on domestic consumption, including food and other household necessities. Around 19% spent money on new house construction, while 15% invested in livestock, 14% bought new land, 13% invested in shops, and 6% invested in their children's education.

Medicinal plant income and poverty measures

In 2014–2015, 40% of households fell below the national poverty line, with a poverty depth of 13% and 6% living in severe poverty. If medicinal plant income is excluded, these figures increase to 83%, 29%, and 16%. In the same year, the Gini coefficient decreased from 0.51 to 0.39 when medicinal plant income was included, indicating that medicinal plant income had a strong positive effect on income equality. The lack of total household income data for earlier years does not allow a comparison of Gini coefficients across time.

The economic importance of medicinal plants to total household income

The average share of income from commercial medicinal plants in total annual household income in 2014–2015 was 58%, ranging in a narrow band from 55–60% across the terciles. This is outside the previously reported range of 3–53% for all environmental products in studies that report total household incomes (Rayamajhi et al 2012; Meilby et al 2014; Chhetri et al 2015; Charlery et al 2016; Oli et al 2016). This is likely to reflect our in-depth focus on the medicinal plant portfolio, the rising prices for many products, and the tendency for households to allocate more labor to medicinal plant collection over time. The average share of commercial medicinal plants in total annual household cash income in 2014–2015 was 78%; this is within the 53–92% contribution of O. sinensis to total household cash income reported in recent studies from Nepal (Childs and Choedup 2014; Shrestha and Bawa 2014; Laha et al 2018; Pouliot et al 2018).

In absolute terms, medicinal plant income was significantly lower in the poorest tercile than in the more well-off terciles. This also translated into a marginally lower relative share for the poorest households. This differs from the previously recorded and consistent pattern of higher relative economic importance in the poorest households (Olsen and Larsen 2003; Rijal et al 2011; Shrestha and Bawa 2014; Shrestha et al 2019). This new pattern could reflect local responses to market dynamics as households are pulled into medicinal plant collection through rising prices (Winkler 2008). This could be a new general pattern in the high mountains throughout Nepal, where medicinal plants are widely collected (Olsen 2005; Olsen and Helles 2009; Pyakurel et al 2018).

In our low-income study area, all households invested in medicinal plant collection, and the success of more well-off households, in terms of higher absolute income, may reflect a lower dependency ratio and hence more available labor: we know they had significantly fewer children. Larger families may not have more collectors, because of an increase in demand for childcare, but the size of the family lowers the values per AEU. To further understand this, we need data on adult labor availability (eg the number of elderly across wealth groups). Another possible explanation could be improved medicinal plant access for more well-off households, though this was not mentioned in interviews.

The studied population had a very low level of education. This could explain the absence of remittances: only 5.6% of households had long-term outmigrating members who were otherwise important in similar areas (Rayamajhi et al 2012; Chhetri et al 2015). Mobility was instead characterized by movement out of the area before the snow arrived, to trade in the lowlands, before returning in spring (Shrestha-Shipper 2010). Lack of human capital could push households toward higher environmental-product reliance, while the high medicinal plant income could also serve to limit the outmigration that has been documented in other high-altitude areas (Childs et al 2014). Respondents mentioned that migrants had started to return during the medicinal plant collection season to get a share of the valuable harvest. Local expenditure patterns for in-migrants are not known, including how much medicinal plant income is exported out of the area.

Other studies have found that medicinal plant collection is mainly undertaken by low-caste groups (Larsen 2002; Kunwar et al 2008) and, more generally, that caste membership influences livelihood strategies (Nielsen et al 2013). This pattern was not found in our study, reflecting the limited caste variation and dominance of the Chhetri caste (high caste), which made up 82% of the sample, including more than 90% of the poorest tercile. Paudel (2007) also did not find a correlation between caste and medicinal plant income in Jumla District. This indicates that lower castes are not prevented from accessing medicinal plant resources, whether in communal areas in the VDC or in government-owned habitats outside the VDC that are de facto considered open access. We did find, however, a significant negative relationship between dominant caste membership and total annual household income, indicating that there is not a simple relationship between castes and wealth groups. Asset levels were constant across the wealth groups (eg in terms of livestock ownership and education) with the only significant difference being that better-off households have significantly more land than the poorest households. Hence, it may be that this 2014–2015 finding reflects households in transitory poverty (rather than chronic poverty; Nielsen et al 2012), pushed there temporarily by unrecorded income shocks (Møller et al 2019) but with assets that could facilitate upward income mobility.

Incomes increased from 2006–2007 to 2014–2015, and 97% of respondents gave increasing medicinal plant income—almost doubling per household in the period—as the main explanation. According to 95% of respondents, the driver was higher unit prices, especially for O. sinensis, which increased by 2300% from 2001–2011 (Shrestha and Bawa 2014). The fungal commodification of the rural economy in Tibet (Winkler 2008) is also occurring in the Himalayan range: O. sinensis trade accounted for 60% of medicinal plant income in 2014–2015. An increase in price for other high-altitude species has also been noted elsewhere (Pyakurel et al 2018). The average income per collection day significantly decreased in the observation period, while the number of collection days and total medicinal plant income per collector significantly increased.

Interviews showed a widespread belief that medicinal plants were becoming scarcer, reflecting the lower income per day. This could, however, be due to the increasing number of collectors and collection days rather than a supply decrease. We did not find evidence to support the assumption by Hopping et al (2018) that collectors can distinguish between shortages deriving from total shortage versus a higher number of collectors. In addition, while collection income per day decreased, community-level income almost doubled in the observation period.

While many authors have argued that commercial harvesting is negatively impacting O. sinensis productivity (eg Winkler 2009; Shrestha and Bawa 2013; Shrestha et al 2014; Negi et al 2015; Hopping et al 2018; Laha et al 2018), for example through premature harvesting and overharvesting, no studies have documented that harvesting is unsustainable. Such studies are complicated by the complex life cycle of the caterpillar fungus: are critical production limits related to the dispersal of fungal spores or larvae production rates, for example? The determinants of harvesting sustainability are not known (Pouliot et al 2018).

The role of medicinal plants in providing a pathway out of poverty

The results suggest that medicinal plant income reduces income measures of poverty while also contributing to increased income equality. This is in line with existing findings for both medicinal plants (Shrestha et al 2019) and environmental income (Rayamajhi et al 2012; Chhetri et al 2015; Walelign et al 2016). The increased medicinal plant income has reinforced the trade component of high-altitude Himalayan livelihood strategies, providing households with hitherto unseen amounts of cash and thus new opportunities to diversify or specialize livelihood strategies.

Investments made possible by medicinal plant income have led to the emergence of a new cash-dependent economy, including gap-filling nonfarm jobs in the winter season, reducing the need for circulation to the lowlands. Medicinal plant collection in the high Himalayas in Nepal, rather than being an employment of last resort or a poverty trap (Angelsen et al 2014), appears to be an environmental-income-led pathway out of poverty. This is evidenced by the reliance on commercial medicinal plant collection in total household income and for cash, as well as the increasing importance of this income source over time, as seen in the increased labor allocated into collection (eg in number of collection days or in decisions to stop rice cultivation). However, as commercial medicinal harvesting may be unsustainable for many species, even if this is not yet documented, collecting households are vulnerable to decreasing supplies, indicating an urgent need to invest in determining sustainable harvest rates and the development of sustainable management practices.