Of those, the watershed development program has been the most employed strategy since the 1990s and has focused on socio-economic development through modernizing agricultural production in India’s drylands.The annual investments in WDPs were approximately 4 billion USD between 2009 and 2012 , indicating the program’s magnitude.Currently, it is amalgamated into a larger program under the ministry of land resources with a similar scale of funding.WDPs encourage the promotion of intensive production systems , which are perceived as ideal for developing countries.WDPs, therefore, have been identified as key drivers triggering rapid transitions in farming systems.A recent study mapped the transitions in agriculture and farming systems in a region in Telangana.It also analyzed the effect of transitions in farming systems on smallholder livelihoods in the last 20 years.The study region was subject to various development programs, of which WDPs were the predominant one.That study, in line with other literature , indicate that farming systems before 1997 were primarily subsistence-oriented, with mixed crop-livestock production and where livestock had diverse functions.However, the subsistence farming system dwindled between 1997 and 2015, and specialized and market-oriented production systems emerged.The role of livestock became limited to the food production function.This was also accompanied by a significant change in land use, where croplands increased by 45% at the expense of wastelands that decreased by 75%.Increased regional production with land-use change led to groundwater scarcity in the region.In the end, the study also reported that some HH broke the socio-economic and cultural barriers to climb the “livestock ladder” to engage in, e.g., dairy farming.However, other HHs became marginalized and/or dropped out of the agricultural sector.

A plethora of articles have been published in the early 2000s on the agricultural and economic benefits of WDPs.Later articles indicated sub-optimal program outcomes due to various social, technical,4x8ft rolling benches and institutional issues.However, there is little information about the characteristics of the emergent farming systems and their economic and/or environmental performance.As indicated by Kuchimanchi et al., the fact that rapid transitions have occurred in the area, that more intensive forms of agriculture with altered crop-livestock interactions have led to higher input use, production costs, and investments ; or that development usually entail multiple trade-offs and undesired effects , calls for further research.Therefore, the aim of this study is to gain insight into the characteristics of emerging farming systems and their economic performance in a dryland region of Telangana, India, that has undergone rapid transitions in farming systems.This knowledge will help enhance the customization of WDPs and other development programs and ensure that their impact is sustainable.The two study watersheds are in the Rangareddy and Nagarkurnool districts of Telangana, India.For this research, we considered the administrative boundaries of the villages falling within the watershed, given that the secondary data are aligned with administrative boundaries.The first watershed covers four villages inhabited by 1820 HHs, and the second covers three villages inhabited by 1186 HHs.The HHs in the region are primarily agrarian , and 8.5% are engaged in non-agricultural activities due to higher education or acquiring non-farm skill sets.The predominant land category in the study region is cropland.The study region falls within the Deccan Plateau and Eastern Ghat agro-ecological sub-region 7.2.The area is characterized by deep loamy and clayey mixed red and black soils, with medium to very high available water capacity and a growing season duration of 120–150 days.The climate is characterized by hot, moist summers and mild, dry winters, with an aridity index of 0.2 ≤ AI <0.5.It is therefore classified as a semi-arid region.These districts are drought-prone, with an annual rainfall of 500–700 mm.We collected data in a stepwise approach to characterize the farming systems existing within the study region.The first step consisted of conducting a HH survey in both study watersheds in 2015.It involved face-to-face interviews with household heads using a structured questionnaire.

The data collected in the survey provided an overview of the population: types of livestock reared and herd sizes; farm sizes and categories of farming HHs ; and caste groups present in the region.The caste system in India is a social hierarchical classification of communities based on occupation, which has evolved since ancient times.Based on the government classification, we considered the four main groups: forward castes , backward castes , scheduled castes , and scheduled tribes.Of the 3006 HHs surveyed, 241 had no cropland or reared livestock, and they were excluded from the study.In the second step, we used the data from the household survey to classify the HHs according to the farming system.Our classification method was adapted from studies by Ser´e and Steinfeld , Kruska et al., Notenbaert et al., Robinson et al., and Alvarez et al..The classification was based on two variables: ownership of cropland and dominant livestock species reared.We identified the following farming systems: crop without livestock , crop with dairy , landless with livestock , crop with small ruminants , and crop with diverse livestock , as described further in the Results section.We conducted five focus group discussions, in the third step, one for each farming system.The participating HHs were randomly selected from the survey list.These discussions were intended to gather information on various qualitative characteristics of each farming system.Within each farming-system category, 30 HHs were randomly selected, and 1–2 members of each household were invited to participate.The gender composition of the focus groups was mixed, with participation varying from 25 to 30 people per group.Measures were taken to ensure proper representation from all farm-size categories and social groups mentioned above.If the representation of one of these categories was lacking, we substituted a randomly selected household from the over-represented group with one from the under-represented group.Each focus group discussion lasted 2–3 h and was conducted in the native language.To ensure that the objective of the discussions would be met, a detailed list of questions was used to guide the discussions.The key questions involved characteristics of current farming systems, including cropping and livestock-holding practices, farm infrastructure and use, off-farm jobs, access to fodder and water resources, livestock markets, and animal healthcare.The presence of different groups in each focus group discussion allowed to contrast potential divergent views between groups ‘in situ’.However, and to avoid domination by the wealthy, elderly, or socially forward groups, and to ensure that sufficient time was allocated for documenting information,each focus group discussion was moderated by an experienced facilitator.

All discussions were documented on charts to maintain transparency and enhance interaction with participants.We ultimately combined the quantitative data from the household survey with qualitative data from the five focus-group discussions to characterize the different farming systems.In the final step, we collected data on the economic performance of various farming systems in the study region.Although five farming systems were identified and characterized, we limited economic data collection to the three systems with consistent income from agriculture , based on the information derived from the focus-group discussions.The HHs to be surveyed were selected according to a two-stage sampling process.In the first stage, we selected the village in each watershed with the highest presence of all farming systems.From the selected village i.e.Thalakondapalle, 75 HHs were randomly selected from the complete household list.The selection for the survey was finalized only after HHs expressed their willingness to participate.Those declining to participate were replaced by new HHs until a sample of 25 HHs per farming system was reached.We compared the distribution of castes and farm size among selected HHs to ensure that they were representative of the total regional population.Data on the economic performance of the 75 HHs were monitored once every fortnight from August 2015 to August 2016 across all agricultural seasons in India: monsoon season , winter , and summer.Each household was provided with a data-collection booklet to record data, which data collectors cross-checked at regular intervals.Herd size was expressed in tropical livestock units.The conversion factors were cattle , buffalo , sheep/goats , and poultry.Labor was analyzed according to family type , assuming two working units for nuclear families and five working units for joint families.For the caste grouping prevalent in India, we considered the four main groups: FC, BC, SC, and ST,flood and drain table based on the Indian government classification.We estimated total revenue, costs of production, and gross margins at the household level as follows.The total revenue earned by a household was calculated based on the total quantity of different crop and livestock products sold multiplied by the market price, as obtained from the survey.The total costs of production were calculated based the total input costs for crop or livestock production, hired labor, and rented farm machinery costs, but excluding capital costs.The total GM was obtained by subtracting total costs of production from total revenue.All economic calculations were performed in Indian rupees and then converted to US dollars at an exchange rate of 71 INR.

The statistical analysis was to determine differences among farming systems and to gain insight into factors determining economic performance of those systems.The statistical analyses were performed using the statistical program GenStat , with a significance level of 0.05.To explore whether land use or herd size of HHs would differ across the farming system and/or between watersheds, we performed some preliminary statistical analysis.That analysis showed relevant differences across farming systems only, but not between watersheds.Hence, we performed an ANOVA with a post hoc Tukey test to identify differences in terms of land use and herd size across farming systems.To explore the economic performance of the three farming systems and gain insight into the factors contributing to it, we undertook a two step approach: ANOVA with post hoc Tukey test to identify differences in economic performance across farming systems, and a general linear model for each farming system to identify factors contributing to the GM.In the GLM model, the dependent variable was GM, with the independent variables of herd size, farm size, caste, and family type , along with all two-way interactions.Caste, labor, and their two-way interactions were not significant in any of the farming systems, and they were therefore removed from the model.Given the skewed distribution indicated by the GLM, we followed the approach described by Kuchimanchi et al.for the statistical analyses, converting the values into their natural logarithms.To ensure that values of 0 would also be transformed, we added one unit to all values.Once the tests were run, the mean values and confidence intervals were then back-transformed and one unit was subtracted from each value.This system accounted for the highest proportion of HHs.Most of these HHs were either marginal or small farmers , with medium farmers constituting 21% and large farmers constituting only 2%.The HHs in this system owned very few bore wells, which were seasonally functional.Cropping was thus predominantly rain-fed, with limited irrigated crop production.Low water availability limited crop farming in this system to one agricultural season per year.Mono-cropping of cash crops was predominant.Farmers reported that cropping practices were intensive and required higher investments, as they grew mainly cash crops, rented farm machinery, and used hybrid seeds and inorganic fertilizers.They added that cash crops were preferred to food crops, given their higher market value.These HHs mainly sold the produce, and the crop residues were either burnt or tilled back into the soil.Participants in the focus groups added that increased drought conditions over the years had led to failed crops, reduced yields, and increasing debts.These HHs owned no livestock due to diminishing common property resources for grazing, less family labor, or limited capacity to hire labor or invest in leasing lands for grazing or bore wells.The limited availability of water resources further inhibited them from taking up dairy production.A lack of livestock resulted in greater use of inorganic inputs in crop farming, as livestock manure was unaffordable.Many HHs therefore opted to take off-farm jobs in order to earn income, although they noted that such jobs were not adequately available.Most of the HHs also depended on the public food-distribution system to meet their food needs, as they cultivated mainly cash crops.The second most prevalent category was the CD system.Most of the HHs in this system were medium and small farmers , with marginal farmers constituting 15% and large farmers constituting 12%.The HHs in this system grew cash crops in combination with food crops and perennial green fodder.