3 Use of GIS and surveys to identify the recommendation domain

Introduction

The recommendation domain (RD) is the geographic area targeted for the output of the research: improved dual-purpose sorghum and millet varieties. The objective in defining the RD is to determine where this new technology is most likely to be desired and adopted by farmers. Two approaches were taken to define it: 1) entering secondary data (source: Indian Council of Agricultural Research (ICAR)/ICRISAT 13-state district database) within a GIS to map out and overlay particular spatial areas of interest; and 2) collecting primary farm-level data in some of the districts identified through the GIS analysis to verify, or ‘ground-truth’ the importance and uses of dual-purpose varieties within these districts.

Use of secondary data and GIS to identify a recommendation domain

The objective of the GIS analysis of secondary data was to identify the districts in India where:

• pearl millet and sorghum are important crops, covering a reasonable proportion of total cropped area
• pearl millet and sorghum residues contribute a significant proportion of the estimated total roughage available to livestock in the district.

Determining where the sorghum and millet are being grown was relatively easy.  However, to assess the importance of sorghum and millet stover in the diets of cattle and buffalo, availability of different feeds (crop residues, grasses, concentrates, agro-industrial by-products etc) had to be estimated by district. The availability of each type of feed was estimated using relevant extraction ratios developed by Singh et al (1997).

Since sorghum and crop residues ‘compete’ with other sources of roughage in an animal’s diet, the estimated production of crop residues and green fodder (i.e. fodder from communal grazing and forests, converted to a dry weight equivalent) were added up to generate estimates of total roughage available for livestock feed in each district. The share of sorghum and millet residues of total available roughage was then calculated.

One hundred and four districts were identified where:

• more than 10% of cropped area was sown to millet and/or sorghum
• the share of total available roughage made up of sorghum and/or millet residues was greater than 10%.

These 104 districts make up the overall RD for the new dual-purpose varieties (Figure 3.1).

Characterisation and division of the recommendation domain

The 104 districts of the RD targeted for the improved dual-purpose varieties being developed account for 33% of India’s area and support 24% of the human population, some 220 million people. Twenty-five per cent of the bovine and 35% of the country’s small ruminant population are located within the RD (Table 3.1). Twenty-nine per cent of India’s annual milk production, equalling 16 million tonnes, comes from these 104 districts. If we assume that the amount of beef produced is proportional to the number of animals (in the absence of this information at the district level), 25% of India’s total beef production of 1.06 million tonnes, or 265,200 t of beef, are produced within the RD.

Figure 3.1. Recommendation domain for improved dual-purpose millet and sorghum varieties.

The RD falls largely in areas with an average annual rainfall of less than 1000 mm; only 15 districts out of the 104 received more than 1000 mm. The proportion of irrigated cropped area in 76 RD districts is lower than the Indian average of 35%. In general, in much of the RD, crop production is uncertain, yields are relatively low, and livestock tend to be more important compared to higher rainfall areas and areas with more extensive irrigation.

It was estimated that 130 million tonnes of total roughage are available in the RD.  Sorghum and pearl millet residues together contributed 50% of the total roughage produced, while other crop residues accounted for 31%. Green fodder made up the remainder (19%), although in the north of the RD, the contribution of green fodder is generally higher than in the south. The estimated annual availability of oilseed cakes and concentrates to ruminant livestock, net of exports and poultry feed, was 1.75 million tonnes.

The number of livestock in terms of total tropical livestock units (TLU) found in the RD is approximately 77.8 million. This translates into an average availability of 1.67 t/year per TLU of roughage and 0.02 t/year per TLU of concentrates + oilseed cakes (excluding the forage obtained through grazing on fallow and other land that is not farmed).

Since millet is more important in the north, and sorghum in the centre and south, the recommendation domain was divided into two (RD1 and RD2) according to the relative importance of each crop (Figure 3.1). That is, RD1 and RD2 were identified based on the relative contributions of sorghum and millet residues to total available roughage (determined from secondary data) and on the relative importance of sorghum and millet straw in animal’s diets (information taken from the results of the primary survey).

RD1 is made up of 46 districts. In these districts, both sorghum and millet residues are important feeds (Table 3.1). Of the various sources of roughage available for livestock feed in RD1, millet makes up 35% and sorghum only 3% of the total. Buffalo and buffalo milk are relatively more important than cows and their milk here, with 7.1 million female buffalo compared to 5.1 million female cattle. With 7.9 million tonnes of milk from both species produced per year, this implies an average of 650 kg/female per year.

In the 58 districts of RD2, the share of sorghum and millet residues in total available roughage is 49% and 5%, respectively. Primary survey results indicated that sorghum stover is preferred to millet, whenever the sorghum is available. There are 9.3 million female cattle and 6.8 million female buffalo in RD2, producing 8.3 million tonnes of milk annually, or 516 kg/female per year, 20% less than in RD1.

Since a large part of the impact we are valuing in this analysis concerns the potential boost in output of milk with higher quality crop residues available as animal feed, the location and density of female cattle and buffalo within our RD is of interest. In India there is on average one milking bovine for every 2 ha of cropland. In almost one-half of the RD1 districts and one-quarter of the RD2 districts, the concentration of female animals was greater than the national average (Figure 3.2).

In terms of availability of milk per person, in most northern (RD1) districts the per capita availability of milk was greater than the Indian average of 200 g/day (73 kg/year) (Figure 3.3). In the south (RD2), only 16 out of the 61 districts produced more than 200 g of milk per person per day.  

Use of primary data to identify herd composition and feed use patterns

Description of survey and districts surveyed

A farm-level survey was designed and carried out to collect data on herd composition, feed availability and feeding practices in 16 districts—8 in each RD (Figure 3.1). In four of the districts in the north (Pune, Beed, Bhavnagar and Pali) both millet and sorghum are significant crops, each occupying over 10% of the cropped area. In the remaining districts,

Figure 3.2. Total adult female cattle and buffalo per hectare of cropped area in RD1 and RD2.

sorghum is found on more than 25% of the cropped area in 5 districts and millet on more than 20% of cropped area in 5 districts (Table 3.2). Six of these districts  (Solapur, Bijapur, Nanded, Akola, Mahabubnagar and Pune) were chosen from a group of selected districts included in the Natural Resources Institute (NRI)/ICRISAT/National Research Centre for Sorghum (NRCS) collaborative study on the factors affecting utilisation of sorghum in India, funded by DFID’s (Department for International Development, UK) Crop Post-Harvest Programme.

Figure 3.3. Per capita milk availability in RD1 and RD2.

Two villages were visited within each district, and five farmers were visited per village. A total of 160 farm households were interviewed. Households with different resource

endowments were selected purposively and only households with livestock were chosen.

While not a random survey, it was felt that even with such a small sample, much of the major variation in feed use patterns and herd composition was captured.

The districts surveyed in each RD were grouped into ‘dry’ and ‘wet’ potential target areas for the new varieties, where the ‘dry’ districts were the four with the lowest average rainfall levels and the ‘wet’ districts were those with the four highest rainfall levels. However, several of the districts categorised as ‘dry’ have some irrigated farm land (Table 3.2), thus the productivity levels found in the dry north will be higher than in the case with no irrigated districts included. The hypothesis is that herd composition, feed strategies, and productivity will differ according to overall rainfall levels. This hypothesis is explored in Tables 3.3–3.6.

Herd composition

Herd structures identified in each district are summarised in Table 3.3. Across the surveyed area, lactating buffalo represented 17% of the herd, lactating cows 18%, oxen 17% and others (young animals and dry bovines) 48%. Crossbred cows were few, mostly in the wetter north districts, where they represented 16% of the total herd.

In general, larger herds were found in the drier districts in the north, where an average herd size of greater than 15 head was found in three of the eight districts. In the south, all mean herd sizes were less than 7 head.  

The proportion of buffalo in the herd was greater in the north, and the proportion of lactating bovines was highest in the drier areas of RD1 (41% of the total herd). Oxen were more important in the south, making up over 30% of the herd, compared to less than 5% in the northern districts.  

Feed composition

Estimates of proportional contributions to animal intake from different sources, including grazing, are shown in Table 3.4. While it was apparent that grazing represented a significant source of feed in the south (42% in the drier districts and 44% in the wetter ones), grazing

contributed very little in the higher rainfall districts in the north (3%). Lower availability of grazing in RD1 was associated with greater use of cut-and-carry grass forage and concentrates.

Despite the low proportion (5%) of cropped area under sorghum in the surveyed districts of RD1 (Table 3.2), sorghum contributed 23% of the feed, much more than millet which covered 28% of crop land. Millet is a short-cycle crop that is harvested before the end of the rains; this explains the low contribution in the wet north. The ratio between sorghum and millet residues in the south corresponds to the planted acreage (Table 3.2).

As would be expected, millet residues contributed most significantly to livestock feed in the drier districts in RD1 (29%, Table 3.4). The higher contribution of other crop residues

(wheat, barley etc) in the wetter north is likely linked to the high level of irrigation (54%, Table 3.2) and cut-and-carry grass forage (13%, Table 3.4).

Feed allocation

The survey data were examined to determine how sorghum and millet stover was allocated to different groups of bovines. In the low rainfall districts of RD1, lactating cows and buffalo each make up 25% of total live weight of the herd, yet receive 41% and 32%, respectively, of sorghum residues (Figure 3.4). In contrast, non-lactating females and young stock received 27% of the stover, while accounting for 50% of the bovine live weight (Figure 3.5). In the wet north districts, lactating animals are also preferentially fed sorghum and millet residues.

Figure 3.4. Feeding of sorghum residues to lactating cows and buffalo.

In the southern districts, oxen comprise over 35% of the bovine mass, receiving 40% of the sorghum (Figure 3.5) and 32% of the very limited millet (data not shown). Lactating bovines receive on average around 30% of the sorghum fed in the south.

Figure 3.5. Feeding of sorghum residues to oxen and other animals.