62 C.J. Pilbeam et al. Agriculture, Ecosystems and Environment 79 2000 61–72
1995. The grazing of animals in the forest, the col- lection of forage from the forest to feed housed ani-
mals, and the use of forest litter for animal bedding, mean that there is a net movement of nutrients from
the forest to the cropped areas via livestock produc- tion Turton et al., 1995. In addition to the use of ma-
nures and composts, soil fertility of the cropped areas is maintained by a variety of other techniques, for ex-
ample diverting nutrient-rich flood waters on to khet land, tethering animals on cropped land, and growing
legumes either as green manures or in rotation. Crop yields in the traditional systems are low and insuffi-
cient to support a burgeoning population. Intensifying agricultural production increases productivity, which
is necessary, but simultaneously results in more open nutrient cycles Jodha, 1990 thereby increasing the
susceptibility of the system to larger nutrient losses and, perhaps, reducing its long-term sustainability.
A nutrient balance quantifies the input of a particu- lar nutrient to an area of land and subtracts from this
the output of the same nutrient from the same area of land Stoorvogel et al., 1993. The balance may
be positive, if inputs exceed outputs, or negative, if outputs exceed inputs, and is a quantifiable indicator
of sustainability Smaling et al., 1996. Nutrient bal- ances not only indicate sustainability but, by chang-
ing the inputs and outputs, also allow the outcome of different future scenarios to be compared Stoorvo-
gel et al., 1993. While these scenarios are generally compared on the basis of the particular nutrient, their
socio-economic impact can also be evaluated if rele- vant economic data are available de Jager and Smal-
ing, 1997. Nutrient balances can be constructed at a range of scales, for example supra-national Stoorvo-
gel et al., 1993 or district Smaling et al., 1993 but it is at the household level that nutrient fluxes are most
directly affected by management decisions Smaling et al., 1996. Furthermore, the concept of sustainabil-
ity combines balanced resource use with economic vi- ability, and it is at the level of the household that the
biological, social and economic considerations are in- tegrated.
This paper aims to determine a N balance for a hypothetical household in the mid-hills of Nepal and
thereby highlight gaps in the quantitative knowledge of N flow. Using the quantitative cycle, sustainability
was assessed with respect to N of the current farming systems in the hills of Nepal, along with the impact of
perturbations to the system caused by either a change in the supply of N to or from parts of the system, or
an increase in the efficiency of N use Pandey, 1996.
2. Defining a typical household in the mid-hills
Survey data Turton et al., 1995 show that house- holds in the mid-hills of Nepal vary considerably in
terms of their size, cropping pattern, livestock num- bers and access to trees, as a consequence of physical
and social factors. Regional differences in the charac- teristics of households are also evident, although the
variation found within a region is probably as great as the differences between regions Tollervey, pers.
commun., 1998. In fact this assertion is supported by the results of an unpublished socio-economic
survey comparing 256 households stratified by food self-sufficiency in two village development commit-
tees VDCs in the western region with 256 house- holds in two VDCs in the eastern region of Nepal
Mathema, pers. commun., 1999. For this reason, discrepancies in household characteristics between
regions will be ignored and, by amalgamating data from diverse sources, a hypothetical household in
the mid-hills will be defined. Although from diverse sources much of the data originates from either Agri-
cultural Research Station ARS Lumle in Kaski District in the western region, or ARS-Pakhribas in
Dhankuta district in the eastern region.
In a survey published in 1986, approximately 1 ha was owned per household LRMP, 1986b. Thapa
1989 observed an average farm size of 0.5–1.0 ha, which had fallen to 0.5 ha in surveys reported in
the mid 1990s Turton et al., 1995; Sharma, 1996a suggesting that in the last decade the average amount
of land owned by a single household has declined. In contrast to this apparent decline, a recent sur-
vey Mathema, pers. commun., 1999 of households shows that each owns on average 1 ha of land. Tak-
ing account of this finding, and for convenience, it will be assumed that the hypothetical household con-
sists of 1 ha. Land within a household is distributed between rainfed land bari, which is predominantly
cropped with maize Zea mays L., and irrigated land khet supporting rice Oryza sativa L. cultivation.
On average, a household owns approximately twice 1.8-times as much bari land as khet land Turton
C.J. Pilbeam et al. Agriculture, Ecosystems and Environment 79 2000 61–72 63
et al., 1995, although wealthier households own relatively more khet and less bari land than poorer
ones. It is common for khet and bari land for a single household to be widely dispersed. Sharma 1996a
reported that, on average, land holdings were divided into 3.9 different pieces of land. This was a reduction
from the values between 4 and 9 reported earlier by LRMP 1986b. It is assumed that the hypothetical
household owns four pieces of land.
Currently, there is an average of two crops per year grown on bari land. Production on bari land during
the monsoon season is dominated by maize LRMP, 1986a which yields from 1.4 Gurung and Shrestha,
1996 to 2.6 Mg ha
− 1
Vaidya and Gurung, 1995 in the eastern and western hills, respectively. Finger mil-
let Eleusine coracana, which is either relayed or grown sequentially with maize yields 1.16 Mg ha
− 1
on average Katuwal and Tiwari, 1997. In general, fewer crops are grown on khet land an average of
1.5–2.0 crops per year, but dependent upon altitude, where production is dominated by rice, which yields
2 Mg ha
− 1
on average Sherchan et al., 1999. Rice is commonly grown in rotation with partially irrigated
wheat Triticum sp., which yields 2–3 Mg ha
− 1
on av- erage Subedi, 1994; Tiwari et al., 1997. All of the
average yields reported here are derived from exper- imental plots associated with either ARS-Lumle or
ARS-Pakhribas. It is assumed that the grain yields from the hypothetical household are the average of
those presented in the preceding text. Assuming a har- vest index of between 0.4 and 0.45 for the major cereal
crops grown in the mid-hills of Nepal, approximate straw yields range from 1.4 Mg ha
− 1
for finger millet to 3.5 Mg ha
− 1
for wheat, with 2.5 Mg ha
− 1
of maize and rice straw.
More than half of the animals in Nepal are found in the mid-hills with 3.3 livestock units LSUs
per household, comprising one bullock or ox, one buffalo, two cows, two goats and poultry Sharma,
1996b. The actual numbers and species of animals per household vary by region, but most households
own cattle, buffalo, goats and poultry. Larger estimates of livestock numbers per household two buffalo, two
cows, two oxen, four goatssheep and six chicken, and equivalent to 5.2 LSU were made by Gurung et al.
1989 in the Koshi hills of eastern Nepal, while the averages 2.39 cattle, 1.42 buffalo, 2.32 goatssheep
and 0.23 pigs, or 3.8 LSU produced by Joshi and Panday 1991 are less than the 5.8 and 4.3 livestock
units per farm found by Thapa 1994 and Turton et al. 1995, respectively. Thorne pers. commun.,
1997 observed that numbers of LSU per household in E. Nepal varied from 2.5 to 8. Synthesizing all of
these values it is concluded that a hypothetical house- hold might, therefore, have 4.4 LSU, comprising in
order of decreasing size two buffalo, two oxen, one cow, three goats and some chickens.
Trees provide timber and fuel for the household and fodder for livestock. The numbers of trees per
household vary widely, but have generally increased on bari and non-cultivated land in recent years Carter
and Gilmour, 1989, particularly among wealthier farmers who possess more land. Estimates of the
numbers of trees per farm range from as low as 16.8 Hopkins, 1983 to 300–400 Fonzen and Oberholzer,
1984, although 36–160 Wyatt-Smith, 1982 and 50–80 Thapa, 1994 are perhaps more likely values.
The latter values have been chosen to represent the hypothetical household. 170 tree species are used for
fodder, but only 30 are used extensively Shrestha and Tiwari, 1992. Estimates of fodder production from
trees on farms are generally between 60 and 90 kg fodder per tree per year Hopkins, 1983, although a
more detailed study of annual production of individ- ual trees in different forests in the Central Himalaya
yield estimates of 4–90 kg fodder leaf and twig per tree per year Rana et al., 1989 depending on species
and forest type. This compares well with the estimates by Wyatt-Smith 1982 of 20–86 kg fodder per tree
per year. Less than one third of the fodder require- ments for animals come from trees LRMP, 1986c;
Subba et al., 1994, and as little as 15 of the total digestible nutrients come from fodder trees Shrestha
and Tiwari, 1992. According to Thapa et al. 1990, “although fodder trees and forest fodder provide a
nutritive supplement, the bulk of fodder during the pe- riod of scarcity is provided by crop residues”. The pe-
riod of scarcity occurs immediately prior to the onset of the monsoon season. Reference to the significance
of the forest and trees for fodder is regularly made Carson, 1992; Turton et al., 1995, but the preceding
analysis suggests that its importance is perhaps over- stated.
Table 1 summarizes the characteristics of a house- hold in the mid-hills of Nepal, and provides the basis
for calculating N flows.
64 C.J. Pilbeam et al. Agriculture, Ecosystems and Environment 79 2000 61–72
Table 1 A summary of characteristics selected from the available literature
to define a hypothetical household in the mid-hills of Nepal and subsequently used for calculating the N budget
Characteristic Attributed values
Total land area ha 1.0
Pieces of land 4
Bari : khet
a
2 : 1 Trees
50–80 No. of Animals
Buffalo 2
Bullock Ox 2
Cows 1
Goat 3
Chicken 6
Crops kg ha
− 1
Average grain yield
b
Maize Zea mays L. 2000 3000
Finger millet Eleusine coracana 1160 1740
Rice Oryza sativa L. 2000 3000
Wheat Triticum aestivum L. 2500 3750
a
Bari land is rainfed, while khet land is irrigated.
b
Average straw yield kg ha
− 1
in parenthesis assuming an harvest index of 0.4.
3. Estimating a nitrogen balance