MODELLING THE INFLUENCE OF IRRIGATION ... - Semantic Scholar

5 downloads 0 Views 171KB Size Report
This approach has been successful in Tanzania where CUPPA-Tea, a process- based simulation ... and the year-to-year variability have not been quantified.
C 2003 Cambridge University Press Expl Agric. (2003), volume 39, pp. 181–198  DOI: 10.1017/S0014479702001151 Printed in the United Kingdom

M O D E L L I N G T H E I N F LU E N C E O F I R R I G AT I O N O N T H E P OT E N T I A L Y I E L D O F T E A ( C A M E L L I A S I N E N S I S ) I N N O RT H - E A S T I N D I A By R. K. PANDA†, W. STEPHENS‡§ and R. MATTHEWS‡ †Agricultural and Food Engineering Department, Indian Institute of Technology, Kharagpur 721 302, West Bengal, India and ‡Institute of Water and Environment, Cranfield University, Silsoe, MK45 4DT, UK (Accepted 8 October 2002) SUMMARY This study reports the results of model simulations of the potential yield of tea in north-east India. The CUPPA-Tea model, developed using data from a high-altitude site close to the equator in East Africa, was validated against the yield data from irrigation experiments conducted on contrasting soil types at Siliguri and Tezpur in the tea growing region of north-east India. The close correspondence between observed and predicted yield and yield distribution suggests that the model is applicable in north-east India. The model was used to simulate the yield response of tea to drought and irrigation using daily weather data for seven years (1983–89) at Siliguri and 14 years (1974–85) at Tezpur. On a clay loam soil at Siliguri, with an available water capacity of about 200 mm m−1 , the predicted mean reduction in yield was 1.5 kg ha−1 for each 1 mm reduction in evapotranspiration. However, there appeared to be no reduction in evapotranspiration until the soil water deficit reached about 240 mm. By contrast, at Tezpur on loamy sand, with an available water capacity of about 100 mm m−1 , evapotranspiration was reduced once soil water deficits exceeded about 85 mm. There was then a mean reduction in yield of 2.2 kg ha−1 for each 1 mm reduction in evapotranspiration. At both sites, even when soil water deficits were not limiting, the predicted year-to-year variation in yields was about 500 kg ha−1 . The results highlight the soil-related differences in response to irrigation and the benefits of using process-based simulation models to investigate the potential yields over long periods.

I N T RO D U C T I O N

The annual yield distribution and potential yield of tea (Camellia sinensis) is largely influenced by seasonal fluctuations in weather variables such as rainfall, temperature and saturation deficit (SD), by soil water deficit (SWD) (Squire, 1979; Tanton, 1982a;b; Carr et al., 1987; Stephens and Carr, 1989) and by photoperiod (Barua, 1969; Tanton, 1982a; Matthews and Stephens, 1998c). Large yield peaks often occur following a cool or dry season, with subsequent peaks that may continue throughout the remainder of the season (Fordham and Palmer-Jones, 1977; Tanton, 1981). The timing and amplitude of these peaks depend on the prevailing weather and can vary considerably from year to year. Without a clear understanding of the processes that cause yield fluctuations it is difficult to interpret the results of experiments that modify yield distribution. Simulation models can help to investigate within- and § Corresponding author: [email protected]

182

R . K . PA N D A

et al.

between-season variability resulting from changeable weather and management inputs. This approach has been successful in Tanzania where CUPPA-Tea, a processbased simulation model, was used to investigate the potential yield of tea in different climates and the likely responses to irrigation (Matthews and Stephens, 1998b). In north-east India, irrigation is increasingly used, not only as an insurance against drought, but also to increase tea yields during the cool dry period from November to March. Dutta (1971) estimated that more than 50% of the area under tea cultivation in this area suffered from either water logging or drought, or from both. In Assam, observed annual yield increases due to irrigation ranged from 20% to almost 90%, but the extent of soil water stress was not reported (Hadfield, 1974). During the cool season, maximum potential soil water deficits range from 80 mm in upper Assam to 300 mm in the Terai region (Dabral, 1994). However, the yield responses to drought and the year-to-year variability have not been quantified. This paper describes the use of the CUPPA-Tea growth simulation model to examine these responses in West Bengal and Assam, two tea-growing states in northeast India. The specific objectives of this study were to: (i) validate of the model for north-east India; (ii) simulate temporal variation of average annual yield in response to changes in weather; (iii) simulate the yield responses to irrigation; and (iv) determine the relationships between the estimates of water use and predicted yields. METHODOLOGY

The simulation model The CUPPA-Tea simulation model (Matthews and Stephens, 1998a), developed by the International Centre for Plantation Studies, Cranfield University, Silsoe, U.K., was used for the study. The model simulates the behaviour of a population of shoots in which the components develop and extend independently at different rates representing the variation observed in natural conditions. The development of each shoot is divided into three phases corresponding to the resting, quiescent and bud-burst phases in temperate trees, with the rate of development in each phase being influenced by temperature, photoperiod, SWD and SD. The model assumes that photoperiod influences the onset and release of bud dormancy, and hence the number of actively growing shoots at any time. As shoots are normally harvested at a specific developmental stage or size, the number of shoots plucked at each harvest is generally the main determinant of yield variation. The model is most sensitive to the values of the two critical photoperiod parameters controlling bud dormancy and shoot development. The shoot growth and development parameters were calibrated using data for clone 6/8 (a widely grown Kenyan clone) from irrigation experiments at an altitude of 1800 m asl in the Southern Highlands of Tanzania (8◦ S, 35◦ E; Matthews and Stephens, 1998c). The base temperature for shoot extension for clone 6/8 is 10 ◦ C and the optimum temperature in CUPPA-Tea has been set to 24 ◦ C. Hadfield (1976) showed that in north-east India, leaf temperatures are often 5–10 ◦ C above air temperature. Thus the critical temperature used in CUPPA-Tea is equivalent to a leaf temperature of around 30–35 ◦ C identified as the optimum temperature for

Modelling the influence of irrigation on tea (Camellia sinensis)

183

photosynthesis by Hadfield (1976). Within the model, the critical SD is set at 2.3 kPa (Tanton, 1982a) with a linear decrease in shoot extension at greater SDs. There is no fixed critical SWD as the modelled response depends on the evaporative demand from the atmosphere, the size and distribution of the root system, and the soil water release curve for each soil type. The model has been successfully validated for conditions in Tanzania and Zimbabwe (Matthews and Stephens, 1998a;b) but has not previously been used outside the tropics, in the northern hemisphere, nor in areas with high summer temperatures such as those experienced in north-east India. Study area To validate the CUPPA-Tea model, weather, soils and yield data were used from irrigation experiments located in two major tea growing regions in north-east India, namely Siliguri in West Bengal (26◦ 47 N, 88◦ 21 E; 160 m. asl) and Tezpur in Assam (26◦ 43 N, 92◦ 26 E; 80 m. asl) some 400 km to the east. The sites have similar climates (Table 1) with the average annual rainfall of about 2000 mm falling mostly during the monsoon from May until the end of September (Siliguri) and April until mid-October (Tezpur). The daylength varies from 10.3 h in December up to 13.7 h in June. During the monsoon, heavy cloud reduces the incoming solar radiation to about 40% of the potential for that latitude. Mean daily temperatures increase from about 17–18 ◦ C in December and January to between 25 and 29 ◦ C from April to October. There is a rapid transition, therefore, in February and March as temperatures rise and the evaporative demand increases, exceeding rainfall by about 50 mm in February and 70–80 mm in March. The soil types vary markedly between the two sites. At Siliguri the soil is a clay loam with an available water content of about 200 mm m−1 , whereas at Tezpur the tea was grown on a loamy sand with an available water content of only 100 mm m−1 . Values of other soil characteristics are summarized in Table 2 and are within acceptable limits for tea growth (Othieno, 1992). Model validation Very few data were available to parameterize the model for the validation procedure. Data on the yield of made tea for mature tea (probably grown from seed rather than vegetatively propagated) from irrigated and unirrigated plots were available for 1989 and 1990 at Siliguri, and for 1986 and 1987 at Tezpur (Dabral, 1994; Tripathy, 1988), but harvest dates had to be estimated from graphs of yield distribution as the original data were not presented by the authors. The model runs on a daily time step, but the weather data available for the two sites were only weekly mean values of rainfall, wind run, maximum and minimum air temperatures, maximum saturation deficit and solar radiation. For each site the model was set up to run with the appropriate soils data. As no data were available on shoot growth and development for the tea grown on each site, the

184

Table 1. Mean monthly meteorological data for Tezpur, Assam and Siliguri, West Bengal. Tezpur, Assam

Siliguri, West Bengal

Solar radiation (MJ m2 d−1 )

ETo (mm month−1 )

Rain (mm month−1 )

Mean temperature (◦ C)

Saturation deficit (kPa)

Solar radiation (MJ m2 d−1 )

ETo (mm month−1 )

Rain (mm month−1 )

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

17.6 19.6 23.1 25.3 26.4 28.2 28.7 28.8 28.2 26.0 22.1 18.6

1.6 1.7 2.1 1.9 1.6 1.5 1.6 1.6 1.6 1.7 1.8 1.7

15.0 16.5 20.0 21.3 20.0 18.3 17.7 17.5 16.8 16.7 15.4 14.6

62 73 121 138 130 120 121 118 105 99 78 62

15 23 55 149 306 297 336 313 236 133 23 9

16.8 18.8 22.8 25.9 27.0 28.4 27.9 28.7 27.1 25.6 21.9 18.4

1.8 2.0 2.5 2.3 2.0 1.9 1.5 1.8 1.6 2.0 2.3 2.1

20.2 20.4 21.5 20.8 18.3 15.9 14.1 15.6 17.6 22.2 21.5 20.7

77 75 112 121 119 106 98 107 106 120 96 83

6 31 31 97 202 280 640 399 347 64 11 16

Total Mean

1227

1895

1219

2125

24.4

1.7

17.5

24.1

2.0

19.1

Month

Source: FAO, 1987; Tripathy, 1988; Dabral, 1994.

et al.

Saturation deficit (kPa)

R . K . PA N D A

Mean temperature ( ◦ C)

Modelling the influence of irrigation on tea (Camellia sinensis)

185

Table 2. The physical and chemical properties of the soil at the two sites. Soil property

Tezpur

Siliguri

Sand (%) Silt (%) Clay (%) Dry bulk density (Mg m−3 ) Field capacity (% volume basis at 0.033 MPa) Permanent wilting point (% volume basis at 1.5 MPa) Available water capacity (mm m−1 ) Soil pH (saturation extract 1 : 2 soil : water) Organic C content (%)

74.4 25.6 0.0 1.7 16.8 6.8 100 5.2 1.2

41.0 26.4 32.6 1.1 35.5 15.4 201 4.7 2.1

Source: Tripathy, 1988; Dabral, 1994.

genotype parameters for clone 6/8 were used. In the absence of specific management information from the two irrigation experiments, the model was run from 1 January with all shoots as buds. This represented the common practice in the region of ‘skiffing’ in which the top layer of foliage is removed during the winter season. Harvesting dates estimated from the results of the irrigation experiments were used in the model. At Siliguri, there were 28 and 24 harvests respectively during 1989 and 1990. During May to October, the tea was harvested every 8–10 d with longer intervals of up to 30 d at the beginning and end of the season when temperatures were cooler. At Tezpur, there were 34 harvests during the 1987 season from March to November, with an average interval of about 8 d. The harvest standard (i.e. the instructions given to harvesters as to the shoot size and number of leaves to be removed) at the two sites was not reported by Tripathy (1988) or Dabral (1994) so the model was set to remove all shoots emerging above the horizontal canopy at each harvest, regardless of size or number of leaves. This approach, colloquially known as ‘plucking black’, is commonly used where temperatures are high and shoot growth is rapid, as it reduces the number of small shoots remaining after harvest that become overgrown by the next harvest. As root depth was not recorded in either experiment, maximum root depth was assumed to be 1 m, corresponding to the approximate depth of the water table during the mansoon. The model was run with and without irrigation. In the irrigated scenario, for both sites, 50 mm of water was applied when the SWD reached 50 mm, to ensure that the yield was not limited by water stress at any time. Multiple year simulation The annual variability in yield response to irrigation and drought at the two sites was then investigated by running the model with actual daily weather data for 14 years (1974–85) at Tezpur and seven years (1983–89) at Siliguri, using the same soil characteristics and crop parameters as those used in the validation procedure. The actual harvest intervals in the experiments used for validation were inconsistent between treatments (Tezpur) and between years (both sites). To increase

186

R . K . PA N D A

et al.

the comparability between the two sites, therefore, a common harvest interval was used: 21 d in January and February; 10 d in March; 7 d in April to September inclusive; 14 d in October and November; and 21 d in December. This reflects ease of commercial management where the same field will always be visited on the same day. The model was run continuously over the entire simulation period so that the water balance carried forward to the next year. This means that, with the exception of the first year, the simulated yield is for unskiffed tea since the shoots continued to extend and develop over the winter period. To evaluate the range of responses to soil water status, the model was run with several irrigation regimes in which the soil water deficit was allowed to build up until a ‘trigger value’ of between 50 mm and the total available water was reached before applying sufficient irrigation water to return the soil to the upper drainable limit. This strategy results in a variable irrigation frequency depending on the water use of the crop and the timing and amount of rainfall. R E S U LT S

Model validation The data available for validating the model at the two sites were limited. In particular, the daily weather data had to be estimated from weekly means and totals. Despite this limitation, the yield distribution of made tea predicted by CUPPA-Tea at Siliguri agreed closely with observed yield of made tea per harvest under both unirrigated and irrigated conditions (Figure 1). CUPPA-Tea predicted the start and end of the season correctly and both the timing and the amplitude of the peaks corresponded well, though in 1990 the predicted yields were rather higher than observed values. A paired t-test on yield at each harvest showed no significant difference between means (n = 52). The Pearson correlation coefficients (r) between simulated and observed yields from individual harvests were 0.83 and 0.70 for irrigated and unirrigated tea respectively indicating that the predicted yields accounted for 69% and 49% of the variation in observed yields. At Tezpur there was less inter-harvest variation in yield with no large peaks at either the beginning or end of the season. The timing differences between observed and simulated peaks resulted in no significant relationship between yields at individual harvests. In addition, the observed mean yield per harvest for irrigated tea at Tezpur in 1986 was significantly greater (30 kg ha−1 ; p < 0.05) than the predicted yields. CUPPA-Tea predicted a large peak in October, associated with the longer harvest intervals, which was completely absent in the observed data (Figure 2), shown at the same scale as results for Siliguri. In 1987, there was no significant difference between observed and predicted yields for each harvest, although, once again, predicted peaks in October and November were not recorded in the observed data. In unirrigated tea, the predicted yield pattern matched the observed data better in 1987 than in 1986 but there was no significant difference between predicted and observed yields in either year. Again, the major difference was the predicted yield peak in October each year, which was not observed in the field.

Modelling the influence of irrigation on tea (Camellia sinensis)

187

(a) Unirrigated

Yield per harvest (kg ha–1)

Unirrigated Simulated

(b) Irrigated Irrigated Simulated

1989

1990

Figure 1. Comparison between observed and simulated distributions in the yield of made tea for a) unirrigated and b) irrigated conditions at Siliguri, NE India during 1989 and 1990.

In order to overcome the effect of small differences in the timing of peaks on the correlation between the observed and predicted yield at individual harvests, data were aggregated on a quarterly basis. At this time scale the yields predicted by CUPPA-Tea accounted for 77% of the observed variation. Regression analysis of predicted against observed quarterly yields of irrigated and unirrigated tea at the two sites showed that the regression intercept was not significantly different from zero and that the slope of the regression line did not differ significantly from unity (Figure 3). The observed

188

R . K . PA N D A

et al.

Yield per harvest (kg ha–1)

(a) Unirrigated

Observed Simulated

(b) Irrigated

1986

1987

Figure 2. Comparison between observed and simulated distributions in the yield of made tea for a) unirrigated and b) irrigated tea at Tezpur, NE India during 1986 and 1987.

Predicted yield of tea (kg ha–1)

y = 1.05x r 2 = 0.77

Siliguri irrigated Siliguri unirrigated Tezpur irrigated Tezpur unirrigated

Observed yield of tea (kg ha–1) Figure 3. Comparison of observed and predicted quarterly yields of made tea for irrigated and unirrigated tea at Siliguri (1989 and 1990) and Tezpur (1986 and 1987). The regression line is for the pooled dataset and is forced through the origin (n = 32).

and predicted annual yields of made tea for the two sites are presented in Table 3. The slope of the regression line through these points (forced through the origin as the intercept was not significantly different from zero) was y = 1.07x though the smaller

Modelling the influence of irrigation on tea (Camellia sinensis)

189

Table 3. Observed and predicted annual yields of made tea at Siliguri and Tezpur. Yield (kg ha−1 )

Siliguri Irrigated Unirrigated Tezpur Irrigated Unirrigated

Year

Observed

Predicted

1989 1990 1989 1990

4080 4080 3350 3420

4730 5060 3250 4730

1986 1987 1986 1987

3660 3470 2510 2430

2570 3020 2850 2640

number of observations (n = 8) and the underestimate of yields from irrigated tea in Tezpur meant that the relationship only accounted for 46% of the observed variation at this time scale. Multiple year simulation The agreement between the observed and predicted yield distribution and annual yields for Siliguri and Tezpur conditions indicated that the CUPPA-Tea model could be used with some confidence to investigate the effects of year-to-year variations in weather on the responses of tea to irrigation and drought in north-east India. At Siliguri, the predicted critical SWD that resulted in a decrease in annual yield was about 240 mm, equivalent to more than 90% of the plant-available water (Figure 4). By contrast, for tea grown on very sandy soils at Tezpur the predicted critical SWD was only about 85 mm, though this still represented about 80% of the available water. In all years, irrigation was predicted to increase yields above that of unirrigated tea. However, the year-to-year variability in the yield response to irrigation depended predominantly on the duration of the dry season, since the maximum SWD was predicted to be close to the limit of plant-available water in all years. For example, at Siliguri, in 1980 the SWD did not exceed 200 mm until April and the predicted yield of unirrigated tea was only about 100 kg ha−1 less than from a fully irrigated crop. In 1981, on the other hand, the SWD reached 240 mm by early January and remained at that level until April; the predicted difference in yield was about 600 kg ha−1 (Figure 5). By contrast, at Tezpur, where the total available water in the soil was less than half that in Siliguri, irrigated tea was always predicted to yield at least 400 kg ha−1 more than unirrigated tea (Figure 6). Saturation deficits also limit shoot extension and hence yield. However, it is difficult to separate the effects of SWD and SD as they tend to be correlated. At Tezpur the SD rarely exceeded Tanton’s (1982a) critical value of 2.3 kPa but at Siliguri this happened every year. The relationship between predicted annual yield and annual cumulative SD above the critical value at Siliguri (Figure 7) shows linear relationships with similar

R . K . PA N D A

et al.

Annual yield (kg ha–1)

190

Maximum SWD (mm)

Annual yield (kg ha–1)

Figure 4. Relationship between mean annual yield of made tea and maximum soil water deficit predicted by the CUPPA-Tea model for the Tezpur (squares) and Siliguri (triangles) sites. Error bars indicate the s.e. of the means across years (Tezpur: n = 13; Siliguri: n = 7). See text for details.

Year Figure 5. Predicted annual yields at the Siliguri site for the period 1978 to 1984 for tea where the soil was returned to field capacity when the SWD reached 240 mm (solid line); and unirrigated tea (dotted line).

191

Annual yield (kg ha–1)

Modelling the influence of irrigation on tea (Camellia sinensis)

Year

Figure 6. Predicted annual yields of made tea at the Tezpur site for the period 1974 to 1987 for three irrigation treatments (irrigating to field capacity at 70 (squares), 90 (diamonds) and 100 mm (triangles) SWD) and for unirrigated tea (circles).

Annual yield (kg ha–1)

y = –11.4x + 3491 r 2 = 0.65

Irrigated

y = –11.2x + 3082 r 2 = 0.77

Unirrigated

Annual SD > 2.3 kPa (kPa d) Figure 7. Relationship between predicted annual yield of made tea at Siliguri and annual cumulative saturation deficit > 2.3 kPa.

192

R . K . PA N D A

et al.

Monthly yield (kg ha–1)

(a) Siliguri

< 240 mm SWD

Unirrigated

Monthly yield (kg ha–1)

(b) Tezpur

Irrigated at 80 mm max SWD Irrigated at 100 mm max SWD Unirrigated

Month Figure 8. Mean simulated monthly yield of made tea under irrigated and unirrigated conditions at a) Siliguri (1978– 1984) and b) Tezpur (1974–1987). The error bars are the standard error of the means (Siliguri, n = 7; Tezpur, n = 13). See text for details.

slopes for both irrigated and unirrigated tea, accounting for 65% and 77% of the variation respectively. Extrapolating back to zero SD limitation suggests a potential yield of tea of about 3500 kg ha−1 , very similar to that predicted for fully irrigated tea at Tezpur. Average monthly yield At both sites, the model predicted that irrigation would change the monthly yield distribution by increasing yields during March when temperatures were rising rapidly and potential evapotranspiration substantially exceeded rainfall (Figure 8). At Siliguri,

Modelling the influence of irrigation on tea (Camellia sinensis)

193

4000 3800

Simulated annual yield (kg ha–1)

3600

Tezpur y = 2.22x + 640 r 2 = 0.70

3400 3200

Siliguri y = 1.54x + 1100 r 2 = 0.54

3000 2800 2600 2400 Tezpur Siliguri

2200 0 0

Figure 9.

900

1200 1000 1100 Annual evapotranspiration (mm)

1300

Relationship predicted by the CUPPA-Tea model between simulated yields of made tea and annual evapotranspiration for Tezpur (n = 78) and Siliguri (n = 35), NE India.

there were no significant differences in mean monthly made-tea yields other than in March, but at Tezpur the effects of drought continued to modify the yield distribution until about September. Irrigation was also predicted to decrease the year-to-year variation in monthly yields, as indicated by the standard error of the mean for each month. Interestingly, at Tezpur there was little inter-annual variation in yields of unirrigated tea in March, reflecting the consistently later start to the monsoon at that site. Relationship between evapotranspiration and yield The predicted relationship between predicted yield and simulated crop evapotranspiration varied between the two sites (Figure 9). At Tezpur, the linear relationship accounted for 70% of the observed variation compared with only 54% at Siliguri. Within the range of annual evapotranspiration that occurred over the period for which weather data existed (800 to 1200 mm), the annual yield was reduced by 2.2 kg (ha mm)−1 at Tezpur compared with 1.5 kg (ha mm)−1 at Siliguri. However, neither regression line passed through the origin, suggesting a curvilinear relationship at lower evapotranspiration values. The effect is that the water use efficiency for harvested shoots is greatest when evapotranspiration is least. DISCUSSION

The acid test for a crop model is its performance in entirely different climatic and soil conditions from which it was developed, and its sensitivity to being run with limited

194

R . K . PA N D A

et al.

data. In this exercise, the model used was developed and calibrated for clonal tea grown in the highlands of East Africa. It was applied in a very different environment in north-east India without modification to any of the crop parameters, with weekly mean- instead of daily weather data, and with estimated harvest dates. Low resolution input data inevitably limits the potential accuracy of the resulting model output as short-term temporal variations in weather are not represented well. This would be most critical during cooler and drier periods when small changes in temperature or soil water status can have large effects on shoot growth rates, which, in turn, affect yields at subsequent harvests by altering the shoot population structure. When running CUPPA-Tea, initial assumptions about the shoot population structure and soil water content also need to be made. The authors’ assumption that the fields were skiffed in the months prior to the start of the experiments appears to be valid at Siliguri as the timing of the peak in both treatments is predicted well. The model was run through to the second year without skiffing and, in both irrigated and unirrigated treatments, the over-prediction at the first harvest suggests that this assumption might have been incorrect. However, in the absence of information on the actual management practices the authors felt that making further changes to input parameters was not warranted even if they resulted in a better fit. At Tezpur, CUPPA-Tea over-predicted unirrigated yields at the beginning of the first year, probably because the assumption of the initial soil water status, which was not recorded in the experiment, was incorrect. The resulting growth predicted early in the season then altered the subsequent yield distribution. In the irrigated plots there were more harvests early in the season and, therefore, no large peak. Interestingly, CUPPA-Tea consistently underestimated actual yields from irrigated plots in the first year, except where the harvest interval doubled in October. It was not possible to determine from the available information whether a management practice, common in some parts of India, of leaving an additional leaf on the bush towards the end of the year to replace the main canopy foliage was being followed at Tezpur, though this would explain the discrepancy. The observed yield differences between the two sites are dominated by the large peaks that occurred at Siliguri at the beginning of the first year and the end of the second year. The key difference in management between the two sites was the harvest interval, which was much less regular at Siliguri and was characterized by particularly long intervals prior to the large peaks. Previous work has shown that predicted yields are sensitive to the harvest interval and plucking standard. In this paper, the authors have made the minimum number of assumptions to allow independent validation, but specific management practices, such as allowing the canopy height to increase by an internode, would result in differences between observed and predicted yields. Given the uncertainties in starting conditions and the poor temporal resolution of the weather data, the relatively good correspondence, not only in seasonal yields but also their temporal distribution, is believed to be an indication of the robustness of the assumptions underlying the model. The results presented here suggest that the CUPPA-Tea model can be used with some confidence on different soil types with

Modelling the influence of irrigation on tea (Camellia sinensis)

195

widely differing available water capacities. It has now been validated at latitudes from 22◦ S in Zimbabwe (Matthews and Stephens, 1998b) to 26◦ N in India (this paper) and for mean monthly temperatures from 15 ◦ C in Tanzania to just under 30 ◦ C, again in India. The effect of differences in harvest interval on the seasonal yield distribution of tea is evident when Figures 1 and 2 are compared with Figure 8, though the last figure has been aggregated to a monthly timescale for clarity. The multi-year simulations did not impose a skiff between years and this would result in an earlier peak than in the validation data (where the bushes appear to have been skiffed) as the top layer of shoots is not removed. In most years, the predicted yields of unirrigated tea were low until the start of the monsoon in April/May. Most of the expected benefit from irrigation resulted from alleviating water stress in March, thus enabling shoot growth to commence at least a month earlier than in unirrigated tea. This response is very similar (but six months out of phase) to that observed in Tanzania (Stephens and Carr, 1991a) and Zimbabwe (Matthews and Stephens, 1998b) where temperatures no longer restrict yields at the end of the dry season. Overall, the water-use efficiencies at Siliguri and Tezpur were within the range of 1–4 kg (ha mm)−1 reported previously for clone 6/8 in Tanzania (Stephens and Carr, 1991b). The difference in the predicted yield/evapotranspiration relationship between the sites (Figure 9) suggests that water stress is not the only factor moderating yields in Tezpur and Siliguri. In particular, the estimated annual evapotranspiration accounted for 70% of the predicted variation in yields at Tezpur but only just over half at Siliguri. The other two climatic factors likely to affect yield are air temperature and SD, both of them through direct effects on the rates of shoot extension (Matthews and Stephens, 1998a). In Zimbabwe, year-to-year variations in SD resulted in 0–20% variation in yield depending on the severity of the dry season. At Siliguri, the monthly mean SDs equalled or exceeded the critical SD of 2.3 kPa in March, April and November, suggesting that this could limit the potential yield response to irrigation. It might explain why the large yield peak observed at Tezpur in March was not matched in Siliguri despite the very similar mean temperatures during March/April. Climatically, the two sites in north-east India are similar yet the response to SWD varied considerably. This was predominantly due to differences in soil type, with twice the volume of available water in the soil at Siliguri as at Tezpur. Thus, the critical SWD beyond which yields were reduced was about 240 mm at Siliguri compared with 85 mm at Tezpur. The effect is to give very different responses to irrigation – the estates in Tezpur would see a much greater return on their capital than would those in Siliguri but the former would need to install sufficient equipment to be able to irrigate more frequently than would be necessary in Siliguri. Returning the soil to the upper drainable limit once the SWD has reached 240 mm is not, however, a practicable irrigation schedule. Maximum applications of about 100 mm are possible with large hand-moved sprinkler laterals operating for approximately 22 h. In the current case, it would be appropriate to use a deficitirrigation schedule whereby 100 mm was applied when the SWD reached, say,

R . K . PA N D A

et al.

Cumulative frequency

196

Siliguri Tezpur

Annual yield increase with irrigation (kg ha–1) Figure 10. Frequency distribution of annual yield increases due to irrigation at Siliguri and Tezpur.

200 mm. This would allow better use of any rainfall within the dry season, as there would still be storage capacity in the soil. The actual benefits of irrigation will vary from year to year so managers need information on the likelihood of a given increase in yield. A cumulative frequency graph, of each year’s predicted yield increase attributable to irrigation (Figure 10) shows the predicted differences in response between the two sites. At each site, in the year with most rain during the dry season, the simulated yield increase due to irrigation was only about 200 kg ha−1 . At Tezpur, however, there was only a 20% chance that the yield increase due to irrigation would be less than 520 kg ha−1 whilst, at Siliguri, this increase in yield would only be achieved in three years out of ten. When these yield increments are converted to irrigation water use efficiencies then the differences are even clearer (Figure 11). The range at Siliguri is limited to about 0.4 to 1.1 kg (ha mm)−1 whereas at Tezpur the response is greater ranging from 0.8 to 2.7 kg (ha mm)−1 . The variability in response at Tezpur probably reflects the more frequent rainfall during the dry season from November to April because rain falling just after an irrigation event benefits unirrigated but not irrigated tea and, hence, reduces the incremental water use efficiency. By contrast, the lower irrigation use efficiencies predicted at Siliguri relate to the effects of limiting saturation deficits during the dry season. Indicative running costs of irrigation from north-east India are about US$0.45 (ha mm)−1 (Dabral, 1994). Prices obtained for teas from this region at the Siliguri auction were in the range US$1.3–2.0 kg−1 (Anon, 1998). Using these figures, the

Cumulative frequency

Modelling the influence of irrigation on tea (Camellia sinensis)

197

Siliguri Tezpur

Irrigation water use efficiency (kg (ha mm)–1) Figure 11. Frequency distribution of predicted irrigation water use efficiency (defined as the yield of made tea per mm of irrigation water applied) at Siliguri and Tezpur.

break-even irrigation water use efficiency is in the range 0.23–0.35 kg (ha mm)−1 suggesting that there would be clear benefits from irrigation in all years at both sites. If tea prices dropped to $1 kg−1 , however, then irrigation would be uneconomical in four years out of ten at Siliguri. In Tanzania, irrigation benefits factory managers since the monthly variation in yields is reduced and factory capacity can be used more efficiently (Stephens and Carr, 1991a). This effect was not evident at either Siliguri or Tezpur and, conversely, irrigation at Tezpur could result in an increased requirement for factory capacity because mean yields in March exceeded those in any other month by about 50%. These results suggest that the potential yield responses of tea to irrigation in northeast India are unlikely to be uniform, even across ostensibly similar climatic zones. Differences between sites in soil water-holding capacity, in the seasonal distribution of rain, and in SDs can all affect the yield response to irrigation and hence its economic viability. Acknowledgements . This study was conducted during the visit of the senior author to Cranfield University, Silsoe, UK, under the international exchange programme of the Indian National Science Academy and the Royal Society. The financial support provided by the latter two organizations and the facilities provided by Cranfield University to the senior author during the study period are acknowledged sincerely.

198

R . K . PA N D A

et al.

REFERENCES

Anon. (1998). Tea Digest. Calcutta: Tea Board of India. Barua, D. N. (1969). Seasonal dormancy in tea (Camellia sinensis L.). Nature 224:514. Carr, M. K. V., Dale, M. O. and Stephens, W. (1987). Yield distribution in irrigated tea (Camellia sinensis) at two sites in eastern Africa. Experimental Agriculture 23:75–85. Dabral, P. P. (1994). Water management for optimal tea production – A case study. Unpublished Ph.D. thesis, Indian Institute of Technology, Kharagpur. Dutta, S. K. (1971). Effects of irrigation of tea in Assam. In Water and the Tea Plant, 145–149 (Eds M. K. V. Carr and S. Carr). Kericho, Kenya: Tea Research Institute of East Africa. FAO (1987). Agrometeorological data sets: Asia. Rome: Food and Agriculture Organization of the UN. Fordham, R. and Palmer-Jones, R. W. (1977). Simulation of intraseasonal yield fluctations of tea in Malawi. Experimental Agriculture 13:33–42. Hadfield, W. (1974). Shade in north-east Indian tea plantations. II. Foliar illumination and canopy characteristics. Journal of Applied Ecology 11:179–199. Hadfield, W. (1976). The effect of high temperature on some aspects of physiology and cultivation of the tea bush (Camellia sinensis) in N.E. India. In Light as an Ecological Factor, 477–495 (Ed. G. C. Evans). London: Backwell Scientific Publications. Matthews, R. B. and Stephens, W. (1998a). CUPPA-Tea: A simulation model describing seasonal yield variation and potential production of tea. I. Shoot extension and development. Experimental Agriculture 34:345–367. Matthews, R. B. and Stephens, W. (1998b). CUPPA-Tea: A simulation model describing seasonal yield variation and potential production of tea. II. Biomass production and water use. Experimental Agriculture 34:369–389. Matthews, R. B. and Stephens, W. (1998c). The role of photoperiod in determining seasonal yield variations in tea. Experimental Agriculture 34:1–17. Othieno, C. O. (1992). Soils, In Tea: Cultivation to Consumption, 137–172 (Eds K. C. Willson and M. N. Clifford). London: Chapman & Hall. Squire, G. R. (1979). Weather, physiology and seasonality of tea (Camellia sinensis) yields in Malawi. Experimental Agriculture 15:321–330. Stephens, W. and Carr, M. K. V. (1989). A water stress index for tea (Camellia sinensis). Experimental Agriculture 25:545–558. Stephens, W. and Carr, M. K. V. (1991a). Responses of tea (Camellia sinensis) to irrigation and fertilizer. I. Yield. Experimental Agriculture 27:177–191. Stephens, W. and Carr, M. K. V. (1991b). Responses of tea (Camellia sinensis) to irrigation and fertilizer. II. Water use. Experimental Agriculture 27:193–210. Tanton, T. W. (1981). The banjhi (dormancy) cycle in tea (Camellia sinensis). Experimental Agriculture 17:149–156. Tanton, T. W. (1982a). Environmental factors affecting the yield of tea (Camellia sinensis). II. Effects of soil temperature, day length and dry air. Experimental Agriculture 18:32–63. Tanton, T. W. (1982b). Environmental factors affecting the yield of tea (Camellia sinensis). I. Effects of air temperature. Experimental Agriculture 18:47–52. Tripathy, P. K. (1988). Maximization of tea production by judicious resource management. Unpublished Ph.D. thesis, Indian Institute of Technology, Kharagpur.