Introduction

Environmental deterioration due to water shortages is an urgent problem in arid areas of north-west China (Kang et al. 2004). An example is the lower reach of the Shiyang River Basin (Minqin County) between the Tenggeli Desert and Badanjilin Desert in the Gansu Province. It is a typical irrigation-dependant oasis with abundant sunlight but limited rainfall (annual precipitation of 110 mm). As a result of the low rainfall and intense competition for water between the upper and lower reaches, the only surface water resource, i.e., the Hongyashan Reservoir at the end of Shiyang River, became dried up in June 2004. To maintain agricultural production and human consumption in the lower reach, deep groundwater at 80–300 m depth has been extracted. Utilization of these water resources has led to some serious consequences, e.g., gradually falling groundwater table, shrinking of vegetation areas, soil salinization and desertification. As the Shiyang River Basin is one of the main agricultural areas in the Hexi Corridor of Gansu Province and has large areas of cereal crops (e.g., spring wheat, maize) and cash crops (e.g., cotton, grape and melon), the competition for water resources between agricultural irrigation and ecological maintenance has also become a difficult problem. In recent years, cotton has become the main cash crop for local farmers in the lower reach of the river basin due to its high profitability. In oasis fields, effective use of limited water resources for cotton production will be a challenge in coming years.

The concept of alternate partial root-zone irrigation (APRI) or partial root-zone drying (PRD) has been raised and attracted considerable interest in recent years (Davies et al. 2000; Dry and Loveys 1998; Gu et al. 2000; Kang et al. 1997, 1998). This irrigation method may improve plant water use efficiency (WUE) without significant yield reduction, and is relatively easy to be applied in the field if alternate irrigation of part of the root system is possible. The concept is derived from earlier split-root studies (Blackman and Davies 1985; Zhang and Davies 1987) and requires parts of the root-zone being dried and wetted alternately (Kang and Zhang 2004). The novelty of this method is that it explores the plant response to soil drying and a root-sourced chemical signal from the drying roots can reach the shoots where physiological processes such as stomatal opening/closing and leaf growth can be controlled (Zhang and Davies 1990). So far APRI has been reported mainly on horticulture crops such as grapevine (Gu et al. 2000), tomato (Kirda et al. 2004; Wagdy et al. 2004; Zegbe et al. 2004), pear (Kang and Cai 2002; Kang et al. 2003) and peach (Gong et al. 2004). Experimental results from both greenhouse and field experiments have shown that APRI reduces irrigation volume by 30–50% with no significant yield reduction and in some cases with better fruit quality. For field crops such as maize, APRI has been shown to maintain high grain yield even with a 50% reduction in irrigation amount (Kang et al. 1998, 2000a, b; Pan and Kang 2000). In recent field experiments where cotton was furrow irrigated using an APRI method in Xinjiang, China, the irrigated volume was reduced by 30% with less than 5% reduction in seed cotton yield (Tang et al. 2005). Similarly, a 12.8–24.4% increase in seed cotton yield were also obtained under furrow irrigation using APRI in Gansu, China (Du et al. 2006).

Earlier experiments with roots split between two containers showed that vegetative growth rate and stomatal opening of leaves can be inhibited even though the shoot hydration is maintained by half of the roots in wet soil (Zhang et al. 1987; Gowing et al. 1990). Such a state was achieved when one of the containers was allowed to dry and roots in the other well-watered container were able to supply enough water to the leaves to prevent their water potential from falling. This concept led to the development of PRD on grapevine and other fruit crops (Dry et al. 2000; Cifre et al. 2005; Claudia et al. 2005; Kang et al. 1997, 2003; Kang and Zhang 2004; Mingo et al. 2004; Wakrim et al. 2005; Zhang et al. 2001). It is now clear that stomatal regulation under such APRI is controlled through the chemical signals from the roots exposed to the drying cycles. The increased concentration of abscisic acid in the xylem flow from roots to leaves triggers closure of stomata (e.g., Blackman and Davies 1985; Tardieu et al. 1993; Zhang and Davies 1987, 1990). APRI is therefore designed to expose part of the root system to drying soil and produce the root signal of drying, while the remaining roots in wet soil can maintain the water supply so that leaves are kept hydrated (Kang and Zhang 2004).

An extra benefit of APRI is that the signal from roots in drying soil may regulate the plant vegetative growth in addition to its effect on stomata. In crops such as cotton, maintaining an appropriate balance between vegetative growth and reproductive development and keeping a low abscission rate of flower buds and bolls is a key task in field management (Li 1979; Shi et al. 1987). Furthermore, the chemical methods that inhibit vegetative growth may contaminate the environment. Since PRD (or APRI) on grapevines have shown a potential to balance the vegetative growth and reproductive growth (Dry and Loveys 1998), it is hypothesized that these same benefits may be obtained in cotton crops.

Drip irrigation has been practiced for many years for its effectiveness in reducing soil surface evaporation. It has been used widely in horticultural crops in both greenhouses and the field. With the development of cheap and durable drip systems, it is expected that drip irrigation may be used with APRI for field crops such as cotton. In 2004 and 2005, we carried out field experiments to investigate the effects of partial root-zone drip irrigation on the soil moisture variation, physiological response, plant growth and yield of cotton crop in an oasis field where irrigation is virtually the only dependable source of water for the crops. The objective of this study was to validate the feasibility of using drip irrigation for APRI and to evaluate the consequences on the water use and yield of cotton.

Materials and methods

Experimental site

The field experiment was carried out during 2004–2005 at the Xiebai Experimental Station of Minqin Agricultural Extension Center of Gansu, on the lower reach of the Shiyang River Basin in the oasis region of north-west China (latitude 38°05′N, longitude 103°03′E, altitude 1,340 m). This station is located in a typical continental temperate arid zone with an average annual sunshine duration of 3,010 h, annual accumulated temperature (>10°C) of 3,148°C, mean annual precipitation of 110 mm (the precipitation during June to September occupies about 60% of the annual precipitation) and a mean annual evaporation from a free water surface of 2,644 mm. The groundwater table depth is consistently below 13–18 m. The soil has a sandy loam topsoil and a clay–loam subsoil with moderate permeability and organic matter content, average field capacity of approximately 0.308 cm3/cm3 in the upper 1.0 m of the soil profile and bulk density of approximately 1.4 g/cm3.

Experimental design

Two irrigation methods, i.e., conventional drip irrigation (CDI, both sides of plant row watered) or alternate partial root-zone drip irrigation (ADI, both sides of plant row alternatively watered), were applied. All irrigations were applied under plastic film mulching. Three irrigation levels, irrigating the ADI and CDI at the same levels (15.0, 22.5 and 30 mm for each irrigation in 2004 and 12.0, 18.0 and 24 mm in 2005, coded with 1, 2 and 3, respectively, and attached to CDI or ADI) were applied. The drip irrigation system used an in-line labyrinth drip-tape with one dripper per 30 cm and irrigation rate of 3 l/h. The timing of irrigation followed the local commercial practice and the volume applied was selected as 50, 75 and 100% of the product of evapotranspiration and crop coefficients under drip irrigated cotton under plastic film mulching carried out at the neighboring experimental station 3 km away in 2001 (Zhang et al. 2004). The field experiment had a total of 18 plots, arranged in a randomized block design with three replicates per treatment. Each plot area was 28.6 m2 (2.6 m × 11 m) in 2004, and 44.0 m2 (4.0 m × 11 m) in 2005, with a 30 cm wide spacing row between the two neighboring plots to eliminate the effect of lateral soil water movement. Drip lines were laid out in a ‘one-line two-row’ pattern (i.e., two rows of cotton were irrigated with one drip line), which is common in local cotton production. For the ADI treatment, the first and third drip lines were opened while the second and fourth drip lines were closed. For the following irrigation, the first and third drip lines were closed while the second and fourth drip lines were opened (Fig. 1). For the CDI treatment, all drip lines were open for each irrigation. There were four irrigations during 2004 and six irrigations during 2005 (Table 1). The irrigation volume for each treatment was controlled by a flow meter installed on the gated plastic laterals. All drip lines were placed on the top of the plastic mulch and water infiltrated into the root-zone through holes punched into the plastic film.

Fig. 1
figure 1

Layout of partial root-zone drip irrigation methods in field experiment

Full size image
Table 1 Details of irrigation treatment on cotton grown in the oasis field in 2004 and 2005
Full size table

Crop management

In 2004, the field was ploughed on April 16. Fertilizer [300 kg/ha of (NH4)2HPO4·CO(NH2)2 and 60 kg/ha of KCl] was applied and incorporated into the top 30 cm at ploughing. The soil was covered using plastic film before sowing to reduce evaporation loss. Cotton seeds (Gossypium hirsutum cv Xinluzao No. 7) were sown on April 21 and seedlings started to emerge on May 10. Each plot had eight rows of cotton. The seedling density was similar to the local cotton production with row spacing of 30 cm and plant spacing of 25 cm.

In 2005, the field was ploughed on April 14. Fertilizer [225 kg/ha of (NH4)2HPO4 and 450 kg/ha of Ca(H2PO4)2] was applied and mixed as in 2004. The same cotton variety seeds were sown on April 27 following the same planting practice as in 2004. All of the crop management, except for the irrigation volume, was the same in 2005 as in 2004.

Measurements

Meteorological data

Meteorological data were measured every 15 s and averaged hourly with a weather station (Hobo Weather Station, USA) which was 200 m away from the experimental plots. Meteorological variables measured included air temperature, relative humidity, global radiation, rainfall and wind speed at 2 m above ground. Reference crop evapotranspiration (ET0) calculated using the Penman–Monteith equation (Allen et al. 1998) and rainfall measured during the experimental period are presented in Table 2.

Table 2 Growth stages, accumulative temperature, effective precipitation and reference crop evapotranspiration (ET0) in the cotton field in 2004 and 2005
Full size table

Soil water measurement

To determine the unevenly distributed soil water content, two 1.2 m length PVC access tubes were installed in each plot as shown in Fig. 1. Soil water content was measured at 5 day intervals using a portable soil moisture monitoring system (Diviner 2000, Sentek Pty Ltd, Australia). The vertical profile of soil water content in every tube was determined from measurements of soil water content at 0.1 m intervals. Readings were taken through the wall of a PVC access tube. Data was collected from a network of access tubes installed at selected sites.

The gravimetric sampling technique and steel rings were used to calibrate the Diviner 2000, and the following calibration equation was used:

$$ {\text{SF}} = 0.2746 \cdot \theta ^{{0.3314}} + 0.9876 $$
(1)

where θ is volumetric soil water content (cm3/cm3), as determined by gravimetric sampling and bulk density of every 10 cm depth increment within the soil profile;SF is the scaled frequency (SF) which is calculated from the following equation:

$$ {\text{SF}} = {\left( {F_{{\text{A}}} - F_{{\text{S}}} } \right)}/{\left( {F_{{\text{A}}} - F_{{\text{W}}} } \right)} $$
(2)

where F A is the frequency reading in the access tube while suspended in air, F S is the reading in the access tube in soil at a particular depth and F W is the reading in the access tube in the water bath.

The approximate evapotranspiration (ET, in mm) of each plot was determined using the water-balance equation as follows:

$$ {\text{ET}} = P + I + W_{0} - W_{{\text{h}}} $$
(3)

where P is the rainfall in the growing period calculated as the sum of daily rainfall amounts greater than 5 mm inclusive (Table 1); I is the irrigation water amount (mm); W 0 and W h is the amount of soil moisture stored in 1 m depth at planting and harvesting (mm), respectively, based on the mean value from the two tubes in each plot.

Physiological measurements and sampling

Day to day changes in photosynthesis rate (P n), transpiration rate (T r) and stomatal conductance (g s) of six leaves per treatment were monitored with a portable photosynthesis system (ADC Bio-Scientific Ltd, UK) at 8:00–9:00 h local time from August 1 to August 4 in 2004 and July 5 to August 18 in 2005. Leaf WUE was calculated with the carbon gained per unit of water loss, equivalent to the ratio of P n to T r.

Leaf water potential was measured with a pressure chamber (3005 Plant Water Status Console, Soil Moisture Equipment Corp., Santa Barbara, CA, USA). Sunny days were chosen for monitoring the diurnal changes (every 2 h from 7:00 to 19:00) of leaf water potential in all treatments on August 30, 2004. On September 2, 2005, only treatment of 24 mm water amount per irrigation was measured for leaf water potential.

Plant vegetative growth was quantified by measuring the increment in plant height and leaf area index (LAI) at specific growth stages. Five tagged plants per treatment were randomly chosen for the measurement in 2004 and ten plants per treatment in 2005. LAI was measured with a PAR/LAI Ceptometer (AccuPAR model LP-80, Decagon Devices, Inc., Pullman, WA, USA) during the bolling stage in 2004, and the bloom flowering stage in 2005, respectively.

The seed cotton yield of each plot was monitored by hand harvesting the cotton twice before the frost and once after the frost in 2004 and 2005. All the harvested seed cotton was weighed for each plot as final yield in both years. The cotton water use efficiency (WUEET) and irrigation water use efficiency (WUEI) were calculated using the following formulas:

$$ {\text{WUE}}_{{{\text{ET}}}} = Y/{\text{ET}}_{{\text{c}}} $$
(4)
$$ {\text{WUE}}_{{\text{I}}} = Y/I $$
(5)

where Y is total seed cotton yield of each plot, ETc is the total evapotranspiration calculated from Eq. 3 during the cotton growing season, I is the irrigation volume applied to each plot, including the 75 mm irrigation amount before seeding, which was applied to wash away the salt.

Before harvesting, five plants per plot in 2004 and ten plants in 2005 were randomly chosen to measure the yield components, i.e., the percentage of boll in the shoot dry weight, boll dry weight per plant, fiber length, seed cotton per plant, lint per plant and fiber percentage. Three plants per plot were dug out from an area of 30 × 50 cm2 and a depth of 60 cm and the root and shoot growth and its ratio in both years were measured.

Data analysis

Data were statistically analyzed by a complete randomized model using Statistical Analysis Software (SAS software, SAS 6.12, SAS Institute Ltd, USA). All the treatment means were compared in the same column or row for any significant differences using the Duncan’s multiple range tests at significant level of 0.05.

Results

Soil water content

Temporal and spatial variations of soil water content at 0–60 cm soil profile are shown in Figs. 2 and 3. Soil water content in the two root-zones of the ADI treatment were different during the alternate wetting and drying cycles (Fig. 2a, b). Soil water content of the wetted side was higher than that of the drying side as a result of irrigation, but the soil water content was found to be relatively constant or increased slightly for several days after irrigation in the non-irrigated side. This may be caused by lateral infiltration or redistribution of water through the root systems. However, the soil water content in both sides of the root-zone of the CDI treatment was similar (Fig. 2c, d).

Fig. 2
figure 2

Spatial variation of soil water content at 0–60 cm soil profile after each watering (July 10, 2004)

Full size image
Fig. 3
figure 3

Temporal variations of soil water content at 0–60 cm soil profile of different root-zone drip irrigation in 2005. All data are averaged values of three plots of irrigation water applied at 24 mm per irrigation

Full size image

The temporal variations in soil water content of the wet and dry root-zones under ADI and CDI where 24 mm per irrigation was applied during the 2005 growing season are presented in Fig. 3. The results show that the soil water content in each ADI root-zone alternately increased and decreased.

Plant growth

As expected, the height of cotton plants were very sensitive to water stress applied by partial root-zone drip irrigation. Temporal variations in plant height for both ADI and CDI at different irrigation levels are shown in Fig. 4. The results showed that lower irrigation water application significantly reduced the plant height, and the cotton plant suffered severe water stress. With increased irrigation volumes, the plant height increased in both the ADI and CDI treatments. When irrigated with 15 mm per irrigation in 2004, plant height in ADI was significantly lower than that of CDI, suggesting more severe water stress occurred in ADI and restricted the growth of cotton plant. With the increase in irrigation amount, the plant height in ADI-2 treatments was closer to that of the CDI-2 treatment. However, the plant height in ADI-3 was greater than that of the CDI-3 treatment (irrigated with 30 mm per irrigation). The results also showed that when irrigated with 18 mm per irrigation during 2005, ADI plant height was significantly lower than that of the CDI treatment. No significant difference was found between the 12 and 24 mm per irrigation treatments but the final plant height of the ADI treatment irrigated with 24 mm per irrigation (measured on October 1) was significantly lower than that of the CDI treatment.

Fig. 4
figure 4

Heights of cotton plant grown in the field under all treatments in 2004 and 2005. All data are average values of five measurements in 2004 and ten in 2005 for each treatment. Vertical bars represent ±SE

Full size image

Leaf area index provided information on the cotton vegetative growth (Table 6). The results showed that ADI significantly inhibited the green leaf area per unit ground area during the bolling stage when irrigated with 15 mm per irrigation during 2004. LAI in the ADI treatment was significantly lower than that of CDI at the same irrigation level, but the difference was not significant when irrigated with 22.5 and 30 mm per irrigation. However, LAI measured during the 2005 bloom flowering stage showed that LAI in the ADI treatments were significantly lower than that of CDI at the same irrigation level, suggesting that ADI inhibited the vegetative growth.

Physiological responses

It was hypothesized that partial root-zone irrigation may reduce ‘luxury’ transpiration loss without reducing the photosynthesis rate by slightly limiting stomatal opening as has been found on maize grown in greenhouses (Wu et al. 1999; Zhang and Davies 1990). To prove whether this mechanism works on cotton grown under field conditions, day to day changes of photosynthesis rate (P n), transpiration rate (T r) and stomatal conductance (g s) were measured during the flowering and bolling stage in both years (Table 3). P n in ADI was not reduced significantly when compared to that of CDI at the same irrigation level. However, T r in ADI in four of the seven measurements was significantly lower than that of CDI at irrigation levels of 15 and 12 mm per irrigation (respectively in 2004 and 2005). T r in ADI in five measurements at irrigation levels of 30 and 24 mm per irrigation (respectively in 2004 and 2005) was also significantly lower that those of CDI. It should also be noted that g s in ADI in four measurements showed the similar changes as T r at irrigation levels of 15 and 22.5 mm in 2004, and 12 and 18 mm in 2005. As a result, WUE in ADI was significantly higher than that of CDI in the respective measurements at the three irrigation levels. Also such trend was especially evident in the mean values of the seven measurements.

Table 3 Photosynthesis rate (P n, μmol CO2 m− 2 s− 1), transpiration rate (T r, mmol H2O m− 2 s− 1), stomatal conductance (g s, mol m− 2 s− 1) and leaf water use efficiency (WUE, μmol CO2 mmol− 1 H2O) of all treatments in 2004 and 2005
Full size table

Diurnal variations of leaf water potential measured during the bolling stage (August 30, 2004) and the boll opening stage (September 2, 2005) showed that cotton grown under ADI or CDI had similar diurnal changes and that the differences between them were not significant if the same volume of irrigation was applied (Fig. 5). ADI did not lead to a leaf water deficit that might have contributed to growth and stomatal regulation.

Fig. 5
figure 5

Diurnal variations of leaf water potential of cotton plants subjected to different partial root-zone drip irrigations during the bolling stage (August 30, 2004) and boll opening stage (September 2, 2005). All data were average values of five leaves at the same position (third leaf youngest and mature from the top). Vertical bars represent ±SE

Full size image

Root and shoot response

Cotton root and shoot responses to different partial root-zone irrigation treatments are shown in Table 4. Generally, shoot dry weight is not significantly different between the two drip irrigations in 2004. ADI slightly influences growth of the above ground cotton plant, but root dry weight in ADI was less than that of CDI when irrigated with 15 or 22.5 mm per irrigation. However, this did not result in significant differences in the root–shoot ratio at the same irrigation level. Furthermore, no significant difference was found in root dry weight and root–shoot ratio during 2005. In general, there was also no significant difference in root density between the two drip irrigations in each year. On the other hand, more secondary roots were stimulated in ADI than in CDI when irrigated with 30 mm per irrigation during 2004 and 18 or 24 mm during 2005. Therefore, ADI generally produces more secondary roots compared with CDI at the same irrigation level, and this response is more significant with higher rates of irrigation water application. Apparently such response should enhance the absorptive capacity in ADI.

Table 4 Root and shoot response of cotton plants grown in the field under all treatments in 2004 and 2005
Full size table

Yield components, yield and water use efficiency

Yield components of cotton under different partial root-zone drip irrigations are shown in Table 5. Boll percentage in ADI treatments was higher than that of CDI when irrigated with 30 mm during 2004. ADI also had more bolls per plant than CDI when irrigated with 15 or 30 mm during 2004. Fiber length in ADI was much higher than that of CDI at the same irrigation level in 2005 and the difference was significant when the irrigation amount increased to 18 and 24 mm, suggesting higher fiber lint quality in ADI treatments. The results also show that ADI has more seed cotton per plant and lint per plant than that of CDI at the same irrigation level in both years. However, there was no significant difference in fiber percentage between the two drip irrigations presumably because the fiber percentage is mainly controlled by genetic characteristics.

Table 5 Yield components and harvest index of cotton grown in the field under all treatments in 2004 and 2005
Full size table

The harvest index (HI) for ADI treatments was highest when irrigated with 15 mm per irrigation. The HI in the ADI treatments were all significantly higher than that of CDI at the same irrigation level during 2004 (Table 5). However, HI decreased with increases in irrigation water applied for the same drip irrigation method in both years.

The surprising result is that the ADI treatment has notably increased cotton production when irrigated with the same amount of water (Table 6). Seed cotton yield in ADI was always better than that of CDI at the same irrigation level. The yield increased with increasing irrigation water application for both ADI or CDI treatments in both years. This suggests that ADI is a better method in increasing seed cotton yield than CDI. The highest seed cotton yield was obtained in ADI-3 (i.e., alternate partial drip irrigation with 30 mm per irrigation in 2004 or 24 mm in 2005). The percentage of pre-frost seed cotton yield (i.e., higher quality fibers for better price) under ADI was significantly higher than that of CDI when irrigated with 12 and 24 mm per irrigation during 2005 (Table 6). When compared to CDI, 9.8, 4.9 and 10.6% more pre-frost seed cotton was yielded with ADI applied at 12, 18 and 24 mm, respectively. The highest WUEET (calculated as seed cotton yield over total water use) was obtained in ADI-3 during 2004 and CDI-1 during 2005. On the other hand, the highest WUEI (calculated as seed cotton yield over total irrigation water) was obtained in ADI-2 during 2004 and ADI-1 during 2005.

Table 6 Leaf area index, seed cotton yield and its pre-frost percentage, crop evapotranspiration (ETc) and water use efficiency (WUE) in all treatments in 2004 and 2005
Full size table

Conventional drip irrigation-3 produced the largest canopy in 2004 but did not achieve the highest seed cotton yield (Tables 4, 6). By contrast, a moderate root–shoot ratio corresponded to the highest WUE in ADI-3. The lowest WUEET was found in CDI-1 when irrigated with 15 mm per irrigation during 2004. ADI significantly improved irrigation WUE with higher irrigation applications (i.e., 22.5 and 30 mm per irrigation during 2004 and 24 mm during 2005). However, the effect was not significant at lower irrigation levels in either year (Table 6).

Discussion

Water is almost always the limiting factor to plant growth and agricultural yield in arid areas. Traditional irrigation theory states that yield is sensitive to soil water stress, especially severe water stress. In this study, alternate partial root-zone drip irrigation maintained cotton yield when irrigation amount was reduced, and as a consequence, WUE was substantially improved. Furthermore, pre-frost seed cotton yield in ADI was higher than that of CDI (Table 6), indicating higher quality and a better product price. How could this be achieved? We believe the maintenance of water status in the shoots, better WUE due to reduced stomatal opening and a better balance between vegetative growth and reproductive growth in the cotton development under ADI condition is the explanation. Stomatal conductance and plant height were successfully controlled by the ADI treatment with less irrigation while plant water potential was maintained compared to those with more water.

Partial root-zone irrigation has been tested in recent years (Kang and Zhang 2004). In the same region, yield production and canopy WUE with alternate partial root-zone furrow irrigation were tested for irrigated maize (Kang et al. 1998). Results showed that alternate furrow irrigation (AFI) maintained high grain yield with up to 50% reduction in irrigation amount, while fixed furrow irrigation and conventional furrow irrigation (CFI) all showed a substantial decrease in yield with reduced irrigation. Furthermore, AFI under plastic mulch was also tested on cotton in Xinjiang, northwest China, where cotton almost totally relies on irrigation (Jia et al. 2003; Tang et al. 2005). AFI saved 30% water, gained 92% total seed cotton yield, but produced 12% more pre-frost seed cotton yield than that of CFI (Tang et al. 2005). Our results under drip irrigation also indicated that ADI gained 21 and 5% more total seed cotton yield than CDI at the same irrigation level in 2004 and 2005, respectively. On the other hand, 11% more pre-frost seed cotton yield in ADI was gained than that of CDI when irrigated with 24 mm per irrigation in 2005. Therefore, WUEET and WUEI in ADI were higher than that of CDI at the same irrigation level in both years. ADI-3 increased WUEET and WUEI by 18 and 21%, respectively, when compared to CDI-3 during 2004, and also 7 and 5% during 2005. The results also indicated that ADI could save 31% of irrigation water in 2004 and 33% in 2005 compared with the CDI treatment while producing the same yield. This indicates that alternate partial root-zone drip irrigation had the potential to save water and was a practical irrigation method in cotton production in arid areas.

In this experiment, soil water content was monitored for its temporal and spatial variations. The results showed that ADI provided a completely different soil moisture environment to the plant root system: approximately half of the root system was always in a drying state while the remainder was functional in terms of water extraction. The wetted and drying sides of the root system were alternated each irrigation. The root system of cotton plants under ADI always grew in a soil moisture environment where there was not only a temporal variation, but also a spatial ‘active’ controlled alternate drying and wetting. For the 30 mm per irrigation treatment during 2004, the main wetting layer was approximately 40 cm below the soil surface in both the ADI and CDI treatments. The soil water content of the wetted root-zone was also higher than that of the drying side, with an increase in soil moisture on the drying side as a result of lateral infiltration, or redistribution of water through the root systems.

A previous split-root experiment on maize indicated that partial root-zone drying could improve WUE without a significant decrease of photosynthetic rate and biomass (Kang et al. 1997). Experiments of alternate drying and wetting on maize (Wu et al. 1999) and cotton (our unpublished data) under greenhouse conditions has also indicated that APRI reduced water consumption and transpiration rate, but did not reduce photosynthetic rate significantly, indicating an increased WUE. Our results in this field experiment show that ADI reduces leaf stomatal conductance, which is in agreement with some earlier research on APRI (Davies et al. 2000; Kang and Cai 2002; Wu et al. 1999). In this study, both leaf photosynthetic rate and transpiration rate were also inhibited when stomatal conductance was restricted, indicating stomatal control of photosynthesis at that particular time.

Many earlier reports have shown that adequate water supply can increase the plant height, number of bolls per plant, boll weight and seed cotton yields (e.g., Pace et al. 1999; Xiao et al. 2000). This study has shown that the height of cotton plant was controlled by either reduced irrigation or ADI treatment, but the boll number and boll weight were unaffected by different irrigation methods. Control of vegetative growth and an appropriate balance between vegetative growth and reproductive growth have always been a problem in cotton production. Research on redundancy growth and its application in field management showed that appropriate reduction of redundancy growth might result in increased yield (Han et al. 2005). Our results show that although dry biomass of aboveground cotton plant in ADI was significantly lower than that of CDI, the seed cotton yield in ADI was significantly higher, suggesting that ADI may increase the translocation of photosynthetic product to bolls and reduce the redundancy growth of stem and leaves. Furthermore, the HI in ADI was higher than that of CDI at the same irrigation level, suggesting higher allocation efficiency of photosynthesis product to the economic yield.

Earlier studies showed that APRI had no significant yield reduction in grapevine (Gu et al. 2000) or only marginal yield reduction in tomato (Kirda 2004). Our early study on maize, pot-grown pepper, pear orchard and peach also showed that APRI was capable of increasing WUE with no significant reduction in crop yield (Kang and Cai 2002). In this study, seed cotton yield in ADI, especially before the frost, was higher than that of CDI with similar irrigation amounts, suggesting that alternate partial root-zone irrigation has the potential to improve both yield and WUE.

In summary, this experiment tested the hypothesis that partial root-zone irrigation improves the WUE at leaf and plot level. ADI showed good physiological responses in P n, T r, g s and WUE, which may provide a useful approach to apply the theory of root-to-shoot long distance signaling processes in field cotton production. An extra benefit from this ADI treatment was that the seed cotton yield and WUE were simultaneously enhanced when compared at the same irrigation level. When irrigated with 30 mm water per irrigation during 2004 and 24 mm per irrigation during 2005, seed cotton yield were both the highest in the ADI treatment. ADI under plastic film mulch has great potential for efficient water use and should be adopted for cotton production in arid areas where irrigation is essential and evaporation demand is high.