Effect of flood and drip irrigation and difference of previous crop residue input on morphological and physiological traits in rice root

ABSTRACT Detailed information on the root system under drip irrigation will help in more efficient irrigation scheduling. Although input of previous crop residues is effective in increasing soil fertility, little is known about its effect on the root system. This study aimed to investigate the effects of irrigation methods (drip or flood) and different management of previous crop residues on root morphological characteristics and its distribution and physiological functions of the rice cultivar ‘Hinohikari’. Experiments were conducted in the paddy field on the university farm (input previous crop residues every year) for both drip and flood irrigation and the paddy and upland field (no input previous crop residues) on the campus of the faculty under flood and drip irrigation. Plant root was sampled with a core sampler (ø 5.5 cm × 30 cm), and root length and surface area were determined using image analysis. Drip irrigation did not change root length compared to flood irrigation, but the root system was heavier and individual root thickness tended to increase compared to flood irrigation. The root depth index was significantly deeper in drip irrigation than in flood irrigation. The bleeding rates were significantly higher in flood irrigation than in drip irrigation. Applying previous crop residues significantly improve root length mainly at the upper soil layers. The mechanisms for enhancing root growth by applying previous crop residues need to be explored individually in flood and drip irrigation, in which physical and chemical properties in soil are largely different. Graphical Abstract


Introduction
Optimization of irrigation water is an essential issue in agricultural production for maximizing the return from the limited water availability (Soundharajan & Sudheer, 2009). Lowland rice with continuous flooding is the most water-demanding of all cereal crops: it requires about 3,000 liters of water per kilogram of grain produced from flooded fields (Bouman et al., 2007). The drip irrigation technology saves water by 32% while maintaining a similar grain yield to puddled transplanting (Rajwade et al., 2018a(Rajwade et al., , 2018b and reduces the cost of irrigation water in the cultivation of rice by 2.0-5.6 times in comparison with the traditional flood rice (Kruzhilin et al., 2017). Kato and Okami (2010) pointed out that vigorous root growth is a prerequisite for soil water uptake and maintaining transpiration in aerobic rice culture. The yield reduction in the lowland cultivars in aerobic rice systems might be due to changes in root activity or function as well as root morphology (Matsuo & Mochizuki, 2009). These points are even more important in drip irrigation, which shows a characteristic distribution of soil moisture compared to other irrigation methods. The wetted region of soil under drip irrigation has a threedimensional distribution when viewed from a single emitter, and the horizontal/vertical ratio of the wetted region increases with higher discharge of water, poordrained, and fine-textured soil (Brouwer et al., 1988;Skaggs et al., 2010). The wetted region should match the root zone for effective use of water. Therefore, information on the root system structure is essential for efficient irrigation scheduling in drip irrigation. Some studies have been reported on the root system of rice under drip irrigation, but there are some differences depending on the method of drip irrigation H. B. He et al., 2014;Henry et al., 2012Henry et al., , 2016Wei et al., 2018). Further information is needed on the rice root system by drip irrigation under various environmental conditions. Furthermore, the information on the effects of previous crop residues on root elongation is limited. The recent decline in demand for staple rice and the fall in rice prices in Japan have led to the crop rotation system with diverse crops such as soybean and wheat in upland converted fields from paddy fields. While input of crop residues as a source of organic matter is effective in improving the soil fertility of upland converted field (Ichita et al., 1983), it is known that crop residues plowing inhibits the early growth of succeeding crops. Tanaka (2001) pointed out that the amount of inorganic nitrogen in the soil is not the main cause of the growth inhibition, but aromatic acids are also responsible. However, to our knowledge, there has been no report about the effect of previous crop residue on the root system of succeeding crops in Japan. On the other hand, the beneficial effect of crop residues on pearl millet on acid sandy soil in the West African semi-arid tropics was due to the improvement of phosphorus nutrition by both increase in P mobility in the soil and enhancement of root growth (Hafner et al., 1993;Kretzschmar et al., 1991).
The main objective of the present study was to investigate the effects of irrigation methods (drip or flood) and different management of previous crop residues on morphological characteristics, distribution, and physiological functions of roots of the Japanese rice cultivar 'Hinohikari'.

Location, plant materials, and management
Field experiments were carried out on two sites, inside the campus (34°16'N, 134°7'E, elevation of 21 m) and the University farm (34°16'N, 134°9'E, elevation of 35 m) of the Faculty of Agriculture, Kagawa University, Japan. The strait-line distance between the campus and the farm is 3.8 km. In both sites, all field soils are classified as clay loam. We used two irrigation methods, surface drip irrigation and flood irrigation for both sites. Japonica rice cultivar 'Hinohikari' was sown in a pot tray (Minoru Pot 448, Minoru Industrial Co., Ltd, Okayama, Japan) on 30 May 2019 and transplanted on 20 June for all sites and irrigation plots.
The paddy field on farm used in this experiment has been used for crop rotation of paddy rice and wheat for many years. The preceding crop, wheat, was harvested on 29 May with a combined harvester, and the straw was cut into 5-cm pieces and sprayed on the field together with crop residues. The field was rotary tillage and puddled, and rice seedlings were transplanted with a line spacing of 33 cm and a row spacing of 22 cm (13.8 hills m −2 ) with a rice planting machine. We applied herbicide (Naginata jumbo, Kumiai Chemical Industry Co., Ltd, Tokyo, Japan) after transplant and maintained flood conditions for 10 days to activate the herbicide. After that, we divided the field into two plots, a flood plot (480 m 2 ) and a drip plot (320 m 2 ), with plastic boards inserted to 30 cm depth. After draining water in the drip plot, we installed drip tubes with emitters every 30 cm (DripNet PC AS16, Netafim Japan Ltd, Tokyo, Japan) at 66-cm intervals. We determined the daily irrigation amount for the drip plot based on the daily reference evapotranspiration measured using an atmometer (ETgage, Model A, ETgage Company, Colorado, USA) installed in the drip plot. Almost the same amount of the average daily reference evapotranspiration over the past few days was irrigated at 8:00 h every morning. Irrigation was suspended on rainy days and for several days after heavy rain. We maintained flood condition in flood plot throughout the growth. For the flood plot, we applied a mix of 30% fast-acting and 70% controlledreleased coated fertilizer which is the mix of 100 and 120 days releasing (Sanukino kome ippatsu, Asahi Hiryou Ltd, Kagawa, Japan) at 70 kg ha −1 of N and 42 kg ha −1 of P 2 O 5 and K 2 O before puddling. For the drip plot, we applied a mix of urea and ammonium phosphate sulfide (JCAM Agri Ltd, Tokyo, Japan) at 30 kg ha −1 of N, P 2 O 5 , and K 2 O before puddling as a base fertilizer. We applied supplement fertilizer by fertigation for the drip plot. Liquid fertilizer, Kumiai No.1 (Sumitomo Chemical Ltd, Tokyo, Japan), was applied five times from 12 July to 7 August at a rate of 15 kg ha −1 of N, 6.3 kg ha −1 of P 2 O 5 , and 8.8 kg ha −1 of K 2 O. Furthermore, 15 kg ha −1 of N was applied two times in August with ammonium sulfate by fertigation. The total N input in drip and flood on farm was 135 kg ha −1 and 70 kg ha −1 , respectively.
Regarding the fields on campus, only paddy rice has been cultivated in the paddy field for many years. In contrast, different field crops such as soybean, wheat, and barley have been cultivated in the upland field depending on the year. The preceding crop of the paddy and upland fields of campus was rice and soybean, respectively, which were harvested in the fall of 2018. Because we harvested the aboveground part of these crops and took them out of the field, we did not put these crop residues into the field except for the stump and underground part. After rotary tillage, we paddled the paddy field of campus (200 m 2 ) and transplanted rice seedlings by hand with a line spacing of 30 cm and a row spacing of 20 cm (16.7 hills m −2 ). We applied the mix of compound fertilizer and controlledreleased coated fertilizer of 100 days releasing (Koshi ippatsu, Asahi Hiryou Co., Ltd, Kagawa, Japan) before puddling at 80 kg ha −1 of N, P 2 O 5 , and K 2 O. We maintained the flood plot constantly flooded throughout the cultivation. The upland field of campus (150 m 2 ) was used for drip plot where the same drip tubes used in the farm were installed at 60-cm intervals. We transplanted rice seedlings by hand with a line spacing of 30 cm and a row spacing of 20 cm (16.7 hills m −2 ). Daily irrigation for drip in the campus had been done the same way on the farm using the ETgage atmometer installed in the drip plot on the campus. A mix of ammonium sulfate and a fertilizer for hydroponic culture (OAT House No. 9, OAT Agrio Co., Ltd, Tokyo, Japan) was applied by fertigation once a week after transplant at a rate of 10 kg ha −1 of N and P 2 O 5 and 6.5 kg ha −1 of K 2 O for a total of 8 times. The total N input in drip and flood on campus was 80 kg ha −1 .

Measurement
Air temperature, solar radiation, and rainfall were measured at the meteorological station of the Faculty of Agriculture, Kagawa University, which is located adjacent to the upland field on campus.
We collected soil and crop residues up to 20 cm in depth in a 30 cm square area using garden trowel after rotary tillage before transplant to estimate the amount of previous crop residues. Soil and crop residues were separated by hand after air drying and weighed to calculate the weight percentages of crop residues contained in the soil. Based on these values and the soil bulk density, we estimated the amount of crop residues contained in soil up to 20 cm depth in 1 m 2 square.
We set up three replication blocks for each plot. We counted the tiller number including main stem of eight hills for each replication block and selected four hills with average tiller numbers to measure the bleeding rate (Morita & Abe, 1999), dry weight, and root morphological traits on the same dates of heading and maturity. The dates of heading for flood were 4 September and 3 September and for drip were 12 September and 11 September on farm and campus, respectively. The dates of maturity for flood were 10 October and 9 October and for drip were 16 October and 15 October on farm and campus, respectively. Starting from 9:00 h to 10:00 h, we cut shoots at approximately 10 cm from the soil surface, put cotton on the cut surface, wrapped with polyethylene film, and sealed with a vinyl band. The cotton, film, and vinyl band were removed after 1 h and immediately weighted to quantify the bleeding rate. The cotton, film, and vinyl band used for each hill were weighed before use in the field. The average temperature of the beginning and the end of bleeding measurement at 5 cm below the soil surface was measured using a digital thermometer (CT-419WP; Custom Corp., Tokyo, Japan). Leaf greenness of the fully expanded leaves was measured using a chlorophyll meter (SPAD-502; Konica Minolta Inc., Tokyo, Japan). Aboveground parts were oven-dried at 80°C for 72 h to determine the dry weight. We took one soil core (ø 5.5 cm × 30 cm) from just below the hill and one from between the hills, for each replication block, using a soil sampler (DIK-110C, Daiki Rika Kogyo, Saitama, Japan). We divided soil cores into 0-10 cm, 10-20 cm, and 20-30 cm layers and washed them carefully on a sieve to separate the roots. We took root images at a resolution of 300 dpi using a flatbed scanner equipped with a transparent manuscript device. A macro program for root length measurement [Root Length 1.8win, Kimura and Yamasaki (2003)] developed by Kimura et al. (1999) and Yamasaki (2001, 2003) was run on an image analysis software (Scion Image, Scion Corporation) to measure the root length and root surface area. The root dry weight was measured after the samples were oven-dried at 80°C for 72 h. The root length, root surface area, and root dry weight of the three layers from two soil cores were pooled, and root length density (cm cm −3 ), root surface density (cm 2 cm −3 ), root weight density (mg cm −3 ), and specific root length (m g −1 ) were calculated in each replication block. Specific root length represents the thickness of root: the longer the specific root length, the smaller the root diameter (Banba & Ohkubo, 1981;Comas et al., 2013;Henry et al., 2012;Morita & Abe, 2014). Furthermore, we analyzed the differences in root length density and specific root length by irrigation and site in each soil layer. The root depth index (cm), which indicates an average root depth (Oyanagi et al., 1993), was calculated from the percentages of root length density in each soil layer to the total root length density of the soil surface to 30 cm depth.

Statistical analysis
Data were analyzed with two-way ANOVA using the model for a split-plot design with three replications to evaluate the effects of irrigation, site, and their interaction on all measured variables, except for the data of previous crop residues in the soil. Differences in previous crop residues in the soil were determined using ANOVA and Tukey's HSD test. Regression analysis between soil temperature and bleeding rate was carried out to evaluate the effect of temperature on bleeding rate. All statistical analyses were performed using JMP statistical software (SAS Institute Japan, Tokyo, Japan). Figure 1 shows meteorological data at the campus. The mean daily temperature increased from 19.7°C in May to 27.4°C in August and then decreased to 19.8°C in October. Mean daily temperatures in July and August were below long-term average while in September and October were above long-term average. The total of monthly rainfall was 628 mm, which is 135 mm below long-term average, with high rainfall in July and low rainfall in June and September. The mean daily solar radiation gradually decreased from 22.5 MJ m -2 d -1 in May to 11.0 MJ m -2 d -1 in October and was lower than   long-term average in July and August. Meteorological condition on farm would be similar to that of campus considering the 3.8 km strait-line distance and 14 m difference in elevation. Table 1 shows weight percentages of previous crop residues in the soil, soil bulk density, and estimated amount of crop residues in soil up to 20 cm depth in 1 m 2 square. The paddy field on farm had previous crop residues 5.1 to 7.0 times more than those in the upland and paddy fields on campus, respectively. Soil bulk densities ranged from 0.88 in the paddy on farm and 1.11 in the upland field on campus. The amount of previous  crop residues in the paddy field on farm was 4.3 to 5.4 times higher than those in paddy and upland field on campus, respectively. Table 2 shows the number of tillers, shoot dry weight, panicle dry weight, and SPAD value at heading and maturity. ANOVA detected the significant effect of site on the number of tillers at heading. The number of tillers on campus at heading was larger than on farm. The effect of irrigation was significant in shoot dry weight and SPAD value at heading. The shoot dry weight was heavier in drip than in flood in both sites, but the SPAD value was lower in drip than in flood in both sites. Table 3 shows the root length density, root surface density, and root weight density at heading and maturity. The significant effects of site were detected on root length density at heading and the root surface density at heading and maturity. The root length density and root surface density were higher on the farm than on campus. On the other hand, the effect of irrigation and interaction on root weight density at heading were both significant. The root weight densitywas heavier in drip than in flood in both sites. Table 4 shows the bleeding rate, specific root length, and root depth index at heading and maturity. The effect of irrigation on the bleeding rate was significant both at heading and maturity. The bleeding rates were higher in flood than in drip in both sites and growth stages. The effect of site, irrigation, and interaction on the specific root length was all significant at heading, while that of irrigation was only significant at maturity. The specific root length is longer in flood than in drip on farm but vice versa on campus at heading. The specific root length on farm is longer than on campus both at heading and at maturity. The effect of site and irrigation on the root depth index was significant at heading. The root depth index was deeper in drip and campus than in flood and farm. Figure 2 shows the relationship between soil temperature at 5 cm below soil surface and bleeding rate. There were no significant relationships between soil temperature and bleeding rate at both heading and maturity.

Results
Since the root length density and specific root length significantly differed depending on irrigation and site, we examined in which soil layer those differences occurred (Figures 3 and 4). Table 5 shows the results of ANOVA for the effect of site, irrigation, and interaction on the root length density and the specific root length in each layer. The effect of site on the root length density was significant in 0-10 cm and 10-20 cm soil layers at heading (Table 5). The root length density on the farm was higher than on campus in 0-10 cm and 10-20 cm soil layers at heading (Figure 3). The effects of site, irrigation, and interaction on the specific root length were all significant in 0-10 cm layer at heading. At maturity, on the other hand, the effect of irrigation was significant in 0-10 cm layer and that of site was significant in 10-20 cm layer (Table 5). The specific root length in 0-10 cm layer was longer on farm and flood than on campus and drip, respectively, at heading. At maturity, the specific root length in 0-10 cm layer was longer in flood than in drip and that in 10-20 cm layer was longer on campus than on farm (Figure 4).

Effects of drip irrigation
The effect of irrigation was not significant on the root length density and root surface density, but the root weight density at heading was significantly heavier in drip than in flood (Table 3). Moreover, the specific root length of the whole root system was significantly shorter in drip than in flood at heading and maturity stages except that on campus at heading (Table 4). When compared by soil layer, the same results were observed at the upper soil layer (Figure 2 and Table 5). Thus, the effect of irrigation was significant on root weight density and specific root length but not significant on root length density and root surface density. These results suggest that drip irrigation did not change root length but that the root system was heavier and individual root thickness tended to increase.
The root depth index indicates the average depth of the root (Oyanagi et al., 1993) or the center of gravity of the root system (Morita et al., 1995). This index also represents crop adaptation to external environmental stress (Oyanagi et al., 1993). The root depth index at heading was significantly deeper in drip than in flood in both sites (Table 4). These results are consistent with H. He et al. (2013), who reported that more roots gathered in deep-soil layer in drip irrigation than in conventional flooding. In contrast, Banba and Ohkubo (1981) concluded that the root system distribution is strongly influenced by soil moisture and that the root system distribution is deeper if soil moisture is low and shallower if soil moisture is high. In our experiment, the horizontal/vertical ratio of soil moisture in root zone under flood irrigation would be 1, while it under drip irrigation with drip tubes placed at 60-or 66-cm intervals is estimated to be less than 1. Therefore, drip would have a deeper root system because the roots grow vertically to absorb water, which likely be more abundant vertically than horizontally. Coelho and Or (1999) reported good agreement of the distribution of the root system of corn with the distribution of soil water content under drip irrigation and the pattern of soil water uptake by roots.
The bleeding rate is an expression of root vigor (Kusutani et al., 2000;Yamaguchi et al., 1995). The bleeding rates were significantly higher in flood than in drip in both sites and growth stages (Table 4). A positive correlation between bleeding rate and soil temperature was reported (Yamaguchi et al., 1995), but our results showed no relationship between bleeding rate and soil temperature ( Figure 2). Although the soil moisture in drip at the time of bleeding measurement is unknown, it should be lower than that in flood. Therefore, the root vigor per se or the difference in soil moisture likely affected the bleeding rate. It seems necessary to examine whether the difference in the bleeding rate when the soil moisture is different represents the difference in root vigor.
SPAD values were significantly lower in drip than in flood at the heading stage in both sites (Table 2). A similar trend was observed at maturity although the difference was not significant. A possible reason for the lower SPAD value in drip than in flood is drought stress. Sun et al. (2004) concluded that drought stress significantly reduced total chlorophyll in plants and that the higher the stress level, the greater the reduction. In our results, however, the rice in drip was more likely to experience drought stress than in flood, but there should have been no severe drought stress judging from the heavier shoot dry weight at heading (Table 2). Another possible reason for the lower SPAD value in drip than in flood may be nitrogen availability in soil. Since the soil in drip was aerobic, ammonium nitrogen was nitrified and converted to nitrate nitrogen by soil microorganisms. Nitrate nitrogen might be leached to the lower layer with soil water or released into the atmosphere as nitrous oxide and nitrogen gas after being denitrified by soil microorganisms under anaerobic conditions. Furthermore, we applied 75 kg ha −1 of nitrogen with a liquid fertilizer (Kumiai No.1) to the drip on farm, of which 10.5 kg ha −1 was ammonium nitrogen and the rest was nitrate nitrogen. Since rice is a crop that prefers NH 4 + as a source of nitrogen (Wang et al., 1993), nitrate nitrogen applied to the drip on farm was poorly utilized and leached or denitrified. Despite applying almost twice as much nitrogen fertilizer as flood on farm, the significantly lower SPAD values in drip support this speculation.

Effects of previous crop residues
Before we discuss the effect of previous crop residues, we would like to consider the effect of different planting density between farm and campus. Although the number of tillers per m 2 was significantly larger in campus than in farm (Table 2), the effect of site on the number of tillers per hill (data not shown) and the dry weight of shoot and panicle were not significant (Table 2). These results suggested that the difference in planting density between farm and campus hardly affected on aboveground traits.
Our results showed the positive effect of previous crop residues on root growth. Namely, root length density and root surface density on farm were significantly higher than on campus (Table 3). When compared by soil layer, the root length density was significantly greater on farm than on campus in 0-10 cm and 10-20 cm soil layers at heading and the same trends were observed at maturity although the effect of site was not significant (Figure 3 and Table 5). On the other hand, there was no consistent trend in the effect of site on specific root length. The specific root length of flood on farm in the 0-10 cm soil layer was significantly greater than that of flood on campus, whereas that of both flood and drip on campus in the 10-20 cm soil layers was significantly greater than those on farm ( Figure 4 and Table 5). Kretzschmar et al. (1991) showed in pot experiment that crop residues (incubation of millet straw) increased the early root growth of pearl millet due to increased lateral root growth. Hafner et al. (1993) showed that crop residue increased root length density in pearl millet, especially in the topsoil at the earlier growth stages. Our results also showed an increase in root length density in the upper layers on farm but no effect on specific root length. A decrease in specific root length, i.e. a decrease in average diameter of roots, may associate with the increase of the percentage of lateral roots. Therefore, our results did not show the positive effect of previous crop residues on lateral root growth. A significantly shallower root depth index on farm than on campus (Table 4) may support the positive effects of crop residues on root growth in the upper soil layers. Hafner et al. (1993) also suggested that the beneficial effects of crop residues on P uptake were primarily attributed to higher P mobility in the soil due to decreased concentrations of exchangeable aluminum and enhancement of root growth. Since these mechanisms cannot be directly applied to our results, the mechanisms for the enhancement of root growth by applying previous crop residues are explored individually in flood and drip irrigation, in which physical and chemical properties in soil are largely different.

Conclusion
Drip irrigation did not change root length compared to flood irrigation, but the root system was heavier and individual root thickness tended to increase compared to flood irrigation. The root depth index in drip irrigation was significantly deeper than in flood irrigation. The possible reason may be differences in soil moisture between drip and flood irrigation or the specific movement and distribution of water from the drip tube emitter. The bleeding rates were significantly higher in flood than in drip. However, it is necessary to verify whether this difference reflects root vigor per se. Significantly lower SPAD values in drip than in flood suggest nitrogen loss by leaching or release into the atmosphere.
Applying previous crop residues significantly improve root length mainly at the upper soil layers. The mechanisms for enhancing root growth by applying previous crop residues need to be explored individually in flood and drip irrigation, in which physical and chemical properties in soil are largely different.