Drought-induced changes in rooting patterns and assimilate partitioning between root and shoot in upland rice Folkard Ascha,*, Michael Dingkuhnb, Abdoulaye Sowc, Alain Audebertb aUniversitat �� Bonn, Institut fur �� Pflanzenernahrung, �� Karlrobert-Kreiten-Str. 13, D-53115 Bonn, Germany bCIRAD-amis, Programme Agronomie, BP 5035, 34032 Montpellier Cedex 1, France cThe Africa Rice Center (WARDA), BP 96, St. Louis 01, Senegal Received 21 February 2004 received in revised form 24 September 2004 accepted 8 October 2004 Abstract Drought is a major stress affecting rainfed rice systems. Root characteristics such as root length density, root thickness, and rooting depth and distribution have been established as constituting factors of drought resistance. Deep rooting cultivars are more resistant to drought than those with shallow root systems. The present study sought to quantify the effects of different levels of drought on dry matter partitioning and root development of three rice cultivars CG14 (Oryza glaberrima), WAB56-104 (O. sativa tropical japonica, improved) and WAB450-24-3-2-P18-HB (CG14 WAB56-104 hybrid). Two experiments on assimilate partitioning under different levels of drought stress were conducted under rain shelters at the West Africa Rice Development Association, Mbe, Ivory Coast. PVC tubes (diameter = 0.2 m, height 0.6 m) containing about 25 kg of sandy loam were used for the drought stress experiments. For rooting depth and root distribution studies, the tubes were subdivided into four compartments of 0.15 m each. In the first trial, tubes with WAB56-104 were gradually droughted to five levels of soil moisture content that were kept constant thereafter. In the second trial, plants of all cultivars were subjected to three drought treatments: (1) constant soil moisture content at field capacity (about 22% moisture content), (2) constant soil moisture content of 14% (about 0.5 MPa soil matrix potential) and (3) constant soil moisture content of 9% (about 1 MPa soil matrix potential). Rice reacted to drought stress with reductions in height, leaf area and biomass production, tiller abortion, changes in root dry matter and rooting depth and a delay in reproductive development. Assimilate partitioning between root and shoot, determined from changes in dry matter, was not affected by drought when the plants were gradually stressed. In no case, additional biomass was partitioned to the roots on the contrary, dry matter partitioning to the root completely ceased under severe stress. Due to the irrigation technique used, vertical soil moisture distribution varied little, but roots grew deeper under drought stress. This was particularly the case for the upland adapted WAB56-104. Implication for modeling of drought responses in upland rice systems is discussed. # 2004 Elsevier B.V. All rights reserved. Keywords: Dry matter partitioning Soil moisture content Root distribution Root���shoot relationships www.elsevier.com/locate/fcr Field Crops Research 93 (2005) 223���236 * Corresponding author. Tel.: +49 228 731678 fax: +49 228 732489. E-mail address: fa@uni-bonn.de (F. Asch). 0378-4290/$ ��� see front matter # 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.fcr.2004.10.002

1. Introduction Upland rice with about 2 Mha land under cultiva- tion is one of West Africa���s most important crops. Drought is a major abiotic stress in rainfed rice systems and recent research efforts focused on improving the performance of rice under water- limited conditions (Dingkuhn et al., in press). Studies in drought-prone environments have shown that the African indigenous species, Oryza glaberrima, per- forms better under drought, than the more widely grown O. sativa japonica cultivars (Dingkuhn et al., 1999 Yeo et al., 1997). Advances in interspecific hybridization between O. glaberrima and O. sativa provide breeders with new opportunities to develop drought-resistant cultivars for rainfed systems in sub- Saharan Africa (Asch et al., 1999 Dingkuhn et al., 1998 Jones et al., 1997). At the same time, rice growth models are being developed considering system- inherent competition for water, nutrients and light to support management options for rice-based upland systems (Dingkuhn et al., 2003). Root characteristics such as root length density, root thickness, and rooting depth and distribution have been established as constituting factors of drought resistance (Minabe,1951 MatsuiandSingh,2003).Rootgrowthis an important component of the adaptation of rice to drought-prone environments, and the size of the root system varies substantially between the different ecotypes of rice (Nemoto et al., 1998 Price et al., 1997). Tropical japonica types have been reported to have larger root systems than indica types (Ingram etal., 1994). Not only the size of the root system, but also the distribution of the root biomass or length in the soil profilewas shown to be important for drought resistance (Kondo et al., 2003 Price et al., 2002). Deep-rooting cultivars are more resistant to drought than shallow- rooting ones (Minabe, 1951 Nemoto et al., 1998) because of the plant���s ability to exploit water retained in the deep soil layers (Kondo et al., 2003 Yoshida and Hasegawa, 1982). Characteristics of plants tolerant or resistant to drought are thus well described, but to what extent can a given plant adapt its rooting pattern during drought (e.g., through changes in partitioning towards the root) to increase its drought resistance? It has been shown that drought effects on partitioning of assimilates to the roots differ among the gramineae. For example, in barley drought stress had no effect on assimilate partitioning between root and shoot (Shone et al., 1983), whereas in droughted Panicum maximum assimilate partitioning towards the root was increased (Tsiung, 1975). In sorghum with increasing drought stress, less assimilate was parti- tioned to the root (Rice, 1979), and in rice osmotic stress induced by PEG (polyethyleneglycol) increased the assimilate partitioning towards the root, whereas low humidity stress increased dry matter partitioning to the shoots (Hirai et al., 1994). In rainfed lowland rice, assimilate partitioning to the root system decreased but the share of assimilates attributed to deeper layers increased (Azhiri-Sigari et al., 2000). Rice growth models for rainfed conditions need to take into account chances in partitioning due to moisture availability in the top soil, in order to simulate correctly crop growth, competition with weeds for water, light and nutrients, and potential yield. For this, it is important to know, if drought changes the plants partitioning patterns between root and shoot in favor of the root or in favor of the shoot and any changes in rooting depth need to be taken into account as well. The present study sought to quantify the effects of different levels of topsoil moisture availability on dry matter partitioning and root development for O. glaberrima and O. sativa cultivars and their inter- specific progenies. We tried to differentiate between two situations: (1) constant low, medium or high moisture availability (different agro-ecological envir- onments) and (2) progressive loss of moisture availability down to low medium or high levels (seasonally occurring drought spells of different severity). 2. Materials and methods Two experiments on assimilate partitioning under different levels of soil moisture availability were conducted in a screenhouse at WARDA���s research station in Mbe, �� Cote �� d���Ivoire (78520N, 5860W, 300 m asl). Mean climatic conditions in the screenhouse were 27.2 8C mean air temperature, 12.2 MJ day 1 solar radiation and 70% rH for the first and 27.5 8C mean air temperature, 9.8 MJ day 1 solar radiation and 79% rH for the second experiment. F. Asch et al. / Field Crops Research 93 (2005) 223���236 224