Water regimes on the development of accessions of the Manihot genus

ABSTRACT The objective of this work was to select water deficittolerant accessions of the genus Manihot, through morphological characters under different water regimes. The experiment was conducted in a randomized block design, using a split-plot arrangement with plots consisting of tree water regimes (simulated rainfall and water depth of 100 and 20% crop evapotranspiration (ETc)), and subplots consisting of eight accessions of the genus Manihot (Gema-de-Ovo and Engana-Ladrão from the species M. esculenta Crantz, and BGMS-115, BGMS-110, BGMS-102, BGMS- 79, BGMS-24, and BGMS-48 from Manihot sp.). The accessions were evaluated considering two crop cycles: the first had 120 and 60 days from the application of the treatments. Plant height, stem diameter, number of leaves, leaf lobe length and shoot dry mass production were evaluated. For each cropping cycle, a split-plot analysis of variance was performed. The highest genotypic means were expressed by the accessions BGMS-115, BGMS-102, BGMS- 79 and BGMS-24 for most of the analyzed variables, regardless of the cultivation cycle. For the characteristic shoot dry mass production, accessions BGMS-102 and BGMS-79 showed the best performances under conditions of limited water regime (20% ETc), regardless of the cropping cycle. Accession BGMS-102 was also grouped in the group with the highest genotypic means, for this trait, in treatments with rain simulation and 100% ETc, in the first cycle, demonstrating that, under stress conditions, this accession is an option to tolerate low water precipitation and responds well when higher precipitation occurs.


INTRODUCTION
The agricultural sector is highly dependent on water for irrigation to meet crop requirements, and occurrence of severe droughts has resulted in large socioeconomic losses due to decreased food production (ZHAO et al., 2017). According to Zhang, Mu and Huang (2016), agricultural production in general increased in recent years, but droughts are the main cause of crop failure, resulting in global instability of food prices and threatening food security. The United States of America estimated losses of almost 30 billion dollars in 2012 from direct losses of agricultural production due to droughts, plus 5 billion when considering livestock and dairy products (ELLIOTT et al., 2018).
Water deficit reduces growth and yield of plants, lowering their productive capacity. However, depending on the intensity and duration of the water stress, plants can develop molecular, cellular, biochemical, and physiological adaptive processes to improve their survival in these environments (PUTPEERAWIT et al., 2017). According to Merwad, Desoky and Rady (2018), responses of plants to water deficit are complex, since they have several mechanisms to grow under low water availability conditions. Identifying and selecting water deficit-tolerant genotypes that can maintain their productive capacity by efficiently using the available water is important for establishing crops in regions with low water availability (MANSOUR et al., 2017). Thus, researchers use different procedures to characterize and identify these genotypes, especially selection by morphological descriptors. Several plant characteristics are used to identify water deficit-tolerant genotypes for different crops, including cassava (M. esculenta Crantz), a commercial relative of the species discussed in this article, such as plant height, leaf lobe length and width, number of leaves (OKOGBENIN et al., 2013), leaf area (CAYÓN; EL-SHARKAWY; CADAVID, 1997) and variables related to vegetative growth (VALE et al., 2012).
According to Ferreira et al. (2009), the aerial part of Euphorbiaceae species can be an alternative to increase the economic viability and productivity of livestock in semi-arid regions, both qualitatively and quantitatively, during the most critical period of the year, that is, in the period of lower precipitation, since it has high nutritional value and good acceptability by animals. In this context, species of the genus Manihot, which are already widely used in agriculture, such as cassava, as well as their wild relatives known as maniçobas (M. glaziovii, M. catingae, M. esculenta ssp. flabellifolia and M. carthaginensis) used in animal feed, are alternatives to increase productivity levels, which are low in most areas with herds in the northeastern semi-arid region, mainly for small ruminants, mainly due to the fact that the feed is based, almost exclusively, on the Caatinga vegetation, which has its forage resources exploited in an extractive way and has low support capacity, with low animal yield (0.08 AU/ha/year) and low productivity (6-8 kg of weight gain/ha/year) (GUIMARÃES FILHO SOARES;RICHÉ, 1995).
In addition, these species differ in terms of resistance to pests and diseases, tolerance to successive cuts and leaf retention, and some of them are arboreal while others are herbaceous, which allows the selection of the most promising depending on the purpose of the breeding program. Although species of this genus are adapted to shallow, low-fertility marginal soils and irregular rainfall conditions and maintain good biomass production even under these conditions, the challenges posed by global climate change (increased temperature and drought severity) make the search for genotypes more tolerant to water deficit, as well as the understanding of its effects on plants, constant in breeding programs.
Therefore, this study aimed to evaluate the effect of different water regimes on the development of accessions of the genus Manihot.

MATERIAL AND METHODS
The experiment was conducted at the Caatinga Experimental Field, of the Embrapa Semiárido, in Petrolina, PE, Brazil (09°03'25''S, 40°28'95''W, and 395 m of altitude). The climate of the region is BSwh', Semi-arid, with high temperatures, according to the Köppen's classification, climatological data during the experimental period can be seen in Figure 1. During the application of the experimental treatments there was no rainfall. The soil of the experimental area is classified as Neossolo Quartzarênico (Entisol) (SANTOS et al., 2018). The soil of the experimental area, whose chemical and physical characteristics are described in Table 1, was prepared according to procedures commonly used by farmers in the Semi-arid region of the state of Pernambuco, Brazil. No chemical fertilization had been applied prior to the experiment. The initial preparation of the soil consisted of one plowing and two harrowing/leveling operations.   from Manihot sp. These accessions were previously selected, as they had better seedling establishment, better forage potential (such as leaf retention and higher crude protein content) and, in the case of M. esculenta accessions, showed greater tolerance to water stress in other works (OLIVEIRA et al., 2015).
The experiment was conducted in a randomized block design with three replications and four plants per plot (spacing of 1.5 m between rows and 1.0 m between plants), two central plants were considered for evaluation, and using a split-plot arrangement, with plots consisting of three water regimes. The subplots consisted of eight accessions of the genus Manihot.
The plant material was obtained from branches of the middle third of healthy plants; these branches were sectioned into 20 cm segments and planted in polyethylene bags with 1 kg of substrate to obtain uniform seedlings. The substrate was prepared using soil, washed sand, and aged goat manure at 2:1:1 ratio. The seedlings were kept in a nursery for two months and then transplanted to the field (May 3, 2017). The holes into which the seedlings were transplanted received 2 L of aged goat manure, applied around the plants, forming a circle, and 2 L at 120 days after harvest. The accessions were evaluated considering two crop cycles, with two harvests being carried out: the first harvest was performed 120 days after transplanting the seedlings, and the second harvest was performed 60 days after the first.
The three water regimes consisted of: 1 -water depths equivalent to the average of the four-month rainy period of the region, with rainfall simulated with micro-sprinklers based on a 30-year rainfall data series; 2 -100% replacement of crop evapotranspiration (ET C ) and 3 -water depth of 20% ET C , these last two treatments being irrigated by drip irrigation. The drip irrigation system consisted of flexible polyethylene drip tapes with nominal diameter of 16 mm, flow rate of 2.1 L h -1 , with emitters spaced 0.50 m apart. The micro sprinkler system consisted of six emitters per plot, with flow rate of 45 L h -1 , spaced 2.5 m apart. The water application efficiency was determined according to Keller and Karmeli (1974). The distribution uniformity coefficients (DUC) found were 68.95% (micro sprinkler) and 96.52% (drip emitters); the systems were classified as good and excellent, respectively, according to Mantovani (2001).
The irrigation management was carried out in two different ways, for the two crop cycles: one with simulated rainfall based on the average rainfall depths of the region, applying the water depths as a simulation of precipitation for the region, based on the climatological normal for the months of January to May, simulating periods with rain and water stress ( Table 2); and the other with irrigation, based on the crop evapotranspiration (ET C ) proposed by Allen et al. (1998), reference evapotranspiration (ET0) obtained from a weather station installed near the experiment area, and the crop coefficients (KC), which were 0.3 (initial phase of leaf production -up to 30 days in the first cycle and 20 days in the second cycle), 1.10 (beginning of the growth phase until the aerial part is harvested), and 0.50 (final phase), and corrected based on the location coefficient (KL) proposed by Keller and Bliesner (1990).
Water from the São Francisco River was used for the irrigations. It showed electrical conductivity of 0.06 dS m -1 and pH of 7.6. The irrigation was performed on alternate days. Soil moisture was monitored during the experimental period using TDR100 (Campbell) probes in the soil layer of 0 to 20 cm. Plots irrigated with 100% crop evapotranspiration (755.64 mm in the first cycle and 593.56 mm in the second cycle) had constant soil moisture (approximately 20%) throughout the experimental period, regardless of the crop cycle. The soil of plots under water deficit of 20% crop evapotranspiration, 151.13 mm in the first cycle and 118.71 mm in the second cycle, had moisture close to 4% in both crop cycles. The evaluated plants were cut at 20 cm from the soil surface, in all plots at the end of the first crop cycle. Altogether, the experiment was conducted in the field for 180 days. In the first 30 days, all plants of all treatments received the same amount of water, to standardize seedling growth. After 30 days the treatments were applied. The historical series used for the application of the rain simulation treatment, along 150 days, is described in Table 2.   The morphological descriptors evaluated for the first and second cycle were: plant height, from the ground level to the base of the terminal bud insertion of the plant, measured with a tape ruler (cm); stem diameter, at the base of the lateral bud insertion, measured with a digital caliper; number of leaves per plant, considering leaves with at least 60% green leaf blade; leaf lobe length, from the lobe insertion to the petiole to the upper end of the central lobe of the leaves, measured with a ruler (cm); leaf lobe width, at the basal part of the lobe, measured with a ruler (cm); and shoot dry mass production (kg.ha -1 ).
For each cropping cycle, a split-plot analysis of variance was performed, considering the effects of the water regime as fixed and the accessions as random. The Scott-Knott method was applied to discriminate accession means within each studied water regime. All analyses were performed using the Exp.Des.pt package of the R Program (FERREIRA; CAVALCANTI; NOGUEIRA, 2018).

RESULTS AND DISCUSSION
Significant effects (p<0.05) were found in the analysis of variance for water regime, accessions and interaction between these two factors on practically all the variables evaluated in both cycles. The only exception was observed for leaf blade width, for which a significant effect was observed only for the accession factor (Table 3). The presence of the interaction between the factors, in the two cultivation cycles, indicates different behavior of the accessions as a function of the water regimes.
Studies involving genotypic characteristics of species of the Manihot genus are scarce in the literature; however, parameters such as plant height can be important for quantification of genetic diversity and selection and identification of water deficit-tolerant plants (OKOGBENIN et al., 2013), because it is essential for the evaluation of yield in several crops. Matos et al. (2016) evaluated M. esculenta under water deficit and found higher plant heights in materials subjected to water stress, especially for the BRS 399 and BRS 398 cultivars. Oliveira et al. (2017) evaluated 49 accessions of M. esculenta and observed that taller plants tend to be more susceptible to water stress. Bergantin et al. (2004) reported that some characteristics of M. esculenta, such as plant height and number of leaves, were affected by both water regime and genotype, and the water stress atrophied the plants, which had fewer leaves when compared to irrigated ones, confirming the results found in the present work (Table 4 and Table 6).
BGMS-102, BGMS-115, BGMS-79, and BGMS-24 had the highest genotypic means for stem diameter in the first crop cycle with simulated rainfall (Table 5). Table 5. Genotypic means for stem diameter (mm) of accessions of the genus Manihot subjected to different water regimes in two crop cycles.
(1) Accessions from the Work Collection of Wild Species of the Manihot Genus of the Embrapa Semiárido. (2) Means followed by same letters in the columns belong to the same group by the Scott-Knott test (P < 0.05). (3) Crop evapotranspiration.
Plants with 100% ET C had similar genotypic means, with average stem diameter above 18 mm. The genotypic means ranged from 12.82 mm (BGMS-115) to 16.62 mm (Engana-Ladrão) with 20% ET C . According to Vale et al. (2012), when the genotype is subjected to a stress condition, decreases in some characteristics are found, even in tolerant genotypes. According to Araújo Filho et al. (2013), the greater water availability to M. pseudoglazovii plants resulted in increased stem diameter. Several other studies confirm increases in stem diameter due to great water availability to the plants (DUTRA et al., 2012), as found in the present study. Therefore, plant growth and development depend on cell division, elongation, and differentiation, whose processes are impaired by water stress because of loss of turgescence (TAIZ et al., 2017). Stem diameter is important for the development of branches and, consequently, for the plant architecture. Thus, a low plant development regarding stem diameter and plant height resulting from water stress may reduce the number of branches for planting and hinder plant establishment (RITCHIE et al., 2010).
Stem diameter was significantly reduced in plants with simulated rainfall in the second cycle, and BGMS-110 had the lowest genotypic mean, denoting less tolerance to this condition. BGMS-24, BGMS-110, BGMS-48, and BGMS-115 had greater stem diameter in treatments with 100% ET C . The genotypic means for stem diameter were relatively similar for all accessions evaluated with 20% ET C .
The highest genotypic means for number of leaves in  (Table 6). For the second cropping cycle, only accession BGMS102 showed a higher genotypic average for this variable when the plants were subjected to the rain simulation regime in this period (Table 6). Its performance under simulated rainfall indicates a genotype with potential for growing as forage in regions with limited water supply. Plant species subjected to great water availability produce larger numbers of leaves, and reductions in growth may by a defense strategy due to drought, mainly by accelerating leaf senescence and abscission, which reduces plant transpiration (ANJUM et al., 2011). Table 6. Genotypic means for the number of leaves of accessions of the genus Manihot subjected to different water regimes in two crop cycles.
(1) Accessions from the Work Collection of Wild Species of the Manihot Genus of the Embrapa Semiárido.
(2) Means followed by same letters in the columns belong to the same group by the Scott-Knott test (P < 0.05).
With the increase of water supply, under the regime with 100% ET C , the accessions BGMS115, BGMS-24 and BGMS-79 stood out with the highest means (Table 6). In the regime with 20% ET C , the highest average results were observed for the accessions Engana-Ladrão, BGMS-110, BGMS-48 and Gema-de-Ovo.
Although BGMS-79 showed similar number of leaves under water regimes with limited water supply in the second crop cycle, this number decreased when the simulated rainfall stopped, indicating the presence of a mechanism of tolerance of the plants to the water stress that causes loss of leaves under limited water conditions.
In addition, differences in shoot yield between accessions and plant age must be considered, since they are factors that affect leaf production, due to increases in the number of leaves, which is important for the survival of the plant and performance of crops, since they capture sunlight for photosynthesis (MORAES et al., 2013;TAIZ et al., 2017). Thus, the number of leaves of plants is dependent on the development stages, for example, dry leaves fall when plants are older or under stress (VANDEGEER et al., 2013).
Engana-Ladrão, Gema-de-Ovo and BGMS115 had the highest genotypic means for leaf lobe length, with simulated rainfall in the first cycle (Table 7). The accessions evaluated showed similar genotypic means with 100% ET C and 20% ET C . BGMS-102, BGMS-115, BGMS-24, and BGMS-79 had the highest genotypic means with simulated rainfall in the second crop cycle. They maintained the leaves even with low irrigation in this period, indicating a mechanism of tolerance to water stress conditions. Gema-de-Ovo and Engana-Ladrão had the highest genotypic means with 100% ET C . The genotypic means of the accessions were similar with 20% ET C ( The accession had greater genotypic means for leaf lobe length with simulated rainfall and 100% ET C in the first crop cycle, and reduced genotypic means with 20% ET C , showing leaf losses as a defense mechanism for mitigating the effects of water stress, allowing its survival. The genotypic means were also reduced in the second crop cycle, mainly in the simulated rainfall. Leaf length and width are also connected to leaf area, whose reduction is one of the most important plant defense mechanisms against water deficit. Considering the morphological characteristics, reduction of leaf area is one of the first reactions of plants to water deficit (TAIZ et al., 2017). Moreover, there is a smaller reduction in the leaf lobe length and, consequently, in leaf area, which decreases transpiration and lessens the risk of permanent wilt in plants under water deficit (ARRUDA et al., 2015). In general, considering the two crop cycles evaluated, the genotypic means of the accessions under the water regimes used were similar, especially those of accessions subjected to the greatest and lowest water availability. It may be due to the sensitivity of the studied materials to water stress regarding leaf lobe length (Table 7).
The accessions showed similar genotypic means for leaf lobe width under the water regimes evaluated in the first crop cycle (Table 8). BGMS-102, BGMS-115, BGMS-24, and BGMS-79 had the greatest genotypic means with simulated rainfall in the second crop cycle, indicating the feasibility of planting these accessions in areas with water shortage conditions. The water availability in the 100% ET C and 20% ET C regimes in the first crop cycle did not affected the genotypic means in the second crop cycle.
In general, the lower water supply reduced the genotypic means for leaf lobe width, and the greater water supply resulted in similar genotypic means for all accessions evaluated in the first and second crop cycles. However, BGMS-48, BGMS-110, Gema-de-Ovo, and Engana-Ladrão had reductions in genotypic means with simulated rainfall in the second crop cycle, denoting genetic materials highly sensitive to water stress.
According to PEZZOPANE et al. (2015), abiotic stresses generate a series of responses of plants through gene expression and cellular metabolism that reduce their cellular osmotic potential, and leaf expansion and overall growth are strongly affected by this reduction in cellular turgor resulting from water stresses. Leaf lobe width is also connected to leaf area and mechanisms of reduction of leaf lobe length and width, and closure of stomata, acceleration of senescence, and leaf abscission are the main responses of plants to water Table 7. Genotypic means for leaf lobe length (cm) of accessions of the genus Manihot subjected to different water regimes in two crop cycles.
(1) Accessions from the Work Collection of Wild Species of the Manihot Genus of the Embrapa Semiárido. (2) Means followed by same letters in the columns belong to the same group by the Scott-Knott test (P < 0.05). ( deficit, in the attempt to maintain the cellular water potential (TAIZ et al., 2017). This was shown mainly by the reduction of the genotypic means of leaf lobe width of accessions subjected to low water supply, regardless of the crop cycle (Table 8). Plants with greater leaf length and width subjected to water stress lose their leaves by accelerating senescence. In addition, the larger the leaf area, the greater the loss of water through transpiration. Table 8. Genotypic means for leaf lobe width (cm) of accessions of the genus Manihot subjected to different water regimes in two crop cycles.
(1) Accessions from the Work Collection of Wild Species of the Manihot Genus of the Embrapa Semiárido. (2) Means followed by same letters in the columns belong to the same group by the Scott-Knott test (P < 0.05).
In the first crop cycle, BGMS-79, BGMS-115, BGMS-102 and BGMS-24 had the highest genotypic means for shoot dry mass production with simulated rainfall (Table 9). Most accessions showed similar genotypic means with 100% ET C , especially BGMS-115, BGMS-102 and BGMS-24. BGMS-102 and BGMS-79 had the highest mean for shoot dry mass production with 20% ET C , denoting genetic materials tolerant to water stress. In the second crop cycle, Gema-de-Ovo had the highest genotypic means for shoot dry mass with simulated rainfall, followed by BGMS-115, BGMS-102 and BGMS-24. BGMS-24 had the highest genotypic means with the 100% ET C , whereas Gema-de-Ovo and Engana-Ladrão had the lowest means. Accessions with 20% ET C had their genotypic means strongly reduced, denoting the importance of water for the survival and development of these plants. In the second crop cycle, BGMS-102, BGMS-79 and BGMS-24 had higher genotypic means with simulated rainfall than the other accessions in the same crop cycle with 20% ET C .
The accessions with greater water supply had higher genotypic means for shoot dry mass production (Table 9), and lower genotypic means with water limitation, in the first and second crop cycles with simulated rainfall and 20% ET C . This can be attributed to the water stress, which may have caused negative effects on cellular turgidity of the plant tissues, and negative alterations in the spatial relationships in membranes and organelles through the reduction of their volume, reducing the hydrostatic pressure inside the cells; this denotes the importance of water to the plant development and production (SANTOS et al., 2012). According to Teodoro et al. (2015), growth analysis through evaluation of dry weight is important to understand the performance of current cultivars, considering that several physiological processes affecting plant development are related to this parameter. Thus, production and development of plants are connected to water use efficiency and must be the basis for the genetic improvement of agricultural-interest species intended for greater productivity with a lower water demand. The biomass production was affected by the water regimes, so the accessions that had the highest genotypic means for shoot dry mass production showed physiological or morphological mechanisms for their development under these conditions, maintaining tissue hydration or active metabolism while dehydrated (TAIZ et al., 2017). However, a higher biomass production requires greater availability of water to the plant due to its increased transpiration (BERNIER et al., 2008). Studies have pointed out the existence of varietal differences in Manihot spp. and differences in their responses to stress situations. Oliveira et al. (2015) evaluated genetic parameters for drought tolerance in Manihot sp. and confirmed this fact, concluding that estimates of genetic variances are higher under water deficit conditions for most agronomic variables for this crop.

CONCLUSIONS
The highest genotypic means were expressed by the accessions BGMS115, BGMS102, BGMS79 and BGMS24, for most of the variables analyzed in this work, regardless of the cropping cycle, indicating high genetic potential to tolerate cultivation under rainfed conditions or with limited water use by irrigation.
For 20% ET C treatment, the accessions with the highest genotypic means for the characteristic dry matter production were BGMS79 and BGMS102, in both crop cycles. These accessions also obtained higher genotypic means for this trait for treatments with rain simulation and 100% ET C .
Considering the genotypes separately and, regardless of the cropping cycle, the characters number of leaves, length and width of the leaf blade and dry mass of the leaves were efficient in indicating variability among the genotypes under the conditions applied in the treatments under study. Table 9. Genotypic means for shoot dry mass production (kg.ha -1 ) of accessions of the genus Manihot subjected to different water regimes in two crop cycles.
(1) Accessions from the Work Collection of Wild Species of the Manihot Genus of the Embrapa Semiárido. (2) Means followed by same letters in the columns belong to the same group by the Scott-Knott test (P < 0.05). (