T he different response of sugarcane genotypes in multiple stress

Research focused on identify abiotic stress-tolerant genotypes is highly desirable since their use may reduce costs of soil and crop management and productivity losses. The aim of this study was to determine the behavior of 24 sugarcane genotypes under high levels of Al3+ and Mn2+ associated with low availability of mineral nutrients. The experiment was carried out under greenhouse condition in a 24 × 2 factorial scheme (24 genotypes × 2 treatments: with and without stress), and four replications in completely randomized design. In the treatment without stress plants were grown in a complete nutrient solution whereas in the treatment with stress a nutrient solution with a high acidity (4.0 ± 0.1) and 5% of its original concentration, as well as a high concentration of aluminum (60 mg L-1) and manganese (700 mg L-1) was used. The genotypes RB966928, RB855443, IACSP96-3060, SP81-3250, RB867515, CTC 21, RB965902, and IAC91-1099 had their biometric characteristics less affected by the stress, possibly due to the ability to continue the process of cell division and elongation and to maintain meristematic viable regions, hence they were considered as the most tolerant. On the other hand, the genotypes RB965917, CTC 15, CTC17, RB855536, CTC 2, CTC 20, and CTC99-1906 were the most sensitive to stress. Root system was the most affected by stress, with most genotypes showing more than 70% reduction in root biomass. No relationship was observed between tolerance level of genotypes and the maturation cycles.


INTRODUCTION
Due to the growing demand from domestic and foreign markets for renewable fuels, sugar, and bioenergy, sugarcane (Saccharum spp.) has become increasingly important in the Brazilian scenario (Unica, 2017).Currently, along with their derivatives, sugarcane compose the second largest source of primary energy in the Brazilian energy matrix (Maia et al., 2018).
As a result, in recent years there has been a great expansion of sugarcane cultivation in Brazil (Goldfray et al., 2010;Caldarelli;Gilio, 2018).With this expansion, sugarcane has advanced to regions of the west of São Paulo State and areas of the Brazilian Cerrado, which are characterized by acid soils, high levels of toxic elements such as aluminum (Al 3+ ) and manganese (Mn 2+ ), and generalized nutrient deficiency (Sousa;Miranda;Oliveira, 2007).
The stress caused by Al 3+ is among the most significant to the crop, damaging mainly its root system and reflecting in a low water and nutrient absorption, consequently reducing plant growth and development (Chen et al., 2010;Silva et al., 2010).
The scientific research focused on identifying and understanding genotypes tolerant to these conditions is highly desirable since their use directly in the field or indirectly in breeding processes allows reducing costs of soil and crop management and productivity losses, which is reflected in an increased agricultural and industrial stability and yield (Too et al., 2014).
When there are no limiting climatic and soil factors, sugarcane develops a large part of its metabolically active roots from the soil surface up to about one meter deep and can reach up to two meters deep in the soil (Luchiari Junior, 1986).However, even correcting and fertilizing the soil arable layer, unfavorable chemical characteristics remain in the subsurface of soil, acting as a chemical barrier that limits root development in depth (Raij, 2011).
Considering that plants are routinely subjected to the interaction of different abiotic stresses, it seems coherent to consider that in these studies the genotypes be submitted to their combined effect (Carlin;Rhein;Santos, 2012).However, most studies with the aim at evaluating the behavior of genotypes to abiotic stresses are limited to studying one factor at a time (Fonseca Júnior et al., 2014;Maia et al., 2018).These studies are very important, but not indicated to choose genotypes for limiting environments such as those in crop expansion areas.In this context, this study aimed to determine the behavior of 24 sugarcane genotypes under high contents of Al 3+ and Mn 2+ , associated with a low nutrient availability, as well as verify the genetic variability related to this joint stress, the relationship between tolerance and maturation cycle (early, medium, and late), and determine the possible tolerant and susceptible genotypes.

Experiments for methodology adjustment
The experiment was carried out in the São Paulo State University, Unesp, School of Agricultural and Veterinary Sciences (FCAV-UNESP), Jaboticabal, SP, located at the geographical coordinates 21°15′22″ S and 48°18′58″ W, with an altitude of 575 m, in the period from April 4 to July 23, 2017.
For carrying out this study, the nutritive solution was provided by means of the system described by Dantas et al. (2001).Briefly, seedlings were obtained from culm billets (3 cm length) containing one bud each and followed the methods described in Carlin, Rhein and Santos, (2012).After cutting, the culm billets were immediately planted in 500 mL capacity plastic containers with holes in the lower part and containing washed and sieved (2 mm) sand.Seedlings were grown without any water restriction for 28 days.After this period, they were selected regarding sanity and homogeneity, being transplanted to 1 L capacity plastic pots (dimensions of 15 × 9 × 9 cm), with small holes at 0.5 cm from the bottom to allow the entrance of the nutrient solution, and containing 750 ml of washed sand.Then, the described plastic pots were placed in a tray with dimensions of 42 × 36 × 11 cm, in which was maintained a layer of 5 cm of nutritive solution, prepared according treatment of each assay.The level of solution was completed daily with distilled water and completely replaced by fresh solution every three days.
To test the mentioned hidroponic alternative system for sugarcane, and to define Al 3+ dose and to find the suitable solution nutrient strength, three preliminary tests were carried out in order to determine the best conditions to work in the main assay with genotypes.All these experiments were carried out in a greenhouse under the same conditions and with the same genotype (IACSP 95-5000).The IACSP 95-5000 is indicated for favorable environments (A1 -C2) and therefore is sensitive to acid and poor soils (Chaves et al., 2015).
The first one aimed to verify the efficiency of hydroponic alternative system in comparison to the traditional hydroponic system.This assay was carried out in a completely randomized design with two treatments (traditional hydroponic systems, and the alternative system) with 10 replications each.In both systems was used a complete nutrient solution (Furlani;Furlani, 1988) with adaptations for sugarcane, prepared with the following stock solutions: 3.1 ml L -1 Ca(NO 3 ) 2 1.64 mol L -1 , 3.1 ml L -1 NH 4 NO 3 0.42 mol L -1 , 2.2 ml L -1 KCl 0.25 mol L -1 , 2.2 ml L -1 K 2 SO 4 0.25 mol L -1 , 2.2 ml L -1 KNO 3 0.24 mol L -1 , 1.6 ml L -1 Mg(NO 3 ) 2 0.96 mol L -1 , 0.2 ml L -1 KH 2 PO 4 0.13 mol L -1 , 0.6 ml L -1 FeDDH 0.16 mol L -1 , 0.6 ml L -1 MnCl 2 0.08 mol L -1 , 0.6 ml L -1 H 3 BO 2 0.03 mol L -1 , 0.6 ml L -1 ZnSO 4 0.005 mol L -1 , 0.6 ml L -1 CuSO4 0.001 mol L -1 , and 0.6 ml L -1 Na 3 MoO 2 0.001 mol L -1 , with final pH adjustment to 5.5 ± 0.1 by using HCl 0.1 mol L -1 or NaOH mol L -1 .For traditional hidroponic system, a layer of 10 cm of nutritive solution was used kipping continuous aeration, while for alternative system, a 5cm layer was maintained in the tray without aeration.In both systems the level of solutions was completed daily with distilled water, and completely replaced by fresh solution every three days.
The other two experiments were conducted with the mentioned alternative system, in a completely randomized design.For the second preliminary assay, treatments consisted of eight Al 3+ concentrations in the complete nutrient solution (0,10,20,30,40,50,60, and 70 mg L -1 ), with three replications.Regarding the assay to define the dilution to be used as low nutrient concentration associated with Al 3+ and Mn 2+ toxicity, the concentrations of 0, 5, 10, 15 and 20% of the complete solution (Furlani;Furlani, 1988) were tested, maintaining fixed the doses of 60 and 700 mg L -1 of Al 3+ and Mn 2+ , respectively, with four replications.
In all three preliminary assays, plants were maintained under respective treatments for 30 days and evaluated for shoot and root dry matter.

Location and experimental design
The experiment was performed in a greenhouse in a completely randomized design, in a 24 × 2 factorial scheme, consisting of 24 genotypes, two treaments (with and without stress), and four replications, totaling 192 experimental units.Seedlings used in this experiment were obtained and selected as mentioned for preliminary assays.

Genotypes tested
The 24 genotypes used in this study belong to three distinct groups regarding maturation, previously identified as early, medium, and late maturation cycle (Table 1).Seedlings were prepared as above mentioned, and propagules were obtained from plants with same age, grown in the FCAV-UNESP experimental farm, thus submitted to the same conditions of management, climate, and soil.

Characterization of treatments with and without stress
For the treatment without stress, plants were grown in a complete nutrient solution (Furlani;Furlani, 1988) with adaptations for sugarcane and a pH of 5.5 ± 0.1.For the treatment with stress, the nutrient solution was only 5% of its original concentration, and with a high acidity (4.0 ± 0.1), containing high concentration of aluminum (60 mg L -1 ) and manganese (700 mg L -1 ), both applied in the form of chloride.For determining the ionic strength of nutrient solution (5%) and the aluminum dose (60 mg L -1 ), the results of the preliminary tests were taken as a basis and manganese concentration (700 mg L -1 ) was defined based on the literature (Benett et al., 2012).
Solution level was corrected daily by adding distilled water, at that time, pH was also corrected.Trays were cleaned and solutions completely replaced every three days when all trays were randomly changed, as well as the seedlings in each tray, in order to keep all experimental units in a completely randomized design.

Analyzed variables
After 50 days, the following non-destructive assessments were performed: number of green leaves, number of dead leaves, stem height, stem diameter, plant height, and leaf area.The live leaves were considered those completely open and with at least 20% green area and the dead leaves were considered those with 20% less green area.Leaf area was determined following the methodology proposed by Hermann and Câmara, (1999).Green color index was determined by using a portable chlorophyll meter model CCM-200 (Opti-Scienses, Inc.), with three measurements in the leaf +1 of each plant, obtaining the green color index (GCI) data of leaf.
After non destructive measurements, plants were harvested, and separated into the shoot, roots, and culm billets.Root volume of plants was determined by immersing the roots in a graduated test tube with distilled water and measuring the displaced volume.All samples were then placed in properly identified paper bags, weighed to determine the fresh matter, and dried in a forced air circulation oven at 65 ± 5 °C until constant weight for determining the dry matter.
Considering that different genotypes have different growth potentials, the comparison of absolute values is not suitable to compare tolerance among genotypes.Because of that, for the main experiment, the relative growth (or relative growth imitation) was used to compare data, and tolerance among genotypes (Lima;Peixoto;Ledo, 2007;Maia et al., 2018).Thus, the average value of replications of each genotype without stress was considered to be 100%.Then, this value was used as reference to calculate relative values of all data obtained in all experimental units belonging to the same genotype.So these relative values were submitted to statistical analysis.

Statistical analysis
The results of the preliminary experiments were submitted to analysis of variance by the F-test and mean comparison by the Tukey's test at a 5% probability level.In the experiment in which Al concentrations were tested, when a significant effect was detected, the polynomial regression analysis was applied by using the software AgroEstat (Barbosa;Maldonado Junior, 2015).
In the main experiment, analysis of variance (F-test) was carried out with relative data in a 24 × 2 factorial scheme (24 genotypes × 2 treatments).Mean comparison was carried out by the Scott-Knott test at 5% probability, also using the software AgroEstat (Barbosa;Maldonado Junior, 2015).
In order to observe the similarity among genotypes, a multivariate exploratory analysis was also carried out by using the software Statistica 7.0 (StatSoft.Inc, 2004), since it is commonly applied statistics to provide additional information on the response of genotypes to different environments (Fox;Crossa;Romagosa, 1997).In order to differentiate the clusters and their relationships with the studied variables, a nonhierarchical method was performed by the k-means grouping.
In addition, the factor analysis was used to identify the processes that would respond to the highest variabilities of the measured variables.For this, we first calculated the load values of provisional factors determined by the principal components analysis, which allowed to obtain components which eigenvalues were not lower than 1, following Kaiser (1958) criterion.
The principal component analysis was used as a technique to extract factors (Seal, 1964;Jeffers, 1978), which is based on the correlation matrix between variables.Factors with eigenvalues ≥ 1 were selected also by following also the Kaiser (1958) criterion.In order to identify the factors, the VARIMAX rotation method was adopted (Kaiser, 1958;Hoffmann, 1992), in addition to providing a better interpretation of factors, this method has as objective to obtain a matrix of loads more identifiable regarding the nature of the measured variables (Maxwell, 1977).

Experiments for methodology adjustment
The hydroponic alternative system used in this study provided similar results (P > 0.05) to those observed for the traditional hydroponic system for all variables of growth and development of Sugarcane plants (Table 2).A better root development was obtained in the alternative system when compared to the traditional hydroponic system.This result is very important in this type of research since this variable is one of the most important to study the tolerance of plants to Al 3+ toxicity (Ecco;Santiago;Lima, 2014;Maia et al., 2018).
The great advantage of this hydroponic alternative system is that it allows assessing a large number of experimental units in a much simpler and economical way since it does not need the oxygenation system of the nutrient solution.In addition, it allows a greater management control on the nutrient solution, as well as changing plant position randomly without the risk of damaging its root.
The results for the shoot and root dry matter with increasing Al 3+ doses showed that the maximum applied concentration (70 mg L -1 ) was not lethal for plants (Figure 1A).For shoot dry matter (SDM), a decreasing linear effect was observed as the Al 3+ concentration increased, with an average weight of 5.19 and 4.38 g for the 60 and 70 mg L -1 doses, respectively.For root dry matter (RDM), the same effect was also observed as the aluminum doses increased, with a mean weight of 3.11 and 1.80 g for the 40 and 70 mg L -1 doses, respectively.
In relation to the experiment that tested nutrient solution, no effect was observed on shoot dry mass in the concentration of up to 20% of the original values proposed by Furlani and Furlani (1998), where Al 3+ and Mn 2+ remained constant.This was probably due to seedling age, i.e. at this stage, only the reserve still available in the culm piece is sufficient for plant development.In addition, symptoms of Al 3+ toxicity in the shoot are not always readily identifiable (Vitti;Mazza, 2002) and, unlike the root system, have a little direct effect of this element, mainly in relatively short-time (Rossiello;Jacob-Netto, 2006).The concentration that less affected root system was that of 5 and 10% of the nutrient solution (Figure 1B).Thus, the concentration of 5% is sufficient to lead to a nutritional stress and be assessed with the effect of Al 3+ and Mn 2+ without causing root death.

Main experiment with 24 genotypes grown with and without stress
As for the main experiment with the 24 genotypes, the analysis of variance showed a significant effect (P < 0.05) for genotypes, treatment, and the interaction genotype × treatment for all 14 biometric variables, except for culm billets fresh, and dry weight.This result shows the existence of genetic variability among the 24 genotypes (Table 3).The significant effect of the treatment on the studied variables also indicates that the conditions were adequate to assess the proposed stress in sugarcane by using this new hydroponic system.
In the principal component analysis (PCA), two components were extracted, which explained 57.46% of the total variation, discriminating the genotypes in four groups (Figure 2A) The first component (PC1) explains about 42% of the total variation of genotypes and has groups represented in green and blue colors, being strongly influenced positively by the variables of the shoot and root biomass (Figure 2A).Genotypes that are furthest from the origin, farther to the right and more aligned with the horizontal axis, are the most tolerant to stress.In turn, the red group, which presents totally opposite behavior of the genotypes of the green and blue groups, should be the most sensitive to the environment under stress.
Table 2: Means, least significant difference (LSD 5%), and standard error of the means (SE) of the biometric characteristics from sugarcane plants (IACSP95-5000) under traditional (HP) and adapted (AS) alternative system.Stem length (SL), plant height (PH), leaf area (LA), root fresh weight (RFW), shoot dry weight (SDW), number of green leaves (NGL), shoot fresh weight (SFW), root dry weight (RDW), root volume (RV), and stem diameter (SD), Means followed by the same letter in the column do not differ statistically from each other by the Tukey's test at 5% probability level.

Systems
Variables      The second component (PC2) is responsible for 15.81% of total variation, presenting more relation with variables whose arrows are more aligned with the vertical axis of the graph.Therefore, the genotypes 12, 20, 18, and 24 are also influenced by stem height (SH), stem diameter (SD), and number of dead leaves (NDL).The variables GCI, number of dead leaves (NDL), and number of green leaves (NGL) did not present a strong influence on the groups discriminated by PC1, despite being indicators of tolerance (Inman-Bamber et al., 2008;Silva et al., 2011) (Figure 2A).The component PC2 was also responsible for dispersing some genotypes of the red group.The genotypes 5 and 22 were strongly influenced by the same variables responsible for the separation of the lilac group, while the genotypes 7 and 21 tended to present an opposite response to that observed by the components of the lilac group.
By analyzing the non-hierarchical k-means clustering results, we decided to define four clusters based on the number of groups determined by principal component analyzes.This analysis allows better observing the separation of groups regarding the genetic variability, as well as their relationships with the evaluated variables (Figure 2B).
As for the pattern of division of genotypes within each group, no relation was observed with maturation cycles, that is, precocious, medium and late.The first cluster, formed only by the genotype 3 (Figure 2B) stood out with the highest variations for the variables of the shoot, root system, and GCI.The second cluster, formed by genotypes 1, 2, 4, 6, 8, 9, 10, 15, 17 and 23, presented the lowest variation and certainly formed by more tolerant and moderately tolerant genotypes to the tested stress, because it was above average for almost all biometric variables analyzed.The third cluster was formed by the genotypes 5, 7, 11, 13, 14, 16, 19, 21, and 22 and was well below the average for the variables.In this cluster, the genotypes 11, 13, 21, 7, and 5, which represent the cultivars IACSP95-5094, CTC2, CTC 15, CTC 17, and RB965917, respectively, were the most sensitive in the treatment with stress.The variation pattern of each cluster is closely related to the average of each group.Therefore, the highest variation was observed for the genotype 3.
In relation to the factor analysis, we first calculated the load values of provisional factors determined by the principal components analysis, which allowed to obtain components which eigenvalues were not lower than 1, following Kaiser (1958) criterion.Approximately 77% of the data variability is explained by four principal factors.Thus, the results of the fourteen original variables were distributed into only four factors, each of them representing an independent physiological process (Table 4).
In the first factor, shoot dry mass, plant height, shoot fresh mass, root volume, root dry mass, and fresh root mass were the variables with the highest factor loads, explaining 41.65% of the total variation of the data (Table 4).This factor is related to the Al 3+ and Mn 2+ toxicity processes and the low nutrient content that must be occurring on the genotypes since the variables SDM, WPH, SFM, RV, RDM, and RFM are positively correlated with this factor (Table 5).
In the factor 2, the variables with the highest factor loads were culm billets fresh and dry weight, which presented an inverse correlation with Factor 2. This factor was responsible for 15.81% of the total variation of the 14 variables measured.Because they were included in a factor distinct from the Factor 1, they should not be related to Al 3+ and Mn 2+ toxicity associated with low nutrient availability.However, these variables might be more related to the process of preparation of the culm billets for seedling production, possibly due to the variation in the diameter of the collected stems and to the imprecision of stem cutting (length) to obtain the culm billet.
In the third factor, the highest loads are related to the variables NGL and GCI, with values of − 0.826445 and − 0.722021, respectively, influencing 11.10% of the total variation.In addition, the fourth factor explained about 8% of the total variation and was composed of the variables NGL and GCI.Similarly, to the second factor, the third and fourth factors also appear to be unrelated to the stress process here studied.However, they are possibly related to  et al., 2010).Moreover, the differential development and/ or growth index may present different values of GCI unit (Fontes;Araujo, 2007) and NGL.
As for the groups formed by the genotypes in the results of Scott-Knot clusters was not found clear relation with the maturation cycles of the genotypes (Table 6), confirming the results obtained by the non-hierarchical grouping method (Figure 2).
Taking into account that tolerance to abiotic stress is the plant ability to maintain stable growth (and values of other physiological parameters) when subjected to stressful conditions (Maia et al., 2018), it is suitable to say that genotypes with a lower variation of growth (and other variables) when comparing different environments can be characterized as tolerant.And, great variation indicates low tolerance.Therefore, the genotypes 3, 2, 12, 9, 23, 8, 4, and 10 had their biometric characteristics less affected under the proposed stress, and hence they were considered as the most tolerant genotypes (Table 6).This tolerance possibly is due the ability to continue the process of cell division and elongation and to maintain meristematic viable regions (Foy,1984).On the other hand, the genotypes 5, 21, 7, 22, 13, 14, and 16 were the most sensitive to the proposed stress, especially the first three genotypes.
The behavior presented by these two groups of genotypes showed great coherence with the reality in the field.For example, the genotypes of the group identified as tolerant are indicated and are actually being cultivated in acid-poor, nutrient-poor and drought-prone environments (Silva et al., 2012;Ridesa, 2017).The genotypes 3, 9 and 23 correspond to almost 50% of the planted area in the northeast region of the country and most of the Brazilian cerrado (Ridesa, 2015a;Ridesa, 2017), regions characterized by acid and nutrient poor soils.On the other hand, the genotypes identified as sensitive are indicated to production environments with good soil and climatic conditions (Ridesa, 2015a;Ridesa, 2017;CTC, 2018).In fact, genotypes 13, 21 and 22 do not reach 3% of the planted area of the Brazilian cerrado, and in the Northeast this percentage is less than 1% (Ridesa, 2015a;Ridesa, 2017).Therefore, it seems that the tolerance level presented by the genotypes may be related to the tolerance to aluminum.Shoot dry weight (SDW), stem diameter (SD), plant height (WPH), culm billet fresh weight (CBFW), culm billet dry weight (CBDW), shoot fresh weight (SFW), leaf area (LA), stem height (SH), root volume (RV), root fresh weight (RFW), root dry weight (RDW), number of green leaves (NGL), number of dead leaves (NDL), and green color index (GCI).

Variables
Factor Among the biometric variables, the root system was the most affected by the stress caused by Al 3+ .In most genotypes, stress limited to root biomass accumulation by more than 70% (Table 6).This result is higher than that found by Watt (2003) when assessing the growth of roots exposed to high Al 3+ concentrations, who observed an inhibition of root growth, with a variation between 36 to 46%.
This high limitation of root biomass production may be associated not only with Al 3+ effect but also with Mn 2+ effect.In fact, although symptoms of manganese toxicity in plants are more pronounced in leaves than in roots, when wheat tolerance to aluminum toxicity was determined together with that of manganese in nutrient solutions, all the genotypes showed a reduction in root growth ranging from 59 to 68% as Mn 2+ concentrations in nutrient solutions increased from 0.11 to 1200 mg L -1 (Camargo, 1995).The reduction of root dry matter due to the toxic effect of Mn 2+ has also been observed in other crops such as Rice (Lindon;Barreiro;Ramalho, 2004) and bean (Soratto et al., 2005).Root volume (RV), root fresh weight (RFW), root dry weight (RDW, culm billet dry weight (CBDW), green color index (GCI), number of green leaves (NGL), and number of dead leaves (NDL) and culm billet weight (CBFW).Means followed by the same letter in the column do not differ statistically from each other by the Scott & Knott test at a 5% probability level.Although no statistical difference has been observed among genotypes regarding the reduction of root dry mass, those less affected also stood out in the other variables.This result reinforces the importance of this variable in the selection of genotypes to the stress caused by Al 3+ , Mn 2+ and low nutrient avalilability.

Genotypes
Regarding GCI readings, no statistical difference was observed among genotypes under stress, even though it is considered as a good tool for genotype selection (Silva et al., 2011).Despite this result, pigment content in leaves was affected by the stress when compared to the treatment without stress, indicating that only GCI reading does not seem to be enough to select genotypes for this variable at that stress level.Therefore, other analyses are recommended in future researches with the same focus.
The genotype 3 stood out with the best results for root volume, root fresh weight, and the second highest average for stem diameter and plant height.In addition, this genotype did not present a reduction in leaf area and shoot dry weight in the treatment with stress.A similar result for this latter variable was also observed for the genotype 2, composing the more tolerant group "A" to stress imposed (Table 7).The most sensitive group "D" consisted of genotypes 5, 11, 13, 19, 21 and 22.The lowest averages for shoot dry weight were presented by the genotypes 21, 11, and 5 and, consequently, they presented the highest growth restrictions, which values that reached 28.62, 27.11 and 22.88%, respectively, while genotypes of the other groups showed reductions less than 20% (Table 7).This result is below that found by Ecco, Santiago and Lima, (2014) studying the interaction between types of abiotic stress (water deficit and soil acidity) in sugarcane under greenhouse conditions, and observed a reduction of 23% in shoot biomass production under stress caused by Al 3+ and 69 % under water stress combined with Al 3+ toxicity.Maia et al. (2018) also found that the stress caused by Al 3+ led to an average reduction of 44% in shoot weight.Restrictions on shoot biomass accumulation due to Al 3+ effect, among other factors, occur due to a reduction in photosynthetic activity.Specifically, Al 3+ interfering with the absorption process of nutrients such as nitrogen and manganese may limit the formation and function of chloroplasts (Konrad et al., 2005;Mendes et al., 2018).
The genotypes 9, 23, 8, 4, and 10, even with a good performance, had a production below 100% for shoot biomass.Similar results were obtained by Maia et al. (2018), that assessed the phenotypic plasticity of 11 sugarcane genotypes under Al 3+ stress and classified the genotype 3 (RB966928) as the most tolerant, even presenting shoot with dry weight below the average, under both cultivation conditions.
The performance of more tolerant genotypes may be associated with their ability to exclude Al 3+ from root apex and/or with its accumulation through mechanisms such as Al 3+ chelation in the cytosol, Al 3+ compartmentalization in the vacuole or in aluminum-protein bonds that are some mechanisms of tolerance to Al 3+ (Hartwig et al., 2007;Inostroza-Blancheteau et al., 2012).
In relation to stem length, the genotypes were classified into two groups.The genotypes 17, 12, and 3 were the least affected, presenting a limitation of only 8.36, 16.55, and 17.73%, respectively, while the genotypes 21, 13, 7, and 8 presented a higher restriction in stem growth (about 50%) in the treatment with stress (Table 7).This variable is considered as excellent indicators of tolerance to the proposed stress since they present a higher correlation with sugarcane production (Silva et al., 2008).

CONCLUSIONS
There is variability regarding the tolerance of the 24 sugarcane genotypes under the stress conditions caused by Al 3+ , Mn 2+ , and low nutrient availability.No relationship was observed between tolerance level of genotypes and the maturation cycles.The genotypes RB966928, RB855443, IACSP96-3060, SP81-3250, RB867515, CTC 21, RB965902, and IAC91-1099 were identified as the most tolerant whereas the genotypes RB965917, CTC 15, CTC17, RB855536, CTC 2, CTC 20, and CTC99-1906 were identified as the most sensitive to high levels of Al 3+ and Mn 2+ , associated to low nutrient availability.

AKNOWLEDGMENTS
The first author thanks to CNPq (National Council for Scientific and Technological Development) for grantinf a graduate scholarship.The second author also are grateful to the CNPq for the granting of research productivity scholarship.

Figure 1 :
Figure 1: Values of the shoot and root dry matter of the genotype IACSP95-5000 submitted to Al doses (A) and different concentrations of nutrient solution (B).

Table 1 :
Identification of genotypes and classification regarding maturation cycles.

Table 3 :
Analysis of Variance for Major Effects and Interaction for all variables with relativized data.

Table 4 :
Eigenvalues and percentage of variation explained by the first four principal components of biometric variables of the 24 genotypes under the stress treatment.characteristics of the genotypes which responses are not related to the studied stress.Several factors can affect GCI index values, including the cultivar(Coelho specific

Table 5 :
Factor matrix determined by the Varimax orthogonal rotation method.

Table 6 :
Mean comparison of genotypes under stress with the data transformed into a percentage for biometric variables by the Scott & Knott test at 5%.

Table 7 :
Mean comparison of genotypes in the treatment with stress with data transformed into a percentage for shoot variables by the Scott & Knott test at 5%.Shoot dry weight (SDW), stem diameter (SD), plant height (PH), shoot fresh weight (SFW), leaf area (LA), and stem height (SH).Means followed by the same letter in the column do not differ statistically from each other by the Scott & Knott test at a 5% probability level.