ASSESSMENT OF LEGUMES WITH POTENTIAL USE AS GREEN MANURE IN THE COASTAL TABLELANDS OF PIAUÍ STATE, BRAZIL

- Green manure may lead to physical, chemical, and biological improvements to the soil. However, the information on its use in the Brazilian Northeast is scarce. The aim of this study was to assess the phenological cycle and phytomass productivity of seven legume species. This study was conducted in the second semester of 2015 in a completely randomized design with seven treatments and four replications. Plant height, fresh mass of shoot (FMS), dry mass of shoot (DMS), fresh mass of root (FMR), and dry mass of root (DMR) were determined at 100 days after sowing. Germination, flowering, and pod maturation were also assessed. The highest averages of FMS were observed in Crotalaria breviflora and Crotalaria mucronata , followed by Canavalia ensiformis , whereas the highest averages of DMS were observed in C. breviflora , C. ensiformis , C. mucronata , and Cajanus cajan cv. IAPAR 43. The highest averages of FMR were obtained by C. breviflora and C. mucronata , followed by C. cajan cv. Fava Larga. The species C. mucronata , C. cajan cv. IAPAR 43, and C. breviflora presented the highest averages of DMR. In addition, C. juncea , C. ensiformis , and C. cajan cv. IAPAR 43 were the earlier species regarding flowering, while C. cajan cv. Fava Larga was the later species. C. juncea was the earliest among all studied species regarding pod maturation. The tested legume species obtained promising initial results during the assessment period, being able to meet the expectations of biomass production and contributing to soil sustainability in this region.


INTRODUCTION
Inappropriate modes of arable land occupation and the need for a rapid food production, coupled with economic interests in the search for profitability in the agricultural sector have contributed to the worsening of environmental degradation and increasing of these imbalances. This dynamic has caused severe changes in soil physical, chemical, and biological attributes, which, together with the acceleration of organic matter mineralization (with consequent reduction of soil fertility), have led different production systems to a decreased production potential in several regions of Brazil and world (CALEGARI, 2014). According to this author, this situation and other aspects related to climate (extreme temperatures and occurrence of floods or prolonged droughts) and the severe and unexpected attack of pests and/or diseases has caused serious risks to food security.
Despite this, modifying the technological trajectory currently dominant in the Brazilian agriculture and considerably increasing the production based on agro-ecological models of production would be possible. Several experiments demonstrate the technical, economic, and social viability of these models, contributing to the establishment of a sustainable agricultural development process (ESPÍNDOLA, 2004;ASSIS, 2005;WUTKE et al., 2007WUTKE et al., , 2008WUTKE et al., , 2009FILHO et al., 2014).
Principles and techniques of organic agriculture are adequate for family agriculture since they contribute to its sustainability with increased opportunities, as well as providing a better nutritional quality (CAMPANHOLA; VALARINI, 2001). However, the significant growth in participation of organic agriculture in the national and international market and the dependence of its certification to standardized norms have pressured farmers in the search for organic technologies, which often lack scientific criteria or validations (PEIXOTO et al., 2008).
Organic agriculture is possible aiming only the substitution of inputs used in conventional agriculture by inputs allowed in organic agriculture. This is a product-based agriculture. On the other hand, the production of food and other agricultural products in a sustainable, economic, social, and environmental way is possible in the ecologically based agriculture. This agriculture is based on processes and seeks to intensify the free natural functions of the ecosystem (AMBROSANO et al., 2014). In this context, there is the importance of using plant species known as green manures capable of improving physical, chemical, and biological characteristics of the production environment; moreover, mineral fertilizers enriched by chemical processes and/or chemically treated to increase solubility, as well as herbicide and soil sterilizing, are prohibited in ecologically based agriculture (BRASIL, 2008).
Sandy soils of the Coastal Tablelands of Piauí State (Brazil) can be subjected to a high degree of degradation unless used and managed properly. The use of legumes for green manuring and soil cover due to its multiple uses may contribute to the preservation of agricultural soils of this region (GARCIA, 2002). The fragility of this ecosystem makes the agricultural activity of this region a challenge since there is a predominance of soils with sandy texture, low organic matter content, and low cation exchange capacity, as well as climate with high temperatures, long periods of water deficit, and constant winds (MEHL et al., 2010). In this scenario, most farmers involved with organic production acquire some inputs, including seeds of green manure, from external sources. There is a parallel market to meet these demands, which increases production costs.
Although Duque (1951) and Haynes (1970) have already mentioned the importance of legumes as a green manure for protecting and restoring soil fertility of the drought polygon and Coastal Tablelands, few studies have been conducted using these species in the Northeast region of Brazil (GARCIA, 2002). This is reflected in agricultural production units, where farmers find it difficult to adjust the use of green manures in their production systems. Knowing and adapting the use of green manuring becomes essential for recommending this management practice in the production units of this region.
Thus, this study aimed to observe the phenological cycle and assess the phytomass productivity of different legume species used as green manure in sandy soils of the Coastal Tablelands of the Piauí State, Brazil.

MATERIAL AND METHODS
This study was conducted in the second semester of 2015 at Embrapa Mid-North/UEP Parnaíba, located at the geographical coordinates 3°5′ S and 41°46′ W and at 46.8 m of altitude. Regional climate is defined as C1dA′a′ (THORNTHWAITE; MATHER, 1955), characterized as a megathermal dry sub-humid climate with small water surplus and concentration of 29.7% of potential evapotranspiration from October to December, with the averages of relative air humidity, precipitation, and temperature of 77.5%, 1,107 mm, and 27.6 °C, respectively (BASTOS; ANDRADE JÚNIOR; RODRIGUES, 2012).
The soil of the experimental area was described as a medium textured Dystrophic Yellow Latosol (Oxisol) of coastal caatinga phase (MELO et al., 2004). Soil samples from a depth of 0-20 m were collected at different points in the experimental area in order to obtain a higher uniformity of substrate. A composite sample was taken from the collected material and analyzed at the Laboratory of Water and Soil in the Embrapa Mid-North/UEP Parnaíba for soil chemical characterization (Table 1).     The experiment was carried out in an experimental area in which no chemical fertilization was used in the last five years. Soil tillage consisted of a plowing and surface harrowing with subsequent leveling, followed by manual leveling.
The experimental design was a completely randomized design with seven treatments and four replications in plots of 2.0 × 2.0 m containing five rows of plants spaced 0.50 m. A high seed density (approximately twice the number of seeds corresponding to each treatment) was used. Plant population was subsequently adjusted by thinning at 20 days after sowing (DAS) ( Table 2). Two manual weedings were carried out. The irrigation system used was a conventional sprinkling with a two-day irrigation shift and an average duration of 1.0 hour. The annual species of tropical climate used in the treatments were T1 -dwarf pigeonpea cv. IAPAR 43 (Cajanus cajan); T2 -smooth rattlebox (Crotalaria mucronata); T3 -sunn hemp (Crotalaria juncea); T4 -jack bean (Canavalia ensiformis L.); T5 -shortflower rattlebox (Crotalaria breviflora); and T6 -pigeonpea cv. Caqui/Fava Larga (Cajanus cajan). Moreover, the perennial species of tropical climate T7 -white hoarypea (Tephrosia candida) was also used.
All the treatments were cut at 0.05 m above soil surface by using a pruning shears at 100 days after test installation. The useful plot consisted of the central rows, discarding the borders, totaling an area of 0.5 m 2 (1.00 m linear × 0.5 m between rows). Plant roots were also removed from a depth of 0.20 m by using a mattock in order to obtain fresh and dry biomass for all the species used in this study. Plant height (PH), fresh mass of shoot (FMS), dry mass of shoot (DMS), fresh mass of root (FMR), and dry mass of root (DMR) were determined. Plants were considered under full flowering when they reached 50% of flowering.
All the botanical material from shoot and root was weighed on a precision scale to obtain the fresh biomass (FMS and FMR). Subsequently, this material was placed in paper bags and conditioned in a forced ventilation oven at 65 °C for 48 hours until constant weight to determine dry biomass production (DMS and DMR).
The data were submitted to analysis of variance by using the software ASSISTAT and Duncan's test at 5% probability level was applied for the contrast of means. Table 3 shows the data related to the averages of emergence (E), flowering (F), plant height at flowering (PHF), and pod maturation (PM). No significant differences were observed regarding the emergence of all the studied treatments. The results obtained for C. juncea (T3), C. ensiformis (T4), and C. cajan cv. IAPAR 43 (T1) showed precocity in relation to flowering when compared to other regions of Brazil. In addition, no significant differences were observed between these plants (p<0.05). In our study, the flowering of C. juncea, C. ensiformis, and C. cajan cv. IAPAR 43 started between 66 and 75 DAS. However, the company Piraí Sementes, specialized in commercializing green manure seeds, presents as a reference for this variable 90 to 120 DAS in south-central Brazil (PIRAÍ SEMENTES, 2014).

RESULTS AND DISCUSSION
In the Coastal Tablelands of Piauí State, Garcia (2002) observed a period of 52 and 75 DAS for the beginning of flowering and the full flowering of C. ensiformis, respectively, whereas, for T. candida, these periods were 120 and 170 DAS, respectively. On the other hand, the company Piraí Sementes presents as a flowering reference for C. breviflora a period from 90 to 120 DAS (PIRAÍ SEMENTES, 2014 1 Table 3. Phenological data of different legume species for green manuring.
Means followed by the same letter in the column do not differ statistically by Duncan's test at 5% level. DAS -days after sowing.
Mascarenhas and Wutke (2014) consider temperature, water availability, and photoperiod, together with biotic and soil fertility factors, as the most restrictive to green manure cultivation. However, in the tropics, where there is rarely a lack of solar radiation, the effects of light on plant development are very little evident (ALCÂNTARA; BUFARAH, 1992). The variation in day and night length throughout the year is an important conditioning characteristic of vegetative development and induction to flowering, especially in legumes. Thus, the critical length of day or night is an inductive factor of change from the vegetative to reproductive or flowering phases (MASCARENHAS; . This phenomenon may have influenced the results obtained for flowering in our study. The cycle of C. juncea is dependent on photoperiod variation, but this species is one of the less sensitive of this botanical genus (AMABILE; FANCELLI; CARVALHO, 2000;AMABILE, 2006). Pigeonpea is usually sensitive to photoperiod, with a quantitative response to flowering on short days and influenced by thermosperiod (MASCARENHAS; WUTKE, 2014), while jack bean has no sensitivity to photoperiod and are appropriately used in the semiarid region of Northeast Brazil (BURLE et al., 2006;CALEGARI et al., 1993;AMABILE, 2006).
A significant difference (p>0.05) was observed for PM. The species C. juncea (T3) was the earliest, with an average of 77.25 DAS, differing statistically from all other legume species, followed by C. ensiformis cv. IAPAR 43 (T1) and C. mucronata (T2), with no significant differences between them (97.75 and 105.25 DAS, respectively). In addition, the species C. breviflora (T5), C. ensiformis (T4), and C. cajan cv. Fava Larga (T6) presented averages of 120.50, 133.75, and 161.25, respectively. The species with later pod maturation was T. candida (T7), with a value of 180.25 DAS. In an experiment conducted in the Coastal Tablelands, Garcia (2002) observed a period for the beginning of pod maturation of 130 DAS for C. ensiformis and 160 DAS for T. candida. Table 4 shows the results of FMS, DMS, FMR, and DMR. Significant differences (p>0.05) were observed considering the production performance of the studied legumes. The species yielding the highest fresh mass were C. breviflora (T5) and C. mucronata (T2), with no significant differences between them, followed by C. ensiformis (T4). On the other hand, those species with the lowest fresh mass were C. cajan cv. IAPAR 43 (T1) and C. cajan cv. Fava Larga (T6), with no differences between them, followed by C. juncea (T3) and T. candida (T7), respectively. Phytomass production of C. breviflora (T5) was higher when compared to the other species assessed in the same period, with a value of 63.1 t ha −1 , but without statistical differences from C. mucronata (T2), which presented a value of 51.5 t ha −1 . Fresh biomass production of C. ensiformis (T4) was 40.40 t ha −1 . No significant differences (p<0.05) was observed between fresh phytomass production for C. cajan cv. IAPAR 43 (T1) (25.45 t ha −1 ), C. cajan cv. Fava Larga (T6) (27.5 t ha −1 ), and T. candida (T7) (17.85 t ha −1 ).
C. breviflora is a native and endemic species of Brazil, being resistant to drought, which may explain its fresh mass productivity. However, the contrast of phytomass production among the crotalaria species causes a great impact. The lowest productivity was obtained by C. juncea, with a value of 14.35 t ha −1 , a species considered as productive or recommended by its productivity in vegetal biomass, producing from 15 to 60 t ha −1 of fresh mass (WUTKE; CALEGARI; WILDNER, 2014).
Crotalaria species are annual plants, very well adapted to sandy textured soils with a low fertility (WUTKE; CALEGARI; WILDNER, 2014). Although the Coastal Tablelands soils present these characteristics, factors such as temperature, water availability, and photoperiod should be taken into consideration in order to obtain an adequate development and more efficient utilization in the different production systems (MASCARENHAS; WUTKE, 2014). According to these authors, although there are few studies on water stress in green manures, particularly in legumes, quantity and distribution of water are decisive factors for plant development, particularly in autumn/winter in regions without precipitation.
In experiments conducted in the Cerrado biome in Goiás State, Amabile, Fancelli and Carvalho (2000) observed an interaction between seasons and species on fresh phytomass production when changed periods from one considered as favorable (beginning of November) to others considered as marginal (January and March), decreasing vegetative stage in species such as C. ochroleuca, C. juncea, and C. cajan. The authors concluded that the delay in sowing time resulted in a linear reduction of fresh mass production for the studied species.
Phytomass production of jack bean was higher than in other studies WILDNER, 2014;OLIVEIRA et al., 2013;TEIXEIRA et al., 2005), proving that, despite a slow initial growth, this species has characteristics which make it adapted for the conditions of this test, including a deep root system, which confers resistance to drought (WUTKE; CALEGARI; WILDNER, 2014). For pigeonpea cultivars, the results corroborate with those obtained by other authors (WUTKE; CALEGARI; WILDNER, 2014), who described the pigeonpea as a species with a vigorous, well developed and deep root system, giving the plant resistance in long periods of drought, such as those faced during the development of this study.
Regarding shoot dry mass, significant differences (p>0.05) were observed between the studied legumes. The species with the highest dry mass were C. breviflora (T5), C. ensiformis (T4), C. mucronata (T2), and C. cajan cv. IAPAR 43 (T1), with values of 21. 75, 14.55, 13.50, and 13.25 t ha −1 , respectively, with no significant differences between them. On the other hand, those with the lowest values of dry mass of shoot were C. cajan cv. Fava Larga (T6), C. juncea (T3), and T. candida (T7), with values of 12.2, 9.8, and 7.25 t ha −1 , respectively, with no significant differences between them. Moreover, the species C. breviflora stood out and presented the highest average of dry mass production when compared to the other species. However, all the studied species presented higher values when compared to those observed by Wutke, Calegari and Wildner (2014) and Amado et al. (2014), under the no-tillage system, and Garcia (2002), under similar edaphoclimatic conditions. In this case, the only exception was observed for C. juncea probably due to the climatic factors observed during the experimental period of our study. Garcia (2002) recommends the use of perennial species just as T. candida for green manure in later phenological stages, since it may maximize phytomass production because this plant can produce annually about 6 to 8 t ha −1 of dry mass (CALEGARI; DONIZETI CARLOS, 2014).
The highest averages of production of DMR were obtained by C. mucronata (T2), C. cajan cv. IAPAR 43 (T1), and C. breviflora (T5), with values of 2.2, 2.0, and 1.95 t ha −1 , respectively, with no significant differences (p> 0.05) between them. The species C. juncea (T3) and C. cajan cv. Fava Larga (T6) had no differences from one another, with DMR values of 1.7 and 1.55 t ha −1 , respectively. On the other hand, the lowest values were observed for T. candida (T7) and C. ensiformis (T4), with values of 0.92 and 0.51 t ha −1 , respectively. In our study, plant roots were manually removed at a depth of up to 20 cm to obtain the fresh and dry mass. According to Calegari et al. (1992), the data on root production may be underestimated due to the difficulty in collecting the entire root system mainly because some legumes can penetrate several meters deep in the soil. However, the quantification of this attribute is valid because it is rarely accounted for in production systems using the green manure management. Wutke, Bulisani and Mascarenhas (1993) described the development of C. juncea roots at depths of up to 4.5 m at 130 days of age, with 79% of its weight within the first 30 cm. In spite of this possibility, an average of 1.7 t ha −1 of DMR for C. juncea was obtained in our study, which is higher than the value reported by Espíndola et al. (1998), who obtained a production of 0.57 t ha −1 of DMR for the same species.
In this study, we searched for a more detailed knowledge of some plants used as green fertilizers and their regional adaptation, where few research results are available for the Brazilian Northeast. This is not the end of the issue; there is still a need for continuity of research on green manuring aiming at recommendations of this management practice in the region's production units.

CONCLUSIONS
Under sandy soil conditions in the region of the Coastal Tablelands of Piauí State, the tested legume species showing the highest production of fresh mass during the experimental period and under local conditions were C. breviflora (63 t ha −1 ) and C. mucronata (51 t ha −1 ), followed by C. ensiformis (40 t ha −1 ). In addition, the highest production of dry mass was obtained for C. breviflora (21.7 t ha −1 ).