Studied area

The experiment was conducted in the municipality of Itapuranga, Goiás, at the Capoeira Grande Farm (15°34'32"S, 50°00'31"W) with an average altitude of 635 m.a.s.l. The climatic conditions in the region are a rainy season from October to April, with an average rainfall of 1600 mm, and average temperatures of 27 and 34 °C in the dry and wet season, respectively.

The climatic data of that region was considered for the development of this study (Figure 1). The meteorological data, from 2013 to early 2014, were obtained from the automatic station of the City of Goiás, located at 47 km from the experimentation site.

Figure 1 Monthly weather data: Total rainfall and average highest and lowest temperature, obtained from the automatic weather station (^{INMET, 2014}). 

The soil of the experimental area was classified as Red Latosol (^{Santos, 2013}) similar to Oxisol (^{Soil Survey Staff, 1999}). The experimental area was formed by irrigated vineyard (small sprinklers), with cv. Isabel grafted on IAC 572 ‘Jales’ grapevine rootstock, in trellis type conduction system spaced 2.5×2.5 m. At the time of the experiment development, the vineyard was two years old after grafting.

Experiment design and treatments

The experiment consisted of a randomized blocks design of six treatments established in a factorial arrange (3×2) with five repetitions. The first factor was cover plants species: Canavalia ensiformis L. DC, Dolichos lablab L. (lablab) and weeds. The second factor was two different pruning times in the vineyards, performed based on cover crop seeding time. That is, the first grapevine pruning season began 25 days after cover crops sowing (DAS) and the second at 55 DAS. Each plot 9 m² (2.0×4.5 m) was contained two grapevine plants.

Period of conduction

The experiment was conducted in two growing seasons. The period called “winter season” started in February and ended in August 2013, and the period called the “summer season” was from August 2013 to February 2014. In the winter season, the grapevine pruning was performed on March 2^nd, 2013 (first pruning, 25 DAS) and April 1st, 2013 (second pruning, 55 DAS) employing a long pruner, keeping five gems per grapevine stick. During the summer season, the grapevine pruning was held on August 31^st (first pruning, 25 DAS) and September 30^th, 2013 (second pruning, 55 DAS) employing a short pruner, keeping two gems per grapevine stick. After each pruning, bud dormancy breaking was conducted with hydrogenated cyanamide (5%), applied with a foam roller.

Three cycles of cover crops were evaluated: (i) sowing was done on February 5^th, 2013, (ii) the plants regrowth was evaluated after their management (mowing April 6^th, 2013), and (iii) a new sowing was conducted on August 6^th, 2013; monitoring their development within sixty days after sowing. Before each sowing, chemical control was conducted for existing weeds in all plots, using 3 L ha^-1 glyphosate. Sowing was done in furrows spaced 0.45 m and approximately 1 to 2 cm deep, performed manually, using five seeds of C. ensiformis and ten seeds of lablab, without any fertilization or seed inoculation. Weeds emerged from the soil seed bank.

The plots composed of the weed cover showed the following species in the first cycle: Bidens pilosa L. > Digitaria horizontalis Willd. > Euphorbia heterophylla L. > Commelina benghalensis L. > Siegesbeckia orientalis L. Weeds in the third cycle were: Bidens pilosa L.> Digitaria horizontalis Willd. > Amaranthus retroflexus L. > Sida rhombifolia L. > Euphorbia heterophylla L. > Commelina benghalensis L. The weeds in the second cycle were not determined.

Crop fertilization was scheduled during the winter and summer seasons (Table 1). In order to manage vineyard health in winter, products containing Metiram + Pyraclostrobin and Metalaxyl-M + Mancozeb and in the summer were sprayed with Metalaxyl-M + Mancozeb, Thiophanate-Methyl and Chlorothalonil + Azoxystrobin + Difenoconazole to prevent and control fungal diseases.

Table 1 Fertilizer application, in grams per grapevine plant, using nitrogen (N), phosphorus pentaoxide (P₂O₅), potassium oxide (K₂O) and micronutrient during winter and summer seasons. 

Nutritional status evaluation of cover crops

The aerial part of biomass samples was dried at 65 °C to determine cover crops’ nutrients at the time of the cuts. The method for chemical analysis was wet digestion of the dry samples according to the methodology of ^{Bataglia et al. (1983)}. Traces elements (N, P, K, Ca, Mg, Cu, Fe, Mn, and Zn) were determined in the aerial part of cover crops, reporting the total accumulation of traces as kg ha^-1.

Evaluation of nutrient release by cover crops

‘Litter bags’ were used to evaluate the plant decomposition (^{Thomas and Asakawa, 1993}). Four litter bags were placed randomly on the soil surface of each plot during the three cycles of cover crops. Sampling was performed at 20, 40, 60 and 80 d, and in each sampling, the litter bags were oven-dried at 65 °C until reaching a constant weight.

The parameters associated with the nutrient release dynamics were calculated based on the weight of dry residue remaining after 80 d of decay and the nutrient concentration in them. To describe the nutrient release from the plant straw, the exponential mathematical model X=X₀ ^(-kt) were adopted, where X is the amount of remaining nutrient that was presented after a time t (d), X₀ is the initial amount of nutrient, and k is a release constant (^{Thomas and Asakawa, 1993}). By reorganizing the equation terms, it is possible to calculate the release constant of nutrients (k) by the material, k=-ln (X/X₀)/t. With the value of k, the half-life (t_1/2) at which half of the nutrients contained in the residue will be released was calculated (T_1/2 =0.693/k) (^{Paul and Clark, 1989}).

Soil fertility evaluation

At the end of 2013 (summer season), four soil sub-samples were collected in each plot, with the help of a Dutch auger, to form a composite sample from the 0-0.20 m layer. In the laboratory, the organic matter (OM), pH, Cation Exchange Capacity (CEC), P, K, Ca, Mg, Al and potential acidity (H+Al) were determined, using the methodology of Embrapa (1997).

Grapevine nutritional status evaluation

The leaf nutritional diagnosis was made using grapevine leave at two different pruning times in each evaluated season. The winter sampling was held on March 31^st and April 30^th of 2013, counting as first pruning and second pruning, respectively. The summer season sampling was held on October 5^th and 27^th, first and second pruning, respectively. For the analysis, five full leave per plant were collected in each plot, totaling ten leave per sample. The leave were collected in the grapevine’s full bloom stage, opposite to the first bunch of the season’s branch. The leave were washed and dried in a forced-air oven at 65 °C until reaching constant weight; they were milled and prepared for the analysis of nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn), according to the methodology of ^{Bataglia et al. (1983)}.

The concentration levels recommended by Campinas Agronomic Institute were used for the vineyard leaf nutritional diagnosis (Table 2) (^{Terra and Tecchio, 2008}).

Table 2 Nutrient levels in gravepine leave collected during blooming. 

Source: ^{Terra and Tecchio (2008)}.

The collected data were submitted to ANOVA, and the means were compared using Tukey test at a 5% level.

Nutritional status of cover crops

The C. ensiformis biomass showed a higher accumulation of N in the first cycle, and P and K in the second cycle (Table 3). ^{Padovan et al. (2011)} found that C. ensiformis was efficient in cycling nutrients, especially immobilized N, K and Ca. It was also determined that C. ensiformis can accumulate N, K and Ca in concentrations of 415, 256 and 327 kg ha^-1, respectively, when evaluating extract nutrient capability from the soil through organic system production in the summer conditions (^{Saminêz et al., 2006}). These results reinforce its great potential as a cover crop and provide basic information for planning the management of plant biomass. ^{Bertin et al. (2005)} found higher total nitrogen content in Crotalaria, C. ensiformis, and lablab, statistically differing from fallow and millet. This result confirms the relationship established by ^{De-Polli and Chada (1989)} in which the N content is superior in more tender species.

Table 3 Nutrient accumulation in the canopy of cover crops, lablab (LB), C. ensiformis (CE) and weeds (W) in three consortium cycles and two grapevine pruning times. 

Lablab did not present statistical difference regarding N accumulation in the first cycle and P and K in the second cycle. Given these results, it is possible to infer that the weeds have benefited from good soil fertility, justifying by the efficient accumulation of nutrients in their biomass (Table 3). In a study of cover crops developed in Cerrado region (State of Maranhão), it was found that lower release of N by the spontaneous vegetation was probably due to the low amount of N in the residue, associated with the low dry plant matter decomposition rate, they also found lower accumulation of P in the biomass (^{Leite et al., 2010}).

^{Silva et al. (2002)} found, in orange orchard, that C. ensiformis and the lablab were the species that showed higher amount of macronutrient levels in the canopy, followed by velvet bean that stood out in N and P levels, Crotalaria spectabilis for K and Ca, and dwarf velvet bean for N and S. According to ^{Souza et al. (2012)}, the content and amount of nutrient uptake by the cover crops species can influence the decomposition of plant material and the performance of the subsequent cultivation. In several studies involving cover crops, the amount of accumulated nutrients depends on the species, the phenological management stage, and the climatic conditions. Regarding the micronutrient accumulation, a significant effect was observed for the canopy corresponding to Cu and Zn levels, the weeds accumulated the lowest amount of these micronutrients, and C. ensiformis presented the highest levels, not significantly differing from lablab (Table 3).

Copper is an essential element for plants, but in high concentrations can cause toxicity, which may extend to man and animals that consume copper contaminated food. Cover crops can be an alternative for the mitigation of copper excess in the soil, which by decomposing provide an increase of the straw amount, among which there is the organic material capable of promoting the immobilization of the available copper and decrease its presence in soil (^{Albarello et al., 2013}).

^{Cavalcante et al. (2012)} found that the plants evaluated for cover crops had, among the micronutrients, high accumulation of Fe and Mn. ^{Duarte and Coelho (2008)} observed that the legumes Crotalaria sp, C. ensiformis and velvet bean (Mucuna sp) extracted higher amounts of P, Ca, Mg, S, Zn, and Fe than weeds. In this study, the levels of P, K, Ca, and Fe were higher in the plants grown in the first pruning time and the first cover crop cycle than the other two cycles.

Release of nutrients by cover crops

The nutrient release parameters are not shown for Mn in the first cycle, and Fe in the three cycles because their contents were higher than the initial after 80 d of decomposition, indicating that sample contamination might have occurred by soil residues. The N, P, K, and Mg release parameters were significantly higher in the first pruning time, showing a higher amount of nutrients released, constant decomposition and shorter half-life (Table 4).

Table 4 Quantity released (kg ha^-1), release constant k (d^-1) and half-life (t1/2) for nutrients of cover crops, lablab (LB), C. ensiformis (CE) and weeds (W), after 80 d in decomposition in the first consortium cycle and two grapevine pruning times. 

In the first cover crop cycle, there was a significant difference in the release constant (k) and half-life of Zn, especially in the C. ensiformis, that presented the highest release constant (0.0203 d^-1) of this nutrient and shorter half-life (35.11 d), but not differing from the lablab, which in turn does not differ from weeds (Table 4).

In the second cycle, there was a significant difference for the amount released and release constant for N, P, K, Ca, Cu, and Zn; the half-life was significant for N, P, K, and Ca (Table 5). Generally, lablab performed better in the release of nutrients for Ca and Zn. The release of P was higher for C. ensiformis (24.99 kg ha^-1), with a half-life of 43.46 d, revealing its efficiency in recycling this nutrient because it also showed higher dry P in the biomass (Table 5). Gamma-Rodrigues et al. (2007) found that N, P, Ca, and Mg release rates were higher in the C. ensiformis compared to Arachis sp, siratro, tropical kudzu, and weeds. ^{Calonego et al. (2012)} also found that the lablab straw was efficient in N, P, and K release.

Table 5 Quantity released (kg ha^-1), release constant k (d^-1) and half-life (t1/2) for nutrients of cover crops, lablab (LB), C. ensiformis (CE) and weeds (W), after 80 d in decomposition in the second consortium cycle and two grapevine pruning times. 

In the second cover crop cycle there were significant differences between pruning times for the decomposition constant (k) of Mg, Cu, and Zn, and for the half-life of Mg and Cu. In the third cycle was found a significant difference only for the half-life of Cu. The lablab showed higher resistance to the release of this nutrient, with a half-life of 46.37 d. The half-life for Cu was also influenced by the grapevine pruning times, as the first time presented a long duration, 42.48 d. For K, the grapevine pruning time influenced the amount released with higher amounts in the first season (Table 6).

Table 6 Quantity released (kg ha^-1), release constant k (d^-1) and half-life (t1/2) for nutrients of cover crops, lablab (LB), C. ensiformis (CE) and weeds (W), after 80 d in decomposition in the third consortium cycle and two grapevine pruning times. 

In the first and third cycles, there is a similar behavior in the nutrient release parameters where the plants do not differ significantly for most nutrients. These results can be explained by the fact that the plants of the second cycle grew longer until management (105 d), while in the first and third cycles the plants grew no more than 60 d. Therefore, the age of the plant can be associated with its nutrient composition (^{Souza et al., 2012}). While the young plants are more tender and have a higher decomposition rate, older plants have most of their parts lignified and, therefore, are more resistant to decomposition and present a consequent lower release of nutrients. Besides, the first and third cycles coincided with periods of the year with higher temperatures and higher rainfall, favoring the decomposition of the straw. For the second cycle, although there was irrigation, and the thermal amplitude was lower, which may decrease the decomposition activity of the straw.

Soil fertility

There was no significant difference among the cover crops for soil chemical attributes after three cultivation cycles of two grapevine crops (Table 7).

Table 7 Chemical attributes of Red Latosol (Oxisol) before experiment installation (collection conducted March 8^th, 2013) and after three cycles (collection on December 16^th, 2013) with the cultivation of grapevine intercropped with cover crops. 

Comparing the values before and after the cover crop, the increase of P, K, Ca, and H+Al level is clear. According to ^{Collier et al. (2011)}, this is due to possibly intake of these nutrients after decomposition of the previous straw. ^{Silva et al. (2002)} also observed Ca increase after the implementation of intercropped leguminous family species in an orange-pear orchard, compared to the ground situation before the experiment. ^{Faria et al. (2004)}, using leguminous cover crops with vineyards under Ultisol in Petrolina (state of Pernambuco) after eight years, noted several improvements in the chemical characteristics of the soil, including an increase the exchangeable Ca in the 0-10 cm depth compared to the control without cover crops. ^{Rosa et al. (2009)} found that acidity and nutrient availability in the soil were influenced by cover crops associated with the grapevine, in the mountainous region of the state of Rio Grande do Sul.

^{Nascimento et al. (2003)} studied the effect of several tropical herbaceous legumes, cultivated as cover crops, on the chemical characteristics of a degraded Luvisol. According to their findings, it was observed significant effects of the legumes on soil fertility with significant increases in pH and exchangeable bases, positively reflecting on the CEC and base cation saturation index. Despite the cover crops promoting discrete soil acidification by raising H+Al levels and reducing organic matter levels, there was no increase in exchangeable Al that remained null in all soil samples.

The K content in soil increased 19.7 mg dm^-3 in the plots cultivated with lablab, 24.4 mg dm^-3 with C. ensiformis and 6 mg dm^-3 with weeds. The P content in the soil also increased reporting values of 5.03 mg dm^-3, 5.53 mg dm^-3, and 1.23 mg dm^-3 in the plots cultivated with lablab, C. ensiformis, and weeds, respectively. On the other hand, ^{Cardoso et al. (2013)} found that the P content in the soil increased by 0.6 mg dm^-3 when cultivated with C. ensiformis and millet. According to the authors, this P increase may be related to the ability of these plants to absorb the P subsurface soil layers and make it available on the surface, after the decomposition of straw. Such association can also be attributed to an element of easily leaching such as K, and plants with deeper roots can cycle this nutrient.

Negative effects were observed for Mg, soil organic matter, and basic cation saturation; showing decreasing levels in the soil at the end of the experiment. The organic matter content was reduced in the plots with weeds from 38 g dm^-3 to 13 g dm^-3. For lablab, the reduction was lower, reflecting the effect of higher biomass production by these plants, thereby maintaining a good level of organic matter in the soil. ^{Collier et al. (2011)} found, in treatment with C. ensiformis, decrease of organic matter in soil because of a positive priming effect, to stimulate the soil biota in the decomposition of the existing organic matter. The activating effect (priming) is defined as the rapid change of the organic carbon and nitrogen content of the soil. It can be positive (mineralization of C and N) by adding low C/N ratio materials or nitrogen mineral fertilizers. Otherwise, this effect can be negative (net immobilization) by the addition of high C/N material (^{Buso and Kliemann, 2003}).

Nutritional status of grapevine

Cover crops did not affect the nutritional status of the grapevines at blooming in both crop seasons (Table 8). A different outcome was noticed by ^{Zalamena et al. (2013)}, who verified lower content of P and K in the leave of grapevines planted with species of cover crops compared to the control treatment (weeds). This reduction can be attributed to the higher absorption and accumulation of both elements in the tissue of cover crops, reducing the availability in the soil for grapevine plants (^{Celette et al., 2009}; ^{Brunetto et al., 2011}).

Table 8 Nutrients in grapevine canopy intercropped with, lablab (LB), C. ensiformis (CE) and weeds (W) and grapevine pruning times. 

According to ^{Celette et al. (2009)}, along with the grapevine cycle, the cover crop plants also absorb the water and nutrients from the soil solution, especially N, which may even reduce the availability of this element to the grapevine. Thus, increase in total N content in the leave of grapevines intercropped with annual cover crops is not normally expected, and may even be the opposite, as ^{Wheeler et al. (2005)} noted.

N, P and K levels were higher in the summer season compared to winter. This increase can be attributed to the fact that grapevine plants may have benefited from the nutrients released by the decomposition of previous cover crops and soil organic matter at the second harvest since there had already been two cycles of cover crops. It can be inferred that the use of cover crops over time provides better nutrient availability in the soil, with consequent benefits for the main crop. ^{Faria et al. (2004)} studied changes in soil characteristics after eleven legume cycles with nine grapevine crops seasons and noticed a soil fertility improvement in the sixth and ninth seasons.

For grapevine, and most of the crops, the standard levels of nutrients that are correlated with the higher production are not well established, but it is possible to work with a concentration range for the interpretation of the results. The concentration ranges recommended by the Agronomic Institute of Campinas for grapevine plants are divided into five levels: deficiency, slight deficiency, normal, slight excess and excess (Table 2) (^{Terra and Tecchio, 2008}).

The grapevine plants showed a slight N deficiency in the winter season, however, in the summer season, the N content was in the optimal (normal) range, reinforcing the use of this nutrient arising from the decomposition of cover crops and soil organic matter. The plant P content in the winter and summer season showed a slight excess. According to ^{Mafra et al. (2011)}, the grapevine has a low demand for P, that is attributed to the association of grapevines with mycorrhizal fungi present in the roots of plants in poor soils, which exploit little soluble forms of this element. However, this is not the case of the present work, because the P content in the soil presented as low to medium, according to ^{Sousa et al. (2004)} (Table 6), so if there was a mycorrhizal association, it might have contributed to increasing the absorption of this nutrient.

The K content in the grapevines, for the two crops, was framed within the slight deficiency range. Grapevine leave showed excess of Ca in the two seasons, except for treatment with the weeds in the winter crop, which was in slight excess range. The scarcity level was observed for Mg in the two seasons, regardless of cover crops. The dynamics of these three nutrients in vineyards are very important, and the relationship between the nutrients, such as K/Mg and K/(Ca+Mg) should be considered. When there is an inverse relationship between these elements, especially high content of K and low Mg and Ca, an abiotic anomaly known as “desiccation of the rachis” can occur (^{Fráguas et al., 1996}; ^{Miele et al., 2009}). According to ^{Silva et al. (2005)}, high K, high Ca and low Mg levels also contributed to the emergence of desiccation of the rachis.

For micronutrients in cover crop treatments, the Cu was framed as under slight deficiency in two crops seasons. The Fe content was in the excess range in the winter season and a slight excess in the summer season. The Mn content fell in the slight excess range in two crop seasons. A slight deficiency was observed for Zn content in grapevine leave, in two crops seasons.