Introduction

Cover cropping with nitrogen (N) fixing forage legumes is a technology increasingly being used in grapevine (Vitis vinifera L.) production systems in various wine producing regions of the world, such as the USA, Australia, Italy, France and South Africa (Miller et al. 1989; Masson and Bertoni 1996; Nieddu et al. 2000; Masson and Gintzburger 2000; Fourie et al. 2006). Cover cropping in vineyards can potentially provide a range of benefits. These include a reduced risk of erosion, enhanced biological activity in soil, increased soil organic matter content and nutrient availability, and improvements in soil porosity and aggregate stability, and increased soil water holding capacity (Frye and Blevins 1989). There have also been reports of cover crops reducing weed incidence and improving the control of some nematodes species (Aballay and Insunza 2002). Cover cropping has only recently been introduced into vineyards in Chile, and their effects have not been widely studied. In particular, little is known about how important N derived from legume cover crops may be as a means of either replacing or supplementing annual applications of N fertilizer to vines. Predicting the impact of an increased N supply on grapevines is difficult due to a poor understanding of the quantitative response of vines to N, and the temporal patterns of both the partitioning of N within the annual above-ground vegetative plant parts and fruits, and the storage and remobilization of N from perennial structures during the growing season (Peacock et al. 1989; Wermelinger and Koblet 1990; Treeby and Wheatley 2006).

Forage legumes species differ in their capacity to symbiotically fix atmospheric N2, and in the N content of their stems, leaves and roots (e.g. Ledgard and Steele 1992; Peoples and Baldock 2001; Campillo et al. 2003; Peoples et al. 2004), and consequently different legume cover crops may also vary in the amounts of N that could potentially be returned to the soil in a vineyard. There is a wide range of productive fodder legumes that could be suitable as cover crops in Chile. For example, in low rainfall areas (<400 mm annual rainfall), hualputra (Medicago polymorpha L.) has a wide variability in phenology, biomass production and N2 fixation (del Pozo et al. 2000, 2002; Ovalle et al. 2006). In areas of high rainfall, late maturing cultivars of subterranean clover (Trifolium subterraneum L.), balansa clover (Trifolium michelianum Savi), yellow serradela (Ornithopus compressus L.), and pink serradela (O. sativus Brot.) could be utilized with this purposes (Avendaño et al. 2005; Ovalle et al. 2005).

Numerous studies have demonstrated improvements in crop N status and production benefits from using annual (e.g Dou et al. 1994; Guldan et al. 1997a, b; Rochester and Peoples 2005) or perennial legumes (e.g Ovalle et al. 2007a, b; Sanchez et al. 2007) as cover crops, mulch or green manure. Rhizodeposition of N via exudates and secretions from living legume roots is one potential pathway that vines could benefit from cover crops (Wichern et al. 2008), and the direct transfer of N via mycorrhizal hyphae connections between the roots of legumes and their neighboring vines might be another (Haystead et al. 1988). However, while some researchers have found evidence of a transfer of N from legumes to a companion species during growth (Høgh-Jensen and Schjoerring 2000; Xiao et al. 2004), others have failed to detect direct below-ground transfer (Ledgard et al. 1985; Giller et al. 1991; Trannin et al. 2000). The main source of plant-available forms of legume N is generally believed to result from the decomposition and mineralization of organic legume residues at the end of the growing season following the death of the legume (Danso et al., 1993; Trannin et al. 2000; Fillery 2001). This would particularly be the case for forage legumes in the absence of grazing animals (Peoples et al. 2004).

The rate by which legume organic residues are mineralized to plant-available forms of soil mineral N (nitrate and ammonium) depends on the chemical composition of the legumes, residue management and soil properties (Ladd et al. 1985; Fillery 2001; Peoples et al. 2004). In annual cropping systems <20–25% of the N in forage legume residues or green manures is commonly recovered in the short-term, although legume N can sometimes represent up to 30–33% of the N requirements of a following crop (Fillery 2001; Crews and Peoples 2005; Peoples et al. 2009). However, much less information is available in systems with perennial cropping systems. In an organic raspberry (Rubus idaeus L.) production system where white clover (T. repens) was used as a cover crop, recoveries of legume N by the raspberry plants were estimated to have ranged from 2.5–7.0% in fruits to 18–22% in canes and leaves using an indirect 15N dilution method (Ovalle et al. 2007a, b). In a coffee (Coffea arabica L.) plantation, 15N studies also suggested that 30% of the N2 fixed by a legume cover crop may be transferred to the coffee plants (Snoeck et al. 2000).

Ovalle et al. (2007c) has previously demonstrated that differences in biomass production can be achieved by mixtures of early maturing cultivars (EMC) of subterranean clover and burr medic or mixtures of late maturing cultivars (LMC) of subterranean clover and balansa clover. These authors also reported that annual legume cover crops could significantly increase the concentrations of plant-available soil N in vineyards in the sub-humid Mediterranean zone of Chile, but no data were reported concerning their impact on vine productivity. The objective of the present study was to evaluate the relative effect of EMC and LMC legume cover crops on a number of growth and production variables of the vineyard, to estimate the size of the contributions of the legumes to the N nutrition of grapevines, and to compare this to current practices of applying 40 kg fertilizer-N ha-1 to vineyards each year, usually in the form of ammonium sulfate.

Materials and methods

Experimental site and treatments

The experiment was carried out in the Experimental Center of Cauquenes-INIA (35°58’ S, 72°17′ W; 140 m above sea level), Maule Region, in the eastern part of coastal mountain range of Chile. Long-term average minimum temperatures of the coldest month (July) is 4.8°C and maximum temperature in warmest month (January) is 29°C. Data of rainfall was obtained from an automatic meteorological station located in the same Experimental Center: total rainfall was 806, 745 and 406 mm, in spring (September-December) was 88, 110 and 47.6 mm, and in summer (January-April) was 28, 58 and 80.1 mm, in 2005, 2006 and 2007, respectively.

The study was undertaken between May 2005 to June 2007, in a 5-yr old vineyard of cv. Cabernet Sauvignon, planted at 2.5 m row width × 0.5 m spacing between vines on a shallow, hilly granitic soil (fine, kaolinitic, mesic Typic Palexeralfs). At the beginning of the study soil pH (water) in the top 10 cm was 6.7, organic matter (OM) was 1.5%, and the levels of availability for N (2 M KCl), P (0.5 M NaHCO3) and K (1 M NH4OAc) were 1.93 mg kg-1, 5.47 mg kg-1, and 0.53 cmol(+) kg-1, respectively.

The experimental treatments were: control without cover crop and the ground kept bare (C), legume cover crops consisting of either a mixture of early maturing cultivars (EMC) of subterranean clover (Trifolium subterraneum L.) and burr medic (Medicago polymorpha L.), or late maturing cultivars (LMC) of subterranean clover and balansa clover (Trifolium michelianum Savi). Four replicate plots (25 m x 4 m) for each legume treatment were sown on 31 May 2005 in a randomised block design. Table 1 presents details of cultivars, seed rate, pasture establishment and management. The vineyard received supplementary drip irrigation (30 L plant-1 week-1) from September to March in each year. All treatments were irrigated with the same water source. Although no analyses of the irrigation water were undertaken during the experiment, the water came from rivers flowing from the mountains that are not subject to nitrate contamination and irrigation water was believed not to represent a significant external input of N for the vines.

Table 1 Treatments, legume species and cultivars, and seeding rate used for cover crops
Full size table

Labeling the soil N pool with 15N

To allow the uptake of legume N by vines to be calculated the soil N within “micro-plot” areas (3 m x 2 m on the row) was labeled using 15N enriched (NH4)2SO4 (10% atom excess 15N). The 15N-enriched fertilizer was added at a rate of 40 kg N ha-1 as two applications: 50% at leaf emergence (September 2006), and 50% during fruit set (December 2006). Labeled fertilizer was applied to four vines inside each micro-plot as a solution diluted 1% (weight / volume) in water to the soil surface in the area of soil moistened by drip irrigation (2 m x 0.5 m) on either side of the row of planted vines. Micro-plots were surrounded by a raised soil barrier to prevent run-off and lateral movement of 15N, and each micro-plot area was delineated with wooden stakes.

Legume biomass harvest and tissue analyses

The above-ground biomass production of the legume cover crops was determined using five quadrants of 1 m × 0.5 m per plot, distributed at both sides of the vineyard rows. In the first growing season cover crops were sampled on 31 October 2005, and in the second season above-ground biomass was determined on 15 August, 21 September and 3 November 2006. Samples were dried in forced ventilation oven at 65°C for 48 h and the dry weight recorded. After sampling, the remainder of the cover cropped area was mown to a height of around 5 cm and the cut foliage was left as a mulch. The N concentration of the legume dry matter was determined using Kjeldahl digestion.

Vine sampling, analyses and calculations

Vine leaves were sampled in December 2006 and February 2007 (at the time of maximum canopy development), dried in an oven with forced ventilation at 65°C for 48 h. Number of bunches and grape production was evaluated for five plants per replicate at the time of fruit harvest in April 2006 and 2007. A destructive sampling of the whole plant (included roots) was undertaken in five plants per replicate at the end of the growing season in June, 2007 for determination of total biomass and N content. The 15N/14N isotopic ratio was determined for various plant parts sampled from the two central vines from each micro-plot area. Total N in vine tissues were determined using Kjeldahl digestion, while analyses for 15N composition were undertaken using an optical emission spectrometer 15N analyzer NOI-6PC (FAN, Germany), as described by Axmann (1990).

The recovery of legume N by grapevines was estimated using an indirect 15N dilution approach that has been applied in similar studies in other cropping systems (e.g. Ofosu-Budu et al. 1995; Trannin et al. 2000; Hood 2001). With this technique the 15N concentration in the tissue of vines growing within the 15N-labeled micro-plots with legumes were compared to vines growing in micro-plots without cover crop, and the percentage of the vine N derived from the legume (Ndfl%) was calculated using Eq. 1:

$$ Ndfl\% = 100 \times \left[ {1 - \left( {Vine{ }atom{\%^{15}}N{ }excess{ }with{ }legume/Vine{ }atom{\%^{15}}N{ }excess{ }without{ }legume} \right)} \right] $$
(1)

Where atom %15N excess represents a measure of the enrichment of vine samples in 15N relative to the background concentration of 15N occurring naturally in atmospheric N2 (0.3663 atom %15N):

$$ Vine{ }atom{\%^{15}}N{ }excess = \left( {Vine{ }sample{ }atom{\%^{15}}N} \right) - 0.{3663} $$
(2)

The calculation using Eq. 1 essentially assumes that all of the N generated by the legume came from symbiotically fixed atmospheric N2 which by definition has an atom %15N excess value of zero. In the case of the current experiment with grapevines where the demand for N by the various annual above-ground plant parts may be partially met from the remobilization of N reserves from perennial structures in addition to soil mineral N (Treeby and Wheatley 2006), it was assumed when comparing 15N-enrichments of individual plant organs of vines that the extent of utilization of older, unlabeled N reserves for growth was identical in all treatments.

The amount of vine N originating from legume cover crops was calculated from Ndfl% and measures of N accumulated by each of the individual plant parts (Vine N) using Eq. 3:

$$ Vine{ }N{ }from{ }legume = Vine{ }N \times \left( {Ndfl\% } \right)/100 $$
(3)

The 15N enrichment data could also be utilized to estimate the proportion of the vine N derived from the 40 kg fertilizer-N ha-1 applied to the micro-plots (Ndff) in both the absence and presence of cover crops using Eq. 4:

$$ {\hbox{Ndff}}\% = {1}00 \times \left[ {\left( {Vineatom{\%^{15}}N{ }excess} \right)/\left( {atom{\%^{15}}N{ }excess{ }of{ }applied{ }fertilizer} \right)} \right] $$
(4)

Since the 15N-enriched fertilizer was applied as 10 atom%15N excess ammonium sulfate Eq. 4 could be simply re-expressed as:

$$ Ndff\% = 10 \times \left[ {\left( {Vine{ }atom{\%^{15}}N{ }excess} \right)} \right] $$
(5)

The amount of vine N originating from fertilizer was calculated from Ndff% and measures of amounts of N accumulated in different plant parts of the vines using Eq. 6:

$$ Vine{ }N{ }from{ }fertilizer = Vine{ }N \times \left( {Ndff\% } \right)/100 $$
(6)

Statistical analyses

The experimental design was complete random blocks with 3 treatments and 4 replications. The statistical analysis was carried out using the statistical program SAS system. T Student’s test and ANOVAs were performed, on pasture and vine biomass, tissue N concentration and 15N data.

Results

Legume cover crops

Leguminous material represented 95–98% of the total above-ground dry matter (DM) that grew in the cover crop treatments in 2005-06 and 99-100% in 2006-07. In both growing seasons, DM production by LMC was significantly higher (P ≤ 0.05) than the EMC (Table 2). The N concentration of the legume biomass was 2.90 and 3.15%, in the EMC and LMC, respectively, but this difference between the two legume mixtures was not significant (P ≥ 0.05). Above-ground inputs of N by the legume cover crops represented 68 and 128 kg N ha-1 in the EMC in 2005-2006 and the 2006-2007 growing seasons, respectively (average 112 kg shoot N ha-1 yr-1), and 157 and 194 kg N ha-1 in the LMC treatment (Table 2, average 161 kg shoot N ha -1 yr-1).

Table 2 Above-ground dry matter (DM) and N content of legume cover crops in a vineyard established in a granitic soil, in two growing seasons
Full size table

Biomass production, N content in vine plants

Grape production for vines growing in association with EMC was lower than those in the control in the first growing season (2005-2006), but was 48–61% higher than the control for both the EMC and LMC treatments in the second growing season (P < 0.05, Table 3). However, despite this improvement in fruit dry matter in response legume cover cropping, determinations of cane, trunk and root biomass were similar (P < 0.05) regardless of whether vines were grown with cover crops or not (Fig. 1). As a net result total vine biomass under cover cropping was only around 15% greater than the control (Fig. 1). The annual components of vines (leaves, cane and fruit) represented 41–44% of the total recovered biomass.

Table 3 Number of bunches per plant and grape biomass production for grapevines growing with or without cover crops in two growing seasons
Full size table
Fig. 1
figure 1

Dry matter production (g plant-1) of the components of vine plants growing with or without cover crops in 2006-2007. Leaves were sampled in February (at maximum canopy development), fruits on April (at harvest) and canes, trunk and roots in June 2007. Means followed by different letters in a column are significantly different according to Duncan test (P ≤ 0.05)

Full size image

The N concentrations measured in 2007 were higher than the control in grape bunches under both cover crops (P ≤ 0.05), and in the canes and trunk of the LMC treatment (Table 4). The total amount of N accumulated by vine plants by the end of the second growing season, was also significantly (P ≤ 0.05) higher in treatments with cover crops (8.0 and 9.1 g plant-1 in EMC and LMC mixtures, respectively) than in control (6.9 g plant-1; Table 4).

Table 4 Determinations of N concentration and N content in above-ground and root biomass for grapevines grown with or without legume cover crops in 2006-2007
Full size table

The annual components of the vines represented 46–51% of the total plant N (Table 4). The largest amounts of N accumulated in all treatments were found to be in the leaves and trunk (each containing between 2.1–2.8 g N plant-1), with slightly lower amounts being recovered in the roots (1.3–2.0 g N plant-1; Table 4).

Recovery of legume N by vine plants

The overall enrichment of 15N measured in the roots and leaves were lower than that detected in the canes and fruit in all treatments (Table 5). This presumably reflected the presence of unlabelled N accumulated in root structures during previous growing seasons, and suggested that leaves had a higher dependence upon remobilization of older sources of N from perennial tissues for growth than uptake of newly assimilated soil N relative to other plant parts (Treeby and Wheatley 2006).

Table 5 Measures of 15N enrichment (atom% 15N excess) in various plant parts of grapevines and estimates of the proportion of the N in those plant parts derived from the legume cover crop (Ndfl)
Full size table

The significantly lower 15N-enrichments detected in vines growing with cover crops than the control provided evidence that some legume N was utilized for growth by all plant parts in the cover cropped treatments (Table 5). Estimates of the proportion of the vine N derived from the legumes (Ndfl%) showed that vine plants growing with the late maturing mixture of annual legumes (LMC) generally had a higher reliance upon legume N for growth than vines with the EMC cover crop (Table 5), which was consistent with the larger inputs of legume N in the LMC treatment (Table 2). Analyses indicated that Ndfl ranged from 7–13% of the N in grape bunches, to 39–56% of the N in roots in the EMC and LMC treatments, respectively. Legume N provided between 14 and 20% of the N in vine canes and leaves (Table 5).

The total contribution of legume cover crops to the N nutrition of vine plants was calculated from the 15N data of individual plant parts to be 1.46–1.92 g N plant-1 or about 12–15 kg N ha-1 with no significant differences between the two cover crop treatments (Table 6). This represented between 18–21% of the N requirement of the grapevine (compare the total plant data in Table 4 with Table 6), and a recovery of <10% of the N annually returned to the soil in above-ground legume residues. The N contribution of legume cover crops can also be estimated by N-difference for comparative purposes from the total plant N data presented in Table 4 (see Hood 2001). Using this procedure the N provided by legumes was calculated by subtracting the N accumulated by the control vines in the absence of legumes (6.92 g N plant-1) from the N content of vines grown with the EMC (8.03 g N plant-1) and LMC treatment (9.09 g N plant-1; Table 4). The resulting additional N in the cover cropped vines attributable to legumes (1.11 and 2.17 g N plant-1) was equivalent to 14% and 24% of the N requirements of EMC and LMC grown vines, respectively.

Table 6 Estimates of the amounts of N recovered by grapevines from legume cover crops or fertilizer
Full size table

Recovery of fertilizer N by vine plants

Since the same amount of 15N-enriched fertilizer was applied to micro-plots in all treatments as is currently applied to vineyards, the 15N data presented in Table 5 could also be used to determine the proportion of the vine N derived from fertilizer (Ndff%) that might be achieved following standard fertilizer recommendations. Estimates of Ndff% indicated that between 8% (LMC roots) and 27% (Control fruits) of the N present in different plant parts came from fertilizer. Total contributions fertilizer N were calculated to be between 1.34 and 1.50 g N plant-1, equivalent to 11-12 kg N ha-1 in both the absence and presence of legume cover crops (Table 6). These values represented 27–30% recovery of the 40 kg N ha-1 fertilizer N applied during the growing season.

Discussion

Productivity and N inputs by legume cover crops

The amount of biomass accumulated by the legume cover crops in the second year of the project was comparable to, and in some cases, greater than, DM yields previously reported for the same species under permanent pasture conditions in the same agro-ecological area (Avendaño et al. 2005; Ovalle et al. 2005) and represented above-ground inputs of 157–194 kg legume N ha -1 for the EMC and LMC treatments, respectively. Since up to 40–50% of the total N in a forage legume may be below-ground associated with rhizodeposition of N and the N contained in the nodulated roots (Peoples and Baldock 2001; Unkovich et al. 2010), the total annual input of legume N by the cover crops in 2006-2007 could have represented up to 300–400 kg N ha-1.

Three possible sources of N could have contributed these large amounts of N generated by the legume cover crops: (a) soil mineral N, (b) N-fertilizer, and (c) biologically fix atmospheric N2. While specific measures of the reliance of the legumes upon N2 fixation for growth relative to soil or fertilizer N were not undertaken, similar studies in the granitic soils of the region have previously indicated that >90% of the N contained in subterranean clover and burr medic can be derived from N2 fixation (Ovalle et al. 2006). Data collected in other forage legume systems also suggest that rates of N fertilizer up to 50 kg N ha-1 can have relatively little effect N2 fixation (Ledgard and Steele 1992), so it would appear unlikely that the two split applications of 20 kg fertilizer-N ha-1 supplied to the vineyard would have been sufficient to significantly suppress N2 fixation by cover crops assimilating several hundreds of kg of N. Therefore, it seems reasonable to suggest that the majority of the N accumulated by the legumes in the present study also came from N2 fixation.

Impact of cover crops on the N-economy of vines

The increased grape production and N concentration of vines growing in association with legume cover crops described here has been also reported elsewhere in the world. For example, a long-term (10 year) experiment undertaken in South Africa with Chardonnay/99 Richter vines growing in association with single leguminous species (either Vicia faba, Medicago spp., or T. subterraneum), found greater grape production, and higher N concentrations in leaf petioles and in grape juice than those in vines grown without cover crop (Fourie et al. 2006). The results are also not dissimilar to that reported for organic raspberry after 3 years growing in association with a white clover cover crop, where fruit yield tended to be higher, although not statistically, from the control without cover crop (Ovalle et al. 2007a). However, there can be delays in the vine’s response to the increased inputs of N so the long-term production benefits of cover-cropping could larger than reported here. For example, an Australian study indicated that at least two growing seasons were required before grape yields were improved by increased applications of fertilizer N (Treeby and Wheatley 2006).

Between 27–35% of the vine N was partitioned in leaves. This was similar to the findings of a grapevine investigation undertaken in Europe (Wermelinger and Koblet 1990), but contrasts with a North American study conducted in a 23 year old Pinot noir vineyard which found relatively larger amounts of N in the roots and trunk than leaves (Schreiner and Scagel 2006), indicating that older and bigger vines may have the capacity to store more N in root and trunk reserves than measured in the present trial in a 5 year old vineyard.

The rate of N uptake by vines increases from bud break and reaches a maximum close to bloom and fruit set during late spring—early summer (Wermelinger and Koblet 1990; Schreiner and Scagel 2006). The soil temperature and water conditions most favorable for rapid mineralization of the N in organic matter are also likely to occur in spring (Mengel 1996; Fillery 2001; Peoples et al. 2009). As a consequence, the supply of plant-available N from the legume residues might be expected to coincide with the time of peak N demand by the grapevine (Ovalle et al. 2007c; Fourie et al. 2007), and should theoretically be conducive with high efficiencies of N uptake by vines and result in a lowered risk of N losses from the system (Crews and Peoples 2005). Certainly the concentrations of soil mineral N (on the top 20 cm), measured on four occasions between August and November 2006, were found to be significantly higher under the LMC (14–194 mg N kg-1) than in EMC treatment (22–63 mg N kg-1), with soil mineral N under both cover crop treatments being greater than the control (2–10 mg N kg-1; Ovalle et al. 2007c). The maximum concentrations soil mineral N where observed in September i.e. during budbreak and early canopy development of grapevine.

The indirect 15N dilution method applied here to determine the recovery of legume N by grapevine has also been used in other cover crops studies (Snoeck et al. 2000; Ovalle et al. 2007b) including grapevine (Patrick et al. 2004). However, it should be noted that the estimates of the relative importance of legume N to the N nutrition of the various vine tissues presented in Tables 5 and 6 calculated using Eq. 1 will inevitably be conservative since it was assumed that all the legume N taken up the vine had been derived from atmospheric N2. Based on past experience with legumes grown on similar soil types one might expect that most, but certainly not all, of the cover crop N to have come from N2 fixation (Ovalle et al. 2006). So while it seems likely that the 15N-enrichment of the legume N would have been low, it was unlikely to have been zero. Consequently, the data in Tables 5 and 6 could undervalue the true contribution of legume N to vines. There may also be inherent methodological problems with the use of the indirect 15N dilution technique, and with the application of 15N-enriched fertilizer to label the soil N pool, which can lead to the uptake of legume N being underestimated (Høgh-Jensen and Schjoerring 2000; Hood 2001; Peoples et al. 2009). Nonetheless, it was encouraging in the present study that similar estimates of the contributions of legume N to vine nutrition were obtained using both the 15N dilution approach (18–21% of the vine’s N requirements) and an alternative, non-isotopic, N-difference method (14–24%).

The percentage of the vine leaf N calculated to have been derived from the cover crops (14–20%, Table 5) was considerably higher than reported by Patrick et al. (2004) who estimated only 0.28% of N in vine leaves to have come from legume N using similar 15N methodologies. However, much less legume N was applied (81 kg ha-1) in the study described by Patrick et al. (2004) and the rainfall during their experiment was only 45 mm. The drier soil would undoubtedly have restricted decomposition and mineralization of the organic legume residues to a greater extent than under the soil conditions experienced in the current experiment.

Importantly, the data in Table 5 indicated that despite the fruit being the main organ to increase biomass (Fig. 1) and N accumulation (Table 4) in response to legume cover cropping, it was the plant tissue that the 15N data suggested had the lowest reliance upon of legume N for growth (7–13% compared to 14–20% for leaves, 15–19% for canes and 39–56% for roots; Table 5). It is interesting to speculate on possible explanations for this result. Since the leaves and canes largely assimilate much of their N prior to flowering and the developing grape bunches only become a sink for N after flowering (Wermelinger and Koblet 1990; Treeby and Wheatley 2006), one possible explanation for the observation may be that the relative contribution of legume N to the soil mineral N pool changed over the course of the growing season and more of the soil mineral N recovered by the leaves and canes was derived directly from legume residues than was subsequently then mineralized and taken up during fruit development.

An alternative explanation may be that the fruit had a heavier reliance on N remobilized from other structures from within the vine than upon the uptake of N assimilated from the soil. This is consistent with the data presented by Treeby and Wheatley (2006) who concluded that remobilized N provided about two-thirds of the N measured in grape berries at harvest. However, this is not a totally satisfactory explanation for although the 15N-enrichments of the fruit did resemble those detected in the canes (Table 5), they were not identical, and they differed markedly from the other plant parts. Perhaps the truth lies in a combination of these two possibilities.

Comparison of recoveries of legume and fertilizer N by vines

The total amount of legume N taken up by vines in the cover cropped treatments (12–15 kg N ha-1) was remarkably similar to the amount of fertilizer N used by vines (11-12 kg N ha-1: Table 6). The apparent recovery of legume N by vines was estimated to represent <10% of the above-ground N and possibly <5% of the total N accumulated by the cover crops each year. While these recoveries are low, such low short-term recoveries of legume N by other plants are not uncommon (Fillery 2001; Crews and Peoples 2005; Peoples et al. 2009). Comparative recoveries of the 15N-enriched fertilizer by vines were calculated to be 27–30% of the total fertilizer N applied, which was not too dissimilar from values previously reported for other crops (Crews and Peoples 2005; Gardner and Drinkwater 2009; Peoples et al. 2009). However, the estimates of the proportion of vine N derived from fertilizer observed here for various plant parts (ranged from 8–13% in roots to 24–27% in fruit) were generally higher than detected by Peacock et al. (1989) when investigating the impact of the time of application of fertilizer N to vineyards on N uptake (Ndff <10% of vine tissue N).

Legume and fertilizer N not recovered by vines

More than 90% of the N inputs provided by the self-replenishing legume cover crops, and 70% of the fertilizer N was not taken up by the grapevines. Unfortunately, the experimental design was such that it was not possible to distinguish between the contributions of below-ground pools of N or the above-ground mulch as sources of legume N for vines, or determine the fate of the legume N that was utilized by the vines. Nor was the soil profile sampled to determine now much residual 15N remained at the end of the 2006-2007 growing season to assess the extent of N losses from the cover cropped or bare soil treatments. However, it is possible to speculate based information reported in the scientific literature about the N-dynamics of legume and fertilizer N in other agricultural systems.

There are data that suggest below-ground legume N may be a more important source of N for crops than above-ground residues (Peoples et al. 2009). Other research indicates that rhizodeposits of N from living roots and other labile forms of legume N are readily immobilized by soil micro-organisms (Murphy et al. 1998; Wichern et al. 2008). Legume sources of N are usually considered to be less susceptible to loss processes than fertilizer and large proportions of legume N inputs tend to remain as organic N or are incorporated in soil microbial biomass to be slowly released for plant use over subsequent years (Fillery 2001; Crews and Peoples 2005; Gardner and Drinkwater 2009; Peoples et al. 2009). Consequently, it would be anticipated that the continued use of legume cover cropping in vineyards would result in a build-up of soil organic carbon and N, and there would be a net reduction in green-house gas emissions due to the substitution of legume sources of N for fertilizer N (Peoples and Baldock 2001; Boddey et al. 2009; Peoples et al. 2009).

Conclusions

The use of cover crops in a vineyard growing in low fertility granitic soils of the Mediterranean region of Chile presented clear advantages over the traditional N fertilized, bare soil management system in terms of improved N nutrition for grapevines. This study demonstrated that mixtures of annual legumes could produce large amounts of dry matter of high N content, resulting in large inputs of N into the system that can contribute similar amounts of N to vines as the current fertilizer recommendations. Grape production was increased by this additional N supply by legume cover crops, but no differences were observed in the leaf, cane, trunk or root biomass. The transfer of legume N to the leaves, roots and woody components of vines were verified using a 15N technique which indicated that around 20% of the total N accumulated by the grapevines were provided by the legume cover crops. However, the 15N data also suggested that legume N was not directly responsible for the observed improvements in grape production in the second year of the experiment. It is believed that this finding was a consequence of N accumulation by the vegetative components and fruit occurring at different times during the growing season. It is proposed that either there were dynamic changes in the dominant sources of organic N mineralized in the soil over time, and/or that the fruit relied more heavily on N remobilized from elsewhere in the vine than on N assimilated from the soil.