Evidence from field nitrogen balance and 15N natural abundance data for the contribution of biological N fixation to Brazilian sugarcane varieties 2 Introduction The Brazilian Pro-Álcool program to produce ethanol as a vehicle fuel from sugarcane started in 1975. At that time the area under this crop was just less than 2 million ha (Mha) and the mean national yield was 47 Mg millable cane ha-1 (MAPA, 2007). The area and yield increased steadily for the first 10-12 years of the program reaching over 4 Mha in 1987 with a mean yield of 62 Mg ha-1. At this time world crude oil prices remained low and direct subsidies for the program were phased out. Over the next 15 years the area under sugarcane only increased by about 1 Mha but in 2001 a law was passed making obligatory to add between 20 and 24 % of anhydrous ethanol to all gasoline (Federal Law No. 10,203 of 22nd February). In 2002 the area passed the 5 Mha mark and the mean national yields exceeded 70 Mg ha-1 (Boddey et al., 2008). With the commercialization of the first Flex-Fuel motors which can run on any mixture of ethanol or gasoline, and the rise in international crude oil prices, the area under cane grew more rapidly to almost 10 Mha today and mean yields have increased to almost 80 Mg ha-1 (IBGE-SIDRA, 2011). Despite the very significant increase in yields since the inception of the Pro- Álcool program, an exceptional aspect of the Brazilian cane crop is that the N fertilizer inputs remain low compared to other regions of the world which have similar yields. As N fertilizer manufacture has a high fossil fuel (generally natural gas) input, this enhances the mitigation of greenhouse gas emissions derived from the use of this biofuel (Boddey et al., 2008). In countries such as India, the USA, Columbia and 1 Australia, N fertilizer is usually added at between 150 and 200 kg N ha-1 yr-1 while in Brazil mean applications are between 60 and 70 kg N ha-1 yr-1. Boddey et al. (2003) calculated that for a crop with yield of 70 Mg ha-1 (mean national yield at that time) approximately 100 kg of N ha-1 are exported from the field either as harvested cane or lost during burning. As only 60 to 70 kg N ha-1 yr-1 are added as fertilizer this means that there is a deficit of 30 to 40 kg N ha-1 yr-1 even assuming that there is no N loss from the system as ammonia volatilisation, denitrification, leaching or erosion. Long-term decreases in yield have not been observed and this constitutes support for the idea that there may be an input of biological N fixation 2 (BNF) to this crop (Boddey et al., 2003). Basanta et al. (2003) conducted a N balance study in the São Paulo production region (Piracicaba) and estimated that over 3 crops (plant crop and two ratoons) an annual mean 103.6 kg N ha-1 were exported to the mill as cane and 83.5 kg of N was lost on burning. In this experiment N fertilizer was only added at planting (63 kg N ha-1) such that the net N export for the three crops was 163 kg N ha-1 yr-1. Despite this the third ratoon crop, 36 months after planting, accumulated 155 kg N ha-1, again indicative of a considerable input from BNF or other external sources. Further evidence for BNF is derived from microbiological studies. In early studies N -fixing bacteria of the genera Beijerinkia were found to be associated with 2 soils in cane fields (Döbereiner, 1961). Subsequently high numbers of Gluconacetobacter diazotrophicus (Li and Macrae, 1992; Reis et al., 1994; Reis Junior et al. 2000) and other diazotrophs of the genera Herbaspirillum and Burkholderia (amongst others) have been found in the interior of cane roots stems and even leaves (James, 2000; Boddey et al., 2003; Perin et al., 2006a). 2 To quantify the contribution of BNF to different Brazilian cane varieties the team at Embrapa Agrobiologia has used the 15N isotope dilution technique combined with short-term N balances on plants grown in large (60 L) pots or a 120 m2 concrete tank filled with 15N-labelled soil (Lima et al., 1987; Urquiaga et al., 1992). In these studies the cane varieties showing the lowest 15N enrichment also showed higher N accumulation and N balance in the soil/plant system, which was powerful evidence for a contributions of BNF to the former varieties. In a field sampling conducted by Boddey et al. (2001) in 11 different plantations in four States of Brazil showed that at all but one site weeds lants were significantly higher in 15N abundance suggesting input of BNF from 0 to 70 % of shoot N. A long-term study (16 years) by the same team performed on one cane variety (CB 45-3) on a plantation in Pernambuco State, NE Brazil demonstrated that there was a positive N balance in the treatments not receiving N fertilizer, again indicating significant contributions of BNF to the cane crop (Resende et al., 2006). However in studies conducted in Australia (Biggs et al., 2002) and South Africa (2005) the 15N abundance of non-N -fixing reference plants was not found to be higher 2 than in sugarcane and the authors concluded that BNF inputs were insignificant. While two other studies have utilized the 15N natural abundance technique to evaluate possible inputs of N fixation to sugarcane, this technique has not been used 2 concurrently with a long-term N balance study. The objective of this present study was to estimate the contribution of BNF to nine cane varieties planted in the field by investigating the long-term N balance of the crop/soil system and from the 15N abundance of the cane varieties at the last five annual harvests. 3 Material and methods Site and experimental layout The experiment was planted on the field station of Embrapa Agrobiologia in Seropédica, Rio de Janeiro State (22o49’S, 43o39’W) and was not irrigated. The soil (Itaguaí series - sand, silt and clay, 32, 26 and 42 %, respectively) at this site is classified as an Acrisol (FAO) or Typic Hapladult (USDA, Soil Taxonomy). Results of soil analyses(0-15 cm depth) were: pH (water), 5.4, exchangeable Al3+, Ca2+ and Mg2+ 0.0, 1.4 and 2.2 cmol dm-3, respectively; exchangeable K, 22 mg dm-3; available c (Mehlich 1) P, 4.9 mg dm-3 and total N 0.57 g kg-1. Nine varieties of sugarcane (Table 1) were planted as setts (stem pieces) in May of 1989 in plots of 5 rows, 5 m in length, and spaced 1.1 m between rows. The experiment was laid out in a randomized complete block design with four replicates. Just prior to planting the experiment the area was fertilized with 125 kg K and 66 kg P ha-1as potassium chloride and single superphosphate, respectively, and 50 kg ha-1 of fritted trace elements (FTE) type BR12 (http://www.nutriplant.com.br/).Each year after harvest 66 kg P and 132 kg K and 50 kg ha-1 of fritted trace elements were added in the same forms as at planting. After 9 years of growth (8 harvests), in May 1999, the area was ploughed and the experiment was re-planted (2nd cycle) with the same varieties in the same plots. Harvests The first plant crop of each cycle was harvested after 18 months of growth (standard practice in Brazil) in December 1990 and December 2000, for the two cycles respectively. Thenceforth the ratoon crops were harvested every 12 months in all three crop cycles. At each harvest the aerial tissue of the plants from the central three rows was 4 separated into fresh stems, senescent leaves (trash) and flag leaves (green leaves left after decapitating the stems). Subsequently all above-ground biomass was removed from the whole area of each plot. The fresh weight of each fraction was determined using a platform balance and subsamples were taken to determine dry weight content in a forced air oven (65 oC for >72 h). For the harvests during the second cycle (2000 to 2004) bulked samples of the third-emergent leaf of 20 cane plants from each plot were also taken for analysis of 15N abundance. Previous work of Boddey et al. (2001) indicated that the 15N abundance of the third-emergent leaf would closely correspond to that of the above-ground biomass. In 2000, 2003 and 2004 whole shoots of non-leguminous weeds growing in the plots were collected to serve as non-N -fixing reference plants for the application of the 2 15N natural abundance technique. For each weed species a composite sample of all aerial tissue was taken from the whole experiment. All samples were ground with a Wiley mill to <1 mm and aliquots for analysis for total N and 15N natural abundance were further ground to a fine powder using a roller mill similar to that described by Arnold and Schepers (2004). Soil sampling At the original planting of the experiment in May 1989 soil samples were taken from three points in each plot with an auger at the depth intervals 0-15, 15-30, 30-45 and 45-60 cm. Table 1 Cane yield (Mg fresh weight stems ha-1) of nine sugar cane varieties and total accumulated rainfall between harvests from 1990 to 1998 (First cycle). Cane variety Year 1990 1992 1993 1994 1995 1996 1997 1998 Mean CB 47-89a 72.3 abb 48.2 a 55.2ab 56.2a 67.9ab 52.5ab 51.0ab 65.5ab 58.6ab CB 45-3a 74.8 ab 38.6 ab 55.1ab 57.7a 78.9ab 64.5ab 62.3a 80.0a 63.9ab SP 70-1143a 82.8 a 41.0 ab 106.0a 71.8a 97.9a 72.6a 61.3a 85.0a 77.3a SP 79-2312a 102.7 a 28.6 c 68.7ab 72.7a 79.0ab 53.3ab 44.8ab 59.5ab 63.6ab SP 71-1406a 99.0 a 48.6 a 55.8ab 46.5a 65.3bc 48.3b 44.4ab 49.8bc 57.2ab SP 71-6163a 80.4 a 49.2 a 58.6ab 45.4a 70.5ab 47.6b 42.3b 63.7ab 57.2ab SP 70-1284a 87.8 a 43.4 ab 54.0ab 48.9a 67.3ab 46.9b 42.5b 66.2ab 57.1ab Krakatauc 39.6 bc 30.5 ab 52.1ab 47.1a 49.3bc 46.2b 45.8ab 58.5ab 46.1bc Chuneed 32.7 c 50.4 a 23.8 b 23.0a 23.3c 17.0c 19.3c 28.2c 27.2c CV (%)e 21 13 19 39 18 23 30 23 20 Rainfall (mm)f 1387 1372 1284 1112 1363 1880 1187 1318 a Commercial interspecific hybrids (Saccharum spp.) b Means followed by the same letter in the same column are not significantly different at P<0.05 (Tukey´s HSD test). c Saccharum spontaneum, not planted commercially but used in plant breeding. d Saccharum barberi, not planted commercially but used in plant breeding. e Coefficient of variation. f Total rainfall between planting and the first harvest, or between consecutive harvest of the ratoon crops. The samples were bulked for the same depth interval in each plot. The sampling depth was restricted to 60 cm because between 60 and 70 cm depth there was hard clay pan, through which almost no roots were observed to penetrate. Simultaneously in four plots of each block trenches were excavated to permit the evaluation of soil bulk density using bevelled rings (length 7.4, diameter 5 cm = 145 cm3), four bulked samples for each depth in each trench. After the final harvest of the second cycle (December 2004) the sampling and determination of soil density were repeated exactly as before. All augered soil samples were air dried, passed through a 2 mm sieve removing all visible root material. Subsamples were then taken for fine grinding using the roller grinder described by Arnold and Schepers (2004). The bulk density samples were dried at 110oC for subsequent weighing and determination of bulk density. Analyses of plant material and soil Finely ground samples (200 mg aliquots) of the third-emergent leaves of the cane and the shoot tissue of the weeds were analysed for total N using the semi-micro Kjeldahl procedure as described by Urquiaga et al. (1992) and aliquots containing between 20 and 40 µg N were analysed for 15N abundance using an automated continuous-flow isotope-ratio mass spectrometer (Finnigan MAT, Bremen, Germany) in the “John M Day Stable Isotope Laboratory” at Embrapa Agrobiologia (Ramos et al., 2001). The ground soil samples were analysed for total N using the same semi-micro Kjeldahl procedure on 1 g aliquots. Determination of mineral N precipitated in rainfall Commercial rainfall collectors (20 cm diameter) were installed at five sites 1 within a distance of 2 km from the field experiment in 2008. Rainfall water samples were collected within 24 h of each precipitation event from August 2008 to May 2010. The volume was measured to estimate rainfall and then a sub-sample of each sample was stored at -15 to-20oC. The samples were thawed and the concentrations of ammonium and nitrate estimated usingthe salicylate method of Kempers and Zweers (1986) for ammonium and using ultraviolet adsorption for thedetermination of nitrate concentration (Norman et al. 1985). For the determination of nitrate the technique used is based on the linear relationship between NO - concentration and the difference in absorbance at 210 and at 3 270 nm wavelengths. The sensitivity of both techniques was better than 0.05 µg N ml-1. Investigation of variation in 15N of plant-available N with depth In order to examine the change in 15N natural abundance of the plant-available N with depth in the soil in this experiment, the procedure of Ledgard et al (1984) was adopted. Soil samples were taken in October 2003 from each of three blocks at depth intervals of 0-10, 10-20, 20-30, 30-40, 40-50, 50-60 and 60-75 cm and all samples were planted to four weed species. The experiment consisted of 4 weed species x 7 depth intervals and 3 replicates (representing the 3 replicate blocks of the field experiment) arranged in a completely randomized design and planted in the greenhouse at Embrapa Agrobiologia. Each pot was filled with 400 g soil, and all were fertilized with 44 mg P, 83 mg K kg soil-1 as single super phosphate and potassium chloride, respectively, and 50 mg of FTE kg soil-1. Each pot was planted to eight seeds of each of the weed species, Commelina erecta (slender dayflower), Eleusine indica (Indian goosegrass), Sorghum bicolor (grain sorghum) and Panicum miliaceum (proso millet). After one week excess plants were removed to leave 4 plants per pot. The shoot tissue of these plants was 2 harvested after 75 days of growth when all were showing signs of severe N deficiency. The plant material was dried, weighed, ground and analysed for total N and 15N abundance as described above for the samples of sugarcane. In order to visualise the rooting depth of the different weed species and the sugarcane varietiespits were opened in 2009 in plots of the continuation of Experiment 2 around the roots of these plants. The profile wall was washed with a spray to show the clean roots. The profiles of roots of the weeds Emilia sonchifolia and Commelina erecta and the sugarcane variety CB 47-89 are shown in Fig. 1. Calculations N balance of soil/plant system The N balance for each sugarcane variety was calculated from the sum of the total N stock in the soil (0-60 cm) at the final harvest for each plot plus the total N accumulated by the cane variety over the 13 harvests minus the initial total N stock in the soil to the same depth at planting. The mass of N in each depth interval was calculated from the N concentration and the bulk density of the soil. The total N stock was calculated by correcting the mass of soil in the profile at the final sampling to that at the first sampling as described by Ellert and Bettany (1995). 15N abundance of plant-available N with depth The shoot tissue of the weed reference plants grown in the soil samples taken from different depths was analyzed for total N and 15N abundance as described above. To simulate a deep rooting plant which removed all plant-available N to a depth of 75 cm the weighted mean 15N abundance of reference plant (rp) was calculated using the following equation: Weighted mean 15N = ( 15N * NT )/ (NT ) ……………….Eqn. 1. rp rp rp 3 where 15Nrp and NTrp are the 15N abundance and total N accumulated by the weed species grown in the soil from each depth interval. Statistical analyses The statistical procedures followed for all plant parameters (cane yield), DM, N total, 15N) were standard analysis of variance using the software SAEG 9.1 of the Federal University Viçosa, MG. Normality and homogeneity of variance of errors were analysed using the Lilliefors and Cochran & Bartley tests, respectively. The required conditions were met in all cases. The analysis of variance was then performed with the application of the F test. Differences between means were separated by the use of the Tukey HSD test at p=0.05. The Student “t” test was used to compare the four replicates of the initial and final soil C stocks, and the N balance, for each variety. Results First Cycle - 1989 to 1999 The commercial SP varieties all yielded over 80 Mg ha-1 of fresh stems at the first harvest (plant crop) (Table 1). Plant crops in Brazilian practice are almost always harvested after 18 months which allows the plants to establish well and, as was observed in this case, the varieties obtained their highest yield during the cycle. Krakatau (S. spontaneum) is a vigorously growing species but produces only thin canes with little sugar. Likewise Chunee (S. barberi) is not a productive cane, and like Krakatau is a component of the interspecific commercial hybrids. The first ratoon harvest in 1991 was lost when the whole experiment caught fire shortly before harvest time, but subsequently ratoon yields recovered over the years. The stem yields over the 8 harvests of the commercial varieties did not differ