Agriculture, Ecosystems and Environment 136 (2010) 148–161

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee

Possibilities of carbon and nitrogen sequestration under conventional tillage and no-till cover crop farming (Mekong valley, Laos) A. de Rouw a,*, S. Huon b, B. Soulileuth c, P. Jouquet a, A. Pierret a,d, O. Ribolzi e, C. Valentin a, E. Bourdon a,d, B. Chantharath f a IRD (Institut de Recherche pour le De´veloppement), UMR 7618 Bioemco (Bioge´ochimie et Ecologie des Milieux Continentaux), Universite´ Pierre et Marie Curie, case 120, 4 place Jussieu, 75252 Paris cedex 05, France b Universite´ Pierre et Marie Curie, UMR 7618 Bioemco, 4, place Jussieu, 75252 Paris cedex 05, France c MSEC programme, (Managing Soil Erosion Consortium) Ban Lak Sip, Luang Prabang district, Lao Democratic People’s Republic d IWMI (International Water Management Institute) P.O. Box 811, Vientiane, Lao Democratic People’s Republic e IRD, UMR 5563 LMTG (Laboratoire des Me´canismes de Transferts en Ge´ologie), Universite´ de Toulouse, 14 avenure Eduard Belin, F-31400 Toulouse, France f PRONAE (Programme National en Agroe´cologie Laos), Vientiane, Lao Democratic People’s Republic

A R T I C L E I N F O

A B S T R A C T

Article history: Received 23 July 2009 Received in revised form 7 December 2009 Accepted 9 December 2009 Available online 15 January 2010

There is limited information, particularly in the tropics, of farming systems that loose or accumulate carbon in their soils. We compared no-till with a mulch-providing cover crop with conventional tillage without cover crop. Side effects were also investigated, weeds, surface crusting, soil macrofauna, infiltration, porosity and roots. The study site was a flat sandy clay loam. Treatments were maintained over five years; within this period, the time between the first and last soil sampling was exactly four years. Both times the same profile locations and exactly the same depths were sampled thereby greatly reducing inherent soil variability. Soil was sampled at five increments from 0 to 40 cm depth. The biomass contributions of maize, cover crop and weeds were measured. The main findings were: (1) The cover crop that was alleged to supply extra inputs to the no-till system failed to do so because the weeds in the tillage treatment became as efficient in accumulating biomass as the planted cover crop. (2) With equal organic inputs over four years (43.0 Mg dry weight ha!1 incorporated into the soil under conventional tillage, and 44.2 Mg dry weight ha!1 remaining on the soil surface as mulch under no-till), the tillage system stored (0–40 cm) significantly soil carbon (+590 g C m!2), whereas the no-till lost carbon (!133 g C m!2). The difference between the systems was significant. Carbon accumulated just below the plough layer. Nitrogen stocks remained unchanged. A very significant lowering of the C:N ratio occurred under no-till. The process of transforming the available biomass on the soil surface into organic matter is apparently too slow to avoid direct losses under no-till. Alternatively, ploughing plant residues into the soil enables to capture some of what would otherwise be lost as CO2 through decay, thereby increasing soil carbon. (3) In the last three years of the experiment, maize grain yields and crop residues stabilized at a lower level but were significantly higher under no-till, 16% and 34%, respectively. Higher yields were attributed to more soil water under no-till due to improved soil structure, though bulk density was not affected. The mulch layer protecting the soil surface favoured infiltration by keeping it crust-free. Water availability was further promoted by a better connectivity of pores and more macrofauna. However, the no-till system depended heavily on fertilizers and herbicides. The lack of effectiveness of herbicides against shifting weed communities threatens the continuation of the system. ! 2009 Elsevier B.V. All rights reserved.

Keywords: Maize Weeds Ruzi grass Earthworms Surface crusts Roots

1. Introduction No-till as an alternative to conventional tillage began in the 1940s with the discovery of hormonal herbicides that allowed

* Corresponding author. Tel.: +33 0 1 44 27 72 82; fax: +33 0 1 48 47 55 34. E-mail address: [email protected] (A. de Rouw). 0167-8809/$ – see front matter ! 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2009.12.013

farmers to control weeds without resorting to ploughs or hoes. Initially, no-till was not recommended to sequester carbon, but to limit erosion. No-till, by developing a litter layer, successfully reduced erosion and this has been the principal factor for increased soil organic matter in no-till systems compared to conventional tillage (Needelman et al., 1999; Valentin et al., 2008). Renewed interest in no-till arose in the 1990s as a mean to build-up organic matter in those soils that had lost large portions of their organic

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

matter content. The advantages were twofold: an increase in soil organic matter would sustain or increase the productivity of the land, and the soil would sequestrate atmospheric CO2. No-till systems agree in keeping the soil surface permanently covered by an organic layer that protects the soil physically from sun, rain and erosion, while suppressing weed growth and feeding soil organisms. The organic layer is derived from crop residues with additional mulch material commonly obtained from a cover crop. Conventional tillage contrasts with no-till in that the soil is bare at the time of sowing and tends to remain so over the cropping period to reduce competition from weeds. Under conventional tillage, clearing, seed bed preparation and weed control results from tillage. In contrast, no-till inevitably requires herbicides to prepare the land for sowing because no effective alternative methods for weeding have been established. Nye and Greenland (1960) demonstrated that the effect of notill on soil organic matter depends as much on the initial organic level of the soil in relation to the equilibrium level as on the properties of the no-till farming system itself; hence high rates of carbon sequestration are only to be expected in soils a long way below their equilibrium level. The greatest long-term potential for carbon storage would lie in the tropics because there the cultivated soils have lost on average 42% of their original carbon stock found under native vegetation (Ogle et al., 2005). West and Six (2007) predicted by meta analysis with a high degree of uncertainty that soil carbon should increase by approximately 20% with the cessation of tillage in the dry and moist tropics, and with an additional 20% in the moist tropics if more than one crop was produced per year. Needelman et al. (1999) suggested that effects of no-till on carbon storage are associated with biomass production, because no or little carbon storage occurs unless more biomass is produced compared to a plough system. More biomass is usually correlated with higher crop yield. However, in the absence of yield increases, which is usually the case under no-till (Corbeels et al., 2006; Giller et al., 2009), the extra biomass benefits from no-till would come from a second crop grown annually (Kuipers, 1991). Such cases were confirmed in later publications (West and Post, 2002; Wright et al., 2007; Spargo et al., 2008). Additional organic inputs could also be obtained by replacing annuals by long-lived species, for instance perennial forages. A subsequent increase in soil carbon was demonstrated in some experiments (Paustian et al., 1997; Needelman et al., 1999; Baker et al., 2007) but the conclusions in Manley et al. (2005) are mixed: superior organic inputs are no guarantee for soil carbon sequestration and overall no-till practices stored little or no carbon. It remains unclear to what extent additional biomass inputs contribute to carbon sequestration mainly because most studies give only rough estimates of biomass inputs, while weed contributions are commonly overlooked. Extra inputs under no-till occurred when weeds were maintained in the no-till plots but eliminated periodically by glyphosate in the tillage plots (Diekow et al., 2005). There is a need for systematic sampling of weed biomass when it represents a noticeable proportion of residues returned to the soil (Barthe`s et al., 2004). In other experiments, the tillage treatment included the export of crop residues whereas in the no-till treatment residues were maintained (Ogle et al., 2005; Razafimbelo et al., 2006), or no-till was supplemented with extra mulch collected outside the plot (Affholder et al., 2008). Soil carbon can be stored under no-till independently of biomass inputs because losses are reduced, for instance erosion has stopped and microbial activity is less. In this study erosion was avoided by using flat land. Consequently, impact of tillage and mulching on soil carbon within the profile was examined. The measurements of microbial activity and content were not part of the study.

149

No-till is part of a « package » because applied in isolation it does not give the expected results. Commonly, the body of practises to be applied includes the cultivation of a mulch-providing cover crop, herbicides, fertilisers and crop rotations. The integrated nature of no-till gives rise to side effects and these need scientific concern. Some effects are overall positive; better structured topsoil, increased infiltration, reduced crusting (Franzluebbers, 2002) and a more abundant soil macrofauna under no-till (Hulugalle et al., 1994). Other effects tend to be negative, such as increased weed problems (Locke et al., 2002), and continuous high inputs. Since herbicides are applied in the same pattern year after year, no-till selects for late emerging, often perennial weeds that are increasingly difficult to control (Kobayashi et al., 2003; Blackshaw, 2005). Herbicide costs go up and increased herbicide use leads to environmental pollution (Monneveux et al., 2006). Secondly, crops grown in untilled land require more nitrogen (Kuipers, 1991; Deen and Kataki, 2003; Monneveux et al., 2006). Previous studies insist on three points. There are too few studies from tropical regions while hopes of carbon sequestration are highest there. Secondly, it remains unclear to what extent additional biomass inputs contribute to carbon sequestration. Thirdly, side effects of the no-till system have to be studied because the system is not only responding in terms of soil carbon, but also in other ways and it is judicious to expect a feed-back if enough time is allowed. In this study we examined crops, weeds, soil macrofauna, soil crusting, water infiltration, porosity and roots. By addressing this matter, the study investigates issues that are often taken for granted: (1) a cover crop adds biomass to the system; (2) weed problems build-up due to herbicides use; (3) mulches increase water infiltration and reduce crusting; (4) plants in no-till systems develop different root systems, shallower rooting than in tillage systems; (5) soil macrofauna activity and diversity improves after mulch application. It is highly desirable that these aspects (1–5) have a direct experimental verification. 2. Materials and methods The study was part of a larger programme on carbon sequestration in tropical cultivated soils. To facilitate comparison across sites all experiments agreed in choice of crops and crop sequence, only two treatments, and meticulous soil sampling at the start and at the end of the experiment. 2.1. Study site and land preparation The study was conducted at the National Agricultural Research Centre (NARC) in the flat valley of the Mekong River near Napok village, 20 km east of Vientiane, Laos (18.8 N, 102.44 E, at 170 m above sea level). The soils at the site are sandy clay loams (47% fine sand, 23% silt, 22% clay) with clay content increasing gradually with depth to 30% at 40 cm. Soils are acidic (pH – H2O 5.7, 0–15 cm depth) and of moderate fertility (CEC 7.0 cmol kg!1, 0–15 cm depth). Such soils, alluvial Ferric Acrisols, occupy a significant part of the Mekong valley in Laos and Northeastern Thailand and are widely cultivated. The climate is tropical with a wet season from May to September and a relatively dry season from October to April. The mean annual precipitation is 1600 mm. In the year preceding the experiment, the following operations were carried out at the experimental site to promote uniformity. In April 2001 at the end of the dry season, the spontaneous vegetation was cut and ploughed into the soil. In June 2001 at the onset of the rainy season, lime (3 Mg ha!1 of CaCO3) was applied followed by repeated cross-wise ploughing. In August 2001 the forage grass Eleusine coracana Gaertn. was sown to cover the site uniformly until the start of the experiment in May 2002. The Eleusine was then ploughed into the soil in the tillage treatment plots, and kept

150

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

at the soil surface after herbicide application in the no-till treatment plots. 2.2. Experimental design and treatments The experiment was a single factor experiment, testing conventional tillage vs. no-till. The treatments were maintained over five seasons (2002–2006). Treatments were applied randomly to plots (25 m by 15 m). There were four replicates, hence a total of eight plots. In May, in the tillage treatment, the plots were disk ploughed down to 20 cm in one direction to control weeds followed by harrowing to prepare the seed bed. At the same time in the no-till plots herbicides were applied in the no-till plots to control weeds, followed by direct sowing in the dead mulch. A single spraying of glyphosate (5 L ha!1) was applied mixed with 2.4D (1.5 L ha!1) and H2SO4 (75 ml ha!1). At the onset of the rains, May or June, manual hill planting of maize was carried out, the same in all plots (0.75 m between rows, 0.5 m within rows, 2 seeds per hill). The popular because cheap maize variety LVN-10 was used, a hybrid animal feed developed in Vietnam but produced in Laos. In the no-till plots, hill sowing of the cover crop Ruzi grass (15 kg seed ha!1, Brachiaria ruziziensis Germ. & Evrard) took place about one month after the sowing of the maize. Ruzi grass was chosen because of its extensive root development allowing it to grow in the relatively dry season. Secondly, the persistence of a large quantity of dead sod after the herbicide spraying is known to stimulate earthworm activity, to reduce surface crusting and to favour infiltration (Lal et al., 1978). In the no-till plots Ruzi grass covered the soil surface part of the rainy season and throughout the dry season, to be killed by herbicides at the onset of the following cropping season (Table 1). In the conventional tillage plots the soil was mostly bare except for Maize during the rainy season and gradually covered by weeds during the dry season. Identical quantities of mineral fertilizers were applied in both treatments: a first application of 300 kg ha!1 NPK 8–24–24 two weeks after the sowing of the maize, and a second application of 100 kg ha!1 urea in July. Harvest of the maize occurred in August or September. The third year, June 2004, patches of excessive weed growth in the notill plots were controlled locally by spraying atrazine (75 ml ha!1) and in the tillage plots by hand hoeing. The fourth and the fifth year, in June, patches of weeds in the no-till plots were controlled by local spraying of gramoxone (1 L ha!1). Daily rainfall was measured at the site. Seasonal rainfall between sowing and harvest of the maize amounted 1090, 1333, Table 1 Principal operations, agricultural station Ban Napok, Laos.

1022, 1320 and 1035 mm in 2002, 2003, 2004, 2005 and 2006, respectively. In all five years the maize did not experience water stress. The rainfall in the relatively dry season was more variable, 502, 567, 366, 516 and 356 mm in 2002, 2003, 2004, 2005 and 2006, respectively. 2.3. Soil sampling Soil samples for carbon and nitrogen studies were collected at the onset of the experiment, May 2002, and at the end of the experiment, May 2006. In each plot nine soil profiles were sampled in a cross arrangement: two ropes crossed diagonally: one sample was collected in the centre and four samples were collected along the diagonals at 2 m and 10 m from the centre, respectively. The same locations were sampled at the end of the experiment. Each profile contained five sample depths (0–5, 5–10, 10–20, 20–30, 30– 40 cm). Bulk density was determined in 2002 and in 2006 for each of the 72 soil profiles using a 100 cm3 density cylinder at the following depths: 0–5, 5–10, 10–20, 20–30 and 30–40 cm. All samples were analysed separately. Samples were dried at 110 8C in an oven until constant weight and weighted for bulk density measurements. 2.4. Carbon and nitrogen analysis All soil samples were softly hand-crushed using a mortar. Twenty-gram subsamples were sieved at 2 mm. The coarse >2 mm size fraction was entirely composed of gravel not exceeding 1% of the total soil weight and of root fragments. Coarse roots were handpicked, dried and weighted. Total organic carbon concentration (TOC in mg C g!1), and total nitrogen concentration (TN in mg N g!1) were measured with a Carlo Erba NA-1500 NC Elemental Analyzer facility of the UMR Bioemco. The C stocks were calculated using the equation: SC ¼ TOC # r # d where SC is the total organic carbon in g C m!2; TOC is the concentration of carbon in g C g!1 with respect to a constant soil sampling volume, r is the soil bulk dry density in g cm!3; d is the thickness of the soil sampled (i.e. 5 or 10 cm in our study). Equivalent calculations were performed for TN stocks. In the soil profiles the bulk density had changed in 2006 with respect to 2002, therefore soil carbon and nitrogen stocks were corrected in 2006 by calculations on an equivalent soil mass basis

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

following Ellert and Bettany (1995). Corrections were made in fixed sample depths and not in genetic horizons because the morphology of these soil profiles showed only very gradual changes with depth. 2.5. Maize harvest At harvest, eight subplots (2 m # 1.5 m) were laid out on the diagonals in each plot. In each subplot, mean height of maize, number of hills, number of cobs, fresh weight of cobs and fresh weight of crop residue were determined. A sample of 50 cobs was randomly collected in these eight subplots to determine dry weight of cobs, grain yield, and single grain yield (from two countings of 200 grains). A random lump crop residue sample was taken from the same eight subplots to determine dry biomass weight. Grain yield, yield components, total above ground biomass and crop residues were determined for each plot in each year.

151

estimated from a grab sample collected from the upper 1 cm of soil below the layer of sand. Values of hydraulic conductivity, mean functional pore size and sorptivity were calculated following standard procedures (White and Sully, 1987). Because of the very dry conditions, the sorptivity calculations were refined following Vandervaere et al. (1997), a procedure that eliminates the earlytime effect in water flow due to the presence of the initially dry contact of the sand layer. Destructive fine root sampling was carried out in February 2006, by collecting 54, 100 cm3 soil samples using standard soil bulk density rings. Three soil depths, 13, 25 and 45 cm were sampled under both conventional tillage and no-till treatments. There were nine replicates all taken in the same treatment plot. Roots were washed free of soil and the root parameters root length per unit soil volume and average root diameter were computed according to Pierret et al. (2007). 2.9. Soil macrofauna

2.6. Weeds, Ruzi grass and litter Floristic inventories (releve´s of 9 m2, Braun-Blanquet, 1964) were made twice a year in each plot: in May to evaluate dry season growth (October–May), and in October to assess rainy season growth (June–October) (Table 1). Each releve´ included height and total cover estimates of the vegetation, a full species list with each species receiving a cover score (1 = 75%) and the sampling of litter and above ground biomass (1 m2). Fresh biomass was split for separate dry weight determination into Ruzi grass and weeds. Dried voucher specimens were made of all the species occurring in the plots and brought to the Chiang Mai University herbarium (Thailand) for verification by Dr. J.F. Maxwell. One specimen of each species was deposited at the National Herbarium of the Laos National University. 2.7. Soil surface conditions The soil surface conditions were characterised and visually quantified in February 2006 by the method used in Southeast Asia (Podwojewski et al., 2008). Percentages of surface covered with weeds, plant residues, algae, free aggregates, structural crusts worm casts and termite constructions as well as mean surface cracks width were measured in two 1 m2 plots randomly selected within each of the eight plots (Valentin and Bresson, 1992). 2.8. Infiltration, porosity and roots Infiltration tests were performed at the end of the dry season (February and March 2006) in order to capture the long-term treatment effects and not the short-term effect of the tillage operation. These tests characterised three topsoil attributes: hydraulic conductivity which is the gravimetric component of infiltration, sorptivity, which is the capillary component of infiltration, and functional pore size, which estimates the fraction of total pores that is active in water transport. Measurements were conducted in randomly selected plots, one no-till and one tillage plot, and repeated six times for each water tension. A disk permeameter (SDEC SW 080 B) was placed on the undisturbed soil surface and a thin layer of sand was added to ensure hydraulic contact between the soil and the disk. The applied pressure heads were set at !60, !20 and !5 mm. The initial volumetric water content of the soil was determined from a bulk density sample collected before each infiltration test (100 cm3 core collected from the upper 5 cm of the soil and oven dried at 105 8C for 24 h), the volumetric moisture of the soil at the imposed pressure head was

Sampling took place once at the end of the experiment in 6–8 June 2006, and included two operations: estimation of macrofauna abundance, and estimation of their activity. Macrofauna was extracted from soil monoliths of 25 cm # 25 cm wide and 0–20 cm depth following standard procedures (Anderson and Ingram, 1993). Two monoliths were extracted per plot from randomly selected sites, in total three no-till and three tillage plots were thus sampled. Soil macro-invertebrates were counted and classified into taxonomic groups and identified at the morphospecies level (i.e. diptera, coleopteran larvae, termites, scolopenders, ants and earthworms). Because biological activity was low at the end of the dry season, an attempt was made to simulate the onset of the rainy season by pouring 10 L of water over 1 m2 of soil surface at the end of the afternoon. There were three replicates in the no-till and three in the tillage plots. Fresh earthworm casts produced overnight at the soil surface were collected the next morning in plots that received water or not, dried at 60 8C during 24 h and weighted. 2.10. Statistical analysis 2.10.1. Agronomic data The agronomic data were analysed using Anova. A single factor (Farming system) analysed the differences in productivity of maize, weed and cover crop in every year using the F test (F13). A strong year effect was observed which raised the question whether the productivity data of all five years could be tested together in one Anova because the errors of various sample sets could be correlated in time. The Bartlett’s test of homogeneity of variances was applied to verify this. For the body of maize data (comprising 11 sample sets, i.e. five years, two treatments, and four replicates), the Bartlett’s test was not significant (x98 < 124.12) and the same accounted for the weed and cover crop data (comprising 10 sample sets, i.e. four years, two treatments, and four replicates, x68 < 88.38). Thus, the homogeneity of variances was assumed and we continued with a two factor Anova (farming system and year). Differences in maize productivity were investigated using the F test (F130 for farming system, F430 for year, and F430 for interaction system # year). Because yields stabilized in the last three years of the experiment, maize was analysed over this period using the F test (F117 for farming system F217 for year). Weed and cover crop productivity were analysed in the same way. Finally we compared years two by two in all possible combinations. For this we used a single classification (year) Anova (F116) comparing the sources of variance within groups (years) and among groups.

152

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

2.10.2. Reducing the number of carbon and nitrogen analyses in the 2006 set In 2002, all 72 soil profiles were analysed but in 2006 analysing the same amount of samples was not possible. Reducing the number of carbon and nitrogen analyses was done in the following way. We took randomly two items from the 2002 sample set and computed the mean carbon stock with the coefficient of variance (CV). Then we added a third item randomly and again computed mean carbon stock and coefficient of variance, then a fourth and so on up to 50 items. Number of items (sample size) was plotted against coefficient of variance. As expected, the variability of carbon stocks decreased as sample size increased; however, beyond a certain sample size it stabilized. This indicated that the degree of heterogeneity in the soil is rather high and also that, beyond a certain number, adding more items to a sample set does not help to reduce the standard deviation. Thus we obtained an estimate of the minimum number (n) of samples required for analysis. The n varied somewhat with soil depth, n was 8 in 0–5 and 5–10 cm soil layers corresponding to CV of 19%, and n was 10 in 10–20, 20–30, and 30–40 cm soil layers corresponding to CV of 20%, 21%, and to 25%, respectively. On the basis of these data, the minimum number of the soil profiles to be analysed for carbon and nitrogen in the 2006 sample set was set at 10 for each treatment. This number was increased to 11 because in one treatment the same profile was selected twice. 2.10.3. Soil data Variance in the soil data analysis was reduced by considering the observations as paired, since in 2002 and in 2006 the same sample location had been used. For the analysis of bulk density and coarse root quantity the data from 72 paired soil profiles were used of which 36 pairs in the tillage plots, and 36 pairs in the no-till plots. For the analysis of carbon and nitrogen stocks, 21 paired soil profiles were available of which 10 pairs in the tillage plots, and eleven pairs in the no-till plots. The Student test of paired observations was used to compare the initial with the final situation. Pairing was effective in reducing the variance by about a third. The Student test was also used to differentiate treatments (test of difference between the means of two independent samples). 2.10.4. Floristic data The two-way indicator species analysis was used for classification (TWINSPAN, Hill, 1979a) and the detrended correspondence analysis was used for ordination (DECORANA, Hill, 1979b). Input data were 64 releve´s (eight plots, two releve´s per year, four years) and all 68 species recorded with their cover scores. The first four divisions calculated by TWINSPAN were considered meaningful (Eigenvalues between 0.450 and 0.382) as well as the first two axis of DECORANA (Eigenvalues first axis l = 0.417, second axis l = 0.282). 2.10.5. Other data The macrofauna data were log(x + 1) transformed. Species richness was defined as the total number of morphospecies and abundance as the number of individuals per m2. Means were compared by t-test. The influence of water application and soil management on earthworm cast production were analysed using a single factor (Farming system) Anova. Surface crusts, hydraulic conductivity, sorptivity and various root data typically do not follow a normal distribution. We therefore used: (1) box plots for data summary, (2) the median as the estimator of central tendency, (3) non parametric Mann– Whitney test to evaluate the hydraulic data and the Welsh two sample t-test for root data. Otherwise Anova was used.

Fig. 1. Carbon stocks in corresponding soil profiles and layers in 2002 and 2006 with line indicating 1:1 relationship; (a) conventional tillage; (b) no-till.

3. Results 3.1. Changes in soil carbon and total nitrogen Soil carbon stocks at the start of the experiment were compared with those at the end of the experiment four years later. The stocks in corresponding sample locations were plotted with a line indicating the 1:1 relationship (Fig. 1). The pattern in Fig. 1a shows that under conventional tillage most of soil samples had increased in carbon stock, whereas under no-till the opposite occurred (Fig. 1b). Regarding the whole profile, 0–40 cm, carbon increased on average by 590 g C m!2 (P < 0.05) under conventional tillage, but diminished by 133 g C m!2 (NS) under no-till (Table 2). Looking at individual soil layers, the maximum increase in soil carbon occurred under conventional tillage in the 20–30 cm layer (P < 0.05), i.e. just below the plough layer. In all other soil layers of both systems, the changes in carbon stock were not statistically significant (Table 2). Comparing the two systems, conventional tillage had stored carbon and no-till had not (P < 0.05, Table 2) and the contrast was maximum in the 20–30 cm soil layer (P < 0.05).

153

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

Table 2 Changes in carbon and total nitrogen stocks since start of the experiment in different soil layers under conventional tillage and no-till, Ban Napok, Laos,, ns = non significant. Farming system

Difference between systems

Conventional tillage Mean change in four years TOC stocks (g C m!2) 0–40 cm +590 0–5 cm +46 5–10 cm 0 10–20 cm +124 20–30 cm +260 30–40 cm +161 TN stocks (g TN m-2) 0–40 cm 0–5 cm 5–10 cm 10–20 cm 20–30 cm 30–40 cm

+20 0 !3 +5 +14 +4

C:N ratio 0–40 cm 0–5 cm 5–10 cm 10–20 cm 20–30 cm 30–40 cm

+0.78 +0.83 +0.48 +0.64 +0.70 +1.10

* **

No-till Significant

Mean change in four years

Significant

*

!133 +67 !83 !53 !57 !8

ns ns ns ns ns ns

+9 +10 !4 +1 0 +1

ns ns ns ns ns ns

ns ns ns ns ns ns

*

**

ns ns ns *

ns ns ns ns ns *

ns *

* ns ns ns ns

!0.50 !0.56 !0.75 !0.65 !0.57 !0.11

Significant *

ns ns ns *

ns

ns

*

*

*

ns ns ns

ns ns

*

P < 0.05. P < 0.01.

A similar graph represents the changes in total nitrogen stocks in the two systems (Fig. 2a, b). Under conventional tillage, the 0– 40 cm profile had stocked on average some 20 g TN m!2 (NS); under no-till it was only 9 g TN m!2 (NS). Inspection of individual soil layers revealed that only under conventional tillage and in the 20–30 cm layer nitrogen had accumulated significantly (P < 0.05). Comparing the systems with each other, the nitrogen stocks were not significantly different (Table 2). Under conventional tillage, the C:N ratio increased significantly (Fig. 3a and b) from 9.8 to 10.6 (P < 0.05%), whereas under no-till the ratio decreased significantly from 10.7 to 10.2 (P < 0.05%). Under conventional tillage, changes were most pronounced in the 0–5 cm soil layer (P < 0.05%), under no-till in the 5–10 cm layer (P < 0.5%, Table 2). Significant differences between the two systems were found in all soil layers down to 20 cm (P < 0.05) and for the profile 0–40 cm (P < 0.01). Overall, the greatest contrast between no-till and conventional tillage was found in C:N ratio indicating that the systems had evolved in opposite directions. 3.2. Biomass production 3.2.1. Maize Significant differences in yield between the systems occurred only for the second year of the experiment, with higher values under conventional tillage (P < 0.01). The analysis of yields over the entire experimental period gave no significant differences between the systems but a strong year and a moderate system # year interaction effect (Table 3). Grain and biological yields showed a downward trend (year effect P < 0.005). Productivity declined in the second and again in the third year at a similar rate in both systems, and then stabilized (Table 3). The Interaction effect suggests that the experimental period can be divided into two parts, a period where yields are consistent with treatment i.e. the last three years, and a period in which this is not the case, i.e. the first two years. Though productivity had fallen compared to the beginning of the experiment, grain yields and biological yields

were consistently higher the last three years of the experiment under no-till compared to conventional tillage (P < 0.05 and P < 0.01, respectively, Table 3). The analysis of yield components demonstrates that higher grain yields under no-till are due to heavier single grain weight indicating superior growth conditions in these plots in the second half of the cropping cycle (Table 3). By contrast, growth conditions before and after flowering including the period of grain initiation, should be similar in both systems because number of cobs per plant and number of grains per cob are very similar. However, the taller and heavier plants produced under no-till indicate better overall growth conditions in these plots. 3.2.2. Cover crop and weeds The dry season biomass production in the no-till plots (Ruzi grass and weeds) was significantly higher (P < 0.01), compared to the biomass produced by weeds only in the conventional tillage plots (Table 4). Dry season biomass declined over the years (year effect P < 0.005) under conventional tillage and notill at a similar rate, falling sharply between the second and the third year and stabilizing afterwards. The biomass produced by Ruzi grass in the dry season declined over-proportionally because increasingly suppressed by the spontaneous vegetation. Ruzi grass contributed 95% to total dry season biomass the first year and 55% the last year. For the rainy season the opposite situation occurred, more biomass was produced in the conventional tillage plots by weeds alone than in the no-till plots by Ruzi grass and weeds together (Table 4). A diminishing biomass trend was observed over the years (year effect P < 0.005), which was similar in both systems. Weed and Ruzi grass biomass fell significantly between the second and the third year (P < 0.05) and then stabilized. The contribution of Ruzi grass to total non-crop biomass in the rainy season showed no obvious trend, from 47%, 81%, 40% to 54% for the first year to the fourth year, respectively. Competition of Ruzi grass or weeds with maize was not evident because maximum biomass occurrences coincided with maximum yield.

154

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

Fig. 2. Total nitrogen stocks in corresponding soil profiles and layers in 2002 and 2006 with line indicating 1:1 relationship; (a) conventional tillage; (b) no-till.

3.2.3. Cumulative biomass production Cumulative biomass production was computed for no-till and conventional tillage in order to estimate all organic inputs in the period between start and end soil sampling (Table 4). Cumulative maize residues were equivalent for the two systems, 18.39 Mg ha!1 dry weight produced in four seasons under conventional tillage and 18.82 Mg ha!1 under no-till. During the dry season more biomass was produced in the no-till plots than in the conventional tillage plots but it was the other way round for the rainy season. Computing the annual production, this difference disappears (Table 4), so that cumulative productivity by other plants than maize was very similar in the two systems, 22.5 Mg ha!1 dry weight under conventional tillage and 22.9 Mg ha!1 under no-till. Hence the cover crop Ruzi grass did not supply the system with extra biomass. The total quantity of biomass, i.e. maize residues and weeds, that was mostly incorporated into the soil under conventional tillage (40.9 Mg ha!1) can be considered identical to the total quantity of biomass, i.e. maize residues, Ruzi grass mulch and weeds, that remained on the soil surface under no-till (41.8 Mg ha!1). 3.3. Changes in weed vegetation The two-way indicator species table (TWINSPAN) has a blockwise arrangement along a diagonal because both releve´s and species have been sorted so that they form groups. The size of the

Fig. 3. C:N ratio in corresponding soil profiles and layers in 2002 and 2006 with line indicating 1:1 relationship; (a) conventional tillage; (b) no-till.

table was reduced by joining the four replicate plots into one mean value of cover score and by deleting 24 rare species (Table 5). The following plant communities emerge from the releve´s classification: a community of no-till, a community of conventional tillage and a community corresponding to the start of the experiment. Both no-till and conventional tillage plant communities were subdivided according to their dry or rainy season aspects. The dry season releve´s of the two systems are placed far apart because their respective species compositions were very different. On the other hand the rainy season releve´s of both systems had more species in common and were placed close to each other. The species group of the rainy season (Aeschynomene and others, Table 5) are mostly annual weeds which are unaffected by weed control, ploughing or herbicides. The species group that is associated with conventional tillage (Borreria and others, Table 5) comprised both annuals and perennials. This group is largely absent from the no-till plots and could suffer from the herbicide treatment applied there, or decline under the competition with the cover crop Ruzi grass, or both. However, looking at sample year and cover score for individual species (Table 5), a trend of progressive dominance of grasses in the conventional tillage plots (Paspalum,

155

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161 Table 3 Maize productivity affected by farming system and year, analysis of all five years of the experiment and the last three years, Ban Napok, Laos, , ns = non significant. Mean over five years

Grain yield (Mg ha!1 year!1) Plants ha!1 harvested Nb cobs plant!1 Nb grains cob!1 Single grain weight (g) Mean height at harvest (cm) Total biomass at harvest (Mg ha!1 year!1) Maize crop residue on soil (Mg ha!1 year!1)

Mean over last three years

Tillage

No-till

F-ratio and significant System

Year

Interaction

2.65 31900 0.97 354 0.264 193 7.27 4.10

2.78 33400 0.97 348 0.280 202 7.60 4.26

1.1 ns 1.9 ns 0.0 ns 0.2 ns 5.7* 8.4** 1.5 ns 0.4 ns

103.3*** 92.5*** 44.8*** 100.6*** 17.3*** 40.8*** 213.0*** 124.2***

0.7 ns 13.0*** 5.5*** 1.2 ns 0.7 ns 3.8* 3.5* 2.6*

Tillage

No-till

F-ratio and significant

1.65 29200 0.92 252 0.274 185 3.99 2.03

1.90 31700 0.99 256 0.295 201 5.02 2.71

System

Year

5.2* 0.8 ns 1.7 ns 0.0 ns 4.3 nsa 17.3*** 12.4*** 9.2**

20.9*** 26.7*** 43.2*** 14.9*** 6.9** 69.1*** 6.9** 0.5 ns

a

Almost significant P-value 0.05 is 4.45. P < 0.05. P < 0.01. *** P < 0.005. *

**

Table 4 Seasonal and cumulative dry weight production in the period between beginning and end soil sampling, with contribution of Maize, cover crop and weeds in conventional tillage and no-till plots, Ban Napok, Laos, ns = non significant. Effecta

Farming system Conventional tillage Mg ha Maize 2002–2005 Grain yield exported Crop residue Weeds and cover crop 2002–2005 Weeds dry season Cover crop dry season Total dry season

!1

year

!1

No-till P Mg ha!1

!1

Mg ha

year

!1

SMg ha

System

Year

!1

2.99 $0 64 4.60 $ 1.72

11.96 18.39

3.08 $ 0.56 4.70 $ 1.39

12.26 18.82

ns ns

***

2.05 $ 0.64 – 2.05 $ 0.64

8.21 – 8.21

0.74 $ 0.32 1.95 $ 0.93 2.80 $ 0.69

2.95 7.79b 11.21

***

*

Weeds rainy season Cover crop rainy season

3.58 $ 0.93 –

14.31 –

1.13 $ 0.45 1.80 $ 0.80

4.53 7.21

Total rainy season

3.58 $ 0.93

14.31

2.93 $ 0.89

Total non-crop biomass

5.63 $ 1.16

22.52

5.74 $ 1.39

***

–

–

*

***

*

***

–

–

11.74

ns

***

22.94

ns

***

a

Test on annual biomass productions. Only three dry seasons because dry season 2002 Ruzi grass not yet planted. **P < 0.05. P < 0.05. *** P < 0.005. b *

Pennisetum) and their encroachment into the no-till plots can be observed. Table 5 also includes a species group indifferent to treatment and season. A detrended correspondence analysis of species (DECORANA) was performed using the same input data. In Fig. 4 the group membership of the typical species of the three communities (Table 5) has been indicated. The first axis represents farming system: no-till is associated with low or negative scores, conventional tillage with higher scores and the vegetation at the start of the experiment with very high scores. Fig. 4, as it includes rare species, confirms the greater biodiversity under conventional tillage. The second axis represents the effect of season on species, low or negative scores are associated with the rainy season, and high scores with the dry season. Axis scores for all species are given in the Annexe. The species with a low axis 2 score are essentially shortlived annual weeds finishing their life cycle during the rainy season, whereas species with a high axis 2 score are either perennial species or annuals that germinate later in the cropping season finishing their life cycle at the end of the dry season (Annexe). 3.4. Other changes in the soil 3.4.1. Bulk density During the experiment, the bulk density increased significantly in both systems (P < 0.001). The process occurred at every soil

depth (P < 0.001) and at the same rate in both systems. From the initial density of 1.43 $ 0.102 g cm!3 (0–40 cm), the soil compacted under no-till to 1.51 $ 0.052 g cm!3 and under conventional tillage to 1.56 $ 0.067 g cm!3. The plough layer (0–20 cm) compacted slightly less than the corresponding layer under no-till. The differences between the farming systems were not statistically significant. We examined whether high organic matter content at the start of the experiment could have avoided soil compaction. A regression analysis of initial carbon concentration on changes in bulk density in the same soil sample gave no relation, for samples of all layers (n = 1440) and for distinct layers (n = 72). Although the positive effect of organic matter on maintaining soil structure is well known, the present carbon concentration was likely too low to produce such an effect. 3.4.2. Surface conditions Significant differences in surface conditions were found among the two treatments. During the dry season, Ruzi grass and weeds covered twice the surface in the no-till plots compared to weed cover in the conventional tillage treatment (75% and 30%, respectively, P < 0.01) and litter nearly threefold (73%, P < 0.001). Structural crusts were very rare under no-till (4%), but covered half of the plot surface under conventional tillage (P < 0.001). These crusts were associated with algae (16 $ 15% in the tilled plots, 2 $ 3% in the no-till plots). Cracks, size 7 $ 8 mm, in the tilled plots were nearly absent in the no-till plots (P < 0.05). No

156

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

Table 5 Species by releve´ table showing weed species groups associated with farming systems and initial state (four replicate releves were combined, 24 rare and accidental species were excluded, mean cover scores 1 = 75%).

§

d = dry season, r = rainy season. #Ruzi grass, the planted cover crop. $fodder grass planted in the year preceding the experiment to uniform experimental conditions.

significant difference was found among the other conditions examined, i.e. worm casts and termite constructions. 3.4.3. Soil macrofauna Species richness was low with 1.8 $ 1.17 and 3.3 $ 1.50 morphospecies extracted per monolith in the conventional tillage and in the no-till plot, respectively. No significant difference was demonstrated between the systems regarding total species richness, total abundance, and abundance for all the six morphospecies. These low values could have been induced by the time of sampling, at the end of the dry season, or indicate natural low levels of macrofauna

activity in these soils. The application of water to mimic the onset of the rainy season led to a significant (P < 0.05) production of earthworm casts, namely on average 65 g m!2 of fresh cast weight on the soil surface in the no-till plot but had no consequence in the conventional tillage plot where only 14 g m!2 of fresh casts was produced overnight. These findings indicate a much larger population of earthworms in the soil under no-till. 3.4.4. Roots Both farming systems lost coarse roots during the experimental period (P < 0.01). At the start of the experiment, in the 0–40 cm

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

157

Fig. 4. Detrended correspondance diagram (DECORANA) of the 68 species recorded in the permanent plots showing clustering of species, first axis reflecting land use, second axis low scores associated with rainy season, high scores with dry season.

profile, coarse roots averaged 2740 g m!3 of soil. Under no-till, coarse roots decreased to 1240 g m!3 soil, and under conventional tillage to 2220 g m!3 soil. Under conventional tillage, these losses were concentrated in the 0-5 cm layer (P < 0.001) and the 5–10 cm layer (P < 0.01) and no changes in root weights were measured in deeper layers. Under no-till losses in root weight occurred not only in the top layers 0–5 cm (P < 0.001) and 5–10 cm (P < 0.001), but also further down, 10–20 cm (P < 0.05). Fine root colonization, expressed as root length density i.e. the root length per unit volume of soil, decreased with soil depth under both conventional tillage and no-till but there was no significant difference between the two treatments at any of the three soil depth increment (Fig. 5). There was however a tendency for more roots in the plough layer compared to undisturbed soils under notill while below the plough layer, at 25 cm depth, the inverse occurred: more fine roots were found under no-till than under conventional tillage. Roots were significantly thicker, 0.56 mm on average (P < 0.01, data not shown) under no-till at a soil depth of 10 cm, probably reflecting the abundance of Ruzi grass roots therein the sub-surface soil layer.

Further we determined the soil medium preferred for the colonization by fine roots, macropores, cracks or dense matrix. The distribution of these categories was different across treatments. Large macropores (diameter > 1 mm) predominated under no-till but were virtually absent under conventional tillage. The inverse occurred for cracks (data not shown). Irrespective of tillage treatment or depth of sampling, fine root colonization tended to occur preferably in the presence of cracks (P < 0.1 at 25 cm depth). Cracks cumulate three advantages for roots: low mechanical resistance, good aeration and relatively stable water supply whereas macropores that are either filled or empty, only offer two: low mechanical resistance and stable water supply. 3.4.5. Infiltration and porosity The hydraulic behaviour of the two systems was similar across the range of water pressure head !60 < h0 < !20 mm (Fig. 6) but towards saturation, that is for the water pressure head h0 = !5 mm, the soil under no-till demonstrated a higher hydraulic conductivity (P < 0.05). However, total pore space (data not

Fig. 5. Box–whisker plots of root length density as a function of tillage treatment, at three soil depth increments. Central line indicates median value; upper and lower edges (hinges) of the box indicate 25th and 75th percentile values; whiskers indicate maximum and minimum values.

158

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

Fig. 6. Box plots of hydraulic conductivity (K0), sorptivity (S0) and mean effective pore size (lm) as a function of water pressure head for conventional tillage and notill. Central line indicates median value, upper and lower edges of the box indicate 25th and 75th percentile values; whiskers indicate maximum and minimum values.

cropping period during which the maize is well fertilized. Fast growing, long-lived annuals or perennials capable of rapid accumulation of biomass throughout the dry season were thus favoured. Typical examples are the tall grass Pennisetum polystachion and the stout herbs Hyptis suaveolens, Ludwigia hyssopifolia and Malvastrum coromandelianum, of which many individuals reach 1.5–2 m height. At the end of the dry season the biggest individual plants withstand the herbicide treatment. Under conventional tillage, the efficiency of the weed community to capture available water and nutrients approached that of the cover crop. Thirdly, the biomass inputs on an annual base give similar rates in both systems. Ruzi grass produced a modest surplus of biomass in the dry season, and suppressed weed growth in the rainy season. Under conventional tillage, the inverse was the case: more weed biomass in the rainy season, less so in the dry season. Added to this the declining influence of Ruzi grass over the years so that the cumulative biomass production is very much the same in both systems. In other studies but not in this experiment, no-till had adverse effects on maize yields (Deen and Kataki, 2003) and on biological yields of both maize and legume cover crop (Monneveux et al., 2006). When a perennial cover crop is added to the cropping cycle and subsequent carbon sequestration occurs, the latter is commonly attributed to the increased time under which the soil was vegetated, secondly to the relative high allocation of belowground carbon that is typical of perennials and thirdly to the formation of stable aggregates within the extensive network of their roots (Paustian et al., 1997; West and Post, 2002; Bernoux et al., 2006; Wright et al., 2007). In this study, the advantages of perennials to accumulate above and below ground biomass were rather equally distributed among the systems. Ruzi grass was not the only perennial in the experiment; the weed community in the control plots comprised many perennials and annual species that became perennial under favourable conditions (Annex). 4.2. Carbon and nitrogen storage

shown), bulk density and mean functional pore size were similar in both systems. The superior infiltration under no-till can only be explained by a better connectivity of pores, which logically results from a superior biological activity in the no-till plots. 4. Discussion 4.1. Biomass inputs The biological yields under conventional tillage were equal to those under no-till, in spite of a cover crop that was alleged to supply extra inputs of biomass. The assumption that a cover crop adds pure gains of biomass to the system is based on little proof. In many experiments the production of the cover crop was roughly estimated and the weedy biomass produced by the control underestimated (Barthe`s et al., 2004), or unknown (e.g. Bayer et al., 2006; Metay et al., 2007). Sainju et al. (2007) found that weeds could equal or out yield the biomass production of most of the cover crops that were screened in the experiment including Ruzi grass. At least three factors explain why the cover crop in this study failed to yield superior inputs compared to the control. Firstly, the productivity of the cover crop was not constant but decreased over the years. Ruzi grass met with growing competition with weeds as high stature weeds shaded out the low statue cover crop. This competition reduced the vigour of Ruzi grass particularly in the dry season. A similar decline in productivity of the cover crop in the second year compared to the first year was measured by Affholder et al. (2008) and Sainju et al. (2007). Secondly, the weed population adapted. Weeds responded to the relative short

The point with our findings is that despite the equal contribution of inputs and the absence of erosion, the conventional system stored significantly carbon, whereas the no-till system lost carbon, and above that, the difference in storage between the two systems was significant. In this study most carbon accumulated in the layer just below the plough depth. A similar result was observed by Angers and Carter (1996), Deen and Kataki (2003), Wright et al. (2007) of which Angers et al. (1997) found a plausible explanation. At the first ploughing, here 20–25 cm, some residues are placed at a maximum depth. With continued ploughing, residues are mixed throughout the plough layer, however, the residues buried at the deepest level of ploughing have less chance of being mixed and they may persist over the period of experimentation. Under no-till, carbon accumulation typically occurs in the surface layers. For instance, carbon was found to accumulate to within a few centimetres of the soil surface (Franzluebbers, 2002; Manley et al., 2005). This was not evident in this study. Either the 0–5 cm surface layer was too thick to show accumulation, or the period of experimentation was too short to develop such a distinctive layer. In our study no-till retained some nitrogen (9 g TN m!2 in 0–40 cm soil) while losing carbon. Many studies confirm that under no-till relatively more nitrogen than carbon is stored. Alvarez (2005) observed that the more nitrogen is available, the more carbon is stored and concluded that nitrogen fertilization is necessity for carbon storage. It is common that the C:N ratio declines when tilled farmland is converted to no-till pasture, for

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

instance from an initial 16.9 to 9.4, 12 years later (Franzluebbers and Stuedemann, 2009). Generally, the C:N ratio is higher under conventional tillage than under no-till (Wright et al., 2007; this study). Where this is not the case, for instance in the long-term study of Sa´ et al. (2001), this could be due to the regular removal of crop residues for forage in the tillage system thus depriving the plough layers of fresh biomass that has a high C:N. 4.3. Why no-till lost carbon Soil loosening as performed by tillage aerates the soil thus accelerating microbial oxidation of organic matter leading to CO2 production (Kuipers, 1991). At the same time, the carbon losses are compensated for by the restitution of crop residues. However, by incorporating the crop residues into the soil their availability for loss increases because they are placed closer to the decomposers (Angers et al., 1997; Bernoux et al., 2006; Spargo et al., 2008). What is measured under conventional tillage is the balance between losses of carbon due to stimulated decomposition – aeration, fresh material availability under favourable conditions – and gains of carbon through the regular incorporation of residues. In this study, after four years of treatment, the balance is positive for carbon. The massive input of fresh material into the plough layer has resisted to drastic oxidation and has only partly been worked by organisms. No-till avoids these losses from microbial decomposition but simultaneously lacks the inputs of fresh organic material deep in the soil. There is no compensation other than the mulch layer at the soil surface. What is measured under no-till is the balance between some losses of soil carbon due to organic matter oxidation despite relative unfavourable conditions deeper in the soil, and gains of carbon in superficial layers. In our study the balance is neutral to slightly negative for carbon. Only part of the mulch is successfully worked and incorporated in the soil by the soil fauna, the rest of plant residue is apparently lost through open-air decay. 4.4. Functioning of the maize crop 4.4.1. Soil improvement Significantly less crusting was demonstrated under no-till, while in the conventional treatment algae developed at the soil surface along with crusts and both hamper infiltration (Malam Issa et al., 2009). At the onset of the rains, crusts and algae are destroyed by tillage every year but reappear because the soil lacks sufficient cover. A higher hydraulic conductivity and a better pore connectivity were also demonstrated under no-till, all this favouring water infiltration. Both systems had similar fine root mass which should be unsurprising because the above ground biomass was also similar. The coarse roots in the soil were probably a remnant of the preexisting vegetation that included woody and long-lived species that were replaced by herbaceous plants during the experiment (Annex). The loss of 55% of coarse roots under no-till compared to only 29% under conventional tillage suggests a greater macrofauna and microbial community that fed on them. This suggestion is supported by a sharp decline in C:N ratio in those soil layers where most coarse roots disappeared. A pronounced decline in C:N ratio indicates that rough biomass is being thoroughly digested by soil fauna and bacteria. The presence of earthworms in the subsoil under no-till, while these were near-absent under conventional tillage also indicate a better soil structure and better conditions for plant growth under no-till. A similar soil improvement was reported by Hulugalle et al. (1994) but without effect on maize yields – 30% increase in earthworm activity, greater infiltration rates and better interconnectivity of pores after five years of no-till.

159

4.4.2. Yield and yield components This study demonstrated, from the third year onward, significantly taller maize plants, and more grain yield and biomass under no-till. In maize, climate and soil relationships have a profound influence on the sequence of developmental events that culminate in the formation of the grain (Norman et al., 1995). Because climate and fertility conditions were equal in both treatments, soil water availability should be responsible for higher yields under no-till. Under favourable moisture conditions, extra dry matter accumulation and stem internode elongation can occur in all growth stages up to grain filling (C¸akir, 2004). In this study, a difference in soil moisture between the treatments would have been large enough to allow a differentiation in biomass accumulation but was too small to cause negative effects in yield components, i.e. fewer cobs per plant and fewer grains per cob under conventional tillage. The yield component mainly responsible for higher yield under no-till was single grain weight. For grain filling, water availability is of minor importance, instead, source or sink limitations determine the process (Tollenaar, 1977; NeSmith and Ritchie, 1992). If sink limitation occurs, the individual grains in a cob compete with each other; hence heavy grains are correlated with few grains per cob. If source limitation occurs, the supply of assimilates to fill the sinks is determinant. In this experiment, single grain weight did not vary with grain number, thus single grain weight was limited chiefly by source capacity. Borra´s and Otegui (2001) found that the supply of assimilates for grain filling depended on plant weight and post-flowering biomass accumulation. This means that every maize plant contributes to grain filling according to its ability, heavier plants contribute more than lighter ones. The slightly smaller maize plants of the conventional tillage plots were not capable to produce as much assimilates to fill the grains to the same level as the larger maize plants in the no-till treatment did. 4.4.3. Weeds No-till commonly results in increased weed pressure (Locke et al., 2002; Giller et al., 2009). In this study weed problems increased every year, requiring other types of herbicides after the second year to supplement the glyphosate treatment. Weed problems were due to shifts in weed populations. Worldwide, perennial species and grassy weeds are more strongly associated with no-till than with conventional tillage (Derksen et al., 2002; Locke et al., 2002; Kobayashi et al., 2003; this study). Perennials were found to increase in less than one year after the last tillage (Blackshaw, 2005). A perennial plant, once it has escaped a first herbicide treatment, accumulates such an amount of biomass that killing it with herbicides the next year is very difficult. The reduced weed diversity under no-till observed in this study is probably due to less species present in superficial layers than in the plough layer because seeds under no-till can only infiltrate into the soil via soil fauna activity and cracks. Secondly, only a limited number of species is capable of emergence at the soil surface, and this excludes most hard-seeded and large seeds that need a greater soil–seed contact (Blackshaw, 2005). Very little is known about these traits in tropical weeds. 5. Conclusion Mandal et al. (2007) predicted that in the tropics because of high temperature more oxidation products per unit organic matter is produced compared to cool and temperate regions. Therefore expectation of carbon storage under no-till should not be too high. In this study, no-till did not store soil carbon despite high inputs of crop residues and mulch and the subsequent stimulation of soil biological activity. The process of transforming the available mulch

160

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161

biomass into soil organic matter is apparently too slow to avoid direct losses. Under conventional tillage some of these losses were avoided by working crop residues and weeds into the soil using tillage and subsequently, soil carbon increased. More study is needed to assess the stability of this carbon, e.g. by estimating turnover rates and microbial activity. After one preparatory year and two experimental years, maize yields stabilized at a moderate level. In this latter period, grain yields and crop residues of maize were 16% and 34% respectively higher under no-till compared to conventional tillage. Higher yields are attributed to more soil water available under no-till due to an improved soil structure although bulk density was not affected. The permanent mulch layer protecting the soil surface, favoured infiltration by keeping it crust-free. Water availability was further promoted by a better connectivity of pores and more macrofauna. A negative point is that the no-till system depends heavily on herbicide use. The lack of effectiveness of herbicides against shifting weed communities threatens the continuation of the system. Acknowledgements We are greatly in debt to the group of Laotian agronomists devoted to independent science who saved the paper, in particular Mrs Phirnphit Soysouvanh. Thanks are due to Mark Newman and Kate Armstrong, Royal Botanic Garden Edinburgh, and Peter van Welzen and Frits Adema, Herbarium Leiden for ‘nasty’ plant identifications. Additional financial support was brought by FFEM (Fonds Franc¸ais pour l’Environnement Mondial). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.agee.2009.12.013. References Affholder, F., Jourdain, D., Morize, M., Quang, D.D., Ricome, A., 2008. Eco-intensification dans les montagnes du Vietnam, contraintes a` l’adoption de la culture sur couvertures ve´ge´tales. Cahiers Agricultures 17 (3), 289–296. Alvarez, R., 2005. A review of nitrogen fertilizer and conservation tillage affects on soil organic carbon storage. Soil Use and Management 21, 38–52. Anderson, J.M., Ingram, J.S.I., 1993. Tropical Soil Biology and Fertility. A Handbook of Methods. CAB International, Wallingford, Oxon, U.K.. Angers, D.A., Carter, M.R., 1996. Aggregation and organic matter storage in cool, humid agricultural soils. In: Carter, M.R., Stewart, B.A. (Eds.), Structure and Organic Matter Storage in Cool, Humid Agricultural Soils. Advances in Soil Science, Lewis–CRC Press, Boca Raton, USA, pp. 193–211. Angers, D.A., Bolinder, M.A., Carter, M.R., Gregorich, E.G., Drury, C.F., Liang, B.C., Voroney, R.P., Simard, R.R., Donald, R.G., Beyaert, R.P., Martel, J., 1997. Impact of tillage practices on organic carbon and nitrogen storage in cool, humid soils of eastern Canada. Soil & Tillage Research 41, 191–201. Baker, J.M., Ochsner, T.E., Venterea, R.T., Griffis, T.J., 2007. Tillage and soil carbon sequestration – what do we really know? Agriculture, Ecosystems and Environment 118, 1–5. Barthe`s, B., Azontonde, A., Blanchart, E., Girardin, C., Villenave, C., Lesaint, S., Olivier, R., Feller, C., 2004. Effect of a legume cover crop (Mucuma pruriens var. utilis) on soil carbon in an Ultisol under maize cultivation in southern Benin. Soil Use and Management 20, 231–239. Bayer, C., Martin-Neto, L., Mielniczuk, J., Pavinato, A., Dieckow, J., 2006. Carbon sequestration in two Brazilian cerrado soils under no-till. Soil & Tillage Research 86, 237–245. Bernoux, M., Cerri, C.C., Cerri, C.E., Siqueira Neto, M., Metay, A., Perrin, A.-S., Scopel, E., Razafimbelo, T., Blavet, D., de Piccolo, M.C., Pavei, M., Milne, E., 2006. Cropping systems, carbon sequestration and erosion in Brazil, a review. Agronomy and Sustainable Development 26, 1–8. Blackshaw, R.E., 2005. Tillage intensity affects weed communities in agroecosystems. In: Invasive Plants: Ecological and Agricultural Aspects. Inderjit, Birkha¨user Verlag, Switzerland, pp. 209–221. Borra´s, L., Otegui, M.E., 2001. Maize kernel weight response to post-flowering source-sink ratio. Crop Science 49, 1816–1822.

Braun-Blanquet, J., 1964. Pflanzensoziologie. Grundzu¨ge der Vegetationskunde, third ed. Springer-Verlag, Vienna. C¸akir, R., 2004. Effect of water stress at different development stages on vegetative and reproductive growth of corn. Field Crops Research 89, 1–16. Corbeels, M., Scopel, E., Cardoso, A., Bernoux, M., Douzet, J.M., Neto, M.S., 2006. Soil carbon storage potential of direct seeding mulch-based cropping systems in the Cerrados of Brazil. Global Change Biology 12, 1773–1787. Deen, W., Kataki, P.K., 2003. Carbon sequestration in a long-term conventional versus conservation tillage experiment. Soil & Tillage Research 74, 143– 150. Derksen, D.A., Anderson, R.L., Blackshaw, R.E., Maxwell, B., 2002. Weed dynamics and management strategies for cropping systems in the Northern Great Plains. Agronomy Journal 94, 174–185. Diekow, J., Mielniczuk, J., Knicker, H., Bayer, C., Dick, D.P., Ko¨gel-Knabner, I., 2005. Carbon and nitrogen stocks in physical fractions of a subtropical Acrisol as influenced by long-term no-till cropping systems and N fertilisation. Plant and Soil 268, 319–328. Ellert, B.H., Bettany, J.R., 1995. Calculations of organic matter and nutrients stored in soils under contrasting management regimes. Canadian Journal of Soil Science 75, 529–538. Franzluebbers, A.J., 2002. Water infiltration and soil structure related to organic matter and its stratification with depth. Soil & Tillage Research 66, 197–205. Franzluebbers, A.J., Stuedemann, J.A., 2009. Soil-profile organic carbon and total nitrogen during 12 years of pasture management in the Southern Piedmont USA. Agriculture Ecosystems and Environment 129, 28–36. Giller, K.E., Witter, E., Corbeels, M., Tittonell, P., 2009. Conservation agriculture and smallholder farming in Africa: the heretics’ view. Field Crops Research 114, 23– 34. Hill, M.O., 1979a. TWINSPAN—a FORTRAN programme for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Cornell University, Ithaca, New York, USA. Hill, M.O., 1979b. DECORANA—a FORTRAN programme for detrended correspondence analysis and recoprocal averaging. Cornell University, Ithaca, New York, USA. Hulugalle, N.R., Ezumah, H.C., Leyman, T., 1994. Changes in surface soil properties of a no-tilled tropical Alfisol due to intercropping maize, cassava and ‘egusi’ melon. Field Crops Research 36, 191–200. Kobayashi, H., Nakamura, Y., Watanabe, Y., 2003. Analysis of weed vegetation of notillage upland fields based on the multiplied dominance ratio. Weed Biology and Management 3, 77–92. Kuipers, H., 1991. Agronomic aspects of ploughing and non-ploughing. Soil & Tillage Research 21, 167–176. Lal, R., Wilson, G.F., Okigbo, B.N., 1978. No-till farming after various grasses and leguminous cover crops in tropical alfisol. I. Crop performance. Field Crop Research 1, 71–84. Locke, M.A., Reddy, K.N., Zablotowicz, R.M., 2002. Weed management in conservation crop production systems, review. Weed Biology and Management 2, 123– 132. Malam Issa, O., De´farge, C., Trichet, J., Valentin, C., Rajot, J.-L., 2009. Microbiotic soil crusts in the Sahel of Western Niger and their influence on soil porosity and water dynamics. Catena 77, 48–55. Mandal, B., Majumder, B., Bandyopadhyay, P.K., Hazra, G.C., Gangopadhyay, A., Samantaray, R.N., Mishra, A.K., Chaudhury, J., Saha, M.N., Kundu, S., 2007. The potential of cropping systems and soil amendments for carbon sequestration in soils under long-term experiments in subtropical India. Global Change Biology 13, 357–369. Manley, J., van Kooten, G.C., Moeltner, K., Johnson, D.W., 2005. Creating carbon offsets in agriculture through no-till cultivation: a meta-analysis of costs and carbon benefits. Climatic Change 68, 41–65. Metay, A., Alves Moreira, J.A., Bernoux, M., Boyer, T., Douzet, J.-M., Feigl, B., Feller, C., Maraux, F., Olivier, R., Scopel, E., 2007. Storage and forms of organic carbon in a no-tillage under crop cover crops system on clayey Oxisol in dryland rice production (Cerrados, Brazil). Soil & Tillage Research 94, 122–132. Monneveux, P., Quille´rou, E., Sanchez, C., Lopez-Cesati, J., 2006. Effect of zero tillage and residues conservation on continuous maize cropping in a subtropical environment (Mexico). Plant and Soil 279, 95–105. Needelman, B.A., Wander, M.M., Bollero, G.A., Boast, C.W., Sims, G.K., Bullock, D.G., 1999. Interaction of tillage and soil texture: biologically active soil organic matter in Illinois. Soil Society of America Journal 63, 1326–1334. NeSmith, D.S., Ritchie, J.T., 1992. Maize (Zea mays L.) response to a severe soil waterdeficit during grain-filling. Field Crop Research 29, 23–35. Norman, M.J.T., Pearson, C.J., Searle, P.G.E., 1995. The Ecology of Tropical Food Crops, second ed. Cambridge University Press, USA. Nye, P.H., Greenland, D.J., 1960. The soil under shifting cultivation. Technical Communications of the Commonwealth Bureau of Soil Science 51, Harpenden, UK. Ogle, S.M., Jay Breidt, F., Paustian, K., 2005. Agricultural management impacts on soil organic carbon storage under moist and dry climatic conditions of temperate and tropical regions. Biogeochemistry 72, 87–121. Paustian, K., Andre´n, O., Janzen, H.H., Lal, R., Tian, G., Tiessen, H., van Noordwijk, M., Woomer, P.L., 1997. Agricultural soils as a sink to mitigate CO2 emissions. Soil Use and Management 13, 230–244. Pierret, A., Latchackak, K., Chathanvongsa, P., Sengtaheuanghoung, O., Valentin, C., 2007. Interactions between root growth, slope and soil detachment depending on land use: a case study in a small mountain catchment of Northern Laos. Plant and Soil 301, 51–64.

A. de Rouw et al. / Agriculture, Ecosystems and Environment 136 (2010) 148–161 Podwojewski, P., Orange, D., Jouquet, P., Valentin, C., Thiet, N.V., Janeau, J.L., Toan, T.D., 2008. Influence of land use changes on runoff and erosion from micro-plots and a catchment in northern Vietnam. Catena 24, 109–118. Razafimbelo, T.M., Albrecht, A., Basile, I., Borschneck, D., Bourgeon, G., Feller, C., Ferrer, H., Michellon, R., Moussa, N., Muller, B., Oliver, R., Razanamparany, C., Seguy, L., Swarc, M., 2006. Effet de diffe´rents syste`mes de culture a` couverture ve´ge´tale sur le stockage du carbone dans un sol argileux des Hautes Terres de Madagascar. Etude et gestion des sols 13, 113–127. Sa´, J.C.M., Cerri, C.C., Dick, W.A., Lal, R., Venske Filho, S.P., Piccolo, M.C., Feigl, B.E., 2001. Organic matter dynamics and carbon sequestration rates for a tillage chronosequence in a Brazilian Oxisol. Soil Society of America Journal 65, 1486– 1499. Sainju, U.M., Schomberg, H.H., Singh, B.P., Whitehead, W.F., Tillman, P.G., LachnichtWeyers, S.L., 2007. Cover crop effect on soil carbon fractions under conservation tillage cotton. Soil & Tillage Research 96, 205–218. Spargo, J.T., Alley, M.M., Follett, R.F., Wallace, J.V., 2008. Soil carbon sequestration with continuous no-till management of grain cropping systems in the Virginia coastal plain. Soil & Tillage Research 100, 133–140. Tollenaar, M., 1977. Sink–source relationships during reproductive development in maize, a review. Maydica 12, 49–75. Valentin, C., Bresson, L.M., 1992. Morphology, genesis and classification of soil crusts in loamy and sandy soils. Geoderma 55, 225–245.

161

Valentin, C., Agus, F., Alamban, R., Boosaner, A., Bricquet, J.P., Chaplot, V., de Guzman, T., de Rouw, A., Janeau, J.L., Orange, D., Phachomphonh, K., Phai, Do, Podwojewski, P., Ribolzi, O., Silvera, N., Subagyono, K., Thie´baux, J.P., Tran Duc Toan, Vadari, T., 2008. Runoff and sediment losses from 27 upland catchments in Southeast Asia: impact of rapid land use changes and conservation practices. Agriculture, Ecosystems & Environment 128, 225–238. Vandervaere, J.-P., Peugeot, C., Vauclin, M., Angulo-Jaramillo, R., Lebel, T., 1997. Estimating hydraulic conductivity of crusted soils using disc infiltrometers and minitensiometers. Journal of Hydrology 188-189, 203–223. West, T.O., Post, W.M., 2002. Soil organic carbon sequestration rates by tillage and crop rotation: a global data analysis. Soil Society of America Journal 66, 1930– 1946. West, T.O., Six, J., 2007. Considering the influence of sequestration duration and carbon saturation on estimates of soil carbon capacity. Climate Change 80, 25– 41. White, I., Sully, M.J., 1987. Macroscopic and microscopic capillary length and time scales from field infiltration. Water Resources Research 23, 1514– 1522. Wright, A.L., Dou, F., Hons, F.M., 2007. Soil organic and N distribution for wheat cropping systems after 20 years of conservation tillage in central Texas. Agriculture, Ecosystems & Environment 121, 376–382.