at SciVerse ScienceDirect

Crop Protection 53 (2013) 23e28Contents lists availableCrop Protection

journal homepage: www.elsevier .com/locate/croproWeed control in conservation agriculture systems of Zimbabwe:Identifying economical best strategies

Tarirai Muoni a,*, Leonard Rusinamhodzi b, Christian Thierfelder b

aUniversity of Zimbabwe, Faculty of Agriculture, Crop Science, P.O. Box MP 167, Mount Pleasant, Harare, Zimbabweb International Maize and Wheat Improvement Centre (CIMMYT), Southern Africa Regional Office, P.O. Box MP 163, Mount Pleasant,12.5 Km Peg Mazowe Road, Harare, Zimbabwea r t i c l e i n f o

Article history:Received 26 November 2012Received in revised form31 May 2013Accepted 4 June 2013

Keywords:Conservation agricultureWeed managementHerbicide useSmallholder farmersEconomic benefits* Corresponding author. Tel.: 263 774311136.E-mail address: tarirai.muoni@gmail.com (T. Muon

0261-2194/$ e see front matter 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.cropro.2013.06.002a b s t r a c t

Weedmanagement under conservation agriculture (CA), especiallywhenmanually controlled is one of themajor setbacks for thewidespread adoption of CA in southern Africa. This studywas conducted at three on-station and three on-farm sites: CIMMYT-Harare, Domboshawa Training Centre and Henderson ResearchStation (on-station sites), Hereford farm, Madziva communal area and Shamva communal area (on-farmsites). The evaluation focused on the effect of initial herbicide application and succeedingmanual weedingwhenever weeds were 10 cm tall or 10 cm in length for grasses with stoloniferouserhizomatous growthhabit. Weeds counts, weeding time and grain yields were collected at all on-station sites. At the on-farmsites, weed counts were done before weeding and a number of farmers were timed during weeding. Theresults showed that herbicides use reduced the weed density and time taken on weeding at all sites.Combining herbicides e.g. atrazine, glyphosate andmetalachlor had the lowest weed density andweedingtime at all sites. However, the treatments had no effect on maize grain yields suggesting that appropriateand timelymanualweeding reduced crop/weed competition.Herbicides treatments had higher input coststhan manual weeding due to the additional cost of herbicide but the treatment with manual weeding onlyhad more overall labour days compared to the mixture of three herbicides. In order to achieve economicbenefits, smallholder farmers may use the time for value addition e.g. expand cropped land area, use timefor value addition, or sell newproducts on themarket. Herbicides use reduces themanual labour needed tocontrol weeds andminimise total crop failure due to untimelyweeding hence, herbicides are an importantbut not the only weed control option under CA systems in Zimbabwe.

2013 Elsevier Ltd. All rights reserved.1. Introduction

Conservation agriculture (CA) is a crop management systembeing promoted in southern Africa due to its potential to conserve,improve and make efficient use of water and nutrients (FAO,2002). CA is based on three principles: a) zero or minimum soildisturbance, b) maintenance of permanent organic soil cover andc) diverse crop rotations and associations (Kassam et al., 2009).Potential benefits of CA include reduced soil erosion and waterrun-off (Derpsch et al., 1986; Thierfelder and Wall, 2009),increased rainfall use efficiency (Thierfelder and Wall, 2009), earlyplanting (Haggblade and Tembo, 2003), increased soil quality andbiological activity (Thierfelder and Wall, 2010), and savings in on-farm labour (Sorrenson et al., 1998).i).

All rights reserved.Smallholder farmers in southern Africa face a number of chal-lenges including low income, insufficient technical knowledge,small land sizes (average of 2.1 ha) and poor farming equipment(Chamunorwa, 2010). The elimination of conventional tillagepractices results in increased weeding pressure especially in theearly years (Vogel, 1994b) and the need for effectiveweed control islimited by insufficient farm labour. No-tillage combined with res-idues under CAmay also lead to a change in themicro-environmentleading to a shift in weed flora intensifying the weed managementproblems (Derpsch, 2008).

Many smallholder farmers in Zimbabwe rely on manual weedcontrol using hand hoes (Mandumbu et al., 2011), which is labourintensive and is slowed down by the presence of residues (Vogel,1994b). Hand hoeing may require up to four weeding times duringthe cropping season for effective weed control (Mashingaidze et al.,2012). Therefore, there is need for effective weed control strategiesthat reduce labour requirement while being feasible within thefarmers circumstances. Weed management practices such asmanual weeding and herbicides facilitate the decrease in weed

Delta:1_given nameDelta:1_surnameDelta:1_given namemailto:tarirai.muoni@gmail.comhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.cropro.2013.06.002&domain=pdfwww.sciencedirect.com/science/journal/02612194http://www.elsevier.com/locate/croprohttp://dx.doi.org/10.1016/j.cropro.2013.06.002http://dx.doi.org/10.1016/j.cropro.2013.06.002http://dx.doi.org/10.1016/j.cropro.2013.06.002

T. Muoni et al. / Crop Protection 53 (2013) 23e2824pressure (Norsworthy and Frederick, 2005). Integration of highlycompetitive green manure cover crops into the farming systemsmay also reduce weed pressure (Caamal-Maldonado et al., 2001).However, the availability of appropriate knowledge on integratedweed management is not available at farm level. Use of greenmanure cover crops e.g. velvet beans (Mucuna pruriens L.) is notperfectly adapted to the smallholder farmers circumstances, as thegrain cannot be easily consumed without significant processing.

Herbicides are a potential strategy for effective weed control un-der CA. Glyphosate [N-(phosphonomethyl) glycine], paraquat (1,10-dimethyl-4,40-bipyridinium), metalachlor (2-chloro-N-(2-ethyl-6-methylphenyl)-N-2-methoxy-1-methylethyl) acetamide and atra-zine (2-Chloro-4-ethylamino-6-isopropylamino-1,3,5-triazine) arereadily available products at most chemical agro-input suppliers ofZimbabwe. However, most of these enterprises operate in urbanareaswhichandcannotbeeasilyaccessedby farmers in remote areas.Some herbicides such as atrazine can bemixedwith other herbicides(e.g. glyphosate andmetalachlor). This increases the controlledweedspectrum and reduces herbicide failure.

Although the use of atrazine is relatively common and widelypromoted under maize production, its residual effects are likely toaffect the succeedingbroadleavedcrops (e.g. a legumecrop followingmaize in a maize-legume rotation). In addition, the use of atrazineraised concerns over human and animal toxicity when it was foundin drinking water but currently the cumulative risk assessment ofthis herbicide is not yet conclusive (Williams et al., 2010).

Few economic studies have been carried out to provide evidenceand support the widespread use and feasibility of herbicides in CAunder smallholder farmers conditions in southern Africa. Theobjective of this study was to evaluate the effectiveness and eco-nomic benefits of manual and chemical weed control strategiesunder CA in Zimbabwe.

2. Materials and methods

2.1. Site description

The experiments were established at six sites; three on-stationand three on-farm sites that receive rainfall in one main croppingseason between November and March in a unimodal pattern. Thesites generally have a long dry period (April to October) where Aprilto July is cool and August to October is warm. All sites are in theZimbabwean agro-ecological region II characterised by annualrainfall of between 700 and 1000 mm. Mean maximum daily tem-peratures during summer can exceed 32 C. The region is suitable forintensive crop and livestock production.

2.1.1. Research station sitesCIMMYT-Harare (17800S, 31500E and 1503 m.a.s.l.), is a

research station located at the University of Zimbabwe Farm on redclay soil, classified as Chromic Luvisol. The soils are characterised byhigh clay content of up to 40% and organic matter content with apronounced granular structure (Nyamapfene, 1991). DomboshawaTraining Centre (DTC) (17370S 31100E and 1560 m.a.s.l.) site ischaracterised by moderately deep Arenosols and Luvisols under FAOclassification (Vogel, 1994b) that have 5% clay content and arederived from granite parent material. Henderson Research Station,HRS, (17 340S, 30 540E and 1136 m.a.s.l.) is on sandy soils (Are-nosols) of poor fertility (>80% sand) derived from granite parentmaterial (Thierfelder and Wall, 2009).

2.1.2. On-farm sitesThe on-farm sites include Hereford farm (17420S, 31450E and

1077 m.a.s.l.), a former commercial farm located in Bindura district,characterised by heavy red clay soils of up to 40% clay contentand rich in organic matter (Nyamapfene, 1991), classified as Chro-mic Luvisols. Madziva communal area (17010S, 31410E and1181 m.a.s.l.) in Shamva district, is characterised by sandy soilsclassified as Arenosols derived from granite parent material(Thierfelder et al., 2012). Chikato School (17190S, 31490E and1161m.a.s.l.) also in Shamva district has Chromic Luvisols rich in clayand organic content. Farmers at all the three on-farm sites growmaize as a staple crop and some cash crops such as cotton (Gos-sypium hirsutum L.) and tobacco (Nicotiana tabacum L.). Cowpeas(Vigna unguiculata (L.) Walp), soyabeans (Glycine max (L.) Merr.)and groundnuts (Arachis hypogaea L.) are also common in theseareas but grown on small pieces of land. Most of the farmers inthese areas own small numbers of cattle and farming is their mainsource of livelihoods. Unconventional small-scale gold mining iswidespread too as a source of income.

2.2. Description of experiments

2.2.1. On-station experimentsThe experiments were established at three on-station sites in

2009e10 season. At all three sites, maize was grown under rain-fedconditions, no-tillage and 2.5 t ha1, maize residues uniformlyspread in each plot. In 2009e10 season, the maize hybrid ZS261 (amediummaturing maize quality protein maize hybrid, which takesapproximately 135 days to reach maturity) was planted at all siteswhich was changed in 2010e11 season to the maize hybrid Pristine601 (a medium maturity hybrid which matures in 135 days aftersowing). The change in variety was necessary because ZS261 wasprone tomaize streak virus. Each site had six treatments as follows:

a. Manual weeding with hand hoe only, whenever weeds were10 cm tall or 10 cm in length for grasses with stoloniferouserhizomatous growth habit (when weeds are still young, acommon practice by smallholder farmers).

b. Paraquat at 1.0 l ha1 (0.25 l ha1 a.i (active ingredient)) atseeding plus manual weeding whenever weeds were 10 cm tallor 10 cm length of grasses with stoloniferouserhizomatousgrowth habit.

c. Glyphosate at 2.5 l ha1 (1.025 l ha1 a.i) at seeding plusmanual weeding whenever weeds were 10 cm tall or 10 cmlength of grasses with stoloniferouserhizomatous growthhabit.

d. Atrazine at 3.6 l ha1 (1.80 kg ha1 a.i) at seeding plus manualweeding whenever weeds were 10 cm tall or 10 cm length ofgrasses with stoloniferouserhizomatous growth habit.

e. Glyphosate atrazine 2.5 l ha1 (1.025 l ha1 a.i) plus 3.6 l ha1(1.80 kg ha1 a.i) respectively at seeding plus manual weedingwhenever weeds were 10 cm tall or 10 cm length of grasseswith stoloniferouserhizomatous growth habit.

f. Glyphosate atrazinemetalachlor 2.5 l ha1 (1.025 l ha1 a.i)plus 3.6 l ha1 (1.80 kg ha1 a.i) plus 1.0 l ha1 (0.96 l ha1)respectively at seeding plus manual weeding whenever weedswere 10 cm tall or 10 cm length of grasses with stoloniferouserhizomatous growth habit.

The experiment was laid out as a randomised complete blockdesign (RCBD) with six treatments replicated three times at eachexperimental site. Maize received a basal application of 150 kg ha1

(11 N, 21 P2O5 and 11 K2O), and 150 kg ha1 ammonium nitrate(52 N) applied as a top dressing, split applied at four and sevenweeks after emergence.

2.2.2. On-farm experimentsTen demonstration plots that were established in the 2004/05

season at the three on-farm sites were used in this experiment. The

T. Muoni et al. / Crop Protection 53 (2013) 23e28 25demonstration plots measured 0.3 ha and each demonstration plotwas subdivided into three tillage treatments, sub-plots of 0.1 ha(10 m 100 m) each. The tillage treatments were: a) conventionalmouldboard ploughing, b) ripper, where a ripper tine was attachedto an ox-drawn plough to make rip lines; maize seed and fertiliserwas dropped into the rip lines by hand, and c) direct seeding, an ox-drawn direct seeder was used to plant maize. Maize varieties wereZS261 in 2009e10 season and Pristine 601 in 2010e11 seasonexcept in Madziva where ZM525 was planted in 2009e10 andZM521 in 2010e11. Basal fertiliser was applied at a rate of165 kg ha1 (12 N, 23 P2O5 and 12 K2O) at planting. Ammoniumnitrate was applied as a top dressing at a rate of 150 kg ha1 (52 N),equally split applied at four and seven weeks after emergence.Glyphosate was applied at all sites at a rate of 2.5 l ha1 for initialweed control during the research.

2.3. Field measurements

At the on-station sites weed measurements were taken from a0.5 m 0.5 m quadrant; the quadrant was randomly placed fourtimes in each plot. Inside the quadrant, weed density was recorded.The time required to weed each plot was also recorded. Grain yieldwas estimated from 4 rows 5 m net plots after physiologicalmaturity. The maize cobs were dried, shelled and converted to kgper hectare at 12.5% moisture content.

Weeding was done with eight farmers in Madziva, sevenfarmers in Hereford and five farmers in Shamva and the time takenwas recorded. A relationship between weed density and weedingtime was established by plotting weed density m2 against weed-ing time (days ha1). This relationship was used to determine thelabour requirements using weed counts at on-station sites for amore realistic economic analysis. All variable costs for each weedcontrol strategy were used in the gross margin analysis.

2.4. Statistical analysis

Weed density, weeding time and grain yield data was subjectedto a test of normality and an analysis of variance using STATISTIX9.0 for personal computers (Statistix, 2008). Mean separation wasdone using the least significant difference (LSD) test at P 0.05 onall significant data. Gross margin analysis was done to evaluate thefinancial benefits of weed control strategies tested in this experi-ment by finding the difference between gross revenue and totalvariable costs of each treatment.

3. Results

3.1. Effects of weed control strategies on weed density for all sites

At CIMMYT Harare, weeding was done three times in bothseasons. The results showed that weed control strategies had aTable 1Effect of weed control strategies on weed density (m2) at first, second and third weedin

Treatment 2009e10

First (18) Second (4

Manual weeding (control) 90a 177aParaquat manual weeding 67ab 163aGlyphosate manual weeding 33bc 123abAtrazine manual weeding 21bc 94bcAtrazine glyphosate manual weeding 10.1c 80bcAtrazine glyphosate Dual manual weeding 4c 57cP-value 0.014 0.05

Means followed by a different letter in columns are significantly different from each otheweeding was done.significant difference on weed density, at first, second and thirdweeding (P 0.014, P 0.05 and P 0.001 respectively) in 2009e10 season (Table 1). In 2010e2011 season weed control strategiesalso had significant differences on the weed density at first andthird weeding (P < 0.005 and 0.001 respectively). Manual weedingonly had larger weed densities than treatments with herbicides andthe least weed density was in the combination of the three herbi-cides atrazine, glyphosate and metalachlor. The results obtained in2009e10 season at CIMMYT Harare showed an increase in weeddensity of up to 273% (i.e. from 33 to 123 weeds m2) in theglyphosate treatment after first weeding but decreased at the thirdweeding to 33 weeds m2 (Table 1).

At DTC weeding was done four times in both seasons (Table 2)and had significant differences onweed density at first, second andfourth weeding (P 0.004, 0.003 and 0.003 respectively) in 2009e10 season. At third weeding, treatments had no significant differ-ence on weed density. In 2010e11 season weed control strategiesalso had significant differences on weed density (P 0.003, 0.002,0.013 and 0.005 respectively) at all four weedings. The smallestweed density (12weedsm2) was observed in the treatment with acombination of glyphosate and atrazine at fourth weeding in2009e10 season (Table 2). A large weed density was recorded inmanual weeding and it was significantly different from all othertreatments at all the four weeding. Weed densities increased atsecond weeding in 2010e11 season and later decreased by 44% inmanual weeding at third weeding and were even smaller at lastweeding (Table 2).

At HRS the weed control strategies had a significant differenceonweed density at first and second weeding only (P 0.02, 0.05) in2009e10 season (Table 3). There was an increase inweed density atthirdweeding and as well as a decrease at fourthweeding. In 2010e11 season therewas a sharp increase inweed density on themanualweeding treatment by 295% (from 69 to 273 weeds m2) at fourthweeding and low weed densities were recorded in glyphosate plusmanual weeding and in the combination of three herbicides.Treatments had significant differences on weed densities in thisseason at all weeding (P 0.01, 0.03, 0.02 and 0.02, respectively).Large weed densities were recorded in the manual weedingtreatment.

3.2. Effects of weed control strategy on weeding time

Increased weed densities naturally increased weeding time. Therelationship between weed density and time taken was welldescribed by the linear model: y 0.015x 1.4 which was highlysignificant (P < 0.05) and correlation coefficient of 0.87 (Fig. 1). Inthe model, y is the time taken to weed (ha1) and x is the weeddensity (m2).

The results from the on-station sites showed that more time isrequired with manual weed control and using herbicides reducesthe number of weeding days per hectare. At CIMMYT Harare,g over two seasons at CIMMYT Harare.

2010e11

5) Third (73) First (18) Second (55) Third (88)

48a 81a 71 113a41ab 53ab 56 74b33bc 32bc 44 62bc28c 23bc 37 60bc18d 32c 15 32cd17d 18c 4 10d0.001 0.005 NS 0.001

r, separated by LSD-test numbers in parenthesis are the days after planting at which

Table 2Effect of weed control strategies on weed density (m2) at first, second, third and fourth weeding over two seasons at DTC.

Treatment 2009e10 2010e11

First (13) Second (31) Third (48) Fourth (93) First (16) Second (40) Third (82) Fourth (104)

Manual weeding (control) 347a 159ab 54 51a 186a 363a 202a 185aParaquat manual weeding 282ab 138b 45 42ab 179ab 229b 176ab 167abGlyphosate manual weeding 269ab 152ab 58 32b 160bc 161bc 140ab 139bcAtrazine manual weeding 153c 176a 35 30bc 154c 158bc 158ab 142bcAtrazine glyphosate manual weeding 177bc 96c 42 12d 149cd 117c 112bc 109cAtrazine glyphosate Dual manual weeding 84c 56d 58 15cd 128d 65c 69c 65dP-value 0.004 0.003 NS 0.003 0.003 0.002 0.013 0.005

Means followed by a different letter in columns are significantly different from each other, separated by LSD-test numbers in parenthesis are the days after planting at whichweeding was done.

T. Muoni et al. / Crop Protection 53 (2013) 23e28262009e10 season, weed control strategies had a significant differ-ence on the weeding time where the least time (3.5 days1) wasrecorded in treatment with atrazine, glyphosate and metalachlor(Fig. 2). In 2010e11 season, treatments had a significant difference(P 0.04). At DTC treatments had significant differences onweeding time in both seasons 2009e10 (P 0.0001) and 2010e11(and P 0.0002) and generally more time was spent whencompared to CIMMYT Harare. At HRS, treatments had a significantdifference on weeding time in 2009e10 (P 0.01), and 2010e11season (P 0.01). Sole atrazine had more days in 2010e11,12 days ha1, which is higher than the recorded at other sites.

3.3. Effects of weed control strategies on grain yields

The results showed that the weed control strategies had nosignificant effect on maize grain yield. In 2009e10 season, highestyields at CIMMYT Harare were recorded in atrazine plus glyphosateplusmanual weeding treatment (4261 kg ha1) and the least was ina combination of three herbicides (3398 kg ha1). However, moreyields were recorded in 2010e11 season where the least yield was4146 kg ha1 at CIMMYT Harare. Largest yields were observed inatrazine plus manual weeding (8254 kg ha1) at DTC in 2010e11season and atrazine plus manual weeding treatment had thesmallest yields (5352 kg ha1) in 2009e10 season. Smaller yieldswere measured in HRS for both seasons. The lowest yields(347 kg ha1) were recorded in 2009e10 season in paraquat plusmanual weeding treatment. Overall, yields ranged between347 kg ha1 and 8254 kg ha1.

3.4. Effects of weed control strategies on gross margin

The results from CIMMYT Harare and DTC in both seasonsshowed variations on the gross margin per hectare and the largestgross margin (US$ 1704) was recorded in the atrazine plus manualweeding treatment at DTC (Table 3). There were no significantdifferences on the gross margin obtained per hectare at all sites. AtCIMMYT Harare manual weeding and the combination of threeTable 3Effect of weed control strategies on gross margin of continuous sole maize production o

Treatment Gross margin, US$ ha1

CIMMYT Harare

2009e10 2010e

Manual weeding (control) 523 676Paraquat manual weeding 596 776Glyphosate manual weeding 581 764Atrazine manual weeding 604 684Atrazine glyphosate manual weeding 614 693Atrazine glyphosate Dual manual weeding 427 714herbicides resulted in lowest gross margin per hectare (US$ 523and US$ 427), respectively in the 2009e10 season. Despite highergross margin at CIMMYT Harare and DTC, a negative gross marginof US$355 ha1 was recorded in glyphosate treatment at HRS(Table 3).4. Discussion

4.1. Effects of weed control strategies on weed density

Weed control strategies had a strong influence on weed densityduring the cropping season. Increases in weed densities at secondand third weeding at all sites was due to shallow cultivation usinghand hoes which encouraged new weed flushes by moving weedseeds closer to the soil surface where conditions are favourable forgermination (Kolb, 2011). The increase in weed density could alsobe attributed to breakage of weed seed dormancy as the seasonprogresses (Mishra and Singh, 2012). At some sites, decreases inweed densities were attributed to efficient weed managementwhere herbicides and manual weeding controlled weeds beforesetting seed. This concurred with Mandumbu et al. (2011) whofound that one single strategy is often ineffective in long-termweed management and also herbicides followed by one or twomanual weeding can efficiently control weeds thus ensuring abettermaize yield at the end of the season (Vogel,1994a). Practisingminimum soil disturbance also reduces the weed populations inthe long run due to creation of unfavourable conditions and theabsence of bringing old and dormant weed seeds to favourablegerminating conditions at the surface (Baral, 2012).

The presence of previous crop harvest residues suppressesweeds. This is due to crop residues physically impeding weedseedlings from emerging and shedding seeds from accessing lightnecessary for germination (Chhokar et al., 2007). In addition, con-servation agriculture practices promote microbial activity andaccumulation of seeds on the surface where they are exposed topredation by micro and macro-organisms thereby reducing theweed seed bank (Mwale, 2009). Lower weed densities at the clayver two seasons at CIMMYT Harare, DTC and HRS.

DTC HRS

11 2009e10 2010e11 2009e10 2010e11

1099 1163 215 2941026 1211 351 282968 1210 355 322935 1704 274 296

1016 1143 346 227959 1383 317 186

Fig. 1. The relationship between weed density per square meter and weeding timeneeded in days per hectare; data measured at three on-farm sites (ten demonstrationplots) in Zimbabwe.

T. Muoni et al. / Crop Protection 53 (2013) 23e28 27soil sites at CIMMYT Harare could be attributed to long-term op-timum weed management on this researcher-managed location.

The suppression of weeds through herbicides at the start of cropgrowth gives crops a comparative advantage over weeds. Paraquat,a non-selective herbicide suppresses both broadleaved weeds andperennial grasses and is able to suppress weeds better than manualweeding although its effectiveness is short term as it is deactivatedby clay and organic matter particles that limit its activity on suc-ceeding weeds (Watts, 2011). Thus, paraquat controls the weedsthat are actively growing. However, paraquat is very toxic tohumans because it is rapidly absorbed by inhalationwhichmakes ita health risk herbicide for smallholder farmers who often do nothave protective clothing (Ashford and Reeves, 2003).We

ed

in

g tim

e (d

ay

s h

a-1

)

0

5

10

15

20

25

CIM

MYT H

ARARE

2009-1

0

CIM

MYT H

ARARE

2010-1

1DTC 2009-1

0

DTC 2010-1

1

HRS 2009-1

0

HRS

2010-1

1

Manual weeding Paraquat Glyphosate Atrazine Atrazine + Glyphosate Atrazine + Glyphosate + Dual

Fig. 2. The effect of weed control strategies on weeding time at CIMMYT Harare,Domboshawa Training Centre (DTC) and Henderson Research Station (HRS), 2009/10e2010/11.The results showed a similar effect of paraquat and glyphosateinweed control. This is because glyphosate (non-selective, systemicherbicide) is rapidly deactivated by clay particles thereby limitingits effectiveness on emerging weeds suggesting that glyphosate canbe an alternative to paraquat or vice versa to reduce the chances ofherbicide resistance by weeds as suggested by Watts (2011).Glyphosate cannot be used on soils that have more than 95% sandcontent (Thierfelder, personal communication) because the herbi-cide remains active in such soils and affect germination of thecrops. This suggests that other herbicides such as metalachlor (pre-emergent herbicide), are needed on sandy soil to control grassseedlings and some broadleaved weeds.

Atrazine application in combination with manual weeding wasmore effective than paraquat and glyphosate because atrazine is aresidual herbicide which remains active in the soil for up to ninemonths or more (Croplife, 2006). It is a selective herbicide generallyeffective only against broadleaved weeds. Atrazine use faced criti-cism in the past e.g. in Europe; as it leaches to groundwater it hasraised concerns about human health (Helling et al., 1988). However,after cumulative risk assessment by the United States EnvironmentProtection Agency the herbicide was reregistered in 2006 in theUSA (Williams et al., 2010). Mixing atrazine with metalachlor orglyphosate reduced risk of herbicide failure and increased weedspectrum controlled thereby significantly decreasing weed den-sities but at an increase in cost. It is generally recommended to mixatrazine with other herbicides that are compatible with it toimprove their effectiveness on weeds (Williams et al., 2011). In amixture of glyphosate and atrazine, glyphosate controls all peren-nial grasses and broadleaved weeds while atrazine works best onbroadleaved weeds and is persistent in the soil for some time(Pfeiffer, 2009). Even smaller weed densities were observed whenatrazine was mixed with glyphosate and metalachlor unlike withmanual weeding and sole herbicides. Metalachlor is a chlor-oacetamide (Rivard, 2003), which acts as a growth inhibitor whichsuppresses synthesis of chlorophyll, fatty acids, lipids and proteinsin both grasses and broadleavedweeds. Hence, in combinationwithglyphosate, atrazine and manual weeding the chance of total weedcontrol is feasible. Under CA there is need to practice all yearweeding to prevent weeds from setting seed. Glyphosate applica-tions or manual weeding can be done at the end of the croppingseason to keep weed density at low levels. However, this will addanother cost to cash constraint smallholder farmers which there-fore may not be a realistic option. Nevertheless, improved weedmanagement using herbicides increases the chances of depletingthe weed seed bank.

4.2. Effects of weed control strategies on weeding time, grain yieldand gross margin

Results from all sites confirmed that using herbicides saved timespent on manual weeding. Manual weeding alone using hand hoesrequired more time on weeding. Paraquat application combinedwith manual weeding, and glyphosate with manual weeding had asmall difference in number of days needed for weeding during theseason. Mwales (2009) study shows that 25 days ha1 (i.e.203 h ha1) are required for weeding under conventional tillage.The results of this study also showed that the higher the weeddensity, the longer the time of weeding and vice versa.

Although weed control strategies had significant effects onweed density and weeding time, they showed no significant effecton grain yield. This shows that when manual weeding is doneeffectively, high maize grain yields are achievable and are compa-rable with herbicide treated plots. At HRS, a site with predomi-nantly sandy soils, smaller yields were obtained which was a resultof poor soil fertility at this particular site, therefore, the benefit of

T. Muoni et al. / Crop Protection 53 (2013) 23e2828using herbicides was small. At CIMMYT-Harare and DTC, highergross margins were achieved because the soil fertility is greatercompared to HRS and therefore, larger grain yields were obtained.Hence, with low grain yields the use of any herbicides used in thisstudy is not attractive to farmers who may abandon the use ofherbicides under CA plots.

Using atrazine plus manual weeding and combining it withglyphosate and metalachlor reduces the weeding time therebycompensating for higher input costs of the treatment. The resultsshow that it is economical to use herbicides under CA becausefarmers save at least US$388 worth of time to be used on other off-or on-farm activities. On sandy soils (HRS, low soil fertility) usingherbicides was not beneficial because the yields obtained were toolow to compensate for the costs of inputs and other operations.Hence, it is critical to focus on improving the initial soil fertility andboost grain yields for farmers to realise higher economic return ofherbicides use.

The time savings by using herbicides under CA can only be abenefit if farmers use this additional time meaningfully for othertasks. Farmers who choose to use herbicides are likely to havemoretime to commit to other farm operations such as growing vegeta-bles in their gardens for sale, value addition to their farm productsand some may also sell their labour off-farm to improve their in-come. The use of herbicides under CA systems reduces the labourconstraints during the peak labour demand periods of the season.The complexity of weed management is eliminated with the use ofherbicides if chosen and applied correctly. With improved weedmanagement through use of herbicides, smallholder farmers canincrease their yields and recover the costs of herbicides use.

5. Conclusions

The use of herbicides as a weed control strategy under conser-vation agriculture in Zimbabwe was tested in two consecutivecropping season in 2009e10 and 2010e11. The results show thatapplication of herbicides is effective in controlling weeds thusreducing the challenges of weed control under conservation agri-culture systems in Zimbabwe. Herbicides reduced the labour bur-dens during the cropping season when labour requirements werehighest. Best results were achieved when herbicides were com-bined (e.g. mixing atrazine and glyphosate or atrazine, glyphosateand metalachlor), which increased the effectiveness of the indi-vidual herbicides while controlling complicated weed species.Herbicides are generally accessible in Zimbabwean towns butbudget constraints and limited knowledge and capacity on how touse the different products currently prevent farmers from suc-cessfully applying them. Manual weeding on the other hand did notreduce maize grain yields when weeding was timely and effective.While the costs of inputs were higher on treatments with herbi-cides, the overall labour was reduced such that the additional costof inputs was balanced out in the gross margins. The soil type andfertility status of the site was however crucial in achieving positivegross margins. The use of herbicides in conservation agriculturesystems can be recommended in most farming circumstances; itcontrols weed species that are difficult to manage, reduces theweeding time for farmers and is seen as a viable option even forsmallholder farmers in Zimbabwe.

Acknowledgements

This research was carried out under the umbrella of the MAIZECGIAR Research Program (MAIZE CRP). We wish to acknowledgefinancial support from the International Fund for AgricultureDevelopment (IFAD) as well as logistical support from CIMMYT.Special thanks go to Mr Nyasha Chipunza, for assistance with theeconomic analysis and Mr Sign Phiri, Herbert Chipara, and the fieldstaff of Henderson Research Station, Domboshawa Training Centreand at CIMMYT-Harare for technical support with establishing theexperiments and data collection. Particular thanks go to ourcolleague Blessing Mhlanga for encouragement during theresearch. We appreciate the work of three anonymous reviewersfor their time and constructive criticism of this manuscript.References

Ashford, D.L., Reeves, D.W., 2003. Use of a mechanical roller-crimper as an alter-native kill method for cover crops. Am. J. Altern. Agric. 18 (1), 37e45.

Baral, K.R., 2012. Weeds management in organic farming through conservationagriculture practices. J. Agric. Environ. 13, 60e66.

Caamal-Maldonado, J.A., Jimenez-Osornio, J.J., Torres-Barragan, A., Anaya, A.L., 2001.The use of allelopathic legume cover and mulch species for weed control incropping systems. Agron. J. 93, 27e36.

Chamunorwa, A., 2010. Livelihoods after Land Reform in ZimbabweWorking Paper 8.Chhokar, R.S., Sharma, R.K., Jat, G.R., Pundir, A.K., Gathala, M.K., 2007. Effect of tillage

and herbicides on weeds and productivity of wheat under riceewheat growingsystem. Crop Prot. 26 (11), 1689e1696.

Croplife, 2006. Zimbabwe Croplife Chemical Handbook, vol. 2, pp. 1e83.Derpsch, R., 2008. No-tillage and conservation agriculture: a progress report. In:

Goddard, T., Zoebisch, M., Gan, Y., Ellis, W., Watson, A., Sambatpanit, S. (Eds.),2008. No-till Farming Systems. World Association of Soil and Water Conser-vation, Bangkok, p. 544.

Derpsch, R., Sidiras, N., Roth, C.H., 1986. Results of studies made from 1977 to 1984to control erosion by cover crops and no-tillage techniques in Paran, Brazil.Soil Till. Res. 8, 253e263.

FAO, 2002. Conservation Agriculture: Case Studies in Latin America and Africa. FAOSoils Bulletin, p. 70.

Haggblade, S., Tembo, G., 2003. Conservation Farming in Zambia.Helling, C.S., Zhuang, W., Gish, T.J., Coffman, C.B., Isensee, A.R., Kearney, P.C.,

Hoagland, D.R., Woodward, M.D., 1988. Persistence and leaching of atrazine, al-achlor, and cyanazine under no-tillage practices. Chemosphere 17 (1), 175e187.

Kassam, A., Friedrich, T., Shaxson, F., Pretty, J., 2009. The spread of conservationagriculture: justification, sustainability and uptake. Int. J. Agric. Sust. 7,292e320.

Kolb, L.N., 2011. Alternative Weed Management Strategies for Organic Cereals:Enhanced Crop-weed Interference and Physical Weed Control. In: Departmentof Plant, Soil, and Environmental Sciences, D Phil. thesis. University of Maine.

Mandumbu, R., Jowah, P., Karavina, C., Tibugari, H., 2011. Integrated weed man-agement in Zimbabwes Smallholder Sector, where are we?: a review. Mod.Appl. Sci. 5 (5), 111e117.

Mashingaidze, N., Madakadze, C., Twomlow, S., Nyamangara, J., Hove, L., 2012. Cropyield and weed growth under conservation agriculture in semi-arid Zimbabwe.Soil Till. Res. 124, 102e110.

Mishra, J.S., Singh, V.P., 2012. Tillage andweed control effects on productivity of a dryseeded riceewheat system on a vertisol in Central India. Soil Till. Res. 123,11e20.

Mwale, C.D.W., 2009. Effect of Tillage Practices on Weed Populations and SeedBanks in Maize Based Production Systems in Malawi (Masters thesis). ISARA,University of Lyon.

Norsworthy, J.K., Frederick, J.R., 2005. Integrated weed management strategies formaize (Zea mays) production on the southeastern coastal plains of NorthAmerica. Crop Prot. 24 (2), 119e126.

Nyamapfene, K.W., 1991. The Soils of Zimbabwe. Nehanda Publisher.Pfeiffer, M., 2009. http://www.ptrpest.com/pdf/glyphosate_moa.pdf.Rivard, L., 2003. Environmental Fate of Metolachlor. Environmental Monitoring

Branch.Sorrenson, A., Duarte, C., Lopez Portillo, J., 1998. Economics of No-tillage Compared

to Traditional Cultivation on Small Farms in Paraguay. MAG/GTZ Soil conser-vation project, Asuncion, Paraguay.

Thierfelder, C., Cheesman, S., Rusinamhodzi, L., 2012. A comparative analysis ofconservation agriculture systems: benefits and challenges of rotations andintercropping in Zimbabwe. Field Crops Res. 137, 237e250.

Thierfelder, C., Wall, P.C., 2009. Effects of conservation agriculture techniques oninfiltration and soil water content in Zambia and Zimbabwe. Soil Till. Res. 105(2), 217e227.

Thierfelder, C., Wall, P.C., 2010. Investigating conservation agriculture (CA) systemsin Zambia and Zimbabwe to mitigate future effects of climate change. J. CropImpr. 24, 113e121.

Vogel, H., 1994a. The need for integrated weed management systems in smallholderconservation farming in Zimbabwe. Tropenlandwirt 96, 35e36.

Vogel, H., 1994b. Weeds in single-crop conservation farming in Zimbabwe. Soil Till.Res. 31, 169e185.

Watts, M., 2011. http://wssroc.agron.ntu. edu.tw/note/Paraquat.pdf.Williams, M.M., Boerboom, C.M., Rabaey, T.L., 2010. Significance of atrazine in sweet

corn weed management systems. Weed Technol. 24 (2), 139e142.Williams, M.M., Boydston, R.A., Ed Peachey, R., Robinson, D., 2011. Significance of

atrazine as a tank-mix partner with tembotrione. Weed Technol. 25 (3), 299e302.

http://refhub.elsevier.com/S0261-2194(13)00149-X/sref1http://refhub.elsevier.com/S0261-2194(13)00149-X/sref1http://refhub.elsevier.com/S0261-2194(13)00149-X/sref1http://refhub.elsevier.com/S0261-2194(13)00149-X/sref2http://refhub.elsevier.com/S0261-2194(13)00149-X/sref2http://refhub.elsevier.com/S0261-2194(13)00149-X/sref2http://refhub.elsevier.com/S0261-2194(13)00149-X/sref3http://refhub.elsevier.com/S0261-2194(13)00149-X/sref3http://refhub.elsevier.com/S0261-2194(13)00149-X/sref3http://refhub.elsevier.com/S0261-2194(13)00149-X/sref3http://refhub.elsevier.com/S0261-2194(13)00149-X/sref4http://refhub.elsevier.com/S0261-2194(13)00149-X/sref5http://refhub.elsevier.com/S0261-2194(13)00149-X/sref5http://refhub.elsevier.com/S0261-2194(13)00149-X/sref5http://refhub.elsevier.com/S0261-2194(13)00149-X/sref5http://refhub.elsevier.com/S0261-2194(13)00149-X/sref5http://refhub.elsevier.com/S0261-2194(13)00149-X/sref6http://refhub.elsevier.com/S0261-2194(13)00149-X/sref6http://refhub.elsevier.com/S0261-2194(13)00149-X/sref7http://refhub.elsevier.com/S0261-2194(13)00149-X/sref7http://refhub.elsevier.com/S0261-2194(13)00149-X/sref7http://refhub.elsevier.com/S0261-2194(13)00149-X/sref7http://refhub.elsevier.com/S0261-2194(13)00149-X/sref8http://refhub.elsevier.com/S0261-2194(13)00149-X/sref8http://refhub.elsevier.com/S0261-2194(13)00149-X/sref8http://refhub.elsevier.com/S0261-2194(13)00149-X/sref8http://refhub.elsevier.com/S0261-2194(13)00149-X/sref9http://refhub.elsevier.com/S0261-2194(13)00149-X/sref9http://refhub.elsevier.com/S0261-2194(13)00149-X/sref10http://refhub.elsevier.com/S0261-2194(13)00149-X/sref11http://refhub.elsevier.com/S0261-2194(13)00149-X/sref11http://refhub.elsevier.com/S0261-2194(13)00149-X/sref11http://refhub.elsevier.com/S0261-2194(13)00149-X/sref11http://refhub.elsevier.com/S0261-2194(13)00149-X/sref12http://refhub.elsevier.com/S0261-2194(13)00149-X/sref12http://refhub.elsevier.com/S0261-2194(13)00149-X/sref12http://refhub.elsevier.com/S0261-2194(13)00149-X/sref12http://refhub.elsevier.com/S0261-2194(13)00149-X/sref13http://refhub.elsevier.com/S0261-2194(13)00149-X/sref13http://refhub.elsevier.com/S0261-2194(13)00149-X/sref13http://refhub.elsevier.com/S0261-2194(13)00149-X/sref14http://refhub.elsevier.com/S0261-2194(13)00149-X/sref14http://refhub.elsevier.com/S0261-2194(13)00149-X/sref14http://refhub.elsevier.com/S0261-2194(13)00149-X/sref14http://refhub.elsevier.com/S0261-2194(13)00149-X/sref15http://refhub.elsevier.com/S0261-2194(13)00149-X/sref15http://refhub.elsevier.com/S0261-2194(13)00149-X/sref15http://refhub.elsevier.com/S0261-2194(13)00149-X/sref15http://refhub.elsevier.com/S0261-2194(13)00149-X/sref16http://refhub.elsevier.com/S0261-2194(13)00149-X/sref16http://refhub.elsevier.com/S0261-2194(13)00149-X/sref16http://refhub.elsevier.com/S0261-2194(13)00149-X/sref16http://refhub.elsevier.com/S0261-2194(13)00149-X/sref17http://refhub.elsevier.com/S0261-2194(13)00149-X/sref17http://refhub.elsevier.com/S0261-2194(13)00149-X/sref17http://refhub.elsevier.com/S0261-2194(13)00149-X/sref18http://refhub.elsevier.com/S0261-2194(13)00149-X/sref18http://refhub.elsevier.com/S0261-2194(13)00149-X/sref18http://refhub.elsevier.com/S0261-2194(13)00149-X/sref18http://refhub.elsevier.com/S0261-2194(13)00149-X/sref19http://www.ptrpest.com/pdf/glyphosate_moa.pdfhttp://refhub.elsevier.com/S0261-2194(13)00149-X/sref21http://refhub.elsevier.com/S0261-2194(13)00149-X/sref21http://refhub.elsevier.com/S0261-2194(13)00149-X/sref22http://refhub.elsevier.com/S0261-2194(13)00149-X/sref22http://refhub.elsevier.com/S0261-2194(13)00149-X/sref22http://refhub.elsevier.com/S0261-2194(13)00149-X/sref23http://refhub.elsevier.com/S0261-2194(13)00149-X/sref23http://refhub.elsevier.com/S0261-2194(13)00149-X/sref23http://refhub.elsevier.com/S0261-2194(13)00149-X/sref23http://refhub.elsevier.com/S0261-2194(13)00149-X/sref24http://refhub.elsevier.com/S0261-2194(13)00149-X/sref24http://refhub.elsevier.com/S0261-2194(13)00149-X/sref24http://refhub.elsevier.com/S0261-2194(13)00149-X/sref24http://refhub.elsevier.com/S0261-2194(13)00149-X/sref25http://refhub.elsevier.com/S0261-2194(13)00149-X/sref25http://refhub.elsevier.com/S0261-2194(13)00149-X/sref25http://refhub.elsevier.com/S0261-2194(13)00149-X/sref25http://refhub.elsevier.com/S0261-2194(13)00149-X/sref26http://refhub.elsevier.com/S0261-2194(13)00149-X/sref26http://refhub.elsevier.com/S0261-2194(13)00149-X/sref26http://refhub.elsevier.com/S0261-2194(13)00149-X/sref27http://refhub.elsevier.com/S0261-2194(13)00149-X/sref27http://refhub.elsevier.com/S0261-2194(13)00149-X/sref27http://wssroc.agron.ntu.%20edu.tw/note/Paraquat.pdfhttp://refhub.elsevier.com/S0261-2194(13)00149-X/sref28http://refhub.elsevier.com/S0261-2194(13)00149-X/sref28http://refhub.elsevier.com/S0261-2194(13)00149-X/sref28http://refhub.elsevier.com/S0261-2194(13)00149-X/sref29http://refhub.elsevier.com/S0261-2194(13)00149-X/sref29http://refhub.elsevier.com/S0261-2194(13)00149-X/sref29

Weed control in conservation agriculture systems of Zimbabwe: Identifying economical best strategies1 Introduction2 Materials and methods2.1 Site description2.1.1 Research station sites2.1.2 On-farm sites

2.2 Description of experiments2.2.1 On-station experiments2.2.2 On-farm experiments

2.3 Field measurements2.4 Statistical analysis

3 Results3.1 Effects of weed control strategies on weed density for all sites3.2 Effects of weed control strategy on weeding time3.3 Effects of weed control strategies on grain yields3.4 Effects of weed control strategies on gross margin

4 Discussion4.1 Effects of weed control strategies on weed density4.2 Effects of weed control strategies on weeding time, grain yield and gross margin

5 ConclusionsAcknowledgementsReferences