lable at ScienceDirect

Crop Protection 66 (2014) 1e7

Contents lists avai

Crop Protection

journal homepage: www.elsevier .com/locate/cropro

Effect of herbicide application on weed flora under conservationagriculture in Zimbabwe

Tarirai Muoni a, *, Leonard Rusinamhodzi b, Joyful T. Rugare a, Stanford Mabasa a,Eunice Mangosho c, Walter Mupangwa b, Christian Thierfelder b

a University of Zimbabwe, P.O. Box MP 167, Mount Pleasant, Harare, Zimbabweb CIMMYT, P.O. Box MP 163, Mount Pleasant, Harare, Zimbabwec Ministry of Agriculture, Weed Research Team, Henderson Research Institute, Private Bag 2004, Mazowe, Zimbabwe

a r t i c l e i n f o

Article history:Received 16 December 2013Received in revised form5 August 2014Accepted 12 August 2014Available online

Keywords:Conservation agricultureWeed species diversityWeed densityHerbicide applicationWeed flora

* Corresponding author. Tel.: þ263 774311136.E-mail address: tarirai.muoni@gmail.com (T. Muon

http://dx.doi.org/10.1016/j.cropro.2014.08.0080261-2194/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Increased challenges of weed control in the smallholder farming sector of southern Africa have oftenresulted in small yields. The objective of this study was to evaluate the effects of different weed controlstrategies on weed flora and composition under conservation agriculture (CA) systems in Zimbabwe. Thisstudy was conducted at three on-station trial sites namely Domboshawa Training Centre (DTC), Uni-versity of Zimbabwe farm (UZ farm) and Henderson Research Station (HRS) in a maizeesoybean rotationfor four seasons from 2009e2010 to 2012e2013 seasons. Hand weeding was done whenever weeds were10 cm tall or 10 cm in circumference for weeds with a stoloniferous growth habit. Weed identificationwas done up to the weed species level, and the ShannoneWeiner diversity and evenness index was usedto determine the response of weed flora to herbicides. Results showed that there were more weeds in theearly years which decreased gradually until the final season. Weed species diversity was not affected byherbicide application and the results indicated that weed species diversity was small in CA systems.Annual weed species constituted a greater proportion of species, and species richness decreased with theduration of the study. Richardia scabra L. and Galinsoga parviflora Cav. were the most common dominantweed species at all sites and in all seasons. Moreover, herbicide application had no effect on the evennessof weeds in the plots but site characteristics had a significant effect on the distribution of weed species(weed species evenness). The results presented in this study suggest that herbicide application facilitatesa depletion of weed seed bank/number of weeds over time. Thus, herbicide application in CA has po-tential to reduce weed density, species richness and species diversity in the long termwhich may lead tomore labour savings and larger yields.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Weed management challenges in the smallholder farmingsector have been reported as one of the major causes of low grainyields in southern Africa. Maize grain yield from smallholder farmsaverages less than 1 t ha�1 and this is often not sufficient to supportan average farming family (USAID/Zim-AIED, 2013). Weeds aremore efficient in competing with crops for nutrients, water andspace, and harbour pest and diseases that all have negative effectson yields obtained at the end of the season (Shrestha et al., 2002).Weed management by smallholder farmers has been practisedusing the mouldboard plough, for families with access to draft

i).

power, and through hand hoes by resource poor farmers (Muoniet al., 2013). Most of the smallholders in Zimbabwe use conven-tional tillage practices for field preparations and for weed control(Vogel, 1994). Although manual weeding using hand hoes is acommon practice within smallholder farming, it is labour intensiveand is often delayed leading to reduced crop yields (Mashingaidzeet al., 2012). Conventional tillage practices often increase soilerosion rates leading to reduced soil quality such as poor soilporosity, nutrient loss and low organic matter content (e.g.Thierfelder and Wall, 2012). Poor soil nutrient statuses in combi-nation with poor weed management practices often contribute todecreased yields. To alleviate this challenge, researchers havesuggested a more sustainable method of farming, commonlyreferred to as Conservation Agriculture.

Conservation Agriculture (CA) is defined as a farming systembased on three interlinked principles which are (a) maintenance of

T. Muoni et al. / Crop Protection 66 (2014) 1e72

a permanent soil cover through crop residues, (b) diverse crop ro-tations and (c) minimum soil disturbance (FAO, 2010). Conservationagriculture has potential to make more efficient use of natural re-sources through integrated management of soil, water and bio-logical resources combined with use of external inputs (FAO, 2010).The use of crop residues helps retaining soil moisture which re-duces the negative effects of mid-season dry spells common insouthern Africa (Thierfelder andWall, 2010). Residues can suppressweeds during the growing season if applied in sufficient quantity.Minimum soil disturbance and retention of crop residues reducethe rate of soil loss and increase soil biological activities (e.g. Dubeet al., 2012). However, the complexity of weed control in CA sys-tems increases due to an increase in perennial weed species (Ganet al., 2008). This has resulted in a general recommendation forincreased use of herbicides in the early years of CA adoption (Wall,2007).

Herbicides have been reported to be effective and economicallyfeasible in the smallholder farming sector where CA is beingpractised (Muoni et al., 2013). Herbicides have the ability to reducesubstantially the weeding pressure but there are potential toxicside effects for humans and the environment (Kolpin et al., 1998).Among the recommended herbicides are glyphosate [N-(phos-phonomethyl) glycine], atrazine [2-chloro-4-(ethylamino)-6-(iso-propylamino)-s-triazine] and metolachlor (2-chloro-N-(2-ethyl-6-methylphenyl)-N-2-methoxyl-1-methylethyl) that have differentmodes of action. Glyphosate is a non-selective systemic herbicidecapable to control weeds that have underground rhizomes. Atra-zine and metolachlor are selective herbicides that are appliedbefore emergence of weeds and atrazine can also be applied afteremergence of weeds and are both effective on broadleaved weedsand some grasses (Croplife, 2006). Rugare and Mabasa (2013) re-ported that the use of herbicides in CA reduced the variable cost ofweed control by at least 21.8% and increased the marginal rate ofreturns by 306% compared to hand hoe weeding. Although manyadvantages of using herbicides have been documented, there islittle information available on the longer term response of weedspecies to herbicide weed control strategies in CA systems underZimbabwean conditions. Increasing the intensity of hand hoeweeding reduces the total weed density and the number of weedspecies that are observed in the plots (Mashingaidze et al., 2012).Crop rotations also facilitate weed suppression and there may be adifferent weed species response due to different rotational crops.Several tools can be used to investigate weed species diversity andevenness in a community such as the ShannoneWeiner index (Hindex for species diversity and E index for species evenness) (Griceet al., 2009). The ShannoneWeiner indices combine species rich-ness (i.e. the number of weed species per area) and species equi-tability (i.e. how even is the number of species) (Nolan andCallahan, 2006). The hypothesis of this study was that herbicideapplication in combinationwith no-till, mulching and crop rotationwill decimate the weed species and their density over time. Thusthe objective of this study was to evaluate the effects of herbicidestrategies on weed flora under conservation agriculture (CA) sys-tems in Zimbabwe.

2. Materials and methods

2.1. Site description

The experiments were established at three research locationsnamely Domboshawa Training Centre (DTC), Henderson ResearchStation (HRS) and University of Zimbabwe farm (UZ farm). All thethree sites are located in natural region II of Zimbabwe and rainfallpattern is unimodal averaging 700e1000 mm per growing season.Rainfall starts in November and ends in April, and mid-summer

temperature ranges from 15.5 �C to 25.0 �C. DomboshawaTraining Centre (17�37

0S, 31�10

0E and 1560 m above sea level

(m.a.s.l)) is located on highly variable soils that are classified asmoderately deep Luvisols and Arenosols, and these soils haveapproximately 5% clay content. Henderson Research Station(17�34

0S, 30�54

0E and 1136 m.a.s.l) soils are classified as Arenosols

according to FAO classification originating from granite rocks(Nyamapfene, 1991). The soils at HRS have a high sandy content(>83%) and are generally low in soil organic matter content(Thierfelder andWall, 2012). University of Zimbabwe farm (17�80

0S,

31�500E and 1503 m.a.s.l) is located on clay soils that have high soil

organic matter and are classified as Chromic Luvisols under FAOclassification (Nyamapfene, 1991).

2.2. Experimental design

The experiment commenced in the 2009e2010 cropping seasonat all sites with maize as the test crop. The experiment was laid in arandomised complete block design (RCBD) with six treatments,replicated three times at all sites. The treatments were;

i. Hand hoe weeding only.ii. Paraquat at 0.25 L ha�1 a.i (active ingredient) at seeding plus

hand hoe weeding.iii. Glyphosate at 1.025 L ha�1 a.i at seeding plus hand hoe

weeding.iv. Atrazine at 1.80 kg ha�1 a.i at seeding plus hand hoe

weeding.v. Glyphosate (1.025 L ha�1 a.i) þ atrazine (1.80 kg ha�1 a.i) at

seeding plus hand hoe weeding.vi. Glyphosate (1.025 L ha�1 a.i) þ atrazine (1.80 kg ha�1

a.i) þ metolachlor (0.96 L ha�1 a.i) at seeding plus hand hoeweeding.

The recommended application rates for the different herbicideswere used in this study and treatments with more than one her-bicide where tank-mixed and applied at the same time. Manual hoeweeding was done whenever weeds were 10 cm tall or 10 cm inlength for stoloniferous weeds, in circumference. A maizeesoybeanrotation was deployed through the trial period. In 2009e2010 and2011e2012, a uniform maize crop, using the maize variety Pristine601, was seeded, whereas soybean (variety Safari) was grown in the2010e2011 and 2012e2013 cropping season after the maize phase.In the maize phase, maize was grown using planting basins at UZfarm and rip lines at HRS and DTC, andmaize harvest residues wereused as ground cover at approximately 2.5 t ha�1 in seasons 1, 2 and4. In the third season, soybean crop harvest residues were retainedand used as ground cover at approximately 1.5 t ha�1. During themaize phase weeding was done up to four times at DTC and HRS in2009e2010 season only whilst in 2011e2012 season and at UZ farmweeding was done only three times throughout the growing sea-son. In the soybean phase weeding was done twice only. Maize wasseeded at 0.9 m � 0.25 m plant spacing to achieve a target plantpopulation of 44,444 plants ha�1. 150 kg ha�1 of Compound D(11 kg N: 21 kg P2O5: 11 kg K2O ha�1) was applied as a basaldressing at seeding and 150 kg ha�1 of ammonium nitrate(52 kg N ha�1) was split applied as top dressing at four and sevenweeks after emergence. In the soybean phase, inoculated soybean(inoculated with Bradyrhizobia japonicum) was seeded at0.45 m � 0.05 m which translated to a target plant population of444,444 plants ha�1 and no herbicide was applied as initial weedcontrol measure in soybean. A basal application of 150 kg Com-pound D (11 kg N: 21 kg P2O5: 11 kg K2O ha�1) was applied bydribbling 90 g in every 10 m row. No top-dressing was applied tothe soybean.

Fig. 1. Effects of weed control strategies on weed density in four seasons (2009e10, 2010e11, 2011e12 and 2012e13) at Henderson Research Station, Domboshawa Training Centreand University of Zimbabwe. Bars with different letters mean that the weed control strategies were significantly different (P < 0.05) for each season.

T. Muoni et al. / Crop Protection 66 (2014) 1e7 3

2.3. Field measurements

In all plots, measuring 6.0 m � 6.3 m, weed counts were doneprior to each weeding by randomly placing a 0.5 m � 0.5 mquadrant four times in each plot at all sites. Weed counts were doneby identifying the weeds in each quadrant then their total in thefour quadrants was summed up for weed density. Weed identifi-cation was done at species level using guidelines outlined by Botha(2001) and, Makanganise and Mabasa (1999). Cyperus esculentus L.and Cyperus rotundas L. were classified as Cyperus spp due to dif-ficulties in identifying them when they were still young. Cynodonnlemfuensis L. was recorded by counting the number of shoots thatwere observed in the quadrants.

2.4. Calculations and statistical analysis

The ShannoneWeiner diversity (H) and evenness (E) indiceswere used to assess the species composition in each plot. TheShannoneWeiner diversity index was calculated using thefollowing formula:

H ¼ N ln N �X

ðn ln nÞ=N (1)

where H is the measured species diversity through proportionalabundance of species, N is the total population density (m�2), n isthe population of each weed species (m�2).

The ShannoneWeiner evenness index was calculated using thefollowing formula

E ¼ H=ln N (2)

where E is evenness of weed species and H and N are as explainedearlier.

The ShannoneWeiner diversity index value of zero indicatesthat there is only one species available (Nolan and Callahan, 2006).The larger the value, the higher the diversity of species within thearea. Species evenness ranges from 0 to 1 and values close to 1show that the species are uniformly distributed in the plots.

All weed density, species diversity, richness and evenness datawas subjected to normality and homogeneity of variance test andthe data was normally distributed. The data was then subjected tothe analysis of variance using Genstat 6th edition to assess thetreatment effects on weed density, species richness, species di-versity and species evenness at each site in all seasons. To assesstreatment, crop, sites and season interactions a linear mixed modelin Genstat 6th edition was used. In the model, treatments and siteswere treated as fixed factors and season was treated as a randomeffect. Seasonwas treated as a random factor because the quality ofthe season could not be determined experimentally. Mean sepa-ration was done using the least significant difference (LSD) test atP � 0.05 on all significant data.

T. Muoni et al. / Crop Protection 66 (2014) 1e74

3. Results

3.1. Effect of different weed control strategies on weed density

Weed control strategies and site characteristics had a stronginfluence (P < 0.001) on the weed density. A strongcrop� treatment interaction (P¼ 0.008) was observed in the linearmixed model analysis. At DTC, weed density was high in the firstseason and it decreased over the time (Fig. 1). Weed control stra-tegies had a significant effect (P ¼ 0.0001) on weed density in2009e2010 season. More weeds were observed in 2009e2010season and manual weeding only had the largest weed density inthat season (706 weeds m�2). In the paraquat plus manual weedingtreatment, most weeds were found in treatments with herbicidesbut the weed density in 2012e2013 decreased by at least 77% of theweed density recorded in the 2009e10 season. In the 2010e2011season, weed control strategies had no significant effect on weeddensity under a soybean crop. In 2011e12 season, weed controlstrategies had a significant effect (P¼ 0.0002) onweed density. Thetreatments atrazine plus manual weeding, andatrazine þ glyphosate and metolachlor plus manual weeding bothhad low weed density (60 and 63 weeds m�2 respectively).Comparably more weeds were recorded in paraquat plus manualweeding and the atrazine þ glyphosate treatment (Fig. 1). In2012e2013 season weed control strategies had no significant dif-ferences on weed density but weed density was very low in alltreatments compared to other seasons. Furthermore, manualweeding, paraquat plus manual weeding, and glyphosate plusmanual weeding had now similar weed densities.

At HRS, weed control strategies had a significant effect (P < 0.05)on weed density in 2009e2010 season only and manual weedingonly had the highest weed density (280 weeds m�2) recorded andthe lowest weed density was 69 weeds m�2 inglyphosate þ atrazine plus manual weeding treatment (Fig. 1). Inthe 2010e2011, 2011e2012 and 2012e2013 season, weed controlstrategies had no significant effect (P> 0.05) onweed density at thissite. It was noted that the 2012e2013 season had generally thelowest weed counts compared to other seasons.

At UZ farm, weed control strategies exerted significant differ-ences (P ¼ 0.0003) on weed density in 2009e10 season only andthe largest weed density was recorded in manual weeding only(Fig. 1). Glyphosate plus manual weeding had 189 weeds m�2 andthe smallest weed density was recorded inglyphosate þ atrazine þ metolachlor plus manual weeding(78 weeds m�2) (Fig. 1). The data for the 2010e2011, 2011e2012and 2012e2013 seasons closely followed the data obtained at HRSfor the similar seasons.

3.2. Effects of different treatments, crops and sites on weed speciesdiversity and evenness under CA

The results showed that site, crop grown and their interactionshad a significant effect (P < 0.001, P ¼ 0.015 and P < 0.001respectively) on weed species diversity in CA plots. Weed speciesdiversity was low at all sites and in all seasons and the treatmentshad no significant differences on weed species diversity at all sitesand in all seasons. At DTC the species diversity index was higher inthe maize phase than in soybean phase. The same trend wasobserved at HRS. At UZ farm the maize phase had higher weedspecies diversity index than the soybean phase, as observed at allother sites. Soybean phases at UZ farm had higher weed speciesdiversity indices than those recorded at DTC and HRS in all seasons.The results from the combined linear mixed model indicated thatsite had a significant effect on species evenness (P < 0.001). How-ever, treatments and different crops grown (maize and soybean)

had no effect on the evenness of weeds in the plots. A significantsite � crop interaction was observed. The results indicated thatspecies were not uniformly distributed in the plots at all sites and inall seasons.

3.3. Effects of different treatments and sites on weed speciesrichness at all sites

Weed species richness was greatly affected by site and treat-ments in the combined linear mixed model analysis (P < 0.001 andP ¼ 0.001 respectively). However, within sites and seasons, theanalysis showed no treatment effects on species richness in allyears except in 2011e2012 season at DTC. More weed species wereobserved in the early years of the study and the weed speciesdecreased with the duration of the experiment. In the 2011e2012season, atrazine plus manual weeding, glyphosate þ atrazine plusmanual weeding and glyphosate þ atrazine þ metolachlor plusmanual weeding had similar species richness. In the 2009e2010season, the dominant weed species observed were Amaranthushybridus L., Leucas martinicensis (Jacq.) R. Br, Cyperus spp, Conyzaalbida Spreng., Galinsoga parviflora Cav., Eleusine indica L., Richardiascabra L. and Commelina benghalensis L. In 2010e2011 season spe-cies richness and the number of dominant weed species decreased.In this season only five weed species were observed in largenumbers and these were: A. hybridus L., G. parviflora Cav., E. indicaL., and R. scabra L. In the 2011e2012 season, manual weeding,paraquat plus manual weeding and glyphosate plus manualweeding had similar number of weed species as recorded in2010e2011 season. A further decrease in dominant species was alsoobserved in 2011e2012 season where E. Indica L., G. parviflora Cav.,R. scabra L. and C. benghalensis L. were the commonweed species. Inthe 2012e2013 season the species richness in manual weeding,paraquat plus manual weeding and glyphosate plus manualweeding decreased by more than 50% and atrazine plus manualweeding had the highest number of species. Only two dominantspecies (G. parviflora Cav. and R. scabra L.) were recorded undermanual weeding.

At HRS weed control strategies had no significant differences onspecies richness in all the four seasons. Manual weeding, paraquatplus manual weeding and atrazine plus manual weeding hadalmost the same number of weed species (8 weed species). Theherbicide combinations (glyphosate þ atrazine plus manualweeding and glyphosate þ atrazine þ metolachlor plus manualweeding) had the same number of weed species (6 weed species).Dominant weed species in this season were Cyperus spp, Digiteriasanguinalis L., E. Indica L., Bulbostylis hispidula Vahl and R. scabra L.The species richness decreased in 2010e2011 season and thesmallest number of weed species (4 weed species) was recorded inglyphosate þ atrazine þ metolachlor plus manual weeding.Glyphosate plus manual weeding had higher species richness thanall the other treatments. The dominant weed species in 2010e2011season were D. sanguinalis L. and R. scabra L. In 2011e2012 seasonthere was an increase in species richness and more weed specieswere observed in paraquat plus manual weeding (9 weed species).Also the number of dominant weed species also increased whichincluded Dactylocterium aegyptium L., D. sanguinalis L., B. hispidulaVahl and R. scabra L. In 2012e2013 species richness decreased andonly two species were dominant (D. aegyptium L. and R. scabra L.).Manual weeding only had the lowest species richness and atrazineplus manual weeding had more weed species.

At UZ farm, weed control strategies had no significant effect onspecies richness in all seasons. In 2009e2010 season the number ofdominant weed species was five which included A. hybridus L.,G. parviflora Cav, Foeniculum vulgare Mill, L. martinicensis andR. scabra. The dominant species recorded in this season were lower

T. Muoni et al. / Crop Protection 66 (2014) 1e7 5

than the species recorded in 2009e2010 season, namelyL. martinicensis (Jacq.) R. Br, F. vulgare Mill, G. parviflora Cav andR. scabra L. Although the species richness decreased in 2010e2011season, it increased in the succeeding season (2011e2012) in alltreatments except in paraquat plus manual weeding. Bidens pilosaL. was an additional dominant weed species that was observed.Species richness decreased in 2012e2013 season by more than 50%when compared to all the other seasons in all treatments and onlyL. martinicensis (Jacq.) R. Br, G. parviflora Cav and R. scabra L. weredominant weed species.

R. scabra L. was a commonweed species that was observed in alltreatments and at all sites in all seasons. At DTC, G. parviflora Cavadditionally was common andwas observed in all seasons and in alltreatments. At UZ, L. martinicensis (Jacq.) R. Br and G. parviflora Cavwere the most common weed species besides R. scabra L. and allwere observed throughout the study period.

4. Discussion

4.1. Effect of different weed control strategies on weed density at allsites and in all seasons

The results showed that different weed control strategies(manual weeding and herbicides) reduced the weed density overtime no matter which strategy was used. In manual weeding thedecrease in weed density may be due to the suppressing effect ofcrop residues during the growing season. Crop residues helpreduce weed seeds germination by impeding the light to reach theseeds and also facilitate suppression of the emerged weeds(Chhokar et al., 2007). Also weed seeds at the soil surface degradefaster under CA due to increased soil biological activities. Weedseeds at the soil surface are too exposed to predation hence, theirchances of emerging are reduced thus reducing the weed density(Mwale, 2009). The decline in weed density over time can also beattributed to continuous weeding before weed plants set seed.This may have reduced the weed seed bank, which may haveotherwise germinated under CA systems. The results concurredwith Mashingaidze et al. (2012) who reported that increasinghand hoe weeding intensity significantly reduces weed density inthe fields over time. The maizeesoybean rotation additionallycontributed to the decline inweed density at all sites. Soybean hasa high plant population per hectare and provides more groundcover and shading, which gives it a competitive advantage overweeds enabling twoweeding to be sufficient for the whole season.Hence, rotational systems that include a legume with narrow rowspacing have a greater potential of suppressing weeds in the longrun as reported by Wall (2007) even in the absence of herbicidesuse.

Residual herbicides such as atrazine provide longer weed con-trol that help better reduction of broadleaf and annual grassesweed density within the seasons. Paraquat and glyphosate have noresidual control effect hence their suppressive effect control is onlyeffective if they are continuously applied during the season e.g. byusing specialised equipment such as weed wipes or sprayers withshields. Controlling small weeds contributed to the decrease inweed density at all sites over time as weeds did not set seed, whichmay have substantially depleted the weed seed bank at the soilsurface. This was previously observed byMwale (2009). As no otherseeds from the sub-soil are ploughed up under CA, there is hopethat the use of chemical products like herbicides may only betemporarily necessary. Thus providing continuous year roundweedcontrol under CA is recommended to avoid weeds setting seed(ZCATF, 2012). A decrease in weed density also suggests that yieldlosses due to crop-weed competition is reduced and also less timeis spent on weeding (Muoni et al., 2013).

4.2. Effects of different treatments, crops and sites on weed speciesdiversity and evenness under CA

The results showed that site and crop as well as their in-teractions had significant effects on the weed species diversityindex. This may be due to differences of soil types and thedifferent ground cover between maize and soybean. Under soy-bean, fewer weeds were observed hence, the weed species di-versity decreased. However, under the maize phase with reducedground cover, weed species diversity increased. Herbicide treat-ments had no significant differences on weed species diversitybecause herbicides such as paraquat, glyphosate and metolachlorare non-selective and will control all present weeds. Althoughatrazine is selective against broadleaved weeds and some annualgrasses, most controlled species that were recorded in this studywere annual weed species. The presence of crop residues com-bined with effective hand weeding also suppressed weeds lead-ing to low weed species diversity (Jones et al., 1999). Reducedweed species diversity may also be caused by minimum soildisturbance practised under CA, which reduces the chances ofdeveloping more complicated perennial weed species. This is dueto reduced cultivation that favours weeds with protractedgermination pattern (Chivinge, 1988). The results also indicatedthat sites had a significant effect on the evenness of weed specieswhich may be due to different management practices that werepractised before the establishment of the trials. The weeds weregrowing freely and allowed to set seed before the trial wasconducted. The reduced weed species diversity means there isreduced crop-weed competition for water, space, nutrients andharbouring of pests that may lower the farmers' yields. A furtherdecrease in weed species diversity may also suggest that the seedbank is getting continuously smaller with improved weed man-agement, thus reduced herbicide dosages may be used to controlweeds. The results also indicated that sites had a significant ef-fect on the evenness of weed species which may be due todifferent management practices that were practised before theestablishment of the trials. The weeds were growing freely andallowed to set seed before the trial was conducted. The un-evenness of weeds in the plots enables spot weeding to be donein the fields. This also reduces the labour requirements and theherbicide quantity that may be needed in situations whereweeds are evenly distributed.

4.3. Effects of different treatments and sites on weed speciesrichness at all sites

There was a decrease in species richness at all sites as thenumber of the seasons increased. This is due to the decrease inweed density through improved weed management practices thatwere done in the plots. The decrease can also be attributed todecrease of weed seeds in the soil thus some species were effec-tively controlled. The dominant weed species were annual andperennial species. All species that were recorded at all sites wheresmall seeded and shallow germinators, and some species such asCyperus spp reproduce by both seed and tubers (Botha, 2001).Weeds which produce small seeds increase in CA systems becausethey can germinate even when covered by crop residues(Makanganise and Mabasa, 1999). However, these weeds are alsoeasily controlled when they are young because they have a shallowroot system. Also the accumulation of the weed seeds at the soilsurface increases their chance to germinate in one season and theyare exposed to insect predation, fungal and bacterial attack thusdepletion of the weed seed bank is high (Wagner and Mitschunas,2008). In the soybean phase, species richness decreased becausesome weeds emerged later in the season or even failed to

T. Muoni et al. / Crop Protection 66 (2014) 1e76

germinate when the ground cover was high. Hence, few weedspecies were recorded under the soybean phases. It is also possiblethat soybean could have suppressed weeds via allelopathy (Wanget al., 2010). Paraquat plus manual weeding and glyphosate plusmanual weeding had similar species richness to manual weedingbecause paraquat and glyphosate have no residual control, hencethey controlled weeds at planting only (Muoni et al., 2013). Intreatments with atrazine, the species richness was small because ofthe longer residual control characteristic of the herbicides(Williams et al., 2010). This reduced the germination of most weedspecies (especially annual species). The dominant weed species atall sites and in all seasons was R. scabra L. which is a shallow ger-minator and has small seeds (Botha, 2001). Thus it can emerge evenwhen covered by residues or with little soil cover, contributing tohigh weed density during the cropping season. Also R. scabra L. hasprotracted germination pattern that makes it germinatethroughout the season especially after cultivation (Chivinge, 1988).Its first flush in the season is the highest to be experienced in thegrowing season and it only reproduced by seed only. Its appearancein all seasons could be due to its large number of seed in the soilthat was depleted gradually during the course of the study. Theweed is controlled by herbicides such as glyphosate, atrazine andmetolachlor but hand hoe weeding is essential to control that hasescaped the herbicide spray. Succeeding hand hoe weeding isnecessary to control all emerging weeds before they set seed.Avoiding this weed from setting seed creates better chances ofdepleting it in the soil seed bank. R. scabra L. grow better inexhausted soils hence, with gradually increasing soil fertility underCA there are chances that the weed may not be observed in largenumbers (Mavunganidze et al., 2009).

5. Conclusion

The use of herbicides in combination with manual hoe weed-ing and applying all the CA principles facilitates decrease in weeddensity over time. Continuous weeding under CA promotes weedcontrol before the weed species set seed, thus reducing of thenumber of weed seeds in the soil seed bank. Herbicide usetogether with crop residues reduces weed species diversity underCA systems. Rotating maize with soybean also reduces weedspecies diversity through suppressing weeds during the growingweeds. Weed species were not evenly distributed at all sitesindicating dominance of some weed species under different soiltypes and environments. Improved weed management practicesreduced the species richness at all sites. This suggests that her-bicides, crop residues and hand hoe weeding played a key role inreducing the weed species that were observed at the onset of theresearch. Reduction in weed density, species diversity and even-ness suggests that the use of herbicides may be reduced or dis-continued after some time, which will be a great benefit forsmallholder farmers. However, these findings require further as-sessments at on-farm level under farmer management practices.From this study it can be concluded that weed control involvingherbicides are an important aspect in the control of weeds insmallholder farmer systems of southern Africa. However, thereduction in weed density and species composition in all weedcontrol strategies under CA shows that with good and effectiveweed control, the weed seed bank can be depleted, which willreduce the labour requirements for smallholder farmers and costsfor herbicides over time.

Acknowledgements

Gratitude is due to the International Maize and WheatImprovement Centre (CIMMYT) under the CGIAR Research

Program (CRP) MAIZE for logistically supporting and to the In-ternational Fund for Agriculture Development (IFAD) for fundingthe research. Special thanks go to Mr Sign Phiri and Mr HerbertChipara for technical support to make the research a success.Field staffs at Henderson Research Station and DomboshawaTraining Centre are greatly acknowledged for their limitlesseffort during the course of the research. We would like to extendour gratitude to the anonymous reviewers for their constructivecriticism. Special thanks go to Blessing Mhlanga, Regina Hlaty-wayo, Wadzanai Mvundura, Connie Madembo, Givemore Mako-nya, Siyabusa Mkuhlani and Jephias Mataruse for helping on thestudy. To Deniah Fadzai Nyereyegona and Muoni family, you areall special.

References

Botha, C., 2001. Common Weeds of Crops and Gardens in Southern Africa. ARC-Grain Crops Institute¼LNR-Instituut vir Graangewasse.

Chhokar, R.S., Sharma, R.K., Jat, G.R., Pundir, A.K., Gathala, M.K., 2007. Effect of tillageand herbicides on weeds and productivity of wheat under riceewheat growingsystem. Crop Prot. 26 (11), 1689e1696.

Chivinge, O., 1988. A weed survey of arable lands of the small-scale farming sectorof Zimbabwe. Zambezia 15, 167e179.

Croplife, 2006. Zimbabwe croplife chemical handbook. In: Herbicides, vol. 2(Harare).

Dube, E., Chiduza, C., Muchaonyerwa, P., 2012. Conservation agriculture effects onsoil organic matter on a Haplic Cambisol after four years of maizeeoat andmaizeegrazing vetch rotations in South Africa. Soil Till. Res. 123, 21e28.

FAO, 2010. The Status of Conservation Agriculture in Southern Africa: Challengesand Opportunities for Expansion (REOSA Technical Brief).

Gan, Y., Harker, N.K., McConkey, B., Suleimenov, M., 2008. Moving towards no-tillpractices in Northern Eurasia. In: Goddard, T., Zoebisch, M., Gan, Y., Ellis, W.,Watson, A., Sombatpanit, S. (Eds.), No-till Farming Systems (Paraguay).

Grice, E.A., Kong, H.H., Conlan, S., Deming, C.B., Davis, J., Young, A.C., Bouffard, G.G.,Blakesley, R.W., Murray, P.R., Green, E.D., 2009. Topographical and temporaldiversity of the human skin microbiome. Science 324 (5931), 1190e1192.

Jones, E., Jessop, R., Sindel, B., Hoult, A., 1999. Utilising Crop Residues to ControlWeeds. School of Rural Science and Natural Resources, University of New En-gland, Armidale.

Kolpin, D.W., Thurman, E.M., Linhart, S., 1998. The environmental occurrence ofherbicides: the importance of degradates in ground water. Arch. Environ.Contam. Toxicol. 35 (3), 385e390.

Makanganise, A., Mabasa, S., 1999. Field Guide to the Identification of ImportantWeeds of Arable Lands in Zimbabwe. Department of Research and SpecialistServices Publication, Harare, Zimbabwe.

Mashingaidze, N., Madakadze, I.C., Thomlow, S., 2012. Response of weed flora toconservation agriculture systems and weeding intensity in semi-aridZimbabwe. Afr. J. Agric. Res. 7 (36), 5069e5082.

Mavunganidze, Z., Makunde, G., Mashingaidze, A., Chivinge, O., Riches, C., Ellis-Jones, J., 2009. Effect of period of cultivation on the abundance and distributionof weed seeds in the soil profile. Appl. Ecol. Environ. Res. 7 (2), 141e148.

Muoni, T., Rusinamhodzi, L., Thierfelder, C., 2013. Weed control in conservationagriculture systems of Zimbabwe: identifying economical best strategies. CropProt. 53, 23e28.

Mwale, C., 2009. Effect of Tillage Practices on Weed Populations and Seed Banks inMaize based Production Systems in Malawi (Master thesis). University of Lyon,ISARA-Lyon, pp. 1e45.

Nolan, K.A., Callahan, J.E., 2006. Beachcomber biology: the ShannoneWeiner spe-cies diversity index. In: Proc. Workshop ABLE, vol. 27, pp. 334e338.

Nyamapfene, K.W., 1991. The Soils of Zimbabwe. Nehanda Publisher.Rugare, J.T., Mabasa, S., 2013. Efficacy and economics of manual and chemical weed

control strategies in the first year of conservation agriculture adoption in thehighveld areas of Zimbabwe. Glob. Adv. Res. J. Agric. Sci. 2 (9), 231e241.

Shrestha, A., Knezevic, S., Roy, R., Ball-Coelho, B., Swanton, C., 2002. Effect of tillage,cover crop and crop rotation on the composition of weed flora in a sandy soil.Weed Res. 42 (1), 76e87.

Thierfelder, C., Wall, P., 2010. Investigating conservation agriculture (CA) systems inZambia and Zimbabwe to mitigate future effects of climate change. J. CropImprov. 24 (2), 113e121.

Thierfelder, C., Wall, P., 2012. Effects of conservation agriculture on soil quality andproductivity in contrasting agro-ecological environments of Zimbabwe. Soil UseManag. 28 (2), 209e220.

USAID/Zim-AIED, 2013. Monthly Update Overview. www.zim-aied.org/docs/Zim-AIED_Monthly_May2013.pdf (accessed 18.09.13).

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

Wagner, M., Mitschunas, N., 2008. Fungal effects on seed bank persistence andpotential applications in weed biocontrol: a review. Basic Appl. Ecol. 9 (3),191e203.

T. Muoni et al. / Crop Protection 66 (2014) 1e7 7

Wall, P.C., 2007. Tailoring conservation agriculture to the needs of small farmers indeveloping countries: an analysis of issues. J. Crop Improv. 19 (1/2).

Wang, Y., Wu, F., Zhou, X., 2010. Allelopathic effects of wheat, soybean and oatresidues on cucumber and Fusarium oxysporum f. sp cucumerinum Owen.Allelopathy J. 25 (1), 107e114.

Williams, M.M., Boerboom, C.M., Rabaey, T.L., 2010. Significance of atrazine in sweetcorn weed management systems. Weed Technol. 24 (2), 139e142.

ZCATF, 2012. Farming for the Future: a Guide to Conservation Agriculture inZimbabwe, second ed. Zimbabwe Conservation Agriculture Task Force,Harare.