620 Agronomy Journa l • Volume109, I s sue2 • 2017

R ice is a principal food source for more than half the world’s population, especially in Asia (Kraehmer et al., 2016; Rao et al., 2007). Rice is grown on 161 million

ha in 114 countries, with an annual production of about 678.7 Tg, and about 90% of the world’s rice is grown and produced in Asia (Matloob et al., 2015). In Asia, rice planting is tradition-ally done by transplanting seedlings manually into puddled soil. In recent years, manual transplantation in many Asia countries has been replaced by MTR and DSR as a response to the increasing costs of labor and/or water (Watanabe, 2011; Matloob et al., 2015). In Asia, 20% of rice fi elds were using the DSR system (Rao et al., 2007). In some parts of Asia, MTR is also adopted by many rice growers for being more stable in rice yields, compared with DSR, such as in Japan (Watanabe, 2011), South Korea, and northeastern and eastern China.

Weeds are one of the major biotic constraints in rice production, in particular in DSR. Experiences over the last three decades have revealed that, in the absence of appropriate weed control measures, the losses caused by weeds in DSR are much higher than those observed in transplanted rice (see review by Matloob et al., 2015). Th e worldwide estimated loss in rice yield from weeds is around 10% of the total production (Oerke, 2006), while yield loss in the DSR system due to weed competition in the absence of any control measures may range from10 to 100% (see review by Rao et al., 2007). Weed management in rice for much of Asia (particularly South Asia) relies on hand weeding, in addition to herbicides (Watanabe, 2011; Kraehmer et al., 2016). However, the overuse of herbicides may threaten sustainable agriculture (Shaner, 2014) via herbicide resistance, crop injury, and environmental pollution. Moreover, hand weeding is becoming problematic for the increasingly labor costs, in particular for big farms. Reduction in herbicide application and integrated weed management are emphasized in many Asian countries, including China.

It is important to understand the characteristics of weed seedbanks in diff erent rice planting systems, for the integrated weed management. Weed seedbanks are the main source of weed infestation in crops, and seedbank dynamics regulate the communities of many of the most important weed species (Barberi and Lo Cascio, 2001). Th erefore, understanding seedbank composition is important to develop effi cient weed

ComparisonofWeedSeedbanksinDifferentRicePlantingSystems

GuoqiChen,QinghuLiu,YuhuaZhang,JunLi,andLiyaoDong*

Published in Agron. J. 109:620–628 (2017)doi:10.2134/agronj2016.06.0348Received 15 June 2016Accepted 24 Oct. 2016Available freely online through the author-supported open access option

Copyright © 2017 American Society of Agronomy5585 Guilford Road, Madison, WI 53711 USATh is is an open access article distributed under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

aBstraCtMachine-transplanted rice (Oryza sativa L.) (MTR), water direct-seeded rice (WDSR), and dry direct-seeded rice (DDSR) are three important alternatives to traditional manual transplantation of rice. Weed infestation is a pervasive problem in all rice planting systems. Th e weed seedbanks under diff erent rice planting systems have seldom been compared. Th us, we sampled weed seeds in fi elds employing MTR, WDSR, and DDSR consecutively for at least 5 yr in Wujin County, eastern China. Seeds of 26 companion weed species of rice, comprising 16 families, were observed. Most weed seeds, 82.5% in MTR, 75.3% in WDSR, and 81.7% in DDSR, were distributed in soil 0- to 10-cm deep. As soil depth increased, the seedbanks of total weeds, broadleaf weeds, grasses, and sedges all signifi cantly decreased under the diff erent rice planting systems, except for sedges under WDSR. Th e DDSR tended to maintain larger seedbanks of sedges and grasses, as well as some upland weeds, such as Digitaria sanguinalis (L.) Scop. and Eleusine indica Gaertn. Th e WDSR system contained the smallest weed seedbank overall but tended to have larger seedbanks of several weeds, such as Ammannia arenaria H. B. K. and Lindernia procumbens (Krock.) Philcox. Weedy rice and Cyperus diff ormis L. tended to maintain larger seedbanks in DSR fi elds. Th e MTR fi elds tended to have larger seedbanks of broadleaf weeds and some traditional rice weeds, with signifi cantly lower richness of weed species in the seedbank. Th erefore, to downsize the weed seedbank, WDSR or MTR should be employed rather than DDSR when possible. Moreover, the infl uences rotation of diff erent rice planting systems on the weed seedbank merit more studies.

College of Plant Protection, Nanjing Agricultural University, Nanjing 210095, China. Th e fi rst two authors contributed equally to this work. *Corresponding author (dly@njau.edu.cn).

Abbreviations: CCA, canonical correspondence analyses; DDSR, dry direct-seeded rice; DSR, direct-seeded rice; MTR, machine-transplanted rice; WDSR, water direct-seeded rice.

Core ideas• Weed seedbanks were compared in three rice planting systems:

machine-transplanted rice, water direct-seeded rice, and dry direct-seeded rice.

• Weed seedbanks were mainly distributed in soil within a depth of 10 cm.

• Dry direct-seeded rice tended to maintain larger seedbanks of sedges, grasses, and some upland weeds.

• Water direct-seeded rice contained the smallest weed seedbank overall.

• Machine-transplanted rice had larger seedbanks of broadleaf weeds and some traditional rice weeds.

Pest interaCtions in agronoMiC systeMs

Published March 9, 2017

Agronomy Journa l • Volume109, Issue2 • 2017 621

management strategies (Feng et al., 2015; Li et al., 2012). Although weed community succession in rice fields under different planting systems has been studied in many countries (Kraehmer et al., 2016; Rao et al., 2007; Watanabe, 2011), weed seedbanks under different rice planting systems have seldom been compared.

Rice cultivation in Jiangsu Province, China, is a useful model system for understanding weed seedbank composition as a response to different rice planting systems. Jiangsu is one of the main rice planting regions in China with an annual planting area of 2.27 million ha (National Bureau of Statistics of China, 2015), and contains large areas of MTR and DSR, including WDSR and DDSR. In 1986 DSR planting area was 7000 ha in Jiangsu Province, and increased to 387,000 ha in 2012 (Sun et al., 2014). Among the DSR fields, about 62% were DDSR and 38% were WDSR (unpublished data, 2012). The popularization of MTR in Jiangsu Province started in 1999. In 2001 the MTR planting area was <10,000 ha, and by 2012 there were 1.10 million ha using the MTR system (Sun et al., 2014). The principal weed species in Jiangsu fields include many of the worst lowland weeds, such as Echinochloa Beauv. spp., weedy rice, Leptochloa chinensis (L.) Nees, Digitaria sanguinalis, Fimbristylis miliacea (L.) Vahl, Cyperus difformis, C. iria L., Monochoria vaginalis (Burm. F.) Presl ex Kunth, A baccifera L., Sagittaria pygmaea Meq., Ludwigia prostrata Roxb., and Alternanthera philoxeroides (Mart.) Griseb. (Holm et al., 1979; Kraehmer et al., 2016; Li, 1998). Therefore, a comparison of weed seedbanks among MTR, WDSR, and DDSR methods in Jiangsu may improve integrated weed management for rice cultivation. We investigated the weed seedbanks of 18 rice fields in one county of Jiangsu Province. All of the fields were planted for at least five consecutive years using the same planting method (MTR, WDSR, or DDSR). The purpose of the study was to determine the characteristics of the weed seedbank under MTR, WDSR, and DDSR.

Material and Methodsstudy area

To better compare the composition of weed seedbank under DDSR, WDSR, and MTR, we selected study area with the following criteria: (i) having MTR, WDSR, and DDSR fields in the same county; and (ii) having villages planting rice for at least five consecutive years using the same planting method (MTR, WDSR, or DDSR). Wujin is the only county in Jiangsu province that matches these criteria. Moreover, in Jiangsu province, there are rice growers renting >50 ha fields and form large farms. These growers tend to have uniformed planting and management practices on the most of their fields, and seldom use hand weeding. Also, fields with uniformed planting and management practices could be good examples for studying the seedbank characteristics. Therefore, we contacted to farmers with fields (>50 ha) for this study.

Field sampling was conducted in rice fields in Wujin County (31°20¢ to 31°54¢ N, 119°40¢ to 120°12¢ E), Jiangsu Province, China, in 2014. Wujin county, located in the plain of the Yangtze River delta in East China, has a subtropical climate with an average annual temperature of about 20.2°C, average rainfall of about 1090 mm, and an average frost-free period of 220 to 230 d (data provided by local Plant Protection Station, a government organization that is responsible for assisting farmers and introducing pesticides to farmers). Paddy fields in Wujin County are commonly used to grow rice from June to October, during the hot and moist season (Fig. 1), and to grow wheat from November to May.

field site description

In 2013, with the assistance of the Wujin Plant Protection Station, we contacted four growers from four villages in Wujin County to request permission to sample in fields during 2014 (Fig. 2). For each village sampled, the majority of fields were rented by the grower we contacted, and most fields used the same planting system with irrigation water from nearby streams and pools. For each MTR, WDSR, and DDSR system, six rice (O. sativa subsp. japonica) fields (each with an area of 1000–1400 m2), located in

Fig.1.Dailymaximumtemperatures,minimumtemperatures,andprecipitationduringriceplantingseasoninWujinCounty,JiangsuProvince,China,2014.

622 Agronomy Journa l • Volume109, Issue2 • 2017

four villages, were sampled to study weed seedbanks 5 to 10 d before harvest. Thus, a total of 18 fields were sampled (Table 1), each of which employed the same planting system (MTR, WDSR, or DDSR) consecutively for at least 5 yr. All the fields sampled were typical paddy soil. Soil texture, soil fertility status, organic matter content, and soil pH were not determined for each field. Typical paddy soil in Wujin County is with pH 5.89 to 6.25, soil organic matter content about 26.0 g kg–1 at the end of rice growing season, and ratio of sand: silt: clay = 18: 45: 37 (data provided by local Plant Protection Station). Herbicides applied during the rice growing seasons for these 18 fields included an application of pre-emergence herbicide, an application of post-emergence herbicides to kill weeds at the three- to five-leaf stage, and another application of post-emergence herbicides to control broadleaf weeds. For both WDSR and DDSR fields, pre-emergence herbicides were applied 3 to 4 d after sowing; first application of post-emergence herbicides were applied about 20 d after sowing; and second application of post-emergence herbicides were applied about 40 d after sowing. For MTR fields, pre-emergence herbicides were applied 5 to 6 d after transplanting; first application of post-emergence herbicides were applied about 20 d after transplanting; and second application of post-emergence herbicides were applied about 40 d after transplanting. Before each herbicide application, water was applied to the solute and/or to distribute herbicides (300 L water for broadcasting 1 ha fields).

Rice was planted 5 to 10 June for DSR and 15 to 25 June for transplanting with MTR to the study area during different years. Rice varieties planted in these fields during the previous 5 yr included Wuyunjing 23 and Nanjing 46. The growth duration of Wuyunjing 23 and Nanjing 46 are about 158 and 165 d, respectively. Both this two varieties were commonly planted in Wujin County. In 2013 and 2014, all the 18 fields sampled were planted with Wuyunjing 23. The growers we contacted bought rice seeds from seed companies that guaranteed seed quality and

purity. Differences in flooding strategies among DDSR, WDSR, and MTR fields occurred mainly prior to the four-leaf stage of rice. For DDSR, fields were moist (but not flooded) before seeding to facilitate the germination of rice and were flooded at the two- to three-leaf stage to a depth of 4 to 5 cm. For WDSR, pre-germinated seeds were sown in fields with water at a depth of about 1 cm, and fields were flooded at the two- three-leaf stage to a depth of 4 to 5 cm. For MTR, rice seedlings at the three- to four-leaf stage (seedling age: 18–20 d) were transplanted in fields containing a water depth of about 2 cm. The MTR fields were tilled to a depth of about 15 cm before rice planting, and DSR fields (including WDSR and DDSR) were tilled to a depth of about 8 cm. Rice fields were flooded for both DSR and MTR for rice from two- to three-leaf stage to the end of tillering stage (about 20 d). Farmers irrigated water onto fields to a depth of 4 to 5 cm, then stopped irrigating; and re-irrigated water until the soil in fields was in saturated conditions. Wheat cultivating systems in the fields sampled were similar. After rice harvested, fields were tilled to a depth of about 8 cm, then wheat seeds (variety Yangmai 16) were sown to field. Herbicides applied during the wheat growing seasons for these 18 fields included a winter application and a spring application. Isoproturon [3-(4-Isopropylphenyl)-1,1-dimethylurea] (281 g a.i. ha–1) + chlortoluron [3-(3-Chloro-4-methylphenyl)-1,1-dimethylurea] (281 g a.i. ha–1) were applied as winter chemical control, during wheat seedlings at the one- to two-leaf stage; and thifensulfuron-methyl {methyl 3-[(4-methoxy-6-methyl-1,3,5-triazin-2-yl) carbamoylsulfamoyl]thiophene-2-carboxylate} (2.0 g a.i. ha–1) + fenoxaprop-P-ethyl {ethyl (2R)-2-[4-[(6-chloro-1,3-benzoxazol-2-yl)oxy]phenoxy] propanoate} (44.6 g a.i. ha–1) + tribenuron-methyl {methyl 2-[[(4-methoxy-6-methyl-1,3,5-triazin-2-yl)- methylcarbamoyl]sulfamoyl]benzoate} (7.9 g a.i. ha–1) were applied as spring chemical control just before wheat jointing stage (1–10 March).

Fig.2.LocationofthefourvillageswithricefieldssampledinWujinCounty,JiangsuProvince,China.DDSR:drydirect-seededrice,WDSR:waterdirect-seededrice,andMTR:machine-transplantedrice.

Agronomy Journa l • Volume109, Issue2 • 2017 623

Table1.Ricesystems(consecutiveyearsforthecommittedsystem),locationsofthevillagewherefieldssampletakenandherbicides(dose,ga.i.ha–1)usedfortherecenttworicegrowingseasonsofthe18fieldsstudied.

Field Village Ricesystem† Pre-emergenceherbicide Post-emergenceherbicide1 Jiuli DDSR(5yr) BS(30)+PE(120) PX(20)+CY(100)2 Jiuli DDSR(5yr) BS(30)+PE(120) PX(20)+CY(100)3 Jiuli DDSR(5yr) BS(30)+PE(120) PX(20)+CY(100)4 Jiuli DDSR(7yr) BS(30)+PE(120) PX(20)+CY(100)5 Jiuli DDSR(5yr) BS(30)+PE(120) PX(20)+CY(100)6 Jiuli DDSR(5yr) BS(40)+PR(360) PX(20)+CY(100)7 Daqiao MTR(6yr) BS(40)+PR(360) ME(115.5)+BE(584.5)8 Daqiao MTR(5yr) PY(50)+PR(500) ME(115.5)+BE(584.5)9 Daqiao MTR(7yr) BS(40)+PR(360) ME(115.5)+BE(584.5)10 Daqiao MTR(5yr) BS(40)+PR(360) PX(20)+CY(100)11 Daqiao MTR(5yr) PY(50)+PR(500) PX(20)+CY(100)12 Daqiao MTR(5yr) BS(40)+PR(360) PX(20)+CY(100)13 Wuyang WDSR(5yr) BS(40)+PR(360) PX(20)+CY(100)14 Wuyang WDSR(6yr) BS(40)+PR(360) PX(20)+CY(100)15 Wuyang WDSR(6yr) BS(40)+PR(360) PX(20)+CY(100)16 Wucheng WDSR(5yr) BS(40)+PR(360) PX(20)+CY(100)17 Wucheng WDSR(5yr) BS(40)+PR(360) PX(20)+CY(100)18 Wucheng WDSR(5yr) BS(40)+PR(360) PX(20)+CY(100)

†DDSR:drydirect-seededrice,WDSR:waterdirect-seededrice,MTR:machine-transplantedrice,BS:bensulfuron-methyl,PE:pendimethalin,PR:pretilachlor,PY:pyrazosulfuron-ethyl,PX:Penoxsulam,CY:cyhalofop-butyl,ME:metamifop,BE:bentazone.Note:asecondtimeofpost-emergencechemicalcontrolwithMCPA(565.9ga.i.ha–1)+carfentrazone-ethyl(34.1ga.i.ha–1)wasappliedforeachfield.Note:thelocationsofthevillagesareshowninFig.1.

Table2.Family,frequency,andabundance(averagenumberofseedsperlitersoil,Mean±SE)ofrice’scompanionweedspeciesamong72soilsamples.

Family Species Frequency Abundance%

Poaceae Leptochloa chinensis (L.)Nees 86.1 66.6±12.2Echinochloa crus-galli (L.)Beauv. 62.5 13.7±2.7Digitaria sanguinalis (L.)Scop. 38.9 21.7±9.4Eleusine indica (L.)Gaertn. 29.2 16.2±5.2Weedyrice 13.9 0.7±0.2

Juncaceae Juncus bufonius L. 5.56 0.21±0.11Cyperaceae Cyperus difformis L. 100.0 63.6±6.7

Cyperus iria L. 58.3 11.1±2.2Fimbristylis dichotoma (L.)Vahl 30.6 14.2±7.4Eleocharis yokoscensis (Franch.etSavat.)TangetWang 23.6 5±1.8Pycreus globosus Retz. 16.7 3.1±1.6Scirpus juncoides Roxb. 13.9 2.8±1.2Fimbristylis miliacea (L.)Vahl 18.1 4.4±2

Lythraceae Ammannia baccifera L. 98.6 94.6±19.5Scrophulariaceae Lindernia procumbens (Krock.)Philcox 91.7 22±2.2Pontederiaceae Monochoria vaginalis (Burm.F.)PreslexKunth 88.9 26.5±4.2Scrophulariaceae Veronica anagallis-aquatica L. 63.9 4.6±0.8Onagraceae Ludwigia prostrata Roxb. 44.4 6.3±2Lythraceae Rotala indica (Willd.)Koehne 40.3 33.1±8.5Lythraceae Ammannia arenaria.H.B.K. 20.8 11.5±3.5Compositae Eclipta prostrata L. 11.1 0.7±0.3Cruciferae Rorippa islandica (Oeder)Borbás 9.72 1.17±0.45Potamogetonaceae Potamogeton distinctus A.Benn. 6.94 2.79±2.58Leguminosae Aeschynomene indica L. 1.39 0.34±0.34Polygonaceae Polygonum hydropiper L. 1.39 0.10±0.10Ranunculaceae Ranunculus sceleratus L. 1.39 0.07±0.07

624 Agronomy Journa l • Volume109, Issue2 • 2017

seedbank samplingSoil samples were taken in October of 2014. In each field,

35 evenly spaced markers were placed. A 20-cm soil plug was collected using a soil sampler (3.5 cm in diameter) at each marker and split into four layers according to depth: 0 to 5, 5 to 10, 10 to 15, and 15 to 20 cm. All the soil from the same layer of the same field was combined into one sample.

The soil samples were air-dried and pulverized. Each soil sample was divided into 10 subsamples, of which three were used to determine the weed seedbank (Feng et al., 2015). Soil from each subsample was placed in a small nylon mesh bag (pore diameter of 0.1 mm) and washed with tap water to remove the clay. The residue, which contained the weed seeds, was air-dried and sieved through increasingly finer meshes (10, 20, 40, 60, 80, 100, and 120 mesh). The residue of each sifting was placed in a Petri dish and examined under a microscope for identification. The species of each seed was identified according to illustrated handbooks (Guan et al., 2000; Yin and Yan, 1997). Viability of seeds was assessed by applying a slight pressure using tweezers (Li et al., 2012;, Vasileiadis et al., 2007).

The average number of seeds of the three subsamples of each main sample was considered the seed abundance for each species in each soil sample. A total of 72 soil samples were collected as follows: three planting systems × six field replications × four soil depths. The exact subspecies of Echinochloa crus-galli seeds washed and sieved from soil samples were unidentifiable and thus the seeds were treated as a single species. Field observation revealed that the E. crus-galli species growing on the sampled fields mainly included E. crus-galli var. crus-galli and E. crus-galli var. mitis.

statistical analyses

Seeds of companion rice weed species (during the rice growing season) were included in the data analysis. We tested the relationship between geographic distance and weed seedbank composition among the four villages sampled. Specifically, pairwise geographic distances, pairwise Bray–Curtis and Jaccard dissimilarities of weed seedbank (with “village–seedbank of different weed species” data matrix) were calculated among the four villages with the “vegan” add-on package (Oksanen et al., 2015) in R 2.12.1 (R Development Core Team, 2015). Then linear regression between spatial distances and dissimilarities

were conducted. To test the effects of different factors on the seedbanks of total weeds, grass weeds, sedge weeds, and broadleaf weeds, general linear mixed effects models were employed, using Mixed Models in the SPSS 16.0 statistical package, with both fixed (i.e., planting system and sampling depth) and random (i.e., herbicide regime nested, within field nested, and within village) effects. In these general linear mixed effects models, sampling depth was treated as covariate. The relationship between seed abundance of different weed groups (total weeds, broadleaf weeds, grasses, and sedges) and soil depth under different rice planting systems were regressed using a logarithmic model. The soil depths were coded in the following way: 1 = 0 to 5 cm, 2 = 5 to 10 cm, 3 = 10 to 15 cm, and 4 = 15 to 20 cm. Canonical correspondence analyses (CCA) were conducted to further test relationships between rice planting systems and the composition of weed seedbanks, using a vegan add-on package (Chen et al., 2013; Oksanen et al., 2015) in the R2.12.1. Seeds of three weed species (Aeschynomene indica L. Polygonum hydropiper L., and Ranunculus sceleratus L.) were only observed in 1 out of 72 soil samples and thus were not included in the CCA.

resultsspecies richness of weed seedbank

Seeds of 26 companion weed species of rice, comprising 16 families, were observed among the 72 total soil samples, with 20 weed species showing a frequency >10% (Table 2). Among these 26 weed species, there were six grass weeds (five from the Poaceae family and one from the Juncaceae family), seven sedges, and 13 broadleaf weeds (including Monochoria vaginalis, a monocot classified as a broadleaf). Among these 26 weed species, six were perennial weed species: Juncus bufonius L., Eleocharis yokoscensis (Franch. et Savat.) Tang et Wang, Pycreus globosus Retz., Scirpus juncoides Roxb., Veronica anagallis-aquatica L. and Potamogeton distinctus A. Bennett. Seeds of A. baccifera, L. chinensis, and C. difformis were the most frequent and abundant species among the soil samples; L. procumbens and M. vaginalis also occurred with high abundance. Seeds of A. indica, P. hydropiper, and R. sceleratus were only observed in 1 out of the 72 soil samples. Species richness of the weed seedbank was significantly lower (p < 0.05) in MTR than in DDSR and WDSR fields (Fig. 3); On average, seedbanks of 15.5, 15.0, and 11.7 weed species were

Fig.3.Numberofrice’scompanionweedspeciesobservedinsoilsampleswithdifferentsoildepthsofdifferentfieldswithdifferentriceplantingsystems.DDSR:drydirect-seededrice,WDSR:waterdirect-seededrice,andMTR:machine-transplantedrice.“F”:Fvalueamongthethreericeplantingsystems.NS:notsignificant,*p<0.05.

Agronomy Journa l • Volume109, Issue2 • 2017 625

observed in each DDSR, WDSR, and MTR field, respectively. Moreover, the weed seedbanks in DDSR showed the highest species richness among the three rice planting systems in each soil layer, in particular at a depth of 0 to 5 cm (Fig. 3).

relationship between geographic distance and weed seedbank Composition

The geographic distance among the four villages ranged from 4.66 to 39.96 km. The Jaccard and Bray–Curtis pairwise dissimilarity values in weed seedbanks of rice fields among the four villages ranged from 0.05 to 0.14, and from 0.38 to 0.66, respectively (Fig. 4). The low Jaccard dissimilarity values suggested that the species composition of weed seedbanks in rice fields of the four villages were very similar. Moreover, the relationships between geographic distance and both Jaccard and Bray–Curtis dissimilarity values were not significant, with

the p values of linear regressions both >0.6 (Fig. 4). Therefore, geographic distance showed no significant influence on the composition of weed seedbank.

Composition of weed seedbanks

According to CCA analyses, the three rice planting systems all influenced the distribution of the weed seedbank. Seeds of Rotala indica (Wild.), Monochoria vaginalis, Ammannia baccifera, E. crus-galli (L.) Beauv., and L. chinensis tended to be more abundant in MTR fields (Fig. 5). Seeds of weedy rice and C. difformis tended to be more abundant in DSR fields. Seeds of A. arenaria and L. procumbens tended to be more abundant in WDSR fields. Seeds of V. anagallis-aquatica showed a slight trend toward abundance in DDSR, and seeds of the other 13 weed species were clearly more abundant in DDSR fields; these species included D. sanguinalis, E. indica, S. juncoides, C. iria, L. prostrata, and F. dichotoma. All seven sedges observed tended to be more abundant in DSR fields. However, seeds of R. indica and M. vaginalis tended to be scarce in DDSR fields (Fig. 5), seeds of E. crus-galli and R. indica tended to be scarce in WDSR fields, and seeds of some weed species that were abundant in DSR fields tended to be scarce in MTR fields, such as sedges, weedy rice, D. sanguinalis, and E. indica.

Vertical distribution of weed seedbank

According to general linear mixed-effects models (Table 3), differences in rice planting systems significantly influenced the seedbanks of overall weeds and broadleaf weeds (p < 0.05) and had weak influence on sedges (p = 0.080), while they showed no significant influence on grasses (p = 0.303). Soil depth had a significant (p < 0.001) influence on seedbanks of four weed groups. With increasing soil depth, all different weed groups

Fig.4.Therelationshipsbetweengeographicdistanceandfloristicdissimilaritiesinthefourvillageswithricefieldssampled.

Fig.5.Canonicalcorrespondenceanalysis(CCA)showingdifferentriceplantingsystems,andthe23companionriceweedspeciesin72soilsamplescollectedfrom18ricefieldsinWujinCounty,JiangsuProvince,China.DDSR:drydirect-seededrice,WDSR:waterdirect-seededrice,MTR:machine-transplantedrice.

626 Agronomy Journa l • Volume109, Issue2 • 2017

showed a significant (p < 0.05) decrease in seed abundance under different rice planting systems, except for sedges under WDSR, and the decreasing slopes of DDSR were all sharper than those of WDSR and MTR (Fig. 6).

Most seeds were distributed in soil at a depth of 0 to 10 cm, including 81.7% in DDSR, 82.5% in MTR, and 75.3% in WDSR. Thus, we further analyzed seedbank composition in the soil within 0 to 10 cm. At a soil depth of 0 to 5 cm, seed abundance of overall weeds under WDSR was significantly the lowest among different planting systems (F value between groups = 5.71, p = 0.014), and seed abundance of sedges under DDSR was significantly the highest (F value between groups = 5.03, p = 0.021), while those of grasses and broadleaf weeds were not significant among different rice planting systems. At a soil depth of 5 to 10 cm, seed abundance of overall weeds under WDSR was significantly the lowest (F value between groups = 7.06, p = 0.007), seed abundance of sedges under DDSR was significantly the highest (F value between groups = 4.22, p = 0.035), and seed abundance of broadleaf weeds was significantly the highest under MTR (F value between groups = 6.59, p = 0.009).

disCussionweed seedbank Characteristics of dry

direct-seeded rice systemsThe challenge of weed management is more serious in

DDSR systems, which contain a larger weed seedbank with higher species richness. The DDSR fields tended to maintain the largest seedbank among the three rice planting systems, with particularly high amounts of seeds of grasses, sedges, and some upland weed species. Many growers have implemented DDSR to deal with the increased costs or decreased availability of labor and water. The risk of grain yield loss due to weeds is greater for DDSR than MTR (Rao et al., 2007). Some weed species infesting upland crops, such as E. indica, D. sanguinalis, Eclipta prostrata, and F. dichotoma, are troublesome in DDSR (Kraehmer et al., 2016; Rao et al., 2007; Watanabe, 2011), and are represented by larger seedbanks. Some principal rice weeds, which are able to produce large numbers of small seeds, become even more problematic in DDSR, such as Cyperus iria (Chauhan and Johnson, 2009), F. miliacea (Chauhan and Johnson, 2009), and Ludwigia prostrata (Li, 1998). Moreover, our results showed that the seedbanks of several traditional rice

Table3.Generallinearmixedeffectsmodelsshowedtheeffectsofdifferenteffectsonseednumbersofdifferentweedgroupsamongthe72soilsamples(takeplantingsystemandsoildepthasfixedeffectsandothersasrandomeffects).

Effect N.df†

Overallweeds Grass Broadleaf Sedge

D.df‡F

value P D.dfF

value P D.dfF

value P D.dfF

value PPlantingsystem 2 15 5.26 0.019 15 1.30 0.303 15 5.57 0.016 15 3.00 0.080Soildepth§ 3 51 50.59 0.000 51 11.11 0.000 51 7.97 0.000 51 19.60 0.000†N.df=Numeratordf.‡D.df=Denominatordf.Randomeffects:herbicideregimenestedwithinfieldnestedwithinvillage.§Soildepth,includingfoursoillayers:0to5cm,5to10cm,10to15cm,and15to20cm.Plantingsystemincludingmachine-transplantedrice(MTR),waterdirect-seededrice(WDSR),anddrydirect-seededrice(DDSR).HerbicideapplicationstrategyandthelocationoffieldsandvillagesarethesameinTable1.

Fig.6.Numberofseedsperm2soilfordifferentweedgroupswithindifferentsoildepths(1=0–5cm,2=5–10cm,3=10–15cm,and4=15–20cm)offieldsunderdrydirect-seededrice(DDSR),waterdirect-seededrice(WDSR),andmachine-transplantedrice(MTR)plantingsystem.NS:theR2foroverallregressionsweresignificant(p<0.05),exceptforsedgesunderWDSR.

Agronomy Journa l • Volume109, Issue2 • 2017 627

weeds tended to be smaller in DDSR fields, such as Monochoria vaginalis and R. indica, both of which are serious rice weeds in Asia (Imaizumi et al., 2008; Rao et al., 2007).

weed seedbank Characteristics of dry direct-seeded rice systems

The WDSR may be employed as a substitute for DDSR to reduce the weed seedbank. The WDSR fields tended to maintain a smaller weed seedbank, especially regarding sedges and grasses, than did DDSR. The initial growth of weeds may be controlled by flooding and pre-emergence herbicides in the WDSR fields. Later, the pre-germinated rice seedlings may compete more effectively with weeds in the WDSR fields than in the DDSR system. Seedbanks of the principal weed species A. arenaria and L. procumbens tended to be large in WDSR and these species were able to produce a large number of seeds (Kraehmer et al., 2016; Li, 1998; Rao et al., 2007).

Cyperus difformis and weedy rice, both of which are serious rice weeds, tended to maintain larger seedbanks in both WDSR and DDSR than in MTR (Kraehmer et al., 2016). Cyperus difformis normally grows in flooded or in very moist soils, and it may be a serious weed in lowland soils used for upland cropping (Chauhan and Johnson, 2009). One C. difformis plant can produce 3000 viable seeds, and the seeds are very small and readily distributed. The ability of the species to complete a vegetative and reproductive cycle within a month during the hot season makes it very competitive with rice, which requires >130 d to reach maturity in DSR (Chauhan and Johnson, 2009). Generally speaking, a rice-growing season using the DSR system is about 15 to 20 d longer than that using MTR. Moreover, shorter and shallower flooding in DSR may also facilitate the infestation of this weed. The infestation of weedy rice in DSR fields is prevalent in many rice-growing areas worldwide, and a shift from DSR to transplanting is frequently recommended to mitigate a weedy rice infestation (Kraehmer et al., 2016).

weed seedbank Characteristics of Machine-transplanted rice system

The MTR fields tended to maintain a smaller sedge seedbank than did DDSR, but they retained larger seedbanks of R. indica, M. vaginalis, A. baccifera, L. chinensis, and E. crus-galli, which are all traditional principal weeds in rice fields in China and other Asian countries (Kraehmer et al., 2016; Li, 1998). In general, the shift from traditional manual transplanting to direct seeding has magnified the weed problem (Kraehmer et al., 2016; Rao et al., 2007).

The MTR system can emulate manual transplanting and thus relieve the problem of weeds. However, some traditional rice weeds are adapted to field conditions under the rice transplanting method and may continue to be a problem. Ordinarily, seedlings at the three- to four-leaf stage are transplanted at 30-cm intervals in the MTR system in East China, and traditional rice weeds may adapt to this niche. For example, E. crus-galli is one of the most devastating rice weeds, as it is well adapted to flooded and non-flooded environments. As it is a C4 grass with a growth habit similar to that of rice, it competes efficiently for water and resources at the expense of rice (Chhokar et al., 2014; Mennan et al., 2012).

Thus, the spacing between rice seedlings in the MTR system may facilitate an infestation of E. crus-galli, which should be immediately controlled (Moon et al., 2014). Leptochloa chinensis is also highly adapted to different rice planting systems. It is a destructive weed species in the rice fields of many Asian countries and has become a troublesome invasive weed in the rice fields of Europe (Benvenuti et al., 2004). During hot seasons, L. chinensis completes its vegetative and reproductive cycle between 30 and 40 d, which makes it very competitive with rice (Dong et al., 2003). Moreover, our results implied that the MTR system tended to hold small seedbanks of weedy rice, sedges, upland weeds (such as D. sanguinalis and E. indica), and perennials. These findings should influence the design of weed management strategies and the rotation schedule of rice planting systems.

ConClusionIn Jiangsu Province, MTR system tends to save 15 to

20 d for wheat cultivation with more stable rice production, compared with DSR. Thus, local governments encourage farmers to implement the MTR system, rather than DSR. Rice growers frequently plant high-quality rice with MTR system in the Jiangsu Province. Nevertheless, to maintain a high production, the MTR system requires more capital for holding or hiring agricultural machinery for land preparation and seedling transplanting. Thus in areas with limited economic level, DSR is more popular. Moreover, many rice growers prefer DSR for lower planting costs, reduced irrigated water consumption, and no additional land for incubating rice seedlings for transplanting. In areas with limited availability of irrigation water, farmers prefer DDSR. Weeds are one of the main constraints of the DSR system. Our results suggest that the DDSR system maintains the largest weed seedbank with particularly high amounts of grasses and sedges. To downsize the weed seedbank in rice fields, WDSR or MTR should be employed in lieu of DDSR. Further, weed seedbanks from fields that employed different rice planting systems showed different compositions. Research on the influences of crop rotations among different planting systems (such as between MTR and WDSR in different growing seasons) on weed seedbanks should be further investigated.

aCknowledgMents

This research was supported by the Special Fund for Agro-scientific Research in the Public Interest of China (201303022) and China Postdoctoral Science Foundation (2015M571763). We thank Longfen Zhu (Wujin Plant Protection Station, Jiangsu Province, China) for providing help with field work. We also thank the reviewers and editors for helpful comments on earlier drafts of the manuscript.

referenCes

Barberi, P., and B. Lo Cascio. 2001. Long-term tillage and crop rotation effects on weed seedbank size and composition. Weed Res. 41:325–340. doi:10.1046/j.1365-3180.2001.00241.x

Benvenuti, S., G. Dinelli, and A. Bonetti. 2004. Germination ecology of Leptochloa chinensis: A new weed in the Italian rice agro-environment. Weed Res. 44:87–96. doi:10.1111/j.1365-3180.2003.00376.x

628 Agronomy Journa l • Volume109, Issue2 • 2017

Chauhan, B.S., and D.E. Johnson. 2009. Ecological studies on Cyperus difformis, Cyperus iria and Fimbristylis miliacea: Three troublesome annual sedge weeds of rice. Ann. Appl. Biol. 155:103–112. doi:10.1111/j.1744-7348.2009.00325.x

Chen, G.Q., Y.H. He, and S. Qiang. 2013. Increasing seriousness of plant invasions in croplands of eastern China in relation to changing farming practices: A case study. PloS ONE 8(9):e74136. doi:10.1371/journal.pone.0074136

Chhokar, R.S., R.K. Sharma, M.K. Gathala, and A.K. Pundir. 2014. Effects of crop establishment techniques on weeds and rice yield. Crop Prot. 64:7–12. doi:10.1016/j.cropro.2014.05.016

Dong, L.Y., J.L. Shen, T.C. Gao, B.J. Zhou and Y. Zhang. 2003. The eco-economic threshold and the critical period for controlling Leptochloa chinensis in direct sowing rice fields. (In Chinese with English abstract.) J. of Nanjing Agric. Univ. 26:41–45.

Feng, L., G.Q. Chen, X.S. Tian, H.M. Yang, M.F. Yue, and C.H. Yang. 2015. The hotter the weather, the greater the infestation of Portulaca oleracea: Opportunistic life-history traits in a serious weed. Weed Res. 55:396–405. doi:10.1111/wre.12151

Guan, G.Q., Y.R. Zhang, G.Y. Sun, S.X. Ding, and Y.B. Wang. 2000. Illustrated handbook of weed seed. (In Chinese.) Science Press, Beijing, China.

Holm, L.G., J.V. Pancho, and J.P. Herberger. 1979. A geographical atlas of world weeds. John Wiley & Sons, New York.

Imaizumi, T., G.X. Wang, T. Ohsako, and T. Tominaga. 2008. Genetic diversity of sulfonylurea-resistant and-susceptible Monochoria vaginalis populations in Japan. Weed Res. 48:187–196. doi:10.1111/j.1365-3180.2008.00622.x

Kraehmer, H., K. Jabran, H. Mennan, and B.S. Chauhan. 2016. Global distribution of rice weeds–A review. Crop Prot. 80:73–86. doi:10.1016/j.cropro.2015.10.027

Li, Y.H. 1998. Weed flora in China. (In Chinese.) China Agric. Press, Beijing.

Li, S.S., S.H. Wei, R.L. Zuo, J.G. Wei, and S. Qiang. 2012. Changes in the weed seed bank over 9 consecutive years of rice-duck farming. Crop Prot. 37:42–50. doi:10.1016/j.cropro.2012.03.001

Matloob, A., A. Khaliq, and B.S. Chauhan. 2015. Chapter five- weeds of direct-seeded rice in Asia: Problems and opportunities. In: L.S. Donald, editor, Advances in agronomy. Academic Press, San Diego, CA. p. 291–336.

Mennan, H., M. Ngouajio, M. Sahin, D. Isik, and E.K. Altop. 2012. Competitiveness of rice (Oryza sativa L.) cultivars against Echinochloa crus-galli (L.) Beauv. In: water-seeded production systems. Crop Prot. 41:1–9. doi:10.1016/j.cropro.2012.04.027

Moon, B.C., J.W. Kim, S.H. Cho, J.E. Park, J.S. Song, and D.S. Kim. 2014. Modelling the effects of herbicide dose and weed density on rice-weed competition. Weed Res. 54:484–491. doi:10.1111/wre.12102

National Bureau of Statistics of China. 2015. Chinese national statistic yearbook 2015. (In Chinese.) China Statistics Press, Beijing.

Oerke, E.C. 2006. Crop losses to pests. J. Agric. Sci. 144:31–43. doi:10.1017/S0021859605005708

Oksanen, J., F.G. Blanchet, R. Kindt, P. Legendre, R.B. O’Hara, G.L. Simpson et al. 2015. Vegan: Community Ecology Package (R package version 2.2-1). https://CRAN.R-project.org/package+vegan (accessed 19 Dec. 2016).

Rao, A.N., D.E. Johnson, B. Sivaprasad, J.K. Ladha, and A.M. Mortimer. 2007. Weed management in direct seeded rice. In: L.S. Donald, editor, Advances in agronomy. Academic Press, San Diego, CA. p. 153–255.

R Development Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Shaner, D.L. 2014. Lessons learned from the history of herbicide resistance. Weed Sci. 62:427–431. doi:10.1614/WS-D-13-00109.1

Sun, T.Q., H.J. Yang, J. Li, and J.P. Deng. 2014. Retrospect and disadvantages analysis of Jiangsu direct-sowing rice and its countermeasures. (In Chinese with English abstract.) China Rice 20:5–9.

Vasileiadis, V.P., R.J. Froud-Williams, and I.G. Eleftherohorinos. 2007. Vertical distribution, size and composition of the weed seedbank under various tillage and herbicide treatments in a sequence of industrial crops. Weed Res. 47:222–230. doi:10.1111/j.1365-3180.2007.00564.x

Watanabe, H. 2011. Development of lowland weed management and weed succession in Japan. Weed Biol. Manage. 11:175–189. doi:10.1111/j.1445-6664.2011.00419.x

Yin, L.P., and Y.S. Yan. 1997. Identification of weed seeds with colored photos. (In Chinese.) China Agric. Sci. and Technology Press, Beijing.