ABSTRACT
Weeds represent one of the major biological constraints to upland rice production in low input agricultural systems. The effects of weeding regimes and rice cultivars on weed growth and rice yield were investigated over three seasons. Four weeding regimes [0 (no weeding control), 1, 2, and 3] and three popular rice varieties (NARIC 2, a local cultivar, and NERICA 4), were tested in 4x3 factorial experiment in a Randomised Complete block with three replicates. The most important weed species recorded were; Biden pilosa, Commelina benghalensis L., Euphorbia hirta L., Micrococca mercurialis Benth., Galisonga parviflora Cav, Sida rhombifolia L., Triumfeta spp, Guizotia scabra, Celocia trigyna, Cyprus rotundus, Panicum Maximum Jacq, and Imperata cylindrica L. Across cultivars, the best weeding regimes for weed control and rice yields were single weeding and weeding twice. Differences among interaction effects between variety and weeding regime were not significant for most traits, except ripening ratio and grain yield in experiment one and experiment two. Across weeding regimes, NERICA 4 out yielded the other varieties in all the three experiments. However, a single well timed hand hoe weeding, together with the use of a cultivar with good adaptation to unfavourable rice growing conditions, such as NERICA 4, would increase land and labour productivity of upland rice-based systems in Uganda.
Key words: Genotype, rice, upland, weeding regimes, yield.
Rice currently ranks as the second most important cereal in Uganda after maize (Anyanga, 2015). Over the last 10 years, production of this crop has more than doubled owing to the expansion of the rice production areas to upland ecology. In addition, the introduction and promotion of upland rice varieties dubbed NERICA (2005 to 2015) has resulted into large scale production of the crop. Indeed, upland rice covers close to 70% of the total area under rice cultivation in the country (Mohamed, 2012). In spite of these developments, the country is still considered a net rice importer, and will perhaps continue unless domestic production increases remarkably to counter the demand by the growing urban population (World Bank, 1993; Hyuha, 2006; Mohamed, 2012). NERICA’s potential yield in sub-Saharan Africa is 5 t/ha; with fertiliser application. However, farmer field conditions in Uganda give a yield of 1.5 t/ha according to several reports (Imanywoha, 2001; Otsuka and Kalirajan 2006). This clearly undermines the status of rice as an important food security and income crop in Uganda. Several reports have highlighted the major causes of this yield gap (Odogola et al., 2006; Hyuha et al., 2007; Kaizzi et al., 2014). Among these, failure to manage weeds and low soil fertility are central to the problem.
Uninhibited weed growth among crops is estimated to cause yield losses in the range of 48 to 100% (Akobundu, 1980; Becker and Johnson, 2001; Johnson et al., 2004; Toure et al., 2013). Specifically, weeds compete severely with rice plants for space, nutrients, air, water and light adversely affecting plant height, leaf architecture, tillering habit, shading ability, growth pattern and crop duration (Miah et al., 1990). In addition, weeds depress the normal yield of grains per panicle and grain weight (Bari et al., 1995). Therefore, a higher rice yield should be a motivation for maintaining a weed free rice crop environment.
Rice is categorised as a weak weed competitive crop. However, several studies have reported the existence of genetic variation in weed competitiveness among rice cultivars (Fofana and Rauber, 200; Haefele et al., 2004; Rodenburg et al., 2009; Saito et al., 2010). Weed competiveness comprises weed tolerance, the ability to maintain high yields despite weed competition, and weed suppressive ability, the ability to suppress weed growth and reduce weed seed production (Zhao et al., 2006a). The number of weed competitive cultivars with high adaptation to African agro-ecosystems reported to date is limited (Wopereis et al., 2008; Rodenburg and Johnson, 2009).
To reduce the effects of weed pressure on crop yields, most upland rice farmers rely largely on pre- and postharvest fires, preparation, and hand or hoe weeding of fields. Labour-intensive hand weeding is often preferred by farmers because the use of herbicides use is costly, and low literacy rates among farmers in Uganda, like the rest of SSA, further limits herbicide use
(Rodenburg and Johnson, 2009). As such, labour availability for weed control becomes limited as weeds multiply with cropping season, resulting into high levels of yield reduction (Saito and Futakuchi, 2014). However, the most commonly accepted approach to manage weeds is to follow an integrated weed management strategy comprising the combined use of two or more available and effective technologies
(Sanyal et al. 2008). Often rice farmers in the uplands of Uganda weed two to three times (by hand or hoe) during the rice growing season, depending on the weed pressure (Alou et al., 2012; Anyanga, 2015). Such weeding interventions should, however, be well timed to optimize weed suppression, grain yield (Dzomeku et al. 2007; Ekeleme et al. 2009), and the time available for the farmer to attend to other non-farm activities (Alou et al., 2012). Therefore, the identification of superior weed competitive rice genotypes would be an attractive, cost effective and safe approach for sustaining rice productivity, particularly for resource-constrained farmers. Weed competitive genotypes would be part of an integrated weed management strategy. The ideal weed competitive genotypes are high yielding under both weedy and weed-free conditions. Therefore, the purpose of this study was to (i) investigate the effect of weeding regimes across popular upland rice varieties and (ii) determine the most weed competitive upland rice variety in the Lake Albert crescent zone of Uganda.
Description of the study site characteristics
Three upland rice experiments were conducted over 3 years (2012 to 2014) at Bulindi Zonal Agricultural Research and Development Institute (BUZARDI) located in the Lake Albertine Crescent Zone of Uganda. The three experiments were conducted in different fields within the research station. This station lies on average longitudes 312954 N and latitudes 12950 E. This institute receives a bimodal rainfall (March to June and August to December) and average rainfall was 1000 mm per year. This experimental site is located in Hoima district which is one of the leading Upland rice producing districts in Uganda (Lodin, 2005). The vegetation in the area is dominantly savannah with short and tall grasses and shrubs with widespread bush burning practiced (Lodin, 2005). Soils in the area are classified as acidic ferralsols (FAO System, 2004) with low exchangeable bases and organic matter. They are dominantly clay (heavy textured), acidic and low particularly in available phosphorus and total nitrogen.
Soil analysis data based on Alou et al. (2012) found that the soils in this experimental station are typically clay loam with sand content of less than 45%. The mean soil pH is below the critical value of 5.2 and below the range of 5.5 to 6.5 considered favorable for plant nutrition. In addition, the mean values of available phosphorus, total nitrogen and exchangeable bases were found to be low.
Experimental design, crop management and data collection
For the study of genotypic response to weeding regimes, a 4 x 3 factorial experiment was established based on randomized complete block design with three replications, and weeding regime and rice variety as factors at four and three levels, respectively. The four levels of weeding regimes were; 1 hoe weeding at 21 days after sowing (DAG), 2 hand hoe weeding at 21 DAG and 42 DAG, and three hand hoe weeding at 21, 42 and 63 DAG, and a no weeding control, while the three levels of rice variety as popular upland rice varieties were; NERICA 4, NARIC2 and Local variety). The plots were of dimensions 5.0 m long × 2.1 m wide. No fertilizers and/or pesticides was applied in this trial. A seeding rate of 50 kg/ha was used to estimate the seed requirement for the trial plots. Seed were dibbled manually in hills 12.5 cm between hills and 30 cm between rows. Sowing was done by the second week of August in each year while harvesting was done by end of November of each year.
Data collection
Data were collected on number of panicles, number of spikelets, spikelet sterility, 1000-grain weight and yield at harvest.
Data analysis
The data collected were subjected to analysis of variance to determine whether the mean squares due to weeding regimes were significant using Genstat version 14 (Payne et al., 2011). An analysis of variance was performed for each trait in each season using ANOVA in Genstat (Payne et al., 2011). Each season (experiment) was analysed separately because experiment × weeding regime × variety interaction was significant. The ANOVA of each experiment was followed by pairwise comparisons using the least significant differences (LSD), when the F-test for the treatment effect was found significant (P ≤ 0.05).
Broad-sense heritability (H) was used to evaluate the stability of performance for rice yield components via estimation of repeatability across replications. The higher the value of H, the greater the genetic stability. H was estimated for rice yield components for each weed treatment for each season, and calculated from variance components, as:
Where,
and
are variety and within – experiment error variances, respectively, and r is the number of replicates. Variance components were estimated using Genstat version 14 (Payne et al., 2011). The estimate of H was biased upward by confounding of variety and variety by experiment variances, but beneficial in approximately comparing the precision for different weed treatments (Zhao et al., 2006b).
Dominant weed species associated with upland rice fields in the Lake Albert crescent zone of Uganda
Several weed species were encountered in this study. These were composed of both broad leaf and grass species. However, their relative abundances depended on the weeding regime applied and previous land use (Nagasawa et al., unpublished). Table 1 shows the major weed species encountered in the fields.
The effect of weeding regimes on selected agronomic traits in rice varieties in the Lake Albert crescent zone of Uganda
In Experiment 1 (second season of 2012 or 2012B), significant differences among weeding regimes were observed for number of panicles per m-2, number of spikelets per panicle and ripening percentage (P<0.05 to P<0.001) (Table 2). Varietal differences were not significantly (P>0.05) different, except number of spikelets per panicle. The differences among interaction effects of weeding regime × variety were not significant (P > 0.05) for most traits (Table 2). The number of panicle per metre-2 (ranged from 149 to 310), number of spikelets per panicle (ranged from 33 to 67), and ripening ratio (ranged from 12.5 to 44.5%) had 254, 52 and 23% as average trait scores for each weeding regime respectively (Table 5). Across varieties, number of panicles per metre-2 (ranged from 34 to 83) with an average trait scores of 61 for variety NERICA 4; ranged from 34 to 62 an average traits score of 62 for NaRIC 2; ranged from 30 to 66 had an average traits score of 48 for Local variety (Figure 1).
In Experiment 2 (second season of 2013 or 2013B), significant differences among weeding regimes were observed for plant height, number of panicles per m-2, number of spikelets per panicles per metre squared, ripening percentage and yield (P<0.05 to P<0.001). Significant differences in variety responses were observed for 1000 grain weight, number of panicles per m-2, ripening percentage and yield (P<0.05 to P<0.001) (Table 3). The differences among interaction of weeding regime × variety interaction was significant for only ripening percentage and yield (P<0.05 to P<0.01) (Table 3). The number of panicle per m-2 (ranged from 112 to 310), number of spikelets per panicle (ranged from 56 to 105), ripening percentage (ranged from 37.5 to 76.9), plant height (ranged from 78.4 to 95.3 cm) and yield (ranged from 714 to 5786 kg/ha) had 239, 87, 66%, 88 cm, 4449 kg/ha as the average trait scores for each weeding regime respectively (Table 6).
Across rice varieties, number of panicles per metre-2 (ranged 275 from to 361) with an average trait score of 275 for NERICA 4; ranged 83 from to 271 and an average traits score of 211 for NaRIC2; ranged from 103 to 284 with an average trait score of 225 for Local. The 1000 grain weight scores ranged from 26.13 to 27 g with a variety average score of 26.7 g for NERICA 4; ranged from 28.87 to 31.1 g with an variety average of 30.1 g for NaRIC 2, ranged from 29.93 to 32.53 g with a variety average of 31.1 g for Local variety. The ripening percentage scores ranged from 40.9 to 82.9 g with a variety average score of 69.2 g for NERICA 4, ranged from 25.4 to 79.4 g with a variety average of 63.2 g for NaRIC 2; ranged from 46.3 to 74.5 g with a variety average of 66.6 g for local variety (Figure 1).
In Experiment 3 (second season of 2014B or 2014B), significant differences among weeding regimes were observed for all traits measured (P<0.05 to P<0.001) (Table 4). Varietal differences were significantly different for 1000 grain weight, number of panicles per metre-2, and yield (P<0.01 to P<0.001). The differences among interaction of weeding regime x variety was not significant for all traits (P>0.05) (Table 4). The number of panicles per m-2 ranged from 111 to 306, number of spikelets per panicle ranged from 56 to 105, 1000 grain weight ranged from 28.31 to 30.01 g, plant height ranged from 66.9 to 86.9 cm, ripening percentage ranged from 37.5 to 76.9% and yield had 237, 86, 29, 81, 66.3%, and 438 kg/ha as average trait scores for each weeding regime respectively (Table 7).
Across varieties, number of panicles per m-2 ranged from 145 to 308 with an average trait scores of 275 for NERICA 4; ranged from 83 to 271 with an average traits score of 211 for NaRIC 2; ranged from 103 to 284 g with an average trait score of 224 for local variety. The 1000 grain weight ranged from 26.3 to 27.0 with an average trait score of 26.7 g for NERICA 4; ranged from 28.87 to 31.07 with an average traits score of 30.1 g for NaRIC 2; ranged from 29.93 to 32.53 g with an average trait score of 31.1 g for local variety. The yield (ranged from 96 to 666) with an average trait scores of 527 kg/ha for NERICA 4; ranged from 29 to 542 kg/ha an average traits score of 3398.5 kg/ha for NaRIC 2; ranged from 90 to 587 kg/ha with an average trait score of 423 kg/ha for local variety as (Figure 1).
Overall, there were increases in the trait responses with each weeding regime and the extent of increases varied by trait and also by season (Tables 5, 6, and 7). In addition, varietal responses also varied with traits and seasons (Figure 1).
Results of broad sense heritability estimation for three upland varieties across weeding treatments are presented in Table 8. Heritability was lower where weeding was done compared to plots where no weeding was conducted for ripening percentage and plant height in all experiments, and for plant height only in Experiment 1.
The heritability for 1000 grain weight, number of panicles per metre squared, number of spikelets per panicle was similar across weed treatments in all experiments, except where weeding was done once in Experiment 2, which had a lower heritability. In contrast, heritability for ripening percentage, plant height and yield varied across weeding treatments.
Weeds are a major biological constraint to upland rice production, especially in low input systems, where resource constrained farmers cannot afford herbicides and therefore rely on manual labour for weed control (Becker and Johnson, 2001; Rodenburg and Johnson, 2009; Saito et al. 2010). In this study, the effect of weed management was investigated on the yield of the associated crop although it could have far reaching consequences on weed species composition and abundances. Effects of management variables on weed flora have been studied for annual crop production systems. Crop rotation and reduced tillage were found to be more important than the amounts of fertilizer and herbicide applied in restraining seed production for both grassy and broadleaf weeds (Kagode et al., 1999). In many parts of Uganda, weed management is not stringent, and is characterised with shallow tillage of weedy fields immediately before sowing, followed by occasional weed removal during the cropping season. As such, weed composition is more related to soil characteristics than cropping system (Ugen and Wortmann, 1997). Towards crop maturation, weeds are normally not removed and after crop harvest, weeds re-establish and continue to grow during the off season period resulting in a dynamic weed flora with an abundant supply of seeds and other propagules (Ugen and Wortmann, 1997).
This study investigated the effect of weeding regimes on selected agronomic traits in farmer preferred upland rice varieties. Generally, there was an increase in yield progressing from no weeding to first weeding, to second and third weeding, although the increases in yield varied with season and the trait considered. This observation agrees with that of Toure et al (2013) and JICA (2010) and confirms the benefit of weeding in rice production. Furthermore, traits such as number of panicle per metre squared, plant height and yield showed more variation compared others such as ripening percentage and 1000 grain weight. This observation corroborates that of Hogue et al (2013).
Given the fact that the small holder upland rice farmers rely on manual labour for weed control which also becomes scarce, it is important to consider the actual number of weeding regimes that would maintain a high rice yield. This study showed that there was almost no difference between the second and the third weeding regimes across the varieties and the traits measured. This observation agrees with that of Akobundu and Ahissou (1985), Alou et al. (2012) and Toure et al. (2011). Therefore, at least one well-timed weeding regime is sufficient to get a good rice yield under upland field conditions of Lake Albert crescent zone of Uganda.
At variety level, the observation of increase in rice yield at each weeding regime level was still evident. However, there was variation in varietal response to each weeding regime. This observation is similar that of several studies Fofana and Rauber (2000); Haefele et al., 2004; Rodenburg et al., 2009; Saito et al., 2010). Moreover, the variations observed across the range of varieties used in this study varied with each trait and season. Traits such as number of panicles per metre squared, ripening percentage, and yield showed the highest variation.
The interaction of weeding regime and variety was not significant for most traits in all experiments, except traits such as ripening ratio and yield in experiment two and three. Yield is quantitative trait which is highly influenced by environment. In the context of our study, the different weeding regimes and seasons caused unique interactions with variety. Modal et al. (2013) observed a similar trend using three Aus rice varieties BR 26, BRRI dhan27, BRRI dhan48 and Pariza; and five weeding treatments viz. no weeding, one hand weeding at 20 DAS (Days after sowing), two hand weeding at 20 DAS and 30 DAS, three hand weeding at 20, 30 and 40 DAS and weed free condition planted using broadcasting method.
It should be noted that our results are unique to the population of upland rice varieties evaluated, and the rice production ecology (in this case, upland ecology). As such, they may not be applicable to other rice production ecologies or water stress conditions, and rice varieties. Therefore, validating our findings, including the identification of clear distinction between the effect of soil characteristics and management (weeding) might be needed before making recommendations based on them.
In conclusion, this study has found that there is a profound effect of weeding regime on upland rice yield in the study region. In addition, the study has also established that different varieties respond differently to the weeding regimes implemented by the farmer. Overall, the farmer may need to apply at least one hand hoe weeding activity on a highly responsive and well adapted variety, such as NERICA 4 to achieve an acceptable yield.
The authors have not declared any conflict of interests.
The authors are thankful to the NERICA Rice Promotion Project under Japan international cooperation agency (JICA) and the Government of Uganda for funding the experiment, to technicians of Bulindi ZARDI who participated in the trials for their cooperation.
REFERENCES
|
Akobundu IO (1980). Weed science research at the International Institute of Tropical Agriculture and research needs in Africa. Weed Science 28:439-445.
|
|
|
|
Akobundu O, Ahissou A (1985). Effect of inter-row spacing and weeding frequency on the performance of selected rice varieties on hydromorphic soils of west Africa. Crop Protection 4:71-76.
Crossref
|
|
|
|
|
Alou IN, Kakeeto R, Akio G (2012). Effect of weeding regimes on yields of Nerica 4 rice in an acidic ferralsol in Lake Albertine crescent. Minia international conference for Agriculture and irrigation in the Nile Basin countries, 26th -29th March. El-Minia, Egypt, pp. 1309-1313.
|
|
|
|
|
Anyanga RT (2015). Effects of varietal differences, plant spacing and weeding regimes on weeding regimes on weed density and yields of upland rice in Uganda. MSc thesis. Kenyatta University.
|
|
|
|
|
Bari MN, Mamun AA, Anwar SMS (1995). Weed infestation in transplant Aman rice as affected by land topography and time of transplanting. Bangladesh. Journal of Agricultural Research 22:227-235.
|
|
|
|
|
Becker M, Johnson DE (2001). Cropping intensity effects on upland rice yield and sustainability in West Africa. Nutrient Cycling Agroecosystems 59:107-117.
Crossref
|
|
|
|
|
Dzomeku IK, Dogbe W, Agawu ET (2007). Response of NERICA rice varieties to weed interference in the Guinea savannah uplands. Agronomy Journal 6:262-269.
Crossref
|
|
|
|
|
Ekeleme F, Kamara AY, Oikeh SO, Omoigui LO, Amaza P, Abdoulaye T, Chikoye D (2009). Response of upland rice cultivars to weed competition in the savannas of West Africa. Crop Protection 28: 90-96.
Crossref
|
|
|
|
|
Fofana B, Rauber R (2000). Weed suppression ability of upland rice under low input conditions in West Africa. Weed Research 40:271-280.
Crossref
|
|
|
|
|
Haefele SM, Johnson DE, M' Bodj D, Wopereis MCS, Mie´ zan KM (2004). Field screening of diverse rice genotypes for weed competitiveness in irrigated lowland ecosystems. Field Crops Research 88:39-56.
Crossref
|
|
|
|
|
Hyuha T (2006). Profit efficiency among rice producers in Eastern and Northern Uganda. Unpublished PhD Thesis. Makerere University.
|
|
|
|
|
Hyuha TS, Bashaasha B, Nkonya E, Kraybill D (2007). Analysis of profit inefficiency in rice production in Eastern and Northern Uganda. African Crop Science Journal 15(4):243-253.
|
|
|
|
|
Imanywoha JB (2001). Rice. In: Agriculture in Uganda, Vol. 2 Crops Mukiibi JK (Ed.), Fountain publishers/CTA/NARO. 70 p.
|
|
|
|
|
JICA (2010). NERICA Rice promotion Project. Rice cultivation Handbook, 46 p.
|
|
|
|
|
Johnson DE, Wopereis MCS, Mbodj D, Diallo S, Powers S, Haefele SM (2004). Timing of weed management and yield losses due to weeds in irrigated rice in the Sahel. Field Crops Research 85:31-42.
Crossref
|
|
|
|
|
Kaizzi KC, Byalebeka J, Semalulu O, Alou I, Zimwanguyizza W, Nansamba A, Odama E, Wortmann CS (2014). Upland rice response to nutrient application in Uganda. African Journal of Plant Science 8:416-425.
Crossref
|
|
|
|
|
Saito K, Futakuchi K (2014). Improving estimation of weed suppressive ability of upland rice varieties using substitute weeds. Field Crops Research 162:1-5.
Crossref
|
|
|
|
|
Lodin JB (2005). An Assessment of Suparica 2 Upland Rice Cultivation and its Impacts on Food Security and Incomes in Hoima District, Uganda. 72 p.
|
|
|
|
|
Miah MHN, Karim MA, Rahman MS, Islam MS (1990). Performance of nitrogen nutrients under different row spacing. Bangladesh Journal of Training and Development 3 (2):31-34.
|
|
|
|
|
Mohamed A (2012). Incentives and disincentives for rice in Uganda. Available at:
View
|
|
|
|
|
Odogola WR (2006). Final survey report on the status of rice production, processing and marketing in Uganda. A report submitted to the Embassy of Japan in Uganda through JICA and Sasakawa Africa Association-Uganda, Kampala.
|
|
|
|
|
Otsuka K, Kaliranjan K (2006). Rice Green Revolution in Asia and its transferability to Africa. Journal of Development Economics 44:107-122.
Crossref
|
|
|
|
|
Payne RW, Murray DA, Harding SA, Baird DB, Soutar DM (2011). GenStat for windows. 17th edition. VSN International, Hemel Hempstead, UK. Plant Biology 38:1016-1023.
|
|
|
|
|
Rodenburg J, Johnson DE (2009). Weed management in rice-based cropping systems in Africa. Advances in Agronomy 103:150-218.
Crossref
|
|
|
|
|
Saito K, Azoma K, Sie´ M (2010). Grain yield performance of selected lowland NERICA and modern Asian rice genotypes under non-fertilized and fertilized conditions in the lowlands of West Africa. Crop Science 50: 281-291.
Crossref
|
|
|
|
|
Sanyal D, Bhowmik PC, Anderson RL, Shrestha A (2008). Revisiting the perspective and progress of integrated weed management. Weed Science 56:161-167.
Crossref
|
|
|
|
|
Terry PJ, Michieka RW (1987). Common weeds of East Africa. Food and Agriculture Organization of the United Nations P 184.
|
|
|
|
|
Touré A, Rodenburg R, Saito K, Oikeh S, Futakuchi K, Gumedzoe D, Huat J (2011). Cultivar and weeding effects on weeds and rice yield in a degraded upland environment of the coastal Savanna. Weed Technology 25:322-329.
Crossref
|
|
|
|
|
Touré A, Sogbedji JM, Gumedzoe YMD (2013). The critical period of weed interference in upland rice in northern Guinea savanna: Field measurement and model prediction. African Journal of Agricultural Research 8:748-1759.
|
|
|
|
|
Promotion of Rice Development Project (2012). Weeds in Upland Field. [Hand book], p. 50.
|
|
|
|
|
Wopereis MCS, Diagne A, Rodenburg J, Sie´ M, Somado EA (2008). Why NERICA is a successful innovation for African farmers: a response to Orr et al. from the Africa Rice Center. Outlook on Agriculture 37:169-176.
Crossref
|
|
|
|
|
World Bank (1993). Growing out of Poverty: A World Bank Country Study (Uganda). World Bank, Washington D.C.
|
|
|
|
|
Zhao DL, Atlin GN, Bastiaans L, Spiertz, JHJ (2006a). Cultivar weed-competitiveness in aerobic rice: heritability, correlated traits, and the potential for indirect selection in weed-free environments. Crop Science 46: 372-380.
Crossref
|
|
|
|
|
Zhao DL, Atlin GN, Bastiaans L, Spiertz JHJ (2006b). Developing selection protocols for weed competitiveness in aerobic rice. Field Crops Research 97: 272-285.
Crossref
|
|
