The Intensity of a Bean Fusarium Root Rot Epidemic is Dependent on Planting Strategies
Abstract
Restricting Fusarium root rot (FRR) epidemics and improving productivity using fewer chemicals is a major concern in bean-growing regions. The main purpose of this research was to identify the planting strategies associated with FRR development and seed production in bean crops. A 2-year study of 122 commercial bean farms in four major producing regions in Zanjan, Iran, was performed to characterize the associations of farming indicators with FRR and productivity. Linear mixed-effect models, contingency tables and correspondence analyses were used to describe the variables relationships. Low disease and high productivity were linked to herbicide and manure applications, fungicidal treatment of seeds, manual sowing and sprinkler irrigation. Furrow irrigation, mechanical sowing, planting on raised seedbeds, the lack of fertilizer and herbicide use accounted for high disease and low seed production. The results of the study provide further evidence of factors contributions to the wider FRR spread over bean-cropping systems. Overall, this suggests that the selected composition of planting strategies can improve bean production and reduce FRR intensity.
Introduction
Although bean crops reduce fertilizer needs via biological N2 fixation and mobilization and recycling of soil phosphorus (Franche et al. 2009), low yields due to root rots (Naseri 2008) are serious obstacles to wider cultivation of this profitable crop in Zanjan, Iran. Fusarium solani f. sp. phaseoli (FSP) is a prevalent and economically important disease in the bean-growing parts of the province and occurs in most Zanjan soils (Naseri 2014). In farms infested with FSP, bean yields have been decreased up to 52% (Naseri and Marefat 2011). It is believed that the development of root rot diseases is relatively dependent on certain agro-ecological conditions (Schwartz and Pastor-Corrales 1989). Indeed, cropping methods can influence the degree to which bean plants are prone to Fusarium root rot (FRR) and can be manipulated to minimize the disease prevalence. In Zanjan, bean production is threatened by the following features presumably increasing the prevailing FRR risk: frequent irrigation, high plant and weed density, deep and early sowing, overuse of urea, growing bean in loamy soils, low soil organic matter, the lack of rhizobial root nodulation and soil treatment with improper fungicides such as benomyl (Naseri and Marefat 2011; Naseri 2014). A build-up of soil and seed infestations and severe disease levels under optimum environmental and plant conditions for FSP may be the consequences of such a cropping system (Naseri 2013). Because of lacking resistance in commercial bean cultivars or local varieties, FRR problems may be worsened by disease-conducive agro-ecosystems in the region (Naseri and Marefat 2011; Naseri 2014). Given that the ten above-mentioned agronomic and soil properties are responsible for the occurrence of bean–FRR across an area, the use of only one or few strategies seems insufficient to control the disease. Thus, the objective of this study was to extend our current epidemiological understanding of the bean–FRR pathosystem to choose more relevant disease descriptors for small-scale experimentation and then develop adequate integrated disease management systems. Attention has also been drawn to those components that their manipulation may lead to a more sustainable bean production and improve organic cultivation. The impacts of animal manures (Sumner et al. 2002), fungicidal treatment of seed (Papavizas and Lewis 1975), herbicides (Mussa and Russell 1977), irrigation system (Miller et al. 1980) and seedbed (Snapp et al. 2003; Valenciano et al. 2006) on the bean–Fusarium pathosystem have been reported previously. However, the contribution of farmers' operations to the establishment of bean–FRR epidemics is still largely unknown. Therefore, this study focused on individual and combined interaction of planting strategies with FRR development and bean productivity under prevailing agro-ecological conditions at Zanjan province.
Materials and Methods
Study area
The sampling survey was conducted on farms in the province of Zanjan that accounts for 12% of bean production in cultivated areas of Iran (Anonymous, 2009). The province is the third largest producer in Iran with 9000 ha distributed among four bean-growing areas: Abhar, Kheirabad (Zanjan), Khoramdareh and Qeydar (Fig. 1). To encompass a large range of production situations, all of the Zanjan bean regions were sampled, as they were very different from an agro-ecological point of view. Cropping systems have been quite varied (Naseri and Marefat 2011), as the techniques have been applied by farmers depending on their experience and investment capacity. The soils were also varied in terms of chemical composition and texture (Naseri 2014). Furthermore, each production region differs greatly from other regions in terms of the disease intensity, soil population and seed infestation by FSP (Naseri and Emami 2013). Given the diversity of pathogen prevalence, environments and cropping systems, a considerable diversity in common bean production situations in the area was expected.
Sampling methods
The methodological approach chosen was a survey which is a proper method to study epidemics affected by a large number of cropping factors (Avelino et al. 2007). This method has been used to investigate disease epidemics in soya bean in the north-central United States (Mila et al. 2003) and coffee in Costa Rica (Avelino et al. 2007). The study was performed on a total of 122 randomly chosen farms across the regions during the two growing seasons, 2008 (35 farms) and 2009 (87 farms). The study area and sampling methods have already been described by Naseri and Marefat (2011). In this article, we particularly describe the examination of data specific to the bean farm–FRR interplay. The farm descriptors assessed during the survey involved certain planting strategies as follows: animal manure usage, cultivation method, fungicidal treatment of seed, herbicide application, irrigation system and seedbed. Crop management variables included practices related to planting strategies, which were generally stable over time.
The sampling survey was designed in collaboration with bean growers who provided details about crop management features (e.g. fertilizer and herbicide usage, seed treatment and sowing method). All bean farms with FRR symptoms were visited directly by the first author to validate survey results and collect further farming data such as irrigation system and seedbed type. For disease assessment, we used the terms such as disease incidence, severity and index (Naseri and Marefat 2011). Disease incidence is a measure of the proportion (or percentage) of diseased plants relative to the total number of plants inspected. Disease severity measures the proportion of symptomatic root tissues using the following scale: 0 = no rot symptoms; 1 = 1–10% covered with lesions; 2 = 10–25% covered with lesions or rots; 3 = 25–50% rots; 4 = 50–75% rots; and 5 = 75–100% rots and plants death (Naseri 2008). Disease index was also defined as follows: disease incidence × disease severity/5. In each farm investigated, bean plants showing FRR symptoms were inspected, counted and sampled three times at V3 (the first trifoliolate leaf fully opened), R6–R7 (from flowering to pod formation) and R9 (pod maturity) growth stages of common bean (Van Schoonhoven and Pastor-Corrales 1987). Samples were assessed for the presence of F. solani in the laboratory. Pathogenicity of isolates was tested in the greenhouse as previously described by Naseri (2008). The potential yield was quantified by counting the number of seeds per plant just before the harvest.
Statistics
The data set was analysed with a linear mixed-effect modelling approach (GenStat; Payne et al. 2002), which is a type of regression analysis that includes both fixed and random effects. The response variables were the disease index and seed numbers per bean plant. Fixed effects were used to account for the influence of cultivation, seed treatment, herbicide, irrigation, manure and seedbed factors in interaction with the growth stage. These planting situations were treated as indicator variables, which are also referred to as dummy variables. Sampling year had no significant interaction with the cropping factors tested (data not shown). The farm term was included as a random-effect variable to account for dependencies in multiple observations (for different sampling times) from the same experimental site. Standard linear regression models assume that each observation is independent and therefore captures correlations among observations from the same farm (Maia et al. 2010). The standard error of difference (SED) test was used for means separation, P = 0.05.
The contingency tables analysis (CTA) evaluated the individual significance of categorical variables that affect either of the response variables. The association of the farming variables was examined twice in terms of their effect on the disease index at pod maturity stage and seed numbers per bean plant. Contingency tables of dependent and independent variables were built to represent the bivariate distribution of the farms. To produce binary-dependent variables, quantitative data for the disease index and seed number were classified into distinct groups so that class boundaries of ≤50 and >50 included nearly equal individuals. Descriptive parameters were used to analyse the index of FRR and seed production in relation to the independent variables (Tables 1 and 2). The variables for which the χ2 value was low (P > 0.10) were considered redundant because of their absence of links to the disease or productivity. The number and size of groups were chosen to respect the condition of validity of the χ2 test performed, so that no more than 20% of the expected values (assuming independence) should be smaller than 5 (Gibbons 1976).
| Variablea | Variable class | Total farms no. | FRR index | χ2 test | ||
|---|---|---|---|---|---|---|
| ≤50% | >50% | Value | P | |||
| Cultivation | Mechanical | 89 | 39 | 50 | 6.5 | 0.011 |
| Manual | 33 | 23 | 10 | |||
| Fertilizer | No fertilizer | 40 | 14 | 26 | 9.6 | 0.008 |
| Chemical | 77 | 43 | 34 | |||
| Manure | 5 | 5 | 0 | |||
| FRR index | ≤50% | 62 | – | – | – | – |
| >50% | 59 | – | – | |||
| Herbicide | No herbicide | 39 | 14 | 25 | 5.8 | 0.056 |
| Trifluralin | 73 | 41 | 32 | |||
| Others | 10 | 7 | 3 | |||
| Irrigation | Flood | 85 | 45 | 40 | 6.4 | 0.041 |
| Sprinkler | 5 | 5 | 0 | |||
| Furrow | 32 | 12 | 20 | |||
| Seedbed | Flat land | 90 | 50 | 40 | 3.1 | 0.079 |
| Raised | 32 | 12 | 20 | |||
- a Only variables with significant associations (P ≤ 0.10) are presented.
| Variablea | Variable class | Seed no./plant | χ2 test | ||
|---|---|---|---|---|---|
| ≤50 | >50 | Value | P | ||
| FRR index | ≤50% | 26 | 36 | 5.6 | 0.018 |
| >50% | 38 | 22 | |||
| Herbicide | No herbicide | 26 | 13 | 7.3 | 0.026 |
| Trifluralin | 31 | 42 | |||
| Others | 7 | 3 | |||
| Irrigation | Flood | 44 | 41 | 5.8 | 0.056 |
| Sprinkler | 0 | 5 | |||
| Furrow | 20 | 12 | |||
- a Only variables with significant associations (P < 0.10) are presented.
The combined associations of the non-metric farm descriptors with FRR at R9 and productivity were analysed using a correspondence analysis (CA) procedure of GenStat. Correspondence analysis is a statistical technique developed to describe the relationships between two or more categorical variables using χ2 matrics (Savary et al. 1995). Three main parameters were considered to interpret CA results as follows: the proportion of total inertia explained by each principal axis, the association of the categorical variables with the axes inertia which convey the axis meaning and finally the cos2 of the angle between the line origin modality and the axes, values close to 1 demonstrate a very good illustration of the modality on the axis (Avelino et al. 2007). Correspondence analysis simplified the large data set by scoring farms according to their cropping patterns and by evaluating the variables' significance in the farming systems studied. A simple CA was performed using a contingency table in which farming variables were in columns and the farms in rows. Then, the coordinates of the farming variables on the selected axes led to the detection of more important cropping properties in the data set. The second CA test used the modalities of FRR and productivity variables in columns and the categories of farming variables in rows. Only the first two axes, which explained the greatest inertia, were considered principal in our CA test. This maintained maximum information linked to the disease and productivity and, at the same time, left loose linkages with unimportant individuals explained well by last CA axes. Only the farm descriptor categories that were well correlated to the principal axes (sum of cos2 > 0.2) were considered for CA interpretations. The purpose of the present analysis was to characterize the properties of bean fields accounted for FRR heterogeneity across the area, which were chosen because they looked indeterminate or suspicious by local farmers and experts.
Results
The two-way interaction of fertilization and growth stage affected bean–FRR index in the farms studied (Wald statistic = 13.2; Df = 4; P = 0.010). At V3 stage, mean FRR index differed between the classes of chemical and no fertilization in being lower for chemicals (Table 3). At R6–R7 and R9 stages, mean FRR index for manure application differed (P < 0.05) from either chemical fertilization or no fertilizer use, with manure usage being lowest. Mean FRR index in chemical- and non-fertilized farms differed (P < 0.05) between the growth stages, being lower at V3. With manure usage, FRR index decreased significantly with increasing time from either V3 or R6–R7 to R9 stage (P < 0.05). The mean number of seeds per plant for no fertilizer application differed (Wald statistic = 5.2; Df = 2; P = 0.075) from either chemical (mean = 64.8) or manure (mean = 77.5) fertilization, with no fertilization being lowest (mean = 45.9; SED = 17.5).
| Factorsa | Factor levels | Growth stages | ||
|---|---|---|---|---|
| Vegetative stage | Flowering–Podding | Pod maturity | ||
|
Cultivation SED = 5.3 |
Mechanical | 33.4 | 49.2 | 53.0 |
| Manual | 17.4 | 29.4 | 30.0 | |
|
Fertilizer SED = 9.8 |
No fertilizer | 36.4 | 46.2 | 60.4 |
| Chemical | 23.3 | 43.1 | 41.9 | |
| Manure | 31.4 | 27.8 | 11.5 | |
|
Herbicide SED = 8.6 |
No herbicide | 39.0 | 46.9 | 59.6 |
| Trifluralin | 24.9 | 42.3 | 40.7 | |
| Others | 19.1 | 41.2 | 40.9 | |
|
Irrigation SED = 11.1 |
Flood | 25.2 | 43.4 | 43.3 |
| Sprinkler | 12.8 | 24.0 | 16.5 | |
| Furrow | 41.5 | 47.2 | 59.7 | |
|
Seedbed SED = 5.3 |
Flat land | 24.7 | 42.6 | 42.1 |
| Raised | 41.5 | 47.2 | 59.7 | |
- SED, standard error of differences.
- a Only factors with significant effects (P < 0.05) are presented.
The most common herbicide used by bean growers was trifluralin, followed by paraquat and bentazon. Herbicide application affected FRR index (Wald statistic = 7.3; Df = 2; P = 0.026) and the number of seeds per bean plant (Wald statistic = 11.1; Df = 2; P = 0.004). At V3 and R9 stages, there were significant differences (P < 0.05) in the mean FRR index between the classes of herbicide application, the disease values being greater for the lack of herbicide use (Table 3). For the three categories of herbicide factor, a lower FRR index (P < 0.05) was detected at V3 in comparison with R6–R7 and R9 stages. There were significant differences (P < 0.05) in the mean number of seeds per plant between the classes of herbicide application, with the yield values being higher for the application of trifluralin (mean = 70.3) compared to miscellaneous herbicides (mean = 38.5; SED = 13.4) and the lack of herbicide application (mean = 43.5).
Irrigation system affected FRR index (Wald statistic = 9.3; Df = 2; P = 0.009) and the number of seeds per bean plant (Wald statistic = 7.5; Df = 2; P = 0.023). Mean FRR index at the three growth stages differed (P < 0.05) between the irrigation systems, being lowest for sprinkler irrigation compared to flood and furrow irrigation (Table 3). Comparing flood and furrow irrigation systems, flood-irrigated beans showed a lower FRR index at V3 and R9 stages. Under flood and furrow irrigation systems, FRR index at V3 was different (P < 0.05) from the disease levels at R9, in being lower at V3. Under sprinkler systems, FRR index differed (P < 0.05) between the growth stages, in being greater at R6–R7. The mean seed number per plant for sprinkler irrigation (mean = 98.0) significantly differed from those for flood (mean = 63.2) and furrow irrigations (mean = 43.0; SED = 19.1).
The type of seedbed affected FRR index (Wald statistic = 6.8; Df = 1; P = 0.009) and the number of seeds per bean plant (Wald statistic = 5.3; Df = 1; P = 0.022). At V3 and R9 stages studied, the mean FRR index differed significantly between the seedbed types in being higher (P < 0.05) for the farms where beans were sown on raised beds (Table 3). There were significant differences in mean FRR index between the sampling times tested, with V3 stage being the lowest. The two-way interaction of seedbed and seed treatment factors affected bean–FRR index in the farms studied (Wald statistic = 6.4; Df = 1; P = 0.011). The lowest disease was detected in the farms where fungicide-treated seeds were sown on raised beds. The mean seed number per plant for flat land (mean = 64.8) was significantly different from that for raised bed (mean = 43.0; SED = 9.5).
The method of cultivation affected FRR index (Wald statistic = 16.9; Df = 1; P = 0.001) and the number of seeds per bean plant (Wald statistic = 5.6; Df = 1; P = 0.018). For both manual and mechanical cultivation in the farms studied, FRR index at V3 was different (P < 0.05) from the disease levels at R6–R7 and R9 in being lower at V3 (Table 3). Mean FRR index at the three growth stages differed (P < 0.05) between the classes of cultivation factor, with manually sown farms being lower. The mean seed number per plant for manual sowing (mean = 75.2) was significantly different from that for mechanical sowing (mean = 53.0; SED = 9.4).
According to the χ2 tests of independence, CA and CTA analyses, FRR index and bean productivity were dependent on the crop management patterns in the farms studied. To avoid complication in interpreting CA and CTA results, we focused on the disease index at pod maturity stage. In addition, the effect of bean growth stage was addressed by REML analysis. The categories of bean-farming characteristics were often linked (P ≤ 0.10) to FRR modalities based on CTA results (Table 1). The productivity modalities were associated with the categories of FRR, herbicide and irrigation factors (Table 2). Based on the first CA, the planting strategies were ordered according to their contribution to the inertia (91.6% and 8.4%) explained by the first two axes (Table 4). The results of second CA showed that the first two axes accounted for 72.3% and 12.7% (85.0% in total) of the variance in the data, which indicated a good fit of the ordination to the data set. Fertilizer and irrigation modalities were best represented on principal axis 1 (cos2 > 0.8; Fig. 2). Herbicide modalities were best illustrated on principal axis 2 (sum of cos2 > 0.4). Principal axis 1 was an axe of decreasing FRR intensity from high (Frr2 > 50%) to low levels (Frr1 ≤ 50%). The contribution of each disease modality to axis 1 was 20%. Principal axis 1 was also an axe of increasing bean productivity from low (Snp1 ≤ 50 seed no./plant) to high levels (Snp2 > 50 seed no./plant). The contributions of Snp1 and Snp2 to axis 1 were, respectively, -0.15 and 0.16. There were no significant contributions (cos2 < 0.1) of Frr and Snp modalities to axis 2. The proximity of the different categories and of the modalities on the graphic representation of the CA could therefore be interpreted as being indicative of an association. The low disease modality (Frr1) and high seed productivity (Snp2) were associated to fertilization of farms with manure (Fer2), fungicidal treatment of seeds (Tre2), herbicide application (Her1 and Her2), manual sowing (Cul2) and sprinkler irrigation (Irr2). High disease modality (Frr2) and low seed productivity (Snp1) were close to furrow irrigation (Irr3), mechanical sowing (Cul1), planting on raised seedbeds (Bed2), the lack of fertilizer (Fer0) and herbicide (Her0) use. Most of these associations were individually described in the previous sections, REML and contingency tables' analyses.
| Variablesa | Coordinates | Sum of cos2b | |
|---|---|---|---|
| Axis 1 | Axis 2 | ||
| Cultivation | −0.01 | 0.04 | 0.05 |
| Fertilizer | 0.53 | 0.29 | 0.82 |
| Herbicide | 0.51 | −0.41 | 0.92 |
| Irrigation | −0.66 | −0.03 | 0.69 |
| Seedbed | 0.15 | 0.06 | 0.21 |
- a Seed treatment factor was excluded from the analysis due to the lack of significant association with the disease and productivity modalities in contingency tables analyses.
- b cos2 = The square cosine of the angle between the line origin modality and the axis.
Discussion
To our knowledge, this is the first attempt to integrate herbicide, irrigation, fertilization and sowing techniques in an approach aimed at analysing bean–FRR epidemics. Furthermore, results suggest multiple disease assessments are needed to accurately describe the interaction of bean and FSP with the farming factors studied.
Because low rates of broiler litter can be as effective as commercial fertilizers in snap beans (Brown et al. 1993), it is of interest to unwrap the interaction of manures with bean root rots for organic production purposes. Although poultry litter had no impact on snap bean–FRR, the use of non-composted litters increased the total soil population of saprophytic fungi (Sumner et al. 2002). This may imply a potential of litter to provide biocontrol of soilborne pathogens. In the present macroscale study, the application of animal manure by bean growers corresponded with FRR reductions over time and productivity improvements, while the reverse trend was observed in chemical- and non-fertilized farms. Provided that plot-scale experiments confirm these regional findings, replacing commercial fertilizers with animal manure may restrict FRR spread in beans without yield loss.
Trifluralin and bentazon increased the virulence of FSP on potted bean plants in the absence of weeds (Mussa and Russell 1977). However, knowledge of the impact of herbicides on pathosystems, including pathogens and weeds, is obviously desirable to minimize harmful effects on crop productivity. The present large-scale findings provide new information on the interaction of the bean–Fusarium pathosystem with farmers' choice herbicides. The lack of herbicide application corresponded with a higher early (V3) and late season (R9) FRR level. Relatively, denser weed populations due to the lack of chemical weeding may have intensified bean–FRR, as Naseri and Marefat (2011) reported high weed densities in severely FRR-affected beans. FRR reduction and increase of productivity in this study also revealed the benefit of trifluralin use for Fusarium-infected beans that merit further plot-scale examinations.
Although extensive production of crops, root disease damage and limited water resources have led to a need for more efficient irrigation strategies, furrow and flood irrigation are widely used by Iranian farmers. Fusarium solani has been detected in open irrigation systems (Shokes and McCarter 1979), which increase the likelihood of pathogen introduction (Zappia et al. 2014) and may induce root rots due to oxygen stress resulting from water accumulation (Miller et al. 1980). A controversial belief is that furrow irrigation may restrict root rots by leaving the surface of raised beds dry and supplying water to roots (Cook and Papendick 1972). Under agroclimatic conditions encountered during this study, sprinkler irrigation had the lowest FRR index over the sampling period. In contrast to flood and furrow systems, FRR index decreased from flowering to pod maturity for sprinkler irrigation. In addition, sprinkler-irrigated farms had 36 and 56% higher seed productivity than flood- and furrow-irrigated farms, respectively. Because FSP is favoured by host water stress (Burke et al. 1972), sprinkler irrigation might have lowered FRR and enhanced bean productivity by maintaining an adequate supply of water to roots. Lower early and late season FRR for flood irrigation compared to furrow system appeared to be the first report of the advantage of flood irrigation for bean–FRR pathosystem. The present findings merit plot-scale examination in the future.
Growing beans on raised beds is known to be beneficial during rainy seasons as they improve soil aeration, promote root system development and possibly increase tolerance to root rots (Buruchara and Rusuku 1992). Snap beans grown on raised beds had a lower FRR and greater yield compared to flat beds in a FSP-inoculated soil (Snapp et al. 2003). Elsewhere, raised beds reduced plant growth and intensified bean root rots, probably because of deeper sowing and delayed emergence (Valenciano et al. 2006). Based on the present findings, a lower FRR index and 50% greater productivity were detected in beans sown on flat lands compared with raised beds. This scientific knowledge obtained from diverse agro-ecosystems not only provided interesting developments of previous plot-scale approaches, but also highlighted the necessity of incorporating this strategy into a FRR-control programme.
The treatment of bean seeds with fungicides reduced root rot risks (Papavizas and Lewis 1975; Valenciano et al. 2006). Although both seed treatment and seedbed operations are known as influential factors in bean-farming systems, the interaction of these two practices has received no consideration. In this regional research, profoundly lower FRR levels were detected for fungicide-treated seeds sown on raised beds. Thus, the effect of seed treatment practice on FRR was dependent on seedbed option. If confirmed in experimental fields, this would have implications for improving bean-cropping programmes.
With a highly diverse combination of agronomic and soil properties assessed in the present research, manually sown farms had 40–48% lower FRR index than mechanically cultivated ones at the three growth stages tested. Moreover, seed production for manual sowing was 30% higher than that for mechanical sowing. To the best of our knowledge, this is the first description of the impact of cultivation strategy on FRR and bean productivity. Such higher FRR and lower productivity for mechanical cultivation might be attributed to soil compaction stress, which is a problem with mechanized crop production (Allmaras et al. 1988). Soil compaction can increase root rots by reducing porosity and hindering aeration, water movement and root growth (Burke et al. 1972).
In conclusion, herbicide, fertilizer, irrigation, seedbed, cultivation and seed treatment strategies were assumed to be involved in FRR control and production improvement according to the present findings. In reality, some of the relations have already been reported, but in a dispersed manner. Thus, our study provides a clearer understanding of this combined selection of planting methods in association with the pathosystem. Further study of the mechanisms that govern bean–FRR epidemics is still required.
Acknowledgements
The authors thank the Iranian Ministry of Agriculture, for funding this research (project no. 4-47-16-88113). Thanks are also due to all field staff of Zanjan Agricultural & Natural Resource Research Center and students of Zanjan University for their help and expertise.
