Search PMN  

 

PDF version
for printing

Peer Reviewed
Impact
Statement




© 2009 Plant Management Network.
Accepted for publication 13 July 2009. Published 31 August 2009.


Relationships Between Yield and Crown Disease of Sweet Corn Grown in the Willamette Valley of Oregon


Nathan L. Miller and Cynthia M. Ocamb, Department of Botany and Plant Pathology, Oregon State University, Corvallis OR 97331


Corresponding author: Nathan L. Miller. millern@science.oregonstate.edu


Miller, N. L., and Ocamb, C. M. 2009. Relationships between yield and crown disease of sweet corn grown in the Willamette Valley of Oregon. Online. Plant Health Progress doi:10.1094/PHP-2009-0831-01-RS.


Abstract

Sweet corn (Zea mays L.) yields in the Willamette Valley of Oregon declined during the 1990s. Severe root rot affected some plants shortly before harvest, but was absent in other plants that showed secondary symptoms of reduced ear yield and leaf death; necrosis of stalk nodes and crown tissues was found instead. Studies were done to determine if there is a relationship among yield and necrosis of crowns, stalk nodes, nodal roots, radicles, or sub-crown internodes. An image analysis program was used to quantify the grayscale value of crown and node tissues. Regression analysis indicates that plants with darker crown tissues have lower ear weights. Rots of the nodal roots, radicle, or sub-crown internode were poor predictors of ear weight at harvest. When either Fusarium oxysporum or F. verticillioides were isolated from crowns of commercial sweet corn plants, these crowns had significantly darker grayscale values than those from which neither species was isolated; ear weights were also lower when F. oxysporum was isolated from the crown or a stalk node.


Introduction

Sweet corn (Zea mays) growers in the Willamette Valley of Oregon reported declining yields during the early 1990s. The decline in yields was initially associated with leaf "firing," where the leaves die prematurely starting at the base of the plant and then progressed up the plant. Root rot has been implicated as the primary cause of this yield decline in western Oregon (3). However, an atypical stalk disease was found in numerous symptomatic sweet corn plants growing in processor fields in the Willamette Valley where a rot of crown as well as stalk nodes could be found as early as the 4-leaf stage with little to no root rot (Ocamb, unpublished). But relationships among root rot, crown and stalk node rot, and yield loss are not clear. The objective of this study was to determine which, if any, disease symptoms are linked to reductions in ear weights in commercial fields. The associations between crown or stalk node necrosis and ear weight to the presence of Fusarium infection were also investigated.


Association of Yield and Symptoms

Ten commercial sweet corn fields in the Willamette Valley were sampled a few days prior to commercial harvest. Each field was divided by visual assessment into ten similarly-sized sections. One transect, chosen randomly, was sampled in each of the ten sections. In each transect, 10 consecutive plants within one row were sampled. Individual, mature ear weights were recorded after removing husks. Root balls were washed clean of soil and the percentage of nodal roots with rot was rated on the following scale: 1 = 1 to 25% of the nodal root ball was rotted, 2 = 26 to 50%, 3 = 51 to 75%, or 4 = 76 to 100%. Rots of the primary root (radicle) and sub-crown internode (mesocotyl) were rated separately using a similar 1 to 4 scale based on 25% increments. Roots were also rated for severity of larval feeding by Diabrotica undecimpunctata undecimpunctata using the following 0 to 3 scale: 0 = no root tunneling observed on any roots, 1 = one to three roots had tunnels, 2 = more than three roots with tunnels but less than half the roots in the root ball had larval feeding damage, or 3 = more than half the roots in the root ball had larval feeding damage.

Stalks and crowns were split longitudinally, parallel to the leaf ranks, and the cut surface was digitally captured on a flatbed scanner (HP scanjet 2400, Hewlett Packard, Palo Alto, CA). Multiple layers of black fabric were used to cover the stalks while scanning to minimize extraneous light. The mean grayscale was calculated for the crown region and the first stalk node immediately above ground that lacked rooted brace roots using the digital image analysis program, ImageJ version 1.34s (US Department of Health and Human Services, National Institutes of Health) (Fig. 1).


 

Fig. 1. Examples of sweet corn crown images (inset) and ImageJ grayscale histogram output of the crown region. Stalks were cut longitudinally at the base and scanned on a flatbed scanner. Each pixel of the specified region is measured on a 256-bit grayscale and the histogram is generated using the number of pixels in each grayscale shade. Larger means indicate lighter grayscale values.

 

The disease measurements were analyzed using mixed linear models (SAS 9.1, SAS Institute Inc., Cary, NC), and transect means were used for statistical analysis of all variables. Statistical models were tested for goodness-of-fit using Bayesian Information Criterion (BIC) (7) to find which models best explained apical ear weight. Sampling took place in several different cultivars on various calendar dates across the geographic area of the valley, so models included site as a random effect. Models with each variable alone, without site, were also included. When models included site as an explanatory variable, all possible combinations of crown, lower stalk node, sub-crown internode, radicle, and nodal root rot as linear variables, without interactions, were tested. Also tested were models containing site and one of the rot variables as well as an interaction between site and that other variable. A total of 43 models were included in the analysis. Information criterion techniques allow many regression models with varying numbers of explanatory variables to be compared for quality of fit to the data. A penalty is imposed for adding explanatory variables to a model to prevent over-fitting. A separate set of analyses were conducted removing single sites, to determine the relative contribution of each site on the outcome of the BIC analysis. Rootworm feeding was not included in the first Bayesian Information Criterion model selection because the first three sites did not include rootworm ratings and the difference in sample size would bias the results. Separate analyses were conducted using only sites where rootworm feeding was quantified. Including rootworm as a variable increased the total number of models tested to 78.

Regression models with lower BIC values are considered better fitting models. BIC values differing by less than 2.0 indicate that there is little difference in the quality of fit between two models (1). All models that include crown grayscale (models 1 through 17) as an explanatory variable had a better fit, as indicated by their smaller BIC values (Table 1), than models which did not include crown grayscale, except for the model that includes only crown (not site). The best-fitting model (Table 1) is a model that estimates a different slope and intercept for the crown effect at each site (model 1). In this model, darker crown grayscales (or more necrotic crowns) are associated with lower ear weights in 9 of the 10 sites sampled. The slopes of ear weight vs. crown grayscale were 3.19*, 0.83, -0.55, 2.21*, 3.18, 3.60*, 4.19*, 0.89, 3.64*, and 1.32* for sites 1 through 10, respectively, and six of the estimated slopes were significant at the P = 0.05 level (indicated by *). When field 3 is excluded, model 1 drops to the 16th best model and model 3 becomes the best-fitting model (slope = 2.1 g decrease in ear weight per darker crown grayscale unit, P < 0.0001).


Table 1. Bayesian Information Criterion (BIC) selection of top 20 best-fitting models to explain variation in mean ear weight per transect (10 plants/transect, 10 transects/field) in 10 commercial sweet corn fields in Oregon’s Willamette Valley.

Modelw BICx Variables in modely Crown NRR PR SCI N1
1 914.1 Site Site•Crown −0.55-4.19*6z
2 917.2 Site Crown NRR 1.57** 14.4
3 918.2 Site Crown 1.44**
4 918.5 Site Crown PR 1.53** 7.2
5 919.1 Site Crown SCI 1.56** 5.4
6 919.3 Site Crown NRR N1 1.69** 14.5 −0.2
7 919.3 Site Crown NRR PR 1.64** 11.9 2.8
8 919.4 Site Crown NRR SCI 1.58** 12.8 1.8
9 920.0 Site Crown PR N1 1.59** 8.6 −0.37
10 920.3 Site Crown N1 1.51** −0.2
11 920.7 Site Crown PR SCI 1.66** 5.5 2.4
12 920.9 Site Crown SCI N1 1.67** 6 −0.28
13 921.1 Site Crown NRR N1 PR 1.56** 10.7 4.3 −0.3
14 921.3 Site Crown NRR SCI N1 1.69** 12.4 2.4 −0.24
15 921.6 Site Crown NRR SCI PR 1.68** 11.6 2.3 0.83
16 922.2 Site Crown SCI N1 PR 1.59** 6.8 2.3 −0.38
17 923.4 Site Crown NRR SCI N1 PR 1.70** 10.4 3.7 1 −0.3
18 935.1 Site
19 935.9 Site N1 0.5
20 936.8 Site NRR 7.2

 w Number of each mixed regression model with mean ear weight as the response variable.

 x Lower Bayesian Information Criterion (BIC) statistics indicate better fitting models.

 y Estimated slopes for the effect of each variable in the model as explanatory variables; Crown = crown mean grayscale, NRR = rating of nodal root rot, PR = primary root (radicle) rot, SCI = rating of sub-crown internode rot, and N1 = mean grayscale of the first stalk node aboveground.

 z For each estimated slope, * = statistically significant at P ≤ 0.05, and ** = statistically significant at P ≤ 0.01. Numbers in superscript represent the number of fields which had significant effects when each field had a different slope due to interaction terms (indicated by •) being included.


The second best model (model 2) based on BIC selection is a model that includes nodal root rot and crown rot with no interactions, so the crown and nodal root rot effects are held constant across all sites (Table 1). In this model, increased nodal root rot has a positive effect on yield (slope = 14.4 g increase in ear weight as nodal root rot increases, P = 0.07), while darker crowns had a highly significant negative effect on yield (slope = 1.57 g decrease in ear weight per darker crown grayscale unit, P < 0.0001). The coefficients (slopes) of disease variables other than crown grayscale are not statistically significant (P > 0.05) in all models that include crown grayscale. Since crown grayscale appears to be the most predictive variable, and only one field (site 3) is affecting the consistency of the slope of crown grayscale, it appears that model 3 is the most useful one for this data set.

Model 3, the parallel lines model for crown grayscale (Fig. 2) estimates a single slope for the crown effect (1.44 g decrease in ear weight per darker crown grayscale unit, P < 0.0001) and a different intercept for each field. In this model, a slope of 1.44 results in a reduction in predicted ear weight values by 6 to 20% over the range of grayscale observed at each site.


 

Fig. 2. Grayscale means of sweet corn crowns versus mean ear weight by field sites. The regression model includes mean crown grayscale and field as explanatory variables without interactions (A) (crown grayscale parallel lines model). The same model is adjusted so all fields fit a common intercept (B) to better show the relationship between crown grayscale and ear weight. Each point represents the transect mean based on 10 plants/transect in 10 transects in each field. Fields 1 to 5 were sampled in 2003 and 6 to 10 were sampled in 2004.

 

When the Bayesian Information Criterion model selection analysis was repeated with only the fields where rootworm feeding was recorded, the best model to explain ear weight included site, crown mean grayscale (slope = 2.3, P < 0.0001), nodal root rot (slope = 23.5, P = 0.06), and rootworm damage (slope = -13.2, P = 0.03) as explanatory variables. This is interpreted as a 2.3 g decrease in ear weight per darker grayscale unit of the crown, a 13.2 g decrease in ear weight associated with each unit of increased rootworm damage severity, and an increase in ear weight associated each unit increase in root rot severity, although not highly significant (P = 0.06). The second best model included only crown mean grayscale (slope = 2.3, P < 0.0001) and rootworm damage (slope = -12.9, P = 0.04) as explanatory variables. Crown grayscale is present in the top 34 models and always has significant slope values (P ≤ 0.05 for slope coefficients), again suggesting that crown grayscale has the most predictive value of ear weight at harvest. Rootworm damage appears to be the next best predictive variable, and is present in 13 of the top 14 models with significant slope values. The addition of disease variables other than crown grayscale add little, if any explanatory value to the models predicting ear weight, and do not have statistically significant slope values in the best-fitting models (top 34 models). Some explanatory variables were highly correlated which can change slope coefficients when two correlated variables are in the same model (Table 2).


Table 2. Pearson correlation coefficients, and associated P values of explanatory variables used for model testing to determine disease symptom and yield association of sweet corn plants in the Willamette Valley.

     SCI NRR RW Crown N1
PR 0.71
<0.0001
0.66
<0.0001
0.24
0.04
0.21
0.37
0.10
0.32
SCI 0.48
<0.0001
0.29
0.016
0.47
<0.0001
0.20
0.045
NRR 0.12
0.34
0.34
0.0007
0.22
0.0338
RW 0.29
0.01
0.089
0.46
Crown 0.54
<0.0001

Abbreviations: PR = primary root rot; SCI = sub-crown internode rot; NRR = nodal root rot; RW = rootworm damage; Crown = mean grayscale of the internal crown tissue; and N1 = mean grayscale of the internal lower stalk node tissue.

Crown and node grayscale values are inversed to make increased disease always positive.

There are 98 transects means for each variable except RW, which has 70, and each transect mean consists of 10 consecutive sweet corn plants in one row. Ten transects were sampled at each of 10 sites.


Association Among Fusarium Species, Crown and Node Necrosis, and Ear Weight

Crowns and lower stalk nodes were selected randomly from 30 to 40 of the 100 plants collected at sites 3 through 10. A portion (< 5 mm × 2 mm × 2 mm) of crown or node tissue was dissected under sterile conditions, dipped in 0.5% NaClO solution, and then rinsed with sterile reverse osmosis (RO) water. The tissue piece was then embedded in Nash-Snyder medium (6) amended with 120 µg/mL chlortetracycline HCl (4). Putative Fusarium colonies were transferred to carnation leaf agar (CLA) (2) and potato dextrose agar amended with streptomycin sulfate (1 g/liter) (SPDA) for microscopic identification to taxonomic species.

There were 159 Fusarium cultures obtained from 136 of the 305 crowns sampled and crown grayscale values ranged from 69 to 168. Isolations from the lowest stalk node without rooted brace roots yielded 137 Fusarium cultures from 120 of the 305 plants sampled and node grayscale values ranged from 87 to 193. The most prevalent Fusarium species isolated from crowns and nodes were F. oxysporum and F. verticillioides. Fusarium oxysporum was isolated from both the crown and first stalk node of the same plant 22 times and this occurred 13 times with F. verticillioides.

Plants sampled from commercial fields and found to have Fusarium oxysporum in the crown tissues (+) were compared to plants in which F. oxysporum was not isolated (-) from the crown. T-tests were used to compare ear weights and crown grayscale of the two groups. The same comparisons were made for F. verticillioides (+/-) and any Fusarium species (+/-) recovered from the crown tissues. Similar tests were conducted for recovery of the Fusarium species from stalk nodes of plants and the associated node grayscale values.

Crowns from which F. oxysporum was isolated were significantly darker (lower grayscale values, P ≤ 0.05) (Fig. 3A), indicative of more severe crown rot, and had significantly lower ear weights associated with these plants (Fig. 3C) compared to plants from which no F. oxysporum or any Fusarium species was recovered from crown tissue. When F. verticillioides was isolated from crowns, these tissues were significantly darker compared to crowns from which the fungus was not isolated or compared to plants from which no Fusarium was isolated. Plants with crowns that yielded any species of Fusarium also had significantly darker grayscale values compared to plants from which no Fusarium was isolated from the crown.


 

Fig. 3. Mean crown grayscale (A), mean stalk node grayscale (B), and ear weights (C, D) of sweet corn plants sampled from commercial fields in the Willamette Valley when Fusarium species were detected or absent in the crown (C) or stalk node (D). Foxy = Fusarium oxysporum, present in 80 crowns and 58 nodes of 305 plants sampled, Fvert = F. verticillioides, present in 39 crowns and 46 nodes, All Fusaria = any Fusarium species, present in 136 crowns and 120 nodes. Within each bar graph, * indicates significant differences at P = 0.05 and ** at P = 0.01 between adjacent black and gray bars, and ŧ indicates significant differences at P = 0.05 and ŧŧ at P = 0.01 compared to the "no Fusarium recovered" group ("All Fusaria," gray bar) based on two-sample t-tests. Error bars represent pooled error from a two-sample test comparing adjacent black and gray bars.

 

When F. oxysporum was recovered from the lowest stalk node without rooted brace roots, ear weights were significantly lower (P ≤ 0.05) but there was no difference in stalk node grayscale values (Fig. 3B, 3D) compared to plants from which this fungus was not recovered from stalk nodes. When F. verticillioides was recovered from the lowest stalk node without rooted brace roots, nodes were significantly darker but ear weights were not reduced. Plants from which any Fusarium species was isolated from the stalk node had significantly darker nodes than plants from which no Fusarium was isolated.


Discussion

Decline of sweet corn yields in the Willamette Valley are probably a result of disease symptoms in addition to, or other than, root rot. The results of the studies reported here indicate that there is a decrease in ear weight associated with increasing crown rot, and that crown rot severity explains more ear weight variation than any other disease variable measured. Other variables measured, such as rot of the radicle, sub-crown internode, or nodal root system, were poor predictors of ear weight at harvest. In fact, nodal root rot, which had the most predictive value of the root-based variables, was positively associated with ear weight in most of our models; higher ear weights were associated with increasingly severe root disease.

Potentially there may be a number of different causes of crown or stalk node rot; however, F. oxysporum and F. verticillioides were recovered frequently from symptomatic sweet corn plants in the Willamette Valley. Crowns of plants from which F. verticillioides or F. oxysporum were recovered were more necrotic than crowns from which neither species was detected. The presence of F. oxysporum within crown and stalk node tissues was also associated with lower ear weights. Since nodes are acting as filters for xylem sap (8) and darker nodes are associated with increased resistance to xylem conductance (5), then it is appropriate to investigate organisms that may inhabit the xylem. The results of pathogenicity studies conducted with several Fusarium species indicate that Fusarium can contribute to crown rot and yield decline (5).


Acknowledgments

Financial support was provided Oregon State University and the Oregon Processing Vegetable Commission. We thank Chris Mundt and Paul Severns for discussion and review of the manuscript. We also thank Karrie Cone, Cassandra Vieville, Rebecca Shala, and Chie Takase for excellent technical assistance.


Literature Cited

1. Burnham, K. B., and Anderson, D. R. 2004. Multimodel inference: Understanding AIC and BIC in model selection. Sociol. Methods Res. 33:261-305.

2. Fisher, N. L., Burgess, L. W., Toussoun, T. A., and Nelson, P. E. 1982. Carnation leaves as a substrate and for preserving cultures of Fusarium species. Phytopathology 72:151-153.

3. Hoinacki, E. V. 2003. Sweet corn decline syndrome in Oregon's Willamette Valley. PhD diss. Bot. and Plant Pathol., Oregon State Univ., Corvallis, OR.

4. Kommedahl, T., Stucker, R. E., and Windels, C. E. 1979. Occurrence of Fusarium species in roots and stalks of symptomless corn plants during the growing season. Phytopathology 69:961-966.

5. Miller, N. L. 2007. Responses and relationships among Fusarium species, sweet corn, and western spotted cucumber beetles. PhD diss. Bot. and Plant Pathol., Oregon State Univ., Corvallis, OR.

6. Nash, S. M., and Snyder, W. C. 1962. Quantitative estimations by plate counts of propagules of the bean root rot Fusarium in field soils. Phytopathology 52:567-572.

7. Ramsey, F. L., and Schafer, D. W. 2002. The Statistical Slueth, 2nd Edn. Pacific Grove, Duxbury, CA.

8. Shane, M. W., McCully, M. E., and Canny, M. J. 2000. The vascular system of maize stems revisited: Implications for water transport and xylem safety. Ann. Bot. 86:245-258.