Effects of Fusarium wilt on Cotton Cultivars with and Without Meloidogyne incognita Resistance in Fields

Small plot replicated tests with commercial cotton cultivars were planted in a producer field at one location in Dawson County (Table 1) in 2004 and 2005, where severe Fusarium wilt had developed in 2003. Small plot replicated tests were also planted in three producer sites (Table 1) in 2019 (Gaines, Hall, and Lynn Counties); two sites in 2020 located in Cochran and Hall Counties; and one site (Hall County) in 2021. Each test had between 24 and 48 entries, with four replications per entry, arranged in a randomized complete block design. Plots were two rows wide (1 m centers) and 10.67 m long. A list of Mi-resistant entries in the tests can be found in Table 2. Plots were irrigated with a center pivot system at all locations, though irrigation capacities and yields differed greatly between sites. All sites were naturally infested with Mi and Fov race(s) 1 and/or 2 in the 2019 to 2021 tests. No race determination was made with the Dawson County test site. Fertilizer, irrigation, and other practices were dictated by the producer’s normal management.

Data collected (Table 1 for dates) included plant stand on either one or both rows once plants had emerged, but before Fusarium wilt symptoms began, and plant stand at harvest, or in the case of the 2004 to 2005 trials, once stands were stable (plants had stopped dying) in July or August. Plots were soil sampled in August or later (Table 1) to assay for root-knot nematode. Samples consisted of six cores per plot collected with a narrow-bladed (40 cm depth, 15 cm width at top, and 8 cm width at the bottom) shovel to a depth of 20 cm, close to the taproot. The top 6 cm of soil was discarded and then soil from 6 cm to 20 cm depth, including some roots, was removed. The soil was mixed in a bucket and then a subsample of 1,000 cm³ soil was removed and placed in a plastic bag. The soil samples were refrigerated for <2 weeks before being assayed for root-knot nematode second-stage juveniles (J2) and eggs. The test in 2005 was not sampled for nematodes.

A pie-pan assay with 200 cm³ soil + root fragments was used to extract J2 over 48 hr (Thistlethwayte, 1970). The circular pie-pans were made of glass and wire mesh (0.64 cm diameter) was placed in the pie-pan. Two pieces of facial tissue (2-ply) were laid on top of the mesh and then the soil sample was placed on the facial tissue. Tap water (250 ml) was gently added to the pie-pan without disturbing the soil, and then the wet facial tissues were arranged around the soil to hold it out of the water. A cover was placed over the pie-pan to eliminate evaporation. The extracted J2 were enumerated by concentrating the extracted liquid to 100 ml and then counting a 5 ml aliquot. A second assay with 500 cm³ soil was used to extract root-knot nematode eggs. The soil + root fragments were placed in a bucket with water (combined volume 3 l of water + soil) and stirred for 10 s. After allowing to settle for 15 s, the contents were poured over a sieve with a pore size of 230 μm and the root fragments caught on the sieve were washed into a beaker in 100 ml tap water and mixed on a stir plate for 5 min in NaOCl (0.525%) (Hussey and Barker, 1973). The mixture was poured through a sieve with a pore size of 230 μm, stacked over a sieve with a pore size of 25 μm. The contents from the bottom sieve were rinsed with tap water, washed into a beaker, dyed with acid fuchsin (Byrd et al., 1983), and the eggs were enumerated from a 5 ml aliquot taken out of the 150 ml total volume.

Plants outside of the test area which exhibited signs of Fusarium wilt were collected and Fusarium was isolated from the vascular system. The isolates were single-spored and stored until species and race typing could be performed. DNA extraction from mycelia was performed using Zymo Quick DNA Fungal/Bacterial Miniprep kit (Zymo Research Corp; Irvine, CA). DNA was used to run PCR using four genetic regions, translation elongation factor (EF-1_α), phosphate permease-like protein (PHO), β-tubulin (BT), and intergenic spacer region (IGS). Primer sequences and thermocycler setting followed Cianchetta et al. (2015). PCR products were sequenced using Sanger Sequencing Platform at Molecular Cloning Laboratories. Bioinformatics used GenBank references to identify the isolates using the references used by Cianchetta et al. (2015). MEGA X software (https://www.megasoftware.net/pdfs/kumar_stecher_2018.pdf) was used for alignment and phylogenic analysis using the MUSCLE algorithm interphase (Tamura et al., 2013). For phylogenic analysis, we used neighbor-joining with 2,000 bootstraps, using all sequences in a concatenated approach (Cianchetta et al., 2015).

The plots were mechanically harvested with a cotton stripper designed to weigh the plot yield on load cells. Stripper plot yields consist of lint, seed, and plant debris. A 1,000 g sample was collected from harvested plots and two replications were ginned from each entry to determine lint percentage of the harvest weights.

Plant mortality was calculated as: ((Initial stand – final stand)/Initial stand) × 100. M. incognita density (Mi) was calculated by the number of egg/500 cm³ soil + (2.5 × number of J2/200 cm³ soil). A transformation of Mi density, LMi = LOG₁₀(Mi + 1) was used for analysis. Yields were adjusted to relative yield on a 0 to 1 scale, so that all six trials from 2019 to 2021 could be combined for analysis. Relative yield = (plot lint yield – MinLY)/(MaxLY – MinLY); where MinLY = lint yield (LY) for the plot with the lowest lint yield at a location, and MaxLY = yield for the plot with the highest lint yield at a location. Cultivar was assigned to a category based primarily on its Mi resistance (based on company description) and company origin. The categories for the period 2004 to 2005 were FM-conventional cultivars; FM-transgenic cultivars; Mi partially resistant cultivars; and all other cultivars. In the period 2019 to 2021, the categories were: Mi susceptible (S); Mi-resistant FiberMax (R-FM); “ST 4946GLB2”; “ST 5600B2XF”; Mi-resistant Deltapine (R-DP); and Mi-resistant Phytogen (R-PHY). Groups of cultivars (based on Mi resistance by company) were used in the analyses rather than individual cultivars because Mi-resistant cultivars are often developed from the same source(s) within a company. ST 4946GLB2 was the Fov and partially Mi-resistant check and was included at all tests (2019–2021). ST 5600B2XF, which was not bred by Stoneville® Cotton, and has an unknown lineage, was also in a separate Mi-resistant category.

The tests in 2004 and 2005 contained two variables, cultivar and aldicarb, in a factorial arrangement. All plots that received aldicarb were eliminated from the analysis. The individual cultivar means from 2004 were presented previously (Wheeler and Gannaway, 2005). The Mi density in Hall Co. in 2019 in replication 4 averaged 9 Mi/500 cm³ soil, and thus that replication was deleted from the data set. All other site-years and replications had sufficient Mi density (average >800/500 cm³ soil) to be utilized.

The various groups were analyzed for percentage mortality, LMi, and lint yield within each test site using mixed model analyses (PROC GLIMMIX, SAS version 9.4; SAS Institute, Cary, NC), where group was the fixed variable and replication, or year and replication were the random variables. Significant differences between categories were determined by t-tests, at P < 0.05. Pearson’s correlation coefficient (PROC CORR) was determined for mortality, Mi density, LMi, and lint yield at each site. Significant relationships between mortality and LMi, the quadratic terms, and the interaction term (mortality × LMi), to describe lint yield, was determined for each site using PROC STEPWISE. For a model to be acceptable, all variables were significant at P < 0.05, and then the highest R² value determined the selected model. This procedure also provided the partial R² for each accepted variable. For data sets from the period 2019 to 2021, all data were combined, and analyzed using a mixed model analysis for percentage mortality, LMi, and relative yield. The fixed variable was group, and the random variables were year, site, and replication.

In the trials conducted during 2004 and 2005, the conventional FM group (FM 819, FM 832, FM 958, and FM 966) had higher mortality (69.2%) than the transgenic FM group (58.7%), or Mi-susceptible group (55.7%) (Table 3). The Mi-resistant cultivars (ST LA887 and ST 5599BR) were intermediate (60.3% mortality) and not different from any of the groups. Transformed M. incognita density (LMi) was lower for the resistant group (1.93) than for all other groups (2.90–3.16, Table 3). Lint yield was higher for the Mi-resistant group (1,448 kg/ha) than for the Mi-susceptible group (1,252 kg/ha) and conventional FM group (941 kg/ha) (Table 2). FM-transgenic group lint yield (1,309 kg/ha) was not different from the yields corresponding to Mi-susceptible or Mi-resistant groups. The conventional FM group had lower yields than any other group. Lint yield was correlated with percentage mortality (r = −0.699, P = 0.0001), but not with Mi or LMi. Lint yield (kg/ha) was best fitted with a quadratic model using percentage mortality (Fig. 1; Eq. 1).

Figure 1

(1) Yield = 1431 + (8.86 × M) − (0.189 × M²), where M = % mortality, P = 0.0001, R² = 0.54.

With the data sets from 2019 to 2021, percentage mortality was significantly affected by group only at the Hall Co. site (Table 4). At that site, percentage mortality was higher for R-DP (22.9%) than for all other groups (8.5%–14.9%) except ST 5600B2XF (20.3%). While the other sites did not have significant group differences, the R-DP group numerically had the highest mortality at Cochran and Lynn Counties, though the susceptible group had the highest mortality at the Gaines County site. When percentage mortality was analyzed across all six data sets, R-DP group had higher mortality (28.1%) than all other groups except for ST 5600B2XF (24.8%) (Table 5). The Mi-susceptible group had 23.3% mortality, and ST 4946GLB2 numerically had the lowest mortality at 16.5%.

LMi was significantly affected by cultivar group for all locations (Table 6). The R-PHY group had significantly lower LMi than all other groups in Lynn County, and numerically the lowest density in Hall County. This group was not planted at the Gaines County site, due to a request from the producer to limit non-dicamba tolerant cultivars to one entry (ST 4946GLB2). R-DP had the lowest LMi at Cochran County and significantly lower LMi than the S group at Gaines and Hall Counties. When analyzed across all six trials, LMi was higher for the susceptible group (LMi = 3.22) and R-FM (3.01) than for R-PHY (1.85), R-DP (2.21), and ST 5600B2XF (2.33) (Table 5). ST 4946GLB2 had higher LMi (2.78) than R-DP and R-PHY.

Lint yield was significantly affected by group at all locations (Table 7). At the Cochran County site, R-FM, R-PHY, and ST 4946GLB2 had higher yields than the susceptible cultivars or R-DP. At the Gaines County site, ST 4946GLB2 had higher yields than ST 5600B2XF and susceptible cultivars. At the Hall County site, the susceptible cultivars had lower yields than all other groups except for R-DP. At the Lynn County site, susceptible cultivars yielded less than R-FM, ST 4946GLB2, and R-PHY. Relative yield was highest for ST 4946GLB2 (0.7058), and significantly higher than all groups except R-PHY (0.6351, P = 0.112) (Table 5). The relative yield for R-DP (0.5303) did not differ from the relative yield for the susceptible cultivars (0.4910). Relative yield for R-FM (0.600) and ST 5600B2XF (0.5782) were intermediate and significantly different from both ST 4946GLB2 and susceptible cultivars.

In Cochran County, lint yield was negatively correlated with mortality (r = −0.24, P = 0.003). Mortality was positively correlated with Mi (r = 0.23, P = 0.004) and LMi (r = 0.20, P = 0.015). At Gaines County, lint yield was negatively correlated with mortality (r = −0.68, P = 0.0001), Mi (r = −0.26, P = 0.010), and LMi (r = −0.38, P = 0.001), and mortality was correlated with Mi (r = 0.45, P = 0.0001). In Hall County, lint yield was negatively correlated with Mi (r = −0.21, P = 0.001) and LMi (r = −0.16, P = 0.009). In Lynn County, lint yield was negatively correlated with LMi (r = −0.27, P = 0.0001).

In Cochran County, lint yield (kg/ha) was fitted to a quadratic term for percentage mortality and a linear term with LMi (Fig. 2; Eq. 2), mortality² had a partial R² = 0.072, and LMi had a partial R² = 0.033.

Figure 2

2) Yield = 835 − (0.0847 × M²) + (46.3 × LMi ), P = 0.0001, R² = 0.11.

In Gaines County, Lint yield (kg/ha) was fitted to percentage mortality (Fig. 2; Eq. 3):

3) Yield = 460 − (4.82 × M), P = 0.0001, R² = 0.45.

Fusarium wilt was most severe in this Gaines County test, and even though the irrigation was terminated prematurely (presumably due to severe disease and subsequent low yield potential), the percentage mortality explained three to six times more of the variation in yield than for the other tests.

In Hall County, Lint yield (kg/ha) was fitted to a quadratic model with percentage mortality (Fig. 2; Eq. 4):

4) Yield = 1924 + (24.8 × M) − (0.596 × M²⁾, P = 0.0001, R² = 0.15.

In Lynn County, Lint yield (kg/ha) was described only by LMi (Fig. 2; Eq. 5)

5) Yield = 994 - (44.5 × LMi ), P = 0.025, R² = 0.07.

Two races of Fov were identified across all locations from 2019 to 2021. In 2019, the Gaines County site isolate was molecularly characterized as Fov race 2, while Lynn and Hall County sites were Fov race 1. In 2020, races 1 and 2 both occurred in the same field at the Cochran County site. For all 3 yr, only race 1 was found at Hall County.

Tolerance and/or resistance to Fusarium wilt (referring only to those races that require M. incognita assistance) has been improved in cotton for many years (Kappelman, 1980, 1982; Zhang et al., 2015). The severe Fusarium wilt problem that occurred in 2003 appeared to be the result of multiple years of planting conventional FiberMax cultivars. Conventional FiberMax cultivars, which were developed in Australia, were introduced into the U.S. around 1999 (Anonymous, 1999). The first observed cases of Fusarium wilt in Australia occurred in 1993 (Kochman, 1995) and isolates of Fov were identified as something unique to Australia and were not races 1 and 2 (Davis et al., 1996). This would mean there was no selection pressure by Fov races 1 and 2 on the germplasm used in developing the conventional FiberMax cultivars grown in the U.S. While these cultivars may have been highly susceptible to Fov, the unusually high mortality for all tested cultivars in 2004 and 2005 suggest that these conventional FiberMax cultivars were also responsible for increasing soil densities of Fov and Mi to levels higher than normal. Fusarium wilt severity is a function of both Mi and Fov inoculum density (Garber et al., 1979; Starr et al., 1989; DeVay et al., 1997). Chawla (2011) planted the Mi-susceptible “FM 9058F” (PVP 200700206) and “ST 4554B2RF” (PVP 200700046) in microplots and sampled the soil for Fov densities for 3 yr. The soil densities of Fov were similar at the start of the experiment (4.6 and 4.4 colony forming units (CFU) × 105/cm³ soil for FM 9058F and ST 4554B2RF, respectively), but much higher for FM 9058F than for ST 4554B2RF after 24 mon (9.5 × 105versus 3.8 × 10⁵ CFU/cm³ soil, respectively). Fusarium wilt incidence for the first, second, and third growing seasons averaged 17.9%, 33.9%, and 69.0% for FM 9058F, and 5.6%, 5.9%, and 4.3% for ST 4554B2RF, respectively. Thus, cotton cultivars can differ both in susceptibility (stand loss, vascular discoloration, and yield loss), and in ability of Fov to reproduce and buildup in the soil. The percentage mortality for all other cultivars tested in 2004 and 2005, while lower than the conventional FiberMax cultivars, were still very high (average >50% mortality) but decreased rapidly in that field when the producer switched to cultivars other than conventional FiberMax for several years (T. Wheeler, personal observations from 4 yr of cultivar trials at that site).

The catastrophic Fusarium wilt problems observed by several producers after the introduction of some M. incognita resistant cultivars in more recent years, was the catalyst for the 2019 to 2021 cultivar trials, and included the exact site in Gaines County where one report occurred. The Gaines County site did have the highest overall Fusarium wilt percentage mortality of all the tested sites (2019–2021); however, there were no statistical differences in mortality between the cultivar groups, indicating no high Fov resistance, even in ST 4946GLB2. This is contrasted with the significant group differences found in all tests with regard to transformed Mi density. The difficulty in testing cultivars for Fusarium wilt mortality in the field is shown in Figure 3 at the Gaines County test site. The spatial variability of the disease, even when Fusarium wilt is relatively severe, resulted in mostly non-significant differences for percentage mortality. The Hall site did have significant mortality differences between groups, but there were also 3 yr of data that could be combined at that site.

Figure 3

The R-DP cultivars possessed more susceptibility to Fusarium wilt than did ST 4946GLB2, R-PHY, R-FM and Mi-susceptible cultivars, even though R-DP cultivars had excellent Mi resistance. ST 4946GLB2, which was thought to be the most Fusarium wilt tolerant cultivar at the start of the 2019 to 2021 trials, did indeed have the lowest percentage mortality and highest overall lint yields in Fusarium wilt/Mi trials, but was not statistically superior to other groups except for R-DP for Fusarium wilt mortality. With regard to Mi resistance, significant separations could be seen between groups, particularly for the R-DP and R-PHY groups compared with more Mi-susceptible groups.

The original source of resistance for Mi in many cotton breeding programs was Auburn 623RNR, which was released by Shepherd (1974). This line was the most Mi and Fov resistant line available at that time in G. hirsutum (Shepherd, 1974). Its resistance to both these organisms greatly surpassed the resistance of Auburn 56, which until then was considered one of the most Fov and Mi-resistant varieties. There are two genes (located on chromosome 11 and 14) associated with this high Mi resistance. Gaudin and Wubben (2021) screened G. hirsutum accessions for resistance to Mi, and while there was a range of resistant phenotypes, the genotypic analyses revealed that all resistant accessions carried either the chromosome 11 (RK1) and/or chromosome 14 (RK2) resistance QTL. It is suspected that the high levels of resistance found in some cultivars (PHY 480W3FE, DP 2141NR B3XF, and DP 2143NR B3XF, as examples) have both Mi resistance genes, homogeneously, but these cultivars do not currently have PVP certificates available. Two other cultivars from Phytogen used in these trials were PHY 332 W3FE (PVP 202000220) and PHY 443 W3FE (PVP 202000221), and molecular markers confirmed they both have the RK1 and RK2 genes homogeneously (based on their PVP certificates). The Mi-resistant gene(s?) in partially resistant ST 4946GLB2 and R-FM group are also presumed as RK1 or RK2 genes (presumed from the work of Gaudin and Wubben [2021]) but may be present heterogeneously given the higher densities of transformed Mi found in the Fov/ Mi trials.

Mi-resistant gene(s) do not by themselves confer resistance to Fov, or there are Mi-susceptible cultivars that may have resistance to Fov (Hyer et al., 1979; Wang et al., 2009; Ulloa et al., 2011). The resistance to Fov observed in Auburn 623RNR was not simply a product of having both the chromosome 11 and chromosome 14 Mi-resistant genes, since this cultivar exhibited high resistance to both Mi and Fov (Shepherd, 1974). Marker-assisted selections have been useful for development of Mi-resistant cotton varieties. However, there is no indication within the development of U.S. cotton varieties (based on PVPs) that molecular markers are being utilized to identify Fov resistance. There have been several studies to determine the genes involved with Fov race 1 resistance in G. hirsutum and potential location of molecular markers (Wang et al., 2009; Ulloa et al., 2011). However, it is important that Fov (race 1 and 2) resistance genes function in the presence of Mi, since these races cause minimal losses in the absence of Mi. Development of markers for Fov (race 1 and 2) resistance and combined utilization of Mi-resistant and Fov-resistant markers would accelerate the development of Fusarium wilt resistant cultivars (to races 1 and 2).