Identification and virulence of Chryseobacterium indologenes isolated from diseased yellow perch (Perca flavescens)

Authors


Correspondence

Julia W. Pridgeon, Aquatic Animal Health Research Unit, USDA-ARS, 990 Wire Road, Auburn, AL 36832, USA. E-mail: julia.pridgeon@ars.usda.gov

Abstract

Aim

To identify pathogen of diseased yellow perch and determine their virulence.

Methods and Results

Fifteen Gram-negative bacterial isolates were recovered from the skin lesions of diseased yellow perch (Perca flavescens). Based on API 20NE test, ten isolates were found to share 67·2–99·9% homologies with Chryseobactertium indologenes. Based on fatty acid methyl ester analysis, 13 isolates were found to share similarities with C. indologenes and other species of Chryseobacterium. Based on sequencing results of partial 16S rRNA gene, 13 isolates shared 99% identities (e value = 2e−50) with the 16S rRNA sequence of C. indologenes (GenBank HQ259684). Based on the 16S–23S rRNA intergenic spacer region (ISR) sequence, the 13 isolates shared 88% identity (e value = 1e−165) with the 16S–23S ISR sequence of C. indologenes (GenBank EU014570). T-coffee multiple sequence alignment revealed that the partial 16S rRNA or the 16S–23S ISR sequence of the 13 isolates shared 100% identity with each other. When healthy yellow perch were exposed to the 15 isolates by bath immersion (c. 6 × 107 CFU ml−1 for 1 h), only C. indologenes isolates killed 10–20% of fish, whereas other isolates were avirulent. When yellow perch were exposed to C. indologenes by intraperitoneal injection, mortality was dose dependent, with LD50 and LD95 values of 1·5 × 108 and 3·2 × 108 CFU per fish, respectively.

Conclusions

Chryseobactertium indologenes could be pathogenic to yellow perch.

Significance and Impact of the Study

This is the first report on the isolation of C. indologenes from diseased yellow perch. Virulence studies suggested that C. indologenes could become pathogenic to yellow perch.

Introduction

The genus Chryseobacterium belongs to the family of Flavobacteriaceae. Several bacteria formally classified in the genus Flavobacterium were reclassified within the genus Chryseobacterium according to Vandamme et al. (1994), including Flavobacterium balustinum, Flavobacterium gleum, Flavobacterium indologenes, Flavobacterium indoltheticum, Flavobacterium meningosepticum and Flavobacterium scophthalmum. Members of the genus Chryseobacterium are widely distributed in various environments, including soil (Benmalek et al. 2010), plants (Cho et al. 2010), fresh water (Kim et al. 2008; Park et al. 2008), marine environment (de Beer et al. 2006), wastewater (Yoon et al. 2007) and food sources, such as milk (Hantsis-Zacharov et al. 2008) and chicken (de Beer et al. 2005). Several species of Chryseobacterium have been isolated from diseased fish. For example, Chryseobacterium scophthalmum was isolated from diseased turbot (Scophthalmus maximus) in Scotland (Mudarris et al. 1994); Chryseobacterium shigense was isolated from rainbow trout (Oncorhynchus mykiss) during disease outbreaks in Spain (Zamora et al. 2012); Chryseobacterium piscicola was reported to be responsible for mortality in farmed Atlantic salmon (Salmo salar) in Chile (Ilardi and Avendaño-Herrera 2008; Ilardi et al. 2009) and Finland (Ilardi et al. 2010). In addition, increasing incidents of Chryseobacterium infections have been reported in hospital environments (Hsueh et al. 1997; Bhuyar et al. 2012).

In January of 2012, mortality among yellow perch was observed in a US fish farm, with symptoms of gross skin lesions below the dorsal fin as well as the caudal peduncle. Diseased fish were sent to Aquatic Animal Health Research Unit, USDA-ARS for pathogen identification. After culturing the skin lesions on Shieh agar plates supplemented with tobramycin (Decostere et al. 1997), majority of the colonies showed yellow pigments resembling the appearance of Flavobacterium, whereas a few of them had no pigment (white colonies). A total of 15 colonies (yellow and white) were then randomly picked from the plates and subjected to bacterial identification. However, at that stage, it was unknown whether the bacteria were pathogenic to yellow perch. Therefore, the objectives of this study were (i) to conduct biochemical identifications of the 15 isolates cultured from the skin lesions of the diseased yellow perch; (ii) to determine whether any of these colonies could cause mortality to yellow perch by bath immersion; (iii) to perform molecular identifications for the isolates that were able to cause mortality among yellow perch by bath immersion; and (iv) to determine the virulence (LD50 value) of the isolate to yellow perch by intraperitoneal injection.

Materials and methods

Bacterial isolates and culture conditions

After mortality was observed in a US yellow perch farm in January 2012, four diseased moribund yellow perch (Perca flavescens) were collected, freshly frozen and sent to the Aquatic Animal Health Research Laboratory, USDA, ARS within 24 h on ice. Fifteen bacterial isolates were recovered from skin lesions of the four diseased fish. Skin lesions on diseased fish were photographed. Based on the symptoms of gross skin lesions below the dorsal fin as well as the caudal peduncle, columnaris disease was suspected. Skin lesion samples were washed in sterile phosphate-buffered saline three times to minimize contamination. A sterile loop was then used to inoculate the washed skin lesion samples onto Shieh agar plates containing 1 μg ml−1 of tobramycin according to Decostere et al. (1997). After overnight growth at 19 ± 1°C, majority of the colonies on the culture plates appeared yellowish, resembling Flavobacterium columnare, whereas a few colonies grew on the plates appeared white. A total of 15 colonies (13 yellow and 2 white) were then randomly picked and restreaked onto Shieh agar plates containing 1 μg ml−1 of tobramycin to obtain single colonies. Single-colony bacterial cultures grown in modified Shieh medium were then subjected to biochemical and molecular identification. Glycerol stock (10% glycerol) of each isolate was prepared in modified Shieh medium and stored at −80 ⁰C.

Gram staining, oxidase test, catalase test and API 20 NE test

Gram staining was performed with Gram staining kit and reagents followed the instruction of the manufacturer (Becton Dickinson, Franklin Lakes, NJ, USA). Oxidase test was performed by adding bacterial smear to filter paper containing BactiDrop oxidase reagent (Remel, Lenexa, KS, USA). Colour development was observed within 1 min. Catalase test was performed by adding a drop of hydrogen peroxide (Fisher Scientific, Pittsburgh, PA, USA) to a slide that contained live bacterial smear. API 20 NE bacterial identification was performed according to manufacturer's instruction (bioMérieux, Durham, NC, USA).

Microbial identification using fatty acid methyl ester profiling

Fatty acid methyl ester (FAME) profiling was performed according to manufacturer's instruction (MIDI Labs, Newark, DE, USA). Briefly, overnight bacterial cultures (25–30 mg) were transferred to 13 × 100 mm glass tubes. Bacterial cells were saponified in 3·75 N NaOH in 50% methanol for 30 min, followed by methylation with 3·25 N HCl in methanol for 10 min. FAMEs were then mixed with hexane and methyl tert-butyl ether (1 : 1) for 10 min. After brief centrifugation, the top phase was collected and mixed with 0·3 N NaOH to remove any free fatty acids and residual extraction solvent. After brief centrifugation, the top phase was removed and subjected to gas chromatography using Agilent 6850 GC system (Agilent Technology, Santa Clara, CA, USA). FAME profiles were then compared to FAME profiles deposited in the standard RCLN50 library, and the similarity indices were then calculated by Sherlock Library Search (MIDI, Newark, DE, USA). An exact match of the fatty acid make-up of an unknown sample with that of a known library entry would result in a similarity index of 1·000. Samples with a similarity of 0·500 or higher with a separation of 0·100 between the first and second choice are considered good library comparison.

Virulence of the 15 isolates to yellow perch by bath immersion

To study the virulence of the 15 bacterial isolates to yellow perch, isolates were cultured overnight in Shieh medium supplemented with 1 μg ml−1 of tobramycin at 19 ± 1°C. The concentration (CFU ml−1) of each isolate was determined through serial dilutions. Briefly, optical density (OD) of the bacterial cultures was measured at 540 nm using a Thermo Spectronic spectrophotometer (Fisher Scientific, Pittsburgh, PA, USA) and adjusted to OD = 1. Serial dilutions were prepared in Shieh medium, and 100 μl of diluted bacterial culture was plated onto Shieh agar plates. After 24-h incubation at 19 ± 1°C, the average number of CFU ml−1 was calculated for each isolate. Healthy yellow perch (2 ± 0·5 g) were obtained from a US yellow perch supplier and acclimated for 30 days prior to challenge. The health status (culture negative) of the yellow perch prior to challenge was evaluated by randomly sampling of selected fish and culturing skin and kidney tissue on Shieh agar plates. Acclimated fish were maintained in 57 l glass aquaria (10 fish per tank) with flow-through dechlorinated tap water and constant aeration with water temperature at 19 ± 1°C. The flow-through water was at a rate of 0·5 l min−1. Dissolved oxygen range was 6 to 8 mg l−1. A 12 : 12 h light/dark schedule was maintained. Fish were fed daily with commercial Aquamax Grower (Brentwood, MO, USA) at 4% of their body weight. For bath immersion, bacteria at beginning concentration of approximately 6 × 107 CFU ml−1 (100 ml of OD540 nm = 1·0 culture was added to 900 ml of water) was used for duration of 1 h in 2 l beakers. After the 1-h immersion, an aliquot of water sample was used to check the viability and the concentration of the bacteria. After the 1-h immersion, fish were returned to 57 l glass aquaria (10 fish per tank, 3 tanks per isolate). Mortalities were recorded daily for 21 days postexposure. For 13 Chryseobacterium indologenes isolates identified by FAME and sequencing, bath immersion was not repeated because these isolates shared 100% identity with each other at the 16S-23S rRNA intergenic spacer region (ISR) region. For non-Chryseobacterium isolates, bath immersions were repeated three times. The anterior kidney of dead fish was cultured on Shieh agar plates for bacterial isolation followed by API 20 NE test. Mean cumulative mortality data were analysed by analysis of variance (anova) using SigmaStat statistical analysis software (Systat Software, San Jose, CA, USA).

Primer design, polymerase chain reaction and sequencing of 16S-23S rRNA ISR

Based on mortality results, only yellow-pigmented bacterial isolates identified as Chryseobacterium spp. showed virulence to yellow perch, whereas other white-pigmented isolates were avirulent to yellow perch. To determine whether these Chryseobacterium isolates were identical, a forward primer Chry-16S-F (5′-GCATCAGCCATGGCGCGGTG-3′) located at 1327 to 1347 bp of the complete 16S rRNA sequence of Chryseobacterium sp. (GenBank accession no. JQ740255) was designed based on the fact that it was conserved among different species of Chryseobacterium. Similarly, a reverse primer Chry-23S-R (5′-CCTTAACGATTTCTTTCCTA-3′) located at 807–787 bp of the 23S sequence of C. piscicola (GenBank accession no. HM131451) was also designed. Polymerase chain reaction (PCR) was performed in a 25 μl mixture consisting of 12·5 μl of 2 ×  Taq PCR Master Mix (Qiagen, Valencia, CA, USA), 9·5 μl of nuclease-free H2O, 1 μl of genomic DNA (10 ng ul−1), 1 μl of forward primer (5 umol l−1) and 1 μl of reverse primer (5 umol l−1). PCR was carried out in a Biometra T Gradient thermocycler (Biometra, Goettingen, Germany). PCR conditions consisted of an initial denaturation step at 94°C for 5 min followed by 35 cycles of 45 s at 94°C, 45 s at 53°C and 2 min at 72°C, followed by a final extension of 10 min at 72°C. PCR products were analysed by 1% agarose gel by electrophoresis. Individual single bands were excised and purified using gel purification kit according to the manufacturer's protocol (Qiagen). Purified PCR products were sequenced using Chry-16S-F primer at the USDA-ARS Mid South Genomic Laboratory (Stoneville, MS, USA) on an ABI 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).

Sequence analysis

Sequences were analysed using the National Center for Biotechnology Information (NCBI) Blast program to search for sequence identities. Multiple sequence alignment for partial 16S rRNA and 16S-23S ISR sequences was performed using T-coffee method (http://tcoffee.crg.cat/apps/tcoffee/do:regular) (Notredame et al. 2000; Di Tommaso et al. 2011).

Determination of LD50 of Chryseobactertium indologenes to yellow perch by intraperitoneal injection

BlastN analysis revealed that all Chryseobacterium isolates shared 99% identities with the 16S rRNA sequence of C. indologenes strain B7 (GenBank HQ259684). T-coffee multiple sequence alignment revealed that the partial 16S rRNA or 16S-23S ISR sequence of the 13 isolates shared 100% identities with each other. T-coffee multiple sequence alignment analysis revealed that all isolates of Chryseobacterium shared 100% homology with each other. Therefore, only a single isolate (isolate 4) was chosen to determine the LD50 value of C. indologenes to healthy yellow perch by intraperitoneal injection. After overnight incubation of isolate 4 at 19 ± 1°C, the average number (CFU ml−1) was calculated. Five different doses (ranging from 6 × 107 to 3 × 108 CFU per fish) that killed 0–100% fish were administered to anaesthetized (10 mg l−1 of tricaine methanesulphonate) yellow perch (2 ± 0·5 g) through intraperitoneal injection. The experiments were repeated three times. After exposing yellow perch to isolate 4, mortalities were recorded daily for 21 days postexposure. The presence of C. indologenes in dead fish was determined by culturing anterior kidney samples on Shieh agar plates followed by API 20 NE biochemical test. LD50 values were calculated using PoloPlus probit and logit analysis software (LeOra Software, Petaluma, CA, USA).

Results

Collection of bacterial isolates from diseased yellow perch

The lesions of the diseased yellow perch appeared yellowish, with very strong smell. Diseased yellow perch had symptoms of gross skin lesions below the dorsal fin as well as the caudal peduncle (Fig. 1). Majority (13/15) of the colonies appeared yellowish, whereas two colonies (isolate 8 and 9) were white.

Figure 1.

Skin lesions of diseased yellow perch, Perca flavescens. (a) Whole body picture; (b) skin lesions under the dorsal fin; (c) skin lesions at the caudal peduncle area.

Gram staining, oxidase test, catalase test and API 20 NE test

All isolates comprised Gram-negative catalase and oxidase-positive rods/cells that produced β-glucosidase hydrolysis (esculin ferric citrate/ESC as substrate) and protease hydrolysis (gelatin/GEL as substrate). Based on API 20NE test results, ten isolates were found to share 67·2% to 99·9% homologies with C. indologenes. Of the 15 isolates, ten were found to share identities with C. indologenes. Three were found to share similarities with Empedobacter brevis.

FAME analysis of the bacterial isolates

FAME analysis revealed that 13 of the 15 isolates shared similar profiles with that of Chryseobacterium species deposited at the RCLN50 database, with first choice similarity indices ranging from 0·246 to 0·395. The 13 isolates had four major fatty acid peaks (Fig. 2a). The biggest peak was 15:0 iso fatty acid at retention time of 2·325 min, which accounted for 40% of the total FAME response. The second biggest peak was summed feature 4 (10 methyl 16:0 and 17:1 iso w9c) at retention time of 2·892 min, which was 28% of the total response. The other two peaks that shared similar percentage (12%) of total response were summed feature 3 (16:1 w7c and 16:1 w6c at retention time of 2·713 min) and 17:0 iso 3OH (at retention time of 3·442 min) (Fig. 2a). Fatty acid profile of isolate 8 had eight major peaks (Fig. 2b), whereas that of isolate 9 had 14 major peaks (Fig. 2c). Both isolate 8 and isolate 9 shared some similarities with Aeromons veronii, with similarity indices of 0·285 and 0·474, respectively.

Figure 2.

Fatty acid methyl ester (FAME) profiles of bacterial isolates cultured from skin lesions of diseased yellow perch. (a) Representative FAME profile of the 13 isolates identified as Chryseobacterium species; (b) FAME profile of isolate 8; (c) FAME profile of isolate 9.

Virulence of the bacterial isolates to yellow perch by bath immersion

Virulence of the 15 bacterial isolates to yellow perch by bath immersion is summarized in Table 1. At the end of the 1-h immersion, water samples were culture positive as input bacteria (Chryseobacterium or Aeromonas) and the concentrations were slightly smaller (4 × 107 CFU ml−1) than the input concentration (6 × 107 CFU ml−1). The isolates that killed 10% or 20% yellow perch were Chryseobacterium. Dead fish were all culture positive on Shieh agar plates. Colonies collected from the posterior kidney of freshly dead yellow perch after bath immersion with isolate 4 were identified as C. indologenes by FAME analysis, with similar index of 0·701. Neither isolate 8 nor isolate 9 was found to be virulent to yellow perch by bath immersion (Table 1).

Table 1. Virulence of the 15 isolates to yellow perch by bath immersion
Isolate no.Immersion concentration (CFU ml−1)Mortality (%)
Medium control00
16·7 × 10710
27·7 × 10710
37·4 × 10710
43·4 × 10720
54·0 × 1070
63·7 × 10710
74·7 × 1070
86·9 × 1070
98·6 × 1070
107·2 × 1070
115·4 × 10710
129·2 × 1070
139·4 × 1070
148·6 × 1070
159·2 × 1070

PCR and sequencing results

Using Chry-16S-F and Chry-23S-R primers and gDNA of the 15 isolates as templates, a single sharp band of c. 800–1000 bp (Fig. 3) showed up in all 13 isolates identified as Chryseobacterium species by FAME. Based on sequencing results of partial 16S rRNA gene, 13 isolates shared 99% identities (e value = 2e−50) with the region from 1352 to 1469 bp of the 16S rRNA sequence of C. indologenes strain B7 (GenBank HQ259684), with only one bp deletion at the 1364 bp of the sequence of HQ259684 (Fig. 4). However, the 13 isolates shared 100% identities with each other (Fig. 4). Based on 16S-23S rRNA ISR sequence, the 13 isolates shared 88% identities (430/489, e value = 1e−165) with the 16S-23S ISR sequence of C. indologenes strain ATCC 29897 (GenBank EU014570). T-coffee multiple sequence alignment revealed that the 16S-23S ISR sequences of the 13 isolates shared 100% identities with each other. The sequence of 860 bp of the partial 16S and ISR of these C. indologenes was deposited at GenBank with accession no. JX649950.

Figure 3.

Electrophoresis of polymerase chain reaction (PCR) products on 1% agarose gel. M: 100 bp DNA marker. Lane 1–15 were PCR products using genomic DNA from isolate 1–15 as template, respectively.

Figure 4.

T-coffee multiple sequence alignment of the partial 16S RNA of the 13 Chryseobacterium isolates with Chryseobactertium indologenes strain B7 (HQ259684).

LD50 of Chryseobactertium indologenes to yellow perch by intraperitoneal injection

Based on sequencing results and T-coffee multiple sequence alignment, the 13 isolates shared 100% identities with each other at the partial 16S rRNA region or the 16S-23S ISR region of C. indologenes, suggesting that all 13 isolates represented essentially the same species. Therefore, only one isolate (isolate 4) was chosen for the injection study. The LD50 value of isolate 4 to yellow perch (2 ± 0·5 g) was 1·5 × 108 CFU per fish, with 95% confidence intervals ranging from 1·2 × 108 to 1·8 × 108 CFU per fish. The LD95 value of isolate 4 to yellow perch (2 ± 0·5 g) was 3·2 × 108 CFU per fish, with 95% confidence intervals ranging from 2·5 × 108 to 5·1 × 108 CFU per fish. After yellow perch were exposed to C. indologenes by intraperitoneal injections, the fish showed a loss of appetite and balance. Another obvious sign presented in diseased fish injected with C. indologenes was a swelling above the anal fin, which was also observed in diseased fish collected from the farm (Fig. 1).

Discussion

Fifteen Gram-negative bacteria isolates were cultured from the skin lesions of yellow perch collected from a disease outbreak in January 2012. Based on API 20 NE bacterial identification test, ten isolates were found to share identities with C. indologenes. FAME analysis revealed that 13 isolates shared similarities with species of Chryseobacterium. Sequencing results of partial 16S rRNA gene or the 16S-23S ISR region revealed that all 13 Chryseobacterium isolates shared 100% identities with each other. These results suggest that both API and FAME could be used to identify bacteria at genus level. However, to have an exact speciation, molecular identification using sequence information might be necessary when API and FAME results are not consistent.

Virulence study by bath immersion revealed that some C. indologenes could kill 10–20% yellow perch at the concentration of c. 6 × 107 CFU ml−1 for 1 h under normal feeding condition. However, some C. indologenes isolates failed to kill any fish at the immersion condition we used in the study. Therefore, it is currently unclear whether higher concentrations and/or prolonged exposure might increase the mortality. It was reported that a prolonged exposure (18 h) of Atlantic salmon to both C. piscicola and Flavobacterium psychrophilum at a low concentration of 2·5 × 106 CFU ml−1 killed 100% fish at 28 days postexposure (Ilardi and Avendaño-Herrera 2008), although neither C. piscicola nor Fl. psychrophilum itself killed any fish at concentration of 5 × 106 CFU ml−1. In the summer of 2012, a disease outbreak occurred again in that farm. Sequencing results of 16S rRNA revealed that all diseased yellow perch during the summer outbreak had Fl. columnare (T. Welch, NCCCWA, USDA-ARS, Leetown, VA, personal communication). Whether a combination of Fl. columnare and C. indologenes will kill more yellow perch merits further study.

At a high concentration of 6 × 107 CFU ml−1 after bath immersion for 1 h, the C. indologenes isolates only killed 10–20% yellow perch. It was previously reported that the presence of extracellular products (ECPs) in the old culture media could inhibit the virulence of Aeromonas hydrophila to channel catfish (Pridgeon and Klesius 2011). At a concentration of 1·65 × 108 CFU ml−1, a highly virulent Aer. hydrophila strain AL09-73 without its ECPs killed 100% of catfish fingerlings within 2 h by bath immersion (Pridgeon and Klesius 2011). However, in this present study, when the old culture media of C. indologenes was replaced with fresh media, the virulence of C. indologenes was not affected (data not shown). The LD50 and LD95 values of C. indologenes isolate 4 to yellow perch by intraperitoneal injection were 1·5 × 108 and 3·2 × 108 CFU per fish, respectively. We previously (Pridgeon and Klesius 2011) reported that the LD50 and LD95 values of Aer. hydrophila isolate AL09-73 to channel catfish (4·6 ± 1·3 g) by intraperitoneal injection were 2 × 102 and 9·4 × 102 CFU per fish, respectively. Taken together, these results suggested that C. indologenes isolated from the diseased yellow perch was not highly virulent by it alone. A Finland isolate of C. piscicola was reported to be only moderately virulent to Atlantic salmon (Ilardi et al. 2010). However, species of Chryseobacterium are consistently isolated from diseased fish in different countries, including C. scophthalmum from diseased turbot in Scotland (Mudarris et al. 1994), C. shigense from diseased rainbow trout in Spain (Zamora et al. 2012) and C. piscicola from diseased Atlantic salmon in Chile (Ilardi and Avendaño-Herrera 2008; Ilardi et al. 2009), suggesting that Chryseobacterium species could become pathogenic to fish under certain conditions.

In summary, 15 Gram-negative bacterial isolates were recovered from the skin lesions of diseased yellow perch. Biochemical API 20NE tests revealed that ten isolates shared homologies with C. indologenes. FAME analysis revealed that 13 isolates shared similarities with species of Chryseobacterium. Sequencing results revealed that all 13 Chryseobacterium isolates identified by FAME shared 99% identities with the 16S rRNA sequence of C. indologenes strain B7 (GenBank HQ259684). Based on 16S-23S rRNA ISR sequence, the 13 isolates only shared 88% identities with the 16S-23S ISR sequence of C. indologenes strain ATCC 29897 (GenBank EU014570). T-coffee multiple sequence alignment revealed that the partial 16S rRNA or 16S-23S ISR sequence of the 13 isolates shared 100% identities with each other. When yellow perch were exposed to the 15 isolates by bath immersion at concentration of c. 6 × 107 CFU ml−1 for 1 h, only C. indologenes isolates killed 10% to 20% of fish, whereas other isolates were avirulent. When yellow perch was exposed to C. indologenes by intraperitoneal injection, mortality was dose dependent, with LD50 and LD95 values of 1·5 × 108 and 3·2 × 108 CFU per fish, respectively. This is the first report on the isolation of C. indologenes from diseased yellow perch. Virulence studies suggest that C. indologenes could become pathogenic to yellow perch.

Acknowledgements

We thank Drs Dunhua Zhang (USDA-ARS) and Victor Panangala (USDA collaborator) for critical reviews of the manuscript. We thank Dr Brian Scheffler and Fanny Liu (USDA-ARS-Catfish Genetics Research Unit) for their excellent sequencing work. We thank Beth Peterman (USDA-ARS) for her excellent technical support. This study was supported by the USDA/ARS CRIS project no. 6420-32000-024-00D. The use of trade, firm or corporate names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement or approval by the United States Department of Agriculture or the Agricultural Research Service of any product or service to the exclusion of others that may be suitable.

Ancillary