Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 23 February 2005
Published in Crop Sci 45:778-783 (2005)
© 2005 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Agricola
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Related Collections
Right arrow Other Forage Crops

Determination of Ergot Alkaloid Content in Tall Fescue by Near-Infrared Spectroscopy

C. A. Robertsa,*, H. R. Benedicta, N. S. Hillb, R. L. Kallenbacha and G. E. Rottinghausc

a Dep. of Agron., Univ. of Missouri, Columbia, MO 65211
b Dep. of Crop and Soil Sci., Univ. of Georgia, Athens, GA 30602
c College of Veterinary Medicine, Univ. of Missouri, Columbia, MO 65211

* Corresponding author (RobertsCr{at}missouri.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Ergot alkaloids are a class of toxic compounds, some of which are produced by a fungal endophyte in tall fescue (Festuca arundinacea Schreb.). The objective of this research was to measure total ergot alkaloid content in tall fescue by near-infrared (NIR) spectroscopy with a calibration developed from immunochemical reference data. Eighty-four tall fescue samples were collected from experiments at the University of Missouri. Samples were from plants that varied in maturity, endophyte status, genotype, growing environment, and storage and preservation treatment. Samples were scanned by NIR radiation and analyzed chemically for ergot alkaloid content using a commercial immunoassay. An empirical prediction equation was developed by regressing NIR reflectance data against absorbance data from the immunoassay. The initial calibrations achieved a 1 – variance ratio (VR) of 0.86 with a mean and standard error of cross validation (SECV) of 0.606 ± 0.12; however, this was possible only when stockpiled tall fescue samples were omitted from the population. Stockpiled samples were checked for chemical accuracy and spectral similarity to the population, and another set of calibrations was developed. The final calibration, which omitted only those stockpiled samples that were infected with a toxic endophyte, had a 1 – VR of 0.89, with a mean and SECV of 0.682 ± 0.11. We concluded that total ergot alkaloid content in tall fescue can be quantified by NIR spectroscopy with a calibration developed from immunochemcial data, though it may not be possible to include samples of stockpiled tall fescue if they are infected with a toxic endophyte. Such a calibration can be robust and precise, reliably predicting an entire class of compounds in a diverse population of tall fescue samples.

Abbreviations: CV, coefficient of variation • HPLC, high performance liquid chromatography • NIR, near infrared • SECV, standard error of cross validation • VR, variance ratio


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
TALL FESCUE is one of the most widely grown pasture grasses in the USA (Sleper and West, 1996). It has a long growing season, is adapted to a wide geographical region, and persists in spite of biotic and abiotic stresses. However, tall fescue fields are infected with the fungal endophyte Neotyphodium coenophialum (Morgan-Jones and Gams) Glenn, Bacon, and Hanlin comb. nov. (Glenn et al., 1996). Most strains of this endophyte produce ergot alkaloids, a class of mycotoxins that causes serious livestock disorders in cattle, sheep, and horses (Sleper and West, 1996).

New cultivars of tall fescue contain patented strains of Neotyphodium that produce little or no ergot alkaloids (Bouton et al., 2002); the strains are introduced into endophyte-free plants to maintain persistence through biotic and abiotic stresses. Despite their benefits to livestock, cultivars with introduced endophytes will not replace cultivars with toxic endophytes for many years (Roberts and Andrae, 2005). In fact, many fields of tall fescue will probably never be replaced because of complex pasture renovation protocols, rugged terrain, and cost of replacement. It is important, therefore, that researchers continue advancing the understanding of tall fescue toxicosis while developing new management practices and technologies (Roberts and Andrae, 2004).

One technology in constant need of advancement is quantitative methodology for ergot alkaloids. Ergot alkaloids have been quantified by immunochemical procedures (Hill et al., 1994; Shelby and Kelley, 1992). In addition, ergovaline, the primary alkaloid among the ergot alkaloids produced by Neotyphodium, has been quantified by high performance liquid chromatography (HPLC) (Rottinghaus et al., 1991; Hill et al., 1993; Craig et al., 1994).

In a follow-up step, HPLC-derived ergovaline data were used to develop an NIR spectroscopic calibration (Roberts et al., 1997). The NIR calibration for ergovaline offered advantages over its corresponding HPLC method. Once the NIR spectrophotometer was calibrated, two technicians could easily analyze 1000 samples within a normal work week. Processing a volume of this magnitude allowed large numbers of samples to be analyzed for ergovaline that would not have been analyzed otherwise because of time and cost restrictions.

Although the ergovaline calibration was successful, it had limitations that eventually relegated it to a mere proof of concept. One limitation was the cost of population expansion. The original equation had been developed with samples from one cultivar from a single study (Roberts et al., 1997). Because NIR technology is empirical, this equation could not be applied to other samples unless the calibration population was expanded. Such expansion would be costly because expansion samples would require the expensive HPLC procedure to quantify ergovaline. Another limitation of the ergovaline equation was its specificity; it did not measure all ergot alkaloids, but ergovaline only.

It is likely that more useful NIR calibrations could be developed for ergot alkaloids, as there are new immunochemical procedures to measure total ergot alkaloids (Hill et al., 1994). An NIR calibration based on an immunochemical data set would offer several advantages over the ergovaline calibration. It would employ a much faster and affordable reference method; this would in turn expedite both calibration development and population expansion. Also, an NIR calibration based on immunochemistry would quantify an entire class of ergot alkaloids, not just ergovaline. The end result could be an NIR equation that is robust, easy to manage, and possibly more accurate than the ergovaline equation. Such an ergot alkaloid calibration could be added to an active NIR equation profile, thereby providing an important parameter to scientists involved in routine forage quality research. Therefore, the objective of this study was the determination of total ergot alkaloid content in tall fescue by NIR spectroscopy with a calibration based on immunochemical reference data.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Sample Population
To develop a robust equation, a diverse population of tall fescue samples was compiled from three independent experiments conducted at the University of Missouri. Samples were from a hay and silage preservation study, a grazing trial, and a winter stockpiling study. All samples had been processed similarly; they had been freeze-dried, ground to pass a 1-mm screen in a cyclone-type grinder, and stored at –5°C.

The population included 84 diverse samples that included two cultivars, endophyte-infected ‘Kentucky 31’, and endophyte-free ‘Phyter’. It also included the experimental germplasms ‘HiMag’, which is endophyte-free, as well as HiMag infected with Neotyphoidum strain HM4; HM4 is an introduced endophyte that produces no ergot alkaloids and is owned by the University of Arkansas. The population also included samples from both clipped plots and grazed pastures; some samples were from monoculture paddocks while others had been collected from grass–legume paddocks. Samples in the population had been collected throughout the growing season as well as the late fall and winter stockpiling season. Also, samples represented the full range of physiological development, including early spring growth, spring and summer regrowth, reproductive maturity, and fall regrowth; in addition, some of these samples were straw aftermath that followed a seed harvest. Finally, the population included samples subjected to a wide range of harvest, preservation, and storage treatments; these treatments included green chop, hay, haylage, and ammoniated hay. Samples were collected at experiment stations near Mt. Vernon and Columbia, MO, over a 5-yr period, between 1998 and 2002.

Immunochemical Procedure for Total Ergot Alkaloids
Samples were analyzed in triplicate for ergot alkaloid concentration (Hill and Agee, 1994) using a commercial ELISA test kit (Catalog no. ENDO899-1, Agrionostics Ltd. Co., Watkinsville, GA). This kit uses a 96-well microtiter plate with a competitive ELISA format. Each well was coated overnight with a protein-lysergic acid conjugate at 4°C and washed the following morning; protein binding sites were blocked with a protein solution included in the kit. Three 0.1-g subsamples were weighed from each sample and placed into 13- by 100-mm borosilicate glass tubes. Eight milliliters of extraction buffer (from kit) was added to each and agitated for 1 h. The particulate matter in the tubes was permitted to settle for 30 min, after which a 1-mL volume was placed into a microcentrifuge tube. The microcentrifuge tubes containing the extraction buffer were centrifuged at 10000 x g for 5 min using a benchtop centrifuge. Fifty microliters of extraction buffer from each sample was pipetted into a microtiter well. Fifty microliters of ergot alkaloid-specific murine monoclonal antibody solution (from kit) was added to the microtiter wells containing the samples and incubated for 2 h. The plates were washed and 50 µL of antimurine polyclonal antibody/alkaline phosphatase conjugate (from kit) was added to each well and incubated for 1 h. The plates were washed, 50 µL of chromogenic buffer (from kit) was added, and the plates were incubated for approximately 10 min. The color reaction was stopped by adding 50 µL of 3 M NaOH. Absorbance of each well was measured at 405 nm. Thus, absorbance represented relative concentration of total ergot alkaloids, hereafter referred to as total ergot alkaloids or relative concentration. At the time of this research, there was no universal standard for conversion of relative concentration to actual concentration of total ergot alkaloids. Absorbance values were therefore used as reference data for NIR calibration development according to the chemometric procedures described below.

Spectra Collection and Initial Calibrations
Samples were scanned with NIR radiation from 1110 to 2490 nm using a Pacific Scientific 5000 scanning monochromator (NIRSytems, Silver Spring, MD) and software from Infrasoft International (Port Matilda, PA). Log 1/reflectance was recorded and converted to second-derivative spectra. Absorbance data were regressed against spectral data using modified partial least squares regression (Duckworth, 1998; Shenk and Westerhaus, 1991). Every fourth sample was reserved for cross-validation. Optimum calibration equations were selected on the basis of high coefficients of determination and low standard errors of calibration and validation.

Four initial NIR calibrations were developed as the diverse population was systematically altered (Fig. 1) . According to this approach, if the first calibration proved imprecise, subsets of samples would be omitted from the population, thereby reducing population diversity and generating additional calibrations. The sequential development of calibrations as populations increase in diversity is sometimes necessary as extremely diverse populations often result in imprecise NIR calibrations (Roberts et al., 2004). Because it is impossible to know which subsets reduce precision, they are omitted or retained on a qualitative basis. In this study, therefore, initial calibrations were based on the whole population and those populations listed below:

  1. all samples (whole population—the most diverse)
  2. all samples except those from the preservation study
  3. all samples except those from the grazing study
  4. all samples except those from the stockpiling study


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 1. Strategy used to develop near-infrared spectroscopic calibrations to measure total ergot alkaloid content in tall fescue.

 
Chemical and Spectral Examination of Omitted Subsets
As the initial calibrations improved with the omission of a subset, the omitted subset was examined further to elucidate reasons for its negative effect on calibration precision, two reasons being lack of chemical accuracy and/or lack of spectral similarity to the population. Chemical accuracy of the omitted subset was checked by analyzing samples for ergovaline concentration according to HPLC procedure described by Rottinghaus et al. (1991) utilizing the sample preparation protocol described by Hill et al. (1993). The HPLC procedure for ergovaline is currently the only other accepted chemical procedure for quantifying ergot alkaloids produced by Neotyphodium. Ergovaline data determined by HPLC were regressed against total ergot alkaloid data measured at 405 nm to produce linear, quadratic, and cubic equations; and the highest-order equation with significant (P < 0.05) regression coefficients was selected to determine goodness of fit between the two analytical methods.

Spectral similarity between the omitted subset and the main population was evaluated by comparing scores calculated in principle component analysis. Using WINISI II software (Infrasoft International, Port Matilda, PA), a library of principle component scores was computed from spectra of all samples. Principle component scores of the omitted subset were plotted in a 3-D graphic hypersphere using the first three principle components. Principle component scores of the remaining population were also plotted as a secondary file within the same hypershpere to permit visualization of symmetric distribution of samples.

Final Calibrations
A series of final calibrations were developed in an attempt to reincorporate samples from the omitted subset. The reincorporation of such samples would allow the final equation to retain the original range in diversity and optimize its robustness. As will be discussed in the results but mentioned here for clarity, the initial equations were successful only when stockpiled samples were omitted. Final calibrations, therefore, would reincorporate all but certain types of stockpiled samples. Final calibrations were performed with the following three populations:

  1. all samples except stockpiled tall fescue with an introduced endophyte
  2. all samples except stockpiled tall fescue with no endophyte (endophyte-free)
  3. all samples except stockpiled tall fescue with a toxic endophyte

Final calibrations were developed with the same regression protocol employed in the initial equations. Also, optimum calibrations were selected according to the statistical criteria specified previously.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Initial Calibrations
The first calibration for total ergot alkaloids was only moderately successful (Table 1). Though it contained all 84 diverse samples and would therefore be considered robust, it yielded unimpressive regression statistics. Its 1 – VR, which may be regarded as the regression coefficient calculated in validation, was only 0.72. Additionally, its coefficient of variation (CV), calculated from the mean and SECV, was 25%. Such statistics indicate far less precision than that reported for the ergovaline calibration (Roberts et al., 1997), which produced a 1 – VR of 0.86 and a CV of 19%.


View this table:
[in this window]
[in a new window]

  		Table 1. Initial calibration data for near-infrared spectroscopic determination of total ergot alkaloid content in tall fescue. Regression statistics were calculated by regressing spectral data against chemical reference data using modified partial least squares regression.

 
The omission of subsets affected the calibration statistics substantially. When samples from the grazing trial were omitted from the population, the calibration lost precision (Table 1); this also occurred when samples from the preservation study were omitted. However, when samples from the stockpiling study were omitted, calibration statistics improved (Table 1), achieving the same level of precision as that reported for the 1997 ergovaline equation. Omission of stockpiled tall fescue samples from the population produced a 1 – VR of 0.86 and a CV of 20%.

Chemical Accuracy of Stockpiled Samples
When the omitted stockpiled samples were analyzed for ergovaline by HPLC, the chromatographic data correlated strongly with the absorbance data originally used in calibration development (Fig. 2) . It should be remembered that in this immunoassay, absorbance is negatively correlated with concentration. Such a strong correlation between these two independent measurements for ergot alkaloids leaves little question that the original immunochemical data for the stockpiled samples were accurate. This accuracy is further supported by characterization of the individual samples, which are known to differ in toxicity. The seven samples that were infected with a toxic endophyte contained ergovaline ranging from 31 to 213 µg kg–1 and gave the lowest absorbance readings (indicating high concentrations of ergot alkaloids) (Fig. 2). All other samples contained no detectable ergovaline or ergot alkaloids, and these samples were HiMag with no endophyte or strain HM4, which produces no ergot alkaloids.


Figure 2
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 2. Ergovaline as quantified by high performance liquid chromatography (HPLC) vs. total ergot alkaloid content (relative concentration) determined by immunochemistry. Samples were stockpiled tall fescue infected with no endophyte, an introduced endophyte, and a toxic endophyte. Abs, absorbance.

 
Spectral Similarity between Stockpiled Samples and Whole Population
Figure 3 displays the 3-D hypersphere containing principle component scores of the stockpiled tall fescue and the other samples in the population. While the plot reveals a region for the stockpiled samples, it also reveals overlap with scores from the main population. This degree of overlap indicates an ideal degree of symmetrical distribution of samples within a single population. Excessive overlap of scores between the two files would have indicated that some samples were redundant. Conversely, the absence of any overlap would have indicated that the stockpiled samples were not merely different from samples in the main population but that they were clustered into a distinct, stand-alone population (Schulz, 2004). In short, Fig. 3 indicates that samples of the stockpiled subset are within the limits of spectral variation considered for this single population. This spectral similarity of stockpiled samples to the whole population was confirmed further in calibration development, as there were no H outliers identified during regression.


Figure 3
View larger version (21K):
[in this window]
[in a new window]

  		Fig. 3. Three-dimensional hypersphere plot of the first three principle component scores calculated from near-infrared reflectance data of 84 diverse tall fescue samples. The plot has been rotated to present maximum separation between two files—those including scores of stockpiled tall fescue (+) and those containing scores of the remaining population ({circ}).

 
Final Calibrations
The final set of calibrations showed that omission of the entire stockpiled subset was not necessary to achieve a precise calibration. When only those stockpiled samples infected with a toxic endophyte were omitted, the calibration was more precise than all other calibrations in this study (Table 2). In addition, this calibration was more precise than the ergovaline calibration reported earlier (Roberts et al., 1997); for this calibration, the 1 – VR was 0.89, and the CV was 16%.


View this table:
[in this window]
[in a new window]

  		Table 2. Final calibration statistics for near-infrared spectroscopic determination of total ergot alkaloid content in tall fescue. Regression statistics were calculated by regressing spectral data against chemical reference data using modified partial least squares regression.

 
It is also worth noting that the R2 of calibration was 0.95, which is unusually high for NIR determination of antiquality components (Roberts et al., 2004). When considered alongside the corresponding 1 – VR and the CV, statistics for the final equation would be considered excellent, keeping in mind that it measured components occurring in the µg kg–1 range (Rottinghaus et al., 1991). The statistics of the final equation would also be considered excellent compared with statistics for NIR determination of plant alkaloids in rangeland crops (Clark et al., 1987).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Though the final equation was precise and potentially useful in many research applications, it raises a question regarding the omission of certain stockpiled samples. Why is it that the omission of only seven stockpiled samples, which were not outliers, increased the 1 – VR from 0.72 to 0.89 and reduced the CV from 25 to 16% (Tables 1 and 2)? This question cannot be answered sufficiently without further experimentation, yet it does invite an explanation to form a working hypothesis for the future.

One explanation would be related to the empirical nature of NIR technology. It is probable that the NIR spectrophotometer is detecting precursors or products of ergot alkaloids that are not measured directly by HPLC or the immunoassay. If not precursors or products, the instrument could be detecting other fungal metabolites that are simply correlated with ergot alkaloids. These precursors, products, or fungal metabolites, hereafter referred to as "detected compounds," would generally be correlated with ergot alkaloids, thereby permitting a precise calibration. As long as their concentrations were proportional to ergot alkaloid concentrations, a precise equation could be expected. However, should their concentrations be proportional to ergot alkaloid concentrations in some but not all of the samples, the correlation would not be consistent across the population. In such a case, calibration precision would decrease; precision could be increased only when samples responsible for the failed correlation were removed from the population.

In this study, precision was increased after omission of one type of samples. The omitted samples were not merely stockpiled tall fescue but were stockpiled tall fescue infected with a toxic endophyte (Fig. 1 and Table 2). These were the only samples in the diverse population whose plants contained ergot alkaloids and were sampled during the winter. As such, they were from plants whose ergot alkaloid concentration is known to decrease over the winter months (Kallenbach et al., 2003), even while the endophyte percentage remains constant.

Ergot alkaloid content in these samples was underestimated by NIR spectroscopy (Fig. 4) . The NIR-predicted absorbance, which is negatively correlated to ergot alkaloid concentration, was twice as high as absorbance measured by the immunochemical procedure, a procedure validated by the HPLC data (Fig. 2). It therefore appears that the spectrophotometer is detecting compounds whose concentrations coincide with ergot alkaloid concentrations until the winter months, at which time they decrease much more so than do ergot alkaloid concentrations.


Figure 4
View larger version (20K):
[in this window]
[in a new window]

  		Fig. 4. Actual and predicted ergot alkaloid content in stockpiled tall fescue infected with a toxic endophyte. Actual absorbance, which is negatively correlated with ergot alkaloid concentration, was quantified by immunochemistry. Predicted absorbance was by near-infrared spectroscopy. Abs, absorbance.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
We conclude that total ergot alkaloid content in tall fescue can be quantified by NIR spectroscopy with a calibration based on immunochemical data. An NIR calibration such as this one has several advantages over the NIR calibration for ergovaline. First, this calibration for total ergot alkaloids can be developed quickly and inexpensively, largely because the immunoassay used in this experiment is more time and cost efficient than HPLC. Second, this calibration predicts a class of compounds rather than a single compound; this class of compounds is the subject of current research questions, and it may prove to be more indicative of livestock toxicosis than is ergovaline alone (Hill, 2005). Third, this calibration is precise. The final calibration in this study had higher correlation coefficients and a lower CV than the corresponding statistics reported for the ergovaline calibration. Its precision was similar to or greater than precision reported for other calibrations that quantify microconstituents. Finally, the calibration for ergot alkaloid content is robust. Although the final population did not include stockpiled samples of Kentucky 31 infected with a toxic endophyte, it included almost every other type of tall fescue sample commonly collected from applied research trials and grown in pastures. The population included samples that ranged in physiological development from vegetative to reproductive maturity, and some samples were straw aftermath from a seed harvest. The population also included several tall fescue cultivars and plant germplasms. In addition, it included samples grown at different locations and over multiple years. The population also included samples from grazed pastures and clipped plots; some were from monocultures and others from mixtures. In addition, the population included tall fescue that had been harvested and stored with various management practices, including green chop, hay, silage, and ammoniated hay. Also, it contained samples with all possible endophyte infection status, including plants that were infected with a toxic endophyte, with an introduced endophyte, and no endophyte. As mentioned earlier, many of these samples were also stockpiled.

A calibration such as this has significant potential in applied research that involves tall fescue breeding, production, and management. Many of these research programs already employ NIR spectroscopy to estimate forage quality, using calibration profiles for simultaneous determination of crude protein, fiber components, and moisture. If these profiles were modified to include an equation for ergot alkaloids, precise estimation of endophyte toxins would be effortless.


   ACKNOWLEDGMENTS
 
This material is based on work supported by the USDA, under Agreement no. 58-6227-3-016. Any opinions, findings, conclusion, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the USDA.

Received for publication July 6, 2004.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in Crop Science:

THIS ISSUE IN CROP SCIENCE

Crop Science 2005 45: vi. [Full Text]  




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Agricola
Right arrow Articles by Roberts, C. A.
Right arrow Articles by Rottinghaus, G. E.
Related Collections
Right arrow Other Forage Crops

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome