HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Bell, G. E. Articles by Solie, J. B. Search for Related Content PubMed Articles by Bell, G. E. Articles by Solie, J. B. Agricola Articles by Bell, G. E. Articles by Solie, J. B. Related Collections Turfgrass Management Turfgrass Plant Analysis

TURFGRASS SCIENCE

Vehicle-Mounted Optical Sensing

An Objective Means for Evaluating Turf Quality

^a Dep. of Horticulture and Landscape Architecture, Oklahoma State Univ., Stillwater, OK 74078
^b Dep. of Biosystems and Agricultural Engineering, Oklahoma State Univ., Stillwater, OK 74078

* Corresponding author (bgregor{at}okstate.edu)

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 
Visual evaluation of turfgrass quality is a subjective process that requires experienced personnel. Optical sensing of plant reflectance provides objective, quantitative turf quality evaluation and requires no turf experience. This study was conducted to assess the accuracy of optical sensing for evaluating turf quality, to compare the rating consistency among human evaluators and optical sensing, and to develop a model that describes a relationship between optically sensed measurements and visual turf quality. Visual evaluations for turf color, texture, percent live cover (PLC), and optically sensed measurements were collected on the National Turfgrass Evaluation Program (NTEP) tall fescue (Festuca arundinacea Schreb) and creeping bentgrass (Agrostis palustris Huds.) trials at Stillwater, OK. Measurements were made monthly for 12 consecutive months from June 1999 through May 2000. Red (R) and near infrared (NIR) reflectance were collected with sensors and converted to normalized difference vegetative indices (NDVI). The NDVI were closely correlated with visual evaluations for turf color, moderately correlated with percent live cover (PLC), and independent of texture. Measurements of turf color and PLC were evaluated more consistently with optical sensors than by visual ratings. Normalized difference vegetation index (Y) could be reliably predicted by the following generalized model for turf color (X) and PLC (Z): Y = B₀ + B₁log10X + B₂Z³. Optical sensing provided fast, reliable turf assessment and deserves consideration as a supplemental or replacement technique for evaluating turf quality.

Abbreviations: NDVI, normalized difference vegetation index • NTEP, National Turfgrass Evaluation Program • NIR, near infrared irradiance (780-nm wavelength) • PLC, percent live cover • R, red irradiance (671-nm wavelength) • VMOS, vehicle-mounted optical sensors

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 
RESEARCHERS AND PRACTITIONERS require specialized techniques to evaluate turfgrass health and make management decisions. Historically, turfgrass color, texture, and PLC have been determined subjectively. Experienced personnel are required to provide an expert opinion on the visual quality of turfgrass stands. This subjective approach has been accepted and used for many years to investigate the visual effects of fertilizer and pesticide applications or genetic variation among cultivars.

Visual turfgrass evaluation is time consuming, can be inconsistent among individual evaluators, and may vary from day to day for the same evaluator (Trenholm et al., 1999). Technology is available that uses optical detectors to determine reflectance from a turf canopy. These measurements may provide an alternative for turfgrass evaluation (Trenholm et al., 1999; Bell et al., 2000b). Optical sensing provides a quantitative value for objective comparison of turf response to treatments or genetic variation. The day-to-day consistency of optical sensing devices appears to exceed that of human evaluators. In addition, minimal training is required to evaluate turf with optical sensors.

Optical sensors measure irradiance reflected from turfgrass over which they travel. Radiation not reflected is either absorbed by the plant or transmitted to the soil. The amount of red (R) radiation (600–700 nm wavelengths) transmitted to the soil surface below a dense turf canopy is small compared with that absorbed by the canopy (Knipling, 1970). Red reflectance from a turf canopy is relatively low because of chlorophyll absorption, but near infrared (NIR) radiation (700–800 nm) is poorly absorbed by plants and more highly reflected (Daughtry et al., 1992). A portion of NIR radiation is reflected from green plant material and a portion is transmitted to the soil (Knipling, 1970; Bell et al., 2000a). Red reflectance increases and NIR reflectance decreases as greenness declines in a dying plant or leaf (Knipling, 1970). This relationship between R and NIR reflectance and living and dead plant material provides a basis for plant health indicators such as normalized difference vegetative indices [NDVI; NDVI = ]. Normalized difference vegetation index is related to absorbed photosynthetically active radiation in wheat (Triticum aestivum L.; Asrar et al., 1984) and has been associated with leaf area index in maize (Zea mays L.) and soybeans [Glycine max (L.) Merr.; Daughtry et al., 1992]. Reflectance measurements have been used to evaluate plant biomass (Walburg et al., 1982; Kleman and Fagerlund, 1987; Wanjura and Hatfield, 1987), plant nitrogen content (Blackmer et al., 1994; Stone et al., 1996), and disease severity (Nutter et al., 1993; Green, 1998). Normalized difference vegetation index has been used to measure drought stress (Fenstermaker-Shaulis et al., 1997), turf chlorophyll content (Howell, 1999), and turf injury and quality (Trenholm et al., 1999; Bell et al., 2000b). Several articles have been written concerning optical sensing for use as a plant status indicator. The practical significance of the use of reflectance detectors to measure turf quality or performance, however, remains unknown. Additional research that investigates optical sensing in relation to known plant indicators or response variables is necessary before decisions can be made concerning the future of this technology in turf management and research. The objectives of this study were to assess the accuracy of optical sensing for evaluating turf quality, to compare the rating consistency of human evaluators and optical sensing, and to determine a model that accurately describes a relationship between optically sensed measurements and turf visual quality.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 
Visual ratings and vehicle-mounted optically sensed (VMOS) measurements were collected on the 1996 National Turfgrass Evaluation Program (NTEP) tall fescue trial and the 1998 NTEP creeping bentgrass trial at the Oklahoma State University Turfgrass Research Center, Stillwater, OK. The tall fescue trial consisted of 390 experimental plots (1.22 by 1.83 m) including three replications of 130 cultivars arranged in randomized complete blocks. The creeping bentgrass trial is the same experimental design and consisted of 30 cultivars and three replications in 1.52- by 1.52-m plots.

The tall fescue trial was mowed 2 to 3 times wk^-1 at 63 mm and fertilized with 116 kg N ha^-1 yr^-1. The silty clay loam (Kirkland silt loam, fine, mixed, superactive, thermic Udertic Paleustolls) soil at this site contained adequate levels of P (90.5 mg kg^-1) and K (334.5 mg kg^-1) and was not supplemented with those or other nutrients. Prodiamine [N3,N3-di-n-propyl-2,4-dinitro-6-(trifluoromethyl)-m-phenylenediamine] was applied in March 1999 and March 2000 to control annual grassy weeds. Azoxystrobin [(E)-2-{2-[6-(6-(2-cyanophenoxy) pyrimidin-4-yloxylphenyl]}-3-methoxyacrylate] was applied for brown patch disease (Rhizoctonia solani Khn) control in August 1999 and the plots were treated with halosulfuron-methyl (methyl 6[((4,6-dimethoxy-2-pyrimidinyl)-amino)carbonylaminosulfonyl]-4-chloro-1-methyl-1H-pyrazole-4-carboxylate] in October 1999 to control yellow nutsedge (Cyperus exculentus L.).

The creeping bentgrass turf received 145 kg N ha^-1 yr^-1 with supplemental applications of P and K on the basis of soil test levels. The turf was grown on sand amended with shredded sphagnum peat moss and mowed daily six times wk^-1 at 4 mm. Curative applications of chlorothalonil (tetrachloroisophthalonitrile) were applied from spring until autumn for dollar spot (Sclerotinia homoeocarpa F.T. Bennett) and brown patch control. Single applications of myclobutanil [-butyl--(chlorophenyl)-1H-1,2,4, triazole-1-propanenitrile], mancozeb (zinc, manganese ethylene bisdithiocarbamate), and azoxystrobin also were made for disease control in 1999 and the trial was treated for black cutworm (Agrostis ipsilon Hufnagel) using lambda-cyhalothrin {[1(S*),(Z)]-(±)-cyano(3-phenoxyphenyl)methyl-3-(2-chloro-3,3,3-trifluoro-1-propenyl)-2,2-dimethyl-cyclopropanecarboxylate} in June and August. March and September applications of dithiopyr [3,5-pyridinedicarbothioc acid, 2-(difluoromethyl)-4-(2-mehylpropyl)-6-(trifluoromethyl)-S,S-dimethyl ester] were used to control annual bluegrass (Poa annua L.), crabgrass (Digitaria spp.), and goosegrass [Eleusine indica (L.) Gaertn.]. Both NTEP trials were weed free except for a small (2%) infestation of creeping bentgrass in the tall fescue.

Vehicle-Mounted Optical Sensing
The instrument used for optical sensing was a commercially available optical sensor (Model PhD 600, Patchen, Ukiah, CA). A downward facing detector measured R (671 nm) and NIR (780 nm) reflectance of illumination originating from sensor-integrated light emitting diodes (Bell et al., 2000a). The radiance originating from the instrument was filtered electronically from ambient sunlight and detected by a phase shift technique (Beck and Vyse, 1994). By means of this technique, the instrument removed the effect of ambient radiation and processed only the energy emitted by the integrated sources. The sensor produced a single value transmitted as a voltage level. Preliminary research indicated that this value was linearly correlated with NDVI but varied with air temperature. A cubic calibration curve was developed to transform the Patchen sensor output to NDVI under varying air temperature conditions. Each sensor was individually calibrated to this model. Calibrations were tested periodically but adjustments were not necessary.

A sensor array was mounted across the front of a model 4100 tractor (Deere, Inc., Moline, IL). The sensors were connected to a processor programmed to collect spectral data as the tractor moved across the turf. Each sensor captured reflected radiation from a 10- by 305-mm sample area. Data collected from a sample area were called a frame. The four-sensor array was mounted 50 cm above the turf and aligned to scan an area perpendicular to the direction of travel 122 cm wide. To avoid data collection from neighboring plots, only the two center-mounted sensors were used for this study resulting in a field of view 61 cm wide. The number of frames collected from each plot varied with tractor speed and ranged from 28 to 156 frames plot^-1 (median = 100–120) during the course of the study. Speed of the tractor was approximately 5 to 8 km h^-1 during data collection. Measurements were collected monthly from June 1999 through May 2000. Data collected were transmitted to an onboard, portable computer as the vehicle was driven across the turf.

Leaves and extraneous debris were removed from the plots prior to spectral collection. Vehicle-mounted optical sensing was performed only when the turf canopy was dry. Other weather conditions varied among and during collection periods. Optical sensors were not sensitive to solar radiation. Therefore, cloud cover or time of day were not important. The tractor was driven over the plots in the same order during each evaluation. Once data were captured by the onboard computer, the collection was processed by a macro (a computer instruction that represents a sequence of instructions in abbreviated form) written in Excel software (Microsoft Corporation, Redmond, WA). Data were assigned to individual plots by a combination of manual techniques performed during collection and calculations performed by the macro.

A manually operated indicator button was included on the vehicle instrumentation. When pushed, this button identified plot data and enabled separation of experimental units in a data set. A shaft encoder mounted on the tractor's right front wheel signaled collection of a frame automatically according to user-specified spatial intervals. Each frame was printed in columnar format to the onboard computer. If the indicator button was depressed during collection of a frame, a 1 was printed in a column identified for that purpose. These indicator integers were printed in the same row as the corresponding frame. The indicator column remained blank when the button was not depressed. The vehicle operator depressed the button when collecting plot data and released it during transition between plots. This procedure enabled the identification of individual plots in a data set. Using this indicator, the Excel macro assembled plots according to location, discarded transition frames, converted Patchen data to NDVI, and calculated a mean for each plot. The tall fescue trial routinely resulted in data sets of 40 000 measurements or more and could be converted to NDVI plot means in less than 20 min.

Consistency among Evaluators and the Relationship of NDVI with Turf Color and Percent Live Cover
Turf was rated monthly for 12 mo by VMOS and three human evaluators. Evaluator 1 (E1) had 12 yr turf rating experience, Evaluator 2 (E2) had 7 yr, and Evaluator 3 (E3) had 1 yr experience. A fourth individual with no turf rating experience operated the optical sensing equipment and compiled optical data. Human evaluators rated turf for color, texture, and PLC using discrete numeric scales. Color (1 = brown, 9 = dark green) and texture (1 = very coarse, 9 = very fine) were rated on an integer scale and PLC was rated on a percentage integer scale (0 = no live cover, 100 = complete live cover). Vehicle-mounted optical sensing reported a single NDVI number on a continuous scale from -1.0 to 1.0 for each frame evaluated. This NDVI number represented an unknown combination of turf characteristics that was believed to be influenced by turf color and PLC. On the basis of historic scientific acceptance of visual evaluation for turf rating purposes, experimental unit means averaged over three human evaluators were deemed accurate measures of turf color, texture, and PLC for this study. Data for human evaluators were compared to determine the consistency among evaluators. Multiple regression equations were developed for tall fescue and creeping bentgrass to predict NDVI from turf color and percent live cover from the mean ratings of three human evaluators for each plot. Turf texture was not related to NDVI and was not evaluated by this method. There were 3240 individual observations of turf color and percent live cover for creeping bentgrass. These data were averaged over the three evaluators to yield 1080 observations that were used to develop the model. There were 14 040 observations of turf color and percent live cover for tall fescue with the average of the three evaluators yielding 4680 observations for model development. Model selection employed the stepwise technique (Draper and Smith, 1966), and included various functions of the independent variables. Throughout this text, r² represents simple regression and R² represents multiple regression coefficients of determination.

Consistency of Rating Techniques over the Growing Season
Data for rating accuracy were collected on or near the 15th of each month. During July 1999, November 1999, and May 2000, additional evaluations were performed by E2 and E3. During these months, E2 and E3 evaluated the tall fescue and creeping bentgrass NTEP trials daily for three consecutive days. Vehicle-mounted optical sensing also was performed on these days and the resulting data were used to measure consistency within evaluators and VMOS. These ratings performed consecutively provided three possible comparisons within evaluators: Day 1 by Day 2, Day 1 by Day 3, and Day 2 by Day 3. Simple regression and correlation were calculated for each of these comparisons (Draper and Smith, 1966).

	RESULTS AND DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 
Vehicle-Mounted Optical Sensing
More time was required to evaluate a trial by VMOS than by visual evaluators. The tall fescue trial required approximately 1.5 h for VMOS evaluation, and the creeping bentgrass trial required approximately 0.5 h. The slowest human evaluator could complete approximately two ratings in the time required for a single VMOS evaluation. The VMOS evaluation, however, required less time for data compilation. By the data techniques and macro described, the tall fescue trial, which routinely resulted in data sets of 40 000 numbers or more, could be converted to NDVI plot means and listed by location in less than 20 min. Manual data entry to a computer spreadsheet required approximately the same amount of time, but had to be checked thoroughly for errors in data entry. The review for errors in data entry almost doubled the time required for manual data compilation. A few checks at random locations were enough to satisfy the VMOS operator that data entry was correct. Overall, VMOS increased the time necessary for evaluation and data entry by about 30%. Collection time could be reduced if vehicle speed was significantly increased during evaluation. With this equipment, the time required for onboard data processing prevented an increase in vehicle speed. Vehicle speed could be increased 10-fold by the use of upgraded data acquisition systems.

Although slower, VMOS evaluation was easier to learn and perform than visual rating. Technician training for VMOS use required approximately 30 to 60 min. Visual evaluation may require several weeks of experience before ratings are acceptable and many months before the evaluator becomes proficient. Vehicle-mounted optical sensing was also more flexible with changing environmental conditions. Visual rating was easier to perform during cloudy conditions and was difficult during early morning and late afternoon when the angle of direct solar radiation was approaching horizontal. Visual evaluators were directed to keep natural light at their backs without allowing a shadow to be cast on the turf under consideration. The VMOS system used for this research was not affected by light conditions because the sensor detected only the integral light sources and solar radiation was ignored. A system of simultaneous up- and down-looking detectors also was effective under most solar conditions in previous research (Trenholm et al., 1999; Bell et al., 2000a, 2002). Finally, VMOS was more convenient to use but required the purchase and maintenance of expensive instrumentation.

Accuracy of Rating Techniques
A visual average, the mean rating determined by three human evaluators on an experimental unit at the same time, was considered an accurate measure of turf status for this study. Correlations between the evaluators' visual average and NDVI indicated that NDVI was closely related to turf color (r² = 0.75 and 0.41 for tall fescue and creeping bentgrass, respectively), independent of texture (r² = 0.01 and 0.04), and moderately related to PLC (r² = 0.39 and 0.34) (Table 1). In most cases, individual human evaluators were more closely correlated with the visual average rating for the three measured parameters than was NDVI with visual average rating (Table 1). Individuals rated turf qualities on a discrete scale, but NDVI was reported on a continuous scale predisposing NDVI data to a wider range of possible measurements and more variability. In addition, each measurement made by a human evaluator was a component of the visual average but NDVI was not. The NDVI rated a different variable than visual evaluators. For NDVI, the rated variable was not color or PLC, but a combination of those and possibly other factors. Correlations assumed a linear relationship between the two variables, but in fact, NDVI best-fit equations for color and percent live color were not linear. These nonlinear equations were utilized to develop a multiple regression equation that predicted NDVI from visual turf color and percent live cover. There was no useful relationship between NDVI and turf texture.

Mathematical Modeling
A model to predict NDVI from turf color using visual ratings on a 1 to 9 scale and PLC on a 1 to 100 scale was developed by the stepwise selection procedure. The relationship of these two variables with NDVI was better for tall fescue than creeping bentgrass (R² = 0.8 and 0.5, respectively) (Fig. 1 and 2) . These are strong correlations considering that the model was based on visual rankings that were more variable than NDVI measurements (Table 2). In the model for tall fescue, the log₁₀ of color accounted for 77% of NDVI variability and the cubic function of PLC accounted for another 3%. The model for creeping bentgrass was not as accurate as that for tall fescue. For creeping bentgrass, the log₁₀ of color accounted for 39% of the variability and the cubic function of percent live cover accounted for an additional 11%. The accuracy of NDVI may have been enhanced on tall fescue due to increased plant biomass. Color rating inconsistencies among visual evaluators on creeping bentgrass probably caused a poorer correlation between the visual average rating and NDVI. These factors combined to enhance a predictive equation for tall fescue and diminish the effectiveness of a predictive equation for creeping bentgrass.

View larger version (15K): [in this window] [in a new window]   Fig. 1. The relationship between normalized difference vegetation index (NDVI) and tall fescue turf color on a 1 to 9 scale with 9 the deepest green and percentage live cover (PLC) on a 0 to 100% scale. Model: NDVI = 0.258 + 0.4867 x log10 turf color + 1.053 x 10-7 x PLC3, R2 = 0.80, P < 0.0001.

View larger version (15K): [in this window] [in a new window]   Fig. 2. The relationship between normalized difference vegetation index (NDVI) and creeping bentgrass turf color on a 1 to 9 scale with 9 the deepest green and percentage live cover (PLC) on a 0 to 100% scale. Model: NDVI = 0.305 + 0.3072 x log10 turf color + 1.1757 x 10-7 x PLC3, R2 = 0.50, P < 0.0001.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 
Approved for publication by the Director of the Oklahoma Agric. Exp. Stn. Research was conducted under Oklahoma Agric. Exp. Stn. Project OKLO 2392.

Received for publication March 23, 2001.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

 

• Asrar, G., M. Fuchs, E.T. Kanemasu, and J.L. Hatfield. 1984. Estimating absorbed photosynthetic radiation and leaf area index from spectral reflectance in wheat. Agron. J. 76:300–306.[Abstract/Free Full Text]
• Beck, J., and T. Vyse. 1994. Structure and method for differentiating one object from another object. U.S. Patent 5 296 702. Date issued: 22 March.
• Bell, G.E., T.K. Danneberger, and M.J. McMahon. 2000a. Spectral irradiance available for turfgrass growth in sun and shade. Crop Sci. 40:189–195.[Abstract/Free Full Text]
• Bell, G.E., D.L. Martin, and R.M. Kuzmic. 2000b. Herbicide tolerance of two cold-resistant bermudagrass cultivars determined by both visual assessment and spectral reflectance. Weed Technol. 14:635–641.
• Bell, G.E., D.L. Martin, M.L. Stone, J.B. Solie, and G.V. Johnson. 2002. Turf area mapping using vehicle-mounted optical sensors. Crop Sci. 42(2).
• Blackmer, T.M., J.S. Schepers, and G.E. Varvel. 1994. Light reflectance compared with other nitrogen stress measurements in corn leaves. Agron. J. 86:934–938.[Abstract/Free Full Text]
• Daughtry, C.S.T., K.P. Gallo, S.N. Goward, S.D. Prince, and W.P. Kustas. 1992. Spectral estimates of absorbed radiation and phytomass production in corn and soybean canopies. Remote Sens. Environ. 39:141–152.
• Draper, N.R., and H. Smith. 1966. Applied regression analysis. John Wiley and Sons, Inc., New York.
• Fenstermaker-Shaulis, L.K., A. Leskys, and D.A. Devitt. 1997. Utilization of remotely sensed data to map and evaluate turfgrass stress associated with drought. J. Turfgrass Manage. 2:65–81.
• Green, D.E., II. 1998. Canopy reflectance as a measure of disease in tall fescue. Crop Sci. 38:1603–1613.[Abstract/Free Full Text]
• Howell, B.M. 1999. The effectiveness of sensor based technology to determine nitrogen deficiencies in turfgrasses. M.S. Thesis. Dept. Plant and Soil Science, Oklahoma State University, Stillwater, OK.
• Kleman, J., and E. Fagerlund. 1987. Influence of different nitrogen and irrigation treatments on the spectral reflectance of barley. Remote Sens. Environ. 21:1–14.
• Knipling, E.B. 1970. Physical and physiological basis for the reflectance of visible and near-infrared radiation from vegetation. Remote Sens. Environ. 1:155–159.
• Nutter, F.W., Jr., M.L. Gleason, J.H. Jenco, and N.C. Christians. 1993. Assessing the accuracy, intra-rater repeatability, and inter-rater reliability of disease assessment systems. Phytopathology 83:806–812.[ISI]
• Stone, M.L., J.B. Solie, W.R. Raun, R.W. Whitney, S.L. Taylor, and J.D. Ringer. 1996. Use of spectral radiance for correcting in-season fertilizer nitrogen deficiencies in winter wheat. Trans. ASAE 39: 1623–1631.
• Trenholm, L.E., R.N. Carrow, and R.R. Duncan. 1999. Relationship of multispectral radiometry data to qualitative data in turfgrass research. Crop Sci. 39:763–769.[Abstract/Free Full Text]
• Walburg, G., M.E. Bauer, C.S.T. Daughtry, and T.L. Housley. 1982. Effects of nitrogen nutrition on the growth, yield, and reflectance characteristics of corn canopies. Agron. J. 74:677–683.[Abstract/Free Full Text]
• Wanjura, D.F., and J.L. Hatfield. 1987. Sensitivity of spectral vegetative indices to crop biomass. Trans. ASAE 30:810–816.

This article has been cited by other articles:

				X. Xiong, G. E. Bell, J. B. Solie, M. W. Smith, and B. Martin Bermudagrass Seasonal Responses to Nitrogen Fertilization and Irrigation Detected Using Optical Sensing Crop Sci., July 30, 2007; 47(4): 1603 - 1610. [Abstract] [Full Text] [PDF]

				Y. Jiang and R. N. Carrow Broadband Spectral Reflectance Models of Turfgrass Species and Cultivars to Drought Stress Crop Sci., July 30, 2007; 47(4): 1611 - 1618. [Abstract] [Full Text] [PDF]

				K. E. Kenworthy, C. M. Taliaferro, B. F. Carver, D. L. Martin, J. A. Anderson, and G. E. Bell Genetic Variation in Cynodon transvaalensis Burtt-Davy Crop Sci., October 2, 2006; 46(6): 2376 - 2381. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Bell, G. E. Articles by Solie, J. B. Search for Related Content PubMed Articles by Bell, G. E. Articles by Solie, J. B. Agricola Articles by Bell, G. E. Articles by Solie, J. B. Related Collections Turfgrass Management Turfgrass Plant Analysis