Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Analysis of Herbage Mass and Herbage Accumulation Rate                           Using Gompertz Equations                 ...
of plant storage organs and amount of stored carbohydrates                                                                ...
State University Donn Scott Airport, Columbus OH (40º04´ N,83º05´ W) in pasture that had been mowed to maintain a heightof...
7.5 cm and 50 kg N ha–1 was applied                                                                                       ...
statistical Analysis                                the literature, and have simpler mathematical computation than   Herba...
table 3. Parameters for instantaneous growth rate (hAri)–herbage mass curves (table 1, eq. [3], hmin, hΔ , and b), their s...
varied considerably during the year and between locations. The                    The asymmetric logistic equations were a...
table 4. Maximum instantaneous growth rate (hAri-max),                     United States. One implication of the HAR i-–he...
some stage, be at the optimum herbage mass. Lax or infrequent                no. 2006-55618-17025; Wisconsin Department of...
Upcoming SlideShare
Loading in …5
×

Analysis of Herbage Mass and Herbage Accumulation Rate Using Gompertz Equations. Agronomy Journal, Volume 102, Issue 3, 2010

912 views

Published on

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Analysis of Herbage Mass and Herbage Accumulation Rate Using Gompertz Equations. Agronomy Journal, Volume 102, Issue 3, 2010

  1. 1. Analysis of Herbage Mass and Herbage Accumulation Rate Using Gompertz Equations David J. Barker,* Fernanda P. Ferraro, Renata La Guardia Nave, R. Mark Sulc, Fernanda Lopes, and Kenneth A. Albrecht AbstrAct Pasture Management Sigmoid equations are recognized as representative of the pattern of herbage accumulation during a growth period; however, the vari- ous equations and their variability among locations and during the growing season have not been well described. The objectives of this study were to find the most suitable, four-parameter sigmoid equations to fit measured herbage mass and to investigate how the patterns of herbage accumulation (i.e., equation parameters) varied with time of year and location. Herbage mass was measured approximately weekly during 11 to 12 growth periods with a rising plate meter (RPM) at three north-central United States locations (Columbus and Coshocton, OH, and Arlington, WI) during 2008, and those data were fit to Gompertz equations. There were four replicates for each growth period. We found predictable relationships between instantaneous herbage accumulation rate (HAR i) and herbage mass for each location and date. Time-independent HAR i vs. herbage mass curves have potential use for pasture management by defining the optimum herbage mass at which HAR i is maximum. The optimum herbage mass varied between 1600 and 4000 kg dry matter (DM) ha–1 depending on location and date. Allowing herbage mass to exceed the optimum point (e.g., delayed harvest), or harvesting to below the optimum point, will reduce the HAR i. The HAR i–herbage mass curves define a range of herbage mass within which pastures can be managed to achieve high HAR i, and maintaining pastures within 90% of the maximum HAR i may be a practical target for producers.S igmoid curves are recognized as representative of the pattern of herbage accumulation during a growth period(Richards, 1959; Landsberg, 1977). Briefly, those curves have logistic equations are those where “acceleration” of growth rate below the point of inflection is identical to the “deceleration” of growth above the point of inflection (Parsons et al., 1988, 2001;an initial period of slow herbage accumulation rate, a period Lemaire and Chapman, 1996). Asymmetric equations (Gom-of accelerating herbage accumulation rate up to a point of pertz and Weibull functions) have a different rate of increasinginflection, after which the herbage accumulation rate deceler- growth (acceleration) than decreasing growth (deceleration)ates toward the ceiling herbage mass (Hmax) (Table 1 Eq. [1], (Cacho, 1993; Belesky and Fedders, 1995). Biologically, asymmet-Fig. 1a). These curves have been applied to perennial pastures ric equations may be a better representation of herbage accumula-(Parsons et al., 1988; Cacho, 1993; Belesky and Fedders, 1994, tion since the processes of accelerating growth rate (mobilization1995; Lemaire and Chapman, 1996; Parsons et al., 2001). In of root and stem reserves, reproductive growth, and tillering) arethe case of perennial pastures, growth is usually from a mini- different from the processes of decelerating growth rate (pasturemum (or residual) herbage mass (Hmin) rather than from a neg- senescence and decomposition, tiller “self-thinning”, and leafligible initial mass such as for seedling emergence after planting shading). In such cases, the HAR i vs. time functions are not sym-(where it is often assumed Hmin = 0). Mathematically, sigmoid metric, but are usually skewed toward the y axis.equations that fit the pattern of accumulation of herbage mass An additional interpretation of herbage mass (Fig. 1a; Tablehave considerable interpretive value, since the derivative (dy/dt) 1, Eq. [1]) and HAR i (Fig. 1b; Table 1, Eq. [2]) equations is,of the herbage mass vs. time relationship defines the relation- for sequential points in time, to plot the time-independentship of HAR i vs. time (Table 1 Eq. [2], Fig. 1b). relationship of HAR i vs. herbage mass (Table 1, Eq. [3]; see The various equations that have been used to fit sigmoid Fig. 2 for an example). The HAR i–herbage mass relationshipgrowth curves (Table 1) fall into two broad categories. Symmetric has considerable practical application in that measurements of herbage mass can be used to predict HAR i (Cacho, 1993;D.J. Barker, F.P. Ferraro, R. La Guardia Nave, and R.M. Sulc, Dep. of Bluett et al., 1998). Furthermore, the HAR i–herbage mass rela-Horticulture and Crop Science, Ohio State Univ., Columbus, OH 43210; F. tionship defines the maximum HAR i. A practical range mightLopes and K.A. Albrecht, Dep. of Agronomy, Univ. of Wisconsin, Madison,WI 53706. Salary and research support provided in part by state and federal be to maintain pastures within the range of 90% of maximumfunds appropriated to the Ohio Agric. Res. and Dev. Ctr. (OARDC), Ohio HAR i. Excessive (or insufficient) forage removal (by grazing orState Univ. Published as OARDC Journal Article HCS 09-16. Received 30 machines) will result in reduced pasture growth rate.Sept. 2009. *Corresponding author (barker.169@osu.edu). In addition to this mathematical justification, the HAR i–Published in Agron. J. 102:849–857 (2010) herbage mass relationship also has a mechanistic basis. ParsonsPublished online 05 Mar., 2010 et al. (1988) showed a curve with similar shape for the relation-doi:10.2134/agronj2009.0381 ship of growth rate and leaf area index (LAI). Given the strongCopyright © 2010 by the American Society of Agronomy,5585 Guilford Road, Madison, WI 53711. All rights re-served. No part of this periodical may be reproduced ortransmitted in any form or by any means, electronic Abbreviations: DM, dry matter; HAR i, instantaneous herbage accumulationor mechanical, including photocopying, recording, or rate; Hmax, maximum (or ceiling) herbage mass of a sigmoid curve; Hmin,any information storage and retrieval system, without minimum (or residual) herbage mass; HΔ, the difference between Hmax andpermission in writing from the publisher. Hmin; LAI, leaf area index; RPM, rising plate meter.A g ro n o my J o u r n a l • Vo l u m e 10 2 , I s s u e 3 • 2 010 849
  2. 2. of plant storage organs and amount of stored carbohydrates that might be available for plant function and growth. The effect of defoliation intensity (residual herbage mass following grazing) on subsequent growth rate has long been recognized (Brougham, 1956), and that relationship can be predicted from Fig. 2. High defoliation intensity, such as graz- ing to a low herbage mass, will slow growth rate more than a less intense defoliation. Implicit in Fig. 2 is a broader relation- ship than only the effect of residual herbage mass following grazing; we hypothesize that when all other factors are con- stant (e.g., climate, pasture species, and soil type), herbage mass can be used to predict herbage accumulation rate throughout the entire regrowth period. The relationship in Fig. 2 also shows the reduction of HAR i when herbage mass exceeds the optimum point, as might occur if forage was not harvested. Although prior work has validated use of sigmoid equa- tions to model forage production, few studies have quantified variation in the equation parameters during a growing season. Radiation, temperature and reproductive development will change over time (and during a regrowth period), and Thornley and France (2005) propose modifications to logistic equations to account for environmental and nutritional factors. The objectives of this study were (i) to find the most suitable (four-parameter) sigmoid equations to fit measured herbage mass accumulation, and (ii) investigate how the patterns of pasture growth (i.e., equation parameters) varied with time of year and location. In contrast to prior modeling work that has followed theFig. 1. (a) A typical Gompertz curve of above-ground herbage pattern of herbage accumulation over time (i.e., confounded withmass (h) for a 180-d growth period (t = days of growth), showing changing temperature, soil moisture, and reproductive status), wefour phases of the sigmoid growth curve, and (b) instantaneous propose to develop equations from plots with different herbageherbage accumulation rate (hAri) (slope of Fig. 1a). mass (and consequently different HARi) on the same date.linear relationship between LAI and herbage mass (Brougham, MAteriAls And Methods1956; Duru, 1989), herbage mass can be used as a surrogate for sitesLAI for modeling and pasture management. The LAI describes Measurements were conducted at three north-central Unitedthe photosynthetic apparatus of a sward and its ability to fix States locations during 2008: Columbus and Coshocton, OH,carbon for growth. In addition, herbage mass describes the size and Arlington, WI. The Columbus site was located at the Ohiotable 1. some common sigmoid equations and their respective “rate of change” functions. equation herbage mass† instantaneous herbage accumulation rate source dy bt  1  = bH ∆ e ae ln  aebt  [2] bt dt e  Richards (1959)Gompertz‡ H = H ∆ e ae + Hmin [1] Draper and Smith (1981) dy  H∆  = b ( H − Hmin ) ln  [3] dt ( H − Hmin )   H∆ dy H ∆ be − a−bt Eq. [4] in Landsberg (1977)Symmetric logistic H= + Hmin = 1 + e ( − a −bt ) dt (1 + e − a −bt )2Symmetric logistic H∆ dy H ∆ abe − bt Eq. [5] in Landsberg (1977) H= + Hmin =(or autocatalytic) 1 + ae − bt dt (1 + ae − bt )2 Richards (1959) dy abt b−1 = H∆ H∆ dt (1 + at b )2Asymmetric logistic H= + Hmin Cacho (1993)   ( H − H min )  H ∆ − H  b − 1 2 1+ at −b dy 1 = −ba b   H−H  b dt H∆  min  H = H (1 − e )+H b b −1 t t b dy  b  t  − Weibull −  a =    e a Hunt (1982) ∆ min dt  a  a † H = herbage mass (or yield); Hmin = the lower asymptote for herbage mass (i.e., minimum residual); Hmax = the upper asymptote for herbage mass (i.e., ceiling mass); H∆= the difference between Hmax and Hmin = Hmax – Hmin; a and b = curvature or shape coefficients; t = time (days of growth).‡ Equations modified by adding Hmin to account for the initial herbage mass.850 Agronomy Journal • Volume 102, Issue 2 • 2010
  3. 3. State University Donn Scott Airport, Columbus OH (40º04´ N,83º05´ W) in pasture that had been mowed to maintain a heightof 10 to 20 cm for the previous 2 to 3 yr. The average botanicalcomposition, determined by physical separation of five samples on12 Aug 2008, was 73% tall fescue [Schedonorus phoenix (Scop.)Holub, formerly Festuca arundinacea Schreb.], 15% Kentuckybluegrass (Poa pratensis L.), 2% white clover (Trifolium repens L.),red clover (T. pratense L.), and 10% other grasses and weeds. Thesoil was a Kokomo silty clay loam, 0 to 5% slope, a fine, mixed,superactive, mesic Typic Argiaquolls. The soil had a pH of 6.8,3.8% organic matter, 86 mg P kg–1 soil, and 233 mg K kg–1 soil.Nitrogen fertilizer was applied on 9 Apr. 2008 at 47 kg N ha–1 asNH4NO3 and on 3 June 2008 at 56 kg N ha–1 as urea. Herbage accumulation at Columbus was measured during 11 Fig. 2. the time-independent relationship between instantaneousgrowth periods, with the first and last periods commencing 8 Apr. herbage accumulation rate (hAri) (from Fig. 1b) and herbageand 9 Sept. 2008, respectively (Table 2, Fig. 3a). Herbage mass was mass above ground-level (from Fig. 1a). the maximum instantaneous herbage accumulation rate (hAri-max) was 33.1measured approximately weekly, beginning 8 Apr. 2008 and end- kg dM ha–1 d–1 and the critical range of herbage mass for >90% ofing 5 Nov. 2008, when all plots were harvested with a flail mower. maximum instantaneous herbage accumulation rate (hAri-90%)Plots were 4.0 by 9.3 m, with four replicates in a randomized (29.8 kg dM ha–1 d–1) was between 2760 and 4170 kg dM ha–1.complete block design. For the first growth period, the first twomeasurements (early April) showed decreasing herbage mass that were mowed to 5.5 cm at commencement of their respectivewas attributed to decay of remnant dead vegetation from winter growth period. The first four periods were harvested after 3 to(dead matter was 77% of herbage mass on 8 Apr. 2008; 41% on 4 mo growth since it was assumed pastures might have reached6 May 2008, n = 5) and those points were omitted from analysis. ceiling herbage mass, but subsequent analysis of the data showedExcept for the first growth period, which was not mowed, all plotsfor subsequent growth periods were mowed to 7.5 cm at com- table 2. starting date and ending date (harvest) for 11 growthmencement of the respective growth period. The first four periods periods, and the total herbage mass above ground level (kg dM ha –1) measured by rising plate meter (rPM) and mow-were harvested after 3 to 4 mo growth since it was assumed pastures er, at columbus and coshocton, oh (mean of four replicates).might have reached ceiling herbage mass, but subsequent analysis Mower starting harvest rPMof the data showed that plots may have been accumulating herbage date (2008) date (2008) total stubble† harvested totalmass after 4 mo, and the last seven growth periods were allowed kg DM ha–1to grow until they were harvested on 5 Nov. 2008, at 7 cm stubble Columbusheight. At Columbus, the four initial growth periods were 86 to 8 Apr.‡ 17 July 4933§ 1821 5190 7011100 d, and subsequent growth periods were 57 to 155 d (Table 2). 22 Apr. 17 July 4719§ 1828 5003 6831 The Coshocton site was located at the USDA-ARS North 6 May 12 Aug. 4714§ 2150 3673 5823Appalachian Experimental Watershed, Coshocton OH, 19 May 21 Aug. 4048§ 2194 2896 5090(40º21´51˝ N, 81º46´56˝ W) in pasture that had been in 3 June 5 Nov. 5081 2652 2445 5097intermittent hay production and grazing for 3 to 4 yr. The 18 June 5 Nov. 4598 2652 1823 4475average botanical composition, determined by physical separa- 2 July 5 Nov. 4188 2652 1148 3800tion of four samples on 6 Nov. 2008, was 76% tall fescue, 4% 17 July 5 Nov. 4035 2652 1100 3752Kentucky bluegrass, 10% white and red clover, and 10% other 30 July 5 Nov. 3542 2652 717 3369grasses and weeds. The soil was a Gilpin silt loam, 0 to 10% 12 Aug. 5 Nov. 3105 2652 646 3298slope, mixed, active, mesic Typic Hapludults. The soil had 9 Sept. 5 Nov. 3290 2652 587 3239a pH of 6.6, 2.8% organic matter, 234 mg P kg–1 soil, and Coshocton117 mg K kg–1 soil. Nitrogen fertilizer was applied as urea on 8 Apr.‡ 20 June 4459§ 1557 3464 502116 Apr. and 5 June 2008 at 47 and 80 kg N ha–1, respectively. 24 Apr. 3 July 4025§ 1128 3975 5103 Herbage accumulation at Coshocton was measured during 9 May 14 Aug. 5284§ 1408 4916 632411 growth periods, with the first and last periods commenc- 20 May 29 Aug. 4163§ 1459 4614 6073ing 8 Apr. and 11 Sept. 2008, respectively (Table 2, Fig. 3b). 6 June 6 Nov. 5422 2536 3180 5716Herbage mass was measured approximately weekly, begin- 20 June 6 Nov. 4581 2536 1907 4443ning 8 Apr. 2008 and ending 6 Nov. 2008, when all plots were 3 July 6 Nov. 4104 2536 1673 4209harvested with a flail mower. Plots were 4.0 by 8.0 m, with four 18 July 6 Nov. 3209 2536 519 3055replicates in a randomized complete block design. For the first 29 July 6 Nov. 2972 2536 414 2950growth period, the first two measurements (early April) showed 14 Aug. 6 Nov. 2700 2536 327 2863decreasing herbage mass that was attributed to decay of remnant 11 Sept. 6 Nov. 3131 2536 400 2936dead vegetation (dead matter was 90% of herbage mass on 8 Apr. † Measured by calibrated RPM.2008; 27% on 9 May 2008, n = 5) from winter and those points ‡ Not mowed from the prior winter (average 2439 and 2796 kg DM ha –1 atwere omitted from analysis. Except for the first growth period, Columbus and Coshocton, respectively).which was not mowed, all plots for subsequent growth periods § Plots lodged.Agronomy Journal • Volume 102, Issue 2 • 2010 851
  4. 4. 7.5 cm and 50 kg N ha–1 was applied as NH4NO3. Herbage accumulation at Arlington was measured during 12 growth periods, with the first and last periods commencing 1 May and 18 Sept. 2008, respectively (Fig. 3c). Plots were 2.0 by 6.0 m, with four replicates in a random- ized complete block design. Except for the first growth period, which was not mowed, all plots for subse- quent growth periods were mowed to 7.5-cm height at commencement of their respective growth period. Herbage mass was measured approximately weekly during the period 1 May to 30 Oct. 2008. At Arlington, the growth periods ranged from 41 to 99 d. Field Methods Herbage mass was measured approximately weekly at each site using a RPM (Ashgrove Pasture Plate, Ashgrove Industries, Ash- hurst, NZ) (Vartha and Matches, 1977). Calibration details are described in detail by Ferraro et al. (2009). Briefly, at each measurement date, 5 to 10 calibration samples were collected that comprised a RPM reading and the vegetation (clipped to ground level) within the 0.1 m2 RPM area. The calibration samples were selected at random to represent the range of vegetation mass present, and included short and tall areas. Subsequent analysis showed no significant differenceFig. 3. Average above-ground herbage mass and the associated Gompertz curves for growth between stubble and leafy vegeta-periods beginning on various dates at (a) columbus, oh, (b) coshocton, oh, and (c) Arlington, tion and a single calibration wasWi. symbols are the average of four replicates. Alternating closed and open symbols are used used for pre- and post-harvestto distinguish sequential growth periods. swards. Clipped samples were driedthat plots may have been accumulating herbage mass after 4 mo, at 60ºC for 48 h. A regressionso the last seven periods were allowed to grow until they were (calibration) equation for each measurement date was calculatedharvested on 6 Nov. 2008, to 7 cm stubble height. At Coshoc- using the calibration data from the sample date and the preced-ton, the four initial growth periods were 73 to 101 d, and subse- ing sample date, to reduce variation. Previous analysis (Ferraro etquent growth periods were 56 to 153 d (Table 2). al., 2009) had shown the intercept was not significantly different The Arlington site was at the University of Wisconsin Arling- from zero, and linear equations were forced through the origin.ton Agricultural Research Station (43º18´ N, 89º21´ W) in a At Columbus and Coshocton, herbage mass was measuredmonoculture of meadow fescue [Schedonorus pratensis (Huds.) using a plot harvester at the conclusion of each growth periodP. Beauv., formerly F. pratensis Huds. cv. ‘Pradel’] that had been (harvest dates in Table 2). At each harvest, herbage mass (aboveseeded in 15-cm rows on 2 May 2007. This pasture was mechani- mowing height) was measured in a 1.1 by 8.0 m strip in thecally harvested three times during 2007. The soil was a Plano silt center of each plot. Harvested mass was calculated from theloam, well-drained, fine-silty, mixed, superactive, mesic Typic harvested FW and the DM percentage of a subsample that wasArgiudoll. The soil nutrient concentrations to 15-cm depth were dried at 60ºC for 48 h. The remaining stubble was measured130 mg K kg–1 soil, 26 mg P kg–1 soil, pH 6.8, organic matter with the calibrated RPM. Total final plot herbage mass was the3.4%. At the start of each growth period, plots were mowed to total of harvested and stubble mass.852 Agronomy Journal • Volume 102, Issue 2 • 2010
  5. 5. statistical Analysis the literature, and have simpler mathematical computation than Herbage mass (average from four replicates for 5 to 10 mea- for other equations. The Gompertz curves were used to showsurement dates after defoliation to a low residual height) was fit the accumulation of measured herbage mass over time (averageto sigmoid equations (Fig. 1a) using PROC NLIN in SAS (SAS for four replicates) (Fig. 3). The slope (HAR i) was calculatedfor Windows V 9.1, SAS Institute, Cary, NC). Models were fit for each experimental unit (plot) at each site (132 equations infor symmetric logistic, Gompertz and Weibull functions (Table total) for use in predicting the HAR i–herbage mass curves.1) with the model having the lowest error mean square being Predicted HAR i and measured herbage mass were fit to theidentified as the best fit to the data. PROC NLIN used the option HAR i–herbage mass equation (Table 1, Eq. [3]) on 25 dates perMethod = Newton, since this had the most reliable convergence; site (Table 3). On approximately 33% of dates, PROC NLIN washowever, Method = Gauss and Method = Marquadt also were unable to converge on a realistic result and a simplified modelalmost as reliable in obtaining convergence. Differences in the final (with two parameters) was used by forcing the equation throughresults of those methods were negligible. Parameter estimation by HAR i = 0 at the average Hmin for each site (1665, 1345, andPROC NLIN had less error when a three parameter model (HΔ, 1360 kg DM ha–1 for Columbus, Coshocton, and Arlington,a, and b) was used (rather than four parameters), and curve fitting respectively) (Table 3). Unreliable parameter estimates werewas simplified by assigning Hmin as the lowest herbage mass mea- obtained for eight dates and were omitted from Table 3. Reasonssured (always within the first three herbage mass measurements). for the inability to obtain parameter estimates included, (i) For each date on which herbage mass was measured (25 dates insufficient data at high herbage mass early in the growing seasonat approximately 1-wk intervals for each site), the measured (April), (ii) insufficient data at low herbage mass late in the grow-herbage mass and the calculated HAR i (calculated for that plot ing season (September), and (iii) the failure of PROC NLIN toon that date using the Gompertz equations determined above) converge (even for a reduced, two-parameter model).were fit to the time-independent, HAR i–herbage mass equa- Some of the parameters for the HAR i–herbage mass equationstion (Fig. 2; Table 1, Eq. [3]) using PROC NLIN in SAS (SAS varied considerably during the growing season (Table 3). Thefor Windows V 9.1, SAS Institute, Cary, NC). Each data point HΔ parameter showed the greatest seasonal variation. Values forcomprised one observation on one plot and all replicates were HΔ were low in spring (mean = 3688 kg DM ha–1), increased toused for the curve fitting (6–31 points per analysis). The param- their maximum during late-May to June (mean = 6305 kg DMeters estimated by PROC NLIN were Hmin, HΔ, and b. The best ha–1), and decreased to their lowest values during August–Sep-model used was the one with lowest error mean square. Approxi- tember (mean = 3242 kg DM ha–1), except for a slight increasemate standard errors for equation parameters were predicted by in September–October at Columbus and Coshocton. The bNLIN. The maximum instantaneous herbage accumulation rate parameter [Table 1, Eq. [3]] described the shape of the Gompertz(HAR i-max), the optimum herbage mass (at which HAR i-max curve and varied seasonally at the three sites. The highest valuesoccurred) and the critical range of herbage mass for >90% of for b occurred in May, when the growth rate was highest (mean =maximum instantaneous herbage accumulation rate (HAR i-90%) 0.089). The lowest values for b occurred in late summer (August)were calculated for each equation using MS-Excel. (mean = 0.013), and increased slightly in autumn (September– October, mean = 0.020). Within each location, the parameter results Hmin did not vary appreciably during the growing season. climate The parameters for the HAR i–herbage mass equations varied Climatic data were measured within 1 km of each site (data among the three sites (Table 3). Values for HΔ were similar fornot shown). Rainfall was adequate for pasture growth at all Columbus and Coshocton, but were slightly higher for Columbussites from April through July, and averaged 125 mm mo–1, 28% during June. The HΔ values were generally lower at Arlingtonabove the 30-yr average (data not shown). Conversely, August than in Ohio. Values for b were similar for the two Ohio sites,to October rainfall averaged 41 mm mo–1, 50% of the 30-yr but were much lower than for Arlington. Values for Hmin wereaverage, and probably limited pasture growth. At Coshocton, slightly greater in Columbus (1765 kg DM ha–1) than Arlingtonthe April to October 2008 mean air temperature equaled the or Coshocton (1360 and 1345 kg DM ha–1, respectively).30-yr average, but Columbus and Arlington were 0.6 and Four important values with practical application were calculated1.1ºC below average, respectively. The average April to October for each week at each site during the growing season (Table 4).2008 air temperature at Columbus, Coshocton, and Arlington The highest values for maximum HARi (HARi-max) at Arling-was 17.8, 17.7, and 14.7ºC, respectively. ton occurred during May (176.8 kg DM ha–1 d–1), and in Ohio occurred during June (86.8 and 66.2 kg DM ha–1 d–1 at Colum- curve Fitting bus and Coshocton, respectively). The HARi-max decreased during Forage accumulation was reliably predicted by all sigmoid the growing season, and the lowest values were usually observedgrowth equations, but was a better fit for the asymmetric equa- during October at each site. The HAR i-max was greatest at Arling-tions than symmetric equations (data not shown). On average ton, intermediate at Columbus and lowest at Coshocton. Thefor 34 dates and locations, the average r 2 and error mean square optimum herbage mass (at HARi-max) also varied between seasonsfor the symmetric logistic equation was 0.88 and 1.38 × 105, and sites, being greatest during summer in Ohio (5400 and 5700and for the Gompertz equation was 0.99 and 6.7 × 104, respec- kg DM ha–1 at Columbus and Coshocton, respectively), and leasttively. There was no appreciable difference in the goodness of in early spring and late fall at all sites (mean = 2835 kg DM ha–1).fit among the asymmetric equations (Gompertz, Weibull, and Of potential interest to pasture managers is the range ofasymmetric logistic). All subsequent analysis was done using herbage mass (maximum and minimum) that ensures HAR iGompertz equations since these are more commonly used in remains within 90% of HAR i-max (Fig. 2, Table 4). This rangeAgronomy Journal • Volume 102, Issue 2 • 2010 853
  6. 6. table 3. Parameters for instantaneous growth rate (hAri)–herbage mass curves (table 1, eq. [3], hmin, hΔ , and b), their standarderrors, and r 2 for three sites and various observation dates during 2008 (n = 6 to 31). date hmin Approx. se hΔ Approx. se b Approx. se r2 kg DM ha–1 Columbus22 Apr. 1665† na 4797.3 19811.4 0.015 0.024 0.8629 Apr. 1665† na 4023.6 4940.2 0.019 0.013 0.906 May 1665† na 4948.9 3146.0 0.020 0.008 0.9415 May 1500.8 119.1 5779.0 2220.3 0.024 0.008 0.9819 May 1475.3 41.3 4521.4 1111.3 0.033 0.006 0.9929 May 1695.9 82.8 6467.4 2799.8 0.031 0.010 0.973 June 1821.0 113.9 8851.4 3660.2 0.024 0.007 0.9611 June 1893.3 130.0 9586.5 5940.1 0.025 0.010 0.949 July 1665† na 6271.8 2118.8 0.018 0.004 0.9630 July 1665† na 9438.0 1577.1 0.011 0.001 0.998 Aug. 1492.6 155.9 8832.6 3002.2 0.010 0.004 0.9912 Aug. 1889.7 34.0 7322.6 516.5 0.013 0.011 0.9821 Aug. 1919.7 44.2 5976.6 547.1 0.016 0.012 0.9727 Aug. 1510.5 34.2 2048.7 86.5 0.027 0.003 0.883 Sept. 1659.7 124.8 3979.0 1064.0 0.020 0.005 0.9710 Sept. 1799.8 156.1 3739.2 775.2 0.022 0.006 0.9119 Sept. 1665† na 4063.6 482.7 0.021 0.003 0.9025 Sept. 1665† na 4421.5 524.0 0.019 0.003 0.9230 Sept. 1329.9 1112.3 6523.2 3017.4 0.012 0.006 0.96 Coshocton24 Apr. 1348.3 133.2 1392.2 300.4 0.034 0.012 0.961 May 1345† na 1784.5 539.7 0.034 0.012 0.849 May 1933.0 160.1 2250.2 867.6 0.045 0.023 0.9413 May 1345† na 3820.1 1716.3 0.028 0.010 0.9320 May 1345† na 5685.3 2596.8 0.025 0.008 0.9627 May 1345† na 7764.2 2160.4 0.021 0.004 0.9713 June 1345† na 7503.2 1581.0 0.023 0.004 0.9720 June 1345† na 11913.3 2305.0 0.015 0.002 0.9727 June 1567.3 98.9 5891.6 798.3 0.024 0.003 0.993 July 1345† na 6056.6 1589 0.022 0.004 0.9711 July 1345† na 6940.8 1874.9 0.019 0.004 0.9718 July 1345† na 9101.3 2209.2 0.014 0.002 0.9824 July 1345† na 9485.6 1879.8 0.013 0.002 0.9829 July 1345† na 7935.3 1468.9 0.014 0.002 0.987 Aug. 1345† na 7745.8 896.1 0.013 0.001 0.9814 Aug. 1320.5 63.4 8283.5 810.4 0.011 0.009 0.9820 Aug. 1416.6 75.2 6040.4 1066.6 0.015 0.003 0.9429 Aug. 1124.1 102.8 7988.2 721.1 0.011 0.001 0.965 Sept. 954.7 142.4 4926.9 902.9 0.017 0.003 0.9811 Sept. 1345† na 3462.5 325.3 0.027 0.003 0.8918 Sept. 1345† na 5295.9 330.9 0.014 0.001 0.9926 Sept. 1345† na 5094.6 485.9 0.016 0.002 0.912 Oct. 1098.8 421.8 6376.7 1192.4 0.011 0.002 0.97 Wisconsin7 May 1360† na 2343.1 324.6 0.189 0.024 0.9914 May 1246.9 17.7 6077.1 749.7 0.079 0.009 1.0021 May 1602.7 70.2 5578.4 1241.7 0.061 0.019 0.9928 May 746.6 252.3 8137.6 1860.0 0.024 0.007 0.964 June 1360† na 4088.2 283.2 0.060 0.012 0.7211 June 1518.6 152.1 4516.6 311.5 0.051 0.008 0.8218 June 1421.7 250.4 4515.9 468.6 0.051 0.009 0.8525 June 1403.5 32.4 3299.3 277.2 0.108 0.017 0.932 July 1418.9 93.5 4444.0 304.5 0.062 0.007 0.989 July 1488.0 0.0 4070.9 169.3 0.052 0.005 0.9616 July 1360† na 3846.2 291.7 0.039 0.006 0.8423 July 1457.3 124.9 3525.2 386.5 0.047 0.012 0.6530 July 1274.4 332.1 4238.0 603.2 0.018 0.005 0.716 Aug. 1360† na 3192.6 208.8 0.017 0.003 0.8413 Aug. 1322.5 42.1 1929.4 84.8 0.035 0.005 0.9120 Aug. 1399.0 1.2 2206.6 72.6 0.025 0.002 0.9027 Aug. 1483.6 75.9 2084.0 160.0 0.022 0.004 0.773 Sept. 1289.1 53.3 1789.0 111.7 0.027 0.003 0.9211 Sept. 1360† na 1815.9 104.4 0.030 0.003 0.869 Oct. 1360† na 1475.2 125.8 0.024 0.005 0.76† HARi-herbage mass equation was forced through a fixed Hmin for that specific site since there was insufficient data for a three parameter model; there was no applicable standard error.854 Agronomy Journal • Volume 102, Issue 2 • 2010
  7. 7. varied considerably during the year and between locations. The The asymmetric logistic equations were a better fit tominimum herbage mass was similar among the three sites (mean measured herbage mass data than the symmetric equations. In= 2590 kg DM ha–1), but was higher in June–July at Columbus every case, the rate of increasing pasture growth rate (below(mean = 3625 kg DM ha–1) than in spring or autumn, or in any optimum herbage mass) was greater than the rate of decreas-season at Arlington. Recommendations for maximum herbage ing pasture growth rate (above the optimum herbage mass).mass varied considerably between seasons and sites, and were Presumably the processes for initial growth following defolia-relatively constant at Arlington (mean = 3340 kg DM ha–1), but tion (use of stored carbohydrates, leaf extension, and initiationwere much higher in summer (mean = 5965 kg DM ha–1) than of new leaves and tillers) were more rapid than the processesspring or fall in Ohio (mean = 3990 kg DM ha–1). leading toward growth suppression (leaf shading, loss of tiller At Columbus and Coshocton, all plots were harvested at the density, and leaf senescence and death). Ecologically, thoseconclusion of the study (Table 2). We found close agreement plants able to show rapid initial growth after defoliation mightbetween the herbage mass estimates from the RPM and the forage have an advantage over their slower neighbors.harvester (harvester–herbage mass = 0.94 × RPM–herbage mass + One practical implication of the asymmetric HAR i–herbage290, r2 = 0.95, P > 0.001), except for the first four growth periods mass relationship (Fig. 2) is that at low herbage mass (below thewhen considerable lodging of reproductive material had been optimum herbage mass), the relationship between herbage massobserved. Since we used a single RPM calibration for all plots, we and HAR i is steeper than at high herbage mass. Thus, belowhad more confidence in the harvester than the RPM data for the the optimum herbage mass, a small change in herbage mass (saylodged plots, and included the total herbage mass measured by the 500 kg DM ha–1) will have a greater effect on HAR i than atharvester (harvested + stubble herbage mass) during curve fitting. high herbage mass. Two implications of this are (i) an error in estimating herbage mass could have a greater effect on HAR i discussion at low than high herbage mass, and (ii) the effect of intensive The primary finding from this study was that herbage mass defoliation could be to reduce HAR i more severely than thecan be used to predict herbage accumulation rate when all effect of failure to control surplus herbage mass.other factors such as climate, pasture species, and soil type areconstant. For every date and location measured, the HAR i– effects of season and locationherbage mass relationship closely fit the time-independent form The Gompertz equation parameters varied during the grow-of the modified Gompertz equation (Table 1, Eq. [3]). The only ing season and among locations. Additional research is requiredexception occurred in April, when the initial growth following to develop a broader suite of parameters for specific locations.winter made it biologically unfeasible to test the effect of high Alternatively, there may be potential for the approach ofherbage mass. Even in this case, the strong positive relation- Thornley and France (2005) to add parameters to a logisticships that were found were consistent with a positive effect of model to specifically accommodate effects such as seasonality.herbage mass on HAR i below the optimum herbage mass. The Gompertz equations are relatively simple, requiring as few These results emphasize the importance for pasture managers as five points to fit a curve and can be developed relatively easilyto monitor farm herbage mass. Herbage mass is a fundamental to predict HAR i for specific locations.measure of a production system. First, measurements of average The values for HAR i-max (Table 4) were consistent withherbage mass for a farm (cover) can be used to ensure herbage is growth rates that occur within the locations measured. Arling-being appropriately utilized and is not being over- or under-uti- ton had the highest HAR i-max of any date or location (176.8 kglized by grazing livestock. Second, measurements of herbage mass DM ha–1 d–1 on 14 May 2008), and had higher average HAR ibefore and after an area is grazed can be used to calculate livestock than the Ohio sites during May and July. Arlington HAR i-maxintake (by the method of forage disappearance) (Macoon et al., was only half the Ohio sites in August and September. A shorter,2003). In addition to these two applications, the HAR i–herbage more intense growing season is typical for more northernmass curves, in conjunction with measurements of herbage mass latitudes. Total potential annual forage production calculatedallow a manager to ensure that pastures are maintained within for each location from HAR i-max (Table 4), the number of daysan acceptable range of herbage mass and avoid any reduction of between HAR i-max calculations, and totaled for all observationsgrowth rate due to excessive, or deficit mass. was 7830, 6880, and 10,080 kg DM ha–1 yr–1 for Columbus, In this study we measured total herbage mass and made no Coshocton, and Arlington, respectively. These yields reflect theconsideration of forage quality. We made no attempt to control relative fertility and forage species of each location. Arlingtonreproductive development during May and June, and the herbage had the best soil with a 1-yr-old meadow fescue pasture, Colum-mass that accumulated for the first four growth periods had sig- bus was of intermediate fertility with an old tall fescue–domi-nificant amounts of stem and dead material. The HAR i–herbage nant pasture, and Coshocton had the lowest soil fertility (lowmass curves have immediate relevance to applications that might soil K) also with tall fescue–dominant pasture.require maximum herbage mass, such as for ligno-cellulosic Seasonal growth curves frequently show a pattern of highenergy production. In many cases, these areas only have a single spring growth rate, a slump during summer, and a flush ofharvest at the end of the season. Belesky and Fedders (1995) have production during fall (Johnson and Parsons, 1985; Denisonshown that Gompertz equations are valid for warm-season (C4) and Perry, 1990). We found highest growth rates occurredspecies, and it is likely that herbage mass will be maximized with in spring, but did not see evidence of any flush of productionseveral harvests rather than a single end-of-season harvest. Mod- during fall. The climatic data (not shown) showed all threeeling could be used to compare the benefit of increased herbage locations had below average rainfall in autumn, that likelymass compared with the additional harvesting costs. prevented the autumn flush usually observed in north-centralAgronomy Journal • Volume 102, Issue 2 • 2010 855
  8. 8. table 4. Maximum instantaneous growth rate (hAri-max), United States. One implication of the HAR i-–herbage massthe optimum herbage mass (at hAri-max), and the minimumand maximum herbage mass for >90% of hAri-max for three curves (Fig. 2, Table 3) might be that high spring growth ratessites and various observation dates (see table 3 for Gompertz might be confounded with higher herbage mass that frequentlyequation parameters and statistics). occur at that time. Conversely, the reported “slump” in sum- optimum Min. herbage Max. herbage mer growth rate is also likely confounded with the low herbage herbage mass mass for >90% mass for >90% date hAri-max (at hAri-max) hAri-max hAri-max mass that usually occurs in summer. The seasonal pattern of forage growth rate observed at any location is not only affected kg DM ha–1 d–1 kg DM ha–1 Columbus by the prevailing climate, but is also the artifact of defoliation22 Apr. 26.1 3450 2700 4300 management and the resultant herbage mass (Johnson and29 Apr. 27.6 3200 2500 3900 Parsons, 1985; Belesky and Fedders, 1994).6 May 36.2 3500 2700 440015 May 51.9 3600 2800 4600 implications for use of Grazing exclosure cages19 May 54.5 3100 2500 390029 May 73.0 4100 3100 5200 One implication of this research relates to the interpretation of3 June 77.8 5000 3700 6400 herbage accumulation within grazing exclosure cages. Exclosure11 June 86.8 5400 4000 7100 cages are frequently used to measure the herbage accumulation rate9 July 41.5 4000 3100 500030 July 36.8 5100 3700 7200 on continually stocked pastures, that is, where herbage growth and8 Aug. 33.5 4700 3400 6000 removal occur simultaneously, such that the net result is a fixed12 Aug. 33.7 4600 3500 5800 herbage mass over time. Where the herbage mass is below the opti-21 Aug. 34.1 4100 3300 5100 mum for HARi-max, it can be concluded from the HARi–herbage27 Aug. 20.1 2300 2000 2600 mass curves that measured HAR within the exclosure cage will3 Sept. 28.8 3100 2600 380010 Sept. 31.4 3200 2700 3800 exceed the actual HAR under continuous stocking. Field et al.19 Sept. 31.9 3200 2500 3900 (1981) and Devantier et al. (1998) compared forage production25 Sept. 31.2 3300 2700 4100 predicted from livestock production with measurements using30 Sept. 27.8 3700 2800 4900 exclosure cages under continuous grazing, and found the measure- Coshocton24 Apr. 14.8 1850 1650 2050 ments overestimated forage production predicted from livestock1 May 22.0 2000 1700 2300 production by 33 and 55%, respectively. The difference between9 May 37.3 2800 2500 3100 measured pasture growth rate within an exclosure cage, and actual13 May 39.5 2800 2200 3400 pasture growth under continuous stocking will depend on the20 May 51.9 3400 2600 450027 May 58.8 4200 3000 5600 relative differences in actual herbage mass present. Using Fig. 3 as13 June 64.3 4100 2900 5500 an example, if pasture mass under continuous stocking was 210020 June 66.2 5700 3900 7800 kg DM ha–1, and average herbage mass within an exclosure cage3 July 48.6 3600 2600 4700 was 3500 kg DM ha–1, the exclosure cage technique could overesti-11 July 48.3 3900 2900 5100 mate the actual growth rate by 100%. An alternate case is possible,18 July 46.2 4700 3300 630024 July 43.6 4800 3400 6500 where exclosure cages could underestimate actual growth rates, in29 July 40.3 4300 3100 5600 the situation where a continuously grazed pasture might be at the7 Aug. 36.4 4200 3000 5600 optimum herbage mass, and accumulation of additional herbage14 Aug. 32.1 4000 2900 5200 mass might slow the measured growth rate.20 Aug. 32.7 3500 2700 450029 Aug. 31.6 3300 2500 4300 implications for rotational5 Sept. 29.9 2800 2100 360011 Sept. 34.7 2600 2100 3200 and continuous stocking18 Sept. 27.3 3300 2500 4200 Among the greatest controversies within the forage industry is26 Sept. 30.5 3200 2400 41002 Oct. 25.1 3600 2600 4700 the debate about the effect of rotational and continuous stocking Wisconsin on forage production. Many recommendations are for pastures7 May 162.9 2200 1900 2600 to be rotationally rather than continuously grazed; however,14 May 176.8 3500 2600 4500 research does not always find a production advantage in support21 May 126.0 3700 2800 4600 of this recommendation (Briske et al., 2008). There are many28 May 70.6 3700 2500 52004 June 89.9 2900 2300 3600 reasons for use of either rotational or continuous stocking man-11 June 84.9 3200 2500 3900 agement, other than maximizing herbage mass (e.g., effects on18 June 84.2 3100 2400 3900 forage quality, avoidance of selective defoliation, etc.); however,25 June 131.4 2600 2120 3200 most managers will aim to ensure high herbage mass production.2 July 100.8 3100 2370 38509 July 77.1 3000 2400 3700 The HAR i–herbage mass curves suggest that pasture growth16 July 55.0 2800 2200 3400 can be maximized by maintaining herbage mass at the optimum23 July 61.4 2800 2220 3370 herbage mass (noting this varies during the season), which could30 July 27.3 2800 2200 3600 be achieved by continuous, but variable, stocking (Johnson and6 Aug. 19.7 2500 2000 310013 Aug. 24.8 2000 1740 2350 Parsons, 1985). However, recommendations should not neces-20 Aug. 20.5 2200 1900 2600 sarily recommend continuous stocking per se, since continuous27 Aug. 16.9 2300 2000 2600 stocking at a herbage mass other than the optimum (either3 Sept. 17.7 1900 1680 2300 over or under) could result in lost production. One benefit of11 Sept. 20.1 2000 1800 2300 rotational stocking is that the variation in herbage mass might at9 Oct. 13.1 1900 1700 2100856 Agronomy Journal • Volume 102, Issue 2 • 2010
  9. 9. some stage, be at the optimum herbage mass. Lax or infrequent no. 2006-55618-17025; Wisconsin Department of Agriculture, Tradeharvesting (allowing high herbage mass) or intensive defoliation and Consumer Protection GLCI grant no. 831-3; and USDA CSREES(resulting in low herbage mass) will both result in lost potential NCR-SARE grant number 2007-38640-18363.for forage production. One conclusion from the HAR i–herbagemass relationship obtained in this study is that it is not so much reFerencesthe forage defoliation method (rotational vs. continuous) that Belesky, D.P., and J.M. Fedders. 1994. Defoliation effects on seasonal productionaffects overall forage production, but the result of defoliation on and growth rate of cool-season grasses. Agron. J. 86:38–45.herbage mass that is the primary issue. Belesky, D.P., and J.M. Fedders. 1995. Warm-season grass productivity and growth The effect of deviations of herbage mass from the optimum rate as influenced by canopy management. Agron. J. 87:42–48.for HAR i-max is clearly shown in the HAR i–herbage mass Bluett, S.J., C. Matthew, G.J. Bishop-Hurley, S.J. Haslett, and J. Hodgson. 1998.curves. Small departures will have a negligible effect on HAR i, The relationship between herbage mass and pasture accumulation rate in win-and allow scope for application of rotational stocking strate- ter. N. Z. J. Agric. Res. 41:299–305.gies that might suit specific management requirements. We Briske, D.D., J.D. Derner, J.R. Brown, S.D. Fuhlendorf, W.R. Teague, K.M. Havs- tad, R.L. Gillen, A.J. Ash, and W.D. Willms. 2008. Rotational grazing onpropose an arbitrary 90% of HAR i-max as being a reasonable rangelands: Reconciliation of perception and experimental evidence. Range-range for herbage mass that might allow for practical guidelines land Ecol. Manag. 61:3–17.of grazing management (Table 4). Of interest is that the upper Brougham, R.W. 1956. Effect of intensity of defoliation on regrowth of pasture.limit for herbage mass is greater than what is usual for grazing Aust. J. Agric. Res. 6:377–387.management recommendations in Ohio. These upper values Cacho, A.J. 1993. A practical equation for pasture growth under grazing. Grass For-do not consider any effect on forage quality. Any accumulation age Sci. 48:387–394.of reproductive seedheads would likely increase herbage mass, Denison, R.F., and H.D. Perry. 1990. Seasonal growth rate patterns for orchard-but be detrimental to forage quality, and additional research is grass and tall fescue on the Appalachian Plateau. Agron. J. 82:869–873.required to determine the dynamics of accumulation of digest- Devantier, B.P., M.G. Lambert, I.M. Brookes, and C.L. Hawkins. 1998. Measur-ible herbage mass rather than total herbage mass. It is likely ing production of continuously grazed hill pastures. Proc. of the N. Z. Grassl. Assoc. 60:157–160.that the herbage mass targets for maximum HAR will vary Draper, N.R., and H. Smith. 1981. An introduction to nonlinear estimation. Ch. 10. p.from the herbage mass targets for maximum digestible-HAR. 458–517. In Applied regression analysis. John Wiley & Sons, Hoboken, NJ. Duru, M. 1989. Variability of leaf area index extension rate on permanent grass- conclusions lands. p. 501–502. In Proc. XVI Intl. Grassl. Congress. Publ. Association Gompertz equations were found to accurately predict herb- Française pour la Production Fourragère.age accumulation patterns throughout the growing season at Ferraro, F.P., R.M. Sulc, D.J. Barker, R. La Guardia Nave, F. Lopes, and K.A. Albre-three north-central locations in the United States. Parameters cht. 2009. Seasonal effects on rising plate meter calibration for forage. In Proc.for the Gompertz equations varied during the growing season Am. Forage and Grassl. Council [CD]. AFGC, Elmhurst, IL.and among locations, and additional research is warranted to Field, T.R.O., D.A. Clark, and M.G. Lambert. 1981. Modelling of a hill country sheep production system. Proc. of the N. Z. Soc. of Anim. Prod. 41:90–94.quantify the factors that affect these terms. A time-indepen-dent expression of the Gompertz equation may have potential Hunt, R. 1982. Plant growth curves. The functional approach to plant growth anal- ysis. Edward Arnold, London.use for pasture management by defining the relationship Johnson, I.R., and A.J. Parsons. 1985. Use of a model to analyse the effects of con-between HAR i and herbage mass. This equation showed the tinuous grazing managements on seasonal patterns of grass production. Grassoptimum herbage mass at which HAR i was maximum, and Forage Sci. 40:449–458.values varied between 1600 and 4000 kg DM ha–1 depending Landsberg, J.J. 1977. Some useful equations for biological studies. Exp. Agric.on location and date. Allowing herbage mass to exceed the 13:273–286.optimum point (e.g., delayed harvest), or harvesting to below Lemaire, G., and D.F. Chapman. 1996. Tissue flows in grazed plant communities.the optimum point, will reduce the HAR i. The HAR i–herbage p. 3–36. In J. Hodgson and A.W. Illius (ed.) The Ecology and management ofmass curves define a range of herbage mass within which pas- grazing systems. CAB Intl., Oxfordshire, UK.tures can be managed to achieve high HAR i, and maintaining Macoon, B., L.E. Sollenberger, J.E. Moore, C.R. Staples, J.H. Fike, and K.M. Portier. 2003. Comparison of three techniques for estimating the forage intake of lactat-pastures within 90% of the maximum HAR i may be a practical ing dairy cows on pasture. J. Anim. Sci. 81:2357–2366.target for producers. The HAR i–herbage mass curves may be Parsons, A.J., I.R. Johnson, and A. Harvey. 1988. Use of a model to optimize the inter-a useful tool for modeling the effect of defoliation patterns on action between the frequency and severity of intermittent defoliation and toherbage accumulation rate, and annual forage production. provide a fundamental comparison of the continuous and intermittent defolia- tion of grass. Grass Forage Sci. 43:49–59. AcKnoWledGMents Parsons, A.J., S. Schwinning, and P. Carrere. 2001. Plant growth functions and pos- We are grateful to the managers at the USDA-ARS North Appalachian sible spatial and temporal scaling errors in models of herbivory. Grass Forage Sci. 56:21–34.Experimental Watershed (Jim Karr), OSU Donn Scott Airport (Greg Richards, F.J. 1959. A flexible growth function for empirical use. J. Exp. Bot. 10:290–300.Foggle, Martin Mussard, and Dale Gelter), and University of Wisconsin,Arlington for providing access and technical support at field sites. We Thornley, J.H.M., and J. France. 2005. An open-ended logistic-based growth func- tion. Ecol. Modell. 184:257–261.thank John McCormick for technical assistance. Partial financial sup- Vartha, E.W., and A.G. Matches. 1977. Use of a weighted-disk measure as an aid inport was provided by the National Research Initiative of the USDA sampling the herbage yield on tall fescue pastures grazed by cattle. Agron. J.Cooperative State Research, Education and Extension Service, grant 69:888–890.Agronomy Journal • Volume 102, Issue 2 • 2010 857

×