Two separate tests were performed, one to train the LAI and LG estimation models, assess genotypic differences, and analyze the relationship between LAI, LB, and pod yield; and the other for validation of the LAI and LG estimation models. Both tests were performed at the Virginia Tech Tidewater Agricultural Research and Extension Center (TAREC) in Suffolk, VA (latitude 36.66 N, longitude 76.73 W) (Fig. 1).Figure 1

Location of the experimental field in 2017. Each image shows the geographic location of the study using red box, which is then zoomed out to the next image. The red box in the last image is the actual esperimental field with peanut plots. The physical maps are

taken from google earth (https://earth.google.com/web) and the aerial image of the plots was created using Pix4Dmapper Version 4.2.26 software (Prilly, Switzerland).

Test 1 was conducted in 2017 using 18 genotypes (Table 1). These genotypes were selected based on economically desirable traits including pod yield, drought tolerance, and disease resistance. Genotypes were planted at a rate of 15 seeds m⁻¹ in 2-row plots, 2.13 m long and 1.83 m wide.Table 1

Genotypes planted in study 1 to estimate the LAI from leaf reflectance.

There were six replications arranged in a randomized complete block design (RCBD); the total plot area was 660 m²; and 108 total plots. At the physiological maturity, pod yield was measured for each plot.

Test 2 was planted on April 30, 2019. Eight peanut genotypes were selected from the US mini-core peanut germplasm collection⁶³ (Table 2). Genotypes were planted at a rate of 20 seeds m⁻¹, in single-row plots, 1.83 m long and 0.9 m wide. Each genotype was replicated 16 times in a RCBD. This test was used for model validation and included ruler-measured and RGB-derived LAI and LG at four times from June 17 to July 18 (Table 3). Each time, a different set of plots were used; therefore, the total number of available plots was 128, with a total area of 290 m².Table 2

Genotypes planted in study 2 for validation of study 1 model.

For both tests, the seed beds were tilled and uniformly raised to 15 cm height before planting. Plots were rainfed and supplemental irrigation was only applied if the rainfall was inadequate over a two-week period. The soil type was Eunola fine-loamy, siliceous, thermic Aquic Hapludults in 2017; and a Kenansville loamy sand in 2019. Both soils being sandy, the water holding capacity at 25 cm depth was 0.10 m m⁻³. Cultural practices, i.e. pest management and fertility, were performed as recommended by the Virginia Peanut Production Guide⁷⁷. Information on the dates of the ground and aerial data collection, the number of images within each flight, cumulative precipitation and growth degree day (GDD) related to the LAI and LG collection dates are presented in Table 3.

LAI measurements started 30 days after planting (DAP) using an AccuPAR® LP-80 PAR/LAI ceptometer (METER Group, Inc. USA). The instrument has two light sensors, one for the above and one for below canopy photosynthetic active radiation (PAR) reading. The below canopy sensor is an 80 cm bar with a total of eight sensors placed at equal distance on the bar. The above canopy sensor was fixed on the operator’s hat and worn flat during data collection always at the same height above the crop. The below canopy sensor was placed at the base of the plant, perpendicular to the row. Two readings per plot were taken from each row and averaged to provide plot LAI. The instrument used the above and below intercepted PAR to estimate LAI. LAI measurements were taken regularly until beginning pod stage at 50 DAP⁷⁸ (Table 3).

Measurements of LG were taken on the same dates as LAI. One peanut plant from each row was randomly selected, and the length of the longest lateral branch was measured from the base of the main stem using a wooden meter ruler. The length of the branches from both sides of the main stem were summed to obtain the LG in centimeters. LG values from both rows were averaged to obtain LG of each plot.

At the physiological maturity (16 WAP), peanut pods were dug using a Sweere C200 peanut digger, windrow dried and combined using Amadas 2110 two row peanut combine for every plot. Pod weight of each plot was measured in grams and then converted to kg ha⁻¹. Pod yield was calculated based on 7% seed moisture.

An AscTec® Falcon 8 octocopter UAV platform (Ascending Technologies, Germany) was used for collection of the RGB images. At the same time with ground LAI data collection, a Sony® α6000 digital camera [24.3-megapixel (6000 × 4000)] was used on the flight campaign to collect aerial images (Table 3). A Sony 20 mm f/2.8 camera lens was used to acquire images in JPEG format and true color bands (red, green, blue). The camera used had 24-bit radiometric resolution; other settings included auto mode for aperture and ISO, and shutter priority mode for shutter speed. The image compression setting was set at ‘fine’ having a 10:1 compression ratio.

The flight plan was based on waypoint navigation, on auto pilot at 20 m altitude with an image overlap of 75% forward and 90% sideways. The flight campaign was created in AscTec® Navigator 3.4.5 software (Ascending Technologies, Germany). The UAV used its built-in GPS (accuracy within 20 cm) to navigate, acquire nadir images, and coordinate recording of individual images. Images were orthomosaic in Pix4Dmapper Version 4.2.26 software (Prilly, Switzerland) to create the RGB field map. We used the ‘reflectance map’ option in ‘index calculator’ under ‘DSM, orthomosaic, and index’ step of Pix4D processing to create individual red, green, and blue reflectance maps (Fig. 2). The orthomosaced reflectance maps had spatial resolution of 0.47 cm.Figure 2

Flowchart of the process for aerial estimation of leaf area index (LAI) and lateral growth (LG). A RGB sensor is used to collect aerial images based on a flight plan. The aerial mages are used to recreate the whole experimental plot orthomosaic^*. Rectangular shapes are created over peanut rows in a fishnet^* layer. The fishnet is used to extract reflectance from each of the reg, green, and blue band. The reflectance are used to derive vegetation indices (VIs). The VIs are subjected to multiple linear regression and artificial neural network regression as predictors for LAI and LG. The different models derived in the process are Reg-1, Reg-2, ANN-1, and ANN-2 for LAI estimation; and Reg-3, Reg-4, ANN-3, and ANN-4 for LG estimation. *Orthomosaic was done using Pix4Dmapper Version 4.2.26 software (Prilly, Switzerland) and Fishnet was created using ArcMap (version 10.6) tool of the ArcGIS (ESRI, Redlands, CA).

The red, green, and blue reflectance orthomosaics were exported to ArcMap (version 10.6) tool of the ArcGIS (ESRI, Redlands, CA) where polygons including entire plant rows were designed, numbered, and collated into a single shapefile to create a fishnet (Fig. 3). The fishnet was used for all orthomosaics, and images from each flight campaign were geo referenced using ground control points in all four corners and in the center of the test. Zonal statistics option was used to extract the DNs. This process averaged the raster information of every pixel within each polygon to give the DN of red, green, and blue rasters (Fig. 2).Figure 3

Red–green–blue (RGB) orthomosaic of the 2019 peanut study plot with ground control points (GCPs) (the black and white checkered objects; 4 at the corners and one in the center). The orthomosaic^* includes the fishnet^* layer (yellow bordered polygons) as well. The panel on the left bottom with various shades from white to black is the reflectance calibration panel. Each individual peanut row is 1.83 m in length and two rows are 0.91 m apart (center to center). * Orthomosaic was done using Pix4Dmapper Version 4.2.26 software (Prilly, Switzerland) and Fishnet was created using ArcMap (version 10.6) tool of the ArcGIS (ESRI, Redlands, CA).

Calibration was performed using a reflectance panel with eight different shades from white to black (Fig. 3). The DNs of the eight shades were recorded for red, green, and blue rasters from each orthomosaic. During every flight the reflectance from each of the eight shades of the panel were measured using ASD HH2 Hand-held VNIR Spectroradiometer (Malvern Analytical, Malvern, U.K.). The DNs and reflectance from the panel were fitted using exponential regression models as suggested in a previous study⁷⁹ (Fig. 4).Figure 4

Regression curves of reflectance (x axis) vs digital numbers (DN) (y axis) from aerially taken red–green–blue (RGB) images over 2017 and 2019.

The models trained for red, green, and blue reflectance for 2017 were:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & {\mathrm{Equation\,A}}1\, \to \,red = 0.1263 \times 1.0091^{DNr} \\ & {\mathrm{Equation\,B}}1\, \to \,green = 0.1263 \times 1.0087^{DNg} \\ & {\mathrm{Equation\,C}}1\, \to \,blue = 0.1144 \times 1.0087^{DNb} \\ \end{aligned}$$\end{document}EquationA1→red=0.1263×1.0091DNrEquationB1→green=0.1263×1.0087DNgEquationC1→blue=0.1144×1.0087DNb

The models trained for red, green, and blue reflectance for 2019 were:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & {\mathrm{Equation\,A}}2\, \to \,red = 0.0212 \times 1.0169^{DNr} \\ & {\mathrm{Equation\,B}}2\, \to \,green = 0.0211 \times 1.0165^{DNg} \\ & {\mathrm{Equation\,C}}2\, \to \,blue = 0.0167 \times 1.0167^{DNb} \\ \end{aligned}$$\end{document}EquationA2→red=0.0212×1.0169DNrEquationB2→green=0.0211×1.0165DNgEquationC2→blue=0.0167×1.0167DNb where red, green, blue is the reflectance from the respective rasters;

DN_r, DN_g, and DN_b are the digital numbers from red, green, and blue rasters, respectively.

Using these equations, reflectance of each row from all orthomosaics were derived. The reflectance of the two rows of each plot was averaged to get the average reflectance value of the plot.

Six RGB-derived VIs were used in this study. They were the blue green index (BGI); red-green ratio (RGR); normalized plant pigment ratio (NPPR); normalized green red difference index (NGRDI); normalized chlorophyll pigment index (NCPI); and plant pigment ratio (PPR) (Table 4). The selection of VIs was based on their connection with leaf pigments and crop physiological traits^61,80–83. The VI, NPPR, was used first time in this study. It is derived using all three reflectances (red, green, and blue) which makes it more useful as rest Vis used have either of the two reflectances.Table 4

Vegetation indices derived from aerially collected red–green–blue (RGB) images of peanut plots in 2017 and 2019.

For the statistical analysis, Statistical Analysis Software (SAS) 9.4 (SAS Institute Inc., Cary, NC, USA.) package was used. Manually measured LAI and LG were correlated to the RGB-derived VIs using Proc CORR statement, and the root mean square error (RMSE) values were determined using Proc REG statement. Proc REG was used to perform multiple linear regression and derive the models for LAI and LG from the VIs. The ‘parameter estimate’ values of each VI from SAS output was used as coefficients of predictors in the models. Stepwise selection was performed using Proc GLMSELECT to select the best predictors for the models. Predicted residual error sum of squares (PRESS) statistic was used to determine the model efficiency from the coefficient of determination (the higher R², the better efficiency), and root mean square error (RMSE), Akaike test criterion (AIC), Bayesian information criterion (BIC), and average square error (ASE) (the lower RMSE, AIC, BIC, and ASE, the better efficiency). Analysis of variance (ANOVA) of measured and derived LAI and LG was performed using Proc GLM. Tukey’s honest significant difference (HSD) was used for genotype means separation at α = 0.05. For regression of estimated LAI and LG with pod yield, Proc REG was used. Graphs were built using graph builder tool of JMP® Pro 15.0.0 (SAS Institute Inc., Cary, NC, USA.).

For the ANN regression, WEKA (Waikato Environment for Knowledge Analysis, version—3.8.4, The University of Waikato, Hamilton, New Zealand) software was used. ‘Use training set’ option was selected in the ‘Multilayer perceptron’ function of the ‘Weka explorer’ to train the models. Three hidden layers were manually added having five, four, and three nodes; learning rate was set at 0.001; momentum at 0.99; and training time was set at 10,000 iterations (Fig. 5). Our methodology was based on previous studies that suggested that increase in number of hidden layers and nodes increase accuracy and enables the network to learn more complex problems⁸⁴. Our hypothesis was having a large first layer and following it up with smaller layers for better performance as the first layer can learn a lot of lower-level features that can feed into a few higher order features in the subsequent layers. LG, LAI, and VIs from 2017 were used for model training. Weka used back-propagation for machine learning of multi-layer classification to train the models and predict outputs. The derived models were saved and are available in a github repository. The derived models were further loaded to validate and re-evaluate the models using 2019 data.Figure 5

Neural network training models for leaf area index (LAI) and lateral growth (LG). The vegetation indices (VIs) are in green boxes as predictors and LAI and LG are in yellow boxes as predicted output. Each column of red dots represents each of the hidden layers, and each dot is a node of that layer. The neural network training was done in WEKA (Waikato Environment for Knowledge Analysis, version—3.8.4, The University of Waikato, Hamilton, New Zealand).

The authors declare that use of plants in the present study complies with international, national and/or institutional guidelines.

The measured average LAI values in 2017 varied from 0.8 to 2.6, whereas the values for 2019 varied from 1.5 to 5.8. The mean LAI was 1.6 in 2017 and 3.7 in 2019. LAI was negatively correlated to the blue reflectance (r = − 0.56; P < 0.0001) (Table 5). Pearson correlation values showed that within several calculated VIs, BGI (r = − 0.89; p < 0.0001), PPR (r = 0.91; p < 0.0001) and NPPR (r = 0.87; p < 0.0001), were best correlated to the ground measured LAI (Table 5). These VIs had RMSE values ranging from 0.27 to 0.28 for the LAI which was lower than for the other VIs. Stepwise selection retained BGI, PPR, NPPR, NGRDI, and NCPI as the best five predictors for LAI estimation.Table 5

Relationship between leaf area index (LAI) and lateral growth (LG) with leaf reflectance and vegetation indices.

The first regression model (Reg-1) was based on the sum of these predictions, i.e. BGI, PPR, NPPR, NGRDI, and NCPI, and had the highest R² (0.91) (Fig. 6), and lowest RMSE (0.33), ASE (0.10), AIC (− 79) and BIC (− 151).Figure 6

Comparison of manually taken leaf area index (LAI) using a ceptometer and derived LAI (y-axis) in 2017 (x-axis) using: Reg-1: LAI = 28.82 × BGI + 13.77 × PPR-7.91 × NGRD + 14.88 × NCPI + 25.86 × NPPR-39.74; Reg-2: LAI = 505.84 × (BGI × PPR × NPPR × NGRDI × NCPI) + 0.134; ANN-1: BGI, PPR, NPPR, NGRDI, and NCPI as predictors of LAI ; ANN-2: product of BGI, PPR, NPPR, NGRDI, and NCPI as predictors of LAI. Each point on the graph represents LAI of every genotype at different days after planting (DAP) averaged over six replications. The box and whisker plots represent the error statistic of estimation models at different DAP. In each box, the central mark is median, and the lower and upper edges denote the 25th and 75th percentile of errors respectively. The whiskers extend to the most extreme data points not considered outliers. Outliers not shown on the chart for clarity.

The first regression model (Reg-1) was based on the sum of these predictions, i.e. BGI, PPR, NPPR, NGRDI, and NCPI, and had the highest R² (0.91) (Fig. 6), and lowest RMSE (0.33), ASE (0.10), AIC (− 79) and BIC (− 151).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\mathrm{Reg}}{\text{-}}1:{\mathrm{LAI}} & = 28.82 \times {\mathrm{BGI}} + 13.77 \times {\mathrm{PPR}} - 7.91 \times {\mathrm{NGRD}} \\ & \quad + 14.88 \times {\mathrm{NCPI}} + 25.86 \times {\mathrm{NPPR}} - 39.74 \\ \end{aligned}$$\end{document}Reg-1:LAI=28.82×BGI+13.77×PPR-7.91×NGRD+14.88×NCPI+25.86×NPPR-39.74

The next regression model (Reg-2) was based on the product of these predictors, i.e. BGI, PPR, NPPR, NGRDI, and NCPI. Reg-2 had R² of 0.87, RMSE 0.37, AIC − 68, BIC − 140, and ASE 0.13.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Reg}}{\text{-}}2:{\mathrm{LAI}} = 505.84 \times ({\mathrm{BGI}} \times {\mathrm{PPR}} \times {\mathrm{NPPR}} \times {\mathrm{NGRDI}} \times {\mathrm{NCPI}}) + 0.134$$\end{document}Reg-2:LAI=505.84×(BGI×PPR×NPPR×NGRDI×NCPI)+0.134

Using ANN for LAI estimation, the accuracy was 97% (R² = 0.97) using the sum of BGI, PPR, NPPR, NGRDI, and NCPI as predictors (ANN-1); whereas it was 91% (R² = 0.91) for the product of BGI, PPR, NPPR, NGRDI, and NCPI (ANN-2) (Fig. 6). The models are available in https://github.com/sayantanhub/LAI_LG_WEKAmodels.

The percentage error of the models Reg-1, Reg-2, ANN-1, and ANN-2 was derived for the individual measurement dates using the formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Error}}\% \, = \frac{{{\mathrm{Manual\, LAI}} - {\mathrm{Estimated\,LAI}}}}{{{\mathrm{Manual \,LAI}}}}$$\end{document}Error%=ManualLAI-EstimatedLAIManualLAI

The average error percentage at 35 DAP was from 0–10%; at 40 DAP was 0–40%; at 45 DAP was from 0–15%; and at 50 DAP was 0–5% (Fig. 6).

The maximum LG of peanut vines varied from 43 to 75 cm at the end of intense LG expansion in 2017 (about 50 DAP), whereas it varied from 66 to 111 cm in 2019 (about 75 DAP). The mean LG was 61 cm in 2017 and 95 cm in 2019 and it correlated to red (− 0.41; P < 0.0001) and blue (− 0.70; P < 0.0001) in 2017 (Table 5). Pearson correlation values showed that within several calculated VIs, BGI (r = − 0.93; p < 0.0001), PPR (r = 0.93; p < 0.0001) and NPPR (r = 0.91; p < 0.0001), were best correlated to ground measured LAI and LG (Table 5). These VIs had RMSE values ranging from 2.3 to 2.7 cm for LG which was lower than for the other VIs. Stepwise selection retained PPR, NPPR, NGRDI, and NCPI as the best four predictors for LG estimation. The third regression model (Reg-3) had the highest R² (0.88) values and lowest RMSE (5.3), ASE (27.5), AIC (400.1) and BIC (310.7). Reg-3 was based on the sum of these predictors (Fig. 7).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\mathrm{Reg}}{\text{-}}3:{\mathrm{LG}} & = 254.26 \times {\mathrm{NPPR}} + 136.76 \times {\mathrm{NCPI}} - 92.73 \times {\mathrm{NGRDI}} \\ & \quad - 82.78 \times {\mathrm{PPR}} - 144.24 \\ \end{aligned}$$\end{document}Reg-3:LG=254.26×NPPR+136.76×NCPI-92.73×NGRDI-82.78×PPR-144.24Figure 7

Comparison of manually measured lateral growth (LG) using a meter scale (x-axis) and derived LG (y-axis) in 2017 using: Reg-3: LG = 254.26 × NPPR + 136.76 × NCPI-92.73 × NGRDI-82.78 × PPR-144.24; Reg 4: LG = 3372.55 × (PPR × NPPR × NGRDI × NCPI) + 19.96. ANN-3: PPR, NPPR, NGRDI, and NCPI as predictors of LG; ANN-4: product of PPR, NPPR, NGRDI, and NCPI as predictors of LG. Each point on the graph represents LG of every genotype at each of the days after planting (DAP) averaged over six replications. The bar chart represents four DAP (x-axis) and P-value (y-axis) derived from paired t-test of manually measured LG using the four models. The P-values lower than 0.05 means the manual LG was different from model derived LG. The box and whisker plots represent the error statistic of estimation models at different DAP. In each box, the central mark is median, and the lower and upper edges denote the 25th and 75th percentile of errors respectively. The whiskers extend to the most extreme data points not considered outliers. Outliers not shown on the chart for clarity.

The fourth regression model (Reg-4) was based on the product of these predictors i.e. PPR, NPPR, NGRDI, and NCPI, and had R² of 0.84, RMSE 6.2, AIC 423.2, BIC 333.3, and ASE 37.9 (Fig. 7).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Reg}}{\text{-}}4:{\mathrm{LG}} = 3372.55 \times ({\mathrm{PPR}} \times {\mathrm{NPPR}} \times {\mathrm{NGRDI}} \times {\mathrm{NCPI}}) + 19.96$$\end{document}Reg-4:LG=3372.55×(PPR×NPPR×NGRDI×NCPI)+19.96

Using ANN to estimate LG, the model accuracy was 94% (R² = 0.94) using the sum of PPR, NPPR, NGRDI, and NCPI (ANN-3); and 87% (R² = 0.87) when using the product of PPR, NPPR, NGRDI, and NCPI (ANN-4). The models are available in https://github.com/sayantanhub/LAI_LG_WEKAmodels.

The percentage error of the models Reg-3, Reg-4, ANN-3, and ANN-4 were derived for the individual measurement dates using the formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Error}}\% \, = \frac{{{\mathrm{Manual \,LG}} - {\mathrm{Estimated\,LG}}}}{{{\mathrm{Manual\,LG}}}}$$\end{document}Error%=ManualLG-EstimatedLGManualLG

The average error percentage at 30 DAP was from 30–40%; 35 DAP was from 0–15%; at 40 DAP was 0–5%; at 45 DAP was from 0–15%; and at 50 DAP was 0–5% (Fig. 7).

VIs derived from the 2019 study were substituted for the corresponding values of the VIs in models Reg-1 to Reg-4. The LAI and LG values derived using these models were correlated with the manual measurements in 2019. Based on the R², the models’ accuracy was 81% for Reg-1, 83% for Reg-2, 80% for Reg-3, and 78% for Reg-4 (Table 6). Model validation with the 2019 data showed that the ANN-1 estimated 73% correctly the manually measured values, and ANN-2 81%. For the LG, ANN-3 estimated 75% correctly the manually measured values and ANN-4 85% (Table 6).Table 6

Validation error statistics, mean error (μ), standard deviation (σ), and coefficient of determination (R²) of the observed and estimated leaf area index (LAI).

Figure 8 presents an example of biomass growth within the first 10 weeks from planting for the peanut genotypes belonging to four market types used for validation in 2019 (Table 2). The picture shows clear visual differences among the genotypes from 45 DAP, i.e. beginning flowering, to 75 DAP, i.e. beginning seed growth stage; and among the dates when ground and aerial measurements were taken, i.e. within 30 days from beginning flowering (at 75 DAP) the ground was completely covered by plants. The picture shows clear distinction between the market types, i.e. the runner and Virginia types were more compact than the Spanish and Valencia that developed distinct main stems from the lateral branches at 75 DAP.Figure 8

Morphological variation among different peanut market types over different days after planting (DAP). The differences are distinct at 75 DAP. The runner types are more spread out; Virginia types are moderately spread; Spanish and Valencia types have more erect main stem than the others.

For models training, in 2017, only Virginia and runner genotypes were used (Table 1). Box and whisker plots of measured and estimated LAI (Fig. 9) and LG (Fig. 10) show the spread of the data for the 18 genotypes measured from 30 to 50 DAP in 2017. Within each date of measurement, the range and the interquartile range (IQR) of the measured and estimated LAI and LG were similar or larger for the estimated traits. This shows that the models are suitable to identify phenotypic variability among peanut genotypes. For example, at 45 DAP, LAI range, i.e. the range from minimum to maximum LAI, was 1.2 for the measured, 1.7 for Reg-1, 2.1 for Reg-2, 1.6 for ANN-1 and 2.1 for ANN-2 estimated data (Fig. 9). Similarly, the IQR range or 50% of the data represented by the box, was 0.3 for measured, 1.1 for Reg-1, 0.7 for Reg-2, 0.6 for ANN-1 and 0.7 for ANN-2 estimated LAI; and the median was at or close to 2 for the estimated LAI corresponding to the manually measured LAI (Fig. 9). Figure 10 shows similar box and whisker results for the LG. Measured and estimated LAI and LG in 2017 were subjected to ANOVA for the effect of genotype within each date of measurement. With the exception of 50 DAP when estimated LAI and LG was not statistically different among the genotype, for all other dates, the measured and estimated LAI and LG showed significant differences among the genotypes, i.e. P-value ranged from 0.002 to < 0.0001. In 2017, the genotype average was 2.9 ± 0.5across the estimated and measured LAI; and 60 ± 3 cm for LG at 50 DAP.Figure 9

The box and whisker plots show the increase in leaf area index (LAI) over days after planting (DAP), where LAI has been derived using the same models Reg-1: LAI = 28.82 × BGI + 13.77 × PPR-7.91 × NGRD + 14.88 × NCPI + 25.86 × NPPR-39.74; Reg-2: LAI = 505.84 × (BGI × PPR × NPPR × NGRDI × NCPI) + 0.134; ANN-1: BGI, PPR, NPPR, NGRDI, and NCPI as predictors of LAI ; ANN-2: product of BGI, PPR, NPPR, NGRDI, and NCPI as predictors of LAI.

Figure 11 shows examples of genotypic variability for the measured and estimated LAI and LG, and includes six genotypes from 2017 at 45 and 40 DAP, respectively. In this example, Wynne and Walton showed an overall smaller LAI than GA09B and breeding line 09X44-2-14-1; and all had overall smaller LAI than Sullivan and line 09X44-2-14-1. Genotypes Walton, 09X37-1-19-2 and 09X44-2-14-1 were overall more spread at 40 DAP than Sullivan, Wynne, and GA09B. The variability of the estimated vs. measured LAI ranged from 5 to 20% and from 3 to 14% for LG; but none of the estimated values were significantly different from the measured data.Figure 11

Bar graph showing Leaf area index (LAI) and lateral growth (LG) of six peanut genotypes. The LAI and LG values of each genotype include manually taken values and derived values using four models (Reg-1, Reg-2, ANN-1, ANN-2 for LAI; and Reg-3, Reg-4, ANN-3, ANN-4 for LG). The LAI measurements are from 45 days after planting (DAP) and LG are from 40 DAP. †LAI and LG values within each genotype are not significantly different using Tukey’s HSD at α = 0.05.

Manually measured and estimated LAI and LG from each measurement date were further used to assess the contribution of early season LAI and LG to peanut pod yield. The relationship fitted cubic regressions for both, LAI and LG, with the highest coefficients of determination (R² from 0.51 to 0.80) when LAI and LG were measured or estimated at 40 and 45, which corresponds with beginning flowering DAP (Table 7).Table 7

Relationship of leaf area index (LAI) and lateral growth (LG) with peanut pod yield at different days after planting (DAP).