Abstract
The use of canopy density for crop production is a useful tool for evaluating management practices for informed decision-making and predicting crop yields. Traditional methods for analyzing canopy density include expensive equipment that requires advanced software and costly repairs, such as leaf area index analyzers, or equipment that can only be used during optimal weather conditions, such as light meters, that quantify photosynthetically active radiation (PAR), thus making it difficult for researchers to maximize the time needed for field research evaluations and data collection. Digital image analyses using technologically advanced cameras, such as smartphone cameras, have allowed new ways of collecting data without the need to purchase specialized instruments. Using a combination of smartphone cameras and ImageJ software, canopy density can be measured in any weather conditions for a much lower cost than that of traditional equipment. This low-cost, digital image analysis method was compared with traditional PAR measurements for ‘Valencia’ sweet orange [Citrus sinensis (L.) Osb.] trees with varying levels of canopy density. A strong positive correlation between the digital image analysis method and standard canopy density measurement method using PAR measurements (r = 0.79; P<0.0001) was found, indicating that this method can be used as an alternative to the PAR method. The digital image analysis method was also consistent when used during different weather conditions, whereas the PAR method was highly variable when quantifying the canopy when clouds were present in comparison with clear sky conditions. This novel method provides researchers and growers with an easy, flexible, consistent, and low-cost option for analyzing canopy density.
Canopy density is a valuable data parameter used for horticultural research to determine the health and productivity of plant systems. Crop production is closely related to plant biomass, and because dry matter content is linearly related to intercepted solar radiation, an improvement in biomass production leads to increased photosynthetic carbon fixation, which translates to increased crop yields (Murchie et al. 2008). For many species of tree fruit crops (e.g., citrus, apple, peach, and pear), yields and fruit qualities are sensitive to changes in canopy density, regardless of whether that change is an increase or decrease in leaf number (He et al. 2008; Khemira et al. 1993; Robinson et al. 1991).
Canopy density can be used to evaluate tree health in response to disease. Currently, in Florida, most commercially grown citrus trees grown are infected with the Gram-negative, phloem-limited bacteria Candidatus Liberibacter asiaticus (CLas), which causes Huanglongbing (HLB; citrus greening). This disease has many devastating symptoms, including increased preharvest fruit drop, small off-flavor fruit, and severe shoot dieback leading to very thin canopies (Bové 2006). There is no cure for this disease. However, the use of enhanced-regimented orchard management programs has shown successful mitigation of disease symptoms (Levy et al. 2023). Recently, it was reported that canopy density is a better and more accurate predictor of HLB severity than CLas titer quantification via quantitative reverse-transcription polymerase chain reaction (Levy et al. 2023; Yu et al. 2022). Providing citrus growers with a tool to measure canopy density would allow them to evaluate new changes in orchard management and predict yields to ensure that their decisions are backed by quantitative evidence rather than subjective (and relative) observations.
Traditionally, equipment used to evaluate canopy density such as canopy analyzers that record leaf area index or photosynthetically active radiation (PAR) meters are expensive to purchase and maintain and/or require proprietary software for analysis that can be difficult to use. PAR meters are relatively simple to use, but they require optimal weather conditions with full sun (clear sky) for accurate measurements; however, these conditions are difficult to obtain during Florida’s long rainy season, which occurs from May through October.
Digital image analyses using technologically advanced cameras, such as those in smartphones, have allowed researchers to obtain more versatile ways of collecting data without needing to purchase specialized equipment. This technology can also be used by growers for grove management evaluations. A digital image analysis method was compared with the traditional method (using PAR) under different canopy and weather conditions.
Materials and Methods
Digital images were obtained with an iPhone 12 mini (Apple Inc., Cupertino, CA, USA) using a 5-foot-long telescopic boom (i.e., “selfie stick”) with a Bluetooth remote that controlled the shutter. The phone was set on the ground midway between trunk and canopy periphery; it was facing the sky to capture an upward image of the canopy using the front-facing camera (Figs. 1A, 1C, and 2). Then, images were uploaded to free image-processing software (version 1.53s; ImageJ) and transformed to 8-bit. Using the thresholding feature, pixels were separated between the “foreground” and “background” (i.e., canopy and sky), respectively (Fig. 1B and 1D). After adjusting the threshold to separate the canopy from the sky, the percentage listed in the thresholding window was recorded. For each tree, %INT derived from each quadrant was averaged to obtain one value that was compared with the PAR measurement.
Images of mild and severe Huanglongbing (HLB)-affected ‘Valencia’ trees. Left column: An original image of (A) mildly affected tree (canopy density: 91.85%) and (C) severely affected tree (canopy density: 68.45%). B and D images were generated after threshold adjustment using ImageJ, which provides a photo with a percentage of the covered canopy (red) for respective A and C images.
Citation: HortScience 58, 7; 10.21273/HORTSCI17112-23
The placement of the phone stick within the canopy for the digital image analysis method. The phone was placed on the ground approximately midway between the tree trunk and canopy periphery, with the front-facing camera facing upward to the sky. One photo of each quadrant of the tree was obtained, resulting in four photos per tree that were averaged to yield one value of the canopy density per tree.
Citation: HortScience 58, 7; 10.21273/HORTSCI17112-23
To determine if the digital image analysis results were consistent in varying weather conditions, PAR measurements and photos of 22 mature field-grown ‘Valencia’ sweet orange trees in Lake Alfred, FL, USA, were obtained during full sun and cloudy conditions.
A Pearson correlation test was performed to determine if the data generated from the digital image analysis was comparable to those of generated using the PAR method. To determine if the digital image method was more consistent during varying weather conditions, a t test (α = 0.1) of full sun conditions and cloudy conditions was performed for each method. A statistical analysis was performed using Sigmaplot (version 14.5).
Results and Discussion
The Pearson correlation test confirmed that digital images analyzed with ImageJ software can be used instead of PAR measurements for measuring the canopy density (Fig. 3). A strong positive correlation (r = 0.78; P < 0.001) was found between the PAR and digital image analysis methods. Because PAR measurements are only accurate during full sun conditions, it is critical for researchers and growers to be able to obtain measurements of the canopy density in varying sun conditions, thus maximizing the time needed for collecting data. Table 1 shows that the PAR method was inconsistent under varying cloudy conditions, whereas the digital image analysis method yielded equivalent results under full sun and cloudy conditions. When comparing the PAR method under full sun or cloud coverage, there was an average 5.47% difference between the two conditions, and statistically different canopy density quantification was observed, whereas the digital image analysis resulted in a small difference of 1.52%, thereby resulting in similar canopy density quantification under the two conditions. This suggests that the digital image method is reliable under cloudy conditions that would otherwise interfere with more traditional canopy density analysis methods.
A correlation plot for canopy density measurements obtained using the “new” digital image analysis method compared with that obtained using the traditional method involving photosynthetically active radiation (PAR).
Citation: HortScience 58, 7; 10.21273/HORTSCI17112-23
Average densities of 22 ‘Valencia’ trees obtained during full sun and cloud coverage conditions using the photosynthetically active radiation (PAR) method and digital image analysis.
Research has shown that disease severity categorized using canopy density of HLB-affected citrus is efficient for predicting yields (Levy et al. 2023; Tang et al. 2019). Figure 1 shows an example of two images of commercially grown ‘Valencia’ sweet orange trees affected by HLB with differing canopy densities. Because of the variability among trees in the same grove using the same management program, decisions based on visual observations may not be reliable for the long-term health of HLB-affected groves. This novel method allows growers and researchers to use the affordable and accessible technology of smartphone cameras to collect invaluable data needed for field site and grove management evaluations.
Conclusion
The digital image analysis method involving the analysis of canopy density has been shown to be accurate, inexpensive, easy, and more accessible than traditional methods. Furthermore, it involves an inexpensive tool (with some learning of Image J software) that citrus growers can use to evaluate grove management practices and yield prediction. This method can also be used by researchers to quantitatively categorize HLB-affected trees according to their health status to achieve optimum field evaluations. Because of the ease and adaptability of this method, researchers can also obtain canopy density measurements quickly and accurately during any time of year.
References Cited
Bové JM . 2006. Huanglongbing: A destructive, newly-emerging, century-old disease of citrus. J Plant Pathol. 88(1):7–37. http://www.jstor.org/stable/41998278.
He FL, Wang W, Wei QP, Wang XW, Zhang Q. 2008. Relationships between the distribution of relative canopy light intensity and the peach yield and quality. Agric Sci China. 7(3):297–302. https://doi.org/10.1016/S1671-2927(08)60069-3.
Khemira H, Lombard PB, Sugar D, Azarenko AN. 1993. Hedgerow orientation affects canopy exposure, flowering, and fruiting of ‘Anjou’ pear trees. HortScience. 28(10):984–987. https://doi.org/10.21273/HORTSCI.28.10.984.
Levy A, Livingston T, Wang C, Achor D, Vashisth T. 2023. Canopy density, but not bacterial titers, predicts fruit yield in huanglongbing-affected sweet orange trees. Plants. 12(2):290. https://doi.org/10.3390/plants12020290.
Murchie EH, Pinto M, Horton P. 2008. Agriculture and the new challenges for photosynthesis research. New Phytol. 181(3):532–552. https://doi.org/10.1111/j.1469-8137.2008.02705.x.
Robinson TL, Lakso AN, Carpenter SG. 1991. Canopy development, yield, and fruit quality of ‘Empire’ and ‘Delicious’ apple trees grown in four orchard production systems for ten years. J Am Soc Hortic Sci. 116(2):179–187. https://doi.org/10.21273/JASHS.116.2.179.
Singh S, Livingston T, Tang L, Vashisth T. 2022. Effects of exogenous gibberellic acid in huanglongbing-affected sweet orange trees under Florida conditions—II. Fruit production and tree health. HortScience. 57(3):353–359. https://doi.org/10.21273/HORTSCI16277-21.
Tang L, Chhajed S, Vashisth T. 2019. Preharvest fruit drop in Huanglongbing-affected ‘Valencia’ sweet orange. J Am Soc Hortic Sci. 144(2):107–117. https://doi.org/10.21273/JASHS04625-18.
Yu Q, Dai F, Russo R, Guha A, Pierre M, Zhuo X, Wang YZ, Vincent C, Gmitter F. 2022. Phenotypic and gene variation in morphophysiological traits in huanglongbing-affected mandarin hybrid populations. Plants. 12(1):43. https://doi.org/10.3390/plants12010042.