REFINING THE RELATIONSHIP BETWEEN CANOPY LIGHT INTERCEPTION AND YIELD IN WALNUT

Transcription

1 REFINING THE RELATIONSHIP BETWEEN CANOPY LIGHT INTERCEPTION AND YIELD IN WALNUT Bruce Lampinen, Shrini Upadhyaya, Vasu Udompetaikul, Greg Browne, David Slaughter, Samuel Metcalf, Bill Stewart, Loreto Contador, Ignacio Porris Gomez, Jedediah Roach, Bob Beede, Carolyn DeBuse, Janet Caprile, Bill Coates, David Doll, John Edstrom, Rachel Elkins, Joe Grant, Janine Hasey, Kathy Kelley, and Bill Krueger ABSTRACT The existing mobile platform light bar was redesigned in order to make it more robust and adjustable to a wider range of tree spacing. In addition, more accurate soil temperature sensors were added. This second generation mobile platform light bar was used during the 211 and 212 seasons and performed well. In addition to the second generation mobile platform light bar, a new harvest supercart equipped with GPS, continuous yield monitoring capability, selfcontained hydraulics and an auto-sampler were also built with partial funding from this project. Light measurements continued on a large number of walnut orchards of varying ages and tree densities. An upper limit of about.5 in-shell tons per acre for every one percent of light that can be intercepted has continued to be supported. When all of the walnut data from the last four years was combined, it was found that midday canopy light interception increased with orchard age until about 1-12 years when it tended to flatten and approach a value of near 93 percent. This results in a mature walnut orchard yield potential above 4 tons per acre. Work was continued on the canopy sensor suite for assessing plant water status. The results were promising and suggest the possibility of further developing this classification technique to assess plant water stress for irrigation scheduling in walnut orchards. In particular, the results using shaded rather than sunlit leaves were very encouraging and this should allow automating of the process onto the mobile platform since it is not critical to assess leaf angle for shaded leaves. PROBLEM AND ITS SIGNIFICANCE Data collected by the authors over the past several years has provided a rough upper limit to productivity in walnut and almond based on the percentage of the available midday canopy photosynthetically active radiation (PAR) that is intercepted. However, most of the data that was collected previously had limitations. The methods of measuring percent PAR interception using a handheld light bar (Decagon Devices, Pullman, WA 99163) were relatively slow and labor intensive. For this reason, much of the light bar data that was used to develop the relationship was based on sampling of relatively small samples of trees. Often the area for the yield and PAR interception data did not match (i.e. PAR data from 5 trees and yield data from either one tree or from an entire row). In the 29 and 21 seasons, we began using the first generation light bar for automated measurement of canopy PAR interception. In the winter of 211, we completed the second generation light bar which was mounted on the Kawasaki Mule in the spring of 211 and used during the 211 and 212 summer seasons. The new mobile platform light bar is continuously adjustable from approximately 8-32 feet. California Walnut Board 147 Walnut Research Reports 212

2 The data generated with the mobile platform light bar are useful for any studies that aim to quantify the impact of treatments on yield. By measuring canopy light interception on a large scale, the impacts of differences in canopy development can be separated out from other treatment impacts allowing more robust data interpretation. OBJECTIVES There were three objectives for this project in 212. The first was to provide support to Shrini Upadhyaya and David Slaughter to complete the light bar modifications to make it sturdier and more adjustable as well as to work toward automating the data processing. The first generation mobile platform light bar was developed with funding from the almond and walnut industries as well as the USDA almond methyl bromide area wide project. An additional part of this objective was added, which was to build a new supercart with load cells and GPS for more automated and grower friendly yield data collection. The second objective was to continue to monitor orchard sites, in cooperation with farm advisors, throughout the walnut growing areas of California for testing the light bar. These sites were chosen in orchards that have a variety of ages and light interception levels. The goal was to find orchards that are well managed and as productive as possible to attempt to place upper limits on the light interception/yield relationship. The third objective was to develop technology for determining plant water status with sensor technology. PROCEDURES Objective 1 The plan was to use the funding from the California Walnut Board to improve the design of the light bar for utilization in walnuts including adding the ability to assess canopy shape using either cameras or LIDAR. This objective was designed to provide support to Shrini Upadhyaya and David Slaughter to modify the existing light bar design to make it more robust, user friendly and able to assess canopy shape. Objective 2 Walnut orchard mapping- The plan was to obtain light bar data on individual rows in a variety of walnut orchards and to then harvest these same rows to obtain yield data to compare to the light bar data. During the 211 season, a total of twenty eight different walnut orchards were mapped with the mule light bar (Table 1). Several of the orchards were mapped multiple times during the season to investigate the impact of time of year and environmental conditions on canopy light interception (Table 1). Data was collected on 18 Chandler orchards, 4 Tulare orchards, 4 Howard orchards, as well as a trial with Forde and Gillet (recent releases from the UC Davis Walnut Improvement Program). In general, light bar measurements were either done in each replication of a particular trial or in 5-1 row middles (depending on orchard size and variability) in representative areas of grower orchards. These rows were then flagged and at harvest time, field weights were taken from each row middle. Subsamples were taken from each row middle, hulled, and dried to estimate in-shell crop weight for each row and samples were submitted for quality analysis. California Walnut Board 148 Walnut Research Reports 212

3 Table 1. Description of sites used for walnut light interception and yield mapping in the 212 season. A total of 23 sites were mapped in 212. Objective 3 A pressure chamber or pressure bomb is often used to measure stem water potential of a plant to determine plant water status. However, measuring plant water status using a pressure bomb is tedious and time consuming, and is not suitable for measuring plant water status if a large number of measurements need to be obtained for irrigation management purposes. The objective of this portion of the project is to develop a sensing system that can determine the plant water status quickly. Based on the energy balance model for a single leaf, the temperature difference between the leaf and the surrounding air can be shown to depend on leaf conductance, net radiation, vapor pressure deficit, and wind speed. Leaf conductance is a major plant parameter limiting evapotranspiration rate that influences leaf temperature. Leaf conductance depends upon many factors. During the midday period, the time that plants experience maximum water stress, stem water potential is the major factor that controls leaf conductance. The other factors that influence leaf temperature are vapor pressure deficit, solar radiation, and wind speed. A sensor suite, consisting of an infrared thermometer, quantum sensor, pyranometer, ultrasonic anemometer, and air temperature and relative humidity probe, has been developed to measure leaf temperature and microclimatic information in the vicinity of the leaf to determine plant water stress. Experiments were conducted to test the validity of this sensor suite. Work on objective 3 was carried out during the summer of 211 and 212 in 2 walnut orchards in Colusa County. Within each orchard three blocks of 5 to 6 trees were established. During a given test, a block was selected at random and leaf temperature, photosynthetically active radiation (PAR), solar radiation, air temperature, and wind speed were measured using the sensor suite on each tree. Stem water potential (SWP) and leaf conductance (g L ) were measured using a California Walnut Board 149 Walnut Research Reports 212

4 pressure chamber and leaf porometer respectively. Five sunlit and five shaded leaves were measured per tree. Each observation consisted of averages of 5 leaf-air temperature (T Leaf -T air ), 5 PAR, 5 solar radiation (R s ), 5 VPD, and 5 wind speed measurements. In addition, stomatal conductance measurements were made on three each of the five sunlit and shaded leaves. One stem water potential measurement was taken per tree. This experimental procedure was repeated between 5 to 12 times for a given orchard. RESULTS Objective 1 Mule platform modification- The existing Mule mounted lightbar setup (Fig. 1a) a b Fig. 1. (a). Design of Kawasaki Mule mounted light bar as used during summer 21. Modifications included adding a branch bumper on front designed to aid in pushing through orchards with many low overhanging branches. (b)-. Over the winter of 21-11, the entire light bar was redesigned and rebuilt and made much more protected, robust and adjustable. California Walnut Board 15 Walnut Research Reports 212

5 The light bar has been modified in order to make it more robust and adjustable to a wider range of tree spacing. This included rebuilding the entire light bar with a much more stable and more adjustable base and a built in protective bumper to push low hanging branches up and over the light bar (Fig. 1b). A more accurate global positioning satellite (GPS) receiver and an encoder that measures distances using the rotation of the axle were added to provide more accurate positional information. In addition, three infrared thermometers with a much narrower angle of view were added for measuring soil surface temperature under the tree canopy. In addition LIDAR was added to the mobile platform light bar (Fig. 2). Fig. 2. Mobile platform lightbar shown set up to use LIDAR to assess canopy shape. A load cell and GPS equipped supercart harvest trailer was also built with partial funding from this project (Fig. 3). The trailer has self-contained hydraulics to operate the augers and rear elevator as well as an autosampler. This allows the trailer to be used with almost any grower harvesters since we do not need to match the hydraulic hookups to the grower s equipment. The trailer also allows our experimental harvest plots to be picked up and weighed on a continuous basis with minimal disruption to grower harvest operations since we can offload the nuts into any type of bank out system the grower uses. The yield data from the load cells on the trailer is logged 1 times each second to the Trimble GPS unit which allows continuous yield data to be compared to continuous light bar data. California Walnut Board 151 Walnut Research Reports 212

6 Fig. 3. Load cell and GPS equipped supercart built with funding from this project. This setup allows us to collect continuous yield data which is then logged to the Trimble GPS unit. Objective 2 A total of 28 orchards were mapped in 211 (Table 1). Midday canopy light interception varied from about 1 to 9 percent in the orchards studied in (Fig. 4). Orchards that produced higher than expected yields one year (for their level of light interception) generally produced less in the following year. Yield data was lost at several sites in all years due to problems with scheduling harvests with growers and/or due to early fall rains making plot harvest difficult to fit within the grower harvest schedule. Measurements were also done throughout the season on an eight year old Howard pruning trial as well as a younger Chandler pruning trial at Nickels Soil Laboratory in Colusa County in Midday canopy light interception was fairly constant through the season for the more mature Howard pruning trial shown in Fig. 5. This suggests that measurements on mature orchards can probably be done from early June to late August with similar results. Seasonal midday canopy light interception through the 211 season for the Chandler pruning trial is shown in Fig. 5. There were no significant treatment differences at any point during the season. However, the large seasonal change in midday canopy light interception in the younger Chandler orchard (Fig. 6) compared to the older Howard orchard (Fig. 5) suggests that the seasonal timing California Walnut Board 152 Walnut Research Reports 212

7 Yield (dry tons/acre) Y Y G Y R Y R Y R Y RR GG R B G R B Y G Sutter Own-rooted trial Rio Oso OR Chandler/Par Sutter Own-rooted trial Rio Oso OR- Chandler/ow Sutter Own-rooted trial Rio Oso OR- Vina/Parado SutterOwn-rooted trial Rio Oso OR- Vina/own-roo Lake Walnut lightbar sites Lake County- Upper L Sutter Walnut lightbar site Rio Oso #2 Sutter Walnut lightbar site Rio Oso #1 San Joaquin Tulare Farmington site Stanislaus Chandler site Modesto Yolo County Chandler site Madison Yuba Tulare Surround trial Colusa Nickels Howard pruning trial 1 G B B B B G B Yield (dry tons/acre) Yield (dry tons/acre) Yield (dry tons/ac) Sutter Walnut lightbar site Rio Oso #2 Sutter Walnut lightbar site Rio Oso #1 San Joaquin Tulare Farmington site 21 Yolo County Chandler site Madison 5 Nickels Howard unpruned Nickels Howard unpruned + nut removal Nickels Howard pruned 4 Nickels Howard pruned + nut removal Nickels Howard pruned 26, hedged 28 Nickels Forde unpruned, unheaded Nickels Forde moderately pruned 3 Nickels Forde heavily pruned Nickels Gillet unpruned, unheaded Nickels Gillet moderately pruned 2 Nickels Gillet heavily pruned Nickels Tulare unpruned, unheaded Nickels Tulare moderately pruned Nickels Tulare heavily pruned 1 Rio Oso Orchard #3 Solano Surround trial untreated control Solano Surround trial Surround treated San Joaquin Tulare Farmington Site 211 Yolo County Chandler site Madison Nickels Howard unhedged 5 Nickels Howard skirted Nickels Howard hedged Nickels Forde 4 Nickels Gillet Nickels Tulare Nickels Chandler minimal pruned 3 Nickels Chandler untrained, unpruned Nickels Chandler heavily pruned Nickels Chandler minimal pruned plus deficit Mature Chandler Lake County 2 Lake County 1 yr old Chandler Lake County 9 yr old Chandler Rio Oso Chandler #1 1 Rio Oso Chandler #2 Tehama mature Chandler #1 Tehama mature Chandler #2 Tehama 5 year old Chandler Tehama 6 year old Chandler Colusa 14 yr old Howard Orchard #1 Midday canopy light interception (%) Colusa 14 yr old Howard Orchard # Midday Canopy Light Interception (%) Tehama 4yr old Tehama 5yr old Tehama 6yr old Tehama 7yr old Tehama 16yr old San Joaquin rootstock Nickels Chandler stress sensing Sutter walnut lightbar site Rio Oso #2 Howard grower hedgerow orchard #1 Howard grower hedgerow orchard #2 Lake County 11yr old Chandler Nickels Forde Nickels Tulare Nickels Gillet Nickels Chandler minimal pruned Nickels Chandler untrained, unpruned Nickels Chandler heavily pruned Nickels Chandler minimal pruned plus deficit Nickels Howard hedged 5' Nickels Howard skirted Nickels Howard unhedged Sutter walnut lightbar site Rio Oso #1 Walnut lightbar Upper Lake site #1 Sutter walnut lightbar site Rio Oso #1 Fig.4. Midday canopy light interception versus yield relationship from various walnut trials from throughout state measured with the Mule mounted light bar in 29, 21, 211 and 212. The upper line is the estimated upper limit and the lower line is the regression for all data (212 only). California Walnut Board 153 Walnut Research Reports 212

8 1 Nickels Howard pruning trial 21 Midday canopy light interception (%) T1 (unpruned) T2 (unpruned w/fruit removal) T3 (pruned) T4 (pruned w/fruit removal) T5 (pruned 24 and hedged 28) Mechanically hedged Date, 21 Fig. 5. Midday canopy light interception for the Nickels Howard pruning trial over the 21 season by pruning treatment. of measurements in young orchards is much more critical. For instance, the minimal pruned treatment increased from about 46 to 56 percent midday canopy PAR interception from late May to late September in 212 (Fig. 6). This is a significantly slower rate of increase in canopy cover compared to the previous year. Midday Canopy Light Interception (%) Heavily pruned (T1) Minimal pruned plus deficit (T2) Minimal pruned (T3) Untrained, unpruned (T4) May Jun Jul Aug Sep Oct Fig. 6. Midday canopy light interception over the 212 season for Nickels Chandler pruning trial. There were no significant treatment differences on any date in 212. Fig. 7 shows the yield per unit PAR intercepted (expressed as in-shell pounds per 1% PAR intercepted) for the 29 through 212 seasons. The line at 1 is the maximum sustained level California Walnut Board 154 Walnut Research Reports 212

9 Yield per unit PAR intercepted (in-shell tons per 1% PAR) Heavily pruned (T1) Minimal pruned plus deficit in (T2) Minimal pruned (T3) Untrained, unpruned (T4) Year Fig. 7. Yield per unit PAR intercepted by treatment and year for Chandler pruning trial at Nickels Soil Laboratory for 29 to 212. Horizontal dotted line shows our current estimate of the highest sustained production level for walnut. of production that might be expected in a mature orchard and we have had one mature Tulare orchard in our trial for three years that has produced very close to the maximum level for all 3 years. It is interesting that the average for the Chandler pruning trial treatments for the seasons is close to the maximum level since this orchard is still filling in the space allotted (Fig. 7). Midday canopy PAR interception increased with orchard age until about 1-12 years when it tended to flatten and approach a value of near 93 percent (Fig. 8). The potential rate of increase 1 Midday Canopy Light Interception (%) Age Fig. 8 shows the midday canopy light interception versus orchard age for walnut sites (each point is the average for an orchard) measured with the Mule light bar during the seasons. The upper boundary to the relationship appears to be about an 8-9% rate of increase from years 1-1 and then a leveling off eventually approaching a level of interception of about 93%. California Walnut Board 155 Walnut Research Reports 212

10 in years 1 to 1 is about 9-1% per year. The rate is dependent on tree spacing, variety, rootstock, water management, etc. These data are useful in evaluating the performance of a given orchard in relation to others of a similar age, planting configuration, etc. Orchard floor temperature- The impact of midday canopy light interception on soil surface temperature is shown in Figure 8. There is a clear relationship between midday canopy light interception on the left side of the light bar and soil surface temperature with the high soil temperatures (approximately 6 C) occurring when midday canopy light interception is low. Even though this orchard is only three years of age, the cool temperatures that are associated with shaded orchards are beginning to show up underneath the tree canopy as shown by the 3 C temperature reached under the tree on the right as it reaches about 9% light interception (Fig. 9). Figure 1 shows similar data for a mature Tulare planting that has produced 4 tons per acre for Fig. 9. The top graph shows fractional midday PAR interception (multiply by 1 to get percent midday PAR interception) while the bottom graph shows orchard floor temperature (in degrees Centigrade) for the corresponding drive row center for a methyl bromide alternatives trial in Kings County, California measured on Aug. 3, 211. the past three years. Note that the high temperatures coincide with the lowest PAR interception and the low temperatures with the highest PAR interception. These data will be useful in assessing impacts of different canopy management regimes on soil surface temperature which can relate to food safety risk since ideal temperature for Salmonella survival in the soil is in the range of 3-35 deg. C which is in the range the shaded areas in walnut orchards generally reach in the Central Valley of California during mid-summer. Because mature walnut orchards tend to be heavily shaded, and this results in near optimal temperatures for Salmonella growth and survival, the potential for food safety risk cannot be ignored. California Walnut Board 156 Walnut Research Reports 212

11 Fig. 1. The top graph shows fractional midday PAR interception (multiply by 1 to get percent midday PAR interception) while the bottom graph shows orchard floor temperature (in degrees Centigrade) for the corresponding drive row center for a mature Tulare walnut orchard in San Joaquin County that has produced an average of 4 tons per acre for the past three years. Data collected on 7/3/211. Objective 3 Determining plant water stress using an IR thermal sensor- Based on the energy balance model for a single leaf, the temperature difference between the leaf and surrounding air can be shown to depend on leaf conductance, net radiation, vapor pressure deficit, and wind speed. Leaf conductance is a major plant parameter limiting the rate of evapotranspiration which influences leaf temperature. Leaf conductance depends upon many factors. During the midday period, when plants experience maximum water stress, stem water potential is the major factor that controls leaf conductance. The other factors that influence leaf temperature are vapor pressure deficit, solar radiation, and wind speed. A sensor suite, consisting of an infrared thermometer, quantum sensor, pyranometer, ultrasonic anemometer, and an air temperature/relative humidity probe, has been developed to measure leaf temperature and microclimatic information in the vicinity of the leaf to attempt to relate these factors to plant water stress (Fig. 1). Experiments were conducted in two walnut orchards in to test the validity of this sensor suite. In the models presented in Table 2, the response variable is the temperature difference between the leaf and the surrounding air and plant water status or its surrogate, stomatal conductance, is the independent or predictor variable. The approach was to use a threshold value of temperature differences corresponding to critical values of SWP. A walnut tree at -8 bars was considered to be stressed (i.e. in need of irrigation). The approach predicted correct plant water status in 95% to 1% of the cases for walnuts (using validation data sets). In terms of rating trees which are stressed as unstressed, which is designated as a critically wrong decision as this has implications for plant growth and yield, California Walnut Board 157 Walnut Research Reports 212

12 Fig. 11. Sensor suite consisting of an infrared thermometer, quantum sensor, pyranometer, ultrasonic anemometer, and an air temperature/relative humidity probe. Also shown is a plant pressure chamber used as a standard to measure midday stem water potential. the results, varied between % to 5% for walnuts. A less serious error that would lead to overirrigation (i.e., detecting unstressed trees as stressed) also occurred in % to 5% of validation datasets for walnut. These results are promising and suggest the possibility of further developing this classification technique to assess plant water stress for irrigation scheduling in walnut orchards. Improvements such as use of a newer, more sensitive IR sensor and more advanced analysis techniques may further improve plant water status prediction in walnuts. The success of the shaded leaf method allows the possibility of automating the process onto the mobile platform since it is not critical to assess leaf angle for shaded leaves. During the 212 growing season, the mobile sensor suite was extensively tested in walnut orchards. The tests revealed that the sensor suite can be used to detect midday stem water potential in walnut. The stepwise selection based multiple linear regression (MLR) models yielded coefficient of multiple determination values of.75 walnuts when shaded leaf temperature was used to develop the model. Moreover, two classification techniques (Stepwise discriminant analysis (SDA) and Canonical discriminant analysis (CDA)) were used to identify stressed and unstressed trees. When these three methods (MLR, SDA, and CDA) were used for plant water stress classification (as stressed and unstressed) we found that in walnuts, critically wrong errors were 12.9, 1., and 1.1% respectively, for MLR, SDA, and CDA methods. Over-irrigation errors were 7.7, 12.7, and 9.8 % respectively, for MLR, California Walnut Board 158 Walnut Research Reports 212

13 SDA, and CDA methods. When the spatial variability in stressed versus unstressed conditions were mapped all three techniques yielded similar results. One major issue that was observed during these tests was that the calibration equation was influenced by the time of measurement (i.e., calibration equation developed at the beginning of the season did not apply well for the end of season data). Moreover, the sensitivity of the inexpensive thermal IR sensor used in this study was also a concern (+.5 C). We decided to address these two issues more thoroughly during the 212 growing season. To address the sensitivity issue, we have developed a multiple (9) IR sensor head unit. This unit should reduce the instrument noise by a factor of three. We have also developed a leaf monitor to continuously measure leaf temperature to address the issue of temporal changes in the calibration. Both of these units are being used in the field this year. Based on the outcome of this year s results, we think it will be possible to retrofit the sensor suite to the mule and this is the plan for 213. In addition, a more portable hand held unit for estimating stress using leaf temperature and environmental variables will also be developed. Plans for 213- The first objective will be to further refine the relationship between midday canopy light interception and yield as well as the impact that alternate bearing has on the relationship. With a better estimate of the maximum productivity per unit light interception, these data can be used to assess potential orchard yield and will allow separating out canopy light interception as a variable in other research projects. For example, if a pruning study is being conducted, this tool will allow the separation of the effect of the pruning treatment on overall canopy light interception as opposed to the effect of the pruning treatment on productivity per unit canopy. It will also allow block to block variability to be assessed before or after a research trial in initiated. The second objective will be to continue to refine using the sensor suite to assess plant stress in walnut. The promising results using shaded leaves means that it should be possible to measure the shaded canopy temperature using the mobile platform lightbar and this is one of the planned activities for 213. The third aspect of this project in 213 will be to continue developing and calibrate a new iphone application that we have developed that can be used to estimate canopy cover from taking a photo of the tree shadows. California Walnut Board 159 Walnut Research Reports 212