Wound reaction to trunk injection of oxytetracycline or water in huanglongbing-affected sweet orange (Citrus sinensis) trees

Trunk injection causes injury to trees and best practices must be established for use of this technology to manage vascular diseases such as huanglongbing. Trunk injection is a technique for applying plant protection compounds that has demonstrated efficacy for management of huanglongbing (HLB), a bacterial disease which has devastated the Florida citrus industry. Recently, the antibacterial oxytetracycline has been approved for trunk injection in citrus in Florida. Despite its potential for restoring health of HLB-affected trees, little information is available on the trees’ ability to heal and compartmentalize the wounds caused by trunk injection and the interaction with season and other factors. In this study, 5-year-old HLB-affected sweet orange (Citrus sinensis) trees were injected bi-monthly to measure seasonal differences in the rate of uptake of injected water and to compare the external and internal wound reaction in both scion and rootstock. Also investigated was the wound reaction after trunk injection of an injectable formulation of oxytetracycline. Injection during spring and summer promoted more rapid wound closure compared to injection in fall or winter, and injection into the scion was less damaging than injection into the rootstock. Although citrus trees effectively compartmentalized wounds inflicted by injection of water, injection of oxytetracycline impeded wound healing and increased internal damage. Application of a fungicide or pruning sealant increased the size of the wound internally and externally. For trunk injection to be adopted for use in commercial citrus orchards, the benefits of injection must outweigh the risks of wounding associated with the application method. This study provides first insight on some of the best practices for effectively utilizing trunk injection in citrus as a crop protection strategy.


Introduction
Trunk injection is a technique for applying plant protection compounds that is typically used as an alternative to foliar spray or soil drench applications (Berger and Laurent 2019;Archer et al. 2021Archer et al. , 2022a. Trunk injection has been tested and used to combat a variety of pests and diseases, including fungal and oomycete diseases such as Dutch Elm Disease (Haugen and Stennes 1999) and laurel wilt of avocado (Crane et al. 2015(Crane et al. , 2020; insect pests and nematodes such as emerald ash borer (Cowles et al. 2009) and pine wood nematode (Sousa et al. 2012); as well as bacterial pathogens and phytoplasmas such as lethal yellowing of palms (McCoy et al. 1982) and huanglongbing (HLB, a.k.a. citrus greening) (Hu and Wang 2016;Archer et al. 2022b. The main advantages that trunk injection provides over conventional methods of pesticide application are an increased efficiency of the compound being delivered, less risk for worker exposure, reduced risk for non-target organisms and the environment, and the ability to treat trees located in populated areas where other methods are not an option (Sánchez Zamora and Fernández Escobar 2000;Wise et al. 2014). Trunk injection techniques have not been optimized for use in commercial crop production; however, there is growing interest in using injection methods to treat perennial crop species on a commercial scale (Archer et al. 2022a). Adoption of trunk injection techniques has historically emerged from devastating pest or pathogen outbreaks. The citrus industry in Florida is currently suffering major production losses due to the presence of HLB, caused by the putative pathogen Candidatus Liberibacter asiaticus (CLas) and transmitted by the Asian citrus psyllid (Diaphorina citri) (Gottwald et al. 2007). The disease was first detected in Florida in 2005 and is now considered endemic (Graham et al. 2020). The location of the pathogen within the vascular system means that therapeutic compounds must be moved systemically throughout the plant; however, the thick cuticle surrounding citrus leaves impedes penetration and uptake of many foliar applied materials (Killiny et al. 2020;Vincent et al. 2022). Additionally, the widespread abundance of the psyllid vector is a challenge for growers to manage, and foliar applications and soil drenches have not substantially reduced the psyllid population or contained the spread of the disease in Florida citrus groves (Graham et al. 2020).
Recent trunk injection experiments have identified a variety of antibiotic compounds which have shown antagonistic effects on the bacterial pathogen associated with HLB, including oxytetracycline, streptomycin, and penicillin (Hu and Wang 2016;Li et al. 2019;McVay et al. 2019), however, little is known about the physiological effects trunk injection has on the tree. Due to differences in tree physiology, target pathogens, and climate, determining the best practices to maximize efficiency and minimize potential negative consequences associated with injection is necessary for each crop/disease system. For example, injection into the roots, rootstock, or trunk of the tree may affect the rate of uptake and distribution of compounds (Tanis et al. 2012;Kondo 1972;Smith 1988;Stipes 1988) and may vary among different tree species (Holmes 1982;Tattar and Tattar 1999). Understanding which part of the citrus tree is most conducive and effective for uptake of crop protection compounds is therefore essential for applying this technology.
The time (season) of injection is another underlying factor that can significantly impact the efficacy of injected compounds. The time of year when injections are most beneficial has been established for some trees and some compounds; however, testing is required for each combination of plant species and compound. It was shown that the speed of compound uptake and distribution after trunk injection increases in rapidly transpiring trees (Aćimović et al. 2020), or rapidly transpiring parts of a tree (Hunt et al. 1974;McCoy et al. 1982). Conditions that favor high transpiration rates are therefore most likely to ensure rapid uptake and distribution of injected materials. For deciduous trees in temperate climates, late spring and summer will likely support the most efficient upward movement of injected compounds because of higher rates of transpiration in the active growing seasons. Fully watered trees, warm temperatures, and slightly windy conditions can also enhance transpiration and therefore increase the uptake rate of injected materials. However, timing of injection must also coincide with the appropriate disease stage or pest lifecycle for effective management (Aćimović et al. 2020).
The position of the injection site on the tree and the season during which injection occurs can significantly impact wounding, and especially, the internal wound compartmentalization (Dujesiefken and Liese 2015). The internal damage associated with tree wounding has been described by the concept of CODIT (compartmentalization of damage/ dysfunction in trees) developed by Shigo and Marx (1977), and expanded by Morris et al. (2020). CODIT is based on the idea that trees are highly sectorialized organisms able to compartmentalize injuries or wounds after they occur to prevent large-scale damage. After an injury, trees respond immediately by blocking decay-causing organisms, mostly fungi and bacteria, from spreading by producing three "walls" (in the axial, radial, and tangential direction) in what constitutes the reaction zone. This zone is characterized by the formation of lignin and other phenolics to protect against oxidative degradation (Nagy et al. 2022). An additional wall is formed during the healing process to create the barrier zone. This wall separates wounded wood from the new wood formed after wounding to restore the continuity of the vascular system (Shigo and Marx 1977). Various fungicidal and petroleum-based products are sold as wound treatments for enhancing wound closure after injury-however, there is limited information on the effects of these materials to promote wound healing after trunk injection.
Most published studies that investigated wounding and compartmentalization have been conducted on forest trees and species growing in temperate climates Shigo and Shortle 1979;Costinis 1980;Santamour 1984;Neely 1988;Smith 1988;Aćimović et al. 2016a). When using trunk injection technologies in agricultural crops, the wounding response by the species of interest need to be established before implementation on a large, commercial scale.
Due to the endemic nature of HLB in Florida (Graham et al. 2020) and the devastation it has wreaked on the citrus industry, there has been growing interest in using trunk injection as a management technique. Injection of oxytetracycline has garnered widespread attention because preliminary trunk injection experiments utilizing oxytetracycline have shown successful reduction in bacterial levels and some improvements in tree health (Hu et al. 2018;Li et al. 2019). More recently, oxytetracycline injection documented notable improvements in citrus tree health and productivity of severely HLB-affected citrus trees (Archer et al. 2022b, and a special need label was approved in October 2022 that allows trunk injection of oxytetracycline for HLB management in the state of Florida. In this context, the objectives of this study were to (1) assess seasonal differences in the rate of uptake of water when injecting into the rootstock trunk or into the scion trunk of HLB-affected citrus trees; (2) investigate wound closure and compartmentalization as they relate to the position of the injection site, the season during which injection is performed, and wound treatment; and (3) compare wound closure and compartmentalization after injection of oxytetracycline or water in grafted sweet orange trees. Developing a baseline understanding of uptake, wounding, and compartmentalization in citrus trees will aid in developing appropriate recommendations for implementing trunk injection as a management tool for HLB and potentially other emerging vascular diseases of citrus.

Plant material and experimental design
Two adjacent blocks composed of 'Valencia' sweet orange (Citrus sinensis) or 'Midsweet' orange (C. sinensis) trees located at the Southwest Florida Research and Education Center in Immokalee, Florida (26.463663 N,81.443892 W) were used to conduct two field trials.
'Valencia' trial. The trees were 5-year-old 'Valencia' grafted on Kuharske (C. sinensis × Poncirus trifoliata) rootstock. At the time of injection, the average tree height was 150 cm. The average rootstock and scion trunk diameter at injection height was 6.7 cm and 5.4 cm, respectively. The experiment was conducted as a randomized complete block design with month of injection as the blocking factor. Nine single-tree replications were used for measuring differences in uptake between scion and rootstock, and differences in uptake and stomatal conductance between seasons. Three individual tree replications per month were used to compare three post-injection wound treatments.
'Midsweet' trial. The trees were 5-year old 'Midsweet' grafted on Swingle citrumelo (C. paradisi 'Duncan' × P. trifoliata) rootstock. At the time of the first injection the average tree height was 130 cm. The average rootstock diameter 5 cm below the graft union and scion trunk diameter at injection height was 6.8 cm and 4.4 cm, respectively. The experiment was conducted as a randomized complete block design with six single-tree replications.
As HLB is endemic in Florida, all trees in both trials were visibly affected by the disease, displaying thin canopies and small chlorotic and/or blotchy mottled leaves typically associated with HLB (Gottwald et al. 2007).

Tree injections
'Valencia' trial. Injections were performed with deionized water every 2 months for 6 months starting in July 2020. Injections were made using Chemjet Tree Injectors (Chemjet ® , Brisbane, Australia) following the manufacturer's instructions. Chemjets are spring-loaded syringes, which release liquid at approximately 25-30 psi after activation. Briefly, a 4.3 mm (11/64 in) brad-point drill bit was drilled to a depth of 15 mm. The Chemjets were inserted directly into the drilled hole at an angle of 20°-30° and were removed once the full amount of compound was taken up by the tree (5-20 min). Two Chemjets, each filled with 20 mL of water, were used per tree. Each tree received one injection of water in the rootstock, halfway between the soil surface and the graft union, and one injection of water on the opposite side of the tree in the scion, halfway between the graft union and the lowest scaffold branch. All injections were made on sunny days between 9 and 10 am. Tree uptake was determined by measuring the amount of time it took to absorb the 20 mL of water provided by each Chemjet and expressed as mL of liquid per minute. Three postinjection wound treatments were compared: (1) Ridomil Gold SL (mefenoxam) fungicide (Syngenta Crop Protection, Greensboro, NC); (2) Tanglefoot wound sealant (liquified petroleum gases, xylenes, and acetylene black) (Tanglefoot ® , Marysville, OH), and (3) no wound treatment (control). Tree height and scion and rootstock circumference at injection height were measured at the time of injection.
'Midsweet' trial. Trunk injections were performed in June 2020 and October 2020 using Chemjet Tree Injectors as described above. Two injection treatments were compared: (1) oxytetracycline (OTC), or (2) deionized water. The OTC formulation was Arbor-OTC (Arborjet, Inc, Woburn, MA; 39.6% oxytetracycline hydrochloride) and was applied at a concentration of 2 g per tree, or 0.792 g active ingredient (0.396 g per side), dissolved in 40 mL deionized water (19,800 ppm). Each tree received two injections (20 mL each) on opposite sides of the trunk at each time point. Chemjets were positioned in the north/south direction in June, and the east/west direction in October. All injections were made on sunny days between 9 and 11 am in the scion of the tree, approximately halfway between the graft union and the lowest scaffold branch. Injections on opposite sides of the trunk were vertically offset by approximately 2 cm to avoid overlapping of injection sites. No post-injection wound treatments were applied.

Stem water potential and stomatal conductance
In the 'Valencia' trial, stem water potential measurements were conducted using a PMS1000 Scholander pressure chamber (PMS Instrument Company, Albany, OR) following the procedures outlined in McCutchan and Shackel (1992). Two fully hardened leaves per tree from the most recent flush were chosen near the trunk or a main scaffold, and in the shade. Each leaf blade was covered for 15 min with a foil-laminated bag (PMS instruments, Albany, OR) before the leaves were removed for measurements. The petiole was re-cut immediately before being placed into the pressure chamber. Stomatal conductance was measured on three leaves per tree at the time of injection (9-11 am) using an SC-1 leaf porometer (METER group, Pullman, WA) following the manufacturer's guidelines. Leaves were chosen from the newest hardened-off flush in direct sunlight with no insect damage and minimal symptoms of HLB or nutritional deficiencies. The porometer was calibrated in the field on the day of injection and measurements were taken using the "Auto" mode on the abaxial side of each leaf, avoiding the midrib or large leaf veins.

External wound size and wound closure
'Valencia' trial External wound measurements were conducted 2, 4, and 6 months after injection following the protocol of Aćimović et al. (2016a), with some modifications, for analyzing wound depth and woundwood formation. Briefly, wound size was measured using a digital caliper and is presented as the average wound diameter, measured from the exterior edges of the woundwood (Luley 2015) in the vertical and horizontal direction (Online Resource 1a). Bark cracking was measured as the extent of cracking above and below the drilled injection site. Wound depth was a direct function of the depth to which the drill was inserted (approximately 15 mm) and remained unchanged in the 6 months following injection. Therefore, each wound was categorized as either "open" or "closed" based on whether a wire (0.8 mm in diameter) could be inserted through the hole in the callus or woundwood encapsulating the injection site. Wound closure was expressed as the percentage of fully closed wounds. Woundwood is irregular in shape; therefore, woundwood formation was expressed as the percentage of wounds with excessive woundwood, with excessive defined as an area of greater than 2 cm 2 . The presence or absence of gumming in the injection site, and the presence or absence of bark cracking greater than 1 cm in length were also determined for each injection site and presented as percentage of occurrence.
'Midsweet' trial External wound measurements and wound closure assessments were conducted 4, 8, and 12 months after injection and at tree take-down in January 2022 following the protocol of Aćimović et al. (2016a) with the modifications described in the previous paragraph. Because of the large differences in woundwood formation and in the length of bark cracking between injection treatments (water or OTC), for this experiment the width of the woundwood at its widest point and the length of bark cracking were directly expressed in millimeters.

Wound compartmentalization
'Valencia' trial Six months after injection, each tree was cut at the soil surface. Following measurements of the external wound size, trees were cut horizontally using a vertical band saw (Rikon 10-305; Rikon Tools, Billerica, MA) in 3 cm-increments above and below each injection site to measure the length of the internal axial discoloration (Online Resource 1b) (Shigo and Shortle 1979;Morris et al. 2020). Each 3 cm-disk containing the drill wound was sliced vertically though the injection site and scanned at 400 dpi on a flatbed scanner (Epson Perfection v850 Pro; Epson America, Los Alamitos, CA) to measure the wound reaction (compartmentalization) area in the axial direction. The area was determined using ImageJ 1.52p software (Rasband, NIH, USA) by converting images to grayscale and manually thresholding each image to represent the area of discoloration (Online Resource 1c) and expressed in square centimeters. The presence of gumming inside the injection ports was visually determined and expressed in percent of occurrence. Following measurements and imaging, all aboveground tree biomass was dried for 2 months at 60 °C to determine dry weights at the time of tree removal for each tree.
'Midsweet' trial In January 2022 trees were cut at soil level and segmented as described for the 'Valencia' trial to calculate the wound compartmentalization area for each injection site. Due to the large differences between the water and OTC treatments for this experiment, the cross-sectional images of each injection site were also scanned to measure the wound compartmentalization area in the radial/tangential direction. Because the wound compartmentalization area was extensive and discoloration extended far beyond in the axial direction, only the compartmentalization area within 3 cm above and below the injection site was measured.

Statistical analysis
All statistical analyses were performed in Rstudio Version 1.4.1717 (R Core Team). Prior to analysis, data were tested for normality and homogeneity of variance.
'Valencia' trial A two-way analysis of variance (ANOVA) was performed to compare the rates of uptake with month of injection (Jan, Mar, May, Jul, Sep, or Nov) and position of injection (scion or rootstock) as factors. Wound closure, excessive woundwood formation, occurrence of gumming, and bark cracking were analyzed using a chi-square test of independence to determine their relationship with month of injection, injection position, and wound treatment (fungicide or pruning sealant). External wound size, internal wound compartmentalization area, and length of internal discoloration were analyzed using a three-way mixed model ANOVA with wound treatment and the interaction of month of injection and injection position as factors. Model selection was based on minimization of the Bayesian information criterion across all response variables. Where differences were significant (p < 0.05), post-hoc comparison of means was performed using Tukey's honest significant difference test. Correlations between scion uptake rate, tree size (circumference and biomass), wound size, internal compartmentalization area, and discoloration were calculated for each month using Pearson's product-moment correlation.
'Midsweet' trial A two-way ANOVA was performed for comparison of differences in wound size with month of injection (June or October) and compound (OTC or water) as factors. Where differences were significant (p < 0.05), a post-hoc comparison of means was performed using Tukey's honest significant difference test. Wound closure and the occurrence of bark cracking were analyzed using a chisquare test of independence to determine the relationship with the month of injection and the injected compound.

Rate of uptake
'Valencia' trial There were significant differences among months and injection position for the rate of uptake of injected water (Table 1). The rate of uptake was slowest in January and March (2.0 mL/min and 1.7 mL/min, respectively) and fastest in May and September (3.6 mL/min and 3.8 mL/min, respectively). Water was taken up significantly faster when injection occurred into the scion (3.2 mL/min) than the rootstock (2.4 mL/min), but there was no interaction of month and injection position. The seasonal uptake pattern was weakly correlated (R 2 = − 0.29, p = 0.002) with the stomatal conductance measured at the time of injection (Online Resource 2). The slowest uptake rate measured in March corresponded to the lowest measured stomatal conductance (162 mmol/m 2 s). While the fasted uptake rates were measured in September and May, the highest stomatal conductance were measured in September and November (370 mmol/m 2 s and 389 mmol/m 2 s, respectively). There was a significant correlation (R 2 = 0.77, p < 0.01) between tree size (circumference and dry weight) and rate of uptake during March, however, this correlation was not significant at other times of the year (Online Resource 3). The stem water potential varied significantly among months (Online Resource 2) and was significantly higher in July (9.7 bars) and May (7.2 bars) than in January, March, and September (5.6-6.1 bars). There was no significant correlation between stem water potential and month of injection or rate of uptake (R 2 = 0.13, p = 0.190).

Wound closure
'Valencia' trial There was a significant effect of month of injection on wound closure after 2, 4, and 6 months ( Table 2). Two months after injection, 33% of wounds inflicted in July were closed compared to 6% of wounds inflicted in September and 0% inflicted in the other months. Four month after injection, 78-100% of wounds inflicted in March, May, July, and September had closed but only 11% and 22% in November and January, respectively. After 6 months, more than 94-100% of wounds created in January, March, May, and July were closed, while 83% and 67% of wounds created in September and November, respectively were closed. There was no significant effect of wound treatment on wound closure at any time after injection, but there was a significant effect of injection position. Two months after injection, 19% of wounds inflicted on the scion had fully closed, compared to 0% in the rootstock (Table 2). Four months after injection, 74% of wounds on the scion and 56% of wounds on the rootstock wounds had closed. After 6 months, 100% of the wounds in the scion were closed, compared to 81% in the rootstock.
'Midsweet' trial There were significant interactions between month of injection and the injected compound on wound closure (Table 3). Four months after injection of water, 100% of wounds were closed when injections had occurred in June compared to only 25% of wounds when injection occurred in October. When water was injected in October, 58% of the wounds closed fully by 8 months, and 100% by 12 months. In contrast, when OTC was injected in June or October, none of the wounds had closed by 4 months after injection. By 8 months, 36% of wounds inflicted by OTC injection in June had closed and by 12 months 83% had closed. When OTC was injected in October, none of the wounds had closed by 8 months, and by 12 months, only 58% of wounds were fully closed.
External wound size, gumming, and bark cracking 'Valencia' trial There was no significant effect of the month of injection on woundwood formation (Table 4). However, the formation of excessive woundwood occurred significantly more often on the rootstock than the scion. Application of a pruning sealant (Tanglefoot) caused excess woundwood formation in 33% of wounds compared to 0% and 6% after application of a fungicide (Ridomil) and water, respectively. There were no significant differences in the occurrence of gumming among different months of injection, between wound treatments, or between injection positions (Table 4). The occurrence of bark cracking at the wound site occurred significantly more often on the scion (31%) than on the rootstock (0%), but there were no differences related to the month of injection or the wound treatment.
The external wound size was significantly affected by the month of injection and the injection position (Table 5). Injection into the rootstock in May resulted in a larger wound size after 2 months (12.5 mm) compared to in the other months (6.4-8.2 mm), regardless of the injection position. After 4 months, injection wounds inflicted in May remained the largest (13 mm), and wounds were larger after injection into the rootstock (10.1 mm) compared to the scion (8.3 mm), but there was no interaction between month of injection and injection position. There were no significant differences in wound size after 6 months related to the month of injection nor the position of the injection. There was a significant effect of wound treatment on wound size after 2, 4, and 6 months ( Table 5). The pruning sealant caused a larger wound diameter (8.5-14.1 mm) than the fungicide (7.2-9.5 mm) or the non-treated control (6.8-9.0 mm) ( Table 5). In July and September, there was a significant correlation between scion trunk circumference and external wound size (Online Resource 3) but there was no correlation for the other months.
'Midsweet' trial Four months after injection the average external size of wounds created in June was 12.9 mm, compared to 7.3 mm for wounds created in October, but differences were significant only at a level of p = 0.071 (Table 6). After 12 months there was no significant difference in wound size based on the month of injection, and wound sizes were 19.2 mm and 19.3 mm for June and October injections, respectively. There were significant differences in wound size depending on the injected compound 8 and 12 months after injection. Injection of OTC caused an average wound size of 21.0 mm and 26.3 mm after after 8 months and 12 months, respectively, while injection of water caused an average wound size of 9.3 mm and 12.2 mm, respectively. Bark cracking was more extensive in June than in October 4 and 8 months after injection, but differences were significant only at a level of p = 0.076 and p = 0.064. Bark crack length was 51.6 mm (Jun) and 40.0 mm (Oct) after 12 months but differences were not significant. The type of compound injected had a significant effect on the length of bark cracking. Injection of OTC caused cracks that were 56.8 mm and 63.2 mm in length after 8 months and 12 months, respectively (Table 6). In contrast, injection of water caused cracks that were only 22.2 mm after 8 months and 28.3 mm after 12 months.

Wound compartmentalization
'Valencia' trial Wound compartmentalization in 5-yearold citrus trees after injection of water followed the principles of CODIT with walls 1-3 of the reaction zone and the barrier zone (wall 4) clearly defined (Fig. 1a). While compartmentalization in the radial and tangential direction was effective in preventing spread of damage past the drill   . 1 Wound compartmentalization in 5-year-old citrus trees injected with water in accordance with the CODIT (compartmentalization of damage/dysfunction in trees) model. a The four walls of compartmentalization in a trunk cross section (left) and a longitudinal trunk section (right). Note that radial and tangential compartmen-talization is more effective than axial compartmentalization. b Strong (left) versus weak (right) axial compartmentalization after injection of water. Note that the outer and metabolically more active sapwood compartmentalized more effectively (red bracket) than the inner wood site, compartmentalization in the axial direction was less effective in some cases (Fig. 1b). Also noticeable was the better ability of the younger (outer) wood to compartmentalize than the older (inner) wood, evident by longer columns of axial discoloration towards the center of the trunk. There was a significant effect of both month of injection and wound treatment on the area of axial compartmentalization (Table 7). Injection in July led to an axial compartmentalization zone of 2.1 cm 2 in area compared to 1.6-1.7 cm 2 after injection in January, September, and November. The application of fungicide after injection increased the axial compartmentalization area to 2.0 cm 2 , compared to 1.7 cm 2 after use of either the pruning sealant or no wound treatment. There was no significant effect of the injection position on the area of axial compartmentalization, and there was no interaction of injection position and month of injection. There was a significant interaction for the length of visible discoloration above and below the compartmentalized wound zone between the month of injection and the injection position (Table 7). Except for July, injection into the rootstock resulted in smaller columns of discoloration than injection into the scion. The longest discoloration (14.0-16.0 cm) was measured when injections occurred in March, May, September, and November into the scion, and the shortest when injections occurred in January into the rootstock, or in July into the scion.
External damage at the wound site manifested as excessive woundwood formation, bark cracking, or gumming, did not result in a concomitant increase the in the area of internal compartmentalization (Online Resource 4).
'Midsweet' trial There were no significant differences in the compartmentalization zone areas related to the month of injection, but the type of the compound injected significantly affected the area of both the axial and the radial/tangential compartmentalization (Table 8; Fig. 2). Injection of OTC resulted in an axial compartmentalization area of 21.1 cm 2  compared to 3.6 cm 2 after injection of water. Similarly, the tangential compartmentalization zone area was 9.4 cm 2 after OTC-injection and only 1.8 cm 2 after water injection.

Uptake rate of trunk-injected water is related to tree transpiration and other factors in HLB-affected citrus trees
The rate at which trunk-injected water was taken up by the tree was related to the transpiration rate, which was estimated by measuring the stomatal conductance. Injection in March resulted in the slowest rates of uptake, which can be attributed to the trees' physiological state and to environmental conditions leading to reduced transpiration. Most notably, in March, the annual spring flush was still in the initial stages of development at the time of injection and trees did not have fully expanded new leaves that drive transpiration. Additionally, in southwest Florida, the rainy season had not yet started and therefore trees may have suffered from mild water stress. The drastic increase in the uptake rate of injected water in May is likely attributable to the new fully expanded, and photosynthetically active leaves, warmer temperatures, and a more saturated root zone due to the start of the rainy season, conditions which persisted through summer and fall (July, September, and November). As described in Aćimović et al. (2014aAćimović et al. ( , 2020 for apple trees and in earlier studies on coconut palms (Hunt et al. 1974;McCoy 1976), this study found that conditions which promote optimum transpiration also increase the rates of uptake of injected materials in citrus trees. In apple trees the transpiration rate was found highest in June and July when the days were longest, the light intensity was highest, and soils contained ample water (Fleancu 2005). Monthly injections in our study occurred on days when environmental conditions were similar to the historic average conditions for that time of year; however, day-to-day variations in rainfall, humidity, wind speed, and the tree physiological state can influence uptake within each season. Additionally, the HLB disease state is likely to influence the rate of uptake. It was shown that water use of HLB-affected trees is significantly lower than water use of healthy trees (Hamido et al. 2017), likely reducing uptake and distribution of injected compounds. As HLB is endemic in Florida (Graham et al. 2020), no healthy trees were available for comparison. When injecting pesticides, the uptake rate will also be influenced by their chemical properties, including water solubility, the acid dissociation constant (pKa), and the normalized organic carbon to water partition coefficient (KOC) (Riederer 2004;Aćimović et al. 2014bAćimović et al. , 2015Aćimović et al. , 2016bArcher et al. 2022a). The comparison of rootstock and scion trunk injections showed minor differences in the uptake rate between the two organs, but major differences in the length of axial discoloration. The discoloration after injection of water in the scion was longer than in the rootstock and may indicate differences in the speed of compound movement after injection. This may have been driven by intra-organismal variation of vascular transport (Jacobsen et al. 2018), species/cultivar-specific structural differences of the vascular conduits between the two organs (Albrecht, unpublished), or graft union effects (Goldschmidt 2014). Whereas most scions used in commercial citrus production in Florida are sweet oranges, rootstock are quite variable and genetically diverse (Rosson 2021;Kunwar et al. 2023). If performing injections into the rootstock, more research would need to be conducted to determine the uptake rate and other factors associated with trunk injections for each rootstock, including the capacity to compartmentalize wounds and prevent decay. Note the extensive area of compartmentalization/discoloration in the axial (left) and radial/ tangential (right) direction

External wound size is affected by wound treatment, injection position, season of injection, and the nature of the injected compound
Use of a wound/pruning sealant noticeably increased the external wound size compared to application of a fungicide or leaving the wound untreated. Our findings support previous reports that show the tendency for wound treatments to inhibit wound closure (Shigo and Shortle 1984;Biggs and Peterson 1990;Dujesiefken and Liese 2015;Holland et al. 2021), possibly due to their inhibitory effect on callus and woundwood formation (Luley 2015).
After injection with water, wound closure was significantly faster on the scion compared to the rootstock and all instances where wounds did not fully close within 6 months were observed on the rootstock. Callus production, followed by the production of woundwood (Luley 2015) is the visible response of a tree to wounding, which, when excessive, may elicit concern among growers. Development of excessive woundwood was more common after injection into the rootstock compared to injection into the scion trunk; however, rates of wound closure were faster in the scion. A slower wound closure in combination with the proximity of the rootstock to the soil may increase the risk of secondary infections with soil-borne fungi such as Phytophthora, which is prevalent in Florida soils. Cambial activity, which is a function of tree metabolic activity (Myśkow et al. 2021), is essential for callus-and woundwood formation and wound closure (Aćimović et al. 2016a, Luley 2015. Additional studies examining the relationship between cambium activity and woundwood formation in the rootstock compared to the scion may provide insight into the reason for the more excessive woundwood formation in the rootstock. According to Luley (2015), callus and woundwood formation are influenced by species, tree health, and the time of the year the wound was inflicted. Hormonal relationships and carbohydrate balance between scion and rootstock in combination with genetic differences likely contributed to the different wound response in the two organs. Despite more excessive woundwood formation in the rootstock, there were no differences in the size of the area of internal compartmentalization.
Cambium cells surrounding the wound site are responsible for generating the barrier zone that separates the new wood from the old, damaged wood. Injury to the cambium initiates callus formation, followed by the suberization and lignification of cells to prevent the movement of decay-causing organisms into the newly created woundwood (Eyles et al. 2003;Morris et al. 2020). New woundwood grows and encapsulates the injection site, leading to a restored continuity of the vascular cambium (Smith 2015). In both field trials, wound closure following injections in fall and winter was slower than in spring and summer. This seasonal effect of wound closure is likely due to a reduction in tree metabolic activity during fall and winter resulting in reduced cambium activity (Rajput and Rao 2001;Budzinski et al. 2016;Bhalerao et al. 2017). Compared with water, injection of OTC resulted in more excessive woundwood formation, slower wound closure, and more extensive bark cracking. This supports previous studies which showed that the injury created from injecting therapeutics may be greater than the physical injury associated with the injection itself (Costinis 1980). Like wound closure, bark cracking appeared to be related to the tree metabolic activity. Injections conducted in spring and summer, when trees are actively growing, resulted in more bark cracking than injections conducted in fall and winter. However, there was no corresponding increase in internal damage, and wound closure was more rapid, corresponding to the better healing capacity of metabolically active tissues. In the 'Valencia' trees that had been injected with water, bark cracking was more frequent when injection occurred during months of high metabolic activity and was only present in the scion. Aćimović et al. (2016a) hypothesized that bark cracking is a response to the interruption of sap flow caused by the injection, leading to cambium and phloem dehydration. Additional studies investigating differential responses to wounding in the cork cambium (phellogen) and vascular cambium may provide insight on why bark cracking is more frequent in the scion of rapidly metabolizing trees. Genetic differences between scion and rootstocks in combination with carbohydrate and hormonal balances between the two organs, as mentioned above, are probable factors contributing to the different response. Like the excessive woundwood formation, and as reported for ash trees (Tanis and Mccullough 2016), bark cracking was not associated with a higher likelihood of internal injury like discoloration of the wood or larger compartmentalization zones and did not impact the ability of the wound to fully close. Despite raising concern among growers, bark cracking is unlikely to increase the risk of pest or pathogen intrusion (Luley 2015).

Internal wound compartmentalization is affected by season of injection and the nature of the injected compound
In this study, the area of internal wound compartmentalization in the axial direction after injection of water was largest when injections were performed in July and smallest when they were performed in fall and winter (September, November, and January). Axial damage or decay is the most difficult to contain because bacteria and fungi introduced into a tree's vascular system can move quickly within the xylem along the transpiration stream (Pouzoulet et al. 2020;Kashyap et al. 2021). Injection in fall and winter might have limited the size of the axial compartmentalization zone because of a reduced xylem flow rate concomitant with a reduced transpiration. The application of a fungicide as a wound treatment also increased the area of axial compartmentalization. This is consistent with previous studies which found that the application of xenobiotics into an injection port will increase the amount of decay or damage caused by wounding Costinis 1980;Stennes and French 1987). In accordance with the observations on other xenobiotics and as previously observed , injection of OTC in the 'Midsweet' orange trees resulted in a large compartmentalization zone area. It is unclear, to what extent the discoloration associated with the compartmentalization zone is due to the production of phenolics associated with the plant defense response (Morris et al. 2020), physical damage associated with phytotoxicity of the OTC in combination with the low pH (2.0) of the formulation, or the color of the formulation (Tanis and Mccullough 2016). Wound compartmentalization and phytotoxicity are likely affected by the formulation of an injected pesticide, and crop protection materials formulated for injection need to be designed to move readily in the xylem (Aćimović et al. 2014b(Aćimović et al. , 2020(Aćimović et al. , 2016b to be effective and reduce damage. The OTC formulation used here is formulated specifically for trunk injection in tree crops including citrus. Although the compartmentalized zone after OTC injection remained firm and the structural integrity of the wood intact, it may be assumed that the vascular conduits in this zone have lost their functionality. When injecting OTC in citrus trees to manage HLB, the potential damage caused by the injection must therefore be evaluated against the expected benefit.
In this study, neither radial nor tangential compartmentalization after injection with water exceeded the width and depth of the drill bit used for creating the injection ports. This is consistent with the CODIT model which states that tangential and radial damage/dysfunction are more effectively blocked than axial damage. The efficacy of wound compartmentalization is directly associated with the arrangement of the ray and axial parenchyma cells in the xylem, which are responsible for plant defense responses, including the production of secondary metabolites that block the spread of decay-causing organisms (Biggs 1985;Morris et al. 2016Morris et al. , 2020. The length of axial discoloration after injection of water, called the 'zone of slightly altered wood' in the CODIT model based on Shigo and Marx (1977), was longer in the scion compared to the rootstock. The rate of uptake of injected water was also faster in the scion. This further supports the notion that the extent of the discoloration is determined by the speed and distance injected materials travel immediately after injection, which is influenced by tree physiological and environmental conditions, and the chemical properties of the injected pesticide (Aćimović et al. 2014b(Aćimović et al. , 2020. A slight discoloration was also evident after application of the black petroleum-based wound treatment; however, this did not result in a larger area of axial compartmentalization. Our studies also showed that even after water injection compartmentalization was less effective in the inner (older) wood, which is in accordance with Dujesiefken and Liese (2015). The reason for this is a lower metabolic activity of the older wood compared to the highly active sap wood located nearest the cambium. Using shallow rather than deep injection devices is therefore recommended for application of trunk injection in commercial citrus production or other tree systems to ensure effective compartmentalization of damage and/or decay and longevity of the tree.
In both trials all trees were affected by HLB, which likely influenced the rate of uptake, wound closure, and the extent of internal compartmentalization/discoloration. Phloem plugging associated with callose deposition in HLB-affected citrus has been shown to reduce translocation and transport of carbohydrates (Etxeberria et al. 2009), while a poor root density common for diseased trees reduces water uptake (Kadyampakeni et al. 2014). A disease-induced lower tree metabolic activity may also contribute to a lessened ability to compartmentalize wounds or develop woundwood essential for effective wound closure (Smith 2015). The imbalances in carbohydrate metabolism associated with HLB may also affect the wound response as the conversion of carbohydrates to secondary metabolites is essential for the plant response to wounding and defense of decay-causing organisms (Morris et al. 2020).

Conclusions
Results from this study can be used to begin establishing recommendations for best practices associated with trunk injection in sweet orange trees in Florida to combat HLB or other emerging vascular diseases. First, we found that the rate of uptake after injection in the scion is more rapid than uptake in the rootstock, and our results support existing literature on other tree species which shows that uptake is most rapid during conditions that promote transpiration. However, uptake and movement are expected to vary depending on the injected formulation and the species. We also found that injection of water in the scion resulted in faster wound closure and less external injury compared to injection in the rootstock. Injection during periods of reduced tree metabolic activity resulted in slower wound closure but reduced internal discoloration, and we found no benefits associated with the application of wound treatments at the injection site. The petroleum-based pruning sealant used in this study resulted in excessive woundwood formation and did not enhance wound closure and internal compartmentalization. The application of a fungicide after injection did not affect the external wound size but increased the area of axial compartmentalization compared to the untreated control. Despite the negative effects of HLB on citrus tree health and metabolism, trees in this study were able to effectively compartmentalize wounds inflicted by using a drill-based trunk injection method. However, a considerably larger external wound size and internal area of compartmentalization was observed when injecting OTC, which may elicit concerns when using this material in commercial production. In October 2022, a FIFRA section 24(c) special local need label was approved allowing trunk injection of OTC in mature citrus trees in Florida. Although application of OTC formulated for injection has proven effective for management of HLB, the long-term impacts of wounding and damage to the vascular conduits associated with injection of OTC will need to be investigated before growers can adopt trunk-injection for HLB management on a large scale.
Author contribution statement LA and UA conceived and designed the study; UA acquired the funding, administered the project, and provided supervision; LA conducted the experiments and collected and analyzed the data; LA wrote the first manuscript draft; UA reviewed and edited the manuscript; LA and UA approved the final manuscript. Data availability Raw data for this study can be made available upon reasonable request.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.