The effectiveness of the neem product TreeAzin® in controlling Cameraria ohridella (Lepidoptera: Gracillariidae: Lithocolletinae)

Infestation by invasive horse-chestnut leaf miner, Cameraria ohridella Deschka & Dimić, permanently lowers the aesthetic and cultural value of horse-chestnut in Central Europe. In 2017–2018, in urban zones in the cities Parchovany and Strážske in the eastern part of Slovakia, we assessed the effi cacy of systemic applications of TreeAzin®, an azadirachtin-based product, in controlling Cameraria ohridella in trials in which it was microinjected into tree trunks. A total of 16 Aesculus hippocastanum trees were treated with 3 ml of TreeAzin® per centimetre diameter at breast height [DBH] and another 17 were treated with 5 ml of the same product per centimetre at DBH, at two study plots. In total, 18 trees were left untreated as controls. In this fi eld experiment, we confi rmed signifi cantly higher effi cacy in the year of application and the following season. Statistically signifi cant differences were found in the average leaf damage caused by C. ohridella, between treated (4.2–24.5% avg. leaf damage) and untreated trees (75.5–94.3% avg. leaf damage). At the end of the fi rst growing season, 81.2–95.0% of the untreated control tree crowns were defoliated while defoliation of the treated trees was 19.2–31.6%. Both the 3 and 5 ml/cm doses were equally effective in terms of crown and leaf damage; no statistical differences were found in average leaf and crown damage between trees treated with doses of 3 ml/cm and 5 ml/cm. Similar results were also obtained the following year. Leaf damage of treated trees was 40.4–16.8% and of untreated trees 67.9%. Crown damage of treated trees was 49.7–59.8% and of untreated trees 78.8%. During the period of this study, the crowns of all the treated trees were statistically and visually healthier and fuller than those of untreated trees. Thus, the effi cacy of this systemic insecticide in controlling C. ohridella in Europe is very promising and provides a suitable treatment for reducing the incidence of this invasive pest.


INTRODUCTION
The horse-chestnut leaf miner, Cameraria ohridella (Gracillariidae: Lithocolletinae), was described by Deschka & Dimić (1986) from Macedonia where it was discovered defoliating European horse-chestnut trees, Aesculus hippocastanum L., near Lake Ohrid. This pest has been recorded in many other European countries since its initial discovery. While many of these new identifi cations may represent populations that have been present, but undiscovered, for many years, there is also strong evidence to suggest that C. ohridella can quickly disperse into new areas (Gilbert et al., 2005). Horse-chestnut leaf miner can disperse over long distances aided by human transport and shorter distances by fl ight (Zúbrik et al., 1999;Gilbert et al., 2004). This insect was recorded in Austria near the city of Linz in 1989and around Enns in 1990(Pschorn-Walcher, 1994. It was discovered in Italy in 1992(Hellrigl, 1998, In an effort to identify more environmentally acceptable control options that are effective for use against invasive insect pests in Europe, we have concentrated our focus on TreeAzin ® . TreeAzin ® Systemic Insecticide is a proprietary formulation of the natural botanical insecticidal group of compounds referred to as azadirachtins. Formulations prepared from neem seed extracts adversely affect a variety of defoliating and wood-boring pests and are rapidly taken up and translocated following stem injection . Azadirachtin is a botanical insect growth regulator and because of its structural resemblance to the insect moulting hormone ecdysone, azadirachtin inhibits PTTH thereby inhibiting moulting, metamorphosis and development of the female reproductive system. Immature insects exposed to azadirachtin (mainly by ingestion) may moult prematurely or die before they complete their development (Rembold & Sieber, 1981). Those insects that survive treatment are likely to develop into deformed adults, incapable of feeding, dispersing or reproducing (Mordue et al., 2000). Besides the well-known insect growth regulating activity, azadirachtin is also a strong antifeedant for many insects (Schmutterer, 1990). Azadirachtins are non-persistent both within trees and in the environment generally and also exhibit relatively low toxicity to mammals, birds, bees and other non-target invertebrates (Kreutzweiser et al., 2011). By the time of senescence, essentially all azadirachtin residues have dissipated from tree leaves , and therefore no negative impacts to detritivores have been observed when fed treated leaves (Kreutzweiser et al., 2011). The azadirachtin-based systemic insecticide (TreeAzin ® ) is being widely used for managing invasive insect pests in Canada and the United States. TreeAzin ® has been proven effective against and is registered for use on a variety of lepidopteran, coleopteran, hemipteran and hymenopteran insect pests. Thus, it was thought that TreeAzin ® could be effective against the horse-chestnut leaf miner, C. ohridella, and was used in this study. The aim of this research was to assess the effi cacy of two doses (3 ml and 5 ml/cm DBH) of TreeAzin ® , (5% azadirachtin) injected into tree stems to control C. ohridella in fi eld trials. We predicted that TreeAzin ® will offer a high, dosedependent measure of control of C. ohridella.

Study plots
Two plots in Eastern Slovakia, Parchovany (48°44´51.0˝N, 21°42´18.4˝E, 110 m a.s.l.) and Strážske (48°52´21.8˝N, 21°50´03.0˝E, 130 m a.s.l.) were selected for assessment. These villages are situated in the eastern part of Slovakia in middle Europa. Predominant type of soil is fl uvisoil. Both were in urban areas of the above-mentioned villages. One alley of trees more than 100 m long borders a road built in 18 th and 19 th century, near each village.

Trees selected for experiments
At Parchovany 24 A. hippocastanum trees were selected and at Strážske 27 trees. These were numbered and marked using forest marking spray. A total of 16 trees were treated with 3 ml/cm of TreeAzin ® at breast height (DBH), 17 others with 5 ml/cm of the The European horse chestnut (A. hippocastanum) is the primary host tree for C. ohridella on this continent (Deschka & Dimić, 1986;Pschorn-Walcher, 1994;Dimić et al., 2005). The moth can also attack and develop on other species of the genus Aesculus (Tomiczek & Krehan, 1998;Freise et al., 2004;Dimić et al., 2005). However, there is evidence to suggest that the closely related Aesculus pavia, which is frequently planted in the same environment, is strongly resistant to C. ohridella (Kobza et al., 2011). It is occasionally reported developing on maple trees (Acer pseudoplatanus and A. platanoides) in which case damage levels may be as high as on horse-chestnut (Pschorn-Walcher, 1997;Hellrigl, 1998;Freise et al., 2004). To date, there is little evidence that C. ohridella represents a major risk for A. pseudoplatanus (Péré et al., 2010).
European horse chestnut is frequently planted in Europe along roads, in urban parks, large gardens and forest environments (Zúbrik et al., 2006). For several reasons, its cultivation in Central Europe was very popular during the 18 th and 19 th centuries, where it was obtained from the Balkan Peninsula (Adam, 1997).
Mining by C. ohridella has the potential to signifi cantly reduce a tree's leaf area by midsummer. Heavy infestations decrease the aesthetical value of trees, resulting in branches dying and repeated sprouting and fl owering in autumn (Percival et al., 2011). Defoliation has also an effect on seed quality (Thalmann et al., 2003). However, widespread dieback of horse chestnut trees has so far not been observed (Butin & Führer, 1994;Kenis & Forster, 1998).
While there are a number of control options for C. ohridella, chemical control measures for the control of this pest are commonly used in urban forest environments (Blümel & Hausdorf, 1996;Zúbrik et al., 2006). Foliar sprays of synthetic and highly toxic insect growth regulators, such as difl ubenzuron, trifl umuron and fenoxycarb are the most popular insecticides; however only difl ubenzuron consistently results in a high level of control (Blümel & Hausdorf, 1996;Gilbert et al., 2003;Głowacka, 2005a;Głowacka et al., 2009). Mechanical control methods such as removing dead leaves, in which pupae overwinter and burning or composting them, remains the most environmentally friendly method used in urban parks and several small cities. This method is recommended by many authors (Kehrli & Bacher, 2003;Pavan et al., 2003;Głowacka, 2005b;Kukula-Mlynarczyk & Hurej, 2007). Classical biological control against C. ohridella also has some potential, but the natural enemy spectrum of this pest is rather small and not very effective (Kenis et al., 2005;Tóth & Lukáš, 2005;Ferracini & Alma, 2007). Systemic insecticides have also been successfully used to control it in the past with good results (Feemers, 1997;Labanowski & Soika, 2003;Pavela & Bárnet, 2005;Ferracini & Alma, 2008;Kobza et al., 2011). There is also the possibility to use glue bands and/or liquid glue on tree trunks (Percival, 2016). Another method is the attract-and-kill technique in urban environments using baited pheromone traps, but results indicate low effi cacy in the case of C. ohridella (Sukovata et al., 2011). same product at DBH level, at both sites. In total, 18 trees were left untreated as controls at these two locations.
Trees with little or no visual symptoms of decline in terms of dead branches, small wounds or dead stem wood were selected for experiments. The tree age varied between 105 and 115 years, average diameter at breast height of these was 57.75 cm (± 5.99 SD) at Parchovany and from 110 to 120 years, average DBH was 62.9 cm (± 17.76 SD) at Strážske. The age of trees was determined based on the city governments' written records. The health status of the trees selected corresponded to their age and some dying branches were present in their crowns.

Treatment
TreeAzin ® is an botanical injectable systemic insecticide formulated with 5% azadirachtin (lot formulation contains 50.90 ± 4.74 azadirachtin A&B), an extract of the neem tree (Azadirachta indica A. Juss.). The main mode of action of azadirachtin is as an insect growth regulator, which reduces insect fecundity and has anti-feeding properties. The product was applied using the EcoJect ® microinjection system, which is a patented technology for the application of systemic insecticides in urban forests and ecologically sensitive areas. Trunk injections were done using a 12-volt battery operated drill with a 15/64" (5.95 mm) drill bit to create a number of holes around the trunk of a tree. Holes were drilled, approximately 13-15 cm apart around the tree, spiralling slightly upwards. The number of holes and volume of product required was determined by the DBH. Each hole was drilled at a 45-degree downward angle and drilled to a depth of about 3 cm beyond the bark. After the holes were drilled, a nozzle was placed in the injection hole and a 20 ml or 8 ml canister was mated to the nozzle and left to empty (Fig. 1). Once the canister had emptied, the canister and nozzle were removed from the trunk. Applications were applied in the fi eld from 24 to 28 April 2017 when the average temperature ranged from 6.7°C (April 24) to 16.4°C (April 26), there was no rain, a S to SE wind direction and wind speed of 7 to 20 km/h.

Design of the fi eld experiment
The experimental design (Table 1) was based on the EPPO Bulletin on Effi cacy Evaluation of Plant Protection Products: Design and analysis of effi cacy evaluation trials (OEPP/EPPO, 2012). Trees were assigned to either the low dose (3 ml/cm DBH) or high dose group (5 ml/cm DBH) of TreeAzin ® or to the untreated control group. These particular dose rates were chosen based on their effi cacy in previous defoliator trials (Bioforest Technologies, 2004, 2005a. Untreated (blank) trees between treated and control trees were not involved in experiments and served as a barrier between the experimental trees. Trees were randomly arranged in blocks: at Parchovany (blocks 1-8) and at Strážske (blocks 11-20). Some selected trees were later excluded from the experimental design because during treatment their health status was found to be unacceptable (indicated by an X in Table 1).

Assessment of pre-treatment trees
A few days before insecticide applications, 4-5 branches were cut from each tree, 2-3 leaves per branch, at each of the four cardinal points of the tree (north, east, south and west) in the lower canopy (where the fi rst generation usually occurs), i.e. 8-12 leaves per tree were used to assess the numbers of eggs and larval galleries in the pre-treatment populations. Branches were put into labelled bags and brought to the laboratory to count the number of eggs and galleries on the upper surface of the leaves using a binocular microscope, Leica M205 C.

Assessment of post-treatment damage
Assessments of post-treatment damage were undertaken fi ve times (May 16, July 11, August 2, August 30 and September 20) in 2017. At every assessment, three leaves at each of the cardinal points (12 leaves per tree) were collected and used to estimate the number of galleries, or the percentage of leaf damage caused by C. ohridella (Fig. 3). For estimating leaf damage, we used the scorecard published by Gilbert & Gregoire (2003) (Fig. 2).
The following season, in 2018, two assessments of the health of the trees were carried out at Parchovany, one on July 23 and second on September 10. The same method of damage assessment was used as in 2017. The Strážske plot was not included in the 2018 assessment, because trees were unexpectedly treated with another insecticide by the city government.

Statistical analysis
For analyses, nonparametric statistical Kruskal-Wallis tests and pairwise multiple comparisons of mean ranks for particular p-values were carried out using STATISTICA 10 (StatSoft).

Assessment of pre-treatment trees
At Parchovany there was only an average of 0.9 (± 1.62 SD) eggs per leaf (103 eggs on 119 leaves) and at Strážske, 0.6 (± 1.59 SD) eggs per leaf (69 eggs on 118 leaves). These results indicate that treatment occurred at the beginning of the oviposition period of the pest. There was no signifi cant difference in the density of eggs laid on trees at both plots (Parchovany: p = 0.9698 and Strážske: p = 0.5280). There was a signifi cantly lower egg density (p = 0.0532) of C. ohridella at Strážske than at Parchovany.

Assessment of post-treatment damage
On the fi rst day sampled in the year of treatment (May 16, 2017), we counted the number of larval galleries caused by the youngest, fi rst instar larvae. Most of them look like a simple 2 mm long tunnel or a 1.5-2.0 mm patch at the end of short petiole (Fig. 3). They were very frequent at Parchovany, showing that infestation by C. ohridella at this locality was very high, whereas at Strážske it was significantly lower ( Table 2).
The level of leaf damage recorded on treated and untreated trees on all the collection dates (July-September) were  N -no. of leaves; SD -standard deviation; averages with different letters differ signifi cantly at a P ≤ 0.05 level. signifi cantly different for both plots in 2017. Average leaf damage was signifi cantly higher on untreated than treated trees. There was no signifi cant differences in the incidence of damaged leaves between trees treated with applications of 3 ml and 5 ml of the pesticide (Table 3).
Overall crown damage increased signifi cantly from the fi rst to the fi nal assessment at both localities in 2017 (Table 4). The difference in the damage to the crowns of treated and untreated trees was statistically signifi cant. There were no signifi cant differences in the results for the 3 ml and 5 ml applications. Variability in the estimates of overall crown damage was not as high as that for damaged leaves.
At the second damage assessment on August 2 an unequal distribution of TreeAzin ® to certain parts of the crowns was noticed, with some branches signifi cantly more damaged by C. ohridella than others. This was expected due to the age and size of the trees included in this study, in which there may have been areas of dead wood or other vascular damage unknown to us at the time of treatment as they were not visible on the exterior of the tree; this may have caused the high variability in the measurements of leaf damage recorded on treated trees (Table 3).
At Strážske, the damage caused by C. ohridella was lower than at Parchovany as discussed previously. But at both Parchovany and Strážske, signifi cant differences were recorded between treated and untreated trees. There were no signifi cant differences in the incidence of damaged leaves on trees treated with the two doses of the pesticide.
The difference in the damage to the crowns of treated and untreated trees was also clearly visible in the fi eld (Figs 4 and 5). From the middle of the season, leaves on untreated trees were all dry and brown, while those of treated trees were still green; except for some branches that were slightly infested. There was a second fl owering and dieback of branches caused by defoliation of the untreated trees.
In the year following the application of TreeAzin ® , the trees at Strážske were unexpectedly treated by the local community with a pesticide, so the results for that site in 2018 were not included in the analysis. At Parchovany, Fig. 3. Photographs of the galleries in the leaves produced by the fi rst instar larvae of the fi rst generation of C. ohridella. They consist initially of a 2 mm long tunnel (left) and then a 1.5-2 mm patch (right) a bit later. In addition, the overall health of the crown (decolourisation and defoliation) was ranked as a percentage of crown damaged at each of the cardinal points for all trees (four data points per tree). leaves from treated trees were signifi cantly less damaged than those from untreated trees in 2018. For both treatments the damage to the crowns of treated trees was signifi cantly less that of untreated trees. The average leaf damage recorded for trees treated with 3 ml/cm DBH and 5 ml/cm DBH was nearly the same and did not differ signifi cantly (Table 5).
No signifi cant differences were recorded in average leaf and crown damage recorded for the trees treated with 3 ml/ cm and 5 ml/cm DBH, which indicates that 3 ml/cm DBH is an acceptable minimum effective concentration of this pesticide for managing this pest. This is also the more costeffective dose. However, in some cases, better results were recorded for trees treated with 5 ml/cm DBH, although the differences were not statistically signifi cant (Tables 3 and  5).

DISCUSSION
Microinjection has been used several times in the past with relatively positive results in Central Europe. In 1997, the systemic insecticide imidacloprid was tested against C. ohridella. However, good results were obtained in preventing defoliation caused by 2 nd and 3 rd generations when applied in July (Feemers, 1997). Krehan (1997), which indicates a much earlier application in April, would prevent defoliation throughout the vegetative season. In 1999, abamectin was used in Hungary with good results (Bürgés & Szidonya, 2001) and later on microinjection was used   Kobza et al., 2011). The biological effi cacy of the systemic insecticide (in our case, TreeAzin ® ) was very good in the fi rst year of treatment. At both localities, Parchovany and Strážske, there were signifi cant differences in the average leaf damage caused by C. ohridella on treated compared to untreated trees. At the end of the fi rst growing season, 95.0% of the crowns of the control trees were damaged at Parchovany (81.2% at Strážske) and only 19.2-31.6% of the treated trees were defoliated. There was no difference in the average leaf and crown damage recorded for trees treated with 3 ml/cm DBH and 5 ml/cm DBH. In the year of application, the crowns of all the treated trees were statistically and visually in much better condition than those of untreated trees.
Despite a good effectiveness confi rmed by the results for some treated trees, sometimes individual branches were infested with C. ohridella. A possible reason for this nonuniformity in crown protection is an unequal distribution of TreeAzin ® to all branches, which may be associated with the age and overall health of the treated trees.
Like other authors (Pavela & Bárnet, 2005;Kobza et al., 2011), we confi rmed that trees older than 100 years can be effectively protected by this pesticide. This is important as many rare old trees growing in parks and gardens have high cultural and aesthetical value and are worth saving. In our experiments, injected trees were not fully protected even though they appeared to be healthy. As translocation often occurs faster in young, healthy trees (Bennett, 1957;Cox et al., 1997Cox et al., , 1998, we assume TreeAzin would more evenly protect the crowns of young trees. Our results confi rm the good effi ciency of neem-based systemic insecticide when applied in April. This confi rms the need for an early application proposed previously by other authors (Feemers, 1997, excepted), in order to target insects before the swarming period of C. ohridella (Krehan, 1997;Pavela & Bárnet, 2005;Ferracini & Alma, 2008;Kobza et al., 2011).
Common criticisms of injecting systemic insecticides include concerns over the potential for: (a) lack of uniform uptake, (b) slow application and (c) wounding of the injected areas (Krehan, 1997). In this trial there was evidence of unequal uptake in some of canopies of treated trees. Despite the overall effi cacy, some branches of treated trees were infested with C. ohridella. In our study, there were no concerns about the speed of application, which was viewed as adequate. It was noted that although the speed was dependent on the time of day (e.g. quicker uptake in the morning), weather, or overall health of the tree (as previously demonstrated by Ferracini & Alma, 2008) it was not a slow or onerous process. During a large-scale experiment in 2006 and 2007 involving abamectin (VIVID®II) and horse chestnut no phytotoxicity was reported (Pavela & Bárnet, 2005;Juhásová et al., 2007). Scar tissue formed around the shallow holes, which were enclosed in the next growing season. Not all authors experience only positive results (Krehan, 1997;Ferracini & Alma, 2008) and some of them highlight that the infl uence of the injection (tree's reaction to drilling) on its health should be investigated more deeply (Krehan, 1997). We did not test this aspect of the application method used (injection) as it was not an aim of the study. As far as we can tell by watching the response of trees to drilling, most of the wounds had healed by the end of the fi rst year, or the beginning of the second. However, roughly 10-20% of the injection sites did not produce suffi cient callus around the wound, where the tissues appeared to be dead. We believe that healing may be closely related to the sharpness of the drill bit, as dull bits may cauterize holes, preventing healing, but this topic was not investigated here. These factors should be considered, and the injection of trees and its infl uence on their health should be evaluated more precisely in the future.

CONCLUSIONS
Our results indicate that the formulation of TreeAzin ® tested protects trees for at least two-years. It is likely that the damage caused by microinjection is outweighed by the reduction in harm caused to these trees by C. ohridella; and using a biologically based insecticide can provide more benefi t to the trees than harm. Indeed, injecting trees reduces the need for spraying the foliage, thereby preventing run-off and spray drift, which may have severe consequences for non-target organisms. On trees treated the previous year the average leaf damage was higher, but still statistically signifi cantly lower than on untreated trees. This is the fi rst report of a two-year effi cacy.Trees were only evaluated in the year of injection in other studies (Pavela & Bárnet, 2005;Juhásová et al., 2007;Ferracini & Alma, 2008). In addition, what was encouraging was that this effi cacy against C. ohridella occurred in very large trees (> 97 cm DBH), which indicates that ancient trees can be protected from attack by this pest.
We demonstrated that systemic insecticide TreeAzin ® can be used to protect horse chestnut trees against C. ohridella. Microinjection has several advantages over traditional chemical methods. For example, small volumes of insecticide are administered, one treatment may protect trees for two years, application is almost independent of weather conditions, it is easy to do, environmentally friendly and precisely targeted (Juhásová et al., 2007;Kobza et al., 2011). Decreasing the damage caused by C. ohridella using imidacloprid results in a fast and long-lasting positive effect on the trees' condition in terms of growth (Jagiełło at al., 2019). Along with these advantages, there are also some open questions, especially the side effect of drilling on the health of the trees. The wound response following trunk injection of green ash (Fraxinus pennsylvanica Marsh.) has been studied over a period of two years by sectioning tree trunks and collecting data on annual radial growth and rate of healing around injection sites. This revealed that wound closure was positively correlated with tree health measured in terms of annual radial growth (Doccola et al., 2011). This fi nding supports earlier research indicating minimal damage and effective compartmentalization by trees when wounded by microinjection, particularly when compared with the wounding caused by increment borers (Shigo et al., 1977). Thus, it will be prudent to investigate how European horse-chestnut (Aesculus hippocastanum) responds to the wounding associated with trunk injections of insecticide.
Azadirachtin is an important natural pesticide and an alternative to conventional insecticides. It has been successfully used against many insect pests. However, as with any broad-spectrum insecticide, it is not without risk to non-target insects (Oulhaci et al., 2018). For example, azadirachtin is slightly to moderately toxic for honeybees although it did not appear to limit their foraging behaviour and is much less toxic than Imidacloprid, which is also often injected into trees (Challa et al., 2019). There is nearly no negative (lethal) effects of azadirachtin on stingless species of bees (Tomé, et al., 2015). Azadirachtin may induce a signifi cant antifeeding effect or a range of sublethal effects on some stingless species of bees, such as, Bombus terrestris or other useful insects. Gontijo et al., 2015;Bernardes et al., 2017); however, as the risks are minor, azadirachtin is still recommended for use in IPM (Challa et al., 2019). We recommend injecting azadirachtin after fl owering in order to limit exposure to spring pollinators. TreeAzin injections pose very little risk to non-target soil-dwelling insects, as there are no residues in the leaves at abscission (Grimault et al., 2011). Soil microbial communities are also not affected by azadirachtin tree injections (Kizilkaya et al., 2012;Suciu et al., 2019). As the active ingredient targets hormones specifi c to insect moulting, other animals are likely to experience little to no direct negative effects due to azadirachtin. The present fi ndings indicate that TreeAzin is a relatively safe pesticide with very low environmental risk and toxicity.

ACKNOWLEDGEMENTS.
We are grateful to anonymous reviewers for their constructive comments on earlier drafts of this manuscript. We also thank D. Fournier (Canada) for linguistic and editorial improvements. This work was funded by the Slovak Research and Development Agency under contract No. APVV-16-0031, APVV-19-0116, APVV-19-0119 and the Ministry of Agriculture and Rural Development of the Slovak Republic under contract No. 08V0301 Research and Development for Innovation and Support of the Competitiveness of Forestry. In addition, this work was also funded by the Ministry of Defence of the Slovak Republic.