Sensor based determination of the fruit bearing capacity in Malus x domestica (Borkh.) aimed at precise crop load management

One major goal of commercial fruit production is a homogenous fruit quality throughout the whole orchard. For this purpose, crop load management practices in apples intend to reduce the number of fruit per tree in order to (i) optimise the carbon supply to demand balance and the (ii) fruit quality in the current season, and (iii) ensure flower bud formation for the following season. Currently, crop load management is routinely performed field uniform without taking into consideration the inter-tree variability in flower or fruit set and the photosynthetic capacity of the trees, which frequently occurs in commercial orchards. Because of the variability among the trees, uniform crop load management can lead to sub-optimal numbers of fruit per tree when targeting homogenous fruit qualities. Currently, no tree-adapted crop load management practices have been developed. One reason is the lack of plant physiological and agronomic models to evaluate the actual fruit set of individual trees and to derive management decisions from these data. For tree-adapted crop load management, the fruit set and growth capacity of all trees within an orchard, often exceeding 2,500 trees per hectare, need to be mapped. Frequent studies of georeferencing and sensing individual trees’ data are available, but the developed approaches lack further application in decision support models. The aim of this thesis was (i) to investigate inter-tree variability in flower set, fruit set, and total leaf area per tree in apple, (ii) to develop a modelling approach to estimate the trees’ capacity to produce fruit of desired diameters, the fruit bearing capacity, which is based on spatially recorded sensor data of individual trees, and finally (iii) to investigate the application of the model, future possibilities, and advantages of tree-adapted crop load management. The variance in flowers per tree was investigated in two commercially relevant apple cultivars ('Elstar': 200 trees in 2011, 2014, 2015; 'Gala': 100 trees in 2014, 200 trees in 2015, 2016). Trees of the cultivar 'Elstar' were more susceptible to alternate bearing in comparison to 'Gala' trees and, therefore, included a higher percentage of trees with a low flower set, unable to meet the desired number of fruit per tree at harvest. Field uniform flower thinning led to yield reductions by over-thinning of trees with low and medium flower set on both cultivars ranging from 1.4 - 7.6 t ha-1. The leaf area per tree was recorded in three commercial apple orchards, taking into consideration the cultivars 'Gala' (996 trees), 'Pinova' (50 trees) and 'RoHo 3615' (100 trees), with a terrestrial 2-D light detection and ranging (LiDAR) laser scanner. A method to estimate individual trees’ photosynthetic and fruit bearing capacity (FBC) has been introduced. The method utilises the total leaf area per tree, monthly recorded gas exchange variables of the fruit and the leaves, fruit growth rates, and weather data and considers these data in a carbon balance model. The leaf area and photosynthetic capacity of the investigated trees was highly variable in all three orchards. In 'Gala', the variance in leaf area per tree and the FBC was similar in two consecutive years. However, the spatial location of cold and hot spots in the FBC, highlighting trees with FBC below or above the average of the surrounding trees, varied between the years. Consequently, for precise crop load management that targets homogenous average fruit diameters in orchards with inter-tree variability in total leaf area, the FBC need to be recorded annually. The FBC can be used to derive optimum fruit numbers per tree to achieve desired fruit diameters. The FBC of individual trees ('Gala' 2018: 100 trees, 2019: 70 trees; 'Pinova' 2018: 35 trees; 'RoHo 3615' 2018: 45 trees) was calculated for the actual average harvested fruit diameter. The modelled FBC of the trees, with little deviation, corresponded to the actual number of marketable fruit per tree. The results additionally revealed that, in the two orchards of the cultivars 'Pinova' and 'RoHo 3615', field-uniform thinning of heterogeneous trees can result in avoidable yield losses on 23%, 31 %, respectively, of the investigated trees. Furthermore, 16 % of the trees of 'RoHo 3615' had numbers of fruit per tree that exceeded the FBC and led to an average fruit diameter below 65 mm, the minimum requirement for fresh market access. In 'Gala' in two consecutive years, the mean fruit mass and mean soluble solids content per tree correlated with the amount of photons that the tree, after the foliage was fully developed, had absorbed per fruit, demonstrating the physiological limitations of the trees to produce fruit of a desired quality. Minimum thresholds of seasonally intercepted photons per fruit to achieve a specific fruit quality were generated. In the period after the foliage of the trees was fully developed until harvest, the total incident photosynthetically active radiation (PAR) was 440 MJ m-2, 508 MJ-2 in 2018, 2019, respective, of which the trees absorbed 19-62 %, depending on the available leaf area and the associated light interception. On average, 7.5 MJ PAR in 2018 and 5.9 MJ in 2019 per fruit was required, that 80% of the fruit reached a marketable fruit diameter. To absorb this amount of PAR, at least 550 cm² of leaf area per fruit were required. For the implementation of tree adapted crop load management, both flower and fruit thinning would be feasible. Flower thinning has two advantages: it is independent of the weather conditions, and the same settings of the thinning device leads to consistent thinning results over several years. However, after flower thinning is performed, late frost can reduce the fruit set of the trees, which may lead to low crop loads below the FBC. Chemical fruit thinning is usually carried out when there is no longer a risk of late frost. However, the thinning efficacy depends on temperature, irradiation, humidity and the physiological condition of the trees. Therefore, thinning results often differ between years. In this work it was confirmed that for successful chemical thinning with the photosynthesis inhibitor metamitron, warm weather conditions (night: 15°C, day: 20°C) in the days before and after application are favourable. At these temperatures, one application of the active ingredient is sufficient to reduce the fruit set of the trees to the desired target values, whereas at lower temperatures further applications may be required. Tree adapted crop load management practices would be beneficial to avoid over-thinning and yield losses on trees with low and medium flower set, to optimise the crop load of all the trees and, consequently, to tap into the full economic potential of an orchard. For future crop load management strategies, the developed model can be utilised to evaluate the actual fruit and flower set in order to make management decisions for individual trees. For this purpose, it is required to utilise tree individual data in plant physiological and agronomic models. The models can potentially serve as algorithms to control machinery that is able to treat trees individually. Tree adapted crop load management would significantly advance the management of apple orchards taking into consideration the desired production targets and the physiological limitations of individual trees.

Author dc:creator

• Penzel, Martin

Advisor dc:contributor.advisor

• Zude-Sasse, Manuela

Licence dc:rights.uri

• https://creativecommons.org/licenses/by-nc-nd/4.0/

Language dc:language.iso

en

Identifier URI

http://dx.doi.org/10.14279/depositonce-15247

OAI identifier oai:identifier

oai:depositonce.tu-berlin.de:11303/16471