Next Article in Journal
Tomato Fruit Growth and Nutrient Accumulation in Response to Blue and Red Light Treatments during the Reproductive Growth Stage
Next Article in Special Issue
Indoor Plant Soil-Plant Analysis Development (SPAD) Prediction Based on Multispectral Indices and Soil Electroconductivity: A Deep Learning Approach
Previous Article in Journal
Improved Recovery of Antioxidant Compounds from Refined Pumpkin Peel Extract: A Mixture Design Method Approach
Previous Article in Special Issue
Development and Testing of an IoT Spectroscopic Nutrient Monitoring System for Use in Micro Indoor Smart Hydroponics
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Horticulturae
Volume 9
Issue 10
10.3390/horticulturae9101112
horticulturae-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott)

by
Fatih Demirel
1,*,
Remzi Uğur
2,
Gheorghe Cristian Popescu
3,
Serap Demirel
4 and
Monica Popescu
5
1
Department of Agricultural Biotechnology, Faculty of Agriculture, Igdır University, Igdir 76000, Türkiye
2
Department of Park and Garden Plants, Nurdagi Vocational School, Gaziantep University, Gaziantep 27000, Türkiye
3
Department of Applied Sciences and Environmental Engineering, National University of Science and Technology POLITEHNICA Bucharest, University Center of Pitesti, 110040 Pitești, Romania
4
Department of Molecular Biology and Genetics, Faculty of Science, Van Yüzüncü Yıl University, Van 65080, Türkiye
5
Department of Natural Sciences, National University of Science and Technology POLITEHNICA Bucharest, University Center of Pitesti, 110040 Pitești, Romania
*
Author to whom correspondence should be addressed.
Horticulturae 2023, 9(10), 1112; https://doi.org/10.3390/horticulturae9101112
Submission received: 15 September 2023 / Revised: 4 October 2023 / Accepted: 7 October 2023 / Published: 9 October 2023
(This article belongs to the Special Issue Smart Horticulture: Latest Advances and Prospects)
Browse Figures
Review Reports Versions Notes

Abstract

:
The primary objective of this research was to ascertain the optimal circumstances for the successful growth of black chokeberry (Aronia melanocarpa (Michx.) Elliott) using tissue culture techniques. Additionally, the study aimed to explore the potential use of machine learning algorithms in this context. The present research investigated a range of in vitro parameters such as total number of roots (TNR), longest root length (LRL), average root length (ARL), number of main roots (NMR), number of siblings (NS), shoot length (SL), shoot diameter (SD), leaf width (LW), and leaf length (LL) for Aronia explants cultivated in different media (Murashige and Skoog (MS) and woody plant medium (WPM)) with different concentrations (0, 0.5, 1, 1.5, and 2 mg L−1) of indole-3-butyric acid (IBA). The study showed that IBA hormone levels may affect WPM properties, affecting the LRL and ARL variables. Aronia explant media treated with 2 mg L−1 IBA had the greatest TNR, NMR, NS, SL, and SD values; 31.67 pieces, 2.37 pieces, 5.25 pieces, 66.60 mm, and 2.59 mm, in that order. However, Aronia explants treated with 1 mg L−1 IBA had the highest LW (9.10 mm) and LL (14.58 mm) values. Finally, Aronia explants containing 0.5 mg L−1 IBA had the greatest LRL (89.10 mm) and ARL (57.57 mm) values. In general, the results observed (TNR, LRL, ARL, NMR, NS, SL, SD, LW, and LL) indicate that Aronia explants exhibit superior growth and development in WPM (25.68 pieces, 68.10 mm, 51.64 mm, 2.17 pieces, 4.33 pieces, 57.95 mm, 2.49 mm, 8.08 mm, and 14.26 mm, respectively) as opposed to MS medium (20.27 pieces, 59.92 mm, 47.25 mm, 1.83 pieces, 3.57 pieces, 49.34 mm, 2.13 mm, 6.99 mm, and 12.21 mm, respectively). In the context of the in vitro culturing of Aronia explants utilizing MS medium and WPM, an analysis of machine learning models revealed that the XGBoost and SVM models perform better than the RF, KNN, and GP models when it comes to making predictions about those variables. In particular, the XGBoost model stood out due to the fact that it had the greatest R-squared value, and showed higher predictive ability in terms of properly forecasting values in comparison to actual outcomes. The findings of a linear regression (LR) analysis were used in order to conduct an efficacy study of the XGBoost model. The LR results especially confirmed the findings for the SD, NS, and NMR variables, whose R-squared values were more than 0.7. This demonstrates the extraordinary accuracy that XGboost has in predicting these particular variables. As a consequence of this, it is anticipated that it will be beneficial to make use of the XGboost model in the dosage optimization and estimation of in vitro parameters in micropropagation studies of the Aronia plant for further scientific investigation.

1. Introduction

The black chokeberry, a member of the Rosaceae family, is a plant native to eastern North America and eastern Canada, and is known for its very modest cultivation needs [1,2]. During the early 20th century, the chokeberry plant was introduced to Russian botanical gardens, afterwards establishing itself in the European region of the nation [3]. In recent times, there has been an increase in cultivation of this crop in eastern European nations and Germany [4,5,6]. Subsequently, the cultivation of the plant expanded to include central and eastern European regions, where it now enjoys extensive cultivation [7]. The cultivation of Aronia, a plant that has been introduced in Türkiye and is regarded as having remarkable properties, is increasingly expanding throughout extensive regions inside the country [2].
Three varieties of chokeberries, namely the black-fruited (Aronia melanocarpa), red-fruited (A. arbutifolia), and purple-fruited (A. prunifolia) varieties, may be found growing in their natural habitat [1]. Aronia is a perennial shrub species characterized by its erect growth habit, reaching heights ranging from 80 to 300 cm. The fruits of this plant are small, with a diameter of roughly 6 mm, and exhibit a dark purple-black coloration. The foliage of the plant has a length ranging from 3 to 7 cm, displaying a glossy surface devoid of hair. The blooming period of its white-pink flowers occurs in the season of spring [8,9,10].
The rapid recognition of the health advantages associated with chokeberry led to the prompt establishment of plantations in the respective areas [1]. Aronia has a much higher concentration of anthocyanins, and possesses a greater antioxidant potential in comparison to many other types of berries [11]. The Aronia fruits include a substantial concentration of vitamins, minerals, and folic acid [12]. Aronia berries are often ingested in many forms, including fresh consumption, dried, in juices and processing into jams, in extracts, or through culinary colorants [13]. The fruits possess a high concentration of anthocyanin and other phenolic compounds [14,15,16]. The presence of polyphenols in Aronia extracts is responsible for their notable antioxidant action [17]. The impact of berry ripeness on the content and characteristics of phenolic compounds, which play a significant role in determining the antioxidative properties of berries, has been documented [18,19]. Therefore, it is in everyone’s best interest to grow and enhance such a helpful plant whenever possible.
In vitro studies on plants are vital for agriculture, plant science, and biotechnology, enhancing cultivation, biodiversity conservation, and sustainability [20,21,22,23,24]. Developing species-specific tissue culture methods is essential due to variations in plant species [24,25,26,27,28]. Although Aronia may be readily reproduced by seeds, this approach is not advised since the resulting plants yield fruit much later than expected and include heterozygotes [29]. The use of in vitro techniques enables the acceleration of plant multiplication, resulting in the rapid production of disease-free plants or removing heterogeneity [24]. Meanwhile, the micropropagation of adventitious shoots generated through organogenesis from somatic tissue explants has been proposed as a viable alternative in the case of Aronia melanocarpa [30]; the multiplication of shoots derived from axillary buds is widely regarded as the most practical and dependable approach for in vitro propagation [31,32]. As a result, it is essential to do research into the most suitable culture media in order to successfully cultivate the plant in vitro [32]. However, analyzing the information that was received as a consequence of the in vitro process with the right statistical methodologies is equally valuable for both optimizing and standardizing the process [33]. The process of optimizing in vitro cultures as a nonlinear, multivariable, and complicated system is distinguished by the fact that it is laborious, that it incurs considerable expenditures, and that it requires a significant investment of time [34]. As a result of this, there is a substantial requirement for the use of innovative computational approaches, such as machine learning algorithms, in order to properly assess and improve the effectiveness of this specific system by reducing the total number of treatments that are utilized [35].
Machine learning (ML) is widely recognized as an implementation of artificial intelligence (AI), a discipline within computer science that endows computers with the capacity to acquire knowledge by means of training datasets [36]. Machine learning (ML), which leads to the development of strong mathematical models, is produced using a dataset that consists of several independent variables or components (referred to as inputs) and dependent variables or responses (referred to as outputs) [37]. Since machine learning algorithms are able to forecast and characterize complicated processes that include numerous factors, they have the potential to become a useful and predictive decision-making tool for the in vitro micropropagation processes that occur in plants [38,39,40,41]. In recent years, it has seen widespread usage in a variety of disciplines relevant to agriculture, and freshly created models are now undergoing adaptation for use in agricultural research such as plant yield estimation [42], animal body weight estimate [43], boron estimation in soil [44], identify the diseases in plants [45], and weather prediction [46].
The objective of this study is to investigate the effects of varying concentrations of IBA (indole-3-butyric acid) hormones in different media (Murashige and Skoog (MS) and woody plant medium (WPM)) on Aronia axillary bud explants. The study aims to achieve the following objectives: (i) develop an efficient protocol for mass production using in vitro application; (ii) assess the predictive performance of various machine learning algorithms; and (iii) identify the most effective predictive feature among the output variables.

2. Materials and Methods

2.1. Plant Material, Preparation of Explant, and Explant Sterilization

The present investigation was carried out in the Tissue Culture Laboratories of the East Mediterranean Transitional Zone Agricultural Research Institute in 2023, with the aim of developing a micropropagation protocol for black chokeberry (Aronia melanocarpa var. Nero). The research used axillary buds from Aronia plants that were 3–4 years old. These plants were taken from the garden of the East Mediterranean Transitional Zone Agricultural Research Institute, from a thoroughly controlled greenhouse environment. In the laboratory, an attempt was made to eliminate physically concentrated dirt and germs on the tissue by subjecting it to agitation with liquid soap in a container filled with water for a duration of 30 min. Following this, the explants that were transferred to the sterile cabinet had a 30-s treatment with 70% ethanol, followed by three washes with sterile distilled water. This procedure was carried out in order to mitigate any potential harmful effects caused by the ethanol. In the subsequent stage, the explants underwent a triple washing with sterile distilled water after a 20-min incubation in a 30% commercial hypochlorite solution (2.5% sodium hypochlorite) supplemented with a small amount of Tween-80 (Tween 80, Sigma-Aldrich, St. Louis, MO, USA). The irregular and superfluous components that had caused harm to the tissues were excised from the sterilized explants under aseptic circumstances, using a sterile cabinet. Subsequently, the top sections of the explants, which contained a single bud, were excised using a scalpel with a number 3 blade. The excised explants, measuring about 1.5–2 cm in length, were then inserted into the growth medium.

2.2. Culture Conditions

The shoot induction medium was created using the established procedure outlined in the MS medium [47] and woody plant medium (WPM) [48]. The medium consisted of salts, vitamins, sucrose (3% w/v), and agar (0.8% w/v). Following the introduction of particular plant growth regulators (PGRs) into the growth media, the pH was then modified to a range of 5.6–5.8. The nutritional media was subjected to boiling, and thereafter transferred to culture bottles with clear covers, with a volume of 375 mL. This was carried out to ensure the even distribution of the agar, with a quantity of 65 mL of medium being used for this purpose. Subsequently, the lids were sealed, and the culture bottles underwent sterilization in an autoclave operating at a temperature of 121 °C and a pressure of 1 atmosphere for a duration of 15 min. The sterilized explants were then transferred to nutritional medium and placed in culture bottles. All of the in vitro cultures were then incubated in a plant culture chamber under controlled conditions, including a 16/8-h light/dark cycle, a light intensity of 40 μmol m−2 s−1, and a temperature of 25 ± 2 °C.
After the completion of surface sterilization, the sterilized explants were then moved to MS (Murashige and Skoog) medium and WPM (woody plant medium) supplemented with IBA (indole-3-butyric acid), in order to facilitate the formation of numerous shoots. The explants were cultivated in a medium that was treated with various concentrations of indole-3-butyric acid (0, 0.5, 1, 1.5, and 2 mg L−1). The in vitro cultures were maintained in the medium for a duration of 60 days (Figure 1). Subsequently, the parameters, denoted as total number of roots (TNR), longest root length (LRL) (mm), average root length (ARL) (mm), number of main roots (NMR), number of siblings (NS), shoot length (SL) (mm), shoot diameter (SD) (mm), leaf width (LW) (mm), and leaf length (LL) (mm), were obtained by measurements conducted on the explants, therefore generating the corresponding dataset.

2.3. Analysis of Variance

A completely randomized design was used in this experiment, with three replications. Each repeat consisted of five jars containing five explants for each treatment combination. The dataset was generated in Microsoft Excel by calculating the mean of twenty-five individual samples for each repeat. Subsequently, this dataset was used for the analysis of variance and machine learning research (Supplementary Table S1). The data acquired from the research was submitted for variance analysis using a suitable statistical software. The data were analyzed using XLSTAT software (Addinsoft, version 2023.1.3), and all analytical results were calculated as the average of three replications. The significant differences were assessed using Duncan’s test conducted at a significance level of 5%.

2.4. Modeling Using Machine Learning (ML) Algorithms

The aim was to estimate the output (observed in vitro parameters) variable using these input (in vitro media and IBA doses) variables for modeling. The study included five machine learning (ML) algorithms, including support vector machines (SVM) [49], random forest (RF) [50], extreme gradient boosting (XGBoost) [51], k-nearest neighbors classifier (KNN) [52], and Gaussian processes classifier (GP) [53]. The input dataset used in this study consisted of two distinct experimental design media, namely MS and WPM, together with varying concentrations of IBA (0, 0.5, 1, 1.5, and 2 mg L−1).
The performance evaluation of the algorithms was conducted based on four primary metrics, namely the mean squared error (MSE), R-squared (R2), mean absolute percentage error (MAPE), and mean absolute deviation (MAD). The coefficient of determination, denoted as R2, quantifies the extent to which the model (Equation (1)) explains the observed data. The mean squared error (MSE) is a measure of the proximity between the predicted values and the actual values (Equation (2)). The mean absolute percentage error (MAPE) is the mean absolute percentage error between expected and actual values (Equation (3)). Additionally, the mean absolute deviation (MAD) characterizes the overall distribution of prediction errors (Equation (4)) [54].
R 2 = 1 − ∑ i = 1 n y i − y i p 2 ∑ i = 1 n y i − y ¯ 2
M S E = 1 n ∑ i = 1 n y i − y i p 2
M A P E = 1 n ∑ i = 1 n y i − y i p y i × 100
M A D = 1 n ∑ i = 1 n y i − y i p
where n is the training sample size in the dataset, y i is the measured real value, y i p is the predicted value, and y ¯ is the measured values mean. The R program was used for the computation of machine learning algorithms and performance metrics [42,55,56].

3. Results

3.1. Analysis of Variance for In Vitro Features

In the context of this research, an analysis of variance (ANOVA) was conducted to assess the total number of roots (TNR), longest root length (LRL), average root length (ARL), number of main roots (NMR), number of siblings (NS), shoot length (SL), shoot diameter (SD), leaf width (LW), and leaf length (LL) parameters derived from the cultivated explants of the Aronia plant (Table 1). The results of the analysis of variance indicate that there was a significant difference in all of the observed parameters between the two media (MS and WPM). Statistically significant changes were seen for all parameters investigated in relation to the impact of the administered dosages of IBA. The analysis of variance demonstrated that the doses of IBA (0, 0.5, 1, 1.5, and 2 mg L−1) were significant for the whole set of investigated parameters. When the interaction between medium and dosage was analyzed, a statistically significant difference was seen for the LRL (p ≤ 0.05), ARL (p ≤ 0.001), and SD (p ≤ 0.05) parameters.
The growth parameters, such as TNR, NMR, NS, SL, and SD, were found to be highest in the medium that had 2 mg L−1 IBA, with values of 31.67, 2.37, 5.25, 66.60, and 2.59, in that order. The highest LW and LL mean values, 9.10 and 14.58, were found in the media containing 1 mg L−1 IBA. The LRL and ARL values that were highest were determined in the media that contained 0.5 mg L−1 of IBA. These values were 89.10 and 57.57, respectively. It was established that Aronia explants grown in WPM exhibited higher success when compared to Aronia explants grown in MS medium according to the average of all of the IBA doses used in the MS medium and the average of all of the IBA doses used in WPM for all of the in vitro parameters that were investigated.
The WPM supplemented with 0.5 mg L−1 IBA exhibited the highest recorded LRL value, measuring 97.93. Conversely, the WPM supplemented with 0 mg L−1 IBA displayed the lowest recorded value, measuring 40.53. The WPM containing 0.5 mg L−1 IBA exhibited the highest ARL value, measuring 63.28, while the WPM containing 0 mg L−1 IBA had the lowest value, measuring 34.78. The WPM with 2 mg L−1 IBA had the maximum SD value, which was 2.85, while the WPM with 0 mg L−1 IBA had the lowest value, which was 1.93.

3.2. Machine Learning (ML) Analysis

Machine learning algorithms are a kind of algorithm that employ statistical and computational methods to detect patterns in data, construct models from those patterns, and then use those models to make predictions or judgments. In this investigation, we utilized the support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBoost), k-nearest neighbors classifier (KNN), and Gaussian processes classifier (GP) algorithms to predict the relationship between inputs and outputs, compare and evaluate the performances of the models, and analyze the data generated from the experiments (tissue culture). The input variables for this investigation consisted of two distinct types of tissue culture media used in the experiments (MS and WPM), as well as the corresponding dosages of IBA (0, 0.5, 1, 1.5, and 2 mg L−1). The output variables included observations derived from in vitro cultivated explants of the Aronia plant. Table 2 provides a summary of the results of the study and displays the outputs of the machine learning models that were used in the research.
Metrics such as the mean squared error (MSE), mean absolute percentage error (MAPE), and mean absolute deviation (MAD) are used in the analysis of an algorithm’s overall performance. When the values of these metrics go down, it indicates that the model’s predictions are becoming closer and closer to the actual values that have been observed. In addition to this, it gauges how well the R-squared (R2) model can explain the variance that exists between the independent variables and the variable that is being investigated.
Based on an evaluation of the mean square error (MSE) and mean absolute percent error (MAPE) metrics, it was ascertained that the XGBoost model exhibited superior performance in predicting the total number of roots (TNR). Conversely, the support vector machine (SVM) model had the maximum performance based on the mean absolute deviation (MAD) metric findings. The R-squared (R2) values for the XGBoost and SVM models were 0.589 and 0.575, respectively (Table 2). Furthermore, Figure 2 presents the actual data collected from the in vitro cultivation of Aronia, with the corresponding values predicted by the models. Based on the results of the linear regression analysis comparing the observed values with the projected values of the models, it was found that the XGBoost model exhibited an even higher R-squared (0.574) value.
The superiority of the SVM model in predicting the longest root length (LRL) was determined by examination of the mean absolute percent error (MAPE) and mean absolute deviation (MAD) metrics. In contrast, the XGBoost model exhibited superior performance as determined by the mean square error (MSE) metric analysis. The R-squared (R2) values obtained for the XGBoost and SVM models were 0.281 and 0.138, correspondingly (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.255) when comparing the observed values with the anticipated values of the models (Figure 2).
The XGBoost model demonstrated higher predictive ability in estimating the average root length (ARL), as determined by the assessment of the mean square error (MSE) and mean absolute deviation (MAD) metrics. On the other hand, the Support Vector Machine (SVM) model exhibited the highest level of performance as determined by the mean absolute percent error (MAPE) metric results. The XGBoost and SVM models yielded R-squared (R2) values of 0.267 and 0.135, respectively (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.241) when comparing the observed values with the anticipated values of the models (Figure 2).
The XGBoost model demonstrated greater performance in estimating the number of main roots (NMR), as determined by evaluating the mean square error (MSE) and the mean absolute percent error (MAPE) metrics. In contrast, the support vector machine (SVM) model exhibited the highest level of performance as determined by the mean absolute deviation (MAD) metric results. The R-squared (R2) values obtained for the XGBoost and SVM models were 0.736 and 0.714, correspondingly (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.727) when comparing the observed values to the predicted values of the models (Figure 2).
The superiority of the SVM model in predicting the number of siblings (NS) was determined based on the assessment of the mean absolute deviation (MAD) and the mean absolute percent error (MAPE) metrics. In contrast, it was shown that the XGBoost model exhibited the highest level of performance as determined by the measure of mean square error (MSE). The R-squared (R2) values obtained for the XGBoost and SVM models were 0.795 and 0.780, correspondingly (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.787) when comparing the observed values with the predicted values of the models (Figure 2).
Based on the assessment of the mean absolute deviation (MAD) and the mean absolute percent error (MAPE) metrics, it was determined that the support vector machine (SVM) model had higher efficacy in forecasting the shoot length (SL). On the other hand, it was observed that the XGBoost model exhibited the highest level of performance when evaluated using the mean square error (MSE) metric. The R-squared (R2) values obtained for the XGBoost and SVM models were 0.571 and 0.534, correspondingly (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.555) when comparing the observed values with the anticipated values of the models (Figure 2).
The XGBoost model demonstrated higher predictive ability in estimating the shoot diameter (SD), as determined by the examination of metrics such as the mean square error (MSE), mean absolute deviation (MAD), and mean absolute percent error (MAPE). The XGBoost model achieved an R-squared (R2) score of 0.743 (Table 2). Additionally, Figure 2 displays the empirical data obtained from the in vitro culture of Aronia, together with the matching values projected by the XGBoost model. The linear regression analysis revealed that the XGBoost model had a favorable R-squared value (0.555) when comparing the observed values to the anticipated values.
The superiority of the SVM model in predicting the leaf width (LW) was determined by the examination of the mean absolute deviation (MAD) and the mean absolute percent error (MAPE) metrics. In contrast, the XGBoost model exhibited the highest level of performance as determined by the mean square error (MSE) metric results. The R-squared (R2) values obtained for the XGBoost and SVM models were 0.583 and 0.535, correspondingly (Table 2). The linear regression analysis revealed that the XGBoost model had a better R-squared value (0.568) when comparing the observed values with the anticipated values of the models (Figure 2).
The XGBoost model demonstrated higher predictive accuracy in estimating the leaf length (LL), as determined by the examination of metrics such as mean square error (MSE), mean absolute deviation (MAD), and mean absolute percent error (MAPE). The XGBoost model achieved an R-squared (R2) score of 0.380 (Table 2). Additionally, Figure 2 displays the empirical data obtained from the in vitro culture of Aronia, together with the corresponding values estimated by the XGBoost model. The linear regression analysis revealed that the XGBoost model had an R-squared value of 0.392 when comparing the observed values to the anticipated values.
Upon comprehensive examination of the ML analysis findings, it is evident that the XGBoost and SVM models exhibit improved prediction capabilities compared to the RF, KNN, and GP models for the observed variables in the context of in vitro cultivation of Aronia explants using MS medium and WPM. The XGBoost model, which has the greatest R-squared value among the models created by XGBoost and SVM, was the model with the most effective prediction power between actual values and predicted values. This model demonstrated notable success in accurately predicting values when compared to the actual values. The linear regression analysis revealed notable findings when comparing the projected values generated by the XGBoost model to the actual values of the observed variables SD, NS, and NMR. These variables exhibited R-squared values over 0.7, distinguishing them from the other variables. In other words, the XGBoost model was able to predict the SD, NS, and NMR variables with higher predictive values more accurately.

4. Discussion

4.1. Evaluation of the In Vitro Micropropagation Ability of Black Chokeberry

This study examined Aronia’s micropropagation potential in vitro using WPM and MS medium and IBA dosages. The research findings indicate that, across all observed variations, the basal WPM exhibited superior performance compared to the MS basal medium in the in vitro cultivation of Aronia. The research demonstrated that the effects of different levels of IBA hormone on the measured parameters inside the WPM may exhibit variations in both the LRL and ARL variables. According to mean IBA doses, Aronia explants media supplemented with 2 mg L−1 IBA demonstrated the highest values for the TNR, NMR, NS, SL, and SD variables. On the other hand, Aronia explants supplemented with 1 mg L−1 IBA displayed the highest values for the LW and LL variables. Lastly, Aronia explants supplemented with 0.5 mg L−1 IBA exhibited the highest values for the LRL and ARL variables. The impact of IBA showed notable efficacy on the rooting-related parameters of explants, thus yielding a substantial contribution to the elongation of shoots. The obtained outcome exhibited a resemblance to the conclusions drawn from several investigations in terms of the impact of IBA [57,58,59]. Indole-3-butyric acid (IBA) hormone has beneficial impacts on the development of both shoot and root structures [60]. The cultivars ‘Melrom’ and ‘Nero’ of Aronia melanocarpa (Michx.) Elliot were examined [31]. According to their findings, the researchers observed a notable disparity in the regenerative capacity between the chokeberry cultivars ‘Nero’ and ‘Melrom’. Specifically, the cultivar ‘Nero’ exhibited a significantly greater ability for regeneration. They reported that the highest number of shoots were obtained when using a basal medium containing MS macroelements, LF microelements, and LF vitamins, supplemented with a concentration of 4.5 mg dm−3 of BA and 0.6 mg dm−3 of IBA.
Various studies have conducted comparisons between basal media. According to the findings of Fallahpour et al. [61], the micropropagation of dwarf sweet cherry rootstock via in vitro culture in MS, DKW, and WPM media revealed that WPM and DKW medium exhibited superior performances as basal media. These media have shown a better efficacy in terms of promoting shoot proliferation and increasing the number of shoots, as compared to the MS medium. In addition, they found that explants grown in WPM, which was one of the media enriched with 2 mg L−1 IBA, demonstrated greater rooting ability in comparison to explants grown in MS and DKW media. The study conducted by Sisko [62] focused on the in vitro propagation of Gisela 5, a rootstock resulting from the crossbreeding of Prunus cerasus and P. canescens. According to the research, it was seen that WPM exhibited the greatest rate of proliferation, with an average of 4.2 shoots per explant. Conversely, the MS medium demonstrated the lowest rate of proliferation, with an average of 3.0 shoots per explant. Hui-Mei et al. [63] found that the WPM showed greater efficacy compared to the MS medium in facilitating shoot regeneration during an assessment of fundamental parameters influencing in vitro shoot regeneration from axillary bud explants of Camptotheca acuminata. In their research on the micropropagation of Meyer lemon (Citrus × meyeri), Haradzi et al. [64] examined the rooting percentage of explants cultivated in MS medium and WPM supplemented with IBA. According to their findings, the percentage of rooting in WPM supplemented with IBA was found to be greater compared to MS medium supplemented with IBA. According to El-Agamy et al. [65], vegetative growth of pomegranate cultivars during in vitro propagation was better in WPM than in MS medium.
While MS medium remains the predominant choice, it is often substituted by media with a lower salt concentration, particularly those with reduced nitrogen content. For instance, the ammonium nitrate level in MS medium is 1650 mg·L−1, while it is 1416 mg·L−1 in WPM. It has been shown that the majority of plants exhibit a preference for nitrate over ammonium as a source of nitrogen [66]. Previous studies have demonstrated that ammonium may have detrimental effects on explants [67,68]. This is attributed to the fact that excessive nitrate can be retained in vacuoles, whereas high levels of ammonium can be hazardous to plants. In contrast, it can be seen that WPM has a greater calcium content in comparison to MS medium [69]. Calcium plays a crucial role in cellular signaling, functioning as a secondary messenger in conjunction with signal transduction proteins [70]. Additionally, it serves to preserve the integrity of the plasmalemma by facilitating the interaction between diverse proteins and lipids on membrane surfaces [71]. The maintenance of plasmalemma integrity is associated with an increase in turgor pressure, resulting in increased water content and improved nutrient retention inside cells [72]. In addition, calcium (Ca) may have a direct influence on cellular and organ development. The role of this particular element encompasses several cellular processes including cell elongation and cell division [73]. Additionally, it has an impact on cellular pH, and serves as a regulatory ion in the translocation of carbohydrates in the source–sink mechanism, affecting both cells and cell walls [70]. The observed increase in growth of Aronia in WPM may be attributed to an improved nutritional status, as previously indicated.

4.2. Usage of Machine Learning Algorithms

The outcome of cultivated cells or tissues in the in vitro regeneration of plants is influenced by many variables. The factors considered include the plant genotype, plant growth regulators (PGRs), components of the culture media, type and age of the explant, as well as enhancer additives and elicitors, among others [74]. The process of optimizing in vitro culture as a nonlinear, multivariable, and complicated system is distinguished by the fact that it is laborious, that it incurs considerable expenditures, and that it requires a significant investment of time [33,34,75]. The capability of machine learning algorithms to evaluate and verify anticipated output variables is reflected in their intrinsic ability to include the input parameters that underlie the outputs [54]. As a result, there is a great requirement for the use of innovative computational methodologies, such as machine learning algorithms, in order to successfully assess and improve the effectiveness of this specific system by reducing the total number of treatments that are utilized [35,76]. In the context of this research, a total of nine distinct observations were conducted using axillary buds of Aronia. Models were developed for each of the nine observations, using five distinct machine learning techniques. The prediction capabilities of these models were afterwards compared. The models that exhibited superior predictive performance were generated using the XGBoost and SVM methods. It was also shown that out of the nine observations, the SD, NS, and NMR parameters exhibited the best level of predictive power. Consequently, it can be said that the values of SD, NS, and NMR parameters of Aronia explants cultivated in MS medium and WPM, with the addition of IBA, may be anticipated via the utilization of the XGBoost and SVM algorithms. These projected values have potential for use in the optimization of dosage. These algorithms have shown effective use in conjunction with in vitro investigations. The research conducted included a diverse range of models, hyperparameters, and performance indicators, successfully covering a wide range of plant tissue culture processes that specifically focused on in vitro germination [77,78], somatic embryogenesis [79], in vitro sterilization [78], in vitro mutagenesis [80], in vitro regeneration [34,54], and the optimization of basal media [81].
Drawing similarities to related studies, the effectiveness of XGBoost in predicting chickpea shoot counts serves as an illustration of its ability to exhibit higher modeling competency [82]. Similarly, the RF (random forest) and XGBoost (extreme gradient boosting) algorithms have been identified as suitable candidates for predicting shoot counts and shoot length in Alternanthera reineckii [83]. In their study, Eren et al. [54] found that the XGBoost model showed superior performance in terms of callus induction (CI%), regeneration efficiency (RE), and the number of plantlets (NP) in wheat. They reported that the R2 scores for these variables accounted for 38.3%, 73.8%, and 85.3% of the variances, respectively. The results of our research are comparable to those found in other studies.

5. Conclusions

The potential for in vitro micropropagation of Aronia was investigated in this study using two basal media, as well as employing varied concentrations of the hormone IBA. According to the findings, Aronia explants grew and developed more successfully in WPM than in MS medium. The WPM with 0.5 mg L−1 IBA had the highest LRL and ARL values. The WPM with 2 mg L−1 IBA exhibited the highest SD value. In order to examine the potential for in vitro micropropagation of Aronia, ML algorithms were used. In the context of in vitro culturing of Aronia explants using MS medium and WPM, it is obvious that the XGBoost and SVM models display better predictive capabilities compared to the RF, KNN, and GP models for the observed variables. This is evident from an in-depth review of the results of the ML analysis, which demonstrates that the XGBoost and SVM models exhibit superior prediction skills. The XGBoost model, which has the biggest R-squared value among the models developed by XGBoost and SVM, is the model with the most effective prediction power between real values and predicted values. It is expected that using the XGboost model would provide advantages in the optimization of dose and for the determination of in vitro parameters in Aronia plant micropropagation investigations, hence facilitating future scientific exploration.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/horticulturae9101112/s1, Table S1: Dataset used for analysis of variance and machine learning research.

Author Contributions

Conceptualization, R.U., F.D. and S.D.; methodology, R.U., F.D. and M.P.; software, F.D. and S.D.; validation, F.D., G.C.P. and S.D.; formal analysis, F.D., S.D. and R.U.; investigation, R.U. and S.D.; resources, R.U.; data curation, F.D., R.U., G.C.P., S.D. and M.P.; writing—original draft preparation, F.D. and S.D.; writing—review and editing, R.U., G.C.P., S.D. and M.P.; visualization, F.D., R.U. and G.C.P.; supervision, F.D.; project administration, F.D., R.U. and S.D.; funding acquisition, R.U., G.C.P., S.D. and M.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Data Availability Statement

All data supporting the conclusions of this research are included in this study.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Ochmian, I.D.; Grajkowski, J.; Smolik, M. Comparison of some morphological features, quality and chemical content of four cultivars of chokeberry fruits (Aronia melanocarpa). Not. Bot. Horti Agrobot. Cluj-Napoca 2012, 40, 253–260. [Google Scholar] [CrossRef]
  2. Yilmaz, A.; Güler, E.; Soydemir, H.E.; Demirel, S.; Mollahaliloğlu, S.; Çiftçi, V.; Karadeniz, T. Miracle Plant: Black Chokeberry (Aronia melanocarpa). MAS J. Appl. Sci. 2021, 6, 83–94. [Google Scholar] [CrossRef]
  3. Šnebergrová, J.; Čížková, H.; Neradova, E.; Kapci, B.; Rajchl, A.; Voldřich, M. Variability of characteristic components of aronia. Czech J. Food Sci. 2014, 32, 25–30. [Google Scholar] [CrossRef]
  4. Jeppsson, N.; Johansson, R. Changes in fruit quality in black chokeberry (Aronia melanocarpa) during maturation. J. Hortic. Sci. Biotechnol. 2000, 75, 340–345. [Google Scholar] [CrossRef]
  5. Esatbeyoglu, T.; Winterhalter, P. Preparation of dimeric procyanidins B1, B2, B5, and B7 from a polymeric procyanidin fraction of black chokeberry (Aronia melanocarpa). J. Agric. Food Chem. 2010, 58, 5147–5153. [Google Scholar] [CrossRef]
  6. Konić–Ristić, A.; Srdić–Rajić, T.; Kardum, N.Đ.; Glibetić, M. Biological activity of Aronia melanocarpa antioxidants pre–screening in an intervention study design. J. Serbian Chem. Soc. 2013, 78, 429–443. [Google Scholar] [CrossRef]
  7. Benvenuti, S.; Pellati, F.; Melegari, M.A.; Bertelli, D. Polyphenols, anthocyanins, ascorbic acid, and radical scavenging activity of Rubus, Ribes, and Aronia. J. Food Sci. 2004, 69, 164–169. [Google Scholar] [CrossRef]
  8. Hardin, J.W. The enigmatic chokeberries (Aronia, Rosaceae). Bull. Torrey Bot. Club 1973, 100, 178–184. [Google Scholar] [CrossRef]
  9. Ara, V. Fachthemen–Schwarzfruchtige Aronia: Gesund––und bald in aller Munde? Fluss. Obs. 2002, 69, 653–657. [Google Scholar]
  10. Scott, R.W.; Skirvin, R.M. Black chokeberry (Aronia melanocarpa Michx.): A semi–edible fruit with no pests. J. Am. Pomol. Soc. 2007, 61, 135. [Google Scholar]
  11. Gonzalez–Molina, E.; Moreno, D.A.; Garcia–Viguera, C. Aronia–enriched lemon juice: A new highly antioxidant beverage. J. Agric. Food Chem. 2008, 56, 11327–11333. [Google Scholar] [CrossRef] [PubMed]
  12. Stralsjö, L.; Åhlin, H.; Witthöft, C.M.; Jastrebova, J. Folate determination in Swedish berries by radioprotein–binding assay (RPBA) and high performance liquid chromatography (HPLC). Eur. Food Res. Technol. 2003, 216, 264–269. [Google Scholar] [CrossRef]
  13. Engin, S.P.; Mert, C. The effects of harvesting time on the physicochemical components of aronia berry. Turk. J. Agric. For. 2020, 44, 361–370. [Google Scholar] [CrossRef]
  14. Shahin, L.; Phaal, S.S.; Vaidya, B.N.; Brown, J.E.; Joshee, N. Aronia (Chokeberry): An underutilized, highly nutraceutical plant. J. Med. Act. Plants 2019, 8, 46–63. [Google Scholar]
  15. Demir, B.; Sayinci, B.; Sümbül, A.; Yaman, M.; Yildiz, E.; Çetin, N.; Karakaya, O.; Ercişli, S. Bioactive compounds and physical attributes of genotypes through multivariate approaches. Folia Hortic. 2020, 32, 189–202. [Google Scholar] [CrossRef]
  16. Demir, B.; Sayıncı, B.; Yaman, M.; Sümbül, A.; Yıldız, E.; Karakaya, O.; Bobuş Alkaya, G.; Ercişli, S. Biochemical composition and shape–dimensional traits of rosehip genotypes. Folia Hortic. 2021, 33, 293–308. [Google Scholar] [CrossRef]
  17. Kokotkiewicz, A.; Jaremicz, Z.; Luczkiewicz, M. Aronia plants: A review of traditional use, biological activities, and perspectives for modern medicine. J. Med. Food 2010, 13, 255–269. [Google Scholar] [CrossRef]
  18. Andrzejewska, J.; Sadowska, K.; Kloska, L.; Rogowski, L. The effect of plant age and harvest time on the content of chosen components and antioxidative potential of black chokeberry fruit. Acta Sci. Pol. Hortorum Cultus 2015, 14, 105–114. [Google Scholar]
  19. Yaman, M. Evaluation of genetic diversity by morphological, biochemical and molecular markers in sour cherry genotypes. Mol. Biol. Rep. 2022, 49, 5293–5301. [Google Scholar] [CrossRef]
  20. Thorpe, T.A. History of plant tissue culture. Mol. Biotechnol. 2007, 37, 169–180. [Google Scholar] [CrossRef]
  21. Haliloğlu, K.; Türkoğlu, A.; Balpınar, Ö.; Öztürk, H.; Özkan, G.; Poczai, P. Mammalian sex hormones effects on in vitro organogenesis of common bean (Phaseolus vulgaris L.). Sci. Rep. 2023, 13, 3337. [Google Scholar] [CrossRef] [PubMed]
  22. Pathak, M.R.; Abido, M.S. The role of biotechnology in the conservation of biodiversity. J. Exp. Biol. 2014, 2, 352–363. [Google Scholar]
  23. Uzun, S.; Yukselgungor, D. Micropropagation of some onobrychis species through in vitro shoot regeneration. Acta Sci. Pol. Hortorum Cultus 2020, 19, 45–52. [Google Scholar] [CrossRef]
  24. Uzun, S.; Ilbaş, A.İ.; Ipek, A.; Arslan, N.; Barpete, S. Efficient in vitro plant regeneration from immature embryos of endemic Iris sari and I. schachtii. Turk. J. Agric. For. 2014, 38, 348–353. [Google Scholar] [CrossRef]
  25. Sarasan, V.; Kite, G.C.; Sileshi, G.W.; Stevenson, P.C. Applications of phytochemical and in vitro techniques for reducing over–harvesting of medicinal and pesticidal plants and generating income for the rural poor. Plant Cell Rep. 2011, 30, 1163–1172. [Google Scholar] [CrossRef]
  26. Bulunuz Palaz, E.; Ugur, R.; Yaman, M. Micropropagation Protocols of New Prunus Hybrids with Significant Rootstock Potential in Fruit Breeding and Cultivation. Erwerbs-Obstbau 2023, 65, 1359–1364. [Google Scholar] [CrossRef]
  27. Twaij, B.M.; Jazar, Z.H.; Hasan, M.N. Trends in the use of tissue culture, applications and future aspects. Int. J. Plant Biol. 2020, 11, 8385. [Google Scholar] [CrossRef]
  28. Monthony, A.S.; Page, S.R.; Hesami, M.; Jones, A.M.P. The past, present and future of Cannabis sativa tissue culture. Plants 2021, 10, 185. [Google Scholar] [CrossRef]
  29. Litwinczuk, W. Propagation of black chokeberry (Aronia melanocarpa Elliot) through in vitro culture. Electron. J. Pol. Agric. Univ. Ser. Hortic. 2002, 5, 1–7. [Google Scholar]
  30. Rusea, I.; Popescu, A.; Isac, V.; Șuțan, A.; Hoza, D. High efficiency shoot multiplication from in vitro cultured meristems of Aronia melanocarpa cv. Nero. Sci. Pap. Ser. B Hortic. 2019, 63, 65–74. [Google Scholar]
  31. Şuţan, N.A.; Isac, V.; Duminică, C.; Popescu, A. Studies on the in vitro micropropagation ability of Aronia melanocarpa (Michx.) Elliot. Curr. Trends Nat. Sci. 2017, 6, 85–92. [Google Scholar]
  32. Kukharchik, N.V.; Kastritskaya, M.S.; Malinovskaya, A.M. Process guide of production of Improved in iv vitro planting material of chokeberry (Arónia melanocárpa). Fruit Grow. 2014, 26, 233–240. [Google Scholar]
  33. Aasim, M.; Katirci, R.; Baloch, F.S.; Mustafa, Z.; Bakhsh, A.; Nadeem, M.A.; Ali, S.A.; Hatipoğlu, R.; Çiftçi, V.; Habyarimana, E.; et al. Innovation in the breeding of common bean through a combined approach of in vitro regeneration and machine learning algorithms. Front. Genet. 2022, 13, 897696. [Google Scholar] [CrossRef] [PubMed]
  34. Hesami, M.; Jones, A.M.P. Application of artificial intelligence models and optimization algorithms in plant cell and tissue culture. Appl. Microbiol. Biotechnol. 2020, 104, 9449–9485. [Google Scholar] [CrossRef]
  35. Silva, J.C.F.; Teixeira, R.M.; Silva, F.F.; Brommonschenkel, S.H.; Fontes, E.P. Machine learning approaches and their current application in plant molecular biology: A systematic review. Plant Sci. 2019, 284, 37–47. [Google Scholar] [CrossRef]
  36. Butler, K.T.; Davies, D.W.; Cartwright, H.; Isayev, O.; Walsh, A. Machine learning for molecular and materials science. Nature 2018, 559, 547–555. [Google Scholar] [CrossRef]
  37. Olden, J.D.; Lawler, J.J.; Poff, N.L. Machine learning methods without tears: A primer for ecologists. Q. Rev. Biol. 2008, 83, 171–193. [Google Scholar] [CrossRef]
  38. Pepe, M.; Hesami, M.; Small, F.; Jones, A.M.P. Comparative analysis of machine learning and evolutionary optimization algorithms for precision micropropagation of Cannabis sativa: Prediction and validation of in vitro shoot growth and development based on the optimization of light and carbohydrate sources. Front. Plant Sci. 2021, 12, 757869. [Google Scholar]
  39. Pepe, M.; Hesami, M.; Jones, A.M.P. Machine learning–mediated development and optimization of disinfection protocol and scarification method for improved in vitro germination of cannabis seeds. Plants 2021, 10, 2397. [Google Scholar] [CrossRef]
  40. Lozano–Milo, E.; Landin, M.; Gallego, P.P.; García–Pérez, P. Machine Learning Deciphers Genotype and Ammonium as Key Factors for the Micropropagation of Bryophyllum sp. Medicinal Plants. Horticulturae 2022, 8, 987. [Google Scholar] [CrossRef]
  41. Sadat–Hosseini, M.; Arab, M.M.; Soltani, M.; Eftekhari, M.; Soleimani, A. Applicability of soft computing techniques for in vitro micropropagation media simulation and optimization: A comparative study on Salvia macrosiphon Boiss. Ind. Crops Prod. 2023, 199, 116750. [Google Scholar] [CrossRef]
  42. Demirel, F.; Eren, B.; Yilmaz, A.; Türkoğlu, A.; Haliloğlu, K.; Niedbała, G.; Bujak, H.; Jamshidi, B.; Pour–Aboughadareh, A.; Bocianowski, J.; et al. Prediction of Grain Yield in Wheat by CHAID and MARS Algorithms Analyses. Agronomy 2023, 13, 1438. [Google Scholar] [CrossRef]
  43. Faraz, A.; Tırınk, C.; Önder, H.; Şen, U.; Ishaq, H.M.; Tauqir, N.A.; Wheed, A.; Nabeel, M.S. Usage of the XGBoost and MARS algorithms for predicting body weight in Kajli sheep breed. Trop. Anim. Health Prod. 2023, 55, 276. [Google Scholar] [CrossRef] [PubMed]
  44. Gökmen, F.; Uygur, V.; Sukuşu, E. Extreme Gradient Boosting Regression Model for Soil Available Boron. Eura. Soil Sci. 2023, 56, 738–746. [Google Scholar] [CrossRef]
  45. Hassan, S.M.; Jasinski, M.; Leonowicz, Z.; Jasinska, E.; Maji, A.K. Plant disease identification using shallow convolutional neural network. Agronomy 2021, 11, 2388. [Google Scholar] [CrossRef]
  46. Dong, J.; Zeng, W.; Wu, L.; Huang, J.; Gaiser, T.; Srivastava, A.K. Enhancing short–term forecasting of daily precipitation using numerical weather prediction bias correcting with XGBoost in different regions of China. Eng. Appl. Artif. Intell. 2023, 117, 105579. [Google Scholar] [CrossRef]
  47. Murashige, T.; Skoog, F. A revised medium for rapid growth and bio assays with tobacco tissue cultures. Physiol. Plantar. 1962, 15, 473–497. [Google Scholar] [CrossRef]
  48. McCown, B.H.; Sellmer, J.C. General media and vessels suitable for woody plant culture. In Cell and Tissue Culture in Forestry: General Principles and Biotechnology; Springer: Dordrecht, The Netherlands, 1987; pp. 4–16. [Google Scholar]
  49. Noble, W.S. What is a support vector machine? Nat. Biotechnol. 2006, 24, 1565–1567. [Google Scholar] [CrossRef]
  50. Breiman, L. Random forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  51. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM Sigkdd International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794. [Google Scholar]
  52. Appelhans, T.; Mwangomo, E.; Hardy, D.R.; Hemp, A.; Nauss, T. Evaluating machine learning approaches for the interpolation of monthly air temperature at Mt. Kilimanjaro, Tanzania. Spat. Stat. 2015, 14, 91–113. [Google Scholar] [CrossRef]
  53. Rasmussen, C.E. Gaussian processes in machine learning. In Summer School on Machine Learning; Springer: Berlin/Heidelberg, Germany, 2003; pp. 63–71. [Google Scholar]
  54. Eren, B.; Türkoğlu, A.; Haliloğlu, K.; Demirel, F.; Nowosad, K.; Özkan, G.; Niedbała, G.; Pour–Aboughadareh, A.; Bujak, H.; Bocianowski, J. Investigation of the Influence of Polyamines on Mature Embryo Culture and DNA Methylation of Wheat (Triticum aestivum L.) Using the Machine Learning Algorithm Method. Plants 2023, 12, 3261. [Google Scholar] [CrossRef] [PubMed]
  55. Kuhn, M.; Wing, J.; Weston, S.; Williams, A.; Keefer, C.; Engelhardt, A.; Cooper, T.; Mayer, Z.; Kenkel, B.; Benesty, M. Caret: Classification and Regression Training: R Package; Cran: Vienna, Austria, 2019. [Google Scholar]
  56. R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria; Available online: https://www.R-project.org/ (accessed on 1 December 2022).
  57. Metivier, P.S.; Yeung, E.C.; Patel, K.R.; Thorpe, T.A. In vitro rooting of microshoots of Cotinus coggygria Mill, a woody ornamental plant. In Vitro Cell Dev. Biol. Plant 2007, 43, 119–123. [Google Scholar] [CrossRef]
  58. Ali, A.; Ahmad, T.; Abbasi, N.A.; Hafiz, I.A. Effect of different concentrations of auxins on in vitro rooting of olive cultivar ‘Moraiolo’. Pak. J. Bot. 2009, 41, 1223–1231. [Google Scholar]
  59. Singh, C.K.; Raj, S.R.; Jaiswal, P.S.; Patil, V.R.; Punwar, B.S.; Chavda, J.C.; Subhash, N. Effect of plant growth regulators on in vitro plant regeneration of sandalwood (Santalum album L.) via organogenesis. Agrofor. Syst. 2016, 90, 281–288. [Google Scholar] [CrossRef]
  60. Hunt, M.A.; Trueman, S.J.; Rasmussen, A. Indole–3–butyric acid accelerates adventitious root formation and impedes shoot growth of Pinus elliottii var. elliottii × P. caribaea var. hondurensis cuttings. New For. 2011, 41, 349–360. [Google Scholar] [CrossRef]
  61. Fallahpour, M.; Miri, S.M.; Bouzari, N. Propagation Of ‘Gisela 5′Rootstock As Affected By Mineral Composition Of Media And Plant Growth Regulators. J. Hortic. Res. 2015, 23, 57–64. [Google Scholar] [CrossRef]
  62. Sisko, M. In vitro propagation of Gisela 5 (Prunus cerasus × P. canescens). Agric. Slov. 2011, 8, 31–34. [Google Scholar]
  63. Hui–Mei, W.; Yuan–Gang, Z.; Feng–Li, D.; Xiao–Ju, Z. Assessment of factors affecting in vitro shoot regeneration from axillary bud explant of Camptotheca acuminata. J. For. Res. 2005, 16, 52–54. [Google Scholar] [CrossRef]
  64. Haradzi, N.A.; Khor, S.P.; Subramaniam, S.; Chew, B.L. Regeneration and micropropagation of Meyer lemon (Citrus × meyeri) supported by polymorphism analysis via molecular markers. Sci. Hortic. 2021, 286, 110225. [Google Scholar] [CrossRef]
  65. El–Agamy, S.Z.; Mostafa, R.A.; Shaaban, M.M.; El–Mahdy, M.T. In vitro propagation of Manfalouty and Nab El–gamal pomegranate cultivars. Res. J. Agric. Biol. Sci. 2009, 5, 1169–1175. [Google Scholar]
  66. Lozzi, A.; Abdelwahd, R.; Mentag, R.; Abousalim, A. Development of a new culture medium and efficient protocol for in vitro micropropagation of Ceratonia siliqua L. In Vitro Cell Dev. Biol. Plant 2019, 55, 615–624. [Google Scholar] [CrossRef]
  67. Glass, A.D.; Britto, D.T.; Kaiser, B.N.; Kinghorn, J.R.; Kronzucker, H.J.; Kumar, A.; Okamoto, M.; Rawat, S.; Siddiqi, M.Y.; Unkles, S.E.; et al. The regulation of nitrate and ammonium transport systems in plants. J. Exp. Bot. 2002, 53, 855–864. [Google Scholar] [CrossRef] [PubMed]
  68. Mansseri–Lamrioui, A.; Louerguioui, A.; Bonaly, J.; Yakoub–Bougdal, S.; Allili, N.; Gana–Kebbouche, S. Proliferation and rooting of wild cherry: The influence of cytokinin and auxin types and their concentration. Afr. J. Biotechnol. 2011, 10, 8613–8624. [Google Scholar]
  69. Reddy, A.S. Calcium: Silver bullet in signaling. Plant Sci. 2001, 160, 381–404. [Google Scholar] [CrossRef]
  70. Ghori, N.H.; Ghori, T.; Hayat, M.Q.; Imadi, S.R.; Gul, A.; Altay, V.; Ozturk, M. Heavy metal stress and responses in plants. Int. J. Environ. Sci. Technol. 2019, 16, 1807–1828. [Google Scholar] [CrossRef]
  71. Thor, K. Calcium—Nutrient and messenger. Front. Plant Sci. 2019, 10, 440. [Google Scholar] [CrossRef]
  72. Aranda–Peres, A.N.; Peres, L.E.P.; Higashi, E.N.; Martinelli, A.P. Adjustment of mineral elements in the culture medium for the micropropagation of three Vriesea bromeliads from the Brazilian Atlantic Forest: The importance of calcium. HortScience 2009, 44, 106–112. [Google Scholar] [CrossRef]
  73. Nakagawa, Y.; Katagiri, T.; Shinozaki, K.; Qi, Z.; Tatsumi, H.; Furuichi, T.; Kishigami, A.; Sokabe, M.; Kojima, I.; Sato, S.; et al. Arabidopsis plasma membrane protein crucial for Ca2+ influx and touch sensing in roots. Proc. Natl. Acad. Sci. USA 2007, 104, 3639–3644. [Google Scholar] [CrossRef]
  74. Niazian, M. Application of genetics and biotechnology for improving medicinal plants. Planta 2019, 249, 953–973. [Google Scholar] [CrossRef]
  75. Hesami, M.; Condori–Apfata, J.A.; Valderrama Valencia, M.; Mohammadi, M. Application of artificial neural network for modeling and studying in vitro genotype–independent shoot regeneration in wheat. Appl. Sci. 2020, 10, 5370. [Google Scholar] [CrossRef]
  76. Niazian, M.; Niedbała, G. Machine learning for plant breeding and biotechnology. Agriculture 2020, 10, 436. [Google Scholar] [CrossRef]
  77. Aasim, M.; Akin, F.; Ali, S.A.; Taskin, M.B.; Colak, M.S.; Khawar, K.M. Artificial neural network modeling for deciphering the in vitro induced salt stress tolerance in chickpea (Cicer arietinum L). Physiol. Mol. Biol. Plants 2023, 29, 289–304. [Google Scholar] [CrossRef] [PubMed]
  78. Hesami, M.; Naderi, R.; Tohidfar, M. Modeling and optimizing in vitro sterilization of chrysanthemum via multilayer perceptron–non–dominated sorting genetic algorithm–II (MLP–NSGAII). Front. Sci. 2019, 10, 282. [Google Scholar] [CrossRef] [PubMed]
  79. Niazian, M.; Noori, S.A.S.; Galuszka, P.; Tohidfar, M.; Mortazavian, S.M.M. Genetic stability of regenerated plants via indirect somatic embryogenesis and indirect shoot regeneration of Carum copticum L. Ind. Crops Prod. 2017, 97, 330–337. [Google Scholar] [CrossRef]
  80. Mirza, K.; Aasim, M.; Katırcı, R.; Karataş, M.; Ali, S.A. Machine learning and artificial neural networks–based approach to model and optimize ethyl methanesulfonate and sodium azide induced in vitro regeneration and morphogenic traits of water hyssops (Bacopa monnieri L.). J. Plant Growth Regul. 2023, 42, 3471–3485. [Google Scholar] [CrossRef]
  81. Arab, M.M.; Yadollahi, A.; Shojaeiyan, A.; Ahmadi, H. Artificial neural network genetic algorithm as powerful tool to predict and optimize in vitro proliferation mineral medium for G× N15 rootstock. Front. Plant Sci. 2016, 7, 1526. [Google Scholar] [CrossRef]
  82. Kirtis, A.; Aasim, M.; Katırcı, R. Application of artificial neural network and machine learning algorithms for modeling the in vitro regeneration of chickpea (Cicer arietinum L.). Plant Cell Tissue Organ Cult. 2022, 150, 141–152. [Google Scholar] [CrossRef]
  83. Aasim, M.; Ali, S.A.; Bekiş, P.; Nadeem, M.A. Light–emitting diodes induced in vitro regeneration of Alternanthera reineckii mini and validation via machine learning algorithms. In Vitro Cell. Dev. Biol. Plant 2022, 58, 816–825. [Google Scholar] [CrossRef]
Horticulturae 09 01112 g001
Figure 1. The Aronia melanocarpa (Michx.) Elliott explants were grown under in vitro conditions (A–C). Explants that were grown for sixty days on WPM (D) and MS (E) medium with progressively higher concentrations of IBA (from left to right; 0, 0.5, 1, 1.5, and 2 mg L−1).
Figure 1. The Aronia melanocarpa (Michx.) Elliott explants were grown under in vitro conditions (A–C). Explants that were grown for sixty days on WPM (D) and MS (E) medium with progressively higher concentrations of IBA (from left to right; 0, 0.5, 1, 1.5, and 2 mg L−1).
Horticulturae 09 01112 g001
Horticulturae 09 01112 g002
Figure 2. Actual values observed for in vitro parameters and values predicted by ML models.
Figure 2. Actual values observed for in vitro parameters and values predicted by ML models.
Horticulturae 09 01112 g002
Table 1. Analysis of variance results of in vitro parameters observed in Aronia melanocarpa (Michx.) Elliott cultured after termination of the experiment.
Table 1. Analysis of variance results of in vitro parameters observed in Aronia melanocarpa (Michx.) Elliott cultured after termination of the experiment.
MediumDose (mg L−1)TNR 1LRL 2ARL 3NMR 4NS 5SL 6SD 7LW 8LL 9
MS0.06.33 ± 2.0153.73 ± 6.46 b1053.12 ± 6.49 abc1.13 ± 0.231.40 ± 0.2038.18 ± 12.852.00 ± 0.00 c5.87 ± 1.6711.60 ± 2.96
0.521.86 ± 3.6380.27 ± 10.01 ab51.87 ± 7.44 abc1.90 ± 0.364.20 ± 0.5352.39 ± 10.242.30 ± 0.26 b7.97 ± 0.5512.80 ± 1.20
1.026.42 ± 3.8752.93 ± 12.62 b44.33 ± 8.14 bcd2.13 ± 0.504.60 ± 0.6049.13 ± 4.712.00 ± 0.20 c8.20 ± 1.3813.13 ± 2.21
1.518.20 ± 6.4847.35 ± 16.02 b40.20 ± 10.82 cd1.87 ± 0.232.93 ± 0.5747.00 ± 12.522.00 ± 0.00 c6.53 ± 1.3610.60 ± 1.40
2.028.53 ± 3.2565.33 ± 13.45 ab46.73 ± 4.81 bcd2.13 ± 0.304.73 ± 0.6460.00 ± 14.202.33 ± 0.31 b6.40 ± 2.2212.93 ± 4.21
Means20.27 ± 8.82 B59.92 ± 15.90 B47.25 ± 8.23 B1.83 ± 0.47 B3.57 ± 1.37 B49.34 ± 12.13 B2.13 ± 0.23 B6.99 ± 1.61 B12.21 ± 2.44 B
WPM0.012.47 ± 1.7240.53 ± 9.42 b34.78 ± 7.57 d1.07 ± 0.111.53 ± 0.3035.33 ± 2.341.93 ± 0.11 c4.93 ± 0.2310.93 ± 1.28
0.526.68 ± 4.4397.93 ± 7.00 a63.28 ± 10.57 a2.32 ± 0.435.12 ± 0.6463.91 ± 12.502.81 ± 0.18 a9.72 ± 0.6715.62 ± 1.46
1.032.23 ± 7.2664.58 ± 15.40 ab54.09 ± 9.92 abc2.60 ± 0.615.61 ± 0.7359.94 ± 5.752.44 ± 0.24 b10.00 ± 1.6916.02 ± 2.69
1.522.20 ± 7.9157.77 ± 16.54 ab49.04 ± 13.20 abcd2.28 ± 0.283.58 ± 0.7057.34 ± 15.272.44 ± 0.00 b7.97 ± 1.6612.93 ± 1.71
2.034.81 ± 7.3379.71 ± 13.27 ab57.02 ± 10.65 ab2.60 ± 0.615.78 ± 0.5573.20 ± 17.332.85 ± 0.37 a7.81 ± 2.7115.78 ± 2.03
Means25.68 ± 9.73 A68.10 ± 23.26 A51.64 ± 13.34 A2.17 ± 0.67 A4.33 ± 1.75 A57.95 ± 16.49 A2.49 ± 0.40 A8.08 ± 2.33 A14.26 ± 2.84 A
Mean Dose0.09.40 ± 3.75 d47.13 ± 10.22 d43.95 ± 11.86 c1.10 ± 0.16 c1.47 ± 0.24 d36.76 ± 8.41 c1.97 ± 0.08 c5.40 ± 1.18 c11.27 ± 2.07 b
0.524.27 ± 4.47 b89.10 ± 12.37 a57.57 ± 10.29 a2.11 ± 0.42 b4.66 ± 0.73 b58.15 ± 12.01 b2.55 ± 0.38 a8.84 ± 1.10 a14.21 ± 1.95 a
1.029.33 ± 6.10 a58.76 ± 14.11 c49.21 ± 9.71 bc2.36 ± 0.56 a5.11 ± 0.81 a54.54 ± 7.56 b2.22 ± 0.31 b9.10 ± 1.69 a14.58 ± 2.71 a
1.520.20 ± 6.83 c52.56 ± 16.97 cd44.62 ± 11.82 c2.07 ± 0.32 b3.26 ± 0.67 c52.17 ± 13.71 b2.22 ± 0.24 b7.25 ± 1.56 b11.77 ± 1.89 b
2.031.67 ± 6.13 a72.52 ± 14.31 b51.87 ± 9.29 b2.37 ± 0.39 a5.25 ± 0.92 a66.60 ± 15.91 a2.59 ± 0.41 a7.10 ± 2.35 b14.36 ± 3.80 a
MSM219.31 ***501.70 *144.55 *0.87 ***4.23 ***555.42 ***1.01 ***8.97 **31.31 ***
MSD464.44 ***1717.94 ***188.19 ***1.65 ***15.29 ***714.46 ***0.41 ***13.49 ***15.06 ***
MS(MxD)1.42 ns225.87 *243.45 ***0.08 ns0.22 ns63.24 ns0.09 *1.97 ns3.52 ns
* Significant at p ≤ 0.05, ** significant at p ≤ 0.01, *** significant at p ≤ 0.001, ns non-significant at p ≥ 0.05, ± standard deviation. MSM, mean square of mediums (M), MSD, mean square of doses (D), MS(MxD), mean square of medium (M) and dose (D) interactions. 1 Total number of roots, 2 longest root length, 3 average root length, 4 number of main roots, 5 number of siblings, 6 shoot length, 7 shoot diameter, 8 leaf width, 9 leaf length, 10 Letters of the same size and notation indicate important items.
Table 2. Algorithms’ goodness-of-fit criteria for prediction of in vitro parameters (total number of roots (TNR), longest root length (LRL), average root length (ARL), number of main roots (NMR), number of siblings (NS), shoot length (SL), shoot diameter (SD), leaf width (LW), leaf length (LL)).
Table 2. Algorithms’ goodness-of-fit criteria for prediction of in vitro parameters (total number of roots (TNR), longest root length (LRL), average root length (ARL), number of main roots (NMR), number of siblings (NS), shoot length (SL), shoot diameter (SD), leaf width (LW), leaf length (LL)).
TraitsML 1 CriteriaSVM 2RF 3XGBoost 4KNN 5GP 6
TNRR20.5750.5230.5890.4980.569
MSE7.0797.5046.9657.6927.131
MAPE26.43134.98125.68637.58729.218
MAD5.1846.1965.2796.2165.523
LRLR20.1380.2450.2810.1000.231
MSE29.08527.23126.56430.37127.472
MAPE23.36031.52330.83335.88230.017
MAD16.76019.31719.21221.12918.747
ARLR20.1350.2070.2670.0380.206
MSE14.21013.60813.08014.99013.616
MAPE20.65323.44322.12927.24523.129
MAD10.39210.87110.35912.32710.824
NMRR20.7140.6730.7360.6060.711
MSE0.3170.3390.3040.3720.318
MAPE13.23116.74013.06918.02314.256
MAD0.2450.3000.2530.3160.267
NSR20.7800.7210.7950.6420.786
MSE0.7830.8800.7550.9970.771
MAPE14.82524.61415.01230.06315.716
MAD0.5460.7390.5720.8460.588
SLR20.5340.5060.5710.4360.544
MSE9.98910.2859.59110.9909.882
MAPE14.65218.14015.43519.74316.592
MAD7.7529.1147.9369.7178.491
SDR20.6480.6640.7430.4390.713
MSE0.2170.2130.1860.2750.196
MAPE6.7127.2945.6379.7756.334
MAD0.1590.1710.1330.2220.150
LWR20.5350.5180.5830.3490.527
MSE1.3731.3981.3001.6251.384
MAPE13.88117.82714.98421.52416.974
MAD0.9901.2131.0511.4141.168
LLR20.1880.3660.3800.2740.368
MSE2.6332.3262.2412.4912.301
MAPE13.52613.88313.23715.55414.085
MAD1.8961.8591.7772.0301.874
1 Machine learning criteria (R2, R-squared; MSE, mean squared error; MAPE, mean absolute percentage error; MAD, mean absolute deviation), 2 support vector machine, 3 random forest, 4 extreme gradient boosting, 5 k-nearest neighbors classifier, 6 Gaussian processes classifier.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Demirel, F.; Uğur, R.; Popescu, G.C.; Demirel, S.; Popescu, M. Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott). Horticulturae 2023, 9, 1112. https://doi.org/10.3390/horticulturae9101112

AMA Style

Demirel F, Uğur R, Popescu GC, Demirel S, Popescu M. Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott). Horticulturae. 2023; 9(10):1112. https://doi.org/10.3390/horticulturae9101112

Chicago/Turabian Style

Demirel, Fatih, Remzi Uğur, Gheorghe Cristian Popescu, Serap Demirel, and Monica Popescu. 2023. "Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott)" Horticulturae 9, no. 10: 1112. https://doi.org/10.3390/horticulturae9101112

APA Style

Demirel, F., Uğur, R., Popescu, G. C., Demirel, S., & Popescu, M. (2023). Usage of Machine Learning Algorithms for Establishing an Effective Protocol for the In Vitro Micropropagation Ability of Black Chokeberry (Aronia melanocarpa (Michx.) Elliott). Horticulturae, 9(10), 1112. https://doi.org/10.3390/horticulturae9101112

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop