JOURNAL OF AGRICULTURAL INFORMATICS AGRÁRINFORMATIKA FOLYÓIRAT

Special Issue on Information Systems: an Agro-Environment Perspective

Co-Editors of this Special Issue

Dr. Zacharoula Andreopoulou Prof. Dr. Milan Mesic Dr. Mariana Golumbeanu

2018. HAAI Vol. 9, No. 1 ISSN 2061-862X

http://journal.magisz.org

The creation of the journal has supported by the New Hungary Development

“Dissemination of research result on innovative information technologies in agriculture”

The project is nanced by the European Union, with the co- nancing of the European Social Fund. Journal of Agricultural Informatics 2018 Vol. 9, No. 1

Journal of Agricultural Informatics

Scientific Journal

Name: Agricultural Informatics

Language: English

Issues: 2-4 per year

Publisher: Hungarian Association of Agricultural Informatics (HAAI), H-4032 Debrecen, Böszörményi út. 138. Hungary

ISSN 2061-862X

ISSN 2061-862X (http://journal.magisz.org/) I

Journal of Agricultural Informatics 2018 Vol. 9, No. 1

Board of Advisors and Trustees

BANHAZI, Thomas - University of Southern Queensland, Australia RAJKAI, Kálmán László - Institute for Soil Sciences and Agricultural Chemistry, Centre for Agricultural Research, HAS, Hungary SIDERIDIS, Alexander B. - Agricultural University of Athens, Greece ZAZUETA, Fedro - University of Florida, USA Editor in Chief

HERDON, Miklós – University of Debrecen, Hungary Associate Editors

ANDREOPOULOU, Zacharoula S. - Aristotle University of Thessaloniki, Greece GAÁL, Márta - Research Institue of Agricultural Economics, Hungary SZILÁGYI, Róbert - University of Debrecen, Hungary Editorial Board

BATZIOS, Christos – Aristotle University of Thessaloniki, Greece BERUTTO, Remigio – University of Torino, Italy BUSZNYÁK, János – Pannon University, Hungary CHARVAT, Karel – Czech Centre for Science and Society, Czech Republic CSUKÁS, Béla – Kaposvár University, Hungary DIBARI, CAMILLA – University of Florence, Italy ENGINDENIZ, Sait – EGE University, Turkey FELFÖLDI, JÁNOS – University of Debrecen, Hungary GACEU, Liviu – University of Transilvania, Romania JUNG, András – Szent István University, Hungary LENGYEL, Péter – University of Debrecen, Hungary TAMÁS, János – University of Debrecen, Hungary THEUVSEN, Ludwig – Georg-August-University Göttingen, Germany VARGA, Mónika – University of Kaposvár, Hungary VÁRALLYAI, László – University of Debrecen, Hungary XIN, Jiannong – University of Florida, USA WERES, Jerzy – Poznan University of Life Sciences, Poland Technical Editors

PANCSIRA, János - University of Debrecen, Hungary

ISSN 2061-862X (http://journal.magisz.org/) II

Journal of Agricultural Informatics 2018 Vol. 9, No. 1

PREFACE

Information technology is an everyday means that is found in all walks of life today. This is also true for almost all areas of agricultural management. The aim of this Journal is to improve scientific knowledge dissemination and innovation process in the agri-food sector. The Journal of Agricultural Informatics has been established in 2009 by the HAAI within a project of the Hungarian National Development Plan Framework. The peer-reviewed journal is operating with international editorial and advisory board supported by the EFITA (European Federation for Information Technology in Agriculture Food and the Environment). This journal serves the publication as well as familiarization the results and findings of research, development and application in the field of agricultural informatics to a wide public. It also wishes to provide a forum to the results of the doctoral (Ph.D) theses prepared in the field of agricultural informatics. Opportunities for information technology are forever increasing, they are also becoming more and more complex and their up-to-date knowledge and utilisation mean a serious competitive advantage. Information Systems for Agricultural and Environmental Sector are continually achieving approval among scientific publications in the past years, since they provide an extensive variety of highly valued innovative applications and methods and can suggest smart services, and up-to-date solutions, that can serve the needs of citizens, enterprises and modern society in general. ICTs can support as an advantage in our “smart-enhanced” society. Volume 9, Number 1 is a special issue under the title “Information Systems: an Agro-Environment Perspective”. It aims to present 6 papers with innovative Information Systems applications and techniques in Forestry, Agriculture, and Environment and, hopefully, to enhance the scientific discussion between scientists, professionals and experts on IT sector with special focus on agriculture, forestry, environment and sustainable development. There are also included papers presented in the International Workshop “Information Technology, Sustainable Development, Scientific Network and Nature Protection" which took place in Edessa, Greece, 8-11 October 2017, during the 18th Panhellenic Forestry Congress, in an updated and extended version. The selection of the papers appearing in this special issue was based on the relevance of their subject to the scope of the Journal of Agricultural Informatics and on the evaluation score of the full papers during the double blind peer review process. We trust that this special issue will encourage scientific research in the area of Agro-Environmental Informatics presenting interesting case studies in the fields of Forestry, Agriculture and Environment within sustainability issues.

Prof. Dr. Miklós Herdon Chair of the Editorial Board and Co-Editors of this Special Issue: Dr. Zacharoula Andreopoulou, Aristotle University of Thessaloniki, Greece Prof. Dr. Milan Mesic, University of Zagreb, Croatia Dr. Mariana Golumbeanu, Institute for Research & Development “Grigore Antipa”,Romania

ISSN 2061-862X (http://journal.magisz.org/) III

Journal of Agricultural Informatics 2018 Vol. 9, No. 1

Content

Christos Liotiris Enhancement of the forest road network accessibility using Information Systems ...... 1

Agni Kalfagianni, Zacharoula Andreopoulou Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs ...... 8

Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas Analysis of logging forest residues as an energy source ...... 14

Aristotelis Martinis , Evgenia Chaideftou , Charikleia Minotou , Κonstantinos Poirazidis Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece ...... 26

Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill...... 41

Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera A Genetic Programming Study on Classification of Cassava Plant...... 47

ISSN 2061-862X (http://journal.magisz.org/) IV

Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol. 9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

Enhancement of the forest road network accessibility using Information Systems

Christos Liotiris 1

I N F O A B S T R A C T Received 10 Jan 2018 Accepted 5 Mar 2018 Accessibility of the forest road network is one of the most important factors in the Available on-line 23 Mar 2018 sustainable management of forests. This study presents the development of an information Responsible Editor: M. Herdon system which enhances the accessibility of the forest road network. This project uses HTML5 and CSS3 for the frontend and WAMP Server with MySQL database as backend. Keywords: End users will be able to receive the necessary information and coordinates via Accessibility, Information notifications on their Android smartphone using an application. This Android application System, Forest Road Network, uses the Google Maps app to display the route that should be followed. Forwarding Open Source Software, Android information to users is operated through Firebase Cloud Messaging. Last but not least, it Application should be highlighted that free and open source software was used to implement the aforementioned system.

1. Introduction

One quarter of the total landmass of Greece is covered by forests, ranking it fourth among the European countries. Forest fires are considered to be one of the most decisive factors affecting forest resources throughout the world and, in particular, Mediterranean countries (i.e., France, Greece, Italy, Portugal, Spain, and Turkey) due to their climate and other factors (Demir et al. 2009). In addition, changes in forest landscapes resulting from road construction, have increased remarkably in recent years. Roads are essential structures to provide access to the forest from the establishment phase to the harvesting stage. Consequently, it is important that roads are properly planned in order to ensure the transportation of forest products as well as the safety, comfort and economy of vehicle operations. The complex structure of the forest road network makes it necessary to use navigation technologies to facilitate the movement of vehicles, citizens and employees working in it. The sustainable management of forest resources can only be achieved through a well-organized road network (Athanasiadis & Andreopoulou 2015).

Global Positioning System (GPS) technology has been increasingly used in many forestry applications such as forest operations, forest transportation, forest fires, etc. Furthermore, it enhances the ability of firefighting vehicles and staff to reach a fire area as quick as possible. Most of those technologies are based on GPS navigation devices or GPS navigation applications for smartphones. Land navigation methods and tools such as Compass with adjustment for magnetic declination, Clinometer, Topographic maps etc., have been used widely so far but their accuracy can be improved by the Information Technologies and the Internet (Wing et al. 2005).

There are many reasons given for the growth of the Internet usage (not specifically retailing) over recent times, including for example, its size as a source of information, increasingly becoming much more accessible and less expensive (Bonn et al. 1999). Especially the latter one, is the main reason why people can access Internet via their smartphones, tablets, laptops etc. mainly for work and entertainment, in daily basis. The purpose of this paper is the development of an Information System which combines different open source technologies and it will be running on Windows machines and Android devices. It aims to facilitate the movement of vehicles and citizens, reducing the unnecessary shifts and providing the fastest and most reliable path.

______1 Christos Liotiris Aristotle University of Thessaloniki [email protected] doi: 10.17700/jai.2018.9.1.432 1 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

2. Materials and Methods

This Information System uses HTML5 and CSS3 for the frontend and WAMP Server with MySQL database as backend. In order to achieve the best performance, users should install Windows 7 or newer edition. It should be pointed out that WAMP Server 2.5 is not compatible with Windows XP, neither with SP3, nor Windows Server 2003. The programming languages and the technologies which were used for the development of this Information System, will be presented below.

WAMP Server refers to a software stack for the Microsoft Windows operating system and consisting of the Apache web server, OpenSSL for SSL support, MySQL database and PHP programming language. The most important part of the WAMP package is Apache (or "Apache HTTP Server") which is used run the web server within Windows. By running a local Apache web server on a Windows machine, a web developer can test webpages in a web browser without publishing them live on the Internet (Bourdon 2012).

Figure 1. Website with Google Maps

PHP (Personal Home Page or Hypertext Preprocessor) is a server-side scripting language designed primarily for web development but also used as a general-purpose programming language. PHP code may be embedded into HTML or HTML5 markup, or it can be used in combination with various web template systems, web content management systems and web frameworks (Lerdorf 2007).

HTML is the World Wide Web’s core markup language. Originally, HTML was primarily designed as a language for semantically describing scientific documents. Its general design, however, has enabled it to be adapted, over the subsequent years, to describe a number of other types of documents and even applications. HTML5 includes detailed processing models to encourage more interoperable implementations. It extends, improves and rationalizes the markup available for documents and introduces markup and application programming interfaces (APIs) for complex web applications. (W3C 2017)

CSS3 (Cascading Style Sheets) is a style sheet language used for describing the presentation of a document written in a markup language. Although most often used to set the visual style of web pages and user interfaces written in HTML. Along with HTML and JavaScript, CSS is a cornerstone technology used by most websites to create visually engaging webpages, user interfaces for web doi: 10.17700/jai.2018.9.1.432 2 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

applications, and user interfaces for many mobile applications (Clark 2014). CSS3 has been split into "modules". It contains the "old CSS specification" (which has been split into smaller pieces). In addition, new modules are added. Some of the most important CSS3 modules are: Selectors, Box Model, Backgrounds and Borders, Image Values and Replaced Content, Text Effects, 2D/3D Transformations, Animations, Multiple Column Layout and User Interface (Lazaris 2010).

MySQL is an open source relational database management system (RDBMS). Its name is a combination of "My", the name of co-founder Michael Widenius's daughter, and "SQL", the abbreviation for Structured Query Language (MySQL, 1995).

Figure 2. WAMPSERVER Homepage

Firebase is a mobile and web application development platform, has grown inside Google and expanded their services to become a unified platform for mobile developers. Firebase now integrates with various other Google services to offer broader products and scale for developers (Firebase 2011).

Google Maps is a web mapping service developed by Google. It offers satellite imagery, street maps, 360° panoramic views of streets (Street View), real-time traffic conditions (Google Traffic), and route planning for traveling by foot, car, bicycle (in beta), or public transportation. Google developed Google Maps App (mapping mobile app) for the Android and iOS mobile operating systems and it uses Google Maps for its information. The Google Maps apps on Android and iOS have many features in common, including turn-by-turn navigation, street view, and public transit information (Google 2017).

Postman is a powerful GUI (Graphical user interface) platform to make your API development faster & easier, from building API requests through testing, documentation and sharing. It was designed from the group up, to support all aspects of API development. Postman’s apps are built on a single underlying layer, ensuring consistent performance and user experience. Postman has features for every API developer: request building, test and pre-request scripts, variables, environments, and request descriptions, designed to work seamlessly together. It supports Mac/Windows/Linux Apps, individual and team options, and multiple integrations, including support for Swagger and RAML formats (Postman 2012).

Android Studio is the official integrated development environment (IDE) for Google's Android operating system, built based on JetBrains' IntelliJ IDEA software and designed specifically for doi: 10.17700/jai.2018.9.1.432 3 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

Android development. It is available for download on Windows, macOS and Linux based operating systems (Android Studio 2013). It is a replacement for the Eclipse Android Development Tools (ADT) as primary IDE for native Android application development.

Figure 3. Postman GUI

3. Results

The project (information system) of this paper consists of a website using WAMP server, MySQL Database (phpMyAdmin), Google Maps, Postman (GUI), Firebase Cloud Messaging and an Android Application which triggers Google Maps application (Figure 4).

Figure 4. Information System structure

In order to describe this Information System step by step, we are going to use a hypothetical scenario. Furthermore, by using this we will be able to comprehend how the Information System interacts with the users so as to store the necessary data and send a notification or more, with the coordinates and other useful information. The scenario says that a logger got injured in the woods and we have to guide a rescue team using the shortest route. So, all we need to do is to send logger’s location to the leader of the rescue team. According to our hypothetical scenario, the Information doi: 10.17700/jai.2018.9.1.432 4 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

System requires at least two end users, User_A and User_B, with very distinct roles in order the system to be functional. User_A creates the marker(s), stores the data and transmits the information. User_B receives the information in his/her Android smartphone.

Initially User_A requests the html page from the server by selecting the file Project1 from the “Your VirtualHost” tab of WampServer’s Home page (Figure 2) and Google Maps is loaded. Then, User_A can click anywhere to the map creating a marker. By clicking on the marker, a pop-up window shows up with empty fields to fill in (Figure 1). All the data inserted by User_A and the coordinates are stored automatically in the database (Figure 7). Afterwards, the selected data from the database are aggregated by using the Postman GUI (Figure 3) in order to create the message which will be sent as notification. User_B receives the notification on his/her Android smartphone (Figure 5). Finally, the delivered notification triggers the Google Maps application (Android) showing the route that should be followed (Figure 6).

Figure 5. Notification Figure 6. Google Maps (Android)

4. Discussion-Conclusions

Information systems (IS) and information technology (IT) are often considered synonymous. In reality, information technology is a subset of information systems. The perception that these terms can be used interchangeably can cause confusion for individuals interested in pursuing a technology- related career. Although both these fields deal with computers, they have distinct characteristics and specific career paths that require different education and training.

Information Systems is an academic study of systems with a specific reference to information and the complementary networks of hardware and software that people and organizations use to collect, filter, process, create and also distribute data. An emphasis is placed on an information system having a definitive boundary, users, processors, storage, inputs, outputs and the aforementioned communication networks (Jessup & Valacich 2007).

Information Technology is considered a subset of information and communications technology (ICT). An ICT hierarchy was proposed where each hierarchy level "contains some degree doi: 10.17700/jai.2018.9.1.432 5 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

of commonality in that they are related to technologies that facilitate the transfer of information and various types of electronically mediated communications" (Zuppo 2012).

Figure 7. Database

Information and Communication Technologies can play a key role in the environmental protection, the environmental sustainability, the environmental education and the rural sustainable development (Andreopoulou 2014).

As previously stated, the forest road network accessibility plays a significant role in the sustainable forest management. Further, enhances forest’s socio-economic development apart from providing a safe and suitable mode of transport. In order to succeed this, an Information System was successfully developed reducing the unnecessary vehicle shifts and ensures the shortest route. Furthermore, it could be used as a guide system for hikers, employees of the Forest Service – Police and Fire Department or even for transporting injured persons. Finally, it should be highlighted that free and open source software was used to implement this system.

5. Acknowledgements

I would like to pay special thankfulness to my supervisor, Professor Andreopoulou Zacharoula, for the guidance and advice she has provided to me without hesitation. I have been extremely lucky to have a supervisor who cared so much about my work, and who responded to my questions and queries so promptly. Her help and constant motivation at every point during my paper helped me to cherish my goal in time.

References

Andreopoulou, S. Z., 2014. Green Informatics: ICT for Green and Sustainability. Agricultural Informatics, vol. 3, no. 2, pp. 1 – 8, ISSN 2061-862X http://www.magisz.org/journal Android Studio, 2013. Android Studio - The Official IDE for Android. [Online]. Available at: https://developer.android.com/studio/index.html [Accessed 28 December 2017]. Athanasiadis, A. & Andreopoulou, Z., 2015. A DSS for the identification of forest land types by the Greek Forest Service. International Journal of Sustainable Agricultural management and Informatics, vol. 1, no. 1, pp. 76-88, https://doi.org/10.1504/IJSAMI.2015.069053 doi: 10.17700/jai.2018.9.1.432 6 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:1-7

Bonn, M. A., Furr, H. L. & Susskind, A. M., 1999. Predicting a Behavioral Profile for Pleasure Travelers on the Basis of Internet Use Segmentation. Journal of Travel Research, vol. 37, no. 4, pp. 333 - 340. https://doi.org/10.1177/004728759903700403 Bourdon, R., 2012. WAMPSERVER, a Windows web development environment. [Online]. Available at: http://www.wampserver.com/en/ [Accessed 28 December 2017]. Clark, S., 2014. Web-based Mobile Apps of the Future Using HTML 5, CSS and JavaScript. [Online]. Available at: http://www.htmlgoodies.com/beyond/article.php/3893911/Web-based-Mobile-Apps-of-the-Future- Using-HTML-5-CSS-and-JavaScript.htm [Accessed 28 December 2017]. Demir, M. et al., 2009. Assessment of forest roads and firebreaks in Turkey. African Journal of Biotechnology, vol. 8, no. 18, pp. 4553 – 4561, ISSN 1684–5315 © 2009 Academic Journals Firebase, 2011. Firebase. [Online]. Available at: https://firebase.google.com/ [Accessed 28 December 2017]. Google, 2017. Google Maps. [Online]. Available at: https://www.google.com/maps/about/ [Accessed 28 December 2017]. Jessup, L. & Valacich, J., 2007. Information Systems Today: Managing in the Digital World. 3rd ed., Prentice Hall Press Upper Saddle River, NJ, USA ©2007. Lazaris, L., 2010. CSS3 Solutions for Internet Explorer. [Online]. Available at:https://www.smashingmagazine.com/2010/04/css3-solutions-for-internet-explorer/ [Accessed 28 December 2017]. Lerdorf, R., 2007. PHP on Hormones – history of PHP presentation by Rasmus Lerdorf given at the MySQL Conference in Santa Clara, California. [Online]. Available at: http://web.archive.org/web/20130729204354id_/http://itc.conversationsnetwork.org/shows/detail3298.html [Accessed 28 December 2017]. MySQL, 1995. History of MySQL. [Online]. Available at: https://dev.mysql.com/doc/refman/5.7/en/history.html [Accessed 28 December 2017]. Postman, 2012. Postman is the most complete API Development Environment. [Online]. Available at: https://www.getpostman.com/postman [Accessed 28 December 2017]. W3C, 2017. HTML 5.2. [Online]. Available at: https://www.w3.org/TR/html52/ [Accessed 28 December 2017]. Wing, M. G., Eklund, A. & Kellogg, L. D., 2005. Consumer-Grade Global Positioning System (GPS) Accuracy and Reliability. Journal of Forestry, June, vol. 103, no. 4, pp. 169 - 173. Zuppo, C. M., 2012. DEFINING ICT IN A BOUNDARYLESS WORLD: THE DEVELOPMENT OF A WORKING HIERARCHY. International Journal of Managing Information Technology, vol. 4, no. 3, pp. 19, doi: 10.5121/ijmit.2012.4302. https://doi.org/10.5121/ijmit.2012.4302

doi: 10.17700/jai.2018.9.1.432 7 Christos Liotiris: Enhancement of the forest road network accessibility using Information Systems Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol. 9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs

Agni Kalfagianni1, Zacharoula Andreopoulou2

I N F O A B S T R A C T Received 28 Jan 2018 Accepted 5 Mar 2018 Nowadays Internet is a successful key for the promotion of Small and Medium Available on-line 23 Mar 2018 Enterprises (SMEs) by offering benefits to e-commerce through Internet marketing. A Responsible Editor: M. Herdon vast majority of SMEs related to agri - food sector has an eminent presence in the Web. This paper studies the characteristics and the connection between Internet and Social Keywords: media Market of Non Wood Forest Products especially of Mushrooms, Truffles and E-commerce, Marketing Herbs of Greek SMEs. The research compares the connection between a social media Mushrooms, Herbs, Non Wood profile of SMEs and its official website in firms in the above sector. As it derives through Forest Products. this study there is not a specific correlation that improves a connection between the SMEs profile of a social media and its website and Social media predominate in the internet market. Therefore, firm owners should focus on social media profile characteristics and features.

1. Introduction Contemporary society is characterized by continuous, growing and fast changes and emerging research applications that require the adaptation of fast communication features, exchange of Information and data and include ICTs in everyday life (Andreopoulou et.al., 2011). Through convenient applications on the Internet, people interact with each other and engage in activities related to daily life, electronic commerce (e-commerce) by using virtual platforms (Piccoli & Ives, 2003; Powell, Piccoli, & Ives, 2004; Andreopoulou; 2013; Turban et al., 2018). Internet is an interactive channel that can be used for informational purposes and by the same time it can eliminate not only the distances but also it can offer its services 24/7. The target consumers audience is extended globally, as e-commerce changed the traditional way of businesses as it provides a new marketing tool and can apply to potential customers for a firm in a more general public as it offers the opportunity to gain customers worldwide (Kalfagianni et al., 2017). The website of a business is crucial for the success of a SME, as it concerns the Internet market of the product and it can be characterised as a dynamic intermediate mean between buyer and seller (Dwivedi, Kappor and Chen, 2015). Websites serve as an important point of contact for most companies, assessing their effectiveness or quality of the website is important as a way to understand whether the company is providing the type and quality of information and interaction to satisfy website users (Tsekouropoulos et al., 2012). This is especially true for companies selling goods and services on their websites such as products derived from forests other than wood (Kalfagianni, 2017). Among the various Internet applications, social media, including social networking sites such as Facebook, Twitter, and LinkedIn have become extremely popular in the past decade. A business social

1 Agni Kalfagianni PhD candidate, Aristotle University of Thessaloniki, Laboratory of forest informatics, School of Forestry and Natural Environment [email protected] 2 Zacharoula Andreopoulou Associate Professor, Aristotle University of Thessaloniki, Laboratory of forest informatics, School of Forestry and Natural Environment [email protected] doi: 10.17700/jai.2018.9.1.434 8 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

profile can built relationship with the customers and achieve win – win deals with them (Töllinen & Karjaluoto, 2011; Andreopoulou et al., 2012). Regarding Non Wood Forest products, since 2009 Food and Agricultural Organization of the United Nations (FAO) mentioned them as a sector of priority. The definition given by FAO and approved by ither organizations, describes them as products that derive from forests, scrubs and tree plantations and differ from the sense of wood (FAO, 1999). The last years, there is a trend for a healthy and mediterranean diet, aroma - therapy and environmentally friendly products that increased the demand of Non Wood Products (Eand Misen, 1995; Pattenella et al., 2006). Mushrooms, Truffles and Herbs are an every day-consumed food especially in the area of Europe (Bharali et al., 2017; Shaaban & Moawad; 2017). They are used since ancient times in traditional medicine and are of great importance for people by the Mediterranean Sea (Jones, 1996; Eddouks et al., 2017; Manju et al., 2017). Moreover, it is though that selling forest product other than wood can be a more relevant source of income (Freed, 2001; Merlo & Croitoru, 2005; Subedi, 2007; Sisak, Riedl and Dudik, 2016; Živojinović et al., 2017). In this paper, the research is focused on Small and Medium Enterprises (SMEs) that are located in Greece and cultivate Non Wood Forest products. There were retrieved firms that both use Facebook and official websites in order to communicate with their customers. In addition the firms were classified according to their marketing and digital characteristics and by the same time there are described their social media network profiles as to their characteristics. Also it is given a focus to the correlation between the promotions of a firm by different digital means.

2. Methodology For the aim of the study, there were collected 16 SMEs that have a presence in Social media, e.g. Facebook and at the same time they promote their products (mushrooms and truffles) through websites. Moreover, for the same reason 17 SMEs that deal with herbs and promote their products both in Social media and Websites were selected. All firms are located in Greece and deal only with Non Wood Forest products, for that reason they are cultivated and packaged by them, in Greece. The research was focused on finding all SMEs that are dealing with Non Wood Forest Products and they have a presence both in Social Media and Websites. Regarding the Social Media profile of SMEs seven (7) marketing and organizational criteria were studied and analysed as to their characteristics as mentioned by the literature (MacCann & Barlow, 2015). All quantitative data were collected from their individual social media sites. Referring to quantitative date there were examined: number of fans (according to the “Likes” that companies have on their profiles), followers (according to people who “Follow” them in Facebook), frequence of postings (that was categorised in three groups: daily, every 2 days and rare), response time to comments of future clients, clear managerial aims and goals (described by their management status on Facebook), promotion techniques that use “like and share” for competition and gaining products though drawing different posts of the firm and general information for the products of the firms and their characteristics as they are mentioned or not from the Facebook ‘s administrator of the firm. For the official websites of the SMEs, there were selected firms that have less than 5,000 visitors, according to the online searching tools “StatShow” and “Similar Web”, aiming to select similar SMEs related to their total digital commercial activities. The aim was to find the number of pages that a user navigates by a visit each time and it was also taken under consideration the qualitative parameters that have to do with the richness of a website (giving information about the location and the goals that the enterprise has set), the importance of information (that should be exact, complete and up to date), the navigation design which should be easy even to people with disabilities, the value of time of customers navigation, the existence of newsletter that should keep a contact between user and firm, product details for each product and promotional techniques or online shop for making ordering easier. At the end, it is described the correlation between the number of fans of the SMEs Facebook page to the visitors that visit the official website per month.

doi: 10.17700/jai.2018.9.1.434 9 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

3. Results Through the analysis of data, it was founded that both Mushrooms and Truffles and Herbs correlated SMEs were distributed non-normally in none of the five quantitative variables (which are: pages per visit, visitors per month and pages views per month regarding webpages and Fans and followers regarding Facebook profiles). In Table 1 are presented the Medians of every variable in each of the two categories of SMEs (Mushrooms and Truffles, Herbs). As Table 1 presents it, Internet users do not visit more than 1 page per visit. Furthermore, there are more visitors to Mushrooms and Truffles Websites than those in Herbs. This is also, followed in Facebook were there is a double numbered for Mushrooms and Truffles pages. Table 1. Medians of quantitative variables. Mushrooms Herbs and Truffles Website pages per visit 0 0.5

Website visitors 465 30 per month Website pages 1035 30 view per month

Facebook Likes 988.5 452

Facebook 985 451 Followers According to the analysis of the qualitative variables Table 2 was conducted. The majority of SMEs related to Mushrooms and Truffles are paying attention to the Design of the Website; as they have created websites that are easy to navigate, while herb’s Website do not pay attention. The majority of the Websites give important information for the products and they specify the goals and objectives of the SMEs. On the other hand, most of them do not promote their products through on- line shops but instead they are proposing websites navigators for a more personal contact with them. What should be mentioned is that firms that are related to herbs are giving great importance to inform future clients about their products but they do not continue having a future contact with them through the subscription in a newsletter.

The qualitative analysis of Facebook profiles showed that none of the SME of both categories promote products or give the opportunity to Facebook users for a discount to the products. Moreover, it should be mentioned that most of them have specific goals that are communicated to future clients.

To conclude there was no correlation between number of likes (Fans) in Facebook profile and Website visitor per month for the SMEs in Herbs (Pearson Correlation Coefficient = -0.9, p=0.7) but there was a small correlation between number of likes (Fans) in Facebook profile and Website visitor per month for the SMEs in Mushrooms and Truffles (Pearson Correlation Coefficient = 0.2, p=0.3).

doi: 10.17700/jai.2018.9.1.434 10 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

Table 2. Qualitative variables of Websites Mushrooms and Herbs Website Truffles Yes No Yes No

Information 93.75% 6.25% 100% -

Importance Information 75% 25% 82.3% 17.6%

Design 68.75% 31.25% 53% 47%

Convenience 62.5% 37.5% 58.8% 41.2%

Value of time for customers 43.75% 56.25% 53% 47%

Newsletter 12.5% 87.5% 11.8% 88.2%

Product details 50% 50% 53% 47%

Promo - shop 12.5% 87.5% 17.7% 82.3%

Table 3. Qualitative variables of Facebook profiles Mushrooms Herbs

Facebook profiles Yes No Yes No Objectives and 62.5% 37.5% 70.5% 29.5% Goals Promotion with - 100% - 100% “Like and Share” General Information 87.5% 12.5% 76.5% 23.5%

Response to posts 50% 50% 53% 47%

Every Every 2 Every Every 2 Frequency of Rare Rare day days day days response to posts 37.5% 64.8% 31.25% 31.25% 23.5% 11.7%

4. Conclusion As it was mentioned above, an effort to find SMEs that are both promoted in Social media, e.g. Facebook and through official Websites was found difficult, as the majority of them prefer only one of these means to communicate with customers. For this reason only 16 SMEs were examined for the Mushrooms and Truffles Market and 17 for Herbs market. doi: 10.17700/jai.2018.9.1.434 11 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

There was mentioned a trend for Mushrooms and Truffles SMEs than for Herbs, regarding the visitors of the Websites and Facebook pages. It seems that the majority of SMEs in both sectors give general information of their firms in both means of communication and most of them make an effort to propound a user friendly website that will attract future customers it will worth the spending time of them to the site. Unfortunately, in both sectors SMEs do not pay attention in promoting there products by giving special characteristics that will make them unique and competitive to others. A vast majority of them is trying to have a contact with the visitors by persuading them to become members or give their e-mails in order to register to their newsletter. By this research it was not mentioned a correlation between the number of fans in Facebook and Website users. It seemed that the presence of social media links in websites do not augment the number of fans in Facebook and vice versa presence of link of website in Facebook profile do not augment the number of website visitors. It is thought that, probably the small number of SMEs that were included in this article may have given a not clear impression of the real situation between the connection of Website and Social Media Profiles. It is evident that social media predominate in the internet market of Non Wood Forest Products. Thus, it is important that the SMEs owners should focus on social media profile characteristics and features, aiming to increase their profit. Additionally the small number of visitors and fans makes this result unclear. We hope that this article will give important information to stakeholders of the areas of Mushrooms, Truffles and Herbs and it will be a guide for entrepreneurs. It is though that it will boost the market of the area of Non Wood Forest Products and it will be an important base for the future creation of Facebook profiles and Websites.

5. Acknowledgement This study is a part of a broader study with the title “Integrated Information System of Non Wood Forest Products: Digital Traceability and Exploitation in Greece” that is founded by General Secretariat for Research and Technology and Hellenic Foundation for Research and Innovation.

References

Andreopoulou, Z., Manos, B., Polman, N. & Viaggi, D. (2011). Agricultural and environmental informatics, governance and management, Emerging Research Applications, IgiGlobal. https://doi.org/10.4018/978-1-60960- 621-3

Andreopoulou, Z., Koliouska, C., & Tsekouropoulos, G. (2012). Strategic planning and decision support in small-medium wood enterprises using database technology., AGRÁRINFORMATIKA/JOURNAL OF AGRICULTURAL INFORMATICS, vol. 3, no. 1, pp. 36-45. https://doi.org/10.17700/jai.2012.3.1.85

Andreopoulou, Z. (2013). Green Informatics: ICT for Green and Sustainability, Journal of Agricultural Informatics, vol. 3, no. 2, pp. 1-8. https://doi.org/10.17700/jai.2012.3.2.89

Bharali, P., Sharma, M., Sharma, C. L., & Singh, B. (2017). Ethnobotanical survey of spices and condiments used by some tribes of Arunachal Pradesh, Journal of Medicinal Plants, vol. 5, no. 1, pp. 101-109. Dwivedi, Y., Kapoor, K. and Chen, H. (2015). Social media marketing and advertising, The Marketing Review, vol. 15, no. 3, pp. 289-309. https://doi.org/10.1362/146934715X14441363377999 Eddouks, M., Ajebli, M., & Hebi, M. (2017). Ethnopharmacological survey of medicinal plants used in Daraa- Tafilalet region (Province of Errachidia), Morocco. Journal of ethnopharmacology, vol. 198, pp. 516-530. https://doi.org/10.1016/j.jep.2016.12.017 FAO (Food and Agriculture Organization of the United Nation) 1999, Towards a harmonized definition of non- wood forest products. [4 March 2018]. Freed, J. (2001). Non-Timber Forest Products in Local Economies. Journal of Sustainable Forestry, vol. 13, no. 3-4, pp. 67-69. https://doi.org/10.1300/J091v13n03_06 doi: 10.17700/jai.2018.9.1.434 12 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:8-13

Kalfagianni, A., ’E-commerce challenges in Mushrooms ant Truffles in Internet Social Networks: The state of Greek SMEs’ Proceedings of the 18th Hellenic Forestry Congress and International Workshop on "Information Technology, Sustainable Development, Scientific Networks and Nature Protection", Edessa, Greece. pp. 320- 326. [8-11 October 2017]. Kalfagianni, A., Andreopoulou, Z., Tsiaras, S., Tsekouropoulos, G., Assessment of the Internet market in Greece; The case of Mushrooms and Truffles, Proceedings of the 6th International Symposium and 28th National Conference on Operational Research, Thessaloniki, Greece. Abstract session 11: Digital Economy pp. 135-139. [8-10 June 2017]. Lober, D. J., & Eisen, M. D. (1995). The greening of retailing: certification and the home improvement industry. Journal of forestry (USA). McCann, M., Barlow, A., (2015) "Use and measurement of social media for SMEs", Journal of Small Business and Enterprise Development, vol. 22, no. 2, pp.273-287. https://doi.org/10.1108/JSBED-08-2012-0096 Manju, K., Jat, R. K., & Anju, G. (2017). A review on medicinal plants used as a source of anticancer agents, International Journal of Drug Research and Technology, vol.2, no. 2, pp. 6. Merlo, M., & Croitoru, L. (2005). Valuing Mediterranean forests: towards total economic value. Cabi Publishing. https://doi.org/10.1079/9780851999975.0000 Pattenella, D., Ciccarese, L., Dragoi, S., Hegedus, A., Hingston, A., Klöhn, S., ... & Thorfinnsson, T. (2006). NWFP&S marketing: Lessons learned from case studies in Europe. Issues affecting enterprise development in the forest sector in Europe. Faculty of Forestry, University of Joensuu, Research notes, vol. 169, pp. 367-403. Piccoli and Ives (2003). Trus t and the Unintended Effects of Behavior Control in Virtual Teams. MIS Quarterly, vol. 27, no. 3, pp. 365. https://doi.org/10.2307/30036538 Powell, A., Piccoli, G. and Ives, B. (2004). Virtual teams. ACM SIGMIS Database, vol. 35, no. 1, pp. 6-36. https://doi.org/10.1145/968464.968467 Shaaban, H. A., & Moawad, S. A. (2017). Chemical composition, nutritional and functional properties of some herbs and spices, Current Science, vol. 3, no. 3, pp. 128-142. Sisak, L., Riedl, M. and Dudik, R. (2016). Non-market non-timber forest products in the Czech Republic—Their socio-economic effects and trends in forest land use, Land Use Policy, vol. 50, pp. 390-398. https://doi.org/10.1016/j.landusepol.2015.10.006 Subedi, B. (2007). Economics of Non-Timber Forest Products and Services, The Initiation, vol. 1, pp. 32-35. Töllinen, A., Karjaluoto, H. 2011.Marketing communication metrics for social media, International Journal of Technology Marketing, vol.6, no.4, pp. 316 – 330. https://doi.org/10.1504/IJTMKT.2011.045911 Tsekouropoulos, G., Andreopoulou, Z., Seretakis, A., Koutroumanidis, T. and Manos, B. (2012). Optimising e- marketing criteria for customer communication in food and drink sector in Greece. International Journal of Business Information Systems, vol.9, no.1, pp.1. https://doi.org/10.1504/IJBIS.2012.044453 Turban E., Outland J., King D., Lee J.K., Liang TP., Turban D.C. (2018) E-Commerce: Regulatory, Ethical, and Social Environments. In: Electronic Commerce 2018. Springer Texts in Business and Economics. Springer, Cham. https://doi.org/10.1007/978-3-319-58715-8_15 Živojinović, I., Nedeljković, J., Stojanovski, V., Japelj, A., Nonić, D., Weiss, G. and Ludvig, A. (2017). Non- timber forest products in transition economies: Innovation cases in selected SEE countries. Forest Policy and Economics, vol. 81, pp. 18-29. https://doi.org/10.1016/j.forpol.2017.04.003

doi: 10.17700/jai.2018.9.1.434 13 Agni Kalfagianni, Zacharoula Andreopoulou: Internet and Market perspectives of Non Wood Forest Products: the case of mushrooms, truffles and herbs of Greek SMEs Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol. 9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

Analysis of logging forest residues as an energy source

Vasileios Filippou1, Ioannis Philippou2, Nikolaos Symeonidis3, Ioannis Eleftheriadis4, Kostas Tsiotas5

I N F O A B S T R A C T Received 19 Dec 2017 Accepted 5 Feb 2018 Available on-line 23 Mar 2018 Forest residue extraction levels in Greece are currently very low, but logging residues Responsible Editor: M. Herdon have the potential to be an important component of the wood energy supply chain. Forests are a major supplier of renewable energy. Exploring the possibilities of utilizing the biomass of logging residues for energy requires analysis and knowledge of its properties. Keywords: In this research work the properties (ash, volatiles, fixed carbon, carbon, hydrogen, forest biomass, logging oxygen, nitrogen and calorific value) of the various constituents of the biomass of fir residues, energy (Abies borisii - regis), pine (Pinus nigra), oak (Quercus frainetto) and beech (Fagus properties, heating value silvatica) logging residues were determined. Bark and ash content increased with decreasing diameter of branches. Ash content was higher in bark than in wood of branches in all species. Ash and nitrogen content was several times higher in bark and in foliage or needles than in wood of branches. Oak branches and twigs of all species had ash and nitrogen content higher than that required by the EN ISO 17225-2 standard for domestic pellets and they should not be used for energy, at least for pellets production. Volatile mater, fixed carbon, carbon and hydrogen content were in the range given by other researchers. Heating value is higher in softwood (fir and pine) than in hardwood (oak and beech) and ranged from 18.72 MJ/kg to 21.00 MJ/kg. In addition, Duncan's multiple range test was used to compare the means between the various constituents in all species.

1. Introduction The growing global energy demand and concerns about the negative effects of growing greenhouse gas (GHG) emissions from fossil fuels call for alternative energy sources, which are low cost, renewable and non-polluting. One such renewable resource is biomass and especially forest biomass (European Commission 2005, Smeets and Faaij 2007, Becker et al. 2011). The remaining biomass in the forests after logging has attracted great interest as an energy source (Malinen et al. 2001, Gan and Smith 2006, Hu, Heitmann & Rojas 2008, Eker, Çoban & Acar 2009, Bouriaud, Ştefan & Flocea 2013, Järvinen and Agar 2014, Filippou et al. 2015a, Filippou et al. 2017a). Forest biomass consists of tops, branches, bark, foliage or needles and stumps (Giuntoli et al. 2015).

1 Vasileios Filippou Aristotle University of Thessaloniki [email protected] 2 Ioannis Philippou Aristotle University of Thessaloniki [email protected] 3 Nikolaos Symeonidis Aristotle University of Thessaloniki [email protected] 4 Ioannis Eleftheriadis Centre for Renewable Energy Sources and savings, Athens [email protected] 5 Kostas Tsiotas Centre for Renewable Energy Sources and savings, Athens [email protected]

doi: 10.17700/jai.2018.9.1.431 14 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

In the past, logging residues were not exploited mainly because their harvest and transport was technically difficult and uneconomic. Currently new harvesting technologies and transportation systems have been developed and in conjunction with the increase in petroleum prices enable their extraction from the forest (Svanaes and Jungmeier 2010). Also, new and more efficient technologies enable conversion of biomass into energy in small units (mainly gasification) or conversion into compressed forms (wood pellets) that can be installed in or near the forests (Kauriinoja and Huuhtanen 2010, Filippou and Philippou 2014). These, further limit the transportation costs and give opportunities for local employment and rural development. Thus, logging residues from final harvest are expected to play an important role in meeting renewable energy goals in many countries (Gan and Smith 2006, Aguilar 2014). Their utilization for energy production could create business opportunities and employments in local populations, generate profit from residual material and provide energy self- sufficiency for rural communities (Philippou 2014). On the other hand, the literature on removal of logging residues from the forest and their exploitation also addresses various environmental and ecological issues and puts some constrains (Hesselink 2010, Abbas et al. 2011, Wall and Hytönen 2011). The main material properties that affect conversion into energy as well the overall energy outcome are moisture content, ash content, volatile content, elemental composition, chemical components and calorific value (Obernberger, Brunner & Bärnthaler 2006, Vassilev et al. 2010, Filippou et al. 2017b). Biomass of logging residues, differ in chemical composition from stem wood (Zeng, Tang & Xiao 2014). There also exists variability in chemical composition between the various constituents of forest biomass (Werkelin et al. 2007, Wang and Dibdiakova 2014). Ash content of biomass is known to vary between tree species and tree components (Rhén 2004, Filippou, Philippou & Sideras 2015b, Hytönen and Nurmi 2015). High ash content can decrease the heating value of biomass. In addition, ash content and its composition affect the proper functioning of the burners and gasifiers (Bryers 1996, Nielsen et al. 2000). The ash adheres to the heat transfer surfaces and cause corrosion. During combustion burning, the elements, mainly K, Na, S and Ca can melt, form sticky particles, adhere to the surfaces of the walls and create a burner malfunction (Raask 1969, Filbakk et al. 2011). The nitrogen content of biofuel is responsible for the formation of nitrogen oxides (NOx) which have an environmental impact (Munalula and Meincken 2009). Biomass pellets, should have low ash and nitrogen content in order to meet quality standards requirements (EN ISO 17225-2:2015). Calorific value of biomass is a function of its chemical composition (Shafizadeh, Chin & DeGroot 1977, McKendry 2002). Various researchers have determined the calorific value of various types of biomass from their elemental composition using proximity regression analysis models (Demirbaş 2003, Friedl et al. 2005, Telmo, Lousada & Moreira 2010, Singh, Singh & Gill 2015). Also, several researchers (Harris 1984, Nurmi 1997, Zeng, Tang & Xiao 2014, Singh, Singh & Gill 2015) have measured the heating value of various tree species and various tree components and found significant differences, both between species and tree biomass components. The aim of this study is to examine the branches of fir, pine, oak and beech that remain in the forest after harvesting as an energy source, while also to determine their properties that influence the efficient conversion into energy. The properties studied included percentage of bark, ash, volatiles, fixed carbon, carbon, hydrogen, oxygen, nitrogen and calorific value. This work is a part of a research program, aiming at exploring the possibilities for the utilization of residues that remain in the forest after logging, for the production of solid biofuels. Within the framework of the program 5 research papers have been published (Filippou and Philippou 2014, Filippou et al. 2015a, Filippou, Philippou & Sideras 2015b, Filippou et al. 2017a, Filippou et al. 2017b).

2. Material and Methods Logging residues of fir (Abies borisii – regis), pine (Pinus nigra), oak (Quercus frainetto) and beech (Fagus silvatica) were taken from forest stands, in northern Greece, during normal logging operations. Representative samples of branches with bark and foliage were taken from five trees of each species. For determining the percentage of bark in branches, transverse discs of different doi: 10.17700/jai.2018.9.1.431 15 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

diameters (from 2 to 9cm) were cut. The percentage of bark was calculated by measuring the diameter of the disc with the bark and after peeling the bark using the formula (1):

(1)

Where: d1=disk diameter with bark d2= disk diameter without bark The branch samples were in total 127 and had diameter from 2 to 9 cm. Measurements were carried out in 4 discs (repetitions) of each branch. For the determination of other properties, the branches were cut into three parts: thick branches (diameter > 5cm), thin branches (diameter of 2 - 5cm) and twigs (branches with a diameter < 2cm including the foliage or needles). Samples of thick and thin branches were debarked in order wood and bark to be tested separately. The samples were air-dried and milled first in a common hammer mill and then in a Willey mill to obtain particle size < 0.420mm (40 mesh). Ash content (% dry weight), the percentage of volatiles, fixed carbon, and the elemental analysis (C, H, N) were determined in accordance with CEN/TS 14775, CEN/TS 15148 and CEN/TS 15104 standards, respectively. The higher heating value (MJ/Kg dry) was determined in accordance with CEN/TS 14918 standard. Three samples of each material were used for the measurements of each property. Afterwards, Variable Analysis between different materials for each property in all species was executed. The Duncan's multiple range test was used to compare the means of chemical components between the various constituents in all species.

3. Results and Discussion

3.1 Bark percentage Table 1 shows the average bark percentage of all branches measured, as well as the average percentage in each of the two class sizes of branches of fir, pine, oak and beech. Bark percentage ranged from 3.55% to 7.85% and it was higher in thin than in thick branches and varied between the species in a descending order: oak>fir>beech>pine. The differences in the percentage of bark between species was more evident when it was calculated at the same branch diameter (d = 5cm) for all species.

Table 1. Bark percentage of branches Branches d= 2-9cm d= 2-5cm d= >5cm d= 5cm Species đ* % đ % % % bark đ a/a a/a** bark a/a bark bark++ 6.04* 6.87 4.01 7.85 8.07 6.05 Fir 6.87 32** (1.19)+ 16 (0.83) 16 (0.41) 5.91 4.09 4.05 4.52 7.97 3.61 Pine 4.49 32 (0.71) 17 (0.83) 15 (0.55) 5.86 6.89 4.02 7.85 7.97 5.85 Oak 7.30 32 (1.32) 16 (1.01) 16 (0.52) 5.93 4.33 4.25 5.00 7.88 3.55 Beech 3.93 31 (1.19) 16 (1.27) 15 (0.22) * Average diameter, **No of samples, +standard deviation, ++ calculated

doi: 10.17700/jai.2018.9.1.431 16 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

3.2 Proximate analysis In Table 2, the ash content, volatile content (VC) and fixed carbon (FC) content of all branches and twigs is shown, as well as the wood and bark of thick (d= >5cm) and thin (d= 2 - 5cm) branches of fir, pine, oak and beech. According to Duncan's multiple range test, the constituents in all species were categorized depending on their residues fraction. Tables 2, 3 and 4 illustrate the results, where each column is followed by figures in different letters (S) that indicate the significant difference by Duncan's multiple range test (P<0.05).

3.3 Ash content Ash content varied between the various parts of branches and between the species from 0.38% in the wood of thick branches of pine to 8.01% in the bark of thin branches of oak. It was multiple higher in bark than in wood of branches and higher in thin than in thick branches. It is obvious that ash content increases with decreasing branch diameter (see also Table 1). Twigs in all species had lower ash content than bark but higher than thick and thin branches. The ash content of the thick and thin branches (wood and bark) was in fir 0.76 to 0.86%, in pine 0.79 to 1.16%, in oak 2.53 to 3.81% and in beech 0.98 to 1.14%, respectively. The above variation of the ash content between the various materials is shown better in Figure 1. Werkelin et al. (2007) found big differences in ash content between wood and bark in branches of spruce, pine, poplar and birch. Zeng, Tang & Xiao (2014) also found significant differences in ash content between different parts of Masson pine trees. Dibdiakova, Wang, & Li (2015) measured ash content of different parts of scots pine tree and found that branch base has ash content of 0.48% and the branch twigs about 1.56%. The EN ISO 17225-2:2015 standard for domestic pellets requires ash content less than 0.7% for A1 class, less than 1.2% for A2 class and less than 2% for B class pellets, thus oak branches and twigs of all species do not meet the above standard requirements and they should not be used alone for pellet production.

Table 2. Proximate analysis of logging residues Property Thick branches Thin branches Twigs (%) All* Wood Bark All* Wood Bark Fir 0.76 0.52 4.50 0.86 0.50 5.00 3.00 Ash ±0.01 ±0.04 ±0.07 ±0.03 ±0.02 ±0.06 ±0.05 78.75d 79.04e 75.24b 78.76d 79.10e 74.16a 76.84c VC ±0.205 ±0.482 ±0.178 ±0.220 ±0.376 ±0.045 ±0.050 20.49b 20.44b 20.26b 20.38b 20.40b 20.84b 20.16a F C ±0.142 ±0.088 ±0.294 ±0.106 ±0.446 ±0.177 ±0.170 Pine 0.79 0.38 2.68 1.16 0.42 3.06 2.27 Ash ±0.01 ±0.03 ±0.04 ±0.04 ±0.02 ±0.02 ±0.06 77.42c 81.24e 73.23b 77.92d 80.17e 72.80a 76.79c VC ±0.332 ±0.120 ±0.455 ±0.630 ±0.105 ±0.390 ±0.191 21.79d 18.38a 24.09e 20.92c 19.41b 24.14 e 20.94c F C ±0.280 ±0.399 ±0.275 ±0.387 ±0.236 ±0.391 ±0.211 Oak 2.53 1.20 6.96 3.81 1.33 8.01 4.14 Ash ±0.04 ±0.03 ±0.06 ±0.64 ±0.42 ±0.49 ±0.69 79.96d 80.32d 75.72a 78.84c 80.83d 75.26a 77.46b VC ±0.187 ±0.520 ±0.177 ±0.386 ±0.435 ±0.859 ±0.501 17.51b 18.57d 17.32a 17.35b 17.84b 16.73a 18.40c F C ±0.310 ±0.433 ±0.463 ±0.455 ±0.196 ±0.337 ±0.503 Beech 0.98 0.80 6.00 1.14 0.90 6.50 5.50 Ash ±0.02 ±0.02 ±0.08 ±0.02 ±0.04 ±0.70 ±0.06 doi: 10.17700/jai.2018.9.1.431 17 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

80.73c 79.51c 76.09b 80.32c 81.13d 77.01b 75.67a VC ±0.050 ±0.190 ±0.156 ±0.390 ±0.320 ±0.230 ±0.190 18.29b 19.69d 17.91a 18.63e 18.02b 16.49a 18.83c F C ±0.263 ±0.194 ±0.155 ±0.407 ±0.185 ±0.113 ±0.140 *All branches (wood and bark at average diameter from Table 1). In each column, figures followed by different letters (S) indicate significant difference by the Duncan's multiple range test (P<0.05).

Figure 1. Ash content in the various types of logging residues

3.4 Volatile and Fixed Carbon content Volatile content (VC) in fir ranged between 74.16% in the bark of thin branches and 79.10% in the wood of thin branches, in pine between 72.80% in the bark of thin branches and 81.24% in the wood of thick branches, in oak between 75.26% in the bark of thin branches and 80.83% in the wood of thin branches and in beech between 75.67% in twigs and 81.13% in the wood of thin branches. In all species, VC of wood was higher than in bark. Fixed carbon (FC) ranged in fir between 20.16% in twigs and 20.84% in the bark of thin branches, in pine between 18.38% in the wood of thick branches and 24.14% in the bark of thin branches, in oak between 16.73% in the bark of thin branches and 18.57% in the wood of thick branches and in beech between 16.49% in the bark of thin branches and 19.69% in the wood of thick branches. FC in all parts of pine was higher in bark than in wood while in oak and beech was higher in wood. In a study (Telmo, Lousada & Moreira 2010) of proximate analysis in 13 wood species, VC varied among the species between 74.7% and 87.1% and FC between 12.4% and 22.5%. In the same study VC and FC of oak wood was 81.70% and 18.00%, of pine 85.80% and 14.10% and of poplar 87.1% and 12.4%, respectively. Demirbas (1997) gives for beech wood 74.00% volatiles content and 24.6% fixed carbon content.

3.5 Ultimate analysis Table 3 gives the carbon, hydrogen, oxygen and nitrogen content of the all branches and twigs, as well as the wood and bark of thick (d=>5cm) and thin (d=2 - 5cm) branches of fir, pine, oak and beech.

Table 3. Ultimate analysis of logging residues Sample Thick branches Thin branches Twigs (%) All* wood bark All* wood bark Fir 49.44c 49.84d 48.69b 48.96b 48.26b 47.87a 49.02b C ±0.705 ±1.397 ±0.428 ±0.262 ±0.266 ±0.269 ±0.160 doi: 10.17700/jai.2018.9.1.431 18 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

6.45b 6.62b 6.14a 6.44b 6.45b 6.39b 6.58b H ±0.146 ±0.131 ±0.030 ±0.158 ±0.135 ±0.227 ±0.036 O 44.01 43.45 44.14 44.42 45.20 44.89 43.89 0.10a 0.09a 0,91e 0.18b 0.09a 0.85d 0.51c N ±0.015 ±0.005 ±0.185 ±0.001 ±0.001 ±0.028 ±0.045 Pine 50.60c 49.92b 50.73c 49.92b 49.02a 49.75b 50.00b C ±0.325 ±0.117 ±0.340 ± 0.272 ±0.125 ±0.310 ±0.480 6.19b 6.81d 6.10a 6.52b 6.64 c 6.36 b 6.58c H ±0.075 ±0.020 ±0.045 ±0.144 ±0.123 ±0.110 ±0.040 O 43.08 43.20 42.68 43.36 44.21 43.48 42.66 0.13b 0.07a 0.49d 0.20c 0.13b 0.41d 0.76e N ±0.011 ±0.005 ±0.042 ±0.020 ±0.026 ±0.020 ±0.020 Oak 46.23b 46.92b 45.12a 46.90b 46.50b 45.22a 47.61c C ±0.519 ±0.285 ±0.254 ±0.577 ±0.345 ±0.143 ±0.238 6.06b 6.35c 6.33c 6.09b 6.27c 5.78a 6.22b H ±0.102 ±0.051 ±0.051 ±0.015 ±0.090 ±0.119 ±0.040 O 47.42 46.52 48.33 46.70 47.01 48.59 45.00 0.29b 0.21a 0.22a 0.31b 0.22a 0.41c 1.17d N ±0.020 ±0.010 ±0.010 ±0.040 ±0.020 ±0.010 ±0.060 Beech 46.85b 47.02c 45.88a 47.62d 49.73f 46.75b 49.09e C ±0.177 ±0.794 ±0.551 ±0.276 ±0.211 ±0.529 ±0.265 6.25b 6.41c 5.95a 6.27b 6.47c 5.97a 6.68d H ±0.058 ±0.015 ±0.152 ±0.148 ±0.035 ±0.079 ±0.055 O 46.65 46.40 47.59 45.85 43.64 46.89 42.66 0.25b 0.17a 0.58d 0.26b 0.16a 0.39c 1.57e N ±0.010 ±0.005 ±0.061 ±0.032 ±0.001 ±0.017 ±0.030 *All branches (wood and bark at average diameter from Table 1). Ιn each column, figures followed by different letters (S) indicate significant difference by Duncan's multiple range test (P<0.05). Carbon content varied in fir between 47.87% in the bark of thin branches and 49.84% in the wood of thick branches, in pine between 49.02% in the wood of thin branches and 50.73% in the bark of thick branches, in oak between 45.12% in the bark of thick branches and 47.61% in twigs and in beech between 45.88% in the bark of thick branches and 49.73% in the wood of thin branches. In fir and pine thick branches had higher carbon content while in oak and beech carbon content was higher in thin than in thick branches. With the exception in pine, wood had higher carbon content than bark, in all species. Ragland, Aerts & Baker (1991) noticed that the carbon content of softwood species varies between 50 and 53%, and that of hardwood species between 47 and 50% mainly due to the varying lignin and extractives content. They also give 52.25% and 54.90% carbon content in oak and pine bark, respectively. Hydrogen content varied in fir between 6.14% in the bark of thick branches and 6.62% in the wood of thick branches, in pine between 6.10% in the bark of thick branches and 6.81%, in the wood of thick branches, in oak between 5.78% in the bark of thin branches and 6.35% in the wood of thick branches and in beech between 5.95% in the bark of thick branches and 6.68% in twigs. With the exception of pine, there is no difference between thick and thin branches (wood and bark) in all species. Hydrogen content was in a descending order: pine>fir>beech>oak. Nurmi (1993) gives for trembling aspen > 5mm branch wood 46.84% carbon content and 5.96% hydrogen content while for the bark of the branch 48.05% carbon content and 5.77% hydrogen content. In the same study gives for scots pine branch wood 53.53% carbon content and 6.03% hydrogen content while for branch bark 54.99% carbon content and 6.7% hydrogen content. Wilén, Moilanen & Kurkela (1996) gives for scots pine logging residues 51.30% carbon content and 6.10% hydrogen content. doi: 10.17700/jai.2018.9.1.431 19 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Oxygen content was determined by subtracting C, H, N and ash content from the whole mass (100%). In fir oxygen content varied from 43.45% in the wood of thick branches to 45.20% in the wood of thin branches, in pine from 42.66% in twigs to 44.21 % in the wood of thin branches, in oak from 45.00% in twigs to 48.59% in the bark of thin branches and in beech from 42.66% in twigs to 47.59% in the bark of thick branches. Oxygen content was in a descending order: oak>beech>fir>pine. Oxygen content was lower than carbon content in all fir and pine biomass components. Nitrogen content in fir varied from 0.09% in the wood of thick and thin branches to 0.91% in the bark of thick branches, in pine from 0.07% in the wood of thick branches to 0.76% in twigs, in oak from 0.21 in the wood of thick branches to 1.17% in twigs and in beech from 0.16 in the wood of thin branches to 1.57% in twigs. In all cases, bark had 2-3 times higher nitrogen content than wood. It is worth mentioning in nitrogen content which is a determinant factor classification of biomass in qualities according to EN ISO 17225-2:2015 standard. According to Duncan's multiple range test it appears to be a grouping trend in softwood species (fir and pine) which have lower nitrogen content in thick than in thin branches and lower in wood than in bark. In hardwood (oak and beech) species there are no differences in thick and thin branches, while nitrogen content is higher in bark than in wood. With the exception of fir, nitrogen content was higher in twigs in all species. The above variation in nitrogen content is shown better in Figure 2. Dzurenda (2013) gives 0.36%, 0.65% and 0.46% nitrogen content for populous branch wood, branch bark and branch chip. Alakangas (2005) gives 0.30% nitrogen content for scots pine whole trees and 0.40% for pine logging residues. Oak and beech twigs have higher nitrogen content than the EN ISO 17225-2:2015 standard for domestic pellets and should not be used alone for energy uses, at least for pellet production.

Figure 2. Nitrogen content in the various types of logging residues

3.6 Heating value Table 4 shows the heating value of the various branch components of fir, pine, oak and beech. The higher heating value (HHV) ranged in fir between 19.57 MJ/Kg in twigs and 20.80 MJ/Kg in the wood of thin branches, in pine between 20.75 MJ/Kg in the wood of thin branches and 21.00 MJ/Kg in the bark of thick branches, in oak between 18.72 MJ/Kg in the bark of thick branches to 19.52 MJ/kg in the wood of thin branches and in beech between 19.31 MJ/Kg in the bark of thin branches to 19.67 MJ/Kg in the wood of thick branches. With the exception of pine HHV was higher in wood than in bark in all species. According to Duncan's multiple range test, with the exception of pine, all species had higher heating value in thin than in thick branches and higher in wood than in bark. The above variation in HHV between the various types of residues is shown better in Figure 3. doi: 10.17700/jai.2018.9.1.431 20 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Table 4. Heating values (MJ/kg) of logging residues Property Thick branches Thin branches Twigs (MJ/kg) All* Wood Bark All* Wood Bark Fir 20.57d 20.61d 20.48b 20.79e 20.80e 20.32b 19.57a HHV2 ±0.04 ±0.03 ±0.01 ±0.02 ±0.02 ±0.025 ±0.015 Pine 20.95c 20.84b 21.00d 20.80a 20.75a 20.84b 20.95c HHV ±0.020 ±0.015 ±0.020 ±0.045 ±0.025 ±0.025 ±0.015 Oak 19.26c 19.15b 18.72a 19.31e 19.52f 19.13b 19.30c HHV ±0.050 ±0.110 ±0.090 ±0.060 ±0.085 ±0.090 ±0.122 Beech 19.58b 19.67c 19.41a 19.63b 19.70c 19.31a 19.42a HHV ±0.020 ±0.040 ±0.020 ±0.025 ±0.015 ±0.04 ±0.02 *All branch (wood and bark at average diameter from Table 1), 2HHV= higher heat value. In each column, figures followed by different letters (S) indicate significant difference by the Duncan's multiple range test (P<0.05). On the average, heating value was in a descending order: pine>fir>beech>oak. Heating value was higher in softwood (fir and pine) than in hardwood (oak and beech). Philippou (1982) found in oak 19.65 MJ/kg and 18.79 MJ/kg, in poplar 19.78 MJ/kg and 19.62 MJ/kg and in pine 20.35 MJ/kg and 21.60 MJ/kg for stem wood and stem bark, respectively. Järvinen and Agar (2014) making pellets from pine and logging residues found that the average caloric content for the whole pine tree was 20.80 MJ/kg and for the residues 21.60 MJ/kg. Telmo, Lousada & Moreira (2010) found an average heating value (HHV) of pellets from beech wood 19.14 MJ/kg. Phyllis2- Database (ECN 2012) gives calorific value for beech bark 21.67 MJ/kg and an average calorific value of beech wood measured by 12 researchers 19.08 MJ/kg with a range between 16.48 – 20.51 MJ/kg. Shafizadeh, Chin & DeGroot (1977), Demirbas (1997) and Gillespie et al. (2013) found a correlation between carbon content and calorific value.

Figure 3. Higher heating value in the various types of logging residues

doi: 10.17700/jai.2018.9.1.431 21 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

4. Conclusions This paper assessed the suitability of logging residues as an energy source. The properties of logging residues of fir, pine, oak and beech differ in some properties that are important for energy usages. Bark percentage was higher in thin than in thick branches and varied between the species in a descending order: oak>fir>beech>pine. Bark and ash content increased with decreasing diameter of branches. Ash content was higher in thin than in thick branches and it was higher in bark than in wood in all species. Fixed carbon, carbon and hydrogen content differed between the various constituents of logging residues. Nitrogen content of all branches varied from 0.10% in thick branches of fir to 0.31% in thin branches of oak. In overall nitrogen content was higher in oak and higher in thin branches in all species. Relatively higher values of ash and nitrogen content found in the biomass of logging residues, are not as high as in the biomass of other lignocellulosic materials such as agricultural residues and annual plants which are being used for energy production. Taking into consideration that the recovery of logging residues may be limited by other constrains such as maintaining long-term soil and site productivity, from the above results we could conclude that the branches of fir, pine and beech could be a proper material for energy usages while the branches of oak and the twigs of all species should be left to provide nitrogen and minerals to the forest soil. However, during their processing to energy, attention should be given to the proper servicing of the burners or gasifiers for avoiding accumulation or melting of the ashes. Acknowledgment: This study was supported in part by the Green Fund, Ministry of Environment and Energy, Greece.

References Abbas, D., Current, D., Phillips, M., Rossman, R., Hoganson, H., & Brooks, K. N., 2011. Guidelines for harvesting forest biomass for energy: a synthesis of environmental considerations. Biomass and Bioenergy, 35(11), pp. 4538-4546. https://doi.org/10.1016/j.biombioe.2011.06.029 Aguilar, F. X. (Ed.). 2014. Wood energy in developed economies: Resource management, economics and policy. Routledge. Alakangas, E., 2005. Properties of wood fuels used in Finland–Biosouth-project. Technical Research Centre of Finland Research Report, VTT Processes, Jyväskylä, Finland, 100. Becker, D. R., McCaffrey, S. M., Abbas, D., Halvorsen, K. E., Jakes, P., & Moseley, C., 2011. Conventional wisdoms of woody biomass utilization on federal public lands. Journal of Forestry, 109(4), pp. 208-218. Bouriaud, O., Ştefan, G., & Flocea, M., 2013. Predictive models of forest logging residues in Romanian spruce and beech forests. Biomass and bioenergy, 54, pp. 59-66. https://doi.org/10.1016/j.biombioe.2013.03.022 Bryers, R. W., 1996. Fireside slagging, fouling, and high-temperature corrosion of heat-transfer surface due to impurities in steam-raising fuels. Progress in energy and combustion science, 22(1), pp. 29-120. https://doi.org/10.1016/0360-1285(95)00012-7 CEN/TS 14775, 2005, Solid Biofuels – Method for the Determination of Ash. CEN/TS 14918, 2005. Solid Biofuels – Method for the Determination of Calorific Value. CEN/TS 15104, 2005. Solid Biofuels – Determination of Total Content of Carbon, Hydogen and Nitrogen. CEN/TS 15148, 2005. Solid Biofuels – Method for the Determination of the Content of Volatile Matter. Demirbaş, A., 1997. Calculation of higher heating values of biomass fuels. Fuel, 76(5), pp. 431-434. https://doi.org/10.1016/S0016-2361(97)85520-2 doi: 10.17700/jai.2018.9.1.431 22 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Demirbaş, A., 2003. Relationships between lignin contents and fixed carbon contents of biomass samples. Energy Conversion and Management, 44(9), pp. 1481-1486. https://doi.org/10.1016/S0196-8904(02)00168-1 Dibdiakova, J., Wang, L., & Li, H., 2015. Characterization of ashes from Pinus Sylvestris forest biomass. Energy Procedia, 75, pp. 186-191. https://doi.org/10.1016/j.egypro.2015.07.289 Dzurenda, L., 2013. Energetic properties of green chips made of branches of wood species Populus x euroamericana clone Pannonia grown on plantations. Annals of Warsaw University of Life Sciences-SGGW. Forestry and Wood Technology, 82. Eker, M., Çoban, H. O., & Acar, H. H., 2009. The procurement problem of logging residues in Turkish forestry. Proceedings in FORMEC, 9, pp. 115-124. EN ISO 17225-2:2015. Solid biofuels ― Fuel specifications and classes. Part 2: Graded wood pellets. European Committee for Standardization. European Commission, 2005. Biomass. Green Energy for Europe. Directorate-General for Research, Sustainable Energy Systems, EUR 21350. Filbakk, T., Jirjis, R., Nurmi, J., & Høibø, O., 2011. The effect of bark content on quality parameters of Scots pine (Pinus sylvestris L.) pellets. Biomass and Bioenergy, 35(8), pp. 3342-3349. https://doi.org/10.1016/j.biombioe.2010.09.011 Filippou, V. and Philippou, I. L., 2014. Harvesting methods of logging residues. Proceedings of 10th National Congress on Renewable Energy. 26-28 November, Thessaloniki Greece, pp 287-297. Filippou, V., Tsanaxidou, Z., Sideras, A., & Philippou, I. L., 2015a. Weight characteristics of logging residues of fir and beech as an energy source. Pro Ligno, 11(4), pp. 663-672. Filippou, V., Philippou, I. L. & Sideras, A., 2015b. Utilization of logging residues for production of biofuels: I. Percentage (%) of bark and ash of branches of fir, pine, beech, poplar and oak. Conference Proceedings, 17th international forestry conference, Nov. 4-7, Kefalonia Greece, pp. 809-819. Filippou V., Philippou L., Eleftheriadis I., Tsiotas K., 2017a. Energy and other properties of fir and beech logging residues. Conference Proceedings, International Workshop, Information Technology, Sustainable Development, Scientific Network and Nature Protection, Okt. 8-11, Edessa Greece, pp. 1340-1348. Filippou V., Philippou I., Eleftheriadis I., Tsiotas K., 2017b. Fuel properties of oak, poplar and pine logging residues. Proligno. 13(4), pp. 570-578. Friedl, A., Padouvas, E., Rotter, H., & Varmuza, K., 2005. Prediction of heating values of biomass fuel from elemental composition. Analytica Chimica Acta, 544(1), pp. 191-198. https://doi.org/10.1016/j.aca.2005.01.041 Gan, J., & Smith, C. T., 2006. Availability of logging residues and potential for electricity production and carbon displacement in the USA. Biomass and Bioenergy, 30(12), pp. 1011-1020. https://doi.org/10.1016/j.biombioe.2005.12.013 Gillespie, G. D., Everard, C. D., Fagan, C. C., & McDonnell, K. P., 2013. Prediction of quality parameters of biomass pellets from proximate and ultimate analysis. Fuel, 111, pp. 771-777. https://doi.org/10.1016/j.fuel.2013.05.002 Giuntoli, J., Caserini, S., Marelli, L., Baxter, D., & Agostini, A., 2015. Domestic heating from forest logging residues: environmental risks and benefits. Journal of Cleaner Production, 99, pp. 206-216. https://doi.org/10.1016/j.jclepro.2015.03.025 Harris, R. A., 1984. Fuel values of stems and branches in white oak, yellow-poplar, and sweetgum. Forest products journal, 34(3), pp. 25-26. Hesselink, T. P., 2010. Increasing pressures to use forest biomass: A conservation viewpoint. The Forestry Chronicle, 86(1), pp. 28-35. https://doi.org/10.5558/tfc86028-1 Hu, G., Heitmann, J. A., & Rojas, O. J., 2008. Feedstock pretreatment strategies for producing ethanol from wood, bark, and forest residues. BioResources, 3(1), pp. 270-294. Hytönen, J., & Nurmi, J., 2015. Heating value and ash content of intensively managed stands. Wood Research, 60(1), pp. 71-82.

doi: 10.17700/jai.2018.9.1.431 23 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Järvinen, T., & Agar, D., 2014. Experimentally determined storage and handling properties of fuel pellets made from torrefied whole-tree pine chips, logging residues and beech stem wood. Fuel, 129, pp. 330-339. https://doi.org/10.1016/j.fuel.2014.03.057 Kauriinoja, A., & Huuhtanen, M., 2010. Small-scale biomass-to-energy solutions for northern periphery areas (Doctoral dissertation, MSc Thesis, Oulu). Malinen, J., Pesonen, M., Määttä, T., & Kajanus, M., 2001. Potential harvest for wood fuels (energy wood) from logging residues and first thinnings in Southern Finland. Biomass and Bioenergy, 20(3), pp. 189-196. https://doi.org/10.1016/S0961-9534(00)00075-1 McKendry, P., 2002. Energy production from biomass (part 1): overview of biomass. Bioresource technology, 83(1), pp. 37-46. https://doi.org/10.1016/S0960-8524(01)00118-3 Munalula, F., & Meincken, M., 2009. An evaluation of South African fuelwood with regards to calorific value and environmental impact. Biomass and bioenergy, 33(3), pp. 415-420. https://doi.org/10.1016/j.biombioe.2008.08.011 Nielsen, H. P., Baxter, L. L., Sclippab, G., Morey, C., Frandsen, F. J., & Dam-Johansen, K., 2000. Deposition of potassium salts on heat transfer surfaces in straw-fired boilers: a pilot-scale study. Fuel, 79(2), pp. 131-139. https://doi.org/10.1016/S0016-2361(99)00090-3 Nurmi, J., 1993. Heating values of the above ground biomass of small-sized trees. Acta Forestalia Fennica 236. The Finish Research Institute. Nurmi, J., 1997. Heating values of mature trees. The Society of Forestry in Finland. Obernberger, I., Brunner, T., & Bärnthaler, G., 2006. Chemical properties of solid biofuels—significance and impact. Biomass and Bioenergy, 30(11), pp. 973-982. https://doi.org/10.1016/j.biombioe.2006.06.011 Philippou, I., 1982. Calorific value of various types of forest biomass. Proceedings of 1th National Congress on Renewable Energy. Thessaloniki. Philippou, I., 2014. Utilization of forest biomass for energy production. Environmental and rural development contribution. Proceedings of 10th National Congress on Renewable Energy. 26-28 November, Thessaloniki. pp. 273-287. Raask, E., 1969. Tube erosion by ash impaction. Wear, 13(4-5), pp. 301-315. https://doi.org/10.1016/0043- 1648(69)90252-X Ragland, K. W., Aerts, D. J., & Baker, A. J., 1991. Properties of wood for combustion analysis. Bioresource technology, 37(2), pp. 161-168. https://doi.org/10.1016/0960-8524(91)90205-X Rhén, C., 2004. Chemical composition and gross calorific value of the above-ground biomass components of young Picea abies. Scandinavian journal of forest research, 19(1), pp. 72-81. https://doi.org/10.1080/02827580310019185 Shafizadeh, F., Chin, P. P., & DeGroot, W. F., 1977. Effective heat content of green forest fuels. Forest Science, 23(1), pp. 81-89. Singh, M. Singh, R. and G. Gill. 2015. Estimating the correlation between the caloric value and elemental components of biomass using regression analysis. Proceedings of 33th IRF International Conference, July 26, 2015, Pune, India, ISBN: 97893-85465-62-8. Smeets, E. M., & Faaij, A. P., 2007. Bioenergy potentials from forestry in 2050. Climatic Change, 81(3-4), pp. 353-390. https://doi.org/10.1007/s10584-006-9163-x Svanaes, J., & Jungmeier, G., 2010. Energy generation techniques. A state of the art overview of available techniques for energy generation through direct combustion of recovered wood. COST. Norsk Treteknisk Institutt. Telmo, C., Lousada, J., & Moreira, N., 2010. Proximate analysis, backwards stepwise regression between gross calorific value, ultimate and chemical analysis of wood. Bioresource technology, 101(11), PP. 3808-3815. https://doi.org/10.1016/j.biortech.2010.01.021 Vassilev, S., & Baxter, D., Lars K Andersen, L., Vassileva, C., 2010. An overview of the chemical composition of biomass. Fuel 89: 913–933. https://doi.org/10.1016/j.fuel.2009.10.022

doi: 10.17700/jai.2018.9.1.431 24 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:14-25

Wall, A., & Hytönen, J., 2011. The long-term effects of logging residue removal on forest floor nutrient capital, foliar chemistry and growth of a Norway spruce stand. Biomass and Bioenergy, 35(8), PP. 3328-3334. https://doi.org/10.1016/j.biombioe.2010.08.063 Wang, L., & Dibdiakova, J., 2014. Characterization of ashes from different wood parts of Norway spruce tree. Chemical engineering transactions, 37, PP. 37-42. Werkelin, J., Johan, B., Skrifvars, B. J., & Hupa, M., 2007. Ash-forming elements in four Scandinavian wood species. Part 1: Summer harvest. Biomass and Bioenergy (29): PP. 451-466. Wilén, C., Moilanen, A., & Kurkela, E., 1996. Biomass feedstock analyses. VTT Publications 282, Espoo. Finland. Zeng, W. S., Tang, S. Z., & Xiao, Q. H., 2014. Calorific values and ash contents of different parts of Masson pine trees in southern China. Journal of Forestry Research, 25(4), PP. 779-786. https://doi.org/10.1007/s11676- 014-0525-3

doi: 10.17700/jai.2018.9.1.431 25 Vasileios Filippou, Ioannis Philippou, Nikolaos Symeonidis, Ioannis Eleftheriadis, Kostas Tsiotas: Analysis of logging forest residues as an energy source Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol.9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece

Aristotelis Martinis1, Evgenia Chaideftou2, Charikleia Minotou3, Κonstantinos Poirazidis4

I N F O A B S T R A C T Received 5 Dec 2017 Accepted 31 Jan 2018 The current study aims to analyze the ecological niche and mapping the distribution of Available on-line 23 Mar 2018 species belonging to Orchidaceae family, in Zante Island and to determine hotspots at Responsible Editor: M. Herdon sites of high species richness. 967 observations were recorded into 110 tracks in 2015 and 120 tracks in 2016, where 47 orchid species were identified belonging to 9 genera. Using Keywords: Maxent, the ecological niche of each species was analyzed and habitat suitability map was Distribution, Suitability map, created using 12 environmental parameters. The suitability maps were transformed into a Altitude, Hydro-lithography, binary format according to the threshold “10th percentile training presence” and the Species richness, Hot-spot transformed maps used to compute hot-spot regions by applying the SMD toolbox using the software ArcGIS. Zante Island is characterized by remarkably high orchid species richness since the synthetic model indicates presence of at least 10 species for the 53.2% of the island area. Topography, geology and landscape openness are determining factors for orchids’ diversity conservation. Altitude is the determining factor of differentiation in species distribution. Hydro-lithology was the following most significant interpreting parameter. The mapping analysis of species abundance units revealed that 12.5% of the island surface could be characterized as hotspots of high value for the conservation of orchid diversity.

1. Introduction Interactions between biotic organisms and environmental variables, such as the relation of heliophyte orchids with light intensity (Naveh 1982), are of crucial importance for ecosystem evolution especially in the Mediterranean landscape. Selecting the proper environmental parameters and collecting field work data can result in the mapping of species distribution and their habitats suitability; such maps depict the predicted spatial distribution of species at wider areas. The combination of environmental parameters and taxa recordings can generate more robust modeling outcomes of species distribution (Elith & Leathwick 2009) principally because proper quantitative methods and tools are valuable for deriving credible conclusions (Tremlová & Münzbergová 2007). Noteworthy is the fact that species occurence modeling can produce multiple maps that are considered suited for floristic analyses because they reveal the existing reliance of species’ presence on biogeochemical requirements (Van der Maaten et al. 2012). The aforementioned are basic tools for

1 Aristotelis Martinis Department of Environmental Technology, T.E.I. of Ionian Islands, Panagoula, 29 100, Zante, Greece [email protected] 2 Evgenia Chaideftou Department of Environmental Technology, T.E.I. of Ionian Islands, Panagoula, 29 100, Zante, Greece [email protected] 3 Charikleia Minotou IFOAM AGRIBIOMEDITERRANEO, Derigny 49, 10434, Athens, Greece [email protected] 4 Κonstantinos Poirazidis Department of Environmental Technology, T.E.I. of Ionian Islands, Panagoula, 29 100, Zante, Greece

protecting species and preserving their diversity in regions with hot spots of high value (Rodríguez et al. 2007). Orchidaceae is one of the prime and most diverse plant families (Tsiftsis et al. 2008) numbering more than 25.000 species of 800 genera worldwide (Popova et al. 2016). Despite its high species number, this family exhibits the highest percentage of extinction rate in the world. In Europe orchid species are reported to have been severely declined the last decades (Kull & Hutchings 2006). In most studies orchids are mainly investigated for their anatomy (Aybeke et al. 2010), medicinal substances and ethnobotanical uses (De Vos 2010), innovations in their breeding (Popova et al. 2016) or their presence is recorded in species lists (Melendo et al. 2003), but less studied are their conservation priorities (Pfeifer et al. 2010). Hence the determination of the ecological niche is considered elementary for the conservation of Orchidaceae species (Tsiftsis et al. 2008). Related issues such as the reduction in the distribution range of orchid species have been investigated for instance in Northern Europe (Kull & Hutchings 2006). It is widely known that the majority of orchids interact with compatible mycorrhizal fungi throughout their lives, due to lack of endosperm in their seeds. Since orchid seed lack an endosperm and consist only of an embryo and a seed coat, this association is obligate during seed germination with the mycorrhizal fungus supplying the critical nutrients for germination and continue to rely on fungi for water and nutrient to some extent at adulthood (Rasmussen 2002; Warcup 1990; Jacquemyn 2016). Mariangela Girlanda and Silvia Perotto (2011), in their research on the of mycorrhizal fungal diversity in four photosynthetic Mediterranean meadow orchids, indicates that meadow orchids may prefer specific fungal partners in natural conditions. In parallel, it is recognized that our knowledge of orchid ecology needs to be enhanced especially in terms of their geological and altitudinal preferences (Hágsater & Dumont 1996), while usually the studies investigating orchid ecological preferences are too few (Tsiftsis et al. 2008, Phillips et al. 2011) and most often such studies are conducted after a decline in the orchid distribution ranges has taken place (e.g. Kull & Hutchings 2006). A major factor resulting in such declines is habitat loss (Fischer & Stöcklin 1997). Consequently, studying the distribution of species’ abundance and how it is affected by geographical and environmental factors, as well as determining their ecological niche, is essential for enhancing the protection planning of the species of Orchidaceae family (Crain et al. 2014). Nowadays the number of threatened species exceeds well our available resources for the protection and conservation of these species, indicating a firm need for prioritizing biodiversity protection with a focus on regions where are detected the earnest protection requirements. A promising approach is to determine 'hotspots' in areas of high endemism threatened by habitat loss and in areas of high species richness or presence of rare taxa (Myers et al. 2000). This may be particularly the case of areas undergoing fragmentation and such areas can be traditionally species-rich areas, entailing habitats that are hotspots, but due to fragmentation (or other threats) their occurring species run the risk of extinction (Piqueray et al. 2011). Species respond differently to threats like fragmentation (Lindborg 2007) and this is a main reason that species preferences are worth-investigating as they can be pivotal in conservation management (Piqueray et al. 2011). The main aims of the study were 1) to analyze the ecological niche and mapping the distribution of species belonging to Orchidaceae family in Zante Island and 2) to determine hotspots at sites of high species richness belonging to Orchidaceae. The determination of ecological factors and the distribution of species as well as the spotting of areas of high values will lead to protection and conservation of orchid. Likewise, this study sets the basis to plan the scientific monitoring of Orchidaceae in the Ionian region.

doi: 10.17700/jai.2018.9.1.430 27 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

2. Methodology 2.1. Study area and floristic recordings The Zante is located in the southwestern of Greece, in the Ionian Sea, between Greece and Italy. It is about 150 km south of Corfu (Kerkira) island and forms with Lefkada, Kefalonia, Ithaki, Paxi, Kythira and many others small islands, the Ionian archipelago (Eptanisa). Zante covers an area of 406 km2 with a population of about 42,000 people (Figure 1). The study area covers all the surface of this small island, in witch the climate is of Mediterranean type, with intense annual precipitation ranging from 800 to 900 mm (average of approximately 850 mm), showing a decreasing trend over time. The wet season starts in September and ends by May (Figure 2). The area is characterized by low and middle altitudes, where the largest altitude is located at the top of the mount Vrachionas and reaches 792 m. As consequence, there is an overlap of vegetation zones and the distribution of species. Thus, a high diversity of habitats is exhibited with a diverse mosaic of varied vegetation types (Pinus halepensis woods, shrubs, meadows, agricultural areas, olive groves, open burnt areas, sand dunes etc.).

Figure 1. Location of Zante Island in the Ionian Sea

Figure 2. Average monthly precipitation and temperature in Zante Island. Temperature in Celsius Degrees (oC) and precipirtation in mm.

doi: 10.17700/jai.2018.9.1.430 28 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

The first recordings of orchids’ distribution in Zante Island took place early in the 19th century (Delforge 1993, Gölz & Reinhard 1995). In our case, a total of 967 recordings were completed between 2015 and 2016, following different tracks with a 2-4 km length and 2-meter belts of either side of each track. Specifically, 110 tracks were trailed in 2015 and 120 tracks were trailed in 2016. All these routes were homogeneously distributed across the habitats of Zante Island (Figure 3). In totally two years of recording, were trailed 230 tracks with an overall length of approximately 650 km.

Figure 3. On the left, recordings in study area (white points: recordings in 2015 and red points: recordings in 2016. The right image shows some indicative recordings of orchids in the mountainous region of Loucha - Gyri (Zante Island - Greece)

Each year (2015 and 2016), orchid recording took place between mid-February and end of May. In sites of high orchid abundance, the recording was repeated (in the beginning and at the end of spring). For each sampling site, a designed protocol was used for the recording of coordinates, observed species, analytical photographical images for species documentation, specific data for vegetation and geo-topographical parameters as well as data and of their conservation status and threats. In particular, the orchid species recording was conducted only using photographs and no orchid individual were sampled for conservation reasons. Orchid species were recorded at all representative biotopes of the island, open and shadowed forest, macquis, phrygana, previously burnt, and wetland ecosystems, with the aim to cover a wide spectrum of environmental parameters so that the suitability models approximate the real species distribution. The orchids were identified based on standard floras (Tutin et al. 1980) and named in line with the binomial concept of Linnaeus (Jarvis, Cribb 2009) following Euro+Med (2006-) and the Flora Ionica Inventory (Flora Ionica Working Group (2016 onwards)) cross-checked with Dimopoulos et al. (2013).

2.2. Data analyses and Individual models of habitat suitability All recordings were introduced into a geographical database and individual models of habitat suitability were developed with the program Maxent (of maximum entropy) (Phillips et al. 2006), using 13 environmental parameters that affect the presence and distribution of Orchidaceae species, in three parameter groups of topography, vegetation and geology (Remm et al. 2008): - The first group includes six parameters of topography: altitude, terrain gradient, effect of the relief exposure (quantitative northbound effect variable and eastbound quantitative variable) and relative topographic location (as a quantitative and qualitative variable). - The second, includes four vegetation indices to characterize the density of the forest area, namely Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), Modified Soil-adjusted Vegetation Index (MSAVI), Enhanced Vegetation Index (EVI) and soil temperature as a surrogate of the openness, recorded by the thermal band of satellite images. doi: 10.17700/jai.2018.9.1.430 29 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

- The third, includes two qualitative geological parameters: lithological and hydro-lithological rock formations, with 10 and 6 classes for the formations correspondingly (Appendix 1).

2.3. Habitat suitability models – Hot spot analysis Habitat suitability models were estimated for each species, using Maxent (Phillips et al. 2006). These models were based on a mean value of 10 replicates, with the method of crossvalidated. The resulted habitat suitability maps, that its scale was firstly continuous between 0 and 1, was transformed into a binary file (presence – absence) using the threshold “10th percentile training presence” to consider the environmental complexity of the region and thus depict the species presence more conservatively, where 90% of the training samples are correctly classified (Figure 4). Considering this threshold, the 10% of excluded presences are potentially represent recording errors, ephemeral populations, or occurrence of unusual microclimatic conditions within an area. The current models for the suitability prediction for species do not require data of verified species absence but they do compare the sites of the species presence with the range of the total change of the niche in the studied area, on the basis of determined factors. However, the statistical models do not recognise the most significant factors that define ecologically the presence of a species. On the other hand, if proper factors are used in the models, it is highly probable that this will result in a proper outcome (for instance a robust distribution of the species presence), though the ecological and management importance of this outcome may only have a trivial meaning. For the above reasons, the selection of the environmental parameters was based on two criteria: a) percentage over a 5% participation in the interpretation of the species distribution, and b) degree of correlation between the selected variables less than 0.8. The best regularization number (at a scale from 0.5 to 3 in increments of 0.5) for each model generality was calculated based on the criterion AICc, and when AICc was the same in 2+ models, then the most proper one was selected, namely the model with the lowest AUC.Diff using statistics in R. Note that the highest generality value corresponding to a spread distribution is the default value of 1 (Phillips and Dudik 2008).

Figure 4. Transformation of the individual suitability map (left) of Anacamptis pyramidalis to a binary file (right). Based on the individual binary models, hot spot analysis was formed by applying the SMD toolbox (Brown et al. 2017) using the software ArcGIS (analysis at scale of 5000-meter squares). Then six spatial units based on species richness were computed via the tool Hot Spot Analysis (Getis-Ord Gi*) and the Euclidean distance.

3. Results and Discussion

3.1. Floristic results and Synthetic maps Totally 47 different orchid species of 9 genera (Ophrys, Orchis, Anacamptis, Serapias, Dactylorhiza, Limodorum, Neotinea, Spiranthes, Himantoglossum) were identified (Appendix 2). Among them, the endemic Serapias neglecta subsp. ionica, Dactylorhiza romana, Ophrys speculum, and the rare Spiranthes spiralis, are indicative of the excessively high species richness of orchids in Zante. Delforge (1993) distinguished 41 species distributed across Zante corresponding to 37 species according to current nomenclature. From these 37 species, 33 were recorded in our study comprising doi: 10.17700/jai.2018.9.1.430 30 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

90% of Delforge database. From the 47 species of our study, 29 of 7 genera were found to have adequate enough data to provide reliable models. Most of the models found to perform well with AUC values more than 0.8 (mean AUC value of all models equal to 0.849, with a standard deviation of 0.07). Combining independent models, a synthetic map of species richness was created (Figure 5a). This map is of particular value for designing and planning the conservation and protection of the Orchidaceae species, demonstrating that at a 53.2% of the island area, the presence of at least 10 orchid species is estimated (Figure 5b). At particular sites (comprising the 4.7% of the island area) it is estimated that at least 20 species up to 27 are present, mainly at the southwest of the Island (Figures 5b).

Figure 5. Synthetic map of species richness of Orchidaceae family (a) and three classes of species richness of Orchidaceae (b) in Zante Island.

3.2. Environmental parameters in orchid distribution analyses (niche analyses) The relation between species abundance and altitude is considered a key factor in floristic studies, as in relatively small geographical distances there are often significant climatic variations that largely determine the biodiversity of a region (Grytnes et al. 2006, Acharya et al. 2011), despite the fact that in many cases and for various taxa, altitude is not a defining factor of species abundance (Tsiftsis et al. 2008, Trigas et al. 2013). Apart from altitude, the range of species dispersion is also associated with other ecological and environmental parameters such as soil pH, aspect, slope, soil temperature, and climatic conditions, mainly precipitation and temperature. Taxa exhibit some degree of physiological tolerance to environmental pressures, while wide variations in species richness are strongly correlated with the climate (Lomolino 2001). Some researchers argue that species’ morphology and abundance are mainly due to an environmental factor, which may be different for each species (McCain 2007, Currie & Kerr 2008). In any case, apart from the altitude, the most important factors seem to be water and energy for many species (O’Brien 1998, Hawkins et al. 2003, Fu et al. 2006). In the particular case of orchids, the species distribution is critically correlated to edaphic and mycorrhizal specialization (Phillips et al. 2011). For rare orchids’ conservation, this implies the importance of microhabitats where present are the suitable endophytes, and fungi are properly distributed within the microhabitats. Hence changes in orchid species composition are not yet known to be driven by changes in endophytes across environmental gradients (Phillips et al. 2011). In the case of Zante Island, despite the small altitudinal differences, altitude is apparently the most determining factor of variation in species distribution, since it was selected as one of the most important variables in 27 individual suitability models out of a total of 29 models. In eleven species the altitude exhibited a bell-shaped adaptation (at 200 meters or at 600 meters as peaks, Figure 6), and in nine species showed an incremental linear adjustment (Figure 7). doi: 10.17700/jai.2018.9.1.430 31 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Figure 6. The impact of altitude, where the peak appears at an altitude of approximately 200 m on Anacamptis pyramidalis (a) and Anacamptis coriophora (b), and the impact of altitude where the peak appears at an altitude of approximately 600 m on Ophrys lutea (c) and Neotinea lactea (d)

Figure 7. The impact of altitude where an incremental linear adjustment is shown on Orchis anthropophora (a), Ophrys ferrum-equinum (b), Ophrys fusca (c), and Ophrys spruneri (d) doi: 10.17700/jai.2018.9.1.430 32 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Topography, geology and landscape openness are determining factors for orchids’ diversity conservation. This is congruent with the findings of Tsiftis et al. (2011) that highlighted the significance of the geological substrate as a surrogate of the soil conditions that inflict on orchid distribution. Hydro-lithology was the following most significant interpreting parameter (17 out of 29 models), with granular deposits dominating as suitable substrate, at south and east aspects of moderate inclination covered by less dense to sparse vegetation. Ophrys lutea exhibits an incremental linear adaptation and the species presence was recorded throughout the altitudinal range of the island. According to recordings, the individuals of O. lutea are more robust and in complete flowering at medium altitudes, while at lower and higher altitudes their flowers are less and smaller. The species exhibits wide distribution also in relation to the geological substrate. Apart from the alluvial deposits, the species is recorded in all other types of geological substrates of the island, while in relation to water permeability, it prefers deposits with moderate, small and very low water permeability (Figure 8).

Figure 8. Geological and hydrological preferences of Ophrys lutea.

Ophrys tenthredinifera was recorded on limestone rocks and marble limestones, while in relation to the altitude it is present along the whole range of the altitude in the island (10-790 m), with robust individuals in full flowering. Anacamptis papilionacea also shows the same tolerance to altitude, while it prefers granular deposits with fluctuating permeability. On the contrary, the distribution of Orchis italica appears to be strongly influenced by the altitude. Orchis italica has robust populations up to altitude of 580 m and beyond a decline is observed in the population and in the individuals’ robustness. Serapias lingua is recorded at large populations in the island, at altitudes of 10-200 m, and besides alluvial deposits, it shows particular tolerance to all other types of rocks and water- permeability. A preference to low-middle altitudes in open areas of the above species has been also reported by Tsiftsis et al. (2008). The erythrocyte deposits, mainly of carst cavities, as well as the limestones and dolomites, form the appropriate geological substrate for Anacamptis pyramidalis (Figure 9a); the limestone, and also marl's limestone, and gypsum substrates are appropriate for Ophrys ferrum-equinum (Figure 9c,d), while Himantoglossum robertianum shows preference to alluvial deposits of low water permeability and to granular deposits of fluctuating water permeability (Figure 9b). While the slope increases, the suitability of the site increases (Figure 9d). Calcareous soils and limestones have been associated with orchid diversity by others (e.g.Tsiftsis et al. 2011) while bedrock reactivity was identified as an important factor for species of Dactylorhiza (Štípková et al. 2017).

doi: 10.17700/jai.2018.9.1.430 33 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Figure 9. Individual models of the effect of abiotic parameters on Anacamptis pyramidalis (a), Himantoglossum robertianum (b) and Ophrys ferrum-equinum (c, d).

In relation to the rest of parameters, the majority of species were recorded at southern and eastern expositions, mainly on medium - and upper - slopes of moderate inclinations with more or less sparse vegetation. The significance of different ecological preferences of species which may be related to the high species richness, has already been identified for Zante Island (Fourth author et al. 2015), suggesting that the landscape heterogeneity is a critical factor in preserving biodiversity. The present study confirmed the previous survey, since the research results of both studies are consistent, and also highlighted the importance of modeling and spatial-analysis that result in useful implications for conservation planners regarding protection of the rarer recorded species.

3.3. Hot Spot analysis Hot Spot analysis revealed that four areas of Zante, which account for a total of 12.5% of the island area, are of great importance for protecting species biodiversity and could be classified as hotspots of high value for the conservation of orchid diversity (Figure 10). It is worth of noticing that in Zante Island have been recorded the rare Ophrys speculum (for the first time at only one site) as well as the rare Dactylorhiza romana, Limodorum abortivum and Spiranthes spiralis the only one species with autumn flowering. Alike, Štípková et al. (2017) also concluded on the usefulness of Maxent tool in determining hotspots of rare orchids in Czech Republick. Emphasis on species extinction, fluctuations and ecological preconditions needs to be enabled (Godefroid 2001). Changes in distributions of orchid species depend largely on local conditions and identification of areas -such as hot spots- on which protection management should focus, is important (Kull & Hutchings 2006) as important is the selection of which management or conservation practices need to be implemented. In our study, environmental parameters were associated with the orchid distribution (niche analyses) for both abundant and rarer orchid species indicating that rare species of limited distribution should not be underestimated in conservation and protection goals (Tsiftsis et al. 2008).

doi: 10.17700/jai.2018.9.1.430 34 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Figure 10. Hot Spot analysis for orchid diversity in Zante Island.

Perspectives and conclusion The results of the present study, that it is the first time it is conducted in Zante Island, confirmed the richness of orchid species in the area, included rare species and recorded the most significant environmental parameters influencing the species distribution; whereas the hot spots of the highest significance for orchid species conservation were detected and mapped. For these species, it is indicated that apart from climatic conditions, the altitude is a determining factor for the majority of orchid species as well as the geological substrate, the slope and the aspect. It was also confirmed that the maps of species distribution which are derived from combining field work with geographical and spatial analysis have the potential to provide safe tools for decision making at spatial scales aiming at biodiversity conservation, through proper prioritization and development of strategies for conservation and protection of sensitive orchid species especially under climate change.

References Acharya, BK, Sanders, NJ, Vijayan, L & Chettri, B 2011. ‘Elevational gradients in bird diversity in the Eastern Himalaya: an evaluation of distribution patterns and their underlying mechanisms’. PLoS One, vol. 6, p. e29097. https://doi.org/10.1016/j.japb.2015.11.002 Aybeke, M, Sezik, E, & Olgun, G 2010. ‘Vegetative anatomy of some Ophrys, Orchis and Dactylorhiza (Orchidaceae) taxa in Trakya region of Turkey’. Flora, vol. 205, pp. 73-89. Brown JL, Bennett JR, & French CM 2017. SDMtoolbox 2.0: the next generation Python-based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses. PeerJ5: e4095; https://doi:10.7717/peerj.4095 Crain BJ, Tremblay RL 2014. Do richness and rarity hotspots really matter for orchid conservation in lightof anticipated habitat loss? Diversity and Distributions, vol. 20, pp. 652-662. https://doi:10.1111/ddi.12179 Currie, DJ, Kerr, JT 2008. ‘Tests of the mid-domain hypothesis: a review of the evidence’. Ecol Monogr., vol. 78, pp. 3-18. https://doi:10.1890/06-1302.1 Delforge, P 1993. Les Orchidées de l'île de Zante (Nomos Zakynthos, Nissia Ioniou, Grèce). Observations et cartographie. Natural. belges 74 (Orchid. 6), pp. 113-172. De Vos, P 2010. ‘European materia medica in historical texts: Longevity of a tradition and implications for future use’. Journal of Ethnopharmacology, vol. 132, pp. 28-47. https://doi:10.1016/j.jep.2010.05.035

doi: 10.17700/jai.2018.9.1.430 35 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Dimopoulos, P, Raus, Th, Bergmeier, E, Constantinidis, Th, Iatrou, G, Kokkini, S, Strid, A, Tzanoudakis, D 2013. Vascular plants of Greece: An annotated checklist. Botanischer Garten und Botanisches Museum Berlin- Dahlem, Berlin; Hellenic Botanical Society, Athens. 372 pp. [Englera 31] Elith, J, Leathwick, JR 2009. ‘Species distribution models: ecological explanation and prediction across space and time’. Annual Review of Ecology, Evolution, and Systematics, vol. 40, pp. 677-697. Euro+Med 2006-. Euro+Med PlantBase - the information resource for Euro-Mediterranean plant diversity. Published on the Internet http://ww2.bgbm.org/EuroPlusMed/ [accessed 21/01/2018]. Fischer, M, Stöcklin, J 1997. ‘Local extinctions of plants in remnants of extensively used calcareous grasslands 1950–1985’. Conservation Biology, vol. 11, pp. 727-737. https://doi:10.1046/j.1523-1739.1997.96082.x Flora Ionica Working Group (2016 onwards). Flora Ionica – An inventory of ferns and flowering plants of the Ionian Islands (Greece). Published at https://floraionica.univie.ac.at; accessed [21/01/2018] Fu, C, Hua, X, Li, J, Chang, Z, Pu, Z & Chen, J 2006. ‘Elevational patterns of frog species richness and endemic richness in the Hengduan Mountains, China: geometric constraints, area and climate effects’. Ecography, vol. 29, pp. 919-927. https://doi:10.1111/j.2006.0906-7590.04802.x Girlanda, M, Segreto, R, Donata, C, Liebel, H, T, Rodda, M, Ercple, E, Cozzolino, S, Gephard, G, and Perotto S 2011. Photosynthetic Mediterranean meadow orchids feature oartial mycoheterotrophy and specific mycorrhizal associations, American Journal of Botany, vol. 98(7), pp.1148–1163. https://doi.org/10.3732/ajb.1000486 Godefroid, S 2001. ‘Temporal analysis of the Brussels Flora as indicator for changing environmental quality’. Landscape and Urban Planning, vol. 52, pp. 203-224. https://doi.org/10.1016/S0169-2046(00)00117-1 Gölz, P, Reinhard, H 1995. Die Orchideenflora der ionischen Inseln Kefallinia und Zakinthos: Neue Beobachtungen und Erkenntnisse. Jour. Eur. Orch., vol. 27, issue 4, pp. 555-621. Grytnes, JA, Beaman, JH 2006. ‘Elevational species richness patterns for vascular plants on Mount Kinabalu, Borneo’. J Biogeogr., vol. 33, pp. 1838-1849. https://doi:10.1111/j.1365-2699.2006.01554.x Hágsater, E, Dumont, V, (Eds.) 1996. Orchids: Status, Survey and Conservation Action Plan. IUCN, Gland, Switzerland and Cambridge, UK. Hawkins, BA, Field, R, Cornell, HV, Currie, DJ, Guégan, JF & Kaufman, DM 2003. ‘Energy, water, and broadscale geographic patterns of species richness’. Ecology, vol. 84, pp. 3105-3117. http://repositories.cdlib.org/postprints/327. https://doi.org/10.1890/03-8006 Jacquemyn, H 2016. Habitat-driven variation in mycorrhizal communities in the terrestrial orchid genus Dactylorhiza. Sci. Rep. 6:37182, https://doi:10.1038/srep37182 Jarvis, C, Cribb, P 2009. Linnaean sources and concepts of orchids. Ann Bot., vol. 104(3), pp. 365–376. https://doi:10.1093/aob/mcp005 Kull, T, Hutchings, MJ 2006. ‘A comparative analysis of decline in the distribution ranges of orchid species in Estonia and the United Kingdom’. Biological conservation, vol. 129, pp. 31-39. Lindborg, R 2007. ‘Evaluating the distribution of plant life-history traits in relation to current and historical landscape configurations’. Journal of Ecology, vol. 95, pp. 555-564. Lomolino, MV 2001. ‘Elevation gradients of species-density: Historical and prospective views’. Global Ecology and Biogeography, vol. 10, pp. 3-13. https://doi:10.1046/j.1466-822x.2001.00229.x McCain, CM 2007. ‘Could temperature and water availability drive elevational species richness patterns? A global case study for bats’. Global Ecol Biogeogr., vol. 16, pp. 1-13. https://doi:10.1111/j.1466- 822x.2006.00263.x Melendo, M, Gimenez, E, Cano, E, Gomez-Mercado, F & Valle, F 2003. ‘The endemic flora in the south of the Iberian Peninsula: taxonomic composition, biological spectrum, pollination, reproductive mode and dispersal’. Flora, vol. 198, pp. 260-276. Myers, N, Mittermeier, RA, Mittermeier, CG, da Fonseca, GAB & Kent, J 2000. ‘Biodiversity hotspots for conservation priorities’. Nature, vol. 403, pp. 853-858. https://doi:10.1038/35002501 Naveh, Z 1982. ‘Mediterranean landscape evolution and degradation as multivariate biofunctions: theoretical and practical implications’. Landscape Plann., vol. 9, pp. 125-146. https://doi.org/10.1016/0304-3924(82)90003- X doi: 10.17700/jai.2018.9.1.430 36 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

O’Brien, EM 1998. ‘Water-energy dynamics, climate, and prediction of woody plant species richness: an interim general model’. J Biogeogr., vol. 25, pp. 379-398. https://doi.org/10.1046/j.1365-2699.1998.252166.x Pfeifer, M., Passalacqua, NG, Bartram, S, Schatz, B, Croce, A, Carey, PD, Kraudelt, H & Jeltsch, F 2010. ‘Conservation priorities differ at opposing species borders of a European orchid’. Biological Conservation, vol. 143, pp. 2207-2220. https://doi.org/10.1016/j.biocon.2010.06.005 Phillips, S. J., Anderson, R.P., & Schapire R.E 2006. ‘Maximum entropy modeling of species geographic distributions’. Ecological Modelling, vol. 190, pp.231-259. https://doi.org/10.1016/j.ecolmodel.2005.03.026 Phillips, S. J., & Dudík, M 2008. ‘Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation’. Ecography 31: 161–175. https://doi:10.1111/j.0906-7590.2008.5203.x Phillips, RD, Brown, AP, Dixon, KW & Hopper, SD 2011. Orchid biogeography and factors associated with rarity in a biodiversity hotspot, the Southwest Australian Floristic Region. Journal of Biogeography, vol. 38, pp. 487–501. https://doi.org/10.1111/j.1365-2699.2010.02413.x Fourth Author et al. 2015. In Proc. 1st Conference in GIS and Spatial Analysis in Agriculture and Environment, 28-29 May 2015, Ed D Kalyvas, Agricultural University of Athens, Greece, ISBN: 978-960-6806-18-6, Copyright 2016. Popova, E, Kim, HH, Saxena, PK & Engelmann, F 2016. ‘Frozen beauty: The cryobiotechnology of orchid diversity’. Biotechnology Advances, vol. 34, pp. 380-403. https://doi:10.1016/j.biotechadv.2016.01.001 Rasmussen, HN 2002. Recent developments in the study of orchid mycorrhiza. Plant Soil, vol. 244, 149–163. https://doi.org/10.1023/A:1020246715436 Remm, K, Linder, M, Remm, L 2008. ‘Relative density of Wnds for assessing similarity-based maps of orchid occurrence’. Ecol Modell. https://doi:10.1016/j.ecolmodel.2008.10.017 Rodríguez, JP, Brotons, L, Bustamante, J & Seoane, J 2007. ‘The application of predictive modelling of species distribution to biodiversity conservation’. Divers Distrib, vol. 13, pp. 243-251. https://doi.org/10.1111/j.1472- 4642.2007.00356.x Štípková, Z, Romportl, D, Černocká, V, Kindlmann P. 2017. Factors associated with the distributions of orchids in the Zeleníky mountains, Czech Republic. European Journal of Environmental Sciences, vol. 7, issue 2, pp. 135-145. https://doi.org/10.14712/23361964.2017.13 Tremlová, K, Münzbergová, Z 2007. ‘Importance of species traits for species distribution in fragmented landscapes’. Ecology, vol. 88, pp. 965-977. https://doi.org/10.1890/06-0924 Trigas, P, Panitsa, M, Tsiftsis, S 2013. ‘Elevational gradient of vascular plant species richness and endemism in Crete–the effect of post-isolation mountain uplift on a continental island system’. PLoS One, vol. 8, no 3, p. e59425. https://doi:10.1371/journal.pone.0059425 Tsiftsis, S, Tsiripidis, I, Karagiannakidou, V & Alifragis, D 2008. ‘Niche analysis and conservation of the orchids of east Macedonia (NE Greece)’. Acta oecologica, vol. 33, pp. 27-35. Tsiftsis, S, Tsiripidis, I, Trigas, P 2011. Identifying important areas for orchid conservation in Crete. European Journal of Environmental Sciences, vol. 1, issue 2, pp. 28-37. Tutin, TG, Heywood, VH, Burges, NA, Moore, DM, Valentine, DH, Walters, SM, Webb, DA (Eds.; with the assistance of Chater, AO, Richardson, IBK). 1980. Flora Europaea, vol. 5: Alismataceae to Orchidaceae (Monocotyledones). Cambridge University Press, Cambridge. xxxvi+[1]+452pp., 5 map. Van der Maaten, L, Schmidtlein, S, Mahecha, MD 2012. ‘Analyzing floristic inventories with multiple maps’. Ecological Informatics, vol. 9, pp. 1-10. https://doi.org/10.1016/j.ecoinf.2012.01.005 Warcup, J, H 1990. Mycorrhizas in Orchids of South Australia (eds Bates, R. J. & Weber, J. Z.) 21–26 (Flora and Fauna of South Australia Handbook Committee, 1990).

doi: 10.17700/jai.2018.9.1.430 37 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Appendix 1: Lithological and Hydro-lithological formations in the study area Geological Parameters Code Description Lithological Eocene to Oligocene marly limestones with thin marly intercalations. At the contact of the marly series with the 1 E-O.k underlying limestones there is a thin intercalation of coarse - conglomerate with pebbles of Cretaceous to Eocene age limestones with no uncorformity Thin to thick bedded limestones intecalated with 2 Ks-Pc.k microbrecciated limestones 3 Mi Conlomerates, phsammite, brecciated limestone and marles Pliocene deposits of light-grey to white sandstones with clayey- 4 Pl marly intercalations. The lower horizons consist of blue marls and marly sandstones

Pleistocene coastal deposits of loose or cohesive 5 Pt conglomerates with marly cement, blue marls and calcarenites

Triassic breccia. Gypsum and anydrite highly disturbed due to 6 T.br diapirism 7 Ts-Ji,k.d Triassic to Jurassic shallow-water limestones and dolomites Holocene alluvial deposits deriving from the transportation of 8 al various material into the lower parts of the island Gypsum coarse crystalline with well developed crystals and in 9 g beds of 5-20 m thickness Holocene terra rossa consisted of loam and sandy loam of 10 tr intense red colour Hydro-lithological Practically impermeable or with selective water circulation 1 A3 formations, with small or very small permeability Limestones, dolomites, crystalized limestones, marbles with 2 K1 high to moderate permeability Miocene, Pliocene and Pleistocene deposits with moderate to 3 P2 small permeability Granular non sedimetary formations with small to very small 4 P3 permeability 5 P4 Granular sedimentary deposits with variable permeability 6 g Gypsum

doi: 10.17700/jai.2018.9.1.430 38 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Appendix 2: The presence of orchid species in Zante Island and the number of recordings in 2015- 2016 per species are displayed. Species nomenclature in line with Euro+Med (2006-) cross-validated with Dimopoulos et al. 2013 & Flora Ionica Working Group (2016 onwards): (*) asterisk in parenthesis indicates the species is listed as heterotypic synonym of the taxon. # indicates the taxon name as included in Dimopoulos et al. 2013. Nr Orchid species Nr of recordings 1 Anacamptis coriophora subsp. fragrans (Pollini) R. M. Bateman, Pridgeon & M. W. Chase # 257 2 Anacamptis laxiflora (Lam.) R. M. Bateman, Pridgeon & M. W. Chase 5 3 Anacamptis papilionacea (L.) R. M. Bateman, Pridgeon & M. W. Chase 71 4 Anacamptis pyramidalis (L.) Rich. 186 5 Dactylorhiza romana (Sebast.) Soó subsp. romana # 3 6 Himantoglossum robertianum (Loisel.) P. Delforge 130 7 Limodorum abortivum (L.) Sw. 16 8 Neotinea lactea (Poir.) R. M. Bateman, Pridgeon & M. W. Chase 63 9 Ophrys apifera Huds. 4 10 Ophrys umbilicata Desf. subsp. umbilicata 8 11 Ophrys holoserica (Burm. f.) Greuter 137 (*Ophrys bombyliflora Spreng.) 12 Ophrys holoserica (Burm. f.) Greuter subsp. holoserica # 16 (*Ophrys episcopalis Poir.) 13 Ophrys ferrum-equinum Desf. subsp. ferrum-equinum 118 14 Ophrys ferrum-equinum subsp. gottfriediana (Renz) E. Nelson 15 15 Ophrys fusca subsp. cinereophila (Paulus & Gack) Faurh. 29 16 Ophrys scolopax subsp. heldreichii (Schltr.) E. Nelson 2 17 Ophrys holoserica subsp. candica (Greuter & al.) H. A. Pedersen & Faurh. 1 18 Ophrys sphegodes Mill. subsp. sphegodes # 17 (*Ophrys sphegodes subsp. atrata (Rchb. f.) A. Bolòs) 19 Ophrys fusca subsp. iricolor (Desf.) K. Richt. 45 20 Ophrys lutea subsp. melena Renz 7 21 Ophrys lutea subsp. galilaea (H. Fleischm. & Bornm.) Soó 21 22 Ophrys lutea Cav. subsp. lutea 232 23 Ophrys sphegodes subsp. mammosa (Desf.) Soó ex E. Nelson 197 24 Ophrys scolopax Cav. subsp. scolopax 16 25 Ophrys scolopax Cav. 7 26 Ophrys speculum Link 3 27 Ophrys sphegodes Mill. 13 28 Ophrys sphegodes subsp. spruneri (Nyman) E. Nelson 55 29 Ophrys tenthredinifera Willd. 262 30 Ophrys umbilicata Desf. 7 31 Orchis anthropophora (L.) All. 31 32 Orchis italica Poir. 122 33 Orchis simia Lam. 5 34 Orchis mascula (L.) L. 3 35 Orchis pauciflora Ten. 16 36 Anacamptis palustris (Jacq.) R. M. Bateman, Pridgeon & M. W. Chase 3 37 Orchis provincialis Balb. ex Lam. & DC. 178

doi: 10.17700/jai.2018.9.1.430 39 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:26-40

Nr Orchid species Nr of recordings 38 Orchis quadripunctata Cirillo ex Ten. 260 39 Serapias bergonii E. G. Camus 29 40 Serapias cordigera L. 126 41 Serapias lingua L. 77 42 Serapias neglecta De Not. 49 43 Serapias neglecta subsp. ionica # 237 44 Serapias orientalis (Greuter) H. Baumann & Künkele 80 45 Serapias parviflora Parl. 31 46 Serapias vomeracea (Burm. f.) Briq. 20 47 Spiranthes spiralis (L.) Chevall. 3

doi: 10.17700/jai.2018.9.1.430 40 Aristotelis Martinis, Evgenia Chaideftou, Charikleia Minotou, Κonstantinos Poirazidis: Spatial analysis of orchids diversity unveils hot-spots: The case of Zante Island, Greece Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol. 9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill.

Krenaida Taraj1, Ilirjan Malollari2, Fatos Ylli3, Ramiz Maliqati1, Adelaida Andoni1, Jonilda Llupa4

I N F O A B S T R A C T Received 28 Feb 2018 Balkan countries are home to outstanding natural bioresources such as large number of Accepted 17 Mar 2018 herbs, plants and forestry species. They are also, among the most important export regions Available on-line 23 Mar 2018 for medicinal and aromatic plants in Europe. Balkan has various flora with different tree Responsible Editor: M. Herdon species as well. Pinus halepensis, also known as the Aleppo pine, is native to the Mediterranean region. However, its range extends to Balkan countries as well, such as Keywords: Albania, Greece, Croatia and Montenegro. It is one of the many trees that are well known Pinus halepensis Mill., Soxhlet for their medicinal properties, as well as for their economical significance. In this work and Clevenger extraction, FTIR essential oils/extracts of Pinus halepensis needles were acquired by hydro-distillation and UV-Vis. spectroscopy (Clevenger type) and Soxhlet extraction. The pine oils and crude extracts were analysed by FTIR spectroscopy and UV-Vis spectrophotometry aiming the identification of the main chemical constituents of the oil extracts of Albanian pine needles. FTIR analyses indicated presence of caryophyllene and pinene as the main chemical constituents in the essential oil and extract of pine needles.

1. Introduction Balkan countries possess rich ecosystems with enormous natural and biological value (Kathe et al. 2003, Metaj 2007, Pieroni & Quave 2014, Schmiderer et al. 2013). Balkan region has very various flora with a great deal of different tree species as well, including pine species. Pinus halepensis (Pinaceae), also known as the Aleppo pine, is native to the Mediterranean region (Abi-Ayad et al. 2011). It is one of the many trees that are well known for their medicinal properties, as well as for their economical significance (Cheikh-Rouhou et al. 2006). Its range expands from Morocco, Algeria, Spain to France, Italy, Croatia, Albania (Toromani et al. 2015) and Greece (Roussis et al. 1995). In Albania, conifer forests cover an area of 104780 ha with a standing volume of 10.7 million cubic

1Krenaida Taraj University of Tirana, Department of Chemistry, Faculty of Natural Sciences, University of Tirana, Blv. Zog I, 1001, Tirana, Albania [email protected] 2Ilirjan Malollari University of Tirana, Department of Chemical Engineering, Faculty of Natural Sciences, University of Tirana, Blv. Zog I, 1001, Tirana, Albania [email protected] 3Fatos Ylli Insitute of Applied Nuclear Physics, P.O. Box 85, University of Tirana, Tirana, Albania [email protected] 1Ramiz Maliqati University of Tirana, Department of Chemistry, Faculty of Natural Sciences, University of Tirana, Blv. Zog I, 1001, Tirana, Albania 1Adelaida Andoni University of Tirana, Department of Chemistry, Faculty of Natural Sciences, University of Tirana, Blv. Zog I, 1001, Tirana, Albania [email protected] 4Jonilda Llupa Laboratory of Food Chemistry, Department of Chemistry, University of Ioannina, Greece [email protected] doi: 10.17700/jai.2018.9.1.440 41 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

meters, while the Mediterranean pine species occupy 18440 ha. Pinus species (P. halepensis and P. pinea) cover approximately 85% of that area (Toromani et al. 2015). Essential oils and crude extracts from Pinus species are reported to have different therapeutic properties. They are used as fragrances, flavoring additives for food and beverages, and intermediates in synthesis of perfumes (Fekih et al. 2014). In this respect, many authors have reported, the major chemical components of Pinus halepensis extract oils are β-caryophyllene and α-pinene (Roussis et al. 1995, Dob et al. 2005, Cheikh-Rouhou et al. 2006, Abi-Ayad et al. 2011, Fekih et al. 2014). Usually the oil extraction from plants are carried out by water distillation and organic solvent extraction using a Soxhlet technique (Dob et al. 2005, Taraj et al. 2013, Andoni et al. 2014, Ciko et al. 2016, Taraj et al. 2017). Following our previous studies on the essential oils extraction from Albanian plants (Taraj et al. 2013, Andoni et al. 2014, Ciko et al. 2016, Taraj et al. 2017), we further extended this work by utilizing a Clevenger type apparatus (hydrodistillation extraction) and a Soxhlet extractor to obtain essential oil from P. halepensis needles. Chemical composition analyses were carried out by spectroscopy techniques. Spectroscopy methods are effective in assessing the qualitative difference between samples (Andoni 2009, Andoni et al. 2009, Andoni 2014, Schulz et al. 2004, Schulz et al. 2005). The first contribution of this paper is to obtain essential oils or extracts from Albanian P. halepensis by different traditional methods. The second contribution is to determine the chemical composition (major bioactive components) by spectroscopic techniques and subsequently compare the obtained findings with those reported in the literature. The paper is structured as follows. Section 2 presents the methodology used for all the oil extractions. Section 3 presents the results on extractions by different methods and the relevant spectra by different spectroscopic techniques. Section 4 presents a brief discussion on the results. At last some conclusions are drawn in Section 5. 2. Methodology The origin of P. halepensis used in this work is from Western Albania. The pine needles were dried in shadow at room temperature and cut off in small pieces. The water distillation extraction was carried out by means of Clevenger apparatus in a round bottom flask using a ratio of 10:1 water/dried needles. The water-flower mixture was then subjected to distillation for an optimum number of hours, which was determined to be 3 hours. In the first 30 minutes once the oil had started collecting in the collecting column of the Clevenger apparatus, about 1 ml of hexane (oil phase) was put through the condenser to wash down any oil which had stuck to the walls of the condenser. The essential oil (dissolved in hexane) was then separated in a separating funnel (Ciko et al. 2016). In the Soxhlet extraction, pine needles were placed inside a container made of thick filter. The container is located into the main chamber of the Soxhlet extractor. The Soxhlet can be slotted onto a flask which contains hexane (in this work), as extraction solvent. The Soxhlet is afterward equipped with a condenser, whereas the hexane is heated and allowed to reflux. The amount of the herb used for Soxhlet extraction was ~10 g, whereas the amount of the solvent (hexane) used was 300 mL. In the current work the extraction process was allowed to run approximately 3 hours. Hexane is removed, by means of a rotary evaporator, yielding though the extracted compounds (Ciko et al. 2016). FTIR spectra were obtained by Nicolet 6700 spectrometer, manufactured by Thermo Electron. In this study, measurements were carried out in the range mid Infra-Red (4000 – 400 cm-1). The spectra were analyzed using OMNIC software. UV-Vis spectra measurements were carried out by 6800 UV- VIS Jenway spectrophotometer. 3. Results Table 1 displays overall results of the yields for the oils obtained with different methods. It is evident from table 1 that the Soxhlet extraction gives rise to higher yield when compared to the yield obtained from the extraction with the Clevenger apparatus. In this respect Dob et al. 2005, reported yield of 0.52% for the essential oil of P. halepensis extracted by hydro-distillation.

doi: 10.17700/jai.2018.9.1.440 42 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

Table 1. Overall results for the extraction of essential oil and crude extract of P. halepensis. Extraction Amount of Extraction Extraction Extraction Yield method P. halepensis solvent time Temperature of extract Clevenger (distillation) 10 g Water 3 h 1200C 0.30% Soxhlet (distillation) 10 g Hexane 3 h 800C 3.54% Figure 1 exhibit FTIR spectrum of the P. halepensis oil extract obtained with the hydro-distillation method (in the insert) and FTIR spectrum of the P. halepensis crude extract obtained with Soxhlet extractor. The spectra of all extracts revealed very similar features. The acquired essential oil obtained by hydro-distillation had a pale yellow color whereas the crude extract obtained with the Soxhlet apparatus had yellow color.

Figure 1. FTIR spectra of P. halepensis essential oil obtained by a Clevenger type hydro-distillation (in the insert) and Soxhlet type distillation. The main components identified in the FTIR spectrum are indicated by arrows.

FTIR spectrum of the oil extract obtained by Clevenger apparatus (inserted spectrum) indicates peaks positioned at ~1637 cm-1, ~1466 cm-1 and ~1380 cm-1. Other IR signals appear in the region ~3000-2800 cm-1 along with minor peaks evident at ~ 1000 cm-1 and at ~700 cm-1. Additionally, FTIR spectrum of the crude extract (Soxhlet extraction) reveals similar features. IR spectrum indicates peaks positioned at ~1645 cm-1, ~1464 cm-1 and ~1378 cm-1. Other IR signals appear in the region ~3000- 2800 cm-1 along with minor peaks evident at ~ 1038 cm-1 and at ~700 cm-1. Furthermore, there are three distinctive bands in the IR spectrum of the crude extract. An intense band appears in the region 1730-1700 cm-1, whereas two other bands appear at ~1240 cm-1 and at ~1181 cm-1. Additionally, UV- VIS spectrum (Figure 2) of P. halepensis oil extract (Clevenger extraction) indicates peaks in the regions, 200-210 nm, 230-235 nm and a minor peak at 270-290 nm.

doi: 10.17700/jai.2018.9.1.440 43 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

Figure 2. UV-Vis spectrum of P. halepensis essential oil obtained by hydro-distillation.

4. Discussion In the FTIR spectrum of the P. halepensis essential oil appears two peaks positioned at ~1466 cm-1 and at ~1380 cm-1. It is known that isopropyl and gem-dimethyl groups give rise to a split umbrella mode with two peaks in the IR spectrum positioned at ~1385 to 1365 cm-1 (Smith 1999). The splitting is caused by vibrational interaction between the umbrella modes of the two methyl groups. The split of the umbrella modes is of about equal intensity. Meanwhile, t-butyl and isopropyl groups also give rise to a split umbrella mode with two peaks positioned between ~ 1393 to 1366 cm-1 (Smith 1999). However, the approximate intensity ratio in this case is 1:2 (Smith 1999). Additionally, the band at ~ -1 1466 cm can also indicate the presence of a CH3, a CH2 or both groups; whereas CH3 symmetric bend (umbrella mode) shows up at 1375±10 cm-1 (Smith 1999). The chemical structures of caryophyllene consists of isopropyl or isobutyl groups, therefore we assign the peaks at at ~1466 cm-1 and at ~1380 cm-1 to caryophyllene. These findings are in good agreement with IR data reported for oil extracts of different herbs (peak position of the same functional groups) (Schulz et al. 2004, Schulz et al. 2005). In this respect, the same authors (Schulz et al. 2004, Schulz et al. 2005) reported IR spectrum of caryophyllene and indicated presence of three peaks in the region 1500-1300 cm-1. The latter is in good agreement with our findings. The C=C stretches appear at ~1660-1630 cm-1 (Smith -1 1999), therefore the IR at ~1630 cm belong to pinene and caryophyllene (vinyl group, =CH2) (Schulz et al. 2004, Schulz et al. 2005). The FTIR spectrum of the crude extract of P. halepensis features similiar signals as the IR spectrum of essential oil of pinus needles. The FTIR peaks in the region 1464-1378 cm-1 and 1645 cm- 1 suggests presence of caryophyllene and pinene respectively. In comparison to the IR spectrum of the essential oil, the IR spectrum of the crude extract is characterised by three additional signals in the region 1730-1700 cm-1, ~1240 cm-1 and ~1181 cm-1. It has been reported by Mohareb et al. 2017, that crude extracts of pinus species contain fatty acids, resin acids and phthalates as well. In this respect, the additional signals in the IR spectrum are characteristic for the ester or acid functional group (Smith 1999). In addition, UV-VIS spectrum of pine essential oil gives rise to features in 200-290 nm region. These features originate from the double bonds (or conjugated double bonds) of caryophyllene and pinene (Taraj et al. 2013).

5. Perspectives and conclusion The extracts from P. halepensis were obtained by means of Clevenger apparatus (hydro- distillation) and Soxhlet extractor using hexane as solvent. The extraction with Clevenger apparatus gave rise to lower yield compared to the yield obtained with Soxhlet method extraction. IR spectra of P. halepensis essential oil and crude extract indicated presence of caryophyllene and pinene as two doi: 10.17700/jai.2018.9.1.440 44 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

major chemical constituents in the extract. IR signals for the isopropyl groups (1466-1380 cm-1) and vinyl group (1645-1637 cm-1) were in good agreement with IR data reported by Schulz and co-authors. UV-Vis signals supported IR findings for the caryophyllene and pinene presence. To this end, the chemical composition of Albanian P. halepensis oil is similar to the chemical composition reported in the literature for the pine needles (P. halepensis). Extractions by Clevenger and Soxhlet apparatuses could also be combined with the extractions with liquid CO2 at different temperatures (sub-critical conditions). Compressed CO2 is frequently used as it represents an environmentally friendly solvent (Taraj et al. 2013). The extraction by liquid CO2 is a very selective extraction. The lower temperature of the extraction allows better separation of thermally labile compounds (Taraj et al. 2013).

References Abi-Ayad, M., Abi-Ayad, F.Z., Lazzouni, H.A., Rebiahi, S.A, Ziani_Cherif C., and Bessiere, 2011. Chemical composition and antifungal activity of Aleppo pine essential oil. Journal of Medicinal Plants Research, 5(22), pp. 5433-5436. Andoni, A. 2009. A flat model approach to Ziegler-Natta olefin polymerization catalysts Eindhoven, PhD Thesis. Technische Universiteit Eindhoven (Eindhoven University of Technology). The Netherlands. Ch. 7. Andoni, A., Chadwick, J.C., Niemantsverdriet. J.W., Thüne, P.C., 2009. Investigation of Planar Ziegler-Natta Model Catalysts Using Attenuated Total Reflection Infrared Spectroscopy. Catalysis Letters, 130, pp. 278-285. Andoni, A., 2014. High resolution electron energy loss spectroscopy for studying planar model catalyst: A test of NO on Rh(100). Revue Roumaine de Chimie, 59(3-4), pp. 245-249. Andoni, A., Xhaxhiu, K., Taraj, K., Çomo, A., 2014. An adsorption method for characterization of surface area and pore size of solid surfaces. Asian Journal of Chemistry, 26, pp. 6833-6838. https://doi.org/10.14233/ajchem.2014.16991 Cheikh-Rouhou, S., Hentati, B., Besbes, S., Blecker, C., Deroanne, C., and Attia, H., 2006. Chemical Composition and Lipid Fraction Characteristics of Aleppo Pine (Pinus halepensis Mill.) Seeds Cultivated in Tunisia, Food Science and Technology International, 15(5), pp. 407-416. Ciko, L. Andoni, A., Ylli, F., Plaku, E, Taraj, K., Çomo, A., 2016. Extraction of essential oil from Albanian Salvia Officinalis L. and its characterization by FTIR spectroscopy: A Soxhlet method extraction. Asian Journal of Chemistry, 28(6), pp. 1401-1402. https://doi.org/10.14233/ajchem.2016.19658 Dob, T. Berramdane T., Chelgoum C., 2005. Chemical Composition of essential oil of Pinus halepensis Miller growing in Algeria, Comptes Rendus Chimie, 8(11-12), pp. 1939-1945. https://doi.org/10.1016/j.crci.2005.05.007 Fekih, N. Allali, H., Merghache, S., Chaïb, F., Merghache, D., Amine, M.E., Djabou, N., Muselli, A., Tabti, B., Costa, J., 2014. Chemical composition and antibacterial activity of Pinus halepensis Miller growing in West Northern of Algeria, Asian Pacific Journal of Tropical Disease, 4(2), pp. 97-103. https://doi.org/10.1016/S2222- 1808(14)60323-6 Kathe, W., Honnef, S. & Heym A., 2003. Medicinal and Aromatic Plants in Albania, Bosnia-Herzegovina, Bulgaria, Croatia and Romania, BFN (German Federal Agency for Nature Conservation), Bonn. https://www.bfn.de/fileadmin/MDB/documents/skript91.pdf Metaj, M., 2007. Biodiversity and the protected areas systems in Albania, Biodiversity, 8, pp. 3-10. https://doi.org/10.1080/14888386.2007.9712823 Mohareb, S.A., Kherallah, EA.I., Badawy, EI.M., Salem, ZM.M., Yousef,, A.H., 2017, Chemical Composition and Activity of Bark and Leaf Extracts of Pinus halepensis and Olea Europaea Grown in Al-Jabel Al-Akhdar Region, Libya Against some Plant Phytopathogens, Journal of Applied Biotechnology & Bioengineering, 3(3), pp. 67. Pieroni A. & Quave C.L., 2014. Ethnobotany and biocultural diversities in the Balkans. Wild food and medicinal plants used in the mountainous Albanian north, north east and east: A comparison, ch. 10, pp 183-194, ISBN 978-1- 4939-1491-3. https://doi.org/10.1007/978-1-4939-1492-0 Roussis, V., Petrakis P.V., Ortiz A. & Mazomenos B.E., 1995. Volatile constituents of needles of five pinus species grown in Greece, Phytochemistry, 39(2), pp. 357-361. https://doi.org/10.1016/0031-9422(94)00885-W doi: 10.17700/jai.2018.9.1.440 45 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:41-46

Schmiderer, C., Torres-Londoño, P. & Novak, J., 2013. Proof of geographical origin of Albanian sage by essential oil analysis, Biochemical Systematics and Ecology, 51, pp. 70-77. https://doi.org/10.1016/j.bse.2013.08.007 Schulz, H. Baranska, M., Belz, H-H., Rösch, P., Strehle, M.A., Popp J., 2004. Chemotaxonomic characterisation of essential oil plants by vibrational spectroscopy measurements, Vibrational Spectroscopy, 35, pp. 81–86. https://doi.org/10.1016/j.vibspec.2003.12.014 Schulz, H. Özkan, G., Baranska, M., Krüger, H., Özcan M., 2005. Characterisation of essential oil plants from Turkey by IR and Raman spectroscopy, Vibrational Spectroscopy, 39, pp. 249-256. https://doi.org/10.1016/j.vibspec.2005.04.009 Smith, B., 1999. Infrared spectral interpretation: A systematik approach. CRC Press LLC. Taraj, K. Delibashi, A., Andoni, A., Lazo, P., Kokalari, (Teli) E., Lame, A., Xhaxhiu, K., Çomo A., 2013. Extraction of chamomile essential oil by subcritical CO2 and its analysis by UV-VIS spectrophotometer. Asian Journal of Chemistry, 25, pp. 7361-7364. https://doi.org/10.14233/ajchem.2013.14642 Taraj, K. Malollari, I., Andoni, A., Ciko, L., Lazo, P., Ylli, F., Osmeni, A., Çomo, A., 2017. Eco-extraction of Albanian chamomile essential oils by liquid CO2 at different temperatures and characterization by FTIR Spectroscopy, Journal of Environmental Protection and Ecology 18(1), pp. 117–124. Toromani, E. Pasho, E., Alla, A.Q., Mine, V., and Çollaku, N., 2015, Radial growth responses of Pinus halepensis Mill. and Pinus pinea L. forests to climate variability in western Albania, Geochronometria, 42, pp. 91-99.

doi: 10.17700/jai.2018.9.1.440 46 Krenaida Taraj, Ilirjan Malollari, Fatos Ylli, Ramiz Maliqati, Adelaida Andoni, Jonilda Llupa : Spectroscopic study on chemical composition of essential oil and crude extract from Albanian Pinus halepensis Mill. Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Hungarian Association of Agricultural Informatics Journal of Agricultural Informatics. Vol. 9, No. 1 European Federation for Information Technology in journal.magisz.org Agriculture, Food and the Environment

A Genetic Programming Study on Classification of Cassava Plant

Indra Laksmana1, Rosda Syelly2, Nurzarah Tazar 3, Perdana Putera 4

I N F O A B S T R A C T Received 16 Oct 2017 Accepted 7 Jan 2018 Cassava (Manihot esculenta Crantz) is an important plant that is consumed in many Available on-line 26 Mar 2018 forms. It could be processed as vegetable, chips, fodder, or bioethanol through a Responsible Editor: M. Herdon fermentation process. The cyclic acid HCN of cassava varies based on the varieties. Cassava with high HCN is toxic when it is consumed directly. This research designed a Keywords: system to identify the cassava varieties based on HCN content by applying a heuristic Cassava Varieties, Genetic search algorithm, using genetic operations. The identification of HCN content by programming, HCN content, applying Generic programming produced a structured classification rule and represents System identification. in tree form. The experiment in this study used binary code data generated from booleanizing process. Binary code data is divided into training data and test data using 5-fold cross-validation, and then the process of genetic operation. Rules are derived from repeated experiments to get the best rule. The best rule to identify with an average accuracy of 95.26%, obtained on population parameters of 10,000, 20-30 nodes. The node consists of Function set of AND, OR, NOR and 96 terminal sets (attributes / identifiers); in addition, the best classification rules are obtained on the crossover probability of 0.9 and 0.1 mutations of 10 generations. The resulting Rule can be utilized by the community in identifying the class of HCN cassava content.

1. Introduction The problems of classification often occur in daily life, such as choosing a vehicle, diagnosing the disease, looking for foods or drugs. It requires someone’s skilled, so the mistakes in the classification of decisions could be minimized. The limitation of skilled increase the error in classifying, therefore an alternative method is needed in determining a solution to classification problem. The selection of appropriate classifier requires consideration of many factors, namely classification accuracy, algorithm and computational performance (Qurat-ul-ain et al. 2010). According to Wahyudi (2013) Classification is a collection of a record in the form of training data set, where each record contains a set of attributes and one attribute is a class. The concept of artificial intelligence can be used to answer the classification problem. Artificial intelligence has the ability to think, guess an answer or perform the certain tasks such as human behavior that allow beyond human capabilities (Nakamura et al.2017). One of the artificial intelligence solutions that can be used in classification problems is genetic programming. Genetic programming (GP) is used to study patterns of data (Sudharmono. 2012). GP is a variant of the genetic algorithm which uses simulated evolution to discover functional programs to solve some task (Luke 2000). According to Sakprasat and Sinclair (2007), the main motivation for using genetic programming in classification rule mining is robustness and an adaptive search method making it more effective in finding patterns. Laksmana et al (2013) has applied GP programming method in identification of family of medicinal plants with an accuracy of 86.32%, resulting in a hierarchy in identifying medicinal plants.

1 Department Agricultural Engineering, Payakumbuh State Polytechnic of Agriculture [email protected] 2 Computer Engineering Department, Payakumbuh School of Technology [email protected] 3 Department of Food Technology, Payakumbuh State Polytechnic of Agriculture [email protected] 4 Department of Agricultural Engineering, Payakumbuh State Polytechnic of Agriculture [email protected] doi: 10.17700/jai.2018.9.1.413 47 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Cassava (Manihot esculenta Crantz) is the third food crop in Indonesia after rice and corn. Cassava in Indonesia, has many regional names such as singkong, ubi jendral, ubi inggris, telo puhung, kasape, bodin, telo jendral, sampeu, huwi dangdeur, huwi jendral, kasbek and ubi perancis. Cassava is frequently used as industrial raw materials, fodder and bioethanol (Purwono and Heni, 2009). Its leaves are used as vegetables and fodder. Its stem used as a fence and planting materials, its seeds can be used as oil and its tuber can be processed as tapioca flour and as bioethanol through the fermentation process. Cassava could directly process directly as West Sumatra traditional food. For example; the boiled cassava as Getuk Kacimuih, the fried cassava as Sanjai chips or Balado chips, etc. The waste of cassava peel can be used to feed goat/sheep (Hanifah 2010). Cassava has many benefits; it encourages the government of West Sumatra to increase the production and productivity of cassava. There are some types of cassava based on the level of cyanide acid (HCN); low, medium, high and very high. Cassava with a large HCN content of 80 mg / kg fresh bulbs tastes bitter and should not be consumed directly. Generally, this cassava is used as flour (Sundari 2010). Cassava has many varieties with varying levels of HCN (Unigwe et al. 2017). The diversity of cassava varieties in Indonesia is quite high. Bank Gen BB-Biogen Bogor recorded as many as 600 germplasm accessions, 452 of which are in the data base (BB-Biogen, 2010). This condition causes a variety of cassava varieties in the field. Therefore, people have to choose which varieties to plant and to consume. Therefore, there is a need it is need for the research to determine the best rule of classification. This study attempted to apply GP to identify the varieties of tubers based on HCN level of contention. The rule of classification or hierarchy in identifying varieties of yams makes the identification process easier, faster and structured. There is a hope to help people to recognize the varieties of cassava easily so that the selection of cassava varieties to be planted can be adjusted on the allocation.

2. Method C# programming language was used for running GP. Data collected entirely from the field, it directly taken from 15 people who planted cassava. The types of cassava that taken as the data called by the names given by the local farmers and the community. They are ubi roti, ubi sanjai, ubi putih, ubi lantak, ubi keriting, ubi kuning, ubi hijau, ubi mentega, ubi roti tiakar, ubi BW, ubi merah, ubi hitam, ubi thailand merah, ubi tailan putih and ubi kasesat. In this study the researchers used 129 cassava plants; consisting of 15 species of cassavaes from various plant locations. Then they are planted in the same location. 10 samples are taken sistematically as the data and used in this study. The content of HCN (Cyanide acid) of the cassava plants data are calculated by using the following formula:

( ( ) ( )) = 10.000 ( ) 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑚𝑚𝑚𝑚 −𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑚𝑚𝑚𝑚 𝑥𝑥 𝑁𝑁 𝑥𝑥 27𝑥𝑥100

𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 𝑜𝑜𝑜𝑜 𝐻𝐻𝐻𝐻𝐻𝐻 � 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑥𝑥 1000 � 𝑥𝑥 The results of the calculations were classified into three classes of cassava; based on their low HCN content (HCN <= 50 mg / kg), medium (50 mg /kg < HCN < 80 mg / kg) and high (HCN > = 81 mg/kg). There are 96 selected attributes of 129 cassava plants. These characteristic attributes are derived from 5 physical traits based on its morphology such as leaves, stems, tubers, fruits and flowers. The stages performed in the study were shown in Figure 1.

doi: 10.17700/jai.2018.9.1.413 48 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Data Collection

Data processingf

Booleanize

Data division

Low HCN Medium HCN High HCN

Training Data Test Data Training Data Test data Training data Test data

Genetic Programming Genetic Programming Genetic Programming Generate rule Generate rule Generate rule

Fitness Fitness Fitness evaluation evaluation Evaluation

Classification Classification Is stop condition Yes Is stop condition Yes is stop condition Yes fulfilled? rule fulfiled? rule fulfiled Classification

No No No GenetiC Genetic Genetic operation Evaluasi Operation Evaluation operation Evaluation

Rule combination

1st Analyzing

Figure 1. Research stages for identification cassava

2.1. Booleanize The booleanize performs the encoding which changes the attributes of the identifier to X0, X1 through Xn. The information of each cassava plant will be changed to the binary values of 0 and 1. The number 0 indicates the absence of any characteristics in a variety while the number 1 indicates existence of the characteristic. Each identifier is encoded from X0 to X95. Booleanize of all data used in this study can be seen in Table 1

Table 1. Booleanizing of data Physical Sub division Encoding aspect Leaf number leaflet odd (X0), even (X1) Structure Rough or soft (X2) texture clear (X3), very clear (X4), vague (X5) shoot color Purplish green (X6), light/ dark green (X7), dark purple/ black purple (X8) vein color White (X9), yellowish white (X10), redness white (X11), green (X12), purplish yellow (X13), beige (X14) stalk color Yellowish green (X15), green (X16), Purple green (X17), red green, (X18), Brownish green (X19), Red (X20), redness yellow (X21)

doi: 10.17700/jai.2018.9.1.413 49 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

high stalk color Green (X22), Brownish green (X23), Redness green (X24), Yelowish green (X25), Green with slightly purple (X26), Redness yellow (X27), Red (X28) leaf stalk length (PTD) cm PTD <=15.5 (X29), 15.6 >PTD>21.5 (X30), PTD >=21.5 (X31) leaf width (LHD) cm LHD<=3.7 (X32), 3.7>LDH>5 (X33), LHD>=5 (X34) leaf length (PHD) cm PHD<=14.3 (X35), 14.3>PHD>17.4 (X36), PHD>=17.4 (X37) leaf form thick (X38), thin (X39) leaf weight (BD) gr BD<=0.39 (X40), 0.39>BD>0.62 (X41), BD>=0.62 (X42) leaf tip form Wide tapered (X43), wide rounded (X44), taper (X45) Stem stem height (TB) cm TB<=241 (X46), 241>TB>304 (X47), TB>=304 (X48) young stem color light green (X49), dark green (X50), rednes/purplish green (X51) old stem color grey (52), light yellow (X53), dark brown green (X54), whitish/redness brown (X55), silver brown (X56), silver and red (X57) distance of young stem segment JRBM<=35.78 (X58), 35.7>JRBM>45.1 (X59), (JRBM) mm JRBM>=45.1 (X60) distance of old stem segment Tua JRBT<=74.5 (X61), 74.5>JRBT>121.8 (X62), (JRBT) mm JRBT>=121.8 (X63) stem diameter (DB) mm DB>=21.51 (X64), 21.51>DB>28.47 (X65), DB>=28.47(X66) number of branch 1 dan 2 (X67), more than 2 (X68) branch form straight (X69), buckle (X70) tuber outer peel color beige (X71), light brown (X72), Pink (X73), dark brown (X74), light red (X75) inner peel color white (X76), beige (X77), yellowish (X78) flesh color white (X79), beige (X80), yellowish (X81) thickness of peel (TKU) mm TKU<=1.01 (X82), 1.01>TKU>1.33 (X83), TKU>=1.33 (X84) epidermis color brown (X85), dark brown (X86), yellowish (X87) epidermis thicknes (TKA) mm TKA>=0.28 (X88), 1.01>TKA>1.33 (X89), TKA>=0.54 (X90) Fruit and fruitish dan flowerish fruitish (X91), flowerish (X92) Flower Amount of sap a little (X93), medium (X94), much (X95)

2.2. Data Division K-fold Cross-validation is used to conduct training and testing. The data is divided equally into K sections and then perform as much as K iteration. If the amount of data (N) is not divisible by K, then the end of the data will have more data than the previous data (K-1). Each iteration, K alternately will be the test data and the K-1 section is used as training data. (Bramer 2007). The booleanized data set were divided by class into training data and the test data with the proportions are 80% and 20% respectively. The distribution of data uses K-fold cross validation method with K= 5. The data is split into five equal parts, the number 5 is chosen because it is assumed that this number will gives a better result. Training data and test data are divided alternately. Four subsets of training data is used as training input in classification and a subset of test data is used to test the training model. The data division scenarios are given in Tables 2 and 3. doi: 10.17700/jai.2018.9.1.413 50 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Table 2. Data Division Fold Data Subset Training data S1, S2, S3, S4 Fold 1 Test data S5 Training data S1, S2, S3, S5 Fold 2 Test data S4 Training data S1, S2, S4, S5 Fold 3 Test data S3 Training data S1, S3, S4, S5 Fold 4 Test data S2 Training data S2, S3, S4, S5 Fold 5 Test data S1

Table 3. Data Division Scenario Class S1 S2 S3 S4 S5 Total Low 6 6 6 6 6 30 Medium 12 12 12 12 11 59 High 8 8 8 8 8 40 Total 26 26 26 26 25 129

2.3. Genetic Programming The Genetic Programming algorithm is designed based on Charles Darwin's theory of evolution by Jhon R. Koza. He was inspired by John Holland who created the Genetic Algorithm. In 1992 Koza applied GP to create a system or computer program that is able to create its own program (Automatic Programming). The method is called Genetic Programming (Lukas 2008), that creates computer program in computer language Lisp, draft scheme as its solution (Koza 1992). Genetic Programming (Koza 1992) is a search algorithm based on natural system mechanism that is genetic and natural selection (Lukas 2008). The solution variables in GP are encoded into a string structure that represents the gene sequence, which is characteristic of the solution. This set is called population. All individuals in the population are representatives of the solution. Part of the individual is called a chromosome. These chromosomes evolve in a continuous iteration process called a generation. In every generation, the individual is evaluated based on an evaluation function until the genetic programming generation will converge to the best individual. In the hope that this is the optimal solution (Laksmana et al. 2013). Genetic Programming by Poly et al. (2008) is an evolutionary computational technique to automatically solve a problem without the need to be told clearly what to do by determining the shape or structure of the solution at the beginning of the problem. Individuals in this study represent the model or hierarchy of cassava varieties. The population is a number of rules that are formed randomly. Each rule will be evaluated based on a particular fitness. The primitive form of Genetic Programming is the set of functions (AND, OR, NOR) and the set of arguments (terminal set) that is the result attribute of booleanization. The next process is as shown in Figure 2.

doi: 10.17700/jai.2018.9.1.413 51 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Gen :=0

Create Initial Random Population

Termination Criterion Yes Designate Satisfied ? Result No

Evaluate Fitness of Each End Individual in Population

i :=0

Yes Gen := Gen+1 i=M? No

Pr Select Genetic Operation Pm Probabilistically Pc Select One Individual Select Two Individuals Select One Individual Based on Fitness Based on Fitness Based on Fitness

i:=i+1

Perform Reproduction Perform Crossover Perform Mutation

Copy into New Insert Offspring into Insert Mutant into New Population New Population Population

i:=i+1

Figure 2. Genetic Programming diagram (Koza 1992)

2.3.1. Create initial random population Create initial random population process will generate a number of individuals within a population consisting of set functions and terminal sets that are generated randomly. One individual describes a form of model or rule to be sought. An example of the rules is shown in Figure 3.

OR

AND AND

NOR NOR AND AND

X90 OR OR AND X46 X73 OR AND

X11 X83 X16 X83 X26 X72 X26 X15 X86 X93

Figure 3. Sample model or identification rules

doi: 10.17700/jai.2018.9.1.413 52 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

2.3.2. Evaluate fitness Fitness evaluation is the ratio of the number of errors in predicting the actual results. The fewer number of errors in an individual, the better the individual values are formed. In this research, the fitness value search algorithm by inserting data booleanize results to the rules or individuals selected from the process ‘Create initial random population’. For example the rules generated in Figure 3 and the evaluation data in Table 4, the rules are consist of into 26 data. Six individuals in class 1, 12 individuals in class 2 and 8 individuals in class 3. In the evaluation process the rules in Figure 3 do predictions with result 11 is high class. This means that the rule has 3 strokes with an accuracy of 72.73%

Table 4. Example of fitness evaluation Class X11 X15 X16 X26 X46 X72 X73 X83 X86 X90 X93 Prediction 1 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 1 1 1 0 2 0 0 0 0 0 0 1 0 1 1 1 0 2 0 0 0 0 0 0 1 0 1 1 1 0 2 0 0 0 0 0 0 1 0 1 1 1 0 2 0 0 0 0 0 0 1 0 1 1 1 0 2 0 0 0 0 0 0 1 0 1 1 1 0 2 0 0 1 0 0 0 0 0 1 0 0 1 2 0 0 1 0 0 0 0 0 1 0 0 1 2 0 0 1 0 0 0 0 0 1 0 0 1 2 0 0 0 0 0 0 0 0 0 1 1 0 2 0 0 0 0 0 0 0 0 0 1 1 0 2 0 0 0 0 1 0 0 0 0 1 1 0 2 0 0 0 0 1 0 0 0 0 1 1 0 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1 3 0 0 0 0 0 0 0 1 1 0 0 1

2.3.3. Genetic operation Genetic operators commonly used in Genetic Programming are elitism, crossover and mutation (Carvalho et al. 2012). The process of genetic operation begins with the selection of rules using tournament method. This method is done by taking four rules at random then compared to taking one best rule. Operation elitism will take one best rule to be copied into the new population. Crossover operations will take two of the best rules and genetic exchanges are made. This crossover example is shown in Figure 4.

doi: 10.17700/jai.2018.9.1.413 53 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

NOR NOR

OR AND X90 OR

X26 X15 X86 X93 X26 X72

NOR NOR

OR OR X90 AND

X26 X15 X26 X72 X86 X93

Figure 4. Crossover evaluation example The mutation process will take one best rule to make a gene change from the rule. The mutation process can be seen in Figure 5

NOR NOR

OR AND OR AND

X26 X15 X86 X93 X9 X15 X86 X93

X9 Figure 5. Mutation operation example

3. Results and discussion All physical aspects of morphology that have been coded using booleanization process and have been divided into training data and test data using k-fold cross validation according to proportion, then the training process from trainer data according to genetic operation to produce model or classification rules in each class. The parameters used in this training process as shown in Table 5. The results of this process will be done in the next process for examining the data test which has been previously divided.

Table 5. Operation value of Genetic Programming Parameter Trial Number of generations 5, 10, 20 Population Size 1000, 10.000 Crossover 0.9 Mutation 0.1 Depth of the tree 5, 7 Max node 25, 30 Function set AND, OR, NOR

doi: 10.17700/jai.2018.9.1.413 54 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Three classes consisting of 129 cassava plants, 96 attributes of the founder of the training process of Genetic Programming produce the model or classification rules shown in Figure 6 below.

a) High class

AND

AND OR

OR OR AND NOR

NOR NOR X74 X8 OR OR AND OR

X73 X77 X92 X6 X87 X22 X75 X28 X93 X68 X56 X82

Figure 6. High class rule IF The outer peel of the tubers is dark brown (X74) OR The color of shoots is dark purple / purple black (X8) AND The epidermis color is light yellow (X73) NOR is the color of stalk top beige (X77) OR Flowering (X92) NOR the color of shoot of purplish green (X6) AND The outer peel of the tubers is yellowish (X87) OR the color of the top stems is green (X22) AND Peel color outer pink tuber (X75) OR Color of the upper red leaf stalk (X28) OR The sap of the tuber is little (X93) AND number of branches more than two (X68) NOR The color of the old stem is brownish brown (X56) OR thickness of peel tubers is equal to 1.01mm (X82) Then High class The identifier of the high class can be seen in Figure 6 above. There are 14 identifiers with a combination of 3 operators AND, OR and NOR. At the first level there is an AND operator, which means it will be true if the two inputs of the two operators below (AND and OR) are true. At the second level there is a combination of OR and AND operators, the OR operator means that it will be true if one of the below operator inputs (AND and NOR) is true. At the third level there are three combinations of operators (AND, OR and NOR), the NOR operator will be true if the two inputs below are the result of the OR operator (with the characteristic of the old brownish brown stem (X56), the same small thickness of peel tuber of 1.01mm (X82)) and AND (with a small sap bulb (X93), the number of branches over two (X68)) is false. There are two founders on the fourth level of the Outer Peel Brown (X74) and the color is dark purple shoot / blackish purple (X8). These two identifiers with the OR operator indicate that one of them must be true. On the other hand, at level five with the NOR operator, it means that this high class does not have dark brown (X73) the peel is brown (X77), flowering (X92) and the shoot color is purple green (X6). Furthermore, with the OR operator, it is clear that one of the markers should be true. The color of the epidermis is Yellowish (X87), the color of the top stems is green (X22) and the outer color of the tuber is pink (X75), the color of the top leaf is red (X28).

doi: 10.17700/jai.2018.9.1.413 55 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

b) Medium Class

OR

OR AND

OR OR NOR AND

AND AND NOR AND AND X73 NOR X15

X17 X49 X84 X56 X78 X42 X3 X82 X43 X95 X39 X51

Figure 7. Medium class rule

IF The color of the lower stalk leaf yellowish green (X15) AND Thickness of tuber peel higher and equal to 1.33mm (X84) NOR Color of stem is dark silver brown (X56) AND Outer peel of tuber is dark brown (X73) NOR The color of lower stalk green / purple (X17) AND light green stem color (X49)) OR Peel color inside tuber is yellowish (X78) AND weight of leaf higher than or equal to 0.62 g (X42) OR Texture of leaf vein is clear (X3) AND Thickness of peel tuber lower than or equal to 1.01 (X82) OR Leaf form is wide tapered (X43) NOR much sap (X95) OR Leaf form is thin (X39) AND Color of young stem is redness/ purplish green (X51) THEN Medium class The identifier of the medium class can be seen in Figure 7 above. There are 14 identifiers with a combination of 3 operators AND, OR and NOR. At the first level there is an OR operator which means it will be true if one of the two inputs below (OR and AND) is true. At the second level there are OR and AND operators. The AND operator will be true if the two underlying two inputs (NOR and AND) are true. At the third level there is a combination of three operators (AND, OR and NOR). The NOR operator will be true if the two inputs are wrong, as seen on the four levels of the medium-class performers not having a dark brown tuber outer peel (X73). In contrast, the medium-class branding must have the coloring of the lower yellowish green stalk (X15). The identities that are not owned by the medium class are also found at level five, i.e., the thickness of the large tuber peel of 1.33mm (X84) and the old brownish brown stem color (X56). For the characteristic of the green / purple (X17) stem color and light green stem color (X49) must be true one of them but not the true value of both. Leaves of leaf shape (X39), the color of green stems reddish / purple (X51) must be either true value or the characteristic of the tapered fat leaf (X43), gummy (X95) does not have both. As for the identification of peel color in yellowish tubers (X78), the weight of the same large leaves of 0.62 gr (X42) has true value of both or true value both for the identification of the clear leaf bone (X3), thickness of the same small tuber peel of 1.01 (X82) .

doi: 10.17700/jai.2018.9.1.413 56 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

C) Low Class NOR

NOR OR

AND AND OR AND

NOR OR NOR OR X83 X82 OR X90

X88 X45 X58 X61 X81 X36 X37 X32 X57 X78

Figure 8. Low class rule

IF Thickness of epidermis higher than or equal to 0.54mm (X90) AND Colour of inner peel is yellowish (X78) OR Colour of old stem is silver and red (X57) OR Thickness of peel is between 1.01mm to 1.33mm (X83) OR Thickness of peel lower or equal to 1.01mm (X82)) NOR Width of leaf lower than or equal 3.7 cm (X32) OR Height of leaf higher than or equal to 17.4 cm (X37) AND Flesh of tuber is yellowish (X81) NOR Width of leaf is between 14.3 to 17.4 cm (X36) NOR Thickness of epidermis higher than or equal to 0.28 (X88) NOR Form of leaf tip is taper (X45)) AND Distance of young stem segment is lower than or equal to 35.78 mm (X58) OR Distance of old stem segment is lower than or equal to 74.5 mm (X61) THEN Low class

The identifier of the lower classes can be seen in Figure 8 above. There are 13 identifiers with a combination of 3 operators AND, OR and NOR. At the first level there is a NOR operator which will be correct if both inputs below it (NOR and OR) increase incorrectly. At the second level there are NOR and OR operators. On the OR operator will happen either one or both of the inputs are correct. At the third level there are two combinations of AND and OR operators, the AND operator will verify correctly if both of the entries below are correct. Low class grain is a large thickness of the same thickness of 0.54mm (X90), peel color in yellowish bulb (X78) Silver stem color and red (X57), tube peel thickness 1.01mm to 1.33mm (X83), Thickness peel of small bulbs equal to 1.01mm (X82), The width of the same small leaf of 3.7 cm (X32), The length of the leaves of the same large leaves of 17.4 cm (X37), Yellowish yellow flesh color (X81), the width of the leaf between 14.3 and 17.4 cm (X36), (thickness of the large bulb of the same bulb of 0.28 (X88), the shape of the tip of the taper leaf (X45), (small yellow stems of 35.78 mm (X58) same stem segment of 74.5 mm X61).

doi: 10.17700/jai.2018.9.1.413 57 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

4. Evaluation The Cassava data which are divided into three classes based on its HCN content i.e. low class containing less HCN 50 mg / kg, medium class containing HCN between 50 and 80 mg / kg and high class containing HCN more than 80 mg/ kg. Each class is divided into train data and test data using K- Fold Cross Validation with K = 5. In fold 1 there are 104 data train and 25 data are used as test data. The results of each fold can be seen in Table 6.

Table 6 Experiment result of each class High class Fold1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 Fold2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 Fold3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 Fold4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 Fold5 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 1 1 1 1 1 1 1 1 Medium Class Fold1 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 Fold2 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 Fold3 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 Fold4 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 Fold5 0 0 0 0 0 0 1 1 1 1 1 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 Low class Fold1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 Fold2 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Fold3 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Fold4 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Fold5 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Experiments conducted repeatedly, and experiments are best taken as classification rules. There are five rules according to the distribution of K-Fold Cross Validation data generated by each class in each fold. The accuracy level generated on each fold using the confusion matrix table of the model or the rules generated by the genetic programming process in each respectively, as shown in Table 7 to Table 11 below

Table 7. Confusion Matrix Fold 1 Actual Class Fold 1 Outside the Low Medium High classes Low 6 0 1 0 Medium 0 11 0 0 Predicted High 0 0 8 0 Class Outside the 0 0 0 0 classes In fold 1 there were 25 test data consisting of 6 low class data, 11 medium classes and 8 high classes. As seen in Table 7 above there was one mistake, i.e. one high-class sweet potato data also detected as low class. This error occurs because of the similarity of the data of cinnamon plant founder in high class with low class. This error occurs on the type of cassette tuber. The accuracy for identification in fold 1 is 96.15%, obtained from the following calculations:

6 + 11 + 8 1 = 100 = 96.15% 6 + 11 + 8 + 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑥𝑥

doi: 10.17700/jai.2018.9.1.413 58 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Table 8. Confusion Matrix Fold 2 Actual Class Fold 2 Outside the Low Medium High classes Low 6 0 0 0 Medium 0 12 1 0 Predicted High 0 0 8 0 Class Outside the 0 0 0 0 classes There are 26 data of cassava that used as test data on fold 2, 26 data that consist of 6 data of low class, 12 medium class and 8 high class. Seen in Table 8 above there was one mistake, one high-quality cassava data also detected as a medium class. This error occurs because of the similarity of the data of cassava cultivator in high class with medium class. This error occurs in the white Tailan cassava type. The accuracy for identification in fold 2 is 96.30%, obtained from the following calculations.

6 + 12 + 8 2 = 100% = 96.30% 6 + 12 + 8 + 1 𝐴𝐴𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑥𝑥 Table 9. Confusion Matrix Fold 3 Actual Class Fold 3 Low Medium High Outside the classes Low 6 0 0 0 Medium 0 12 0 0 Predicted 0 0 8 0 Class High Outside the 0 0 0 0 classes Test data on the fold 3 contained 26 data of cassavaes consisting of 6 low class yam, 12 medium and 8 high class. In Table 9 above there is no error. The accuracy for identification in fold 3 is 96.15%, obtained from the following calculations 3 = x 100% = 100% 6+12+8 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴Table𝑎𝑎𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 10.𝑓𝑓𝑓𝑓𝑓𝑓 Confusion𝑓𝑓 6+ 12Matrix+8 Fold 4 Actual Class Fold 4 Low Medium High Outside the classes Low 6 0 0 0 Medium 0 12 0 0 Predicted High 0 0 8 0 Class Outside the 0 0 0 0 classes It can be seen in Table 10, fold 4 with 26 cassava data as test data consisting of 6 low class yam, 12 medium and 8 high class, genetic programming process in generating excellent model or rule, yielding 100% accuracy rate with calculation as following Accuration fold 4= x 100% = 100% 6+12+8 6 +12+8

doi: 10.17700/jai.2018.9.1.413 59 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Table 11. Confusion Matrix Fold 5 Actual Class Fold 5 Low Medium High Outside the classes Low 6 5 0 0 Medium 0 9 0 0 Predicted High 0 3 8 0 Class Outside the 0 0 0 0 classes Seen in Table 11 there are 2 types of errors from the 26 test data consist of 6 in low class, 12 in medium class and 8 in high class. Both types of these errors occur in the medium class. The first mistake that there are three data of class yam is identified to the high class, this type of cassava is curly curl whose data is taken from Pekanbaru riau. The second mistake is that five data of cassava is being detected as low class, this type of cassava is yam lantak. The accuracy for identification on this fold is 83.87%, obtained from the following calculations. Accuration fold 5= x 100% = 83.87% 6+9+8 Evaluation of the performance of the system6 obtained+9+8+3+ 5can be calculated from the average accuracy value of all the fold very good result that is equal to 95.26%, with the following calculation

. . . = x 100% = 95.26% 96 15+96 30+100+100+83 87 5. Conclusion 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 5 Cassava contains a toxin called cyanide acid (HCN). In this study the cyanide acid content of cassava was classified into three classes (low containing 50 mg/kg HCN), medium (containing HCN between 50 and 80 mg/kg) and high (containing HCN more than 80 mg/kg). The cassava identification system by applying a heuristic search algorithm using genetic operations produces a simple and structured identification model and can be used to locate classification rules with good accuracy. These three classes are divided into training data and test data by using K-fold cross validation technique with K = 5, genetic programming process using AND, OR and NOR operators and as many as 96 identification is done repeatedly to get the best model or rule, the best performance accuracy were derived at 95.26%.

Acknowledgment The authors would like to thank Ministry of Research Technology and Higher Education of Indonesia for funding this research through applied product research scheme 2017.

References BB-Biogen 2010 ‘Buku Katalog Plasma Nutfah Tanaman Pangan 2010’ Balai Besar Litbang Biogen. Bogor. Bramer M. 2007 ‘Principles of Data Mining. London (GB)’ Springer. Carvalho M.G, Laender A.H, Goncalves M.A, and Silvia A.S. 2012 ‘A Genetic Programming Approach to Record Deduplication’ IEEE Transactions on Knowledge and Data Engineering, Vol. 24, No. 3. https://doi.org/10.1109/TKDE.2010.234 Hanifah VW, Yulistiani D dan Asmarasari SAA. 2010. Optimising the Use of Cassava Pell as Feed for the Empowerment of Enye-Enye Farmers. Seminar Nasional Teknologi Peternakan dan Veteriner

doi: 10.17700/jai.2018.9.1.413 60 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant Journal of Agricultural Informatics (ISSN 2061-862X) 2018 Vol. 9, No. 1:47-61

Koza JR 1992 ‘Genetic Programming On the Programming of Computers by Means of Natural Selection’ London (US): MIT Press Laksmana I, Herdiyeni Y, Zuhud EAM. 2013 ‘Genetic Programming for Medicinal Plant Family Identification System’ Journal Of Research and Applications, vol 4 no (3), pp. 217-234 Luke S. 2000. Two Fast Tree-Creation Algorithms for Genetic Programming. IEEE Transactions on Evolutionary Computation, Vol. 4, No. 3. USA: Department of Computer Science, University of Maryland. https://doi.org/10.1109/4235.873237 Lukas, Iskandar.A. 2008 ‘Permainan Catur Akhir King-Rook-King (KRK) Menggunakan Pemrograman Genetika’ Konferensi Nasional Sistem dan Informatika (2008 Nov 15); Bali, Indonesia (ID)KNS. 328-334 Nakamura K, Morita T and Yamaguchi. 2017. ‘A User-Centric Platform PRINTEPS to Develop Integrated Intelligent Applications and Application to Robot Teahouse’ Procedia Computer Science. Vol. 112, pp 2309- 2318. https://doi.org/10.1016/j.procs.2017.08.266 Purwono and Heni Purnamawati 2009. ‘Budidaya 8 Jenis Tanaman Pangan Unggul’ Penebar Swadaya Qurat-ul-ain, Latif G, Kazmi SB, Jaffar MA, Mirza AM. 2010 ‘Classification and segmentation of brain tumor using texture analysis’ Proceeding AIKED'10 Proceedings of the 9th WSEAS international conference on Artificial intelligence, knowledge engineering and data bases. Wisconsin (US). pp 147-155. Sakprasat S, Sinclair MC. 2007 ’Classification rule mining for automatic credit approval using genetic programming’ IEEE Congress on Evolutionary Computation 7 pp548-555. Sudharmono M. 2012 ‘Pengenalan Varietas kunyit berdasarkan ciri fisik menggunakan Genetic Programming dengan Praproses Booleanizing’. Departemen Ilmu Komputer FMIPA Institud pertanian Bogor Unigwe, Cyprian Robinson, Raji, Ademola Moshood, Ajayi, John Olurotimi, Popoola, Abiola Moshood, Balogun, Fatima Adeola and Adekunle, Femi Olayinka. 2017. ‘Carcass, Organ Weights and Histo-morphology of Internal Organs of Sows Fed Fermented and Enzyme Supplemented Cassava Peels Meal (CPM) Based Diets’ Journal of Plant and Animal Sciences Vol. 2 No 1, pp. 026-036. Wahyudi E.N 2013. ‘Teknik Klasifikasi untuk Melihat Kecenderungan Calon Mahasiswa Baru dalam Memilih Jenjang Pendidikan Program Studi di Perguruan Tinggi’ Jurnal Teknologi Informasi DINAMIK. I vol 18. pp 55- 64.

doi: 10.17700/jai.2018.9.1.413 61 Indra Laksmana, Rosda Syelly, Nurzarah Tazar, Perdana Putera: A Genetic Programming Study on Classification of Cassava Plant