Volume 34, Issue 4 p. 446-456
AWARD PAPER
Open Access

Perspectives on biodiversity informatics for ecology

Takeshi Osawa

Corresponding Author

Takeshi Osawa

Graduate School of Urban Environmental Sciences, Tokyo Metropolitan University, Tokyo, Japan

Correspondence

Takeshi Osawa, Graduate School of Urban Environmental Sciences, Tokyo Metropolitan University, Minami-Osawa 1-1, Hachiouji, Tokyo 192-0397, Japan.

Email: [email protected]

Search for more papers by this author
First published: 04 July 2019
Citations: 19
*Takeshi Osawa is the recipient of the 21st Denzaburo Miyadi Award.

Abstract

Biodiversity informatics is the application of informatics techniques to ecology and biodiversity sciences. The premise is utilizing natural history collections/data, such as specimens and biodiversity observations, with information and communication technology. During the past two decades, biodiversity informatics has improved dramatically and has been applied increasingly in ecology science. In this paper, I review biodiversity informatics for ecology with a particular focus on the treatment of data. First, I discuss the traditional perspective of data collection and its usage in the ecological community. Then, I provide an overview of the trajectory of changing perspectives of data treatment relative to developing biodiversity informatics (i.e., the infiltration of the data reuse and sharing concepts). Specifically, I discuss the significance of data reuse, which offers numerous benefits for research, and data sharing with their supporting mechanisms and case studies. Finally, I discuss potential future developments in biodiversity informatics relative to the field of ecology.

1 INTRODUCTION

In the 21st century, we live in the digital information era. Information and communication technology (ICT) has changed our daily lives drastically, including academic fields. For example, we can now access research articles, derive large scientific data and conduct complex analyses through the Internet from our offices or homes. This new tide is strongly influencing the ecology and biodiversity science fields (Bisby, 2000; Edwards, 2000).

Biodiversity informatics (Canhos, de Sausa, De Giovanni, & Canhos, 2004; Soberón & Peterson, 2004) is an advanced research field with strong contributions from ICT, particularly the Internet (Bisby, 2000; Soberón & Peterson, 2004). Biodiversity informatics is the application of informatics techniques to facilitate the improved management, presentation, discovery, exploration and analysis of biodiversity information (Berendsohn et al., 2011; Canhos et al., 2004; Soberón & Peterson, 2004). Bisby (2000) has stated that the central goal of biodiversity informatics is to develop systems that permit interoperability and knowledge synthesis across a broad range of local systems and to embed such systems in global knowledge architectures. In other words, biodiversity informatics could be the basis of modern biodiversity science, including ecology (Figure 1). In the preliminary development of this research field around the year 2000, numerous researchers discussed the broad potential of this field as an extension of related research (Bisby, 2000; Canhos et al., 2004; Edwards, 2000; Salski & Recknagel, 2003; Soberón & Peterson, 2004).

Details are in the caption following the image
Basic concept of biodiversity informatics for ecology and related fields

Even though the development of biodiversity informatics has been driven by recent ICT advancements, the content of biodiversity informatics is essentially that of natural history, including specimen collections and living collections (botanical and zoological gardens and culture collections), as well as data collections from research, citizen science activities, literature and others (Berendsohn et al., 2011) (Figure 1). In fact, the initial concept for the Global Biodiversity Information Facility (GBIF), one of the central institutions of this field, is a framework for facilitating digitization and making isolated biodiversity records (primarily specimens) more interoperable (Edwards, 2000). Science Museum Net (http://science-net.kahaku.go.jp/?ln=en), one of the largest biodiversity databases in Japan (Osawa, 2017), has integrated specimen records from approximately 100 museums across Japan. Natural history is the basis of natural science, and collecting specimens is inarguably the first step in natural science (Osawa, 2017; Troudet, Vignes-Lebbe, Grandcolas, & Legendre, 2018; Watanabe, 2016). In addition, we currently have access to other natural history data via ICT, such as digital photographs, digital sounds and digital movies. Together with traditional approaches and new technology, we can establish and implement innovative biodiversity informatics approaches into biodiversity science.

In the first two decades of the 21st century, biodiversity informatics has developed remarkably. The types and volume of available biodiversity data have increased dramatically, and we can use multiple datasets to perform case studies. Surprisingly, such case studies cover a broad range of ecology and biodiversity sciences. Thus, biodiversity informatics is now becoming an important infrastructure in ecology and biodiversity sciences. In this paper, I review recent progress, achievements and challenges in biodiversity informatics relative to ecology and biodiversity sciences, with a specific focus on the treatment of data.

2 TRANSITION OF BIODIVERSITY DATA

Biodiversity is distributed across the globe, with the highest concentration of diversity in the tropical regions, especially in developing countries and in the oceans (Edwards, 2000). Researchers have long collected specimens as a primary form of biodiversity data to better understand biodiversity. Such specimens have primarily been methodically collected by taxonomists as an initial study step (Troudet et al., 2018). Biodiversity data from specimens have high quality and reliability because they can be verified based on re-examinations of stored specimens as vouchers (Osawa, 2017; Watanabe, 2016). Many ecologists have also collected specimens of their target species. Such specimens have primarily been collected by major centers in developed countries, such natural history museums, herbaria and microorganismal repositories (Edwards, 2000), rather than in their region of origin.

In recent years, unvouchered observational records (i.e., observations with no link to any tangible material) have primarily been collected rather than actual specimens (Gaiji et al., 2013; Troudet et al., 2018). In fact, in Japan, the number of collected specimens peaked in 2000 for Science Museum Net, which is one of the largest specimen-based biodiversity databases, and the unvouchered national biodiversity web database (http://www.biodic.go.jp/) of the Biodiversity Center of Japan (J-IBIS) was launched at approximately the same time (Osawa, 2017). Unvouchered observations and vouchered specimens are biodiversity occurrences with different fundamental natures, each having their own assets and liabilities (Troudet et al., 2018). In other words, each type of dataset has strong points and weak points. For example, well-curated specimens facilitate a wide range of studies and analyses, such as multimedia analyses (Kano, Nakajima, Yamasaki, Kitamura, & Tabata, 2018) and DNA sequence analyses; however, significant effort is required to collect and maintain such specimens (Edwards, 2000; Troudet et al., 2018) (Figure 2). Conversely, with unvouchered observations, data accumulation is much faster compared to accumulating specimens, and huge amounts of data can be collected via the Internet in, for example, crowdsourcing approaches (Osawa, Baba, Suguro, Naya, & Yamauchi, 2017); however, the data quality can be problematic (Bisby, 2000; Kitchin, 2014) (Figure 2). eBird (https://ebird.org/home), a famous project that gathers bird observations (Sullivan et al., 2009), had approximately 500 million bird observation records as of 2018 (Kelling, 2018). However, unvouchered observational records cannot be used for some studies, such as DNA sequencing, and may include incorrect records due to identification errors (Kremen, Ullman, & Thorp, 2011; Osawa & Wada, 2016) (Figure 2). Even though there are some techniques to relieve misidentification problems (Miller et al., 2011; Molinari-Jobin et al., 2012), identification errors are an inherent weak point of unvouchered records. Ideally, collecting data using vouchered specimens is required for specific purposes and to ensure reliability (Osawa, 2017; Watanabe, 2016). In practice, however, unvouchered observations have several advantages, including the ease of collection, storage (a large archival space is not required), and simple maintenance (Troudet et al., 2018). Currently, many specimens and unvouchered observational records have been digitized and these data can be accessed over the Internet (Figure 2).

Details are in the caption following the image
Conceptual diagram of differences between vouchered (specimen) records and unvouchered observational records

3 NEW BIODIVERSITY DATA USAGE PARADIGM

Prior to the current information era, field ecologists conducted observation-based research and several well-known Japanese ecologists achieved notable results (Inoue & Marsura, 1983; Kato, Takimura, & Kawakita, 2003; Kikuzawa, 1991; Nakano & Murakami, 2001). Field observation has always been an important method in ecology, and, historically, ecologists have collected and managed their own research data. However, with the advent of the information era, a new paradigm has gradually emerged (i.e., data reuse and sharing). Sharing data eliminates the need to repeatedly collect data, thereby providing an additional return on the original investment (Parr & Cummings, 2005).

Data reuse and sharing were initially adopted in the molecular domain, for example, molecular genomics (Insel, Volkow, Li, Battey, & Landis, 2003), and then gradually spread to ecology (Parr & Cummings, 2005). Some ecologists accustomed to collecting and managing their own data were initially reluctant to share and reuse data (Figure 3). In addition, other than summaries posted in publications, most data were not made publicly available and, over time, many datasets were lost, even to the original collectors (Whitlock, 2011). However, recently, many ecologists have become increasingly aware of the importance of data reuse and sharing and have launched some supporting systems such as data repositories (Costello, 2009; Costello, Michener, Gahegan, Zhang, & Bourne, 2013; Whitlock, 2011).

Details are in the caption following the image
Differences between traditional and new data treatment methods

4 ECOLOGICAL RESEARCH BASED ON EXISTING DATA

Even though data reuse and sharing provide several benefits (Costello, 2009; Whitlock, 2011), the most obvious advantage is increased data availability. In addition, advances in computational ability and analysis techniques have expanded the scope of ecological studies both spatially and temporally.

4.1 Spatially expanded studies

Macroecology research, which deals with ecological patterns and processes at large spatial scales (Gaston & Blackburn, 2008), can benefit from greater volumes of available data. It is virtually impossible for individuals or relatively small research groups to conduct field observations on a national scale. However, national- and global-scale data can be obtained by integrating multiple datasets collected and managed by different researchers (Figure 3). For example, García-Roselló et al. (2015) have predicted global distribution patterns of marine fish and Padalia et al. (2014) have predicted the potential invasion range of alien invasive species in India. Together with my colleagues, I have studied the relationships between the distribution of threatened plants and agricultural land use in Japan (Osawa, Kamauchi, Hosoya, & Ito, 2013; Osawa, Kohyama, & Mitsuhashi, 2013; Osawa, Kohyama, & Mitsuhashi, 2016a; Osawa & Mitsuhashi, 2017). For these studies, we used threatened plant distribution data collected by the Japanese Ministry of Environment to establish a national red list (https://ikilog.biodic.go.jp/Rdb/booklist). In the 2010s, the number of macroecological studies (based on integrated reusable data) covering the entire Japanese archipelago increased dramatically (Amano et al., 2011; Kadoya & Washitani, 2011; Naoe et al., 2015; Saito, Kurashima, & Ito, 2016; Suzuki-Ohno, Yokoyama, Nakashizuka, & Kawata, 2017; Yoshioka, Fukasawa, Mishima, Sasaki, & Kadoya, 2017).

4.2 Temporally expanded studies

Generally, time-series data provide more information compared to single-time data. For example, time-series data with manipulation of studied system can be used to evaluate hypotheses (Walters & Green, 1997). From a conservation perspective, time-series monitoring data can provide information about a system's response to management actions (Yoccoz, Nichols, & Boulinier, 2001). However, time-series observations, such as monitoring, require significant effort. By combining original data with available reusable data, time-series data can be obtained relatively easily (Figure 3). For example, Moritz et al. (2008) quantified the impact of approximately one century of climate change on a small mammal community by resampling census data collected in the early 1900s. Cameron et al. (2011) evaluated the decline of bumblebees using historical specimen records (1900–1999) and a current intensive nationwide North American survey (2007–2009). My colleagues and I conducted a study that combined our specimen records and historical specimen data to evaluate habitat stability for Carabidae insects, which are frequently used as an environmental indicator species in the Kanto region of Japan (Osawa, Jinbo, & Iwasaki, 2014; Osawa, Watanabe, Ikeda, & Yamamoto, 2014). In that study, we collected specimen records and reused historical records from the 1940s to the 1970s (Yoshitake, Kurihara, Yoshimatsu, Nakatani, & Yasuda, 2011). Note these records were originally collected before I was born.

4.3 Biodiversity monitoring data by government and private business sectors

Currently, biodiversity conservation and sustainable use for ecosystem services are important social issues (Balmford et al., 2005). Therefore, these issues are of interest to researchers, government agencies and private businesses, that is, contractor and/or corporate social responsibility businesses (MacDonald, 2010). Recently, governments and private businesses have frequently been involved in monitoring biodiversity, for example, to evaluate biodiversity loss (Yoccoz et al., 2001) and to conduct alien invasive species census surveys (McNeely, 2001). Data collected in these activities are also used for ecological research. For example, along with my colleagues, I have evaluated the dispersion characteristics of alien invasive plants in a riparian area based on a local government dataset (Osawa, Mitsuhashi, & Niwa, 2013). This dataset was comprised of distributional records for 10 alien invasive plants that covered riparian vegetation along 680 km of alluvial river sections (Osawa, Mitsuhashi, et al., 2013). We also performed a distribution analysis of alien plants in part of the Fuji-Hakone-Izu National Park in the Hakone region (Akasaka, Osawa, & Ikegami, 2015) and an expansion risk analysis on an alien snail Helix aspersa (Osawa & Omae, 2016). In addition, based on data collected by government agencies and private businesses, we evaluated vegetation damage caused by Sika deer (Osawa, Igehara, Ito, Michimata, & Sugiyama, 2015; Osawa, Kadoya, & Kohyama, 2015). However, Yoccoz et al. (2001) pointed out that government agencies and private businesses do not always pay sufficient attention to several fundamental aspects of the monitoring process, such as (a) the motivation for the monitoring, (b) the species to be monitored and (c) the monitoring method. Currently, several techniques can be used to analyze the monitoring data. We can conduct conservation research in collaboration with governments and/or private businesses, and the monitoring data collected by governments and private businesses can provide valuable resources for ecological research.

4.4 Citizen science data

Historically, natural history work, such as specimen collection and curation, has been performed by nonacademics (Balke et al., 2013; Hopkins & Freckleton, 2002). ICT has expanded the work of amateurs, particularly the collection of regional biodiversity records (Kobori et al., 2016; Osawa, Yamanaka, et al., 2017; Silvertown, 2009). Recently, citizen science approaches (e.g., field observations by members of the public) have generated increasing interest in ecology and biodiversity sciences (Kobori et al., 2016; Osawa, Yamanaka, et al., 2017). One important research benefit provided by citizen science is the crowdsourcing of large datasets (Osawa, 2015; Osawa, Yamanaka, et al., 2017; Suzuki-Ohno et al., 2017). Suzuki-Ohno et al. (2017) estimated the potential distribution ranges of six bumblebees based on photographs taken by citizens throughout Japan. I have also analyzed broad distribution records collected from six prefectures using more than 1,500 barn swallow nest points collected by more than 100 citizens to identify the nesting preferences of this species (Osawa, 2015). Biodiversity informatics can play an important role in linking amateurs and biodiversity researchers (Osawa, Yamanaka, & Nakatani, 2013). In addition to its scientific merits, an important aspect to consider concerning citizen science approaches is its educational advantages, that is, citizen science provides members of the public opportunities to observe and interact with organisms in nature (Dickinson et al., 2012; Dickinson, Zuckerberg, & Bonter, 2010; Kobori et al., 2016; Miyazaki, Murase, Sahara, Angulo, & Senou, 2017; Sasaki, Ohnishi, & Osawa, 2016).

4.5 Available map data

Relative to data reuse, we can identify several types of nonbiological data that can potentially contribute to ecology and biodiversity research. One of the most important types of nonbiological data is map data, particularly land use and land cover maps. The importance and application of land use and land cover data are particularly emphasized in the estimation of biodiversity and ecosystem services (Akasaka et al., 2014; Turner et al., 2015). One of the most widely applied methods is the use of vegetation or habitat maps to stratify units of diversity (Akasaka et al., 2014; Dalleau et al., 2010). Akasaka et al. (2014) reviewed currently available map data in Japan and identified several types of available map data. However, for such data, integration and detailed, standardized classifications are required (Akasaka et al., 2014). Accordingly, Ogawa et al. (2013) established a comprehensive land use classification map. In addition, my colleagues and I have established and released a detailed agricultural land use map based on statistics (Osawa, Igehara, et al., 2015; Osawa, Kadoya, & Kohyama, 2015). The number of available land use and land cover maps is increasing worldwide (Arino et al., 2012; Bontemps et al., 2011; Miettinen, Shi, Tan, & Liew, 2012; Morton et al., 2011; Sano, Rosa, Brito, & Ferreira, 2010), and such data are contributing to many studies (e.g., Ohnishi, Osawa, Yamamoto, & Uno, 2019) because such datasets can be reused by multiple researchers.

5 A DATA REUSE SUPPORT MECHANISM

As previously discussed, data reuse has several merits for ecology and biodiversity sciences. In addition, data reuse has other benefits, such as making “unmined” data available, avoiding data duplication, reducing the burden on researchers, improving the transparency of research procedures, providing alignment with open access principles and recognizing that the outputs of publicly funded research are public assets (Bishop, 2009; Fry, Lockyer, Oppenheim, Houghton, & Rasmussen, 2008). However, mechanisms are required to support and improve these benefits. In this section, I discuss an important support mechanism for data reuse (i.e., data sharing).

5.1 Establishing a web database

Data integration and the establishment of databases have significant value for ecology and biodiversity sciences (Muto-Fujita et al., 2017). With the development of ICT, the number of web-based databases that provide multiple data types has been continuously increasing. The GBIF is a central biodiversity database project (https://www.gbif.org/) (Edwards, 2000) with more than 1,000,000,000 biodiversity records, including occurrences, taxon names, species profiles and records from ecological fieldworks, and users can derive and use these records as open data (described in greater detail in a later section). The Japanese node of the GBIF (hereafter JBIF) has a portal website (http://www.gbif.jp/v2/) that supports biodiversity data searches of the GBIF database using the local language, that is, Japanese. In addition, JBIF itself opened with more than 6,000,000 biodiversity records of both specimen-based and observation-based international data, that is, in English. These records were contributed by several organizations and participants, such as natural history museums in Japan in collaboration with the Science Museum Net (http://science-net.kahaku.go.jp/), the J-IBIS unvouchered national biodiversity web database (http://www.biodic.go.jp/), researchers (Osawa, Baba, et al., 2017; Osawa, Yamanaka, et al., 2017) and citizen scientists (Osawa, 2013; Osawa & Wada, 2016).

iNaturalist (https://www.inaturalist.org/) is a central social network website for collecting and managing biodiversity data based on human observations. iNaturalist users can pool and share several types of biodiversity observational records, such as photographs, sounds and simple observations, and these records can be easily mapped. In addition, iNaturalist users can establish projects that focus on specific areas or taxa. Even though observational records tend to suffer from quality problems such as misidentification (Kremen et al., 2011; Osawa & Wada, 2016), iNaturalist can be used to divide data into “casual” and “research grade” levels to control the data quality (Bowser, Wiggins, Shanley, Preece, & Henderson, 2014). Therefore, iNaturalist can be used for biodiversity research and nonacademic purposes. Currently, several biodiversity data-sharing projects have been launched not only based on natural history data but also based on other materials such as literature sources, DNA barcoding and animal tracking (Hobern et al., 2013).

5.2 Interoperability of data/database

Even though integrating scattered biodiversity data, for example, establishing a database, is useful for research, scattered databases are ineffective from the user perspective. Therefore, one of the current concepts related to improving databases is interoperability (Berendsohn et al., 2011). An important issue relative to database interoperability is standardization. Standards provide a consistent representation of the data to be shared, enabling data from different sources to be combined while minimizing data loss or duplication (Berendsohn et al., 2011). Note that data records written in the same format are easy to integrate. The Biodiversity Information Standards (TDWG) organization plays a central role in defining and promoting data standards and protocols to support interoperability between disparate and locally distributed systems (Berendsohn et al., 2011). They have established and improved the widely used Darwin Core biodiversity data standard format (Wieczorek et al., 2012). Darwin Core is currently adopted as the data standard for the GBIF community, as well as other communities and organizations, such as the Ocean Biogeographic Information System (http://www.iobis.org/), the Nordic Genetic Resource Center (http://www.nordgen.org) and Biodiversity International (formerly IPGRI) (http://www.bioversityinternational.org). In Japan, IKIMONO LOG (https://ikilog.biodic.go.jp/), a central biodiversity observation database (Osawa, 2017), has recently adopted Darwin Core. To promote the Darwin Core standard, my colleagues and I have published tutorial papers that describe the format in detail (Osawa, Kurihara, Nakatani, & Yoshimatsu, 2011; Osawa & Totsu, 2017). Such promotional activities are important to realize effective data interoperability.

5.3 Data papers

Traditionally, many ecologists have subscribed to the idea that data should be collected, managed and accessed by only the collector. As a result, ecological data other than those used for publication have remained unused. If such data are not made available to the public or are only stored in a permanent archive as unpublished research data, they are likely to be lost over time (Costello, 2009; Costello et al., 2013; Heidorn, 2008). We should encourage and reward the collection of useful data as an achievement in its own right (Kaye, Heeney, Hawkins, De Vries, & Boddington, 2009). By giving credit via full citation to the originators of data, researchers can more appropriately acknowledge the contribution of the actual data (Whitlock, 2011). Publishing the data corresponding to an academic publication (i.e., a data paper) is a practical idea that has the potential to promote data sharing, and this has several benefits for researchers (Costello, 2009; Costello et al., 2013). For example, data publication could increase the visibility of an author's work and improve citation rates (Piwowar, Day, & Fridsma, 2007). In addition, the primary motivation for individual scientists to publish is to demonstrate their contributions to science and the subsequent peer-recognition influences their reputation, employment opportunities, promotion at work and ability to secure funding (Costello, 2009; Costello et al., 2013). Because data papers can promote data sharing and provide several benefits to researchers, I recommend that Japanese ecologists publish original articles and data papers simultaneously (Osawa, 2017). Accordingly, my colleagues and I have published seven data papers (Fukasawa et al., 2016; Osawa, 2013; Osawa, Baba, et al., 2017; Osawa, Igehara, et al., 2015; Osawa, Kadoya, & Kohyama, 2015; Osawa & Wada, 2016; Osawa, Yamanaka, et al., 2017; Voraphab, Hanboonsong, Kobori, Ikeda, & Osawa, 2015) and several of these data papers were published simultaneously with corresponding original articles (Osawa, 2015; Osawa & Inohara, 2008; Osawa, Kohyama, et al., 2013; Osawa et al., 2016b; Osawa, Yamanaka, et al., 2013). Here, a data paper should include all available data, not just the data used in the original article (Osawa, 2017). This is expected to minimize data loss. Ecological Research, a journal published by the Ecological Society of Japan, has released data papers since 2011, and the number of such articles has gradually increased in recent years (Osawa, 2017). In the near future, I hope this strategy of publishing original articles and data papers simultaneously will become the norm in the Japanese ecological community.

5.4 Open data

To promote both data sharing and reuse, database licenses (i.e., the exploitation of intellectual property) is a critical issue because all data rights should essentially be owned by the collector. “Open data” that are shared and worked with by anyone, anywhere, for any purpose (OPEN KNOWLEDGE INTERNATIONAL; https://okfn.org/), is a powerful concept to resolve intellectual property barriers (Osawa & Iwasaki, 2016; Osawa, Jinbo, et al., 2014; Osawa, Watanabe, et al., 2014). With open data, data reuse and sharing can be easily promoted because open data minimize intellectual property regulations. The easiest way to facilitate open data is the use of a standardized and famous open data license, such as the Creative Commons licenses (https://creativecommons.org/) CC BY and CC BY-SA (Osawa & Iwasaki, 2016). Using such standardized, famous open data license could benefit both data providers and users such as such as reduced licensing costs. Once a data provider applies an open data license and releases the data via the Internet, potential users can access, use, reuse and share the data mostly freely. The Creative Commons CC BY and CC BY-SA licenses do not abdicate intellectual property rights but minimize barriers to reuse and sharing; thus, the data owner can maintain their fundamental rights. In summary, with such licenses, a user can use the data freely with proper data citation (similar to literature citations). This practice is easy to understand and implement, particularly for researchers. GBIF and JBIF already apply Creative Commons open data licenses to almost all of datasets (some of these are applied the Creative Commons CC BY-NC); thus, with proper citation, we can use those datasets freely. This idea is strongly related to the data paper concept.

6 CONCLUSION AND FUTURE PERSPECTIVES

In this paper, I have reviewed the history of biodiversity informatics relative to ecology, with a focus on data treatment (i.e., data use, reuse and sharing) according to my studies and experiences. Data treatment is only one component of biodiversity informatics. However, I believe this is the most important issue in this field. Many researchers already value data as the foundation of their sciences, but, to date, the concepts of data reuse and data sharing remain immature (Costello, 2009; Costello et al., 2013; Osawa, 2017; Parr & Cummings, 2005; Whitlock, 2011). Recently, many ecologists have begun to value data reuse and sharing as a foundation of their research, and this perspective is growing. Therefore, I am confident that biodiversity informatics for ecology will continue to be developed in the near future. However, we face a new problem relative to this new paradigm (i.e., we risk discounting the importance of conventional natural history work, such as collecting specimens, detailed field observations and prolonged maintenance of such materials and data). Specimens require storage space, such as museums, and curators, and digitized data, including observational records, require archive space, such as large stable digital storage facilities and information managers. In addition, there is a severe resource shortage for maintaining specimens (Osawa, 2017; Watanabe, 2016). Improving data availability through ICT imposes a risk (i.e., the importance of traditional natural history work that supports data availability may become forgotten). In addition, we face problems related to the financial sustainability of natural history work. Recently, necessary research funds are becoming increasingly competitive; however, many competitive research funds cannot be used for data management, maintenance and curation. Thus, current intensive research funds cannot support this basic natural history work. Biodiversity informatics can enhance the utility of natural history products; however, the basis will always be the practical natural history work. Therefore, combining traditional natural history work and biodiversity informatics will realize a basis for future efforts in ecology and biodiversity sciences. I would like to emphasis for early career ecologists on the importance of natural history works. Biodiversity informatics without natural history work is almost emptiness as a natural science. To further develop biodiversity informatics, we should rethink the importance of natural history work and continue traditional work under the new paradigm of data reuse and sharing with long-term support systems. Biodiversity informatics is a conventional and novel approach to natural history science, including ecology.

ACKNOWLEDGMENTS

I am greatly thankful to receive the 21st Miyadi Award of the Ecological Society of Japan. I also thank my previous supervisors, Dr. G. Kudo, Dr. T. Kubo, Dr. A. Ushimaru and Mr. H. Mitsuhashi for several supportive and useful discussions through my study history. Further, I thank various collaborators and colleagues, especially Dr. M. Akasaka, for several discussions through my study history. Two anonymous reviewers gave me constructive suggestions. Finally, I thank my family, Akiko, Yuuka and Soushi, for being supportive throughout my life.