State of biodiversity documentation in the Philippines: Metadata gaps, taxonomic biases, and spatial biases in the DNA barcode data of animal and plant taxa in the context of species occurrence data

Abstract

Anthropogenic changes in the natural environment have led to alarming rates of biodiversity loss, resulting in a more urgent need for conservation. Although there is an increasing cognizance of the importance of incorporating biodiversity data into conservation, the accuracy of the inferences generated from these records can be highly impacted by gaps and biases in the data. Because of the Philippines' status as a biodiversity hotspot, the assessment of potential gaps and biases in biodiversity documentation in the country can be a critical step in the identification of priority research areas for conservation applications. In this study, we systematically assessed biodiversity data on animal and plant taxa found in the Philippines by examining the extent of metadata gaps, taxonomic biases, and spatial biases in DNA barcode data while using species occurrence data as a backdrop of the 'Philippines' biodiversity. These barcode and species occurrence datasets were obtained from public databases, namely: GenBank, Barcode of Life Data System and Global Biodiversity Information Facility. We found that much of the barcode data had missing information on either records and publishing, geolocation, or taxonomic metadata, which consequently, can limit the usability of barcode data for further analyses. We also observed that the amount of barcode data can be directly associated with the amount of species occurrence data available for a particular taxonomic group and location-highlighting the potential sampling biases in the barcode data. While the majority of barcode data came from foreign institutions, there has been an increase in local efforts in recent decades. However, much of the contribution to biodiversity documentation only come from institutions based in Luzon.

Keywords: Comparative analysis; Conservation; Genetic diversity; Online biodiversity database; Sampling biases; Spatial analysis; Species diversity.

Figure 1. Summary of barcode records associated with specific gene markers and issues encountered while manually parsing through the descriptive information on sampling locality.

For graph A, the genetic summary of the available barcode records focuses on the gene markers of interest used in the examination for metadata gaps, taxonomic biases, and spatial biases in DNA barcode data on animal and plant taxa sampled in the Philippines were the following: cytochrome b (CYTB), cytochrome oxidase c subunit I (COI), internal transcribed spacer 2 (ITS2), ribulose-1,5-biphosphate carboxylase (rbcL), and maturase K (matK). For graph B, the geolocation issues resulted in the descriptions of the sampling location (particularly in terms of administrative units) being unclear or in some cases, inconclusive. The categories include misspelled (incorrect spelling), none (no major issue), mixed (more than one issue), unspecified (somewhat informative but still vague), unknown (completely not informative), multiple (provided more than one location), and mismatch (discrepancies between the administrative units provided). This dataset includes the records with NA entries for country sampled (for A and B) and those that had additional information on the geolocation other than the coordinates (for B only).

Figure 2. Relationship between the percentage of barcode records identified at the species level and the proportion of documented species (represented in species occurrence data) that currently have DNA barcode data available.

This relationship was evaluated for each known animal (orange) and plant (green) taxonomic group represented in the Philippine barcode data at the phylum/division (A), class (B), order (C), and family (D) levels. This dataset includes the records with NA entries for country sampled.

Figure 3. Relationship between the amount of genetic and species data associated with each known animal and plant taxonomic group represented in the Philippine biodiversity data at different taxonomic levels.

This relationship was evaluated for each known animal (orange) and plant (green) taxonomic group represented in the Philippine barcode data at the phylum/division (A), class (B), order (C), and family (D) levels. Values were transformed logarithmically prior to plotting however, taxa with zero (0) records in either genetic or species data were assigned the value of negative one (−1). Dashed lines represent the 5^th and 95^th percentiles for genetic (horizontal) and species (vertical) data. This dataset includes the records with NA entries for country sampled.

Figure 4. Maps of the sampling distribution of barcode and species occurrence data on animal and plant taxa across the Philippines and the relationship between the two datasets in terms of province.

For both maps (A – barcode data and B – species occurrence data), records on marine specimens were assigned to a specific province based on which corresponding centroid has the shortest distance from the given sampling coordinates (if available). Also, values presented in the maps represent the number of records in the thousands. In the scatter plot (C), values were transformed logarithmically and provinces with zero (0) records in either genetic or species data were assigned the value of negative one (−1). Dashed lines represent the 5^th and 95^th percentiles for genetic (horizontal) and species (vertical) data. The barcode dataset includes the records with NA entries for country sampled.

Figure 5. Map of the distribution of barcode data on Philippine animal and plant biodiversity contributed by different countries across the world and their contribution to documenting efforts across the years.

For map A, contribution was based on the institution that holds the copyright to the image associated with the records while for the graphs, it was based on the collection of samples, starting from the 1990s (B) and submission of barcode data, starting from the 2000s (C) by foreign countries (violet) and the Philippines (red). Trendlines in the graphs represent the average, “best” fitted line. This dataset includes the records with NA entries for country sampled.

Figure 6. Heatmap matrix showcasing the relationship between the number of barcode records associated with regions that have been sampled and the regions of local institutions that contributed the data.

There are officially seventeen regions in the Philippines, represented by the Philippine map (A), with non-numerical regions labelled as follows: ca, Cordillera Administrative Region (CAR); mm, National Capital Region (NCR or also referred to as Metro Manila); and br, Bangsamoro Autonomous Region in Muslim Mindanao (BARMM). Regions are also divided based on their island groups – namely Luzon (red), Visayas (yellow), and Mindanao (blue). For matrix B, contribution was based on the institution that holds the copyright to the image associated with the records. Regions along the x- and y-axis are sorted to provide spatial context, with the map as a reference. The diagonal line represents the “ideal” scenario wherein the region serving as the processing center of barcode data can sufficiently sample its own local area. This dataset includes the records with NA entries for country sampled.