Mind the Gaps: Taxonomic, Geographic and Temporal Data of Marine Invertebrate Databases from Mozambique and São Tomé and Príncipe

Abstract

One of the best ways to share and disseminate biodiversity information is through the digitization of data and making it available via online databases. The rapid growth of publicly available biodiversity data is not without problems which may decrease the utility of online databases. In this study we analyze taxonomic, geographic and temporal data gaps, and bias related to existing data on selected marine invertebrate occurrences along the coastline of two African countries, Mozambique and São Tomé and Príncipe. The final marine invertebrate dataset comprises of 19.910 occurrences, but 32% of the original dataset occurrences were excluded due to data gaps. Most marine invertebrates in Mozambique were collected in seagrasses, whereas in São Tomé and Príncipe they were mostly collected offshore. The dataset has a temporal coverage from 1816 to 2019, with most occurrences collected in the last two decades. This study provides baseline information relevant to a better understanding of marine invertebrate biodiversity data gaps and bias in these habitats along the coasts of these countries. The information can be further applied to complete marine invertebrate data gaps contributing to design informed sampling strategies and advancing refined datasets that can be used in management and conservation plans in both countries.

1. Introduction

Museums and herbaria with natural history collections (NHC) are vital sources of scientific knowledge. These collections provide a vast amount of historical and current information on the world’s biodiversity [1] and have proven to be a critical record of ecological change and evolution. One of the best ways to share such information is through the digitization of data and to make it available via online databases. The Global Biodiversity Information Facility (GBIF, http://www.gbif.org, accessed on 10 October 2018) aims at mobilizing biodiversity data from NHC, surveys and other sources by storing data in an online portal [2]. GBIF is not the only online biodiversity database available (e.g., OBIS, iDigBio) but it is the one that provides more information and the most widely used [3,4], with the intention to ‘make the world’s primary data on biodiversity freely and universally available via the Internet’ [5]. Currently, GBIF provides access to 2.2 billion occurrence records (as of August 2022).

This rapid growth of publicly available biodiversity data is not without problems, and the scientific community acknowledges taxonomic, geographical and temporal uncertainty and the existence of data gaps [6]. Differences in funding and data sharing, together with punctual geopolitical conflicts often result in spatial and species biases [2] and scarcity of data availability for some regions, requiring a thorough analysis of data quality [7]. The data available in GBIF has very diverse sources, quality, scope and accuracy [8] provided mostly by NHC, research and monitoring campaigns. The inaccuracies and uncertainties in coverage and quality of data may decrease the utility of online databases [8], and it is essential to minimize these problems through the standardization of data formats such as the Darwin Core [9] and a peer-review system for data publication [10]. Another source of inaccuracy is that the purpose of data collection may differ from the subsequent use by GBIF users [11].

Marine habitats face a wide range of pressures related to human population growth, such as industrial activities, fishing and aquaculture, maritime transportation, tourism, or urbanization of the coastline [12]. These local drivers of degradation are further strengthened with global impacts of climate change, such as sea level rise, cyclones and related floods and strong winds temperature increase and acidification. These combined pressures unbalance biotic communities and ecosystem processes, affecting biodiversity, community structure and species distributions. Marine macroinvertebrates are vital components of marine habitats due to their abundance and diversity, playing an important role in marine ecosystems. In spite of their significance there is a lack of compiled information regarding their biodiversity [13]. Developing countries are particularly vulnerable to biodiversity loss and the degradation of ecosystems. The gaps in data from developing countries are also more pronounced which is explained by limited funding and geopolitical conflicts, such as war. This means that data collection and monitoring programs, aiming to a better representation and knowledge of the natural capital, are insufficient when compared with those of developed countries, causing unbalanced online databases by underrepresentation of geographical regions [14].

The goal of this study was to assess taxonomic, geographic and temporal data gaps and bias related to existing data on selected marine invertebrate occurrences (Annelida, Arthropoda, Cnidaria, Echinodermata and Mollusca) along the coastline of two poorly studied African countries, Mozambique (MOZ) and São Tomé and Príncipe (STP), by compiling scattered data from online databases (namely GBIF and museums’ NHC databases) and scientific literature. This analysis provides baseline information relevant to a better understanding of marine invertebrate biodiversity databases, their limitations and their improvement to accurately address updated biodiversity issues. Further, this dataset can be used to complete marine invertebrate data gaps and to develop informed sampling strategies leading to more refined datasets that can be applied in management and conservation plans in both countries.

2. Materials and Methods

2.1. Study Areas

Mozambique is located in the Western Indian Ocean, South-eastern Africa and has a coastline extension of 2500 km and an Economic Exclusive Zone (EEZ) of 493,000 km² [15]. The three most critical coastal habitats include mangrove forests, seagrass beds and coral reefs, with an approximate area of 2909, 439 and 1860 km², respectively [15].

São Tomé and Príncipe (STP) is an island state located at the Equator in the Gulf of Guinea at 380 km off the continental African coast [16]. It consists of one volcanic archipelago, of which the two main islands are São Tomé and Príncipe, separated by approximately 146 km, and has a coastline extension of 240 km. The EEZ covers an area of 160,000 km² [17]. Mangrove forests, seagrass beds and coral reefs in STP have an area of approximately 0.8 [18], 15 [16] and 2 km² (estimated by comparison of satellite images of mangrove forests and seagrass beds habitats), respectively.

2.2. Compilation of Data

All available data on marine invertebrate occurrences (i.e., the presence of a species, or other taxon, at a particular place on a specified date) from MOZ and STP were downloaded from GBIF (http://www.gbif.org, accessed on 3 December 2018) (DOI’s in Appendix A) into a dataset to assess taxonomic, geographic and temporal data gaps and bias of marine invertebrate data from the different coastal habitats of both countries. A total of 76,945 occurrences were assembled, of which terrestrial and freshwater species comprised more than 60% of the data (47,189 occurrences, mainly Insecta and Gastropoda occurrences). These were manually excluded from the initial dataset, as GBIF does not have an option to select marine occurrences. Data from museums’ NHC databases and scientific literature (see Appendix A) were also incorporated into the dataset, following Darwin Core standards [19] (see fields used in Supporting Information Table S1). Only marine invertebrate occurrences identified at species or genus level were included in the dataset. The surveyed habitats mainly comprised mangrove forests, seagrass beds and coral reefs. Some occurrences were not collected in any of these three habitats, so they were grouped in two categories: coastal area for all other coastal habitats (e.g., rocky shores or sandy beaches) and offshore for occurrences 12 km until the limit of the Economic Exclusive Zone (EEZ).

The dataset was thoroughly reviewed and validated. A taxonomic review was conducted, and identifications were updated using the World Register of Marine Species (WoRMS, http://www.marinespecies.org, accessed on 23 February 2019). Some occurrences did not have any location data (locality and coordinates) and/or date of collection (some of this data was not accessed from the online databases sources). Occurrences without coordinates but with a good locality description were georeferenced using GEOLocate Collaborative Georeferencing tool (CoGe, https://coge.geo-locate.org, accessed on 18 March 2019) and Google Maps (https://www.google.com/maps, accessed on 18 March 2019). Missing dates on the dataset downloaded from GBIF were later retrieved by directly accessing the respective museum or institution database (e.g., Smithsonian-National Museum of Natural History or the California Academy of Sciences). The dataset was saved as an Excel spreadsheet and all basic analysis were performed using Excel tools (Microsoft Corporation, 2018, Microsoft Excel, available at: https://office.microsoft.com/excel).

2.3. Taxonomic Coverage

After a preliminary analysis of the dataset, the invertebrate phyla from three habitats (mangrove forests, seagrass beds and coral reefs) with more occurrences were selected for both countries. The main groups were Annelida, Arthropoda, Cnidaria, Echinodermata and Mollusca. Occurrences were analyzed by taxonomic level namely family, order, and genera.

2.4. Geographical and Temporal Coverage

Geographical marine invertebrate occurrences were analyzed by phyla and habitat along the coasts of MOZ and STP, simultaneously, allowing the comparison between both countries. Phyla occurrences in both countries were also mapped. To construct these maps, the dataset was loaded into ArcGIS ArcMap as a CSV file. Longitude and latitude of point coordinates were mapped to X and Y fields, respectively. The location of mangrove forests, seagrass beds and coral reefs in the coastal zones of MOZ and STP were retrieved as layers from the UN Environment World Conservation Monitoring Centre and the ReefBase websites. The coordinate reference system used was WGS 1984 [20].

Temporal occurrences were analyzed along the coasts of MOZ and STP by phyla and habitat. Several time intervals were considered for the temporal analysis. Occurrences before 1900 were grouped, as only 13 records were registered, while all other occurrences were grouped by decade.

3. Results

3.1. Compilation of Data

The marine invertebrate dataset had almost 30,000 occurrences in both MOZ and STP, but only 68% of these entries were used to assess taxonomic, geographic, and temporal data gaps and bias, as 32% had to be excluded. Exclusion of occurrences without location data (locality and coordinates, as occurrences that only had locality data were georeferenced) and/or date of collection (some of these data were retrieved directly from the museums online databases), represented 21% of the original dataset. The remaining 11% of exclusions were occurrences that did not have taxonomic identification at genus or species level. After data validation, a final dataset was compiled with 19,910 occurrences, of which 19,359 from MOZ and 551 from STP, and it is accessible through the GBIF portal under the Creative Commons Attribution 4.0 International license, at http://ipt.gbif.pt/ipt/resource?r=marineinvertebrate_moz_stp, accessed on 19 October 2022 [21].

3.2. Taxonomic Coverage

The total number of marine invertebrate occurrences per phyla is shown in

Table 1. Marine invertebrate occurrences per phylum and corresponding frequency in Mozambique (MOZ) and São Tomé and Príncipe (STP).

Phylum	MOZ	STP	MOZ	STP
Annelida	45	21	0.23%	3.81%
Arthropoda	11,411	296	58.94%	53.72%
Cnidaria	649	24	3.36%	4.36%
Echinodermata	469	56	2.42%	10.16%
Mollusca	6785	154	35.04%	27.95%
Total	19,359	551	100.00%	100.00%

. In both countries the phylum with more occurrences was the Arthropoda followed by the Mollusca. The phylum with the least occurrences was the Annelida.

Mozambique’s Arthropoda occurrences in the dataset (Figure 1) belonged mainly to the class Malacostraca (96%). Within this the order Decapoda was the most well represented (98%). Most decapod occurrences refer to the swimming crab family Portunidae (38%) and penaeid shrimps’ family Penaeidae (37%).

The Mollusca occurrences in MOZ (Figure 1) were distributed mainly by two classes, Cephalopoda (52%) and Gastropoda (40%). Within the Cephalopoda, most occurrences belonged to the orders Myopsida (48%) and Sepiida (49%). All Myopsida occurrences belonged to the Myopsidae family and almost all of these belonged to the squid genus Loligo (96%). Sepiida occurrences were distributed in two families, Sepiidae (99%) and Sepiolidae (1%) and almost all occurrences of Sepiidae belonged to the cuttlefish genus Sepia (99.9%). Most of Mozambique’s Gastropoda occurrences in the dataset belonged to the order Neogastropoda (55%). The families with more occurrences were the Muricidae (22%), Nassariidae (17%) and Conidae (15%).

Marine Arthropoda occurrences in STP (Figure 1) belonged mainly to the class Malacostraca (71%). Within the Malacostraca, the order Decapoda represented the majority of occurrences (56%) (and most of these belonged to the shrimp family Alpheidae (41%). As for the Mollusca in STP, most of the occurrences (Figure 1) belonged to the classes Bivalvia (41%) and Gastropoda (46%). Bivalvia occurrences belonged mainly to the order Nuculanida (24%) and most of Gastropoda occurrences belonged to the order Nudibranchia (52%).

3.3. Geographical Coverage

Most marine invertebrate occurrences in MOZ (Figure 2) were collected in seagrass beds (56%) followed by offshore habitats (27%) and mangrove forests (12%). In STP (Figure 2) most marine invertebrates were collected in offshore (54%) and coastal area habitats (30%).

Regarding phylum occurrences in each habitat (Figure 2), mangrove forests and seagrass beds in MOZ had a clear predominance of Arthropoda (72% and 70%, respectively) and Mollusca (22% and 29%, respectively). In coral habitats, most of the occurrences were Mollusca (89%) and only 5% of occurrences were cnidarians.

In STP (Figure 2) seagrass beds there was a prevalence of Arthropoda and Mollusca occurrences (34% and 40%, respectively). In coral reef habitats there were only Mollusca occurrences, and in mangrove forests there were only six occurrences (50% Cnidaria, 17% Arthropoda, 17% Echinodermata and 17% Mollusca).

Regarding phyla occurrences along the coastline, in MOZ (Figure 3A) Cnidaria and Mollusca were more represented in the northern provinces (Cabo Delgado and Nampula) while Arthropoda had more incidence in the central (Zambezia and Sofala) and southern provinces (Inhambane, Gaza and Maputo). In STP (Figure 3B) there were more Arthropoda occurrences in São Tomé Island and more Cnidaria and Echinodermata occurrences in the Príncipe Island.

3.4. Temporal Coverage

The dataset has a wide temporal range, the oldest occurrence was collected in 1816 and the most recent in 2019. The majority of occurrences were collected in the last two decades, representing 81% of all records (Figure 4). From 1960 to 1969 there was a slight increase in marine invertebrate occurrences comprising of 8.8% of all occurrences.

Most occurrences in MOZ (Figure 4) were registered in the decades 2000–2009 and 2010–2019. The third time interval with more occurrences was 1960–1969. In STP most occurrences (Figure 4) were registered in 1950–1959, 1970–1979 and 2000–2009.

Regarding phyla occurrences, in MOZ the Mollusca were recorded in each time interval analyzed, with 50% (or over) of occurrences in six-time intervals (1800–1899: 50%; 1900–1910: 67%; 1910–1919: 76%; 1950–1959: 51%; 1970–1979: 84%; 1990–1999: 76%). Arthropoda occurrences were most abundant in 1940–1949 (60%), 2000–2009 (66%) and 2010–2019 (65%). Other phyla occurrences in Mozambique rarely surpassed the 50% threshold, except for Cnidaria occurrences in 1930–1939 (81%) and Echinodermata occurrences in 1800–1899 (50%) and 1980–1989 (50%).

In STP, several time intervals had no marine invertebrate occurrences (1900–1909; 1920–1929; 1930–1939 and 1940–1949). Mollusca occurrences were most frequent in 1910–1919 (100%), 1980–1989 (87%) and 1990–1999 (60%), and Arthropoda in 1800–1899 (67%), 1970–1979 (51%), 2000–2009 (72%) and 2010–2019 (51%). Other phyla occurrences were lower than 50%, except for Echinodermata occurrences in 1960–1969 (50%).

The temporal distribution of marine invertebrate occurrences per studied habitat (mangrove forests, seagrass beds and coral) revealed several differences. In MOZ, occurrences were more distributed through time in mangrove forests, while in seagrass beds and coral habitats most occurrences were collected in more recent decades (2000–2009 and 2010–2019). All Annelida occurrences in mangrove forests were collected in 1950–1959. In STP, in mangrove forests and seagrass beds, most occurrences were collected in two main time intervals, 1950–1959 and 2010–2019. In coral habitats all occurrences were collected from 2010 to 2019.

4. Discussion

Online databases such as GBIF, NHC collections, research data and published scientific papers are useful tools for gathering biodiversity data that are normally dispersed and sharing it in a structured and accessible way. We used these tools to compile a dataset with a total of 19,910 marine invertebrate occurrences from Mozambique and São Tomé and Príncipe, which enabled to assess taxonomic, geographic and temporal data gaps and bias in these two countries. Although GBIF uses Darwin Core as a standardized format for publishing data, many records did not have the complete information. For instance, 21% of the records downloaded from online databases were excluded because they had incomplete date of collection and/or location data. We were able to georeference most of the occurrences that did not have coordinates associated to the locality data, but some were too vague or just mentioned the country (e.g., West Africa, Mozambique). Another 11% of those records were also excluded because they did not have taxonomic identification at genus or species level. While this identification can be made a posteriori (and indeed this is one of the many benefits of NHC) the lack of date of collection and location data can invalidate the usefulness of an occurrence. Increasing the digitization of NHC data is essential to acquire a better insight into data gaps in both countries as many institutions are behind in this process. Furthermore, NHC managers and curators should reinforce the need to register the essential data (such as date and time of collection, coordinates, and locality—as accurate as possible, collector name, expedition name—if available, altitude or depth) when new entries are added to the collections. The data should then be added to GBIF, paving the way for FAIR data (findable, accessible, interoperable and reusable) [22].

A preliminary analysis of the dataset revealed a clear discrepancy in the volume of data between both countries. Mozambique’s occurrences represent 97% of the data and São Tomé and Príncipe’s only 3%. This could be due to several factors. São Tomé and Príncipe is a small country, its coastline represents 10% of that of Mozambique and its EEZ 32%. Higher education in the field of Biology is still recent in São Tomé and Príncipe. The University of São Tomé and Príncipe was established in 2014, while in Mozambique the University of Eduardo Mondlane was established in 1962 and this institution is responsible for the functioning of the Estação de Biologia Marítima da Ilha de Inhaca (Marine Biology Station of Inhaca Island) [23].

In both countries the taxonomic groups with most occurrences represent species that are used as a food source or for decorative purposes, such as Portunidae (crabs), Penaeidae (shrimps), Myopsida (squids), Sepiidae (cuttlefish) and Neogastropoda (sea snails) in Mozambique, and Alpheidae (snapping shrimps), Nuculanida (clams) and Nudibranchia in São Tomé and Príncipe. This bias towards food resources is probably due to the fact that a large portion of the data downloaded from GBIF has been published by official institutions, such as IIP in Mozambique (National Institute of Fisheries) (see Appendix A). In the instance of Nudibranchia in São Tomé and Príncipe the majority were collected by researchers particularly interested in this taxonomic group (e.g., [24]). These gaps and limitation of taxonomic coverage can lead to biases that affect all future analyses of the data [25].

Overall, the geographical coverage in both countries has serious gaps. Considering that MOZ accounts for 60% of mangrove forests in mainland eastern Africa [15] it is surprising that only 12% of occurrences were collected in mangrove forests. In STP the three analyzed habitats were underrepresented in the data downloaded from GBIF, which is due to the sparse coverage these habitats have in the country [16,18]. In both countries, Cnidaria occurrences were underrepresented (MOZ) or completely lacking (STP) in coral habitats. This could demonstrate a bias in collecting (e.g., the main object of the studies carried out in these habitats were other taxonomic groups).

There is also a gap in historical records in both countries, more pronounced in STP than in MOZ. The decline in the number of occurrences in MOZ between the time intervals 1960–1969 and 1990–1999, can be explained by the political instability between 1964 and 1992 [26]. In STP, in the last decade (2010–2019) a decline in occurrences is noticeable (from 40% of occurrences in 2000–2009 to 7% in 2010–2019), despite the establishment of the Parque Natural Obô de São Tomé and Parque Natural Obô do Príncipe in 2006 and the Island of Príncipe Biosphere Reserve in 2012.

Marine macroinvertebrate biodiversity data gaps can only be filled by creating local databases and institutionalization of the data collection methods in an equative and sustainable way for each country. Its main justification would be the need to manage these resources, especially the endangered, unique and endemic ones. Both countries could adopt a similar approach to the Marine Strategy Framework Directive (MSFD) developed for the European Union. An inventory of the temporal occurrences, abundance and spatial distribution of marine macroinvertebrates as well as information regarding species composition, biomass and annual/seasonal variability [27] is essential.

This study provides baseline information relevant to a better understanding of marine invertebrate biodiversity data gaps and bias in mangrove forests, seagrass beds and corals along the coasts of Mozambique and São Tomé and Príncipe. This information can be further used to design informed sampling strategies leading to more refined datasets that can be used in management and conservation plans in both countries allowing future research to focus in completing data gaps.