Mapping endangered mammals for conservation

June 12th, 2018, Published in Articles: EE Publishers, Articles: PositionIT, Featured: PositionIT

The Endangered Wildlife Trust (EWT) and SANBI, in collaboration with multiple partners, recently revised the 2004 national Red List of Mammals, a list which helps to identify regional populations that are of global importance for the conservation of the species. Mapping is a key aspect of the red listings process, with the below article providing insight into the mapping methodology.

We must understand the risk of each species becoming extinct to prioritise conservation efforts and allocate scarce resources effectively. The International Union for the Conservation of Nature (IUCN) established the Red List in 1963 to objectively categorise the probability of extinction for every species on the planet. Assessments are carried out through extensive networks of stakeholders pooling their expert knowledge. Red Lists have become the backbone of global species conservation as a unified and standardised protocol to measure biodiversity loss and inform policy decisions. Extinctions, however, occur regionally before occurring globally. National (or regional) Red Lists are thus needed to help prioritise regionally-threatened species and to feed into global assessments.

The Endangered Wildlife Trust (EWT) produced the previous national Red List of Mammals in 2004 [1]. For the 2016 revision, the trust has partnered with the South African National Biodiversity Institute (SANBI), supported by collaborations with MammalMAP (a partnership between the Animal Demography Unit, University of Cape Town and the Mammal Research Institute, University of Pretoria) and the Species Survival Commission (SSC) of the IUCN. Key stakeholders and contributors also included South African National Parks (SANParks), provincial conservation agencies, universities, museums and the private sector. The assessment region included South Africa, Lesotho and Swaziland, as well as the footprint of all transfrontier parks bordering these three countries.

The project began in February 2013. The first phase concentrated on networking with experts and collating empirical data on distribution and population size. We identified the initial set of assessors and data contributors from a chain-referral survey, which then grew on an ad hoc basis. Data contributors included museums, university researchers, statutory conservation agencies, environmental consultancies, private protected areas, landowners and citizen scientists. Overall, we amassed 460 931 occurrence records and 41 075 population count records. Cleaning and synthesising this database is an ongoing project. In total, there were 104 primary data contributors from 60 institutions, and in their private capacity. This collated database should be maintained and updated regularly to make future Red List revisions more accurate and efficient.

Data sources, collation and curation

Mammal data in South Africa exists in numerous, often fragmented, sources. To consolidate this data, a data collection process that first identified potential contributors using a chain-referral survey was initiated. We collected additional datasets in an ad hoc manner from additional experts. We began data collection in June 2013. From these contributors the EWT requested information on species distribution patterns, population sizes and population trends, and the types and severity of the threats facing each species. The contributors collected this data using a variety of methods, including field surveys (ground, aerial and acoustic), traps (camera, Sherman and mist nets), informal geo-referenced sightings (direct observations or spoor), and mortalities (e.g. road kills). Provincial conservation authorities were requested to share their data via a working group at the Department of Environmental Affairs.

A full list of all data-providing institutions is provided in Table 1. The game count data received from protected areas was converted into distribution points by taking the centroid of the reserve for each count year per species. This data was also stored in a relational database so that national population trends for the relevant species could be analysed. We also solicited data from all major museums and national parks authorities. Overall, we synthesised 460 931 data points between June 2013 and December 2016. Of these, the majority (99%) were either geo-referenced, or could be geo-referenced based on the locality information provided.

Table 1: List of data-providing institutions for the Red List revision, ranked by number of records provided. The data are unique, vetted and geo-referenced records which underwent several rounds of data cleaning.

When problems with data records were identified, they were vetted by returning them to the source to check information accuracy. Records that could not be vetted were not used in the assessments or maps. Verifying such records remains to be completed. As it is a labour-intensive process to check geo-referencing of records and to update taxonomy of older specimens or records, records for threatened species were prioritised. For threatened species, out-of-range records were identified based on the most recent available IUCN distribution maps, and the georeferencing was either checked and corrected, or these records were flagged as problematic and excluded from the maps as well as the assessments. However, these records are retained in the database for future verification.

The database is not an atlas. Records reflect only verifiable presence and not absence, thus problems with false negatives are probably geographically biased and indices of abundance cannot be calculated as the database contains no measurement of observer effort. Field surveys are not standardised, thus precluding diversity analyses. However, this database represents the first attempt to consolidate South African mammal data from multiple sources into a central database. No data will be made publicly available unless explicitly condoned by the contributors. All data remain the property of the contributors. Centralising the data ensures that future revisions have a baseline to compare against and is an important foundation on which further national biodiversity assessments can be built.

Map production

The assessment region includes South Africa, Swaziland and Lesotho, as well as all transfrontier conservation areas. This is intended to display distribution across functional and connected landscapes and not simply within arbitrary political boundaries (Fig. 1). Due to data resolution varying from point to quarter degree grid cell, most maps have been plotted at the quarter degree grid scale (QDGC level 2). Plotting at QDGC scale was also to protect sensitive species information. It is important to note that the taxon may not occur throughout the QDGC but only at one site. Caution should thus be used in interpreting fine-scale distribution within the QDGC. Planners and managers must therefore combine the distribution maps with data at finer resolutions to improve conservation planning, or request the point data from the data providers. For the Chiroptera, because all data are at point scale, these maps have been plotted as such.

Fig. 1: The number of mammal records per quarter degree grid cell (QDGC) used in the assessment. The highest densities of records are typically found in protected areas, while the lowest densities of records are found in the Northern Cape and Lesotho.

Fig. 1: The number of mammal records per quarter degree grid cell (QDGC) used in the assessment. The highest densities of records are typically found in protected areas, while the lowest densities of records are found in the Northern Cape and Lesotho.

The distribution maps follow a data-driven approach and are based on multiple sources of empirical, expert-reviewed data and thus represent the verified minimum range of the species free from false positives. False negatives need to be rectified through further field surveys. The establishment of the central database and identification of survey gaps that need to be filled will provide the first useful national overview to plan a systematic monitoring programme. When assessments are migrated to an online system, the underlying datasets will be linked to the assessment, thus enabling a more rapid determination of genuine versus non-genuine changes.

Mapping terminology

Historical (pre-2000) records: What is defined as “historical” is not what is normally understood by the term (i.e. pre-anthropogenic disturbance/transformation), but simply refers to data collected before the year 2000 or before the first national Red List assessment, which was conducted from 2002 to 2004 using prior data. Thus, the historical distribution in the current maps is only to mark recent changes in distribution between the 2004 and current Red List projects. However, direct comparisons between Friedmann and Daly [1] and the current maps must be interpreted carefully, as it is likely that differences reflect the underlying data used in the assessments and not genuine range expansions.

Specifically, the maps display the following layers:

Current (post-1999) distribution: All distribution records collected after the year 2000. For ranched species, all available records pertaining to game farms, wildlife ranches, conservancies or private game reserves were included on the maps to represent the extent of homogenisation [2].

Overlap records: A grid cell in which there are both pre-2000 and post-1999 records.

Undated records: This layer represents all the distribution records for which no date of collection was provided or for which no date can be found. Although the majority of these records are likely to be pre-2000, they are assigned their own category on a precautionary basis until they can be further investigated.

Formally protected areas: This layer includes all provincial, national and transfrontier protected areas from the 2011 National Biodiversity Assessment [3] and thus represents the minimum protected area coverage within the assessment region. Private protected areas are not displayed as the complete dataset is not available yet. The layer was not used for any analyses of the data, but simply to provide context, and broadly shows the protected area network in the country. The available information on protected areas is being regularly updated and more up-to-date layers are now available.

Global range: The global distribution range of the species, using the latest IUCN Red List data, is shown as an inset in the map to situate the regional distribution within a global context.

In addition to the layers above, for a subset of 23 ranched and utilised large mammal species, an additional layer was displayed:

Natural distribution range: This layer, produced for the Department of Environmental Affairs by an Intergovernmental Task Team comprising scientists from all provinces from 2012 to 2015 [4], recreated the historical natural distribution of the relevant species over the period 500 years before present to c. 1930 based historical accounts, confirmed archaeological records, and records of current distribution (generally up to 1930, as it is assumed that up till this time distributions were relatively unchanged by translocation) where there is good evidence that the species occur in the same place as during the historical period. Negative records (confirmed absences) were also used to help define distribution ranges. This data was then overlaid with vegetation types [5] to create a historical or natural range. However, this layer is not the same as a habitat suitability map. The maps will be continually refined as new data become available.

Geographical parameter protocol

Estimating geographical parameters and patterns is one of the main ways to list species, especially small mammals. The extent of occurrence (EOO) (Fig. 2) was calculated as the minimum convex polygon around all compiled geo-referenced records [6]. This measure may exclude discontinuities or disjunctions within the overall distribution of a species. Area of occupancy (AOO) (Fig. 2) is the area within a species’ extent of occurrence that is occupied. This measure reflects the fact that a species will not usually occur throughout the area of its extent of occurrence, which may, for example, contain unsuitable habitats. The AOO is the smallest area essential at any stage to the survival of existing populations of a species. The size of the AOO should be at a scale appropriate to relevant biological aspects of the species.

Fig. 2: The EOO, or range, is important in applying sub-criterion B1 and is calculated through a minimum convex polygon, while the AOO can be used to apply sub-criterion B2 and can be calculated through occupied grid cells, both in conjunction with other sub-criteria.

Fig. 2: The EOO, or range, is important in applying sub-criterion B1 and is calculated through a minimum convex polygon, while the AOO can be used to apply sub-criterion B2 and can be calculated through occupied grid cells, both in conjunction with other sub-criteria.

The AOO was estimated by calculating the amount of natural habitat remaining within the EOO using a national land cover dataset from 2013 [7] with the following heuristics:

  • If the species is not a habitat specialist, AOO was estimated as all remaining natural habitat within the EOO.
  • If the species is a habitat specialist (for example, grassland or forest specialists), the relevant vegetation types [5] within the EOO were clipped to the land cover layer [7]and the remaining natural habitat was calculated.

For wetland specialists, the following method was used:

  • Where a home range size or maximum dispersal distance is available, we used this value to buffer wetland patches.
  • In the absence of such ecological information, we buffered the wetlands by both 500 m (strip width used to assess habitat condition around wetlands in the National Biodiversity Assessment, as it provides a good proxy for wetland condition [3]) and 32 m (minimum buffer zone of no development around waterbodies, as set in the National Environmental Management Act, Activity 9 and 11 listing 1 of Government Notice R544 and Activity 16 Listing 3 of Government Notice R546 of 2010).
  • The buffer strips around the wetlands were summed and overlaid with the land cover layer to calculate the remaining natural vegetation around wetlands.

For species heavily impacted by the traditional medicine trade, we used rural area expansion between 2000 and 2013 [8], which is approximately the ten-year duration recommended for small mammals in the Red List guidelines [6], as a proxy for population decline from harvesting. Additionally, we calculated the effective intact AOO as the proportion of the AOO unaffected by harvesting. If home range or dispersal ability was known, this value was used to buffer distribution points as a radius. If not, both distribution points and “huts” (as a proxy of rural development; Eskom Spot Building Count Eskom, 2011) were buffered by a radius of 10 km (feasible walking distance from villages). We then used current land cover data to subtract the amount of transformed land currently contained within the AOO.

Finally, we subtracted the area of the natural AOO that intersects the buffered rural villages and thus within harvesting distance, which left an estimate for AOO that contains natural land at least 10 km away from potential harvesting threats.

A location is a geographically or ecologically distinct area in which a single threatening event can rapidly affect all individuals of the species present. The size of the location depends on the area covered by the threatening event (for example, an entire river might be one location if it was all threatened by the construction of a dam upstream). A location may include part of one subpopulation or many subpopulations. For example, if two or more subpopulations occur within an area that may be threatened by one such event, they must be counted as a single location. Conversely, if a single subpopulation covers an area larger than may be affected by any single event, it must be counted as more than one location. Where a species is affected by more than one threatening event, a location should be defined by considering the most serious plausible threat. For example, where the most serious plausible threat is habitat loss, a location is an area where a single development project can eliminate or severely reduce the population (Fig. 3).

Fig. 3: The last seven remaining subpopulations of a golden gole species exist in fields within the same region (< 100 km2) but isolated from each other by a matrix of unsuitable habitat. If the most plausible threat is a dam that will flood the region and drown all subpopulations, this is considered one location and may qualify for Critically Endangered under B1ab(i).

Fig. 3: The last seven remaining subpopulations of a golden mole species exist in fields within the same region (< 100 km2) but isolated from each other by a matrix of unsuitable habitat. If the most plausible threat is a dam that will flood the region and drown all subpopulations, this is considered one location and may qualify for Critically Endangered under B1ab(i).

Locations will not be applicable to widespread, unfragmented species or to all types of threat; In many cases, it is not possible to estimate the number of locations for a species. Furthermore, locations must be defined by plausible or imminent threats, not all possible future threats (for example, a possible meteorite impact is not a plausible threat).

Advancements on prior work

This Red List revision has made the following important advancements:

Consolidated database: The data collated during this project can enhance the efficiency of future revisions through good data management practices and the establishment of data-sharing agreements with partner institutions. These initiatives are underway through the construction of the National Biodiversity Information Facility at SANBI. The current database can also form the foundation of an atlas project where data gaps are identified and systematically monitored similar to that of the Karoo BioGaps project and to the field surveys conducted for both the national butterfly and reptile assessments [9, 10].

Conservation evidence: We weighted the threats and interventions according to the evidence presented in the scientific literature so as to standardise the evaluation of severity and effectiveness, respectively. This data is being compiled into a database summarising the type and strength of evidence presented by scientific papers regarding mammal conservation and will be used as a resource for future assessments.

Measuring conservation value of managed subpopulations: Through a series of expert workshops, a framework was developed to objectively measure the wildness of managed subpopulations through attributes relating to evolutionary and ecological dynamics. This framework was applied to all relevant species to standardise the inclusion of private subpopulations into the Red List. Refinement of the framework is ongoing and will ultimately link to the IUCN Green List, which is also under development.

Watch-list categories: Three additional qualifying categories were created with the intention to flag species that are in urgent need of additional research or direct conservation interventions. These qualifiers complement the Red List categories and can help to prioritise assessments needing urgent revisions.

Information quality classifications: We standardised the data quality used to assign Red List statuses for each taxon on a spectrum from low to high confidence. These classifications will be used to determine what data is needed to make the assessments more robust and thus will be linked to the conservation evidence framework.

Recommendations going forward

The following are recommendations for future Red List mapping efforts:

  • Maintain the centralised mammal distribution database by collating additional datasets as they are produced. This will significantly reduce the time needed to conduct future revisions.
  • Continue vetting the records to correct errors and to reflect recent taxonomic changes.
  • Source funding to perform systematic mammal surveys, much like what was done in the Butterfly and Reptile Red Lists and Atlas projects [9, 10], as small mammals are particularly reliant on museum records at present.
Fig. 4: The potential relationship and feedbacks between the Red List (solid arrows represent primary relationships; dotted arrows represent potential feedbacks), research institutions, government agencies and the private wildlife sector.

Fig. 4: The potential relationship and feedbacks between the Red List (solid arrows represent primary relationships; dotted arrows represent potential feedbacks), research institutions, government agencies and the private wildlife sector.

Conclusion

The national Red List is not simply a guideline but a tool to synthesise evidence on mammal conservation and to engage stakeholders. While the Red List statuses reflect the relative extinction risk of species (based chiefly on geographical and population data), the process of setting priorities for conservation actions requires several additional considerations, such as ecological significance, cultural value, logistical factors in implementing conservation plans, availability of funding or capacity, and existing legal frameworks to implement conservation plans [11]. Additionally, the global status of the species and the proportion of the global population that occurs within the assessment region should influence the priority list. This will also help to identify regional populations that are of global importance for the conservation of the species.

Regional Red Lists are important as they are used to:

  • Inform conservation policies and legislation (both national and international);
  • Identify research gaps and stimulate monitoring programmes;
  • Monitor the status of biodiversity and report on the state of the environment (through use of indices such as the Red List Index);
  • Regulate the development and use of wildlife resources;
  • Target areas for conservation planning;
  • Increase public awareness of threats to biodiversity;
  • Set priorities for the allocation of limited conservation resources [11].

Users of this Red List include conservation planners, research scientists, managers of conservation organisations, landowners and protected area managers, environmental impact agency workers, civil servants compiling governmental reports, officials involved in land-use planning, environmental educators and concerned members of the public looking to lodge protests against damaging development applications.

This project has brought together scientists, conservation practitioners, government officials, landowners and citizen scientists to produce assessments, databases and frameworks for South Africa’s sustainable future. We have created three legacies that will greatly improve the efficiency and accuracy of future revisions: a synthesised occurrence and population count database; referenced assessments that provide the foundation for future revising rather than recreating; and frameworks that help to adapt the application and standardisation of the IUCN sub-criteria to a South African context.

South Africa is a stronghold for African mammal biodiversity and keeping track of our species together through these legacies will help it to remain so.

Acknowledgement

This article is an excerpt from The Red List of Mammals of South Africa, Swaziland and Lesotho 2016: Introduction and Methodology. The full report is available at www.ewt.org.za/reddata/reddata.html.

References

[1] Y Friedmann, B Daly, editors. 2004. Red Data Book of the Mammals of South Africa: A Conservation Assessment. IUCN SSC Conservation Breeding Specialist Group and Endangered Wildlife Trust, South Africa.
[2] D Spear, SL Chown. 2009. The extent and impacts of ungulate translocations: South Africa in a global context. Biological Conservation 142:353–363.
[3] A Driver, KJ Sink, JN Nel, S Holness, L van Niekerk, F Daniels, Z Jonas, PA Majiedt, L Harris, K Maze. 2012. National Biodiversity Assessment 2011: An Assessment of South Africa’s Biodiversity and Ecosystems. Synthesis Report. South African National Biodiversity Institute and Department of Environmental Affairs, Pretoria, South Africa.
[4] C Birss, I Rushworth, NB Collins, D Peinke, D Buijs. 2015. Inferred natural distribution ranges of large mammals in South Africa. Version 1. Unpublished GIS coverage. Department of Environmental Affairs, Pretoria, South Africa.
[5] L Mucina, MC Rutherford. 2006. The Vegetation of South Africa, Lesotho and Swaziland. South African National Biodiversity Institute, Pretoria, South Africa.
[6] IUCN Standards and Petitions Subcommittee. 2017. Guidelines for Using the IUCN Red List Categories and Criteria. Version 13. Prepared by the Standards and Petitions Subcommittee. Available from http://www.iucnredlist.org/documents/RedListGuidelines.pdf.
[7] GeoTerraImage. 2015a. 1990–2013/14 South African National Land-Cover Change. DEA/CARDNO SCPF002: Implementation of Land-Use Maps for South Africa. Project Specific Data Report, Pretoria, South Africa.
[8] GeoTerraImage. 2015b. Quantifying settlement and built-up land use change in South Africa.
[9] S Mecenero, JB Ball, DA Edge, HL Hamer, GA Henning, M Krüger, EE Pringle, RF Terblanche, MC Williams, editors. 2013. Conservation assessment of butterflies of South Africa, Lesotho and Swaziland: Red List and atlas. Saftronics (Pty) Ltd., Johannesburg & Animal Demography Unit, Cape Town, South Africa.
[10] MF Bates, WR Branch, AM Bauer, M Burger, J Marais, GJ Alexander, MS de Villiers, editors. 2014. Atlas and Red List of the Reptiles of South Africa, Lesotho and Swaziland. South African National Biodiversity Institute, Pretoria, South Africa.
[11] RM Miller, et al. 2007. National threatened species listing based on IUCN criteria and regional guidelines: current status and future perspectives. Conservation Biology 21:684–696.

Contact Dr Lizanne Roxburgh, EWT, Tel 011 372-3600, lizanner@ewt.org.za