MPA Challenge: Coherence

MPA Coherence Data Evaluation (project phase II)
To evaluate the coherence of the MPA network a number of datasets is vital:

  • Location of Marine Protected Areas, preferably with associated data on species and habitats.
  • Species specific distribution data to be evaluated, preferably with additional information on ecology, habitat preference etc..

We performed a coherence analysis for the Arctic Ocean using a matrix-based approach developed by OSPAR (2006,2008). A more technical description is available in the next section. A geographic database with Marine Protected Areas was collated in phase I of this project. This formed the basis of the coherence analysis. The matrix approach includes bioregions. These were implemented as Large Marine Ecosystem as delineated by PAME (2013). While other checkpoint projects were able to rely fairly heavily on EU-related data portals like EMODnet and Copernicus Marine Environmental Monitoring Service, the scope of these data sources generally excluded the Arctic Ocean. Therefore, the project team in this case had to make the best possible use of other –Arctic specific or global– sources of data.

The main remaining challenge was to source datasets on the species and decide which species to include. 

The following data sources were identified and used:

  1. AquaMaps
  2. GBIF
  3. Birds of the World (BirdLife International)
  4. IUCN Red List

The table below summarises species groups covered by each data source, type of data offered, ease of use and accessibility.

Description AquaMaps GBIF Birds of the World IUCN Red List
Species groups covered Sharks and Rays; Bony Fish; Invertebrates and Sea Mammals Sharks and Rays; Bony Fish; Invertebrates and Sea Mammals; Birds Birds Sharks and Rays; Bony Fish; Invertebrates and Sea Mammals; Birds
Species groups not covered Birds - Sharks and Rays; Bony Fish; Invertebrates and Sea Mammals -
Type of data offered Map view; CSV-download (c-squares codes included, offering link to geographic display and analysis) Map view; species dossier; CSV-download a.o. per species (includes Lon/Lat-coordinates per observation point) Map view; species dossier; Download on request ESRI File GeoDatabase received Map view; species dossier; Download on request Shapefile (zipped) received, with additional files
Resulting data in GIS Polygons (vector grid) Points Polygons Polygons
Ease of use Medium Medium High High
Data request Free download Free download Required for geographic data. Granted for full dataset. Initial single species request granted, well within stated turnaround times. Second multi-species request unanswered. Also after reminder sent.

Coherence Analysis
Coherence is an important and desirable characteristic of a network of Marine Protected Areas. A coherent network offers protection to a sufficiently large proportion of species and habitats to ensure their continued survival and existence. It also means that the different elements of the MPA network are within a reasonable distance from each other to support a metapopulation of the species so that migrating or otherwise mobile (e.g. during a larval phase) individuals stand a good chance of reaching a ‘safe haven’ (the next protected area).

To assess the coherence of the MPA network the dataset constructed during phase I of the project has been used in conjunction with additional datasets on species occurrence and sea ice as the one major habitat for the Arctic Ocean. For this assessment an evaluation framework developed by OSPAR was used (OSPAR, 2006, 2008). It attempts to assess aspects such as: Adequacy, Viability, and Replication.  

The analysis was based on 18 Arctic Bioregions, identical to the Large Marine Ecosystems (LME) as delineated for the Arctic (PAME, 2013)

The second component was the analysis of the population, extant etc. across those 18 bioregions for a 25 species and one habitat (sea ice). 

Outcome of the Coherence analysis
While the available data was sufficient to complete the coherence analysis, the process was complicated by the fact that the species data had to be obtained from different sources. This meant that  the methodology had to be altered to achieve comparable results. 

All species included in the coherence analysis are Arctic species meaning that make extensive use of the Arctic and are included in Meltofte (2013). From our list the species with the lowest scores for Replication and Adequacy/Viability are bird species that breed in the Arctic, but that are otherwise not very reliant on the Arctic Ocean and Marine Protected Areas.  

The least protected Arctic marine mammals are

  • the  Polar Bear (Ursus maritimus), low Replication score and
  • the Hooded Seal (Cystophora cristata), low proportion of the population inside MPA.

MPA Ecological Coherence evaluation

Both the Polar Bear and the Hooded Seal have a strong association with sea ice as an important habitat. Sea ice is also important habitat for  the Ringed Seal (Pusa hispida), the Narwhal (Monodon Monoceros), the Ivory Gull (Pagophila eburnea), Polar Cod (Boreogadus saida) and, to a lesser extent the Walrus (Odobenus rosmarus) and Beluga whale (Delphinapterus leucas). 

A single habitat has been included in our coherence analysis: sea ice (based on the min. and max. extent recorded in 2015). This habitat has the lowest Replication score within the analysis and also is poorly protected by/represented in current MPA boundaries. 

The project team did not find any targets set (internationally or nationally) and focussed on the protection of species and/or habitats in the Arctic Ocean. As a result this column within the matrix was left empty. 

All of the results were combined into table 1, with the species/habitat in rows, and, the bioregions and other characteristics to be evaluated in columns. 

Table 1: MPA Coherence Analysis.
Species/habitat with blue background indicated sea ice and/or a strong association with this habitat. In the columns for the bioregions a light blue background (for mammals, fish and invertebrates) indicates year-round presence within that bioregion. Beluga has a blue outline to indicate a year-round presence in the Arctic (#1). For bird species a slightly darker blue background indicated presence within the bioregion outside the breeding season. For a few bird species bioregions have a blue outline, to indicate presence inside the bioregion outside the breeding region, but with decreased reliance on these bioregions during non-breeding season as the species also make use of large areas outside the Arctic. For Brent Goose, the Hudson Bay has a pale yellow background indicating it is used during migration to/from the breeding grounds in the Arctic and wintering areas further south.
Numbers in the bioregion columns indicate (estimated) population sizes for that region.
Feature targets are part of the OSPAR evaluation framework,  however the column remains empty, as no (international) set targets have been identified. The columns for Replication (#2) and Adequacy/Viability (#3)have been colour-coded to aid the interpretation.

Table MPA coherence analysis

#1: Beluga whales have been evaluated using a dataset originating from AquaMaps and within that group it is deviant. Most AquaMap suitability maps have areas where the Overall Probability reaches its maximum value of 1.0. For the Beluga map this is not the case as it peaks at 0.6, and is therefore below the 0.8 threshold used for all other AquaMaps datasets to determine which bioregion are of prime interest to the species.

#2: Colour-coding Replication

<10 >10 >25 >50 >100 >250

#3: Colour-coding Adequacy/Viability

<2.5% >2.5% >5% >10% >25% >50%

The Coherence Analysis has relied on four major sources of information – for the species:

  1. AquaMaps
  2. GBIF
  3. Birds of the World (BirdLife International)
  4. IUCN Red List

The initial plan was to use the distribution maps that are presented on the IUCN Red List website. Unfortunately, after an initial and rewarded request for Little Auk (Alle alle) a second request did not receive any reply. It may have been that our choice to bundle a dozen or so species into a single request did not fit within the administrative processes at IUCN. It seemed to us that not repeating the same request, with all additional text etc. for each species separately would be more efficient. We also waited for a few weeks –as indicated- before an email was sent to a different address to give notice of not receiving an answer. 

AquaMaps was found as an alternative source of data. This dataset offers more or less full cover data on occurrence based on an ecological envelope and overall probability of occurrence. The ecological envelope is extracted from amongst others data points from online collection databases such as GBIF and OBIS and includes minimum, optimal and maximum values for such habitat traits as depths, salinity, temperature etc. The resulting dataset is available for download as a  csv-file and contains a reference to the c-squares system. With proper use of the c-squares information the csv-data can be coupled to the proper polygons thus yielding a geographic dataset. Below a sample map is shown for Walrus (Odobenus rosmarus). 

MPA sample walrus

To estimate the population within an LME (bioregion) the summed results of the Overall Probability multiplied by the area of that polygon was used. 

AquaMaps (Kaschner et al., 2016) has data for amongst others Sharks and Rays; Bony Fish; Invertebrates and Sea Mammals. No data for birds.

GBIF or in full Global Biodiversity Information Facility was also used. This source provides point data on species occurrence. It has been used for species that were not available from AquaMaps. This data source has been used for one (sea) mammal and two bird species. As an example a map for Polar Bear (Ursus maritimus) is shown.

MPA example polar bear

BirdLife International also has much available data, including geographical datasets with polygons for areas used by a species during periods for: 1. resident, 2. breeding, 3. non-breeding, 4. passage and 5. seasonal occurrence uncertain. For consistency the main analysis was based on the breeding season. For most of the bird species that were assessed a large part of the total population breeds strictly within the Arctic (up to 100%) and the habitat is essential. On our request for access to the data BirdLife International replied positively and a copy of the Birds of the World (geodatabase) was made available for the project. The geographical data was also shared with IUCN for amongst others the Red List-website. 

Several species had multiple entries in the data (most often two) for some Large Marine Ecosystems (bioregions). By examining the GIS data, the (assumed) underlying cause was identified as separate and adjoining polygons stemming from different origins. To correct the results the affected datasets where aggregated to a single total results per LME. An example is shown for the Ivory Gull (Pagophila eburnea).

MPA example ivory gull
Processing the data in GIS
The diverse datasets where collated in GIS, where amongst others the data was analysed.

Most datasets did not arrive with a suitable projection already defined. All datasets were projected to a polar projection: WGS 1984 North Pole LAEA Atlantic (WKID 3574). This ensured that further operations that rely on determining distance or overlap between features would work properly. It was found earlier in the project that the software apparently has difficulty handling polar projections correctly.

The C-squares required to geographically interpret the CSV-files from AquaMaps are available for download (size 0.5 degree lon/lat), from the C-squares website. An Arctic subset was selected and then projected (WKID 3574). After this preparation a join operation was all that was needed to correctly show the species distribution on a map. 

To collect the statistics as defined in the OSPAR matrix approach for each species several operations were required to:

  • Determine the overlap with LMEs and estimate a population size (or similar) per bioregion (=LME).
  • For species that have a considerable presence outside the Arctic LMEs during the evaluated season, as was the case for several bird species, an estimate was also made of the numbers per area in use outside the Arctic LMEs. With this the total Arctic population could be extrapolated from the global number. Population sizes were for the most part taken from information presented on the ‘IUCN Red List’-website, as was the IUCN status (e.g Vulnerable or Near threatened).
  • Determine the proportion of the species population that could be considered protected based on an overlap with Marine Protected Areas (Adequacy/Viability). The MPA-database created during Phase I of this project was used for this analysis. The area of overlap was measured for the polygon-based datasets taken from the Bird of the World-database and the AquaMaps-distribution maps. The AquaMaps data has information on how suitable a given c-square is evaluated as and this was used to weigh the area accordingly (highly suitable squares weighted the most). A search distance of 25 km for point data was used so that any observation point up to that distance outside an MPA would still be considered protected. The reasoning behind this was that there is a chance that the observation position is not exact, the observation is subject to coincidence and the species are mobile. Most of the species could, within a fairly short period, reach the relative safety offered by an MPA. Note: the same search distance has also been applied for point data in relation to LMEs.
  • Determine the number of MPAs that contribute to the protection of a species (Replication).