Dynamic distribution modelling

Eva Lieungh

Natural History Museum, University of Oslo, the Norwegian Ministry of Education and Research

Field surveys were carried out in 2009, 2019, and 2020. All vascular plant species inside the 15 habitat patches were recorded in 2009 and 2019, but only the habitat specialists' occurrences are reported in this dataset. In 2020, occurrence was only recorded for the 49 habitat specialists.

Fieldwork in 2009 was carried out by KB, OES, and AKW, all experienced botanists. Potential locations of dry calcareous grassland had been identified by aerial photos and distribution modelling. The patch locations were then visited by the fieldwork team and either discarded or confirmed and delimited. Polygon coordinate points were recorded using a handheld GPS with an estimated accuracy of 1-10 m. All three in the team looked for species in the same patch, one being in charge of writing down occurrences. When no new species were found, all three tried to find one more species each before declaring the list complete. Species lists per patch were noted in a notebook. Fieldwork in the 2019 and 2020 surveys was carried out by EL alone. Patches were found using a map with semi-transparent polygons over a recent aerial photo. Species presences were recorded in a spreadsheet on a smartphone. Patches were inventoried by carefully walking back and forth across the polygon to cover the entire area. EL took pictures of each patch. EL also took pictures of uncertain species with the iNaturalist (iNaturalist 2008) app to aid identification after fieldwork as it is illegal to collect plants inside the protected area.

We took several steps to quality-check the data. Each occurrence was evaluated manually to ensure the highest possible quality of data: after data collection was complete in 2019 and 2020, each observation was checked against the data from the previous year(s), occurrences in GBIF, and pictures taken during fieldwork. Some uncertainty was resolved using pictures stored in iNaturalist, where pictures were stored for individual plants along with their coordinates and other metadata. Some of these pictures were sent to OS, SLO and other experts to confirm uncertain species identifications. Additional trips to Gressholmen by EL along with SLO were carried out in September 2019, and alone in June 2020, to check uncertain presences and suspected false absences.

For 2009 and 2019, there are more data available than described here. These data include all vascular plant species, not just the specialists, and some additional descriptive variables for each patch.

In spite of quality control, some uncertainty remains. Delineating polygons inevitably involves uncertainties, and GPS polygon boundaries were not physically delimited in the field. In addition, changes in patch size may have occurred between 2009 and 2019, due to regrowth or other factors. To a lesser degree, this is also an issue between the 2019 and 2020 surveys. False absences may occur in the dataset, which is common in field surveys (Morrison 2016), either from incomplete survey effort or because the species was not detectable at the time of fieldwork. As flowering phenology varies, early-flowering species (e.g. Veronica arvensis, Draba verna) are especially susceptible to being omitted, even though special care was taken to look for them. Misidentification could also occur, though the availability of previous data and use of trained botanists with expert knowledge of the local flora should minimise this risk.

Calcareous grasslands occur naturally fragmented on shallow soils by the Oslo Fjord, mainly on exposed and/or grazed Cambrian–Silurian marine sedimentary rock. The climate is relatively warm, with an annual temperature of 5.7°C (Aune 1993) and annual precipitation of 763 mm (Førland 1993). The area is amongst the most species-rich in Norway, as the combination of calcareous bedrock and warm summers is rare and hosts species that are mostly found further south in Europe.

The species were recorded in 15 patches of dry calcareous grassland on Gressholmen Island, situated in the inner Oslo fjord (Fig. 1). Habitat patches were identified through a combination of distribution modelling at 5 x 5 m resolution, interpretation of aerial photos, and in the field (Wollan et al. 2011). The mean patch size was 1078 m², range 264 m² to 2382 m². Total habitat area, as the sum of all patch areas, was 48516 m².

Figure 1.  

Study area on Gressholmen (left) and location in northern Europe (right). Numbered habitat patches are shown in yellow on Gressholmen Island.

Humans have shaped and impacted the habitats on Gressholmen, probably for as long as the Oslo fjord has been settled. Today, the Island is artificially connected to two other Islands, Heggholmen and Rambergøya. Most of this Island trio is protected as nature reserves, but has previously housed a shooting range, an airport, a soap factory with connected houses, and a paint factory. It has likely been grazed by domestic animals while the Island was inhabited. From the 1970s until its eradication in 2007, a large rabbit population occupied the Islands and grazed the vegetation heavily. Several summerhouses, a small harbour, and a restaurant are still in use and the Islands are a popular destination for day-trips in the summer.

59.882 and 59.887 Latitude; 10.717 and 10.727 Longitude.

The dataset encompasses 49 habitat specialist vascular plant species. The list of habitat specialist species was developed, based on flora information of habitat requirements and distributional range, expert opinion, and field observations from a large number of polygons (see Evju and Sverdrup-Thygeson 2016). More vascular plant species were recorded in the first two years and are available upon request. Field surveyors used Norsk Flora (Lid and Lid 2005) and Gyldendals store nordiske flora (Mossberg and Stenberg 2012) for species identification. Species names were checked and updated using the Species Nomenclature Database (Norwegian Biodiversity Information Centre 2022) and matched with the best possible species name in GBIF. The list of taxa included follow the GBIF names as published in the dataset. The corresponding list of species names following the Norwegian Species Nomenclature Database is published on GitHub. Common names are given in Norwegian to provide an additional link between the different species lists, including to the original data collection file where some species names were outdated.

This work is licensed under a Creative Commons Attribution (CC-BY) 4.0 License.