Appendix C. Climate Suitability (Biodiversity)
Methodology
To evaluate the potential benefits of newly created woodland for priority woodland species, we used climate suitability maps created from previous work commissioned by Natural England. This report, published in 2015, assessed the impact of climate change on 3,000 species across Great Britain. This analysis was later extended to a total of 4,447 species in a 2022 publication.
Here, we outline the key modelling steps and parameters used by Critchlow et al. (2022) and Pearce-Higgins et al. (2015) to predict future suitability under climate scenarios.
- Species distribution data were acquired from the Biological Records Centre and British Trust for Ornithology at 10 km resolution.
- Four climatic variables were included in the models, using Met Office datasets averaged from 1961 to 1990 UK: 1) mean temperature of the coldest month, 2) growing degree days, 3) the coefficient of variation of temperature and 4) soil moisture.
- Future climate projections for the UK (2070–2099) were derived from UKCP09, based on the Intergovernmental Panel on Climate Change’s (IPCC) low (‘B1’) and medium-to-high (‘A1B’) emission scenarios. The B1 scenario represents 2°C of global warming by the end of the century, while the A1B scenario corresponds to 4°C of warming.
- A Bayesian, spatially explicit Generalised Additive Model (GAM) was used to link species distribution data with climate data and map these relationships across space. This modelling approach also takes into account differences in how much effort people put into observing and recording species.
- Climate suitability maps were created for each species under baseline and future climates.
For this study, we extracted a subset of the species assessed by Pearce-Higgins et al. (2015) and Critchlow et al. (2022). We began by filtering the species list to include only ‘priority species’—those that had been identified as of principal importance in England under Section 41 of the Natural Environment and Rural Communities Act 2006 (British Parliament, 2006) (n = 260). Next, we refined the list further by selecting species significantly associated with habitats being created through the NCF planting, including all woodland and trees outside woodland types. To determine these habitat associations, we relied on expert judgment using Natural England’s Species Evidence Base (SEB). The SEB contains two fields relevant to species’ habitat associations: one listing habitats of critical importance and another listing habitats of secondary importance. If NCF habitat types appeared in either field, we classified the species as being associated with those habitats. This resulted in 30 species being included in the refugia indicator, of which 25 are associated with woodland habitats and 17 with TOW habitats (see Table C1).
Table C1. Summary of bird and plant species included in the refugia biodiversity indicator. These species are of priority conservation concern in England, are associated with woodland and/or TOW habitats, and have baseline and future climate suitability maps available for the UK (as modelled in Critchlow et al., 2022).
Taxonomic group | Number of species | Species |
|---|---|---|
Birds | 13 | Cuckoo Cuculus canorus |
Grasshopper Warbler Locustella naevia | ||
Grey Partridge Perdix perdix | ||
House Sparrow Passer domesticus | ||
Linnet Linaria cannabina | ||
Marsh Tit Poecile palustris | ||
Nightjar Caprimulgus europaeus | ||
Spotted Flycatcher Muscicapa striata | ||
Tree Pipit Anthus trivialis | ||
Tree Sparrow Passer montanus | ||
Turtle Dove Streptopelia turtur | ||
Wood Warbler Phylloscopus sibilatrix | ||
Yellowhammer Emberiza citronella | ||
Non-vascular plants | 4 | Chalk Screw-moss Tortula vahliana |
Lesser Squirrel-tail Moss Habrodon perpusillus | ||
Ribbonwort Pallavicinia lyellii | ||
Thatch-moss Leptodontium gemmascens | ||
Vascular plants | 13 | Annual Knawel Scleranthus annuus |
Copse-bindweed Fallopia dumetorum | ||
Coral-necklace Illecebrum verticillatum | ||
Crested Cow-wheat Melampyrum cristatum | ||
Deptford Pink Dianthus armeria | ||
Grape-hyacinth Muscari neglectum | ||
Juniper Juniperus communis | ||
Small-white Orchid Pseudorchis albida | ||
Spreading Bellflower Campanula patula | ||
Tower Mustard Arabis glabra | ||
White Helleborine Cephalanthera damasonium | ||
Yellow Bird's-nest Hypopitys monotropa | ||
Yellow Centaury Cicendia filiformis |
For each selected species, we converted continuous probabilities of suitability to binary predictions (0-1) according to the threshold that maximised the True Skill Statistic (TSS) (Liu et al. 2005; 2013). TSS is a widely used metric for evaluating model performance, calculated as TSS = Sensitivity + Specificity – 1, where Sensitivity measures the proportion of correctly predicted presences, and Specificity measures the proportion of correctly predicted absences. To calculate sensitivity, we used independent occurrence datasets downloaded from the National Biodiversity Network (NBN) Atlas. We cleaned the NBN data by removing historical records (pre-1961) and records with high coordinate uncertainty (>5000 m). We then visually assessed each cleaned occurrence dataset to ensure that the data was broadly representative of the species range in Great Britain (Csergő et al. 2017) (this step resulted in the exclusion of five species: Coccothraustes coccothraustes, Crepis mollis, Emberiza cirlus, Lullula arborea and Ophrys insectifera). We referenced the following sources for species distributions: BirdLife International range maps (2024) for birds, the Botanical Society of Britain and Ireland Atlas (2020) for vascular plants and the British Bryological Society Atlas (2014) for bryophytes. To calculate specificity, pseudo-absence datasets were created by randomly selecting up to an equal number of points (limited to one per raster grid cell) as the cleaned recorded presences across Great Britain.
The binary distribution maps were stacked and summed to create a heatmap representing the richness of priority woodland and TOW species across Great Britain for baseline climate conditions and the two future climate change scenarios. To estimate the number and composition of priority species that could benefit from the newly created woodland habitats, we compiled a list of species whose climatic ranges intersected with the planting polygons. These outputs allowed us to estimate baseline and future species richness at the planting and LA levels, as well as species gains and losses in the future.
Sensitivity test
To test the sensitivity of the species richness and turnover estimates to the method of binary conversion, we also produced planting- and LA-level outputs following the fixed-threshold approach adopted by Critchlow et al. (2022). This method involves converting continuous probabilities to binary based on a 0.01 threshold (≥0.01 = suitable, <0.01 = unsuitable).