Context: Both climatic extremes and land-use change constitute severe threats to biodiversity, but their interactive effects remain poorly understood. In forest ecosystems, the effects of climatic extremes can be exacerbated at forest edges.

Objectives: We explore the hypothesis that an extreme summer drought reduced the richness and coverage of old-growth forest species, particularly in forest patches with high edge exposure.

Methods: Using a high-resolution spatially explicit precipitation dataset, we could detect variability in drought intensity during the summer drought of 2018. We selected 60 old-growth boreal forest patches in central Sweden that differed in their level of drought intensity and amount of edge exposure. The year after the drought, we surveyed red-listed and old-growth forest indicator species of vascular plants, lichens and bryophytes. We assessed if species richness, composition, and coverage were related to drought intensity, edge exposure, and their interaction.

Results:  Species richness was negatively related to drought intensity in forest patches with a high edge exposure, but not in patches with less edge exposure. Patterns differed among organism groups and were strongest for cyanolichens, epiphytes associated with high-pH bark, and species occurring on convex substrates such as trees and logs.

Conclusions:  Our results show that the effects of an extreme climatic event on forest species can vary strongly across a landscape. Edge exposed old-growth forest patches are more at risk under extreme climatic events than those in continuous forests. This suggest that maintaining buffer zones around forest patches with high conservation values should be an important conservation measure.

The data consists of 60 inventoried old-growth boreal forest patches in central Sweden that differed in their level of drought intensity and amount of edge exposure. In these sites we conducted surveys of red-listed and old-growth forest indicator species of bryophytes, lichens and vascular plants (a subset of the species list from the monitoring program of woodland key habitats, see more in the main article method section). Inventories were conducted between the end of June and mid-August 2019. We subdivided each site into 20 × 20 m subplots. For each present species, we estimated their total cover in each subplot (dm²) and in the case of the orchid Goodyera repens (L) R.Br., we also estimated the percentage of flowering individuals for the first encounter in each subplot. We categorized all surveyed species based on organism groups and substrate association. Organism groups were: lichens (divided into cyanolichens and chlorolichens), bryophytes and vascular plants. For substrate association we separated species as epiphytic (divided into high-pH bark and low-pH bark), epixylic, epilithic and epigeic.

Drought intensity was calculated based on a high-resolution precipitation dataset called HIPRAD (HIgh-resolution Precipitation from gauge-adjusted weather RADar, Berg et al. 2016). We calculated three drought measures for each site: i) total precipitation during the whole summer of 2018 (1^st May - August 31^st); ii) precipitation during the extreme drought period during the summer of 2018 (22^nd June - 27^th July); iii) the summed precipitation before and after this extreme drought period in ii (1^st May - 21^st June and 28^th July - 31^st August).

We calculated two background climatic variables: the average summer precipitation from 1^st May to 31^st August over the years 2010 till 2017, and growing degree days (GDD, measured as days with mean temperatures over 5°C (based on the dataset from Meineri & Hylander, 2017).

Data on the size and vegetation type in the 60 sites (woodland key habitats) from the Swedish Forest Agency. The amount of laying and standing dead wood was counted (trees > 30 DBH), and the canopy cover was determined based on five hemispherical photos, processed in ImageJ.

Edge effects and edge exposure were determined by categorizing the strength of microclimatic edge effects for each subplot in each site as either strong, weak or absent, based on previous measures of how microclimatic edge effects penetrate edges with different edge orientation at forest/clear-cut interfaces. To get a measure of edge exposure for each site, we calculated as the proportion of subplot-area that were exposed to edge effects (including both weak and strong edge effects).

More details can be found in the method section 2.1 - 2.4 in the paper. 

The following datasets are included:

1) Sitelevel data for the 60 woodland key habitats, including: woodland key habitat ID (Swedish Forestry Board, Betecknings_ID), coordinates, Site size, standing and laying dead wood, canopy cover, presense of high pH bark, edge exposure, the drought indices, and the background climatic variables. 

2) A species coverage by site matrix

3) Species richness of the different groups at site level

4) Species richness at subplot level

5) The coverage of the four common species at site level

6) The coverage of the four common species at subplot level (four files, one for each species). 

A list of the species organism group and substrate affility can be found in Online appendix, Table S2.