Species richness is the most commonly used metric to quantify biodiversity. However, examining dark diversity, the group of missing species which can potentially inhabit a site, can provide a more thorough understanding of the processes influencing observed biodiversity and help evaluate the restoration potential of local habitats. So far, dark diversity has mainly been studied for specific habitats or largescale landscapes while less attention has been given to variation across broad environmental gradients or as a result of local conditions and biotic interactions. In this study, we investigate the importance of local environmental conditions in determining dark diversity and observed richness in plant communities across broad environmental gradients. Using the ecospace concept, we investigate how these biodiversity measures relate to abiotic gradients (defined as position), availability of biotic resources (defined as expansion), spatiotemporal extent of habitats (defined as continuity), as well as species interactions through competition. Position variables were important for both observed and dark diversity, some with quadratic relationships, e.g., plant richness showing a unimodal response to soil fertility corresponding to the intermediate productivity hypothesis. Interspecific competition represented by community mean Grime C had a negative effect on plant species richness. Besides position-related variables, organic carbon was the most important variable for dark diversity, indicating that in late succession habitats such as forests and shrubs, dark diversity is generally low. The importance of highly competitive species indicates that intermediate disturbance, such as grazing, may facilitate higher species richness and lower dark diversity.

The dataset consists of three subsets:

1) a dataset with plant (excl. bryophytes) observations from a large number of 5 m radius circular vegetation plots from terrestrial habitats (including many different habitats, also bogs, fens and other wet habitats, but not purely freshwater (lakes and streams) and not marine) in Denmark. These data were both collected as part of the Danish Monitoring of NATURA2000 sites (NOVANA) and as part of the Danish biodiversity research project called Biowide. The data consists of species lists from the aforementioned vegetation plots. This lists can be thought of as complete, i.e. species not in a given plot were most likely absent. The dataset was cleaned by lumping together subspecies and varieties and the species level, removing all non-plants, removing all hybrids and removing observations at a lower than species level (genus, familiy etc.). There are two related works given. One that documents the Biowide data and another one describing every detail of the whole NOVANA national monitoring programme, but this is only available in Danish. The file is called tmp-bionova.tsv.

2) Environmental factors used for statistical analysis. These were recorded as part of the Biowide project. This dataset also holds the calculated response variables for statistical analysis. The file is called responsevar-and-env-bw115.csv.

3) A dataset with all vegetation data from each Biowide site. Each site consisted of four 5 m radius vegetation plots (data for each plot available in (1)). These data were cleaned in the same manner as (1). The file is called biowide-site-matrix-115.csv.

If using this dataset please contact us, we used a lot of time to do the processing. Also please contact the project leader of the Biowide project, see https://bios.au.dk/om-instituttet/organisation/biodiversitet/projekter/data/ .