Environmental drivers of Liparis loeselii decline in Denmark

Vegetation and environmental properties from extant, potential, and historical occurrence sites of Liparis loeselii (Orchidaceae) in Denmark, collected in order to develop habitat suitability models from Ellenberg Indicator Values and field-measured environmental properties, and in order to identify the primary reasons for the observed decline of L. loeselii in Denmark and beyond.

Methods
The data were combined from existing nature surveillance data and targeted collection of from Liparis loeselii populations using identical methods. We extracted data from all 285 sites (i.e. protected areas, variable in size, but typically several to tens of hectares) in The Danish National Monitoring Programme, NOVANA (https://novana.au.dk/) 2004-2012, which included plots (i.e. circular quadrats of 78.5 m2) classified as fen in a broad sense (i.e. rich fen, alkaline spring, floating fen, wet meadow, humid dune slack and alkaline fen with Cladium mariscus (L.) Pohl. In a first step, we pruned the data so that for repeatedly monitored plots, only the latest recording was kept, and monitoring sites with less than four survey plots were excluded. This procedure resulted in a data set of 17,436 vegetation quadrats. In a second step, individual plots of non-fen vegetation, which had been included due to the random placement of recording plots at monitoring stations, were excluded. This was done using a poisson regression of the presence of 27 typical rich fen species (from Ejrnæs, R., Nygaard, B., Fredshavn, J. R., Nielsen, K.-E., & Damgaard, C. (2009). Terrestriske Naturtyper 2007 – NOVANA. http://www.dmu.dk/Pub/FR712.pdf) as a function of community-mean Ellenberg indicator values for moisture, nutrients and soil pH. Plots with effectively unsuitable habitat conditions (too dry/wet, eutrophied or acidic) were identified by having a predicted number of typical fen species smaller than 0.5. No plots occupied by L. loeselii were excluded in this step.

Since monitoring plots are scattered randomly within each site and because L. loeselii is locally scarce, even within sites of occurrence, very few sampling quadrats included L.loeselii (i.e. 30 plots). In order to better cover the habitat variation of the focal species, we collected supplementary data at eight sites with L.loeselii populations (Vandplasken, Nørlev Kær, Ajstrup Kær, Tved Kær, Urup Dam, Helnæs Made, Even, Forklædet). Six of the sites with supplementary plots were already part of the monitoring programme, but without L. loeselii in any sampled quadrats. The supplementary data were collected in June through August 2013. After data trimming and addition of supplementary data, our data set contained 4,479 quadrats distributed over 270 sites of potentially suitable habitat, of which 88 plots from 8 sites contained L. loeselii.

In the monitoring scheme, lists of vascular plant and bryophyte species were recorded along with visual assessment of vegetation structure from standard monitoring plots of 78.5 m2 (circle with radius = 5 m). Assessment of vegetation structure included vegetation height (calculated as an average of four measures in the centre of the plot) and cover (in percentage) of trees and shrubs in two height classes (above and below 1 m tall). An identical protocol was used for the collection of supplementary data. GPS coordinates were available for all plots with corresponding data on elevation above sea level and a calculated Topographical Wetness Index, which is a proportional measure of the water retention potential in a given plot.

In a subset of plots in our main data, the following soil properties had been assessed as part of the national monitoring scheme:

Substrate pH was measured either – at wetter sites – directly in rhizosphere water using a combination glass electrode or – at drier sites – in a 0.01 M CaCl2 suspension of soil at 21°C with a combination glass electrode (available from 352 plots).
Plant available soil P was measured by extracting fresh soil with 0.5 M NaHCO3, filtration, addition of H2SO4 and estimation of P concentration by spectrophotometry at 890 nm (following Banderis et al. 1976) (available from 194 plots).
Nitrate in water was analysed with ion chromatography or flow injection analysis (available from 282 plots). The sample was injected in a stream of ammonium chloride and passed over a cadmium reactor, where nitrate was reduced to nitrite. Nitrite was reacted with sulphanilamide and naphtylethylenediamine and the resulting azo-colour was spectrophotometrically determined at 540 nm.
Conductivity was measured in soil water in the field using a conductivity meter (available from 348 plots).
Total N content in bryophyte tissue was measured using a LECO CNS-2000 analyser (Eurofins, Vejen, Denmark). Dried plant material was incinerated at 1100 °C. After measuring CO2, using infrared spectroscopy, N oxides were reduced to N using a copper catalyst, and CO2 was removed by absorption before measuring N in a thermal conductivity cell (available from 250 plots).
Total P content in bryophyte tissue was measured using “Danish Standard” (DS259:2003/SM3120:2005-ICP-OES), in which plant material is treated with HNO3 and heated in a closed test tube at 120 °C and P content measured using atomic absorption spectrophotometry (Eurofins, Vejen, Denmark) (available from 201 plots).

Shoots of Calliergonella cuspidata (Hedw.) Loesk. was collected as the standard species for tissue nutrient analysis, except in 10 plots, from which C. cuspidata was absent and, therefore, Campylium stellatum (Hedw.) C.E.O. Jensen was collected instead. In total, i.e. data combined from the main data and our own supplementary records, a set of 201 plots had direct measurements of all above listed environmental variables, out of which 31 plots contained L. loeselii.

In addition to direct measurements, we used community-mean Ellenberg Indicator Values as indirect measures of soil pH (EIV-reaction), moisture (EIV-moisture), light availability (EIV-light) and nutrient availability (EIV-nutrient). In addition, we used the ratio EIV-nutrient/EIV-reaction (hereafter referred to as nutrient ratio), which works as an indicator of eutrophication.