Skip to main content
Dryad

Small scale variability in soil moisture drives infection of vulnerable juniper populations by invasive forest pathogen

Cite this dataset

Donald, Flora et al. (2020). Small scale variability in soil moisture drives infection of vulnerable juniper populations by invasive forest pathogen [Dataset]. Dryad. https://doi.org/10.5061/dryad.3xsj3txc9

Abstract

The oomycete plant pathogen, Phytophthora austrocedri, is an aggressive killer of cypress trees causing severe mortality of Chilean cedar (Austrocedrus chilensis) in Argentina since the 1940s and now common juniper (Juniperus communis s.l.) in the UK. Rapid mortality of key UK juniper populations was first observed in the early 2000s; the causal agent of mortality was confirmed as P. austrocedri in 2012 and the pathogen has now been widely detected - but is not ubiquitous - in juniper populations across Scotland and England. Although juniper has a broad distribution across the northern hemisphere, the UK incidence of P. austrocedri remains the only confirmed infection of juniper populations globally. Juniper is an important species for biodiversity, so it is imperative to understand the abiotic and biotic drivers of emergent P. austrocedri infection to inform detection, containment and conservation strategies to manage juniper populations across the full extent of its range.

As management of UK juniper populations is primarily conducted at a local level, we investigated field scale drivers of disease – in three, geographically separate populations with different infection histories. Variation in the proportion of juniper showing symptoms - discoloured or dead foliage – was measured using stratified sampling across along key environmental gradients within each 100-hectare population, including juniper density identified from aerial imagery. Potential predictors of infection included altitude, slope, distance to nearest watercourse, soil moisture (mean percentage volumetric water content), area of red deer browsing damage and area of commonly associated vascular plant species. We assessed support in the data for alternative models explaining the spatial distribution of P. austrocedri symptoms using full subset covariate selection and Deviance Information Criteria (DIC). Despite differences in environmental gradients and infection histories between populations, area of juniper symptomatic for P. austrocedri increased with waterlogging, increasing with soil moisture in sites where soils had higher peat or clay contents, and decreasing with proximity to watercourses where sites had shallower, sandier soils.

These results are consistent with key drivers identified at both local and landscape scale in Chilean cedar. Our approach enables identification of site-specific disease management strategies including prioritisation of inspections in microsites with high soil moisture and promoting conservation measures such as creation of sites for natural regeneration in drier microsites to minimise pathogen spread and maximise the resilience of existing juniper populations.

Methods

Three infected juniper populations from where Phytophthora austrocedri had previously been isolated (Henricot et al., 2017) were selected to best represent the diversity of climatic, topographic and edaphic conditions occupied by juniper in the UK. In all three locations, the juniper population is a component feature of a Special Area of Conservation designated habitat and a qualifying interest of a Site of Special Scientific Interest (SSSI). Two populations are in Scotland: one in Perthshire (P, Glenartney SSSI) and one in the Cairngorms (C, North Rothiemurchus Pinewood SSSI), and one population is situated in the Lake District (LD, Birk Fell SSSI) in the north of England.

Juniper was sampled using 10 x 10 m quadrats from pre-selected locations stratified according to the area and density of juniper, altitude, slope and distance to watercourses. Quadrat sampling was carried out over five days at each location in October 2017. Quadrat centroids were geo-located using ArcPad v. 10.2 on a Panasonic FZ-GI tablet with GPS accuracy to 3 m and recorded as eastings “X” and northings “Y” in the British National Grid geographic reference system (datum = OSGB_1936). Number of sampled quadrats: P = 51, LD = 46, C = 50.

The following predictors were measured in each field quadrat.

“Area.of.symptoms” (m2) was estimated as a fraction of the "Area.of.juniper” (m2) present in each quadrat, where symptoms constituted foliage discolouration and dead needles (retained or dropped) that extended to a minimum of a whole branch and did not result from either browsing or mechanical damage.

Where a distinctive phloem lesion typical of P. austrocedri could be found, a 500 mg tissue sample was collected from one representative symptomatic tree per quadrat. The sampled tissue was stored at - 20 °C until quantitative real-time PCR (qPCR) could be carried out following the protocol described in Mulholland et al. (2013). A positive “qPCR.result” = 1; absence of symptomatic lesions or “not detected” “qPCR result” = 0.  

Area of juniper cover “Juniper.density” across 30 x 30 m including the 10 x 10 m quadrat was estimated as ≤ 20 % (0) or > 21 % (1) to distinguish between quadrats situated in isolated or contiguous juniper stands.

Area of “Berry.bearing” (m2) juniper was used to estimate the area of female juniper.

Area of “Herbivore.damage” (m2) was estimated as the area of bark stripping plus any resulting dead branches / stems (i.e. mechanical breakage from wind or snow damage was excluded).

Soil moisture was measured as % volumetric water content (VWC) using a FieldScout TDR 300 probe at 20 cm depth in Perthshire and the Cairngorms, and 3.8 cm depth in the Lake District. Measurements were collected from i) areas within each quadrat where juniper was absent, ii) under asymptomatic juniper and iii) under symptomatic juniper. An equal number of measurements (minimum four) was collected from each category present, resulting in eight to twelve point sample measurements from which the mean “Mean.soil.moisture” (% VWC) was calculated.

Area (m2) of vascular plant taxa present in each quadrat was recorded according to a target list (Appendix B) of taxa chosen to indicate placement of microsites along soil moisture, nitrogen and pH gradients. Where other tree species were present, canopy cover was included in the estimation of area. “Area.of.trees” (m2) is the combined total area of tree species present in each quadrat.

The Ellenberg F score attributed to each taxon was extracted from Hill, Preston, & Roy (2004). “Mean.Ellenberg.F.value” was calculated as a weighted mean of Ellenberg F scores based on the area of each taxon recorded in each quadrat.

The 50 m digital rivers network (Moore et al., 2000) was clipped to the boundary of each study population and merged with additional watercourses mapped in the field using the tracking function in ARCpad. “Watercourse.proximity” (m) was then calculated by measuring Euclidean distance to the nearest watercourses from quadrat XY coordinates using the gDistance function in the rgeos package (Bivand and Rundel, 2017).

The same methods were used to map deer and sheep tracks and lie-ups across each study site and calculate distance to nearest evidence of “Grazing.activity” (m). Activity was likely underestimated across all sites, but particularly in the Cairngorms and Lake District.

The remaining covariates were obtained from existing GIS datasets.

A digital elevation model (DEM) of each study site was supplied by NeXTPerspectives™ (updated February 2014) at 5 m resolution and averaged to 10 m using the aggregate function in the raster package (Hijmans, 2016) implemented in R v. 3.4.0 (R Core Team, 2017). Layers of slope and aspect were calculated from the resampled 10 m DEM using the terrain function in the raster package (Hijmans, 2016). Altitude (m), slope (°) and aspect (°) were extracted to the XY coordinate of each sampled quadrat.

“Soil.type” was extracted to quadrat centroids from 250 m resolution datasets, obtained from a digitised version of the soil map produced by Forbes (1984), the Soilscapes dataset (Farewell et al., 2011) and the National Soil Map of Scotland (James Hutton Institute, 2011) for the Perthshire, Lake District and Cairngorms populations respectively.

National Vegetation Classification (NVC) community data, supplied by Scottish Natural Heritage (2017), was included as a covariate for the Perthshire and Cairngorms populations. The eight Perthshire communities were simplified to four broad types, amalgamated as follows: acid grasslands (U4 Festuca ovina - Agrostis capillaris - Galium saxatile; U20 Pteridium aquilinum - Galium saxatile; U24 Arrhenatherum elatius - Geranium robertianum), mires (M10 Carex dioica Pinguicula vulgaris; M23 Juncus effusus/acutiflorus Galium palustre) and mosaic communities suggesting transition from drier to wetter soil (U5 Nardus strictaGalium saxatile; M15 Trichophorum germanicum - Erica tetralix wet heath) (Rodwell, 1991).

Topographic wetness index (TWI) is calculated as ln(a/tanb) where a is the specific catchment area and b is the local slope (Raduła, Szymura and Szymura, 2018). There are three stages to TWI calculation:

  1. Preparation of the DEM and slope (and river) rasters
  2. Creation of a flow accumulation raster from the DEM using a flow routing algorithm
  3. Creation of TWI raster using the flow accumulation and slope rasters

As there are so many pre-processing and flow routing algorithm options, a validation step was added to the workflow in order to choose the best TWI layer developed for each population on the strength of the Pearson correlation between the TWI value and mean soil moisture (% VWC) obtained for each sampled quadrat.

The water catchment for each study population was downloaded from the National River Flow Archive (https://nrfa.ceh.ac.uk/, accessed March 2018) and imported to SAGA GIS v. 2.3.2 (Conrad et al., 2015). TWI calculation using the 10 m DEM performed well in the Perthshire and Lake District populations but poorly in the Cairngorms population where 25 m resolution yielded a stronger correlation. Each DEM was clipped to extent of the catchment and prepared using the “fill sinks XXL Wang/Liu” algorithm to prevent the premature termination of drainage that can occur where local elevation minima have no lower neighbours as an artefact of DEM generation and do not relate to terrain features (Reuter et al., 2009). The algorithm progressively increases the elevation values of these “sinks” until the lowest elevation value is found from which the water can “spill” out into the rest of the cell (Wang and Liu, 2006). As the DEM used for the Cairngorms population was of a coarser resolution, explicitly incorporating the river network using the “burn stream network to DEM” tool improved the calculation. This tool lowers the value of each DEM cell that interfaces with a watercourse pixel by the minimum elevation in a neighbouring cell minus one (Conrad et al., 2015). The flow routing algorithm that performed best for the Cairngorms and Perthshire populations was Mass Flux, which divides each pixel into quarters and routes flow at that scale so flow can be routed in different directions where this is ambiguous (Gruber and Peckham, 2009). The recursive implementation of the deterministic eight algorithm, which apportions flow from each grid cell to a single adjacent cell through the steepest upslope gradient (and uses arbitrary assignment where flow direction is unclear), led to the strongest correlation and, therefore, best performing TWI layer for the Lake District population (O’Callaghan and Mark, 1984).

Usage notes

Bivand, R. and Rundel, C., 2017. rgeos: Interface to Geometry Engine - Open Source (GEOS). R package https://cran.r-project.org/web/packages/rgeos/rgeos.pdf

Conrad, O., Bechtel, B., Bock, M., Dietrich, H., Fischer, E., Gerlitz, L., Wehberg, J., Wichmann, V., Böhner, J., 2015. System for Automated Geoscientific Analyses (SAGA) v. 2.1.4. Geosci. Model Dev. 8, 1991–2007. https://doi.org/10.5194/gmd-8-1991-2015

[dataset] Farewell, T.S., Truckell, I.G., Keay, C.A., Hallett, S., 2011. Use and applications of the Soilscapes datasets. http://www.landis.org.uk/downloads/downloads/Soilscapes_Brochure.pdf

[dataset] Forbes, A.R.D., 1984. Glen Artney Juniper Wood: An ecological study. Honours thesis (unpubl.). Stirling University, Stirling.

Gruber, S., Peckham, S., 2009. Chapter 7 Land-Surface Parameters and Objects in Hydrology, in: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, Applications. pp. 171–194. https://doi.org/10.1016/S0166-2481(08)00007-X

Henricot, B., Pérez-Sierra, A., Armstrong, A.C., Sharp, P.M., Green, S., 2017. Morphological and genetic analyses of the invasive forest pathogen Phytophthora austrocedri reveal that two clonal lineages colonized Britain and Argentina from a common ancestral population. Phytopathology 107, 1532–1540.

https://doi.org/10.1094/PHYTO-03-17-0126-R

Hijmans, R.J., 2016. raster: Geographic Data Analysis and Modeling. R package version 3.0-2. https://CRAN.R-project.org/package=raster

Hill, M.O., Preston, C.D., Roy, D.B., 2004. PLANTATT Attributes of British and Irish plants: status, size, life history, geography and habitats. NERC Centre for Ecology and Hydrology, Cambridge.

[dataset] James Hutton Institute, 2011. 1:250,000 Soil map (National soil map of Scotland). https://www.hutton.ac.uk/learning/natural-resource-datasets/soilshutton/soils-maps-scotland/download#soilmapdata

[dataset] Moore, R.V., Morris, D.G., Flavin, R.W., 2000. CEH digital river network of Great Britain (1:50,000). https://catalogue.ceh.ac.uk/documents/7d5e42b6-7729-46c8-99e9-f9e4efddde1d

Mulholland, V., Schlenzig, A., Macaskill, G.A., Green, S., 2013. Development of a quantitative real-time PCR assay for the detection of Phytophthora austrocedrae, an emerging pathogen in Britain. For. Pathol. 43, 513–517. https://doi.org/10.1111/efp.12058

O’Callaghan, J.F., Mark, D.M., 1984. The extraction of drainage networks from digital elevation data. Comput. Vision, Graph. Image Process. 28, 323–344.

R Core Team, 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. URL https://www.R-project.org/.

Raduła, M.W., Szymura, T.H., Szymura, M., 2018. Topographic wetness index explains soil moisture better than bioindication with Ellenberg’s indicator values. Ecol. Indic. 85, 172–179. https://doi.org/10.1016/j.ecolind.2017.10.011

Reuter, H.I., Hengl, T., Gessler, P., Soille, P., 2009. Chapter 4 Preparation of DEMs for Geomorphometric Analysis, in: Hengl, T., Reuter, H.I. (Eds.), Geomorphometry: Concepts, Software, Applications. Elsevier B.V., pp. 87–120. https://doi.org/10.1016/S0166-2481(08)00004-4

Rodwell, J.S., 1991. British Plant Communities Volumes 1:5. Cambridge University Press, Cambridge.

[dataset] Scottish Natural Heritage, 2017. National Vegetation Classification. https://gateway.snh.gov.uk/natural-spaces/dataset.jsp?dsid=NVC

Wang, L., Liu, H., 2006. An efficient method for identifying and filling surface depressions in digital elevation models for hydrologic analysis and modelling. Int. J. Geogr. Inf. Sci. 20, 193–213.

Funding

Scottish Forestry Trust

Forest Research

Scottish Forestry

Scottish Natural Heritage

Royal Botanic Garden Edinburgh

Scottish Forestry Trust

Forest Research

Scottish Forestry

Royal Botanic Garden Edinburgh