Study areas

4.3.1. Palokangas area (I, II, V)

The Palokangas study area, totaling of 34.5 km², is located in Ilomantsi of eastern Finland (62°53′N, 30°54′E, Figure 2). Managed Scots pine forests, owned by the Tornator Ltd are mainly growing on dry and dryish forest site types. The relief within the area is relatively flat with a mean elevation above the sea level of 170 m. In 2009, the majority of the stands were young and middle-aged, with average age of 53 years and diameter-at-breast-height (dbh) of 14.7 cm (Talvitie et al. 2011).

The initial outbreak by the common pine sawfly was already visible at the area in 1999.

The outbreak initiated when the widespread outbreak spread to the east and reached the region (section 4.2.). Within the region, the common pine sawfly caused growth losses and Scots pine mortality within an area of circa 10,000 ha. The population density, damage intensity, and spatial extent have been fluctuating ever since. Peak population densities at gradation phase were observed during 2000-2002, and again at 2005. After 2005, population densities have remained at relatively high postgradation phase, showing some chronic nature.

Last time the area was visited in the spring of 2016, new defoliation from the previous year was apparent (Kantola T., personal observation). The forest owner has conducted large salvage loggings within the area due to the prolonged outbreak. Typically, the infested Scots pine in the Palokangas area were also infested by secondary pests of pine shoot beetle (Tomicus piniperda L.) and lesser pine shoot beetle (Tomicus minor Hartig) (Kantola T., personal observations).

Figure 2. Locations of Palokangas study area and locations of Adaptive cluster sampling plots and clusters within the study area. (Map by ESRI/HERE©)

4.3.2. Outokumpu area (V)

Outokumpu study area is located in eastern Finland (62°46´N, 28°57´E), west of the Palokangas area (Figure 1). The landscapes are generally flat and highly fragmented. The landscape mosaics are comprised by agricultural fields, lakes, and residential components in addition to forests. The forested areas are mainly Scot pine dominated intensively managed commercial forests, targeting for fiber and timber production. Similar to the Ilomantsi area, the main site types comprise poor and dry heath (Calluna type), quite poor and dryish heath (Vaccinium type) (Cajander 1926). However, medium fertile fresh heath (Myrtillus type) is more predominant in Outokumpu than in the Palokangas area. In general, the forest stands in Outokumpu are more fertile than in the Ilomantsi area. A typical size of a forest stand varies between one and three ha. Regarding the most recent pine sawfly outbreaks in these regions, the forests in the Outokumpu study area have a shorter disturbance history than in the Ilomantsi district. The most recent gradation phase of the European pine sawfly initiated in 2008 within an area of circa 50,000 ha. The sawfly population obtained the peak densities in 2011 and started to collapse into postgradation phase since then.

4.3.3. Abisko area (V)

The Abisko study area is located in Northernmost Sweden, near Norwegian border (68.35°N, 18.82°E, Figure 1). These sub-alpine forest landscapes are featured by mountain birch forests, mires, and heath vegetation with high abundance of dwarf shrubs, grasses, and lichens (Wielgolaski 2001). The Abisko region has a history with outbreaks by both autumnal moth and winter moth (Bylund 1995; Tenow 1996). The unmanaged mountain birch stands are characteristically infested by these species in a time interval of 9−10 years. Two most recent moth outbreaks occurred in 2004 and 2012−2013 (Heliasz et al. 2011; Olsson et al.

2016b). The subarctic region, including Abisko area has been subjected to increasing temperatures in the last decades causing, e.g., disappearing of permafrost and changes in the vegetation communities (Callaghan et al. 2010; Johansson et al. 2006). Elevated temperatures during this century have led to mean annual temperatures above 0 °C in the area (Callaghan et al. 2010) Warming climate, particularly less frequent extreme winter temperatures support the overwintering egg survival (Tenow 1972; Callaghan et al. 2010). The magnitude of climate change driven impacts on the population dynamics of these moth species and further their impact on the mountain birch ecosystems are still unknown (Jepsen et al. 2008; Heliasz et al. 2011).

4.3.4. The Linville River Gorge area (III, IV)

Target areas of the sub-studies III and IV are located in the Lower Linville River area, in the Grandfather Ranger District (35˚56’N, 81˚55‘W) of the Pisgah National Forest, Southern Appalachians, NC, USA (Figure 3). In the sub-study III, the study area comprised the Lower Linville River Watershed, with the spatial extent of approximately 60 km². For the sub-study IV, the study area was delineated by available remote sensing data sets (section 4.4.). The area of 40 km² is located in the northern part of the Lover Linville River area and much of it (in the center) overlaps with the study area of the sub-study III.

The Linville Gorge wilderness comprises most of these two study areas. The great majority of the wilderness area have been left unlogged and the wilderness is included to the greatest old-growth forests within the region. The area has suffered from frequent forest fires

(Newell and Peet 1998; Wimberly and Reilly 2007). Topography in the area is very complex and prominent cliff-like bluffs are dividing upper and lower slopes (Wimberly and Reilly 2007). The elevation of terrain ranges from 350 m to 1300 m, from the southern part and the river basin to the upper ridges in the north. Three major ecological zones comprise the area:

Acidic Cove, Xeric Pine-Oak Heath and Oak Heath, and Mesic Oak-Hickory (Simon et al.

2005). The region’s climate is humid and temperate. Characteristics of the area provide a wide range of habitats that are suitable for over 400 vascular plant species and a vast diversity of tree species (Schafale and Weakley 1990; Simon et al. 2005). Both of the susceptible hemlock species, eastern and Carolina hemlocks are abundant in the area (Jetton et al. 2008).

Hemlock woolly adelgid was observed in the area for the first time at the early beginning of this century (Koch et al. 2006). The species have caused substantial hemlock mortality in the region.

Figure 3. Locations of the two study areas, and the overlapping area in the lower Linville River region, North Carolina, USA. (Map by ESRI/HERE©)

4.3.5. Introduced and native ranges of the hemlock woolly adelgid (VI)

The in ENM modeling task (sub-study VI), areas for background evaluation extent and model projections have to be delineated (Figures 4 and 5). Background evaluation extent describes the area, within the niche model is calibrated while the projection areas comprise the extent to with the calibrated models are projected under contemporary or future climatic conditions.

Normally, ENMs for invasive species are calibrated within the native range and then projected into the novel environments. However, in case of limited number of species occurrence observations within the species’ native distribution may fail to cover the extent of suitable habitats and the developed models may lack transferability to new ranges or spatial scales (Menke et al. 2009; Peterson et al. 2011). Further, ENMs calibrated in highly differing areas with different environmental gradients are prone to errors when projected elsewhere (Menke et al. 2009). Errors can be even more pronounced if models of alien invasive species altering their niche requirements are projected under future climate scenarios (Urban et al.

2007). Due to the inadequate, sparse, and clustered pattern of available HWA observations in the continental East Asia, the models were calibrated within the introduced range in the eastern North America, were plenty of information in regard to the species exist. The HWA observations, described in the section 4.5., were utilized in the delineation of the background evaluation extent. It is not yet known if the HWA is yet in equilibrium and have invaded its whole range in the eastern North America. For that reason, we delineated the background evaluation extent with a narrow 20 km buffer from a convex hull polygon of the species occurrence observations (Figure 5). The calibrated models were projected to the eastern North America and reverse-casted to the native ranges of the species in East Asia and western North America. The projection areas in the both eastern and western North America were buffering 800 km and 2000 km in East Asia around the available presence points.

Figure 4. Counties with at least one HWA observation by 2018 and native distribution of eastern hemlock (Little 1971) within the eastern North America.

4.4. Remote sensing data sets