Introduction

Disturbance of subarctic vegetation has increased during past decades due to increased human activities in the high latitudes (Forbes, 2005). Trampling, camping, industrial activities, and use of all-terrain vehicles cause vegetation damages, having substantial ecological effects on plant communities, soil composition, and microclimate of the sites (e.g., Marion and Cole, 1996; Forbes et al., 2005). Vegetation and topsoil disturbance change species richness, floristic composition, and biomass of plants, increase soil erosion and compaction, and reduce water holding capacity of soil (e.g,Tolvanen et al., 2001; Forbes et al., 2005; Tolvanen and Kangas, 2016). High-latitude regions are characterized by short growing season, low temperatures, low availability of nutrients and thin humus layers (e.g., Bliss, 1962; Billings, 1987), restricting plant growth and recovery after disturbance. Besides environmental factors, plants' tolerance to disturbance and regeneration after damage are much determined by the specific characteristics of an individual plant species. In general, graminoids and deciduous dwarf shrubs have better tolerance to trampling and higher regeneration rate after disturbance than cushion plants and evergreen dwarf shrubs, due to their high photosynthesis rate and belowground storages of nutrients (Chapin, 1980; Bryant et al., 1983; Karlsson, 1985; Cole, 1995a, 1995b;Tolvanen et al., 2001).

Despite relatively large-scale research on the effects of trampling and soil removal on vegetation (e.g., Cole, 1995a, 1995b; Gellatly et al., 1986; Marion and Cole, 1996; Hautala et al., 2001; Törn et al., 2006), relatively little is known on the long-term recovery of vegetation in the high latitudes. Most of the studies conducted have examined recovery in the short term, the average time after disturbance being 1–5 years, and there are only a few studies covering a time frame of several decades. Even though some of the short-term studies show relatively rapid recovery of vegetation after damage (e.g., Emanuelsson, 1984; Tolvanen et al., 2001), the few existing long-term studies indicate that the recovery can be incomplete (Wielgolaski ,1998; Jorgenson et al., 2010) or temporary (Willard et al., 2007), or that the plant communities have not recovered to initial communities but have changed toward different stable-state communities (Forbes, 1996; Becker and Pollard, 2016).The long-term studies to date have been conducted in barren mountain regions, and our knowledge on the long-term natural regeneration of vegetation is limited, especially concerning the subarctic mountain birch (Betula pubescens ssp. czerepanowii) zone, which is typical for Fennoscandian mountain areas between barren mountain belt and coniferous forest located at lower altitudes. The upper limit of the mountain birch zone in the southeastern Norwegian mountains is at about 1300 m a.s.l., whereas in the northernmost Fennoscandia (at 71°N) it is almost at the sea level (Aas and Faarlund, 2001).

The aim of this study was to investigate the plant communities at the 64-year-old military sites in Malla Strict Nature Reserve, NW Finnish Lapland. We compared the vegetation of the observation posts in the barren mountain zone and the camps located in the mountain birch zone to the control sites on each zone. Our expectation was that the vegetation cover would be more disturbed in the barren mountain than in the mountain birch zone, because earlier studies indicated that mountain biotopes are more vulnerable to trampling than forested biotopes (Hoogesteger, 1984; Kangas et al., 2007). We also wanted to investigate if there were differences in regeneration in the spots with different disturbance intensity on the camps, where the bases of former huts had inevitably been subjected to more disturbance than their surroundings. These kinds of studies are important in order to understand the resilience and recovery of vegetation in a fragile subarctic environment, and also to follow up and predict changes under changing climate conditions, which may alter substantially environmental circumstances in the high latitudes.

Materials and Methods

Study Area

The Malla Strict Nature Reserve (69°04′N, 20°40′E) is located in the Kilpisjärvi area in western Lapland, representing the most northeastern part of the Scandes mountain range (Fig. 1). The reserve area (30.5 km²), is characterized by subarctic mountains ranging in the altitudes between 472 and 951 m a.s.l. The region is composed of Caledonian bedrock with limestone, and it is known for its rare vegetation. The lower parts of the mountains are covered by mountain birch with the tree line limit at ∼600 m a.s.l. (Kauhanen and Mattsson, 2005).The area is among the coldest regions in Finland with an annual mean temperature of -1.9 °C and growing season of approximately 109 days (Finnish Meteorological Institute, 1998–2002).

The reserve has been mainly out of human use since 1916, when the area was protected, except for one hiking route that traverses the reserve and is open to the public. Nowadays, the route attracts about 10,000 hikers yearly (Ohenoja, 2009). Before protection, the area was used by Sámi reindeer herders as a summer pasture. After establishment of the reserve, grazing was forbidden, and reindeer have been mainly out of the area except for some years during past decades, when reindeer have occasionally escaped into the reserve and grazed there during the summer months Jokinen and Heikkinen, 2005; Heikkinen et al., 2005). The main human activity in the reserve was, however, at the end of World War II, when the German army established military sites in Malla and the other nearby mountains before retreating to Norway. About 1000 men spent seven months in the reserve area in 1944–1945 between mid-September and the end of April. The occupied sites were documented on a map made by the Finnish army in May 1945 (Anonymous, 1945), and they comprised both observation posts in the barren mountain zone and camps where the soldiers lived, located lower, in the mountain birch zone. The observation posts, built of stones and metal, were occupied by 4–5 men at a time. The camps consisted of plywood huts, each of which housed 16–18 men (Urho Wiik, personal communication, 22 June 2009). The vegetation damage caused by the soldiers gives us an opportunity to study natural succession of vegetation in a timeframe of over 60 years. Apart from the collection of wood and metal from the camps immediately after the war and one operation to clean up metal waste in the 1980s, the sites have remained untouched without human interference. We expect that the possible influence of other environmental factors, such as variation in climate and occasional short-term grazing by reindeer have affected equally both the military sites and their controls.

FIGURE 1.

Location of the study area.

Data Analysis

We used nonparametric multi-response permutation procedures (MRPPs) to examine the differences in the plant coverage between military camps and controls within both zones, using Bray-Curtis matrix calculated from plant species percentage cover. The distance matrix was rank transformed to make the analysis more sensitive (McCune et al., 2000; McCune and Grace, 2002). The test-value T shows the difference between groups: the greater the difference is, the more negative the T value. The p-value shows the probability that the T-value was obtained by chance. The test value A expresses the chance-corrected within-group agreement; A = 1 - (observed Δ/ expected Δ), where Δ is the overall weighted mean of within-group means of the pairwise dissimilarities among sampling units.

A max = 1 when all items are identical within groups (Δ = 0)

A = 0 when heterogeneity within groups equals expectation by chance

A < 0 with more heterogeneity within groups than expected by chance

Indicator species analysis (IndVal; Dufrene and Legendre, 1997) was used to detect the indicatorvalue of a plant species or functional group in each study category. The value indicates how often a species occurs in the specific group, and how faithful it is to the group. The Monte Carlo method was then used to test the reliability of the indicator value. In this method, random groups are created from the original data and tested to ascertain whether these give indicator values as high as those in the real data. The procedure is repeated 5000 times, and the resulting p-value expresses the probability of obtaining an equal or greater indicator value from the randomized data (Dufrene and Legendre, 1997).

Nonmetric multidimensional scaling (NMS, McCune and Grace, 2002) was used to detect similarities in the species composition. The NMS analysis was conducted with seven functional groups of plants: (1) bushes and trees, (2) dwarf shrubs, (3) herbs, (4) graminoids (including grasses and grass-like plants such as sedges and Luzula species), (5) mosses, (6) liverworts and (7) lichens. Differences between environmental variables (thickness of humus layer, coverage of exposed soil, stones, litter, metal waste, and altitude and exposition of the site) and between the vegetation coverage of bottom and field layers were examined using the Mann-Whitney nonparametric test, because all the variables were not normally distributed. The environmental variables were also used as explanatory variables in the ordination analysis. Differences in the vegetation cover among the three damage intensity categories of the camps were studied using both MRPP and indicator species analyses. All the analyses were performed with PC-ORD (McCune and Grace, 2002) except for the Mann-Whitney pairwise tests, which were run using SPSS 16.0 for Windows (SPSS incorporated).The number of vegetation plots was equal (5 × 10 plots = 50) in all other analyses except in the analysis of damage intensity categories, where the number of plots was 22 under the huts, 11 in the transition zone, and 17 between the huts.

FIGURE 2.

Camp site in the mountain birch zone, showing a distinct boundary between the dwarf shrub vegetation of an undisturbed area and the camp site covered mainly by lichens of the Cetraria and Cladonia genera, mosses of the Polytrichum genus, and graminoids such as Deschampsia flexuosa.

Results

Vegetation Coverage at the Military Sites and Their Controls

In the mountain birch zone, the camps were clearly visible in the terrain as circles several meters in diameter (Fig. 2). A great deal of mineral soil was exposed at the bases of camps (difference to the controls p = 0.032, Table 1), and metal rubbish was found commonly (Table 1). The mean thickness of the humus layer was more than double at the control sites compared with the camps (Table 1), however, the difference was not statistically significant due to large variation among the control sites. The MRPP analysis showed significant differences in the species composition between the camp sites and their controls (T = -2.406, A = 0.149,p = 0.017). The coverage of dwarf shrubs was lower at the camps than at the controls (IV = 64.1, p = 0.009); especially V.myrtillus (IV = 69.8, p = 0.007) and E. hermafroditum (IV = 66.0, p = 0.040) showed lower coverage (Table 2).

TABLE 1

The mean values (± S.D.) of site variables measured at camp sites and their controls (Control CS) in the mountain birch zone, and at the observation posts and their controls (Control OP) in the barren mountain zone. All values except for humus (cm) are percentages of the maximum coverage (100%). The p-values of the Mann-Whitney (M-W) pairwise comparison between the disturbed sites and their controls are given with the significance limit p = 0.05. Significant differences are in bold.

TABLE 2

Results of indicator analysis of functional plant groups and some individual species and genera comparing the camp sites and their controls in the mountain birch zone. IV = indicator value. The p-value shows the probability of obtaining an equal or greater indicator value by chance. Significant differences are in bold.

The vegetation in the barren mountain zone was generally sparse and open (Fig. 3). The humus layer at the observation posts was significantly thinner as compared with the control sites (22.5 mm vs. 42.3 mm, p = 0.008, Table 1). No significant differences were detected in the vegetation coverage of observation posts and their controls in the MRPP analysis (T = 0.117, A = -0.007, p = 0.454); however, the indicator analysis showed that the coverage of E. hermafroditum was lower at the observation posts than at the controls, the difference being nearly significant (IV = 68.9, p = 0.090, Table 3). Furthermore, the data indicated increased coverage of graminoids Carex bigelowii, Juncus trifidus, and Deschampsia flexuosa at the observation posts (Table 4), and the species had high indicator values (Table 3), even though the differences were not statistically significant. Similarly, Salix herbacea showed higher and Cassiope hypnoides lower coverage at the disturbed sites compared with the control sites (Table 4).

Ordination Analysis

The two main axes explained a total of 79% of the variation in the data in the ordination analysis. The highest explanatory value (58%) was shown by axis one (vertical axis), which correlated positively with altitude and negatively with litter and woody material on the ground (Fig. 4). It divided the sites of the barren mountain zone from those in the birch zone, having a positive correlation with lichen coverage (Pearson r = 0.875) and a negative correlation with liverworts (r = -0.808) and bushes and trees (r = -0.732). The second (horizontal) axis (explanatory value 21%) distinguished disturbed areas from their controls, correlating positively with the coverage of stones and metal waste and negatively with the thickness of the humus layer (Fig. 4) and the coverage of dwarf shrubs (r = -0.695) and mosses (r = -0.648). In general, the disturbed and undisturbed sites were more clearly separated from each other in the mountain birch zone, whereas both the observation posts and their controls of the barren mountain zone resembled more each other in their vegetation communities.

FIGURE 3.

Two collapsed observation posts in the barren mountain zone. The undisturbed vegetation on the steep slope in the foreground consists of dwarf shrubs and lichens mainly of the genera Cladonia and Stereocaulon, whereas the disturbed area in front of and between the remains of the two observation posts consists mainly of graminoids.

Discussion

The results show that the subarctic mountain vegetation that was disturbed at the military sites during World War II had not recovered in 64 years. The MRPP analysis showed distinct differences in the species composition between the disturbed sites and their controls in the mountain birch zone, with reduction of dwarf shrubs Vaccinium myrtillus and Empetrum hermafroditum. Contrary to expectations and the results of some earlier studies (e.g., Hoogesteger, 1984; Kangas et al., 2007), differences in the vegetation as compared to the controls were greater on the camps in the mountain birch zone than at the observation posts above the tree line. This can be attributed to several factors—for example, differences in the intensity of trampling and the physical factors affecting the vegetation coverage of these two habitats. Vegetation in the barren mountains of the area is naturally sparse and open due to frost and harsh climate conditions, containing bare soil and unestablished plant communities even in undisturbed areas (Kauhanen and Mattsson, 2005). The trampling intensity has evidently been lower at the observation posts, which were occupied by 4–5 men at a time, compared to the camp sites, where 16–18 men occupied one hut continuously and several huts were grouped together. Trampling experiments in northern Fennoscandia show a relatively low threshold in the trampling tolerance of mountain heath vegetation: 150–200 tramples caused 30%–40% loss in vegetation (Gellatly et al., 1986; Törn et al., 2006), and even 25 trampling passes caused visible signs in vegetation (Tolvanen et al., 2001). Our study lacks information on the initial coverage of vegetation before the sites were established, but the differences in vegetation around the observation posts and camp sites compared to the present controls indicate that the damage had exceeded the threshold limit and altered the plant communities at the disturbed sites. Furthermore, the thinner humus layer at the military sites in both zones as compared to the control sites indicates long-term effects of disturbance on plant biomass accumulation and decomposition.

TABLE 3

Results of indicator analysis of functional plant groups and some individual species and genera comparing the observation posts and their controls in the barren mountain zone. IV = indicator value. The p-value shows the probability of obtaining an equal or greater indicator value by chance. Significant differences are in bold.

TABLE 4

The mean coverage of vascular plants and genera of lichens and liverworts per site at the observation posts and their controls (Control OP) in the barren mountain zone, and the camp sites and their controls (Control CS) in the mountain birch zone.

Continued.

FIGURE 4.

Nonmetric multidimensional scaling (NMS) ordination of the study sites,based on the environmental variables and functional groups of plants (bushes and trees, dwarf shrubs, herbs, graminoids, mosses, liverworts and lichens). Symbols: 1 = observation post, barren mountain, 2 = control, barren mountain, 3 = camp, mountain birch zone, 4 = control, mountain birch zone.

In general, the vegetation at the camp sites in the mountain birch zone differed significantly from the thick dwarf shrub coverage found at the control sites. Site characteristics such as slope steepness, exposition, and soil texture affect the vulnerability of vegetation to damage (e.g., Törn et al., 2006). Compacted soil, exposed mineral soil, and destruction of the humus layer decrease absorption of water and increase erosion (Marion and Cole, 1996) and also intensify frost and variation in microclimate (Mäkipää, 1999). Even though these physical and chemical factors influence the extent of damage and the recovery rate of vegetation, earlier studies indicate that morphological characteristics of plants may contribute even more to explaining the variation in response to trampling (Cole, 1995a, 1995b). At our study sites, E. hermafroditum and V. myrtillus were the vascular plants most affected in terms of coverage in both vegetation zones, and C. hypnoides showed reduced coverage at the disturbed sites in the barren mountain zone, all these being chamaephytes with their perenniating buds above the ground surface. This life form has been found to be more vulnerable to trampling than hemicryptophytes and geophytes (Cole, 1995b). E. hermafroditum is known to be very sensitive to mechanical disturbance, whereas V. myrtillus and V. vitis-idaea are intermediate in their trampling tolerance (Hoogesteger, 1984; Tybirk et al., 2000; Tolvanen et al., 2001). Vaccinium species are able to grow from belowground stems, which can assist in recovery provided that they have been saved from the damage (Hautala et al., 2001), which may explain their better tolerance to trampling compared with E. hermafroditum. However, in 60- to 70-year-old disturbance areas in Disko Bay, Greenland, Empetrum had reentered the disturbed areas after 60 years, whereas recovery was slower in the Vaccinium and Cassiope dominated dwarf shrub heaths (Tybirk et al., 2000). Our study did not show substantial differences between these species in recovery at the observation posts, but at the camp sites, E. hermafroditum and Juniperus communis had the highest coverage between the huts, indicating some recovery of shrubs and dwarf shrubs at the least disturbed patches of the sites. Unlike other dwarf shrubs, S. herbacea seemed to be favored by the open space created in the barren mountain zone.

TABLE 5

Results of indicator analysis of functional plant groups and some individual species and genera in the three damage intensity categories of the camps: under the huts (Under), in the transition zone between the huts and the vegetation cover between them (Transition), and on visibly less disturbed vegetation between the huts (Between). IV = indicator value. The p-value shows the probability of obtaining an equal or greater indicator value by chance. Significant differences are in bold.

FIGURE 5.

The percent coverage and number of species of graminoids in the three damage categories of the camp sites: under the huts, between the huts, and in the transition zone.

Studies monitoring long-term vegetation recovery in Arctic-alpine regions are few; all those that have been done indicate slow recovery or nonrecovery of original vegetation after disturbance (Forbes, 1996; Wielgolaski, 1998; Willard et al., 2007, Jorgenson et al., 2010; Becker and Pollard 2016; see also review by Tolvanen and Kangas, 2016). Instead, these studies show resistance of, or recovery by, different graminoids such as Deschampsia, Alopecurus, Poa, and Juncus species. In our study, C. bigelowii, J. trifidus, and D. flexuosa were the dominant graminoids at the observation posts, and the latter dominated also at the transition zones of the camp sites in the mountain birch zone. Caespitose, or matted, graminoids are the most resistant of the vascular plants to trampling (Cole, 1995b). Many graminoids (e.g., C bigelowii) have a high regeneration rate due to a high photosynthetic capacity and belowground rhizomes, which serve as energy stores and produce daughter tillers after damage, aiding their establishment at exposed sites (Chapin, 1980; Bryant et al., 1983; Tolvanen et al., 2001). They can also benefit from the possible short-time addition of nitrogen in the surroundings of campsites, unlike dwarf shrubs, which have limited capacity to utilize added nutrients (Chapin et al., 1986; Mäkipää, 1999). J. trifidus, which is a pioneer species in open areas of barren mountain, also has a high seed production capacity and high seed germination percentage, which contribute to its establishment in barren soils (Marchand and Roach, 1980).

The bottom layer of our study sites in the mountain birch zone comprised lichens in the genera Cetraria, Peltigera, and Stereocaulon, and mosses in the genera Polytrichum and Dicranum, showing similar species composition to that in the harvesting holes in Wielgolaski's study (1998). These cryptogams are resistant to trampling and it is assumed that they assist in the stabilization of the soil after disturbance (Hoogesteger, 1984; Wielgolaski, 1998). Lichens, as well as several mosses, are able to grow vegetatively from pieces of their thallus (Studlar, 1980;Törn et al., 2006). Of the mosses, especially Polytrichum species thrived under the huts in the mountain birch zone. Polytrichum commune has been found to invade exposed and disturbed areas, and its shoots have shown almost finite life expectancy, the species being also successful in sexual reproduction and having a proliferating underground system that can compensate for the damages to the aboveground shoots (Callaghan et al., 1978). The comparison of damage intensity categories at the camp sites demonstrated the above-mentioned recovery strategies and potentials of different plant growth forms: dwarf shrubs and mountain birch seedlings were growing between the huts at the least disturbed plots, whereas graminoids prevailed in the transition zone where they could take advantage from the open space. However, they had not been able invade the most disturbed plots under the huts, where only cryptogams prevailed.

Many studies looking at the short term suggest that vegetation recovers relatively rapidly after trampling in mountain heaths (Emanuelsson, 1984; Tolvanen et al., 2001). Nevertheless, such recovery can be temporary. Willard et al. (2007) studied the recovery after trampling damage over a period of 42 years in the Rocky Mountains, U.S.A., and noticed that the regrowth that took place during the first 10 years stopped and even declined during the later decades of the study due to unfavorable climate conditions. Our results show very little ecological recovery (cf. Hylgaard, 1980) of the sites to their original state in the 64 years. Rather, the results suggest stagnation or slight functional recovery; that is, the new vegetation at the sites has started to develop as a functioning plant community with a different species composition than the original (cf. Walker et al., 1987). Similarly, Becker and Pollard (2016) found in a 60-year study on the recovery of vegetation in Canadian High Arctic that the disturbed landscape had not recovered to initial conditions but changed toward a different stable-state community. In a longer time frame, Forbes (1996) found indications of permanent vegetation shift in 800-year-old Inuit settlements on the Canadian Arctic islands; the study shows that the grass vegetation that took over the old settlement areas had persisted for several hundred years. The weak ecological recovery of the sites in our study may be related to the fact that the stress-tolerant, slow-growing dwarf shrubs and cryptogams, which are typical for the low-productive communities in the study region, have been found to respond weakly to the changes in resource levels or changing environmental conditions (Virtanen et al., 2010). At the moment, there are contradictory predictions on the effects of changing climate on the dwarf shrub communities in subarctic ecosystems (see, e.g., Bokhorst et al., 2009; Virtanen et al., 2010). Nevertheless, many recent studies indicate that the development of future climatic and edaphic conditions may shift the vegetation cover of Arctic and subarctic heaths toward increased coverage of deciduous shrubs (e.g, Walker et al., 2006; Elmendorf et al., 2012; Myers-Smith et al., 2011), which may increase gross ecosystem production (see Tolvanen and Kangas, 2016). Different scenarios stress the importance of further monitoring of vegetation changes in old disturbance areas.