In addition to invasibility of the habitats and species traits (i.e.
invasiveness), factors such as propagule pressure and residence time should be considered when predicting and understanding plant invasions (e.g. Milbau et al. 2009). Propagule pressure affects both the invasibility of the habitat and the ability of the species to spread into new areas (e.g.
Colautti et al. 2006, Simberloff 2009) and it is a key element to understanding the success and failure of alien plant invasions (Lockwood et al. 2005).
Residence time represent another dimension of propagule pressure.
It is an important determinant of present geographical range sizes of alien plants (Pyšek and Jarošik 2005, Rejmánek et al. 2005, Williamson et al.
2009). The longer species is present in the area the more propagules are spread, and the higher their chance is to establish and invade over larger range (Pyšek and Jarošik 2005). Generally, it takes at least 150 years for an alien plant species to reach their maximum (Williamson et al. 2009), thus many neophytes (see Table 1 for definitions) have probably not yet occupied all suitable habitats. In addition, residence time is associated with the alien species response to environmental conditions. For instance, in Central Europe neophytes prefer wet, fertile habitats, while archaeophytes are more common in sunny, dry to mesic habitats (Chytrý et al. 2005, Pyšek et al. 2005, Simonová and Lososová 2008).
1.5 IMPACTS OF PLANT INVASIONS
Invasive alien species cause severe ecological hazards, and are widely considered as one of the leading direct causes of biodiversity loss (e.g.
Didham et al. 2005, Vilà et al. 2011) (see however Davis et al. 2011).
Invasive alien species may impact on native species by competing for resources, facilitating the spread of pathogens, and through hybridisation and impacts on higher trophic levels (e.g. Levine et al. 2003, Hulme 2007).
In addition, invasive alien plants can transform the structure and function of the ecosystems by, for example, changing nutrient cycling and disturbance regime (e.g. Mack and D´Antonio 1998, Levine et al. 2003, Rejmánek et al.
2005). However, the impact of invasive plants on biodiversity is less severe than impact of alien pathogens, herbivores and predators, and not a single native plant species has been documented to being driven to extinction by competition from alien plants alone (e.g. Rejmánek et al. 2005). The ecological impacts of invasive alien plant species are largely species-specific and the severity of the impacts depends on the identity of the invading species (Hejda et al. 2009, Vilà et al. 2011).
In addition to ecological impacts, invasive alien plants cause economic, social and health detriments. Many alien plants have become weeds, which cause crop losses and control costs (e.g. Pimentel et al.
2000, Vilà et al. 2010). In addition, alien plant species can reduce availability of pollinators to native species, and decrease the recreational and aesthetic values of the landscape (e.g. Pyšek et al. 2009b, Vilà et al.
2010). For instance, Rosa rugosa Thunb. ex Murray grows in abundant, thorny thickets in the Nordic beaches, reducing recreational use of the beaches (e.g. Weidema 2000, Vilá et al. 2010, MMM 2012). Several alien plants can also cause allergies or other severe health problems (e.g. Vilà et al. 2010), such as burn and blisters induced by Heracleum mantegazzianum Sommier & Levier (e.g. Weidema 2000). Many invaders are known to cause multiple impacts, but the current understanding is often restricted to relatively few dominant species (e.g. Pyšek et al. 2009a, Vilà et al. 2010, 2011). Although invasive alien plant species have severe impacts, the impacts are heterogeneous and vary even within particular impact type (e.g. Vilá et al. 2011).
1.6 RESEARCH NEEDS
Although agricultural habitats are among the most invaded habitats (e.g.
Lonsdale 1999, Chytrý et al. 2005, Pyšek et al. 2010a), the studies of the temporal, spatial and within-habitat variation in the invasion level of alien plants in agricultural habitats are lacking. In addition the factors contributing to the invasion level have been studied insufficiently (but see Thompson et al. 2001, Lake and Leishman 2004, Leishman and Thompson 2005, Thuiller et al. 2006), especially in the boreal region.
Species vary in their response to environmental conditions (e.g.
Richardson and Pyšek 2006) and in their effect on native species diversity (e.g. Stohlgren et al. 1999, Vilà et al. 2011). However, species-specific studies on the effect of environmental conditions and impact on native species diversity are inadequate, especially on less dominant alien plant
species (e.g. Pyšek et al. 2009a, Vilà et al. 2010, 2011). In addition, the studies of alien plant invasion on the terrestrial habitats in Finland are scarce (but see Nummi et al. 2001, Valtonen et al. 2006, Hyvönen and Jalli 2011, Ranta and Viljanen 2011, Ramula and Pihlaja 2012), and the studies on the effects on environmental conditions on alien plant species are lacking.
One of the factors affecting plant invasions is the native species diversity, and several processes have been identified to generate either positive or negative native-alien richness relationships (e.g. Fridley et al.
2007). These processes are related to spatial scale, which emphasize the importance to study native-alien relationships at multiple spatial scales.
However, only few studies have estimated the effect of different processes at multiple spatial scales taking into account also species diversity components operating at multiple scales (e.g. Davies et al. 2005, Stohlgren et al. 2006, Capers et al. 2007, Belote et al. 2008, Veech and Crist 2010).
In addition to features of the habitat and ecosystem, the success of plant invasions depends on the characteristics of the invading plant species.
Although characteristics of a successful plant invader are known to be habitat-specific and affected by environmental features, empirical studies of the characteristics of successful invader have largely ignored the environmental conditions (but see Thompson et al. 2001, Lake and Leishman 2004, Leishman and Thompson 2005, Thuiller et al. 2006).
2 AIMS OF THE STUDY
In this thesis, I will assess the temporal, spatial and within-habitat variation in the invasion level in Finnish agricultural habitats, and examine how environmental conditions and species characteristics contribute to the level of invasion and to the occurrence of alien plant species. In addition, I assess the effect of plant invasion on native plant diversity. I aimed at answering to the following questions:
1. Which are the invasion levels of alien plant species in five different agricultural habitats (I) and four different geographical (II) regions in Finland?
2. How environmental factors affect on the invasion level and the occurrence of alien plant species (I), and what is the impact of alien plant species on the native plant species richness and diversity (II)?
3. What is the relationship between native and alien plant species diversity in agricultural habitats at multiple spatial scales, and which processes contribute to these diversity-invasibility relationships? (III) 4. Which are the characteristics of a successful plant invader and how
these characteristics are related to habitat characteristics? (IV)
3 MATERIALS AND METHODS
3.1 STUDY AREA
In this study, I used a comprehensive data from long-term national monitoring study on the effects of the Finnish agri-environment support scheme (MYTVAS) (Kuussaari et al. 2008). The data were collected from agricultural landscapes situated in four geographical regions in Finland:
south, south-western, western and eastern Finland (Fig. 1). The southern and south-western regions are situated in hemiboreal and southern boreal zones, and western and eastern regions in the middle boreal zone (Ahti et al. 1968). Southern and south-western regions have the most advantageous climatic and edaphic conditions for crop production in Finland (percentage of cultivated field 50.9% and 58.6%, respectively), whereas western and eastern Finland are characterized by cooler climate, a shorter growing season and lower proportion of arable land (42.4% and 27.4%, respectively) than in southern and south-western Finland (e.g.
Kivinen et al. 2006, Kuussaari et al. 2008, Tarmi et al. 2009). In the western Finland, the typical agricultural landscape is dominated by intensively cultivated arable land, surrounded by coniferous forests and mires, whereas eastern Finland is characterized by extensive forest cover (59.3%), and agriculture is based mainly on dairy farming (e.g. Kivinen et al. 2006, Kuussaari et al. 2008, Tarmi et al. 2009).
In Finland, approximately 1 300 vascular plant species are regarded as established, and roughly 550 terrestrial vascular plants are alien to Finland (e.g. Weidema 2000, MMM 2012). Hyvönen and Jalli (2011) assessed the number of agricultural weed species in Finland, and detected 815 alien weed species, most of which (501, 61%) were casual neophytes.
Thus, most of the alien weeds are found in Finland occasionally, and the circumstances for establishment of permanent population have not been favourable (e.g. climate conditions; Hyvönen 2011, Hyvönen et al. 2011), although the propagule pressure is high (Hyvönen and Jalli 2011). In the future, established neophytes are expected to extend their distribution and increase their occupation in agricultural habitats, and climate change may affect the establishment of the casual neophytes (Hyvönen and Jalli 2011).
According to the Finland´s National Strategy on Invasive Alien Species (MMM 2012), 24 alien plant species (~4% of all alien plant species) are regarded as invasive in Finland. Invasive alien plant species are known to have negative effect on native species diversity, species composition and ecosystems structure (e.g. Valtonen et al. 2006, MMM 2012). Semi-natural agricultural habitats (e.g. semi-natural grasslands, wooded pastures) and ruderal habitats are a primary habitat for almost third of the threatened vascular plants in Finland (Rassi et al. 2010).
Figure 1 A) The four geographical regions (A= western Finland, B= eastern Finland, C= south-western Finland and D= southern Finland) and the location of the 1 km2 sites. The green colour indicates the occurrence of agricultural land. B) The aerial photograph provides an example of the sampling design (Nurmijärvi, southern Finland). The 1 km2 sites were divided into four squares, and the vascular plants were recorded from six 50 m × 1 m transects lines (red lines) in each 0.25 km2 square (Kuussaari et al. 2008).
3.2 STUDY DESIGN
The MYTVAS vascular plant data (Kuussaari et al. 2008) comprised a total of 52 sites (1 km2), and each site was divided into four squares (Fig. 1).
Among these squares (0.25 km2) two most divergent squares were selected in order to represent the landscape heterogeneity within the 1 km2 site (Kuussaari et al. 2004). Vascular plants were recorded from six 50 m × 1 m transects lines in each 0.25 km2 square, and from three quadrats (1 m2) along each transects. Thus, the hierarchical data set comprised five spatial scales: 1 m2 quadrats, 50 m2 transects, 0.25 km2 squares, 1 km2 sites and regions.
The transects were situated in five distinct habitat type: (1) field margin (margin between two agricultural fields), (2) forest margin (margin between a forest and an agricultural field), (3) road margin (margin between a road and an agricultural field or road verge within agricultural habitat), (4) grassland (including uncultivated meadows, abandoned fields and cultivated or natural pastures) and (5) other habitat types (including margin between agricultural field and a waterway, cart-tracks and other habitats low in number). In addition to habitat type, environmental variables included several variables measured at different spatial scales. At 1 m2 quadrats, data included local environmental variables (total vegetation coverage, proportion of bare ground and rockiness). Environmental variables measured at 50 m2 transects included spatial variables (longitude
Figure 1 A) The four geographical regions (A= western Finland, B= eastern Finland, C= south-western Finland and D= southern Finland) and the location of the 1 km2sites. The green colour indicates the occurrence of agricultural land. B) The aerial photograph provides an example of the sampling design (Nurmijärvi, southern Finland). The 1 km2 sites were divided into four squares, and the vascular plants were recorded from six 50 m × 1 m transects lines (red lines) in each 0.25 km2square (Kuussaari et al. 2008).
3.2 STUDY DESIGN
The MYTVAS vascular plant data (Kuussaari et al. 2008) comprised a total of 52 sites (1 km2), and each site was divided into four squares (Fig. 1).
Among these squares (0.25 km2) two most divergent squares were selected in order to represent the landscape heterogeneity within the 1 km2 site (Kuussaari et al. 2004). Vascular plants were recorded from six 50 m × 1 m transects lines in each 0.25 km2 square, and from three quadrats (1 m2) along each transects. Thus, the hierarchical data set comprised five spatial scales: 1 m2 quadrats, 50 m2 transects, 0.25 km2 squares, 1 km2 sites and regions.
The transects were situated in five distinct habitat type: (1) field margin (margin between two agricultural fields), (2) forest margin (margin between a forest and an agricultural field), (3) road margin (margin between a road and an agricultural field or road verge within agricultural habitat), (4) grassland (including uncultivated meadows, abandoned fields and cultivated or natural pastures) and (5) other habitat types (including margin between agricultural field and a waterway, cart-tracks and other habitats low in number). In addition to habitat type, environmental variables included several variables measured at different spatial scales. At 1 m2 quadrats, data included local environmental variables (total vegetation coverage, proportion of bare ground and rockiness). Environmental variables measured at 50 m2transects included spatial variables (longitude
and latitude), habitat quality (shadiness, moisture, average vegetation height) and disturbance regime (proportion of bare ground, mowing).
Variables of geographical location and landscape diversity were measured at 0.25 km2 squares, and climatic variables (including total summer temperature sum and precipitation, total number of frost days and starting date of the growing season) were calculated at 1 km2 sites. In addition, I collected species trait data of 17 species characteristics, which represented the ecological and morphological traits, and traits related to invasion history, dispersal and species requirements for environmental conditions, from several databases, such as BiolFlor (Klotz et al. 2002) and LEDA traitbase (Kleyer et al. 2008).
3.3 ANALYSES OF THE DATA
Invasibility can be characterized by the survival rate of invading species, but it is difficult to quantify, because the influence of species invasiveness and propagule pressure on invasion level must be accounted for (e.g.
Chytrý et al. 2008a, Pyšek et al. 2010a, Catford et al. 2012). As a precondition for quantifying invasibility is possibility to compare invasion level across habitats and ecosystems (e.g. Catford et al. 2012). Invasion level of alien species can be used to assess the extent or severity of invasions, and to reveal spatio-temporal trends (Chytrý et al. 2008a, Catford et al. 2012). In addition, invasion level can act as an early warning sign for ecological degradation and as an estimate for the consequences of invasion (Catford et al. 2012). I assessed the invasion level for five different habitat types (I) and four different geographical regions (II) using relative alien species richness (I and II) and alien species diversity (measured as Shannon-Wiener diversity index) (II). The invasion level can change in time depending on the identity of alien species present, propagule pressure and biotic and abiotic conditions (e.g. Catford et al.
2012). I assessed changes in the levels of invasions over a decade (II). In addition, I calculated the frequencies of occurrence of neophytes for each study year.
The invasion level is affected by several environmental factors, such as climate, geographical location, the structure of the plant community and habitat properties (e.g. Chytrý et al. 2008a, Pyšek et al. 2010a). I examined how these environmental factors affect on species richness and occurrence (I, II). Native species, archaeophytes and neophytes were examined separately (I), because their response to environmental conditions varies (e.g. Chytrý et al. 2005, Pyšek et al. 2005, Simonová and Lososová 2008, Pyšek et al. 2010a). Because species vary in their response to disturbance and other environmental variables (e.g. Hobbs and Huenneke 1992, Smith and Knapp 1999), I also determined the effect of environmental conditions on the most common neophyte species (Achillea ptarmica L., Epilobium adenocaulon Hausskn., Galium album Mill.
and Trifolium hybridum L.) (II). These analyses were conducted with a combination of principal components (PCA) and generalized linear mixed models (GLMM) analyses. PCA summarizes the environmental data, and reduce the multicollinearity among the environmental variables (e.g. Kent and Coker 1992), whereas GLMM allows the use of non-normal distributions and the incorporation of random terms that control for spatial non-independence (e.g. Bolker et al. 2009).
I examined the relationship between native and alien species richness using generalized linear models (III). In order to understand the drivers of these relationships we included alpha, beta and gamma diversity in the analyses. The relationship between native and alien species is strongly associated with spatial scale, and driven by species interactions (e.g. competition) and environmental conditions (e.g. Shea and Chesson 2002, Pauchard and Shea 2006, Stohlgren et al. 2006, Pyšek et al.
2010a). Thus, I analysed how the environmental variables describing geographical location, productivity, disturbance regime and landscape structure affected diversity components of alien and native plant species at three spatial scales (1 m2 quadrats, 50 m2 transects and 0.25 km2 squares).
In addition to environmental conditions, the success of an alien plant species depends on the species traits of the invading plant species. In the search for the characteristics of successful invader, I studied the differences between neophytes and native species using Fisher´s exact test with sequential Bonferroni correction (IV), which is more accurate than asymptotic tests of independence for small, sparse data and small expected values (e.g. Mehta and Patel 1999). The sequential Bonferroni correction avoids the probability of type I errors, which may be inflated when performing multiple tests (e.g. Holm 1979). Since species traits are habitat-dependent (e.g. Thompson et al. 1995, Lloret et al. 2005), I assessed the interaction between environmental factors (habitat preferences, climate, geographical location) and characteristics of neophytes by using RLQ analysis combined with Hartigan´s K means clustering method (see e.g. Thuiller et al. 2006). RLQ analysis is a multivariate method (IV), which enables study of relationship between species traits and environmental conditions, and can reveal processes that remain hidden when analyzing environmental factors and species characteristics separately (e.g. Dolédec et al. 1996). The Hartigan´s K means clustering was used to define functional groups of neophytes sharing similar traits and similar responses to environmental conditions. In addition, with Moran´s I randomization test (e.g. Cliff and Ord 1973) I tested whether the clusters were phylogenetically independent.
I used the invasion levels, occurrence of the neophytes and comparisons between invaded and uninvaded sites in order to estimate the effect of neophyte invasion to native species diversity (I, II). I tested the differences in native species richness and native diversity (measured as
most common neophytes in 50 m2 transects using a t-test or a Kruskal-Wallis test if the equality of the variances was not attained (II).
4 RESULTS AND DISCUSSION
4.1 SPATIAL, TEMPORAL AND HABITAT VARIATION IN INVASION LEVEL (I, II)
I found that invasion level of alien plants varied between different semi-natural agricultural habitats (I), geographical regions, and the study years (II). The results were sensitive to the method of measuring the invasion level (either by relative alien species richness or alien species diversity).
Relative alien species richness was highest in frequently disturbed and more intensively managed habitats, such as field margins and road margins in agricultural landscape, whereas infrequently disturbed and managed grasslands were more seldom invaded by the alien plants. This result is consistent with previous studies (e.g. Chytrý et al. 2005, 2008b, Pyšek et al. 2010a) indicating that agricultural and ruderal habitats with human-induced disturbances, high fertility and propagule pressure exhibit highest levels of invasion.
The invasion level was strongly dependent on geographical location (I, II). For instance, relative alien species richness was higher in southern and south-western Finland than in eastern and western Finland. Thus, invasion level decreased northward with decreasing temperature and increased towards east with increasing continentality. This may be partially explained by the more favourable climate, migration history and routes, and land-use history and intensity (e.g. Luoto 2000, Kivinen et al. 2006).
For instance, plant diversity of field margins is lowest in the most intensive cereal production areas of the south-western and southern Finland and highest in areas of mixed farming in the eastern Finland (Tarmi et al. 2002, 2009). Similar geographical trends related to latitude have been detected globally, and alien species richness often decreases towards poles (e.g.
Lonsdale 1999). Consistent with previous studies, I found that the invasion level tended to be lower in northern boreal semi-natural habitats than in agricultural habitats of central and southern Europe (e.g. Vilà et al. 2007, Chytrý et al. 2009) (I). However, in the most disturbed Finnish semi-natural agricultural habitats, invasion level of alien plants may reach the same level as in ruderal habitats in central and southern Europe (e.g. Chytrý et al. 2008b).
In addition to spatial and habitat variation, the invasion level varied between to study years (II). For instance, alien species diversity was lower in 2005 than in other study years. The temporal variation in alien species diversity may be explained by variation in climatic conditions (e.g.
precipitation was higher in 2005 than average, see Kuussaari et al. 2008), disturbance regime and fluctuation in resource availability (e.g. nutrients, water and light) (e.g. Davis et al. 2000, Richardson and Pyšek 2006). To overcome problems related to largely stochastic variation in environmental
conditions, a longer monitoring period would have been needed to detect the temporal changes in the invasion level. In addition, temporal variation was depended on the measure used for invasion level. I discovered no clear temporal variation in relative alien species richness, but in alien species diversity (as Shannon-Wiener diversity index) lower values were detected in 2005 than in the other study years. High alien species richness
conditions, a longer monitoring period would have been needed to detect the temporal changes in the invasion level. In addition, temporal variation was depended on the measure used for invasion level. I discovered no clear temporal variation in relative alien species richness, but in alien species diversity (as Shannon-Wiener diversity index) lower values were detected in 2005 than in the other study years. High alien species richness