Species occurrence data was obtained from FLORKART (Federal Agency for Nature Conservation / Network Phytodiversity Germany; http://www.floraweb.de) for the period 1950–2013. FLORKART includes over 14 million records on species occurrences collected by thousands of volunteers. Species were analyzed at a spatial resolution of grid cells with 10' longitude × 6' latitude (~ on average 130 km² ranging from 117 to 140 km²). A presence/absence matrix was generated for a random sample of 1000 grid cells that contained at least 45 (out of 50) species that can be reasonably assumed to occur in every grid cell and serve as proxy for mapping quality (Kühn et al. 2006). This approach of grid cell selection ensured that chosen grid cells were properly surveyed. Additionally, some grid cells were smaller because they were located at the borders or along the coast. Thus, we excluded cells smaller than 117 km² (which is the size of the smallest grid cell that is not truncated by borders or coastlines). Individual matrices were generated for five groups of plants: native (976 species), archaeophytes (168 species) and neophytes (156 species), with 1,300 plant species in total; neophytes were further divided into (i) species featured in the German-Austrian Black List Information System of invasive species (GABLIS; Essl et al. 2011), with 26 species, and (ii) species not included in GABLIS, with 130 plant species. Following GABLIS (Essl et al. 2011), plants were classified into action black list (invasive with limited distribution) and management black list (invasive and widely distributed species). In our paper, we will refer to the species from GABLIS black list (action and management list) as invasive neophytes and to the ones that are not included in GABLIS as non-invasive neophytes.

Trait data for all plant species were obtained from the Database on Biological and Ecological Traits of the Flora of Germany, BiolFlor (Klotz et al. 2002; Kühn et al. 2004; http://www2.ufz.de/biolflor/index.jsp), and LEDA (Kleyer et al. 2008; https://uol.de/en/landeco/research/leda/data-files). These traits represent morphology, phenology and habitat preferences of all three groups of plant species: SLA, seed mass, height, storage organs, pollination vector, flowering period, urbanity and hemerobic level (Table 1).

Climate data (temperature, precipitation; Table 2) were obtained from the ALARM project (Fronzek et al. 2012) for the period 1961–1990, land cover (Suppl. material 1: Table S2) data from the CORINE database (Bundesamt für Kartographie und Geodäsie, 2012), and geological data (Table 2) from a map of the German Federal Institute for Geosciences and Natural Resources (Bundesanstalt für Geowissenschaften und Rohstoffe, 1993), all scaled to the same resolution as the floristic maps.

We analyzed the relationship between traits and environment across native and alien plant species. For each group (natives, archaeophytes, neophytes, non-invasive and invasive neophytes) matrices of species presence/absence × grid cell were created (S). Correspondingly, environmental matrices (environment × grid cell, E) and trait matrices (traits × species, T) for every status group were compiled. To directly associate matrices S with E and T, we used a fourth corner approach as implemented in the function traitglm ()of mvabund in R (Warton et al. 2015). Fourth corner analysis combines S (first–upper-left–corner), E (second–lower-left–corner) and T (third–upper-right–corner). The fourth (missing–lower right) corner is generated as a matrix that describes the trait-environmental relationships. We checked for collinearity among environmental variables and excluded variables with r > |0.7| (Dormann et al. 2013). The function manyglm presents a multivariate extension of GLM (generalized linear model) and calculates the coefficient estimates of GLMs fitted to all (explanatory) variables simultaneously (Wang et al. 2012). Coefficients describe how environmental predictors can be predicted by changes in traits. Further, we used the function anova.traitglm () based on bootstrapping with 99 permutations, to test for the statistical significance of trait–environment relationships in predicting presence of only non-native species (for computational reasons, see below) on all sites (Suppl. material 1: Table S1a–d). Since the response matrix S was binary multivariate data, we used binomial distribution.

The data analysis was performed using R, version 3.6.1 (R Core Team 2017). The analysis of a larger matrix (e.g. native species) took 19 days on a Dell PowerEdge R930 Server with 4 * CPU E7-8867 v4 2.4 GHz (72 Cores) and 6 TB RAM with Windows 2016.

Native species in Germany showed high heterogeneity in their functional traits and habitat conditions; thus the relationships between traits and environment were weak (ranging from -0.0003 to 0.01; Suppl. material 1: Table S2a).

The frequency of archaeophytes well adapted to urban environmental conditions (urbanity; Fig. 1a; Suppl. material 1: Table S2b) increased with mean temperatures (of both warmest and coldest month), broader precipitation range, across natural and urban areas, and with the number of geological patches. Conversely, their frequency decreased with an increase in annual precipitation, the proportion of calcareous subsoil and total number of Corine Land Cover (CLC) patches per grid cell.

With higher temperatures of the warmest month, species with high seed mass, wind- or self-pollination, high level of naturalness and those beginning to flower early will increase, while those with a long flowering period will decrease. Increasing amounts of precipitation disadvantaged small species that prefer artificial habitats but promoted species with high SLA, seed mass, presence of storage organs and multiple storage, self-pollination, as well as early beginning and late end of flowering.

Mean annual precipitation and number of CLC patches showed a strong positive relationship with multiple storage organs, yet mean temperature of the coldest month negatively affected this trait (Fig. 1b; Suppl. material 1: Tables S2c). Both wind- and self-pollination were negatively influenced by mean annual precipitation, and wind pollination was positively related to temperature (of the coldest and warmest month), sandy substrates and number of geological patches. Increase in the temperature of the warmest month promoted urbanophilic species, while the temperature of the coldest month positively affected the duration and end of the flowering period. Mean annual precipitation showed a negative relationship with plant height, but positive effects on SLA and plants with multiple storage organs.

Increasing winter temperature positively affected wind- and self-pollination and flowering duration, whereas tall urbanophilic species were negatively affected (Fig. 1c; Suppl. material 1: Tables S2d). Conversely, high summer temperatures were positively correlated with the frequency of tall urbanophilic non-invasive neophytes, and negatively with long flowering duration or larger SLA and seed size. An increase in the number of CLC patches favored insect-pollinated, urbanophilic plant species with higher SLA, while negatively affecting the abundance of long-flowering, self-pollinated species.

The temperature of the warmest month was positively related to SLA, multiple storage organs, self-pollination and negatively to duration of flowering (Fig. 1d; Suppl. material 1: Tables S2e). In contrast, the temperature of the coldest month was negatively related to SLA and positively to hemeroby. Annual precipitation negatively affected the beginning of flowering, while the precipitation range was positively associated with SLA and self-pollination. The number of CLC patches had a positive relationship with multiple storage organs and a negative one with hemeroby.

Differences among invasive neophytes (black list) were positively associated with land cover and mostly negatively with geological predictors. Neophytes with a limited distribution in Germany (action list) had positive relationships with all three types of land cover and with number of CLC patches and negative associations with calcareous, sandy substrates and number of geological patches.

Archaeophytes and neophytes showed several contrasting trait–environment relationships (Fig. 1a, b). Specifically, the frequency of self-pollination in archaeophytes increased with the temperature of the warmest month, mean annual temperature and proportion of loess substrates, while under these conditions the frequency of neophytes diminished. Similarly, in archaeophytes we observed a positive relationship between urbanity and temperature of the coldest month, the proportion of natural areas and number of geological patches, and a negative relationship with annual precipitation and number of land cover patches. Neophytes showed opposing trends.

Further, we observed differences between non-invasive neophytes and invasive neophytes (Fig. 1c, d). While the frequency of invasive neophytes with higher SLA increased with temperature of the warmest month and precipitation range, non-invasive neophytes displayed reversed trends. Similarly, urbanophilic invasive neophytes were promoted by increasing temperature of the coldest month, and insect-pollinated invasives by the number of geological patches and temperature of the warmest month, with contrasting tendency in non-invasive neophytes. Finally, insect-pollinated invasive neophytes benefited from increasing annual precipitation and a high number of land-cover patches, although these variables showed to be disadvantageous for non-invasive neophytes.

As to the best of our knowledge, no statistical test allows the formal comparison of results across different fourth-corner analyses; we have to interpret differences among the trait-environment responses of different groups qualitatively. Trait-environment relationships were similar (positive or negative, respectively) for archaeophytes and neophytes in 13 cases but differed in seven cases. Primarily, urbanity expressed contrasting relationships, suggesting human-induced propagule pressure as an important driver. Neophytes tend to be more urbanophilic, thus the increase in temperature was positively related to this trait (urban heat island effect; Ricotta et al. 2009). Urban areas facilitate neophytes (Kühn et al. 2004; Kühn and Klotz 2006), and alien species are often associated with cities (Chytrý et al. 2008b; Knapp et al. 2009; Aronson et al. 2014). Some studies showed that neophytes are becoming a dominant group in urban areas (Chocholoušková and Pyšek 2003; Pyšek et al. 2004), while the association of archaeophytes with this type of environment decreased in recent decades, and they are more common in arable landscapes (Botham et al. 2009). Hence, the increase in the proportion of arable and natural land cover affected urbanophytic neophytes negatively, but the increase in the proportion of urban area increased their abundance (and resulted in a reversed trend in archaeophytes). Neophytes are cultivated in gardens and public parks (Reichard and White 2001; Pergl et al. 2016), and their spread is further facilitated by extensive transportation systems (Seebens et al. 2015). Consequently, cities often present harbors for the spread (von der Lippe and Kowarik 2008) and establishment of newly introduced species (Kühn et al. 2017).

The majority of neophytes (especially invasive) are pollinated either by insects or wind, whereas archaeophytes are often self-pollinated (Pyšek et al. 2011). Many agricultural weeds are self-pollinated archaeophytes, possibly due to a lack of suitable pollinators or because of abiotic stress. Further, in archaeophytes, self-pollination is more common with increases in the proportion of loess. This can be due to loess being very fertile and suitable for agriculture, so self-pollination can be an alternative (Kühn et al. 2006), especially with the increasing scarcity of insects in regions of intensive agriculture (Hallmann et al. 2017).

Flowering phenology is important for the successful spread of invasive species (Knapp and Kühn 2012). Plant species have evolved in tune to local climatic regimes in their native range or colonized such regions naturally. With increasing temperatures (summer and winter), invasive neophytes finish their flowering period later in the year (with overall shorter duration). However, higher summer temperatures had a negative effect on the duration and higher winter temperature caused invasive neophytes to start flowering earlier. Many invasive species in Germany originate from warmer climates and as a result, an increase in winter temperature can act as a switch to earlier flowering. Earlier flowering of invasive species compared to non-invasive may ensure their reproductive success, and higher summer temperatures prolong the flowering season to late summer (Knapp and Kühn 2012). Low precipitation often impedes flowering, and the species that flower earlier can avoid summer droughts (Godoy et al. 2009). The increase in precipitation range (usually resulting from wet winters and dry summers) decreases the duration of flowering and plants were flowering later in the year. Depending on the origin of invasive neophytes, we can expect different responses to current or future climatic conditions. Provided that climate in the introduced area is the same as in the native area, flowering phenology can stay the same. However, if introduced species are subjected to a different climate, the flowering depends on the capability of invasive species to adapt or respond plastically to new conditions.

Alien plants that have often been introduced for their aesthetic features as ornamental plants can attract pollinators (colorful and fragrant flowers) and divert them from native plants (Bjerknes et al. 2007; Muñoz and Cavieres 2008). The majority of tropical and temperate plants are insect-pollinated (Ollerton et al. 2011), invasive neophytes, though, are primarily insect or self-pollinated. Additionally, many invasive species are annual plants and when suitable pollinators are not available they are able to self-pollinate which can be beneficial for the successful invasion of new areas (van Kleunen et al. 2007).

Climatic factors did not have a different effect on the occurrence of invasive species from the management or action black list. Species on the action list are more likely to be found in all three types of land cover than those from the management list. We can, therefore, expect that species which are invasive but still of limited distribution, will spread, especially as habitats become more fragmented (occurrence of action list species shows an increase with CLC number of patches).

Geological bedrocks did not have a major effect on most of the traits in different groups, despite explaining roughly a quarter of plant distribution variability in Germany (Pompe et al. 2008). Archaeophytes often occur on loess, which is highly productive and usually used for intense agriculture. However, in calcareous substrates archaeophytes tend to flower later while invasive neophytes flower earlier and are taller. Species-rich calcareous grasslands used to be common in Germany and are now frequently afforested, suffer from shrub encroachment or are surrounded by agricultural fields (Fischer et al. 1996). Sandy substrates can warm up earlier during winter and spring and can be suitable for neophytes introduced from warmer regions. Additionally, due to its low water-retention property, sandy substrates are frequently colonized by species adapted (i.e. having suitable traits) to drought.

Different land-cover types as well as the number of land-cover patches and geological patches had an effect on most of the traits of invasive neophytes, and very little (or no effect) on archaeophytes. Furthermore, landscape transformation and heterogeneity have an effect on invasive species in different stages of invasion and fragmentation of the landscape may facilitate the spread of invasive species (With 2002). Habitat heterogeneity intensifies invasion and increases dispersal (O’Reilly-Nugent et al. 2016; Dukes and Mooney 1999), and we have recorded a positive relationship with flowering phenology, SLA, height and seed mass of invasive neophytes. However, invasive neophytes with multiple pollination vectors (i.e. having different pollination types) benefited the most whereas wind-pollinated species colonized the least heterogeneous landscapes. These wind-pollinated invasive species are often dependent on specific habitats, for example, Fraxinus pennsylvanica or Acer negundo are often abundant in riparian or urban habitats (Burton et al. 2005).

Many studies have shown that functional traits of alien species are associated with invasiveness (Hamilton et al. 2005; Pyšek and Richardson 2007; Ordoñez et al. 2010; van Kleunen et al. 2010; Gallagher et al. 2015; Divíšek et al. 2018). However, the results were often ambiguous, possibly due to excluding environmental factors from analyses. In our study, we showed that traits, particularly of invasive neophytes, exhibit a strong relationship with the environment. Native species showed fewer associations with environmental factors as their traits may be more conservative in their native habitat and less likely to fluctuate. Yet, we looked at climatic conditions within a limited period (1961–1990) and native species might show significant changes in their functional traits as climate changes. Similar to native species, archaeophytes, the species that have settled in Germany for a long time, showed the least significant trait-environment relationships among alien species, while the traits of invasive neophytes are greatly affected by climate, geology and land cover. As discussed, this might be due to the fact that many invasive species were introduced from areas with different climatic or geological conditions and respond more flexibly to changes in the environment (Hellmann et al. 2008).

Invasive neophytes mainly show positive trait-environment relationships. Since the values for most of the traits increased with the incorporated environmental factors (especially climatic and land cover variables), we can expect future climate and land-cover change to affect invasive neophytes more strongly than other alien groups. We showed that climate may affect in particular SLA, insect pollination and phenology of invasive species, whereas land cover may mainly influence height, seed mass and wind pollination. Climate change could affect archaeophytes as well. They mainly showed positive relationships with climatic variables, and their values increased with the increase in temperature and precipitation. Future studies on the relationship between functional traits and environment of invasive plants are required in order to examine the effects of climate change or land cover changes. There is evidence that climate change may promote invasiveness (Pyšek et al. 2005), thus distinguishing which traits of alien species are benefiting under different climatic scenarios, can be valuable for management implications.