Advertisement

Biological Invasions

, Volume 19, Issue 7, pp 2159–2170 | Cite as

Present and future distribution of three aquatic plants taxa across the world: decrease in native and increase in invasive ranges

  • Morgane Gillard
  • Gabrielle Thiébaut
  • Carole Deleu
  • Boris Leroy
Original Paper

Abstract

Inland aquatic ecosystems are vulnerable to both climate change and biological invasion at broad spatial scales. The aim of this study was to establish the current and future potential distribution of three invasive plant taxa, Egeria densa, Myriophyllum aquaticum and Ludwigia spp., in their native and exotic ranges. We used species distribution models (SDMs), with nine different algorithms and three global circulation models, and we restricted the suitability maps to cells containing aquatic ecosystems. The current bioclimatic range of the taxa was predicted to represent 6.6–12.3% of their suitable habitats at global scale, with a lot of variations between continents. In Europe and North America, their invasive ranges are predicted to increase up to two fold by 2070 with the highest gas emission scenario. Suitable new areas will mainly be located to the north of their current range. In other continents where they are exotic and in their native range (South America), the surface areas of suitable locations are predicted to decrease with climate change, especially for Ludwigia spp. in South America (down to −55% by 2070 with RCP 8.5 scenario). This study allows to identify areas vulnerable to ongoing invasions by aquatic plant species and thus could help the prioritisation of monitoring and management, as well as contribute to the public awareness regarding biological invasions.

Keywords

Brazilian waterweed Climate change Parrot feather RCP scenarios Species distribution models Water primroses 

Introduction

Climate change and biological invasions are two of the main drivers of global change with impacts on health, ecosystems and biodiversity, which induce massive costs for society (e.g., Bradshaw et al. 2016). These drivers are expected to impact all ecosystems, including inland aquatic ecosystems (Dudgeon et al. 2006). Indeed, water temperature are expected to increase, while the modification of precipitation regimes may alter flow regimes (Whitehead et al. 2009; Watts et al. 2015). Moreover, climate change is also expected to have impacts on biological invasions, although the strength and direction of this impact varies between species (Mainka and Howard 2010; Bellard et al. 2013). There are 15 taxa living in inland aquatic ecosystems in the list of the 100 world’s worst invasive species from IUCN (Lowe et al. 2004), while this biotope represents less than 1% of the earth’s surface.

Most of the aquatic invasive plant species have been introduced into their invasive range by humans for their ornamental characteristics or for their use in the aquarium trade. Once they are released into aquatic environments, their spread is facilitated by flows and flooding events that connect water bodies. Propagules can also be dispersed by recreational boats and by birds. Dense waterweed (Egeria densa Planch.; Hydrocharitaceae), water primroses (Ludwigia hexapetala, Ludwigia grandiflora (Michx.) Greuter and Burdet and Ludwigia peploides subsp. montevidensis (Spreng.) P.H. Raven; Onagraceae), and parrot feather (Myriophyllum aquaticum (Vell.) Verdc.; Haloragaceae) are five aquatic plant species native to South America. They have been introduced into Europe, the USA, Australia and New-Zealand, where they increasingly extended their range (Yarrow et al. 2009; Hussner 2012; Thouvenot et al. 2013a). These taxa are considered invasive species by the IUCN because of their impacts on aquatic ecosystems: by forming dense mats on the water surface, they generate, amongst others, lower water flow, dam obstruction, and cause water navigation problems. In addition, the management costs of these species in invaded areas are high (Thouvenot et al. 2013a).

The control and eradication of invasive species are known to be more efficient during the early stages of the invasion, as well as at early stages of the plant life cycle. Thus, the challenge for managers is to detect the presence of invasive organisms as soon as possible, to avoid the colonisation of new environments and to limit their range expansion. To achieve this objective, species distribution models (SDMs) can be used as tools for the early detection of invasive species (Broennimann et al. 2007; Guisan et al. 2013), especially when establishing predictive scenarios, since such models have generally been proven to reflect the correct response to climate change (Stephens et al. 2016). Many studies have investigated invasive plant species distribution through projection by SDM (Kriticos et al. 2003; Peterson et al. 2008; Qin et al. 2014; Thalmann et al. 2015), but only a few of them have used non-native macrophyte species (Heikkinen et al. 2009; Alahuhta et al. 2011). Thus, knowledge about the current potential distribution of macrophytes species could be of interest for managers worldwide. Indeed, invasion fronts are not stabilised yet and the bioclimatic limits of aquatic plant species in their invasive range under current climates are unknown. Information about their potential bioclimatic and biogeographic extent could help to prioritise the monitoring of species presence in high suitability areas, and thus improve species control. For the same reasons, predicting the future potential distribution of aquatic plant species under different climate change scenarios is also of interest. Indeed, climate change is expected to alter the distribution of areas with suitable climate within the next decades, as shown for other invasive species (Bellard et al. 2013).

In this study, we focused on the climate change impact on the potential distribution of three invasive macrophyte taxa that have major impacts in their introduced ranges: E. densa, Ludwigia spp. and M. aquaticum. Previous studies considered the distribution of some of these species, but so far they have only been made at a country scale or based on a limited set of algorithms and general circulation models (GCMs) (Gallardo and Aldridge 2013; Kelly et al. 2014). By applying rigorous modelling methods, we aimed (1) to establish the current potential distribution of five invasive macrophytes and (2) to predict the future climate suitability for these species at world and continental scales.

Materials and methods

Species data

Presence data from the entire range (native and non-native) of species are necessary to model the climatic niches of invasive species accurately (Broennimann and Guisan 2008). Occurrences were collected from internet databases (GBIF, GISIN, AVH, speciesLink) completed by occurrences obtained from personal observations, personal communications and from published and grey literature (see Appendices S1 and S2 in Electronic supplementary material). To ensure correspondence with environmental variables, records before 1950 were discarded, and those after 1950 were aggregated into 0.8° cells, i.e. cells of about 10 × 10 km.

Water primroses have very similar morphologies, high phenotypic plasticity, and apart from the flowering period, they are often mistaken for one another when identifying species (Dandelot et al. 2005). Furthermore, the taxonomy of L. grandiflora and L. hexapetala is controversial. They are respectively called L. grandiflora subsp grandiflora and L. grandiflora subsp hexapetala in Europe, and L. grandiflora subsp hexapetala is sometimes referred to as L. grandiflora. These characteristics confuse the identification of the species, and as we could not be sure that the occurrences collected in databases related to the right species, we chose to consider the three taxa together. Hereafter they are called Ludwigia spp.

Climate data

We used bioclimatic variables from the WorldClim database (Hijmans et al. 2005), averaged for the 1950–2000 period, at a 5 arc-min resolution (≈0.8°). We applied a protocol to select relevant predictive variables for each taxa, selecting uncorrelated bioclimatic variables (Pearson’s r < 0.7) that best predicted the current distribution of the taxa (see Appendix S3 and Leroy et al. 2013; Bellard et al. 2016). The selected variables are shown in Table 1 and S3.3.
Table 1

Effect of selected climatic predictor variables, based on the fitted response curves (Appendix S4)

Variable

Egeria densa

Ludwigia spp.

Myriophyllum aquaticum

Mean annual temperature

 

Limiting < 13 °C

 

Mean diurnal range

Slightly negative

Slightly limiting

 

Annual temperature range

Limiting

<15 and >39 °C

  

Mean temperature of warmest quarter

Limiting

 

<15 and >28 °C

>27.5 °C

<12 and >37.5 °C

Mean temperature of coldest quarter

  

Limiting

<2 and >19 °C

Annual precipitation

Slightly limiting

Slightly negative

 

Precipitation seasonality

Slightly limiting

Slightly negative

 

Precipitation of driest quarter

Slightly limiting

Limiting

>480 mm

>500 mm

Precipitation of coldest quarter

Slightly limiting

Slightly negative

A variable was considered to have a slightly negative effect when the taxa response remained above 0.75, a slightly limiting effect when the response was between 0.5 and 0.75 for some values of the variable, and a limiting effect when the taxa response fell below 0.5. When there is a limiting effect, the variable values that limit the species response under 0.75 are indicated on the line below

To project future changes in distributions with respect to climate change, we used the four representative concentration pathway scenarios of the IPCC (RCP 2.6, 4.5, 6.0 and 8.5), based on different assumptions of greenhouse gas emissions. We chose two future periods: the 2050s and 2070s (30 year periods from 2041 to 2060 and 2061 to 2080). As uncertainty in forecasting future distribution is partially due to GCMs (Buisson et al. 2010), we used three different GCMs that simulated the impact of the different climate scenarios for the two future periods: the Hadley Centre Coupled Model version 3 (HADCM3) (Collins et al. 2001), Coupled Global Climate Model version 3 (CGCM3) (Yukimoto et al. 2012) and Community Climate System Model version 4 (CCSM4) (Gent et al. 2011). CGCM3 and CCSM4 were downloaded from WorldClim at 5 arc-min resolution.

Modelling methods

Species occurrence data with no corresponding climate data were removed so that occurrence points had obligatory bioclimatic correspondence. After this operation, the number of 0.8° cells used to train models was 837 for E. densa, 1454 for Ludwigia spp. and 1422 for M. aquaticum.

We performed the species distribution modelling and ensemble forecasting using the biomod2 package (Thuiller et al. 2009) with R Development Core Team (2015). We used nine different algorithms implemented in biomod2, including: three regression methods, (1) a generalized linear model (GLM) (McCullagh and Nelder 1989), (2) a generalized additive model (GAM) (Hastie and Tibshirani 1990) and (3) multivariate adaptive regression splines (MARS) (Friedman 1991); two classification methods, (4) classification tree analysis (CTA) (Breiman et al. 1984) and (5) flexible discriminant analysis (FDA) (Hastie et al. 1994); and four machine learning methods (6) artificial neural network (ANN) (Ripley 1996), (7) generalized boosted models (GBM) (Ridgeway 1999), (8) maximum entropy (MAXENT) (Phillips et al. 2006) and (9) random forest (RF) (Breiman 2001). These algorithms require datasets with both presence and absence, and we had presence-only records, so we randomly generated three sets of pseudo-absences with equal numbers of presences (Barbet-Massin et al. 2012).

We calibrated the models with 70% of randomly selected data, and evaluated the performance of each model with the remaining 30%. The evaluation stage was performed with three evaluation metrics, the area under the relative operating characteristic curve (ROC), the true skill statistic (TSS) and a similarity index, the Jaccard index (Jaccard 1901). Indeed, Leroy et al. (submitted) suggested that both AUC (area under the curve) and TSS were dependent on prevalence, so they recommended the use of an alternative similarity index to evaluate model performances, as done by Finch et al. (2006) and Ebeling et al. (2008) and also recommended by Li and Guo (2013). We repeated the calibration and evaluation operations three times to obtain an average value of model performance. We evaluated the response of the taxa to climatic predictors with the evaluation strip method (Elith et al. 2005). We used the ensemble forecast method implemented in the BIOMOD platform to combine the nine model outputs, and thus provide a robust forecast of our model taxa distributions. Probability maps were transformed into binary maps of suitable areas or non-suitable areas using the probability threshold that maximised the TSS value (Liu et al. 2005). We generated one current binary distribution map and three future binary distribution maps per scenario and per study period (i.e. 2050 and 2070). Consensus binary maps were obtained by committee averaging, i.e. by attributing presence in a cell when at least two of the three GCMs predicted presence, otherwise we assigned absence (Araújo and New 2007; Gallien et al. 2012).

Suitable aquatic environments

Since our model taxa have aquatic habitat requirements, we filtered our projected suitability maps to only include cells containing aquatic ecosystems adapted to our taxa. We downloaded land use datasets from the Global Lakes and Wetlands Database (GLWD) (Lehner and Döll 2004), and aggregated data into 0.8° cells, corresponding to the resolution of our projection raster. We selected lakes, reservoirs and rivers for E. densa, which is strictly affiliated to water, and added layers corresponding to freshwater marshes, floodplains, swamp forests, flooded forests, coastal wetlands, intermittent wetland and wetlands for Ludwigia spp. and M. aquaticum, which are amphiphyte taxa. For each taxa, we counted the number of cells occupied by suitable aquatic environments, and the number of those cells where models predicted presence of the taxa. Results of potential taxa distribution under current and future climates were expressed as a percentage of suitable aquatic environments, per scenario and per year, at the world scale and for six continents. Variability in future projections has been quantified per scenario and per taxa by calculating the standard deviation across the three GCMs, for the same extents the results of potential taxa distribution.

Results

Model evaluations

The fitted response curves of the three taxa showed a strong response to the selected predictor variables (Appendix S4). For the three taxa, the most limiting determinants for species distribution were temperature variables, especially the mean temperature of the warmest quarter which limited the response of all three taxa (Table 1).

For the three taxa, all the calibrated models had ROC values above 0.8, TSS values above 0.6 and Jaccard indices above 0.3 (meaning that all models predicted correctly at least 30% of presences in the cross-validation dataset) (Appendix S5), with an average ROC value, TSS value and Jaccard Index value higher than 0.9, 0.7 and 0.4, respectively (Table 2). Thus, all the calibrated models were included in the ensemble forecast.
Table 2

Models evaluation metrics

Species

Mean values (calibrated models)

Ensemble modelling values

ROC

TSS

Jaccard

ROC

TSS

Egeria densa

0.94

0.78

0.41

0.97

0.83

Ludwigia spp.

0.92

0.74

0.44

0.96

0.80

Myriophyllum aquaticum

0.95

0.83

0.46

0.98

0.84

Current potential distribution

The current bioclimatic range of the three taxa was predicted to represent 6.6–12.3% of their suitable habitats, with a large variation between continents (Table 3). In their native continent (South America), Ludwigia spp. and E. densa are predicted to have a suitable bioclimatic range in around 30% of the aquatic habitats and/or water bodies, while 14.5% of aquatic habitats are likely to be climatically suitable for M. aquaticum.
Table 3

Results of potential current species distribution at the world and continental scales, expressed in percentage of suitable water bodies for Egeria densa, and in percentage of water bodies and wetlands for Ludwigia spp. and Myriophyllum aquaticum

 

% of suitable environment

Egeria densa

Ludwigia spp.

Myriophyllum aquaticum

World

12.3

9.1

6.6

Africa

20.3

13.5

1.6

Asia

4.1

1.5

2.7

Oceania

69.5

30.3

17.3

Europe

24.6

12.5

17.5

North America

10.4

7.5

6.6

South America

31.6

29.2

14.5

Outside their native continent, Oceania was the most suitable continent with 17.3% (M. aquaticum), 30.3% (Ludwigia spp.) and 69.5% (E. densa) of suitable aquatic habitats. In New-Zealand, a large proportion of the aquatic habitats was predicted to be suitable for the three taxa, except at high altitudes in South Island. In Australia, the suitable areas were located in the north-east, south-east and in Tasmania, while the centre and north of the country were not predicted to be suitable. Asia was the least suitable continent with less than 5% of their aquatic habitats predicted to be suitable for all taxa. Differences occurred between taxa, e.g. Africa was predicted to be almost unsuitable for M. aquaticum (1.6%), but suitable for both E. densa (20.3%) and Ludwigia spp. (13.5%). In Europe and North America, the two continents where the three taxa are invasive, their potential current distribution represented respectively 12–25% and 7–10% of their specific suitable habitats (Table 3; Fig. 1).
Fig. 1

Percentage of suitable areas for Egeria densa, Ludwigia spp. and Myriophyllum aquaticum with current climatic conditions, and for 2050 and 2070 with the four RCP scenarios in Europe, North America and South America. The error bars are standard deviation across the three GCMs, represented for each future scenario

Future potential distribution

Europe and North America are the only two continents where our models predicted future increases in the size of the bioclimatic range for the three aquatic taxa (Fig. 1, Appendix S6), which were proportional to gas emission scenarios. According to the results using the RCP 8.5 scenario, by 2070, the three macrophyte taxa will increase their range up to 2.2-fold in Europe and between 1.4-fold and 1.8-fold in North America (Fig. 1).

In Europe, new areas, such as Iceland, are predicted to become suitable to M. aquaticum after 2050, even with the low gas emission increase scenario (Fig. 2). Although the climatic conditions of southern Norway and southern Sweden are predicted to be currently suitable for E. densa and M. aquaticum, their bioclimatic ranges will probably move further north and inland, independently of the future period or of the scenario considered (Fig. 2, Appendix S7). The bioclimatic range of L. spp., which is currently at lower latitudes than those of the other taxa, is predicted to increase northwards, especially into Ireland, the UK, Germany, the Netherlands and Denmark. Some areas in Central and Eastern Europe are predicted to become suitable for the three taxa, more or less severely, depending on the period and on the gas emission scenario considered (Fig. 2, Appendix S7). On the other hand, the ranges of the three macrophytes are predicted to decrease in the Mediterranean region, particularly under high gas emission scenarios.
Fig. 2

Current and projected future environmental suitability of Europe for Egeria densa, Ludwigia spp. and Myriophyllum aquaticum, according to two different climate scenarios (RCP 2.6 and RCP 8.5). ac Current climate suitability; df changes in the predicted distribution range of the species by 2050 according to RCP scenario 2.6; gi changes in the predicted distribution range of the species by 2070 according to RCP scenario 8.5

In North America, the climate of the Great Lake region is predicted to become suitable for the studied taxa from 2050, even with RCP 2.6, and especially for Lakes Ontario and Erie, and for the south of Lakes Huron and Michigan (Fig. 3). Climates of other large lakes of Canada are not predicted to become suitable for the three taxa (Fig. 3), which explains the relatively low increase of the percentage of suitable areas for future climates for this continent (Fig. 1). Other regions of North America are predicted to become suitable for the taxa, with a global progression northwards, in particular on the coasts, but with differences between taxa. The four scenarios show that water bodies located at the eastern border between Canada and the US are predicted to become suitable for M. aquaticum, while E. densa could even reach water bodies located on the left bank of the St Lawrence River (Fig. 3). The west of British Columbia and South Alaska are likewise predicted to become suitable areas for these two taxa. The range of Ludwigia spp. is also predicted to move further north than at present, particularly in the Great Lakes region, and some aquatic ecosystems in central US will become suitable environments. For the three macrophyte taxa, some areas which are currently suitable, are predicted to become unsuitable, especially in Mexico, Cuba, the central US and southern Florida under some scenarios.
Fig. 3

Current and projected future environmental suitability of North America for Egeria densa, Ludwigia spp. and Myriophyllum aquaticum, according to two different climate scenarios (RCP 2.6 and RCP 8.5). ac Current climate suitability; df changes in the predicted distribution range of the species by 2050 according to RCP scenario 2.6; gi changes in the predicted distribution range of the species by 2070 according to RCP scenario 8.5

In their native range, the distribution ranges of our model taxa will decrease proportionally to greenhouse gas emission scenarios (Fig. 1). Under the RCP 8.5 scenario, E. densa, Ludwigia spp. and M. aquaticum are predicted to lose 26, 55, and 20% respectively, of their current suitable areas. In other continents, future climate change will either decrease the proportion of areas suitable for these taxa, or maintain it, such as for E. densa in Oceania, where suitable environments are predicted to decrease slightly in Australia, but to increase in New Zealand (Appendix S6).

Discussion

In this study, we predicted the global changes in climate suitability for three invasive aquatic taxa which have massive economic and ecological impacts: dense waterweed (E. densa), water primroses (Ludwigia spp.), and parrot feather (M. aquaticum). Our results highlighted that large portions of aquatic ecosystems are predicted to be suitable for these taxa in most regions of the world. In addition, our results showed that climate change is predicted to have negative impacts in their native range, but positive impacts in their invasive range, which has important implications for their management. This work represents the first large-scale study of climate change impacts on invasive macrophyte distribution.

Taxa bioclimatic ranges and impacts of climate change

The predicted responses of taxa to bioclimatic variables (see Table 2) are consistent with known temperature tolerance and preferences (Getsinger and Dillon 1984; Hussner and Champion 2011; Thouvenot et al. 2013a). Our results suggested large differences in the proportion of suitable aquatic ecosystems between taxa for some regions of the world (e.g., Asia, Africa), with E. densa generally having the highest habitat suitability and M. aquaticum having the lowest suitability throughout the world. These differences can be explained by their different climatic niches, but also by the differences between the types of aquatic environments considered suitable for the different taxa. For example, in Oceania, water bodies are rare, and most of them are in areas with suitable climates for E. densa, which explains the high proportion of this taxa in this particular range. Another example is the difference between M. aquaticum and Ludwigia spp. which have similar life forms (amphiphytes) but different environmental preferences. The predicted range of M. aquaticum shows that this species has suitable bioclimatic areas at latitudes which are further north than those suitable for Ludwigia spp., a result consistent with models by Kelly et al. (2014) in Ireland. The higher cold tolerance of M. aquaticum and E. densa may allow them to proliferate in Nordic countries and in southern Alaska from 2050 onwards, contrary to Ludwigia spp. Our results suggest that all three taxa will benefit from climate change in their invasive range, where they could expand their distribution. Unsurprisingly, their range is predicted to shift northwards in Europe and North America, with a more severe progression for high greenhouse gas concentration scenarios. This result is consistent with the predictions of Gallardo and Aldridge (2013) and Kelly et al. (2014) in Ireland and in Great Britain, but also with predictions for other macrophyte species (Heikkinen et al. 2009; Alahuhta et al. 2011). However, the proportion of suitable environments is predicted to decrease in the future in their native range, as well as in other areas of the world where they have not yet been reported. In general, higher greenhouse gas concentration scenarios should create less suitable climatic niches than lower ones. Ludwigia spp. is the taxa which may undergo the most severe loss of environmental niches in South America, and suitable climatic conditions are predicted to almost disappear from parts of the continent, which may threaten the taxa in its native range.

Sources of uncertainties

The protocol applied here was devised with the aim to mitigate uncertainties. The high number of occurrences used to calibrate the models ensured good precision for the niche modelling, even though there is more monitoring and knowledge about taxa presence in their invaded ranges.

Our predictions illustrate the potential bioclimatic range of species based on their modelled environmental niches, and the occurrence of species within their suitable bioclimatic range is conditional upon multiple factors that can impede the colonisation of new environments by these plant taxa (see Fig. 1 in Guisan and Thuiller 2005). Firstly, species have to disperse to a suitable area, which happens mainly by vegetative reproduction and is often human mediated. Prevention of the species dispersion by humans might be more manageable and efficient than controlling the invasive species once they have colonised new sites (see below).

Secondly, our models did not consider local factors such as water body size, water depth, or water quality (pH, nutrients, turbidity, etc.) which can greatly influence the suitability of aquatic environments for the taxa studied (Feijoó et al. 2002; Hussner et al. 2009; Hussner 2010; Bornette and Puijalon 2010). In addition, photoperiod is another factor that was not represented in the predictive variables, even though it influences plant growth, and could limit their spread, especially at high latitudes. Nonetheless, large differences in photoperiod during the year also have consequences on local climates, and thus should be reflected indirectly in the climatic variables used. Include non-climatic variables to the models could allow to improve the spatial accuracy of the predictions.

Thirdly, the spread of the studied taxa throughout wetlands and water bodies might be impacted by competition with other native and invasive plant taxa. For example, several studies demonstrated that species from the Hydrocharitaceae family such as Elodea canadensis or Elodea nutallii recently experienced a northward range shift (Heikkinen et al. 2009; Carey et al. 2016), while the current potential distribution of Hydrilla verticillata is comparable with that of E. densa (Peterson et al. 2003). These species may therefore colonise the same water bodies than our studied species. Nonetheless, as many invasive species, the taxa investigated in this study do not seem to be adversely affected by competition with other species (Stiers et al. 2011; Thouvenot et al. 2013b). Thus, the colonisation of the same sites by these species might be more likely to lead to a succession of dominant invasives, or to communities of invasive plants. Consistent with this idea, Gallardo and Aldridge (2013) showed that the invasive species richness of aquatic plants may increase in Ireland in the future.

In addition, the GLWD database used in this study to limit the potential presence of the taxa to aquatic environments and wetlands, represents current locations of water bodies and wetlands. In the future, a decrease in precipitation and the modification of rainfall frequency could lead to modifications of water bodies and wetland size and location. To be more accurate in our predictions, we need an estimation of the future potential distribution of water bodies and wetlands.

Finally, the modelled distribution of Ludwigia spp. might be either more or less accurate than for the two other taxa studied, as it groups together three different species from the Ludwigia genus. The results presented in this study represent average potential distributions and if the three Onagraceae species do not have the same climatic limits, the results might be under- or overestimated compared with an analysis of individual species. On the other hand, if they present similar bioclimatic niches, considering them together might have enhanced the prediction.

Prioritisation of management policy

Our predictions of the current climate ranges of the three invasive taxa presented in this study do not reflect their current distribution, especially outside their native range. Indeed, taxa have not yet been introduced into every continent, and, among other factors, the time since introduction of the taxa into their invasive range has not always allowed them to be in equilibrium with the environment, so they probably have not reached all of their suitable environments yet. Our results therefore highlight, not only an imminent risk of colonisation of new areas by the taxa, but also an opportunity to control the progression of the invasion front. The taxa spread mainly vegetatively in their exotic ranges, but Ludwigia spp. are also capable of sexual reproduction, which could increase their dispersion capacity (Ruaux et al. 2009). In water bodies already colonised, propagules are mainly dispersed by water flow (Santamaría 2002; Li 2014). However, the colonisation of new watersheds can occur when propagules are transported by birds (Santamaría 2002), even though humans are the main vector of introduction and spread through water bodies (Zhang and Boyle 2010; Gallardo and Aldridge 2013).

In order to control the invasion front, part of the policy could be to sensitise people who are susceptible to use the water bodies to recognise the species, and to report presences to local organisations or to a database. Fishermen would therefore be able to remove invasive plant fragments from their boats when moving from one water body to another. Limiting the trade of these taxa could also be a solution, if coupled with increasing the awareness of people owning aquariums and garden ponds to the necessity of not introducing the species into natural ecosystems. Given the high human influence on the dispersion of aquatic plant species (Gallardo and Aldridge 2013), educating people about the species could be a solution to preclude these macrophytes from reaching new suitable environments.

In conclusion, our models predicted future northward shifts in the bioclimatic ranges of the taxa in their invasive ranges (Europe and North America). The increase in bioclimatic suitability may accelerate the rate of expansion of their northernmost invasion front. In addition, the possible increase in flooding events due to more frequent heavy precipitation events (IPCC 2012) may facilitate the ability of these taxa to disperse into new habitats, both within and outside their invasive ranges. Consequently, future monitoring of species presence should be prioritised both in current and future unoccupied but suitable habitats, within and to the north of current potential distributions. In Europe, the climate of Iceland is currently unsuitable for the three taxa, but is predicted to become suitable for E. densa and M. aquaticum under future climatic conditions. Given the known sensitivity of islands to invasion, any importations of these species to Iceland should be avoided. Water bodies of Australia and New-Zealand are predicted to remain highly suitable for E. densa, but the long distances between some water bodies in Australia could prevent the dispersion of this species across the whole of the southeastern and southwestern parts of the country.

Notes

Acknowledgements

We kindly thank Márcio José Silveira for providing the occurrences of the studied taxa in Brazil, and Aldyth Nys for the English editing of the manuscript. This work was supported by a Ph.D. fellowship from the French Ministry for Higher Education and Research to MG. We would like to warmly thank the two reviewers who evaluated and contributed to improve a previous version of this manuscript.

Supplementary material

10530_2017_1428_MOESM1_ESM.pdf (3.3 mb)
Supplementary material 1 (PDF 3402 kb)

References

  1. Alahuhta J, Heino J, Luoto M (2011) Climate change and the future distributions of aquatic macrophytes across boreal catchments. J Biogeogr 38:383–393. doi: 10.1111/j.1365-2699.2010.02412.x CrossRefGoogle Scholar
  2. Araújo MB, New M (2007) Ensemble forecasting of species distributions. Trends Ecol Evol 22:42–47. doi: 10.1016/j.tree.2006.09.010 CrossRefPubMedGoogle Scholar
  3. Barbet-Massin M, Jiguet F, Albert CH, Thuiller W (2012) Selecting pseudo-absences for species distribution models: how, where and how many? Methods Ecol Evol 3:327–338. doi: 10.1111/j.2041-210X.2011.00172.x CrossRefGoogle Scholar
  4. Bellard C, Thuiller W, Leroy B et al (2013) Will climate change promote future invasions? Glob Chang Biol 19:3740–3748. doi: 10.1111/gcb.12344 CrossRefPubMedPubMedCentralGoogle Scholar
  5. Bellard C, Leroy B, Thuiller W et al (2016) Major drivers of invasion risks throughout the world. Ecosphere 7:1–14. doi: 10.1002/ecs2.1241 CrossRefGoogle Scholar
  6. Bornette G, Puijalon S (2010) Response of aquatic plants to abiotic factors: a review. Aquat Sci 73:1–14. doi: 10.1007/s00027-010-0162-7 CrossRefGoogle Scholar
  7. Bradshaw CJA, Leroy B, Bellard C et al (2016) Massive yet grossly underestimated costs of invasive insects. Nat Commun. doi: 10.1038/ncomms12986 Google Scholar
  8. Breiman L (2001) Random forests. Mach Learn 45:5–32. doi: 10.1023/A:1010933404324 CrossRefGoogle Scholar
  9. Breiman L, Friedman JH, Olshean RA, Stone CJ (1984) Classification and regression trees. Chapman and Hall, LondonGoogle Scholar
  10. Broennimann O, Guisan A (2008) Predicting current and future biological invasions: both native and invaded ranges matter. Biol Lett 4:585–589. doi: 10.1098/rsbl.2008.0254 CrossRefPubMedPubMedCentralGoogle Scholar
  11. Broennimann O, Treier UA, Müller-Schärer H et al (2007) Evidence of climatic niche shift during biological invasion. Ecol Lett 10:701–709. doi: 10.1111/j.1461-0248.2007.01060.x CrossRefPubMedGoogle Scholar
  12. Buisson L, Thuiller W, Casajus N et al (2010) Uncertainty in ensemble forecasting of species distribution. Glob Chang Biol 16:1145–1157. doi: 10.1111/j.1365-2486.2009.02000.x CrossRefGoogle Scholar
  13. Carey MP, Sethi SA, Larsen SJ, Rich CF (2016) A primer on potential impacts, management priorities, and future directions for Elodea spp. in high latitude systems: learning from the Alaskan experience. Hydrobiologia 777:1–19. doi: 10.1007/s10750-016-2767-x CrossRefGoogle Scholar
  14. Collins M, Tett SFB, Cooper C (2001) The internal climate variability of HadCM3, a version of the Hadley Centre coupled model without flux adjustments. Clim Dyn 17:61–81. doi: 10.1007/s003820000094 CrossRefGoogle Scholar
  15. Dandelot S, Verlaque R, Dutartre A, Cazaubon A (2005) Ecological, dynamic and taxonomic problems due to Ludwigia (Onagraceae) in France. Hydrobiologia 551:131–136. doi: 10.1007/s10750-005-4455-0 CrossRefGoogle Scholar
  16. Dudgeon D, Arthington AH, Gessner MO et al (2006) Freshwater biodiversity: importance, threats, status and conservation challenges. Biol Rev 81:163–182. doi: 10.1017/S1464793105006950 CrossRefPubMedGoogle Scholar
  17. Ebeling SK, Welk E, Auge H, Bruelheide H (2008) Predicting the spread of an invasive plant: combining experiments and ecological niche model. Ecography (Cop) 31:709–719. doi: 10.1111/j.1600-0587.2008.05470.x CrossRefGoogle Scholar
  18. Elith J, Ferrier S, Huettmann F, Leathwick J (2005) The evaluation strip: a new and robust method for plotting predicted responses from species distribution models. Ecol Model 186:280–289. doi: 10.1016/j.ecolmodel.2004.12.007 CrossRefGoogle Scholar
  19. Feijoó C, García ME, Momo F, Toja J (2002) Nutrient absorption by the submerged macrophyte Egeria densa Planch.: effect of ammonium and phosphorus availability in the water colum on growth and nutrient uptake. Limnetica 21:93–104Google Scholar
  20. Finch JM, Samways MJ, Hill TR et al (2006) Application of predictive distribution modelling to invertebrates: Odonata in South Africa. Biodivers Conserv 15:4239–4251. doi: 10.1007/s10531-005-3577-z CrossRefGoogle Scholar
  21. Friedman JH (1991) Multivariate adaptive regression splines. Ann Stat 19:1–67CrossRefGoogle Scholar
  22. Gallardo B, Aldridge DC (2013) The “dirty dozen”: socio-economic factors amplify the invasion potential of 12 high-risk aquatic invasive species in Great Britain and Ireland. J Appl Ecol 50:757–766. doi: 10.1111/1365-2664.12079 CrossRefGoogle Scholar
  23. Gallien L, Douzet R, Pratte S et al (2012) Invasive species distribution models—how violating the equilibrium. Glob Ecol Biogeogr 21:1126–1136. doi: 10.1111/j.1466-8238.2012.00768.x CrossRefGoogle Scholar
  24. Gent PR, Danabasoglu G, Donner LJ et al (2011) The community climate system model version 4. J Clim 24:4973–4991. doi: 10.1175/2011JCLI4083.1 CrossRefGoogle Scholar
  25. Getsinger KD, Dillon CR (1984) Quiescence, growth and senescence of Egeria densa in Lake Marion. Aquat Bot 20:329–338. doi: 10.1016/0304-3770(84)90096-2 CrossRefGoogle Scholar
  26. Guisan A, Thuiller W (2005) Predicting species distribution: offering more than simple habitat models. Ecol Lett 8:993–1009. doi: 10.1111/j.1461-0248.2005.00792.x CrossRefGoogle Scholar
  27. Guisan A, Tingley R, Baumgartner JB et al (2013) Predicting species distributions for conservation decisions. Ecol Lett 16:1424–1435. doi: 10.1111/ele.12189 CrossRefPubMedPubMedCentralGoogle Scholar
  28. Hastie T, Tibshirani R (1990) Generalized additive models. Chapman and Hall, LondonGoogle Scholar
  29. Hastie T, Tibshirani R, Buja A (1994) Flexible discriminant analysis by optimal scoring. J Am Stat Assoc 89:1255–1270. doi: 10.2307/2290989 CrossRefGoogle Scholar
  30. Heikkinen R, Leikola N, Fronzek S et al (2009) Predicting distribution patterns and recent northward range shift of an invasive aquatic plant: Elodea canadensis in Europe. BioRisk 2:1–32. doi: 10.3897/biorisk.2.4 CrossRefGoogle Scholar
  31. Hijmans RJ, Cameron SE, Parra JL et al (2005) Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25:1965–1978. doi: 10.1002/joc.1276 CrossRefGoogle Scholar
  32. Hussner A (2010) Growth response and root system development of the invasive Ludwigia grandiflora and Ludwigia peploides to nutrient availability and water level. Fundam Appl Limnol/Arch Hydrobiol 177:189–196. doi: 10.1127/1863-9135/2010/0177-0189 CrossRefGoogle Scholar
  33. Hussner A (2012) Alien aquatic plant species in European countries. Weed Res 52:297–306. doi: 10.1111/j.1365-3180.2012.00926.x CrossRefGoogle Scholar
  34. Hussner A, Champion PD (2011) Myriophyllum aquaticum (Vell.) Verdcourt (parrot feather). In: Francis RA (ed) A Handbook of global freshwater invasive species. Routledge, New York, p 456Google Scholar
  35. Hussner A, Meyer C, Busch J (2009) The influence of water level and nutrient availability on growth and root system development of Myriophyllum aquaticum. Weed Res 49:73–80CrossRefGoogle Scholar
  36. IPCC (2012) Managing the risks of extreme events and disasters to advance climate change adaptation. Cambridge University Press, CambridgeGoogle Scholar
  37. Jaccard P (1901) Distribution de la flore alpine dans le Bassin des Drouces et dans quelques regions voisines. Bull la Société Vaudoise des Sci Nat 37:241–272Google Scholar
  38. Kelly R, Leach K, Cameron A et al (2014) Combining global climate and regional landscape models to improve prediction of invasion risk. Divers Distrib 20:884–894. doi: 10.1111/ddi.12194 CrossRefGoogle Scholar
  39. Kriticos DJ, Sutherst RW, Brown JR et al (2003) Climate change and the potential distribution of an invasive alien plant: Acacia nilotica ssp. indica in Australia. J Appl Ecol 40:111–124CrossRefGoogle Scholar
  40. Lehner B, Döll P (2004) Development and validation of a global database of lakes, reservoirs and wetlands. J Hydrol 296:1–22. doi: 10.1016/j.jhydrol.2004.03.028 CrossRefGoogle Scholar
  41. Leroy B, Paschetta M, Canard A et al (2013) First assessment of effects of global change on threatened spiders: potential impacts on Dolomedes plantarius (Clerck) and its conservation plans. Biol Conserv 161:155–163. doi: 10.1016/j.biocon.2013.03.022 CrossRefGoogle Scholar
  42. Li W (2014) Environmental opportunities and constraints in the reproduction and dispersal of aquatic plants. Aquat Bot 118:62–70. doi: 10.1016/j.aquabot.2014.07.008 CrossRefGoogle Scholar
  43. Li W, Guo Q (2013) How to assess the prediction accuracy of species presence–absence models without absence data? Ecography (Cop) 36:788–799. doi: 10.1111/j.1600-0587.2013.07585.x CrossRefGoogle Scholar
  44. Liu C, Berry PM, Dawson TP, Pearson RG (2005) Selecting thresholds of occurrence in the prediction of species distributions. Ecography (Cop) 28:385–393. doi: 10.1111/j.0906-7590.2005.03957.x CrossRefGoogle Scholar
  45. Lowe S, Browne M, Boudjelas S, De Poorter M (2004) 100 of the world’s worst invasive alien species a selection from the Global Invasive Species Database. Invasive Species Specialist Group, Species Survival Commission, World Conservation Union (IUCN)Google Scholar
  46. Mainka SA, Howard GW (2010) Climate change and invasive species: double jeopardy. Integr Zool 5:102–111. doi: 10.1111/j.1749-4877.2010.00193.x CrossRefPubMedGoogle Scholar
  47. McCullagh P, Nelder JA (1989) Generalized linear models, 2nd edn. Chapman and Hall, LondonCrossRefGoogle Scholar
  48. Peterson AT, Papes M, Kluza DA (2003) Predicting the potential invasive distributions of four alien plant species in North America. Weed Sci 51:863–868CrossRefGoogle Scholar
  49. Peterson AT, Stewart A, Mohamed KI, Araújo MB (2008) Shifting global invasive potential of European plants with climate change. PLoS ONE. doi: 10.1371/journal.pone.0002441 Google Scholar
  50. Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecol Model 190:231–259. doi: 10.1016/j.ecolmodel.2005.03.026 CrossRefGoogle Scholar
  51. Qin Z, DiTommaso A, Wu RS, Huang HY (2014) Potential distribution of two Ambrosia species in China under projected climate change. Weed Res 54:520–531. doi: 10.1111/wre.12100 CrossRefGoogle Scholar
  52. R Development Core Team (2015) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org
  53. Ridgeway G (1999) The state of boosting. Comput Sci Stat 31:172–181Google Scholar
  54. Ripley BD (1996) Neural networks and pattern recognition. Cambridge University, CambridgeCrossRefGoogle Scholar
  55. Ruaux B, Greulich S, Haury J, Berton J-P (2009) Sexual reproduction of two alien invasive Ludwigia (Onagraceae) on the middle Loire River, France. Aquat Bot 90:143–148. doi: 10.1016/j.aquabot.2008.08.003 CrossRefGoogle Scholar
  56. Santamaría L (2002) Why are most aquatic plants widely distributed? Dispersal, clonal growth and small-scale heterogeneity in a stressful environment. Acta Oecol 23:137–154. doi: 10.1016/S1146-609X(02)01146-3 CrossRefGoogle Scholar
  57. Stephens PA, Mason LR, Green RE et al (2016) Consistent response of bird populations to climate change on two continents. Science (80-) 352:84–87. doi: 10.1126/science.aac4858 CrossRefGoogle Scholar
  58. Stiers I, Crohain N, Josens G, Triest L (2011) Impact of three aquatic invasive species on native plants and macroinvertebrates in temperate ponds. Biol Invasions 13:2715–2726. doi: 10.1007/s10530-011-9942-9 CrossRefGoogle Scholar
  59. Thalmann DJK, Kikodze D, Khutsishvili M et al (2015) Areas of high conservation value in Georgia: present and future threats by invasive alien plants. Biol Invasions 17:1041–1054. doi: 10.1007/s10530-014-0774-2 CrossRefGoogle Scholar
  60. Thouvenot L, Haury J, Thiebaut G (2013a) A success story: water primroses, aquatic plant pests. Aquat Conserv Mar Freshw Ecosyst 23:790–803Google Scholar
  61. Thouvenot L, Puech C, Martinez L et al (2013b) Strategies of the invasive macrophyte Ludwigia grandiflora in its introduced range: competition, facilitation or coexistence with native and exotic species? Aquat Bot 107:8–16. doi: 10.1016/j.aquabot.2013.01.003 CrossRefGoogle Scholar
  62. Thuiller W, Lafourcade B, Engler R, Araújo MB (2009) BIOMOD—a platform for ensemble forecasting of species distributions. Ecography (Cop) 32:369–373. doi: 10.1111/j.1600-0587.2008.05742.x CrossRefGoogle Scholar
  63. Watts G, Battarbee RW, Bloomfield JP et al (2015) Climate change and water in the UK—past changes and future prospects. Prog Phys Geogr 39:6–28. doi: 10.1177/0309133314542957 CrossRefGoogle Scholar
  64. Whitehead PG, Wilby RL, Battarbee RW et al (2009) A review of the potential impacts of climate change on surface water quality. Hydrol Sci J 54:101–123. doi: 10.1623/hysj.54.1.101 CrossRefGoogle Scholar
  65. Yarrow M, Marin VH, Finlayson M et al (2009) The ecology of Egeria densa Planchon (Liliopsida: Alismatales): a wetland ecosystem engineer? Rev Chil Hist Nat 82:299–313CrossRefGoogle Scholar
  66. Yukimoto S, Adachi Y, Hosaka M et al (2012) A new global climate model of the Meteorological Research Institute: MRI-CGCM3. J Meteorol Soc Jpn 90A:23–64. doi: 10.2151/jmsj.2012-A02 CrossRefGoogle Scholar
  67. Zhang C, Boyle KJ (2010) The effect of an aquatic invasive species (Eurasian watermilfoil) on lakefront property values. Ecol Econ 70:394–404. doi: 10.1016/j.ecolecon.2010.09.011 CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2017

Authors and Affiliations

  1. 1.ECOBIO, UMR 6553 CNRSUniversité de Rennes 1RennesFrance
  2. 2.IGEPP, UMR 1349 INRAUniversité de Rennes 1Le RheuFrance
  3. 3.BOREA, UMR 7208, Muséum National d’Histoire Naturelle, Université Pierre et Marie Curie, Université de Caen Basse-Normandie, CNRS, IRDSorbonne UniversitéParisFrance

Personalised recommendations