Sensitivity of river fishes to climate change: The role of hydrological stressors on habitat range shifts

Science of the Total Environment 562 (2016) 435–445

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Sensitivity of river fishes to climate change: The role of hydrological stressors on habitat range shifts Pedro Segurado a,⁎, Paulo Branco a,b, Eduardo Jauch c, Ramiro Neves c, M. Teresa Ferreira a a b c

Universidade de Lisboa, Instituto Superior de Agronomia, Centro de Estudos Florestais (CEF), Tapada da Ajuda, 1349-017 Lisboa, Portugal CEris – Civil Engineering for Research and Innovation for Sustainability, Instituto Superior Técnico, Universidade de Lisboa, Lisboa, Portugal Universidade de Lisboa, Instituto Superior Técnico, MARETEC, Avenida Rovisco Pais, 1049-001 Lisboa, Portugal

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Hydrological stressors were retained in habitat the models for most species • Models using hydrology produce less severe predictions of fish habitat range shifts • Hydrological stressors strongly influence projections of habitat range shifts. • Salmo trutta fario was the most sensitive species to climate change

a r t i c l e

i n f o

Article history: Received 2 February 2016 Received in revised form 25 March 2016 Accepted 25 March 2016 Available online xxxx Keywords: Global change High-end scenarios Habitat models Hydrological stressor Hydrological modelling Freshwater fish

a b s t r a c t Climate change will predictably change hydrological patterns and processes at the catchment scale, with impacts on habitat conditions for fish. The main goal of this study is to assess how shifts in fish habitat favourability under climate change scenarios are affected by hydrological stressors. The interplay between climate and hydrological stressors has important implications in river management under climate change because management actions to control hydrological parameters are more feasible than controlling climate. This study was carried out in the Tamega catchment of the Douro basin. A set of hydrological stressor variables were generated through a process-based modelling based on current climate data (2008–2014) and also considering a high-end future climate change scenario. The resulting parameters, along with climatic and site-descriptor variables were used as explanatory variables in empirical habitat models for nine fish species using boosted regression trees. Models were calibrated for the whole Douro basin using 254 fish sampling sites and predictions under future climate change scenarios were made for the Tamega catchment. Results show that models using climatic variables but not hydrological stressors produce more stringent predictions of future favourability, predicting more distribution contractions or stronger range shifts. The use of hydrological stressors strongly influences projections of habitat favourability shifts; the integration of these stressors in the models thinned shifts in range due to climate change. Hydrological stressors were retained in the models for most species and had a high importance, demonstrating that it is important to integrate hydrology in studies of impacts of climate change on freshwater fishes.

⁎ Corresponding author. E-mail address: [email protected] (P. Segurado).

http://dx.doi.org/10.1016/j.scitotenv.2016.03.188 0048-9697/© 2016 Elsevier B.V. All rights reserved.

436

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

This is a relevant result because it means that management actions to control hydrological parameters in rivers will have an impact on the effects of climate change and may potentially be helpful to mitigate its negative effects on fish populations and assemblages. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Since 1976 the rate of global warming has been the steepest of the last 1000 years. This increase in average temperature has predictable impacts on species. Nevertheless, organisms, populations and communities tend to respond not to global averages but to more regional changes (Walther et al., 2002). Climate change and additional humaninduced impacts on rivers have led freshwater biodiversity to plunge at a pace faster than that of terrestrial or marine biodiversity (Vörösmarty et al., 2000; Alcamo et al., 2003; Jenkins, 2003; Dudgeon et al., 2006). Several species have specific temperature thresholds and precipitation tolerances (Woodward, 1987; Hoffman and Parsons, 1997) by which species distributions are shaped in different climatic regimes. Additionally, with climate alterations these specific combinations of temperature and hydrology tend to shift due to changes in precipitation, water temperatures, evaporation and hydrological patterns (Filipe et al., 2013). These changes ultimately alter habitat conditions for freshwater species (Regier and Meisner, 1990; Schindler, 2001; Wenger et al., 2011), and species often follow this environmental shift by accordingly shifting their distribution ranges. In fact, during the last century it became clear that poleward and upward shifts of species ranges were occurring all around the world and for a vast variety of taxa (Easterling et al., 2000; Hughes, 2000; McCarty, 2001; Walther et al., 2001). Freshwater species are severely affected by habitat alterations but their dispersal is strongly constrained by the river network configuration (Grant et al., 2007), e.g. even adjacent basins may show very low fish dispersal rates between them. Hence, river fish ranges have a limited ability to trace the geographical shifts of favourable environments. Freshwater communities are additionally impacted by multiple stressors that act upon the systems (Ficke et al., 2007; Palmer et al., 2008; Olden et al., 2010; Comte et al., 2013) and are sometimes composed by species of limited dispersion capability which seriously hamper their ability to handle habitat and environmental alterations (Woodward et al., 2010). Xenopoulos et al. (2005) estimated that the predicted reduction in river discharge caused by climate change and water abstraction may lead 75% of local fish diversity to extinction by 2070. Although difficult to assess, due to complex abiotic and biotic interactions (Clavero et al., 2010), the impacts of discharge reduction are even more relevant in southern Europe rivers (Maceda-Veiga, 2013). Climate change may also affect fish by operating indirectly on properties that are specific to river networks, such as the hydrological regime and water quality (Palmer et al., 2009). Most studies so far did not evaluate how these factors may operate simultaneously on fish occurrence (e.g. Buisson et al., 2008; Filipe et al., 2013). Planning conservation actions for freshwater fish species is challenging (Beatty et al., 2014) because of their different life cycles, some of which involving the use of different habitats at different life stages (Magoulick and Kobza, 2003), and the inherent longitudinal gradient of rivers, where upstream always influences downstream sections (Gorman and Karr, 1978; Allan et al., 1997). Climate is accepted to have a major control over the natural distribution of most species (Pearson and Dawson, 2003) and, consequently, the change in climate will most likely influence species range shifts (contractions and expansions) (Hughes, 2000; McCarty, 2001; Walther et al., 2002). In fact, modelling the ecological consequences of climate change has become a very prolific field of research in the last decade (Bellard et al., 2012). Although few predictions have been confirmed empirically, there is growing evidence that many species are shifting their range boundaries to track recent climate change in a consistent way both across terrestrial and aquatic environments (Comte and

Grenouillet, 2013). There are strong evidences that anthropogenic climate change is accelerating declines of many freshwater fish species, particularly in regions with arid or Mediterranean climates (Moyle et al., 2013). Freshwater fish are considered especially vulnerable to environmental changes, because their dispersal ability besides being greatly constrained by the river network structure (Grant et al., 2007), it is further limited by artificial barriers (Branco et al., 2014). Factors that are intrinsic to species, such as physiological traits, may also interfere on the potential effects of climate change on fish. Climate alterations can lead to changes in growth, survival, reproduction, or responses to changes at other trophic levels (Beaugrand et al., 2002, 2003). In fact, recent studies have demonstrated a significant effect of species traits on the variability of the observed range shifting trends on stream fishes under climate change (e.g. Alofs et al., 2014). This has important implications on the biotic integrity assessment of rivers that most often rely on indicators based on functional guilds of species. It is important to understand which functional guilds will be more or less resilient to adapt the bioassessment methods to climate change. The majority of work that tries to determine climate change impacts on freshwater fish species has been focused on cold water species (Keleher and Rahel, 1996; Nakano et al., 1996; Rahel et al., 1996) of the Northern hemisphere (Comte et al., 2013) and usually focusing on thermal suitability under climate change, neglecting the influence of other variables on fish species distribution (Eaton and Sheller, 1996; Rahel et al., 1996; Mohseni et al., 2003; Sharma et al., 2007). Widening this knowledge to other species types of different regions and encompassing several variables that are interconnected with climate change would grant us a more comprehensive understanding of the susceptibility of freshwater fish species to climate change (Beatty et al., 2014). For example, hydrology has been shown to play an important role on determining the life-history strategies of freshwater fish species (Mims and Olden, 2012; Chessman, 2013; Sternberg and Kennard, 2013). Nevertheless, and although climate change directly promote hydrological changes, approximately 80% of the studies that aim at determining the impacts of climate changes on distributional ranges of freshwater fish species do not consider the direct effect of hydrology and focus solely on climatic and habitat variables (Comte et al., 2013). Here, we calibrated empirical fish habitat favourability models in the Douro basin, NW Iberian Peninsula, and assessed the importance of considering hydrological stressors in the future fish habitat shift projections within a pilot sub-basin, under a high-end scenario of climate change. We assessed the shifts of habitat favourability for a range of fish species with distinct life-history traits and also the changes in species richness and species turnover. We expect an important impact of hydrological stressors in the distribution shifts of favourable fish habitats, which we hypothesize to be more evident for species that use habitats less prone to hydrological stressors. We also expect that non-native species will benefit from the climate and hydrological projected changes. The interplay between climate and hydrological stressors in the projected geographical shifts of habitat favourability has important implications in river management under climate change because while the implementation of actions to control hydrological parameters may be feasible and with immediate effects, climate mediation actions are harder to implement and their response is more dilated in time. The relative contribution of hydrological changes to fish habitat regression under future climate changes is also relevant to select species that are more vulnerable to hydrological alterations under climate change and hence be used as indicators of water scarcity and drought.

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

437

2. Material and methods

2.2. Data

2.1. Study area

We used fish data from a total of 254 sites in the Portuguese Douro basin sampled between 1996 and 2012 (Fig. 1). The sampling was performed by electrofishing following standard procedures similar to the one adopted by the European Committee for Standardization (CEN norm 14011 March 2003). Each site was sampled only once, and the fishing team progressed upstream in a zigzag pattern with single passes covering all present habitats (riffles, pools) and collecting fish with a dip net. Fish were then placed in a container filled with river water, identified to the species level and returned alive to the river. We selected species with N30 presence records and ended up with a set of nine dominant species (Table 1). The most common species were Pseudochondrostoma duriense and Squalius caroliterti, with 172 and 165 presence records, respectively, and the less frequent species was Anguilla anguilla with 31 presence records. Environmental variables including several site-descriptors, topographical variables, climate and hydrology, were compiled at the segment scale (stretch between confluences), using the CCM2 river network GIS theme as the spatial base (Vogt et al., 2007). Site descriptor variables were based on the variables already included in the CCM2 database. Hydrological stressors were modelled for the time period 2005– 2014 with the process-based modelling approach implemented in the Land module of the MOHID Water Modelling System (www.mohid. com), using meteorological data from MeteoGalicia WRF 12 km model, the Corine Land Cover 2006 (European Environmental Agency, 2010), soil maps from JRC (http://esdac.jrc.ec.europa.eu/) and the topography interpolated from the 90 m resolution Digital Elevation Model provided by the NASA Shuttle Radar Topographic Mission (Jarvis et al., 2008). Meteorological data included hourly temperature, relative humidity, solar radiation, wind and precipitation, between January 2008 and August 2014. MOHID Land is a modular system developed to simulate hydrodynamic processes, water quality and sediment transport in watersheds (Trancoso et al., 2009). The model setup included the following processes: overland flow, infiltration, evapotranspiration, runoff, drainage network transport, infiltration and saturated and unsaturated porous media transport (Braunschweig et al., 2004; Tzoraki et al., 2009). Collinearity among all pairs of variables was previously assessed using a Pearson correlation matrix. Among the pairs of variables with correlations above 0.75, the variables to be retained were selected according to their biological relevance. Fourteen environmental variables were selected for the subsequent analysis, including five sitedescriptors, three climatic variables and six hydrological parameters (Table 2). All GIS compilation of information and layer operations, were performed in ArcGis 10.0 (ESRI, Redlands, CA, USA).

The study area comprises the Portuguese part of Douro River basin (Fig. 1), the largest in the Iberian Peninsula, with a total surface of 97,290 km2, of which 18,643 km2 are within Portuguese territory (http://snirh.apambiente.pt/). This extensive geographic area encompasses a wide range of physical and chemical characteristics, land use and anthropogenic pressures. The climate is predominantly Mediterranean in most of its extension, with an increasingly Atlantic influence towards the Portuguese coastal region. The average annual rainfall varies considerably, from values in the range of b500 mm/yr in the central semi-arid depression to 1400 mm/yr mountainous northern areas. Human density increases downstream, from 21/km2 in Bragança to 701/km2 in Oporto. In general, the land use of the Douro basin is dominated by agricultural activities. Most human pressures associated with urbanization and industries are concentrated near the coast, namely water quality problems associated with nutrient enrichment and high biochemical oxygen demand (BOD) due to industrial effluent discharges, urban development and intensive farming (Cabecinha et al., 2009). The study is especially focused in the Tamega river basin, an important tributary of the Douro River in NW Portugal (Fig. 1). The main river course originates in Galicia, Spain (North of Monterrey, Verín) and flows southwest along 184 km. It occupies territories of both Portugal and Spain, covering ca. 3316 km2, of which 2637 km2 are in Portugal, with an average flow of 70,31m3/s (http://snirh.apambiente.pt/). Climatically, the basin is included in temperate oceanic sub-Mediterranean region (Ninyerola et al., 2005), with average monthly temperatures ranging between 12 °C and 17.5 °C and average monthly precipitations ranging between 700 mm/yr and 1300 mm/yr. An overall decrease in flow has been reported for the last decade, attributed to a general heating in the region (Santos et al., 2014). The basin is occupied by a very heterogeneous land use, including small farming plots, vineyards, orchards, eucalyptus and pine plantations, small oak forests, and scattered human and industrial areas. This river basin is extensively impacted by barriers, including ca. 162 weirs and 9 large dams. Four new large hydropower dams are planned to be constructed in this river basin in the context of the Portuguese National Program of dams with high Hydropower potential.

2.3. Modelling habitat favourability We first calibrated empirical predictive habitat models (Guisan and Zimmermann, 2000) using Boosted Regression Trees (BRT) to describe

Table 1 Number of presence records of the nine dominant species in the Portuguese Douro Basin among the 254 sites.

Fig. 1. Study area and location of fish sampling sites.

Species

Number of presence records

Anguilla anguilla Salmo trutta fario Luciobarbus bocagei Achondrostoma oligolepis Pseudochondrostoma duriense Squalius alburnoides Squalius carolitertii Gobio lozanoi Lepomis gibbosus

31 122 126 79 172 66 165 58 42

438

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

Table 2 Explanatory variables selected to fit the habitat models. Explanatory variables

Units

Description

m3/s m3/s

Monthly average flow Monthly average of minimum flow Number of days with no flow

Days with high flow

Number of days Number of days Number of days Number of days

Climatic Mean temperature

°C

Hydrology Average flow Average minimum flow Days with zero flow Days with null flow Days with low flow

Temperature standard deviation Total annual precipitation Geomorphology Strahler Area Upstream area Altitude gradient Mean elevation Mean slope

°C

Number of days where the average flow was below 0.01 m3/s Number of days where the average flow was lower than the monthly average flow Number of days where the average flow was higher than the monthly average flow Mean long term annual temperature in the primary catchment Standard deviation of the long-term average annual temperature in the primary catchment

mm

Mean long term average annual precipitation in the primary catchment

Strahler number km2 km2 % m %

Order number of the river segment Area of the primary catchment Area of the upstream catchment Relief energy of the river segment Mean elevation in the primary catchment Mean slope in the primary catchment

relationships between environmental variables and fish occurrence data in the Portuguese Douro basin. BRT is an ensemble methodology that fits statistical models in a different way from the conventional single parsimonious model fitting techniques. They combine two modelling techniques, (i) regression trees - that use binary splits to adjust the response to the predictors; and (ii) boosting - a method that combines multiple models to increase predictor ability. BRT has the advantage of being able to deal with collinearity and even non-linear relationships between predictors (Elith et al., 2008). The occurrence of species are usually determined by a great number of factors interacting with each other through complex relationships; a big advantage of boosted regression trees is that it deals automatically with these interactions (Elith et al., 2008). To fit BRT models, we followed the procedures proposed by Elith et al. (2008). To optimize the number of trees of each model, a stepwise procedure based on 10-fold cross validation with the area under the Receiver Operational Curve (AUC; Fielding and Bell, 1997) as the measure of accuracy was undertaken. The AUC varies between 0.5 (random classification) to 1 (perfect classification) to determine how the prediction of species occurrence deviates from random. The number of trees needed for correct predictions are determined by the learning rate (lr – contribution of each tree to the model) and by the tree complexity (tc – number of splits of each tree). Tree complexity was set to 2 and for each species the learning rate was adjusted to ensure that at least 1000 trees were combined into the final model (Elith et al., 2008). The model was then simplified by eliminating sequentially the least contributing variables to model fit, performing a k-fold cross validation at each step to help deciding when to stop the selection process. BRT modelling was performed with packages gbm (Ridgeway, 2007) and dismo (Elith et al., 2008; Hijmans et al., 2013) for R version 3.2.0 (R Core Team, 2015). The BRT models were calibrated using records of species presence/ absence and the set of regionalized environmental variables compiled for the whole Portuguese Douro river network. The fitted values of BRT models correspond to probabilities of species occurrence at each site. These probability values can be considered as surrogates of overall

habitat favourability. To quantify and compare the current and future total area of habitat available, we classified the probabilities in predicted presence or absence using species prevalence (proportion of sample sites containing the species) as the classification cut-off probability. The explanatory and predictive power of the models was expressed, respectively, by the resulting AUC values for the training dataset and for the cross validation procedure (mean AUC of the random 10-fold cross validation subsets). The importance of each environmental variable in the model (variable importance) was estimated by averaging the number of times a variable is selected for splitting a tree in the BRT model and the squared improvement resulting from these splits (Friedman, 2001). The resulting values were scaled to the 0–100 interval, with higher numbers indicating a stronger influence on the response variable. 2.4. Climate and hydrological change scenarios We used the BRT predictive models to extrapolate habitat favourability for all the segments in the Tamega sub-basin, both under the baseline conditions (2005–2014) and under a future climatic and hydrological change scenario (2046–2065). Hence, variables selected in the BRT models had to be computed for all the segments of the Tamega sub-basin both for the baseline and future scenario conditions. Site descriptor variables were considered constant between the baseline and the future scenario. Only climate and hydrology were considered to suffer changes. For the climatic projections, we considered the most pessimistic climate change scenario from the Fifth Assessment Report on Climate Change (IPCC, 2013), the Representative Concentration Pathways RCP8.5, for the time frame 2046–2065. The RCP8.5 assumes a scenario storyline that projects a trajectory of high population increase but moderate technological enhancements, leading to long term high energy demands. RCP8.5 is considered a baseline scenario as it does not involve any specific climate change mitigation and hence it is considered the upper frontier of the RCPs (Riahi et al., 2011). Temperature is projected for four periods, December–February, March–May, June– August and September–November, while precipitation is forecasted for two periods, April–September and October–March. We used the median forecasted values (50%) for the simulations. This climate change scenario was used to simulate changes in hydrological stressor variables with MOHID Land. To generate hydrological variables for the RCP8.5 scenario, the meteorology was processed accordingly to methodology described in Oliveira et al. (2012). The precipitation and temperature reference datasets were averaged for the two distinct areas identified in the RCP8.5 scenario and for the two different periods with distinct changes (summer and winter). The RCP changes were applied to these averages and correction factors were calculated. These factors were applied for each instant in the original datasets, originating a new dataset with the projected climate changes in temperature and precipitation that was used as input to the Tamega watershed model. 2.5. Fish sensitivity to climate change Several measures were computed to express species sensitivity to the climate and the hydrological change scenario. The difference between the modelled probabilities of species occurrence for the future and for the present was computed for each segment. Based on this difference the mean probability variation per segment as well as the percentage of segments with a decrease in probability were computed for each species. The model classification into current and future presence absence was also used to classify each segment into one of the following categories: (1) segments currently classified as presence that are classified as absence under the climate change scenario, (2) segments currently classified as absence that are classified as presence under the climate change scenario, (3) segments that remain unchanged (either presence or absence). Changes in predicted species richness between the present and the future and the species turnover per river segment

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

were also computed. Species turnover was computed as a percentage based on the ratio between the sum of the number of species gained and loss and the sum of species richness and the number of species gained in each segment (Peterson et al., 2002; Buisson et al., 2008). 2.6. Relative contribution of hydrology to habitat loss and gain The relative influence of the hydrological variables on the explanatory power of models was estimated by summing the respective variable importance. We run two additional BRT models, by not considering either climatic variables or hydrological variables as candidate variables. A total of three models were therefore available for each species that included, apart from the site-descriptor variables, the climatic variables (CLIM), the hydrological variables (HYDRO) and the complete set of variables (COMPLETE). Then we compared the models in terms of their accuracy and the future projections of species habitat suitability, species richness and turnover per segment. 3. Results 3.1. Climate and hydrological change scenarios The climate change scenario considered in this work projects an annual increase in mean air temperature (from + 1.9 °C in the west to 2.4 °C in the east region of the Tamega basin). The largest temperature increase is expected during summer (from + 2.3 °C in the west to 3.4 °C in the east region of the Tamega basin). Total precipitation is expected to suffer a decrease in the annual mean (from −10% in the west to 9% in the east region of the Tamega basin), also with the highest variation during summer (from −22% in the west to −17% in the east region of the Tamega basin). In the river segments under study, a mean variation of +2 °C in the Mean temperature and −115% in the Total annual precipitation are expected (Table 3). Under the selected climate change scenario, the hydrological model predicts a mean decrease in the average, maximum and minimum flows, a mean increase in the number of days per year with zero, null and high flows and a mean decrease in the number of days per year with low flow (Table 3). The majority of segments will suffer a decrease in average, maximum and minimum flow. The number of days with null and zero flow will remain unchanged (43.94% and 78.11%, respectively) or increase (56.06% and 21.89, respectively). However, for the majority of segments (70.20%) there will be a decrease in the number of days with low flow. For the number of days with high flow, there will be an increase for most river segments (70.03%). 3.2. Habitat favourability models Overall, the resulting favourability models showed an overall good discrimination and predictive power, according to the resulting AUC values, both considering results for the training data and for the 10fold cross-validations (Table 4). For two species, Squalius alburnoides

439

Table 4 Model precision as measured by the area under the ROC (AUC) and model validation using a 10 fold cross-validation (values between brackets). Species

Climate

Hydrology

Both

Anguilla anguilla Salmo trutta fario Luciobarbus bocagei Achondrostoma oligolepis Pseudochondrostoma duriense Squalius alburnoides Squalius carolitertii Gobio lozanoi Lepomis gibbosus

0.97 (0.84) 0.97 (0.89) 0.97 (0.88) 0.92 (0.75) 0.89 (0.75) 0.87 (0.68) 0.84 (0.69) 0.99 (0.92) 0.99 (0.89)

0.93 (0.82) 0.99 (0.88) 0.98 (0.88) 0.93 (0.78) 0.94 (0.79) 0.86 (0.62) 0.87 (0.68) 0.99 (0.89) 0.94 (0.87)

0.93 (0.84) 0.98 (0.89) 0.96 (0.88) 0.93 (0.79) 0.94 (0.79) 0.86 (0.74) 0.87 (0.67) 0.99 (0.91) 0.96 (0.88)

and Squalius caroliterti, and to a less degree also for Achondrostoma oligolepis and Pseudochondrostoma duriense, the models showed a poorer performance according to the cross-validation estimates of AUC. Interpretation of results regarding these four species should take into account this limitation. No significant differences in the accuracy measures for each species were found among model types (training data, Friedman Chi-square = 0.89, p-value = 0.64; cross validation, Friedman Chi-square = 4.22, pvalue = 0.12). For the models that considered site-descriptor variables and both climatic and hydrological stressors, the number of variables in each model varied between species, from two variables in the Anguilla anguilla model to all 14 variables included in the Gobio lozanoi model (Table 5). The variables that were present in a larger number of models and showing higher mean importance were the mean total annual precipitation and the drainage area in the upstream catchment, with, respectively, a frequency of 9 and 8 models and a variable importance of 21% and 19.1%. Temperature Standard deviation was only selected once and strahler was selected twice but showed the lowest mean importance (Table 5).

3.3. Species sensitivity to climate change According to the percentage of segments with a decrease in probability as predicted by the models that included site-descriptors, climate and hydrological stressors as candidate variables, the species that will lose more overall habitat favourability under the climate change scenario will be Salmo trutta fario, A. anguilla and S. alburnoides (Fig. 2; see also Supplementary data). The species that simultaneously will increase the overall habitat favourability and lose suitability in a lower percentage of segments are A. oligolepis, P. duriense and G. lozanoi. In terms of the percentage of total habitat that is lost under future climate change (see maps from suppl. mat.), S. trutta fario and A. anguilla are also the main loser species with, respectively, 16.3%, and 11.6% of segments of suitable habitat that are lost. Among the winner species, the nonnative G. lozanoi and Lepomis gibbosus showed the highest net gain of

Table 3 Mean difference in the climatic and hydrological variables between the modelled future scenario values and the present modelled values (left column) and the % of segments where there has been a decrease in the average, maximum and minimum flow and an increase in the number of days with null, zero, low and high flows. Variables

Mean difference

% of segments with a decrease

% of segments unchanged

% of segments with an increase

Mean temperature Temperature SD Total annual precipitation Average flow Max. average flow Min. average flow Days with null flow Days with zero flow Days with low flow Days with high flow

2.01 0 −115.36 −0.99 −12.95 −0.09 85.45 13.92 −8.96 7.86

100 0 0 99.83 98.32 79.63 0.00 0.00 70.20 29.29

0 100 0 0.00 0.00 19.02 43.94 78.11 0.67 0.67

0 0 100 0.17 1.68 1.35 56.06 21.89 29.12 70.03

440

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

Table 5 Selected variables, their importance (%) in each species' model, their mean importance and their absolute frequency in the models. Variables Strahler Upstream area Altitude gradient Area Mean elevation Mean slope Precipitation Temperature Temperature SD Average flow Min. avg. flow Days with null flow Days with zero flow Days with low flow Days with high flow Total

Anguilla anguilla

Salmo trutta fario

Luciobarbus bocagei

Achondrostoma oligolepis

Pseudochondrostoma duriense

Squalius alburnoides

12.1

60.8

23.2

2.1 29.8

15.1

62.1

21.2

20.9

37.9

4.0 15.9 15.5

18.2

4.1

2

1.0 4.7

2.6

8.0

1.2

5.2 8.0

26.7

1.4 15.6

27.1 31.0

18.4 2.9

26.8

25.4 18.3

10.8

5.0

5.1

0.6

12.1

6.4

5.0

3.2

10

3

4

Gobio Lepomis lozanoi gibbosus

6.1

5.9 2.8 1.2

33.2

Squalius carolitertii

14

segments with suitable habitat with an increase of, respectively 13.3% and 6.7 of segments.

3.4. Impact of hydrological stressors For most fish species, the hydrological variables resulted as an important component of the models' explanatory power, according to the variables importance in the models that used the complete set of candidate variables (Fig. 3). Only L. bocagei and A. anguilla did not include hydrological variables in their final models. For the remaining species, the contribution of hydrology, as measured by the sum of their importance (Fig. 3), varied between 27% in S. alburnoides and 54% in L. gibbosus. For the species that was predicted to be more affected by the future climate change scenario (S. trutta fario), the hydrological variables represented 31% of the variable importance.

10.9

0.3 21.0

2 8

1.2

3

0.7 16.5

2 6

3.9 19.1 7.8 0.7

4 9 5 1

6.7 9.4

5 5

13.5 12.4

1.7 13.9 20.3 7.5

12.6 4.7

4.0 8.7

10.6

1.9

4.6

5

2.6

0.8

3

8.1

4.1

5

3.3

3

5.5

7.1 4

34.7

Mean Number of importance models

9

15

36.6

17.7 4

There were differences in the habitat favourability, as measured by changes in the predicted probabilities of occurrence, among future projections that, apart from site-descriptors, only considered climatic variables, hydrological stressors and the complete set (Fig. 4). For example for S. trutta fario, in average the habitat favourability loss when only climatic variables were included is much stronger than when both climate and hydrology were considered; no average change was found when only hydrological stressors were included. The predicted habitat favourability loss for S. carolitertii is also stronger when only climatic variables are considered as candidates to inclusion. In contrast, for the non-native species L. gibbosus, there is an increase of habitat favourability when only climatic variables are considered when compared to both the models that only include hydrological stressors and the complete set of variables. Another non-native species, G. lozanoi, show a marked decrease in the gained habitat favourability when only hydrological stressors are considered.

Fig. 2. Predicted percentages of segments that are lost, gained or remains unchanged according to the COMPLETE models. Species are sorted in an increasing order of the net variation of favourable segments.

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

441

Fig. 3. Contribution of hydrological variables to the explanatory power of each COMPLETE model, as measured by the sum of the relative importance values of each variable. Species are sorted by an increasing order of importance of hydrological variables.

Fig. 4. Boxplots showing the distribution of the % of variation of the probabilities of occurrence of the segments, for each model type (climate only, hydrology only and complete variable set). Grey bars indicate non-significant differences (posthoc Friedman Nemenyi tests, p b 0.05).

442

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

3.5. Predicted variations of species richness and turnover Species richness per segment was predicted to show a more overall positive variation when only hydrological variables were considered as candidates (Fig. 5; see also the maps in Supplementary data), although the medium value was close to zero for the three model types. The medium predicted species turnover per segment varied between 33% for the models that only considered hydrological variables and 50% for the models that only considered climatic variables. For the models that considered the full set of variables, the species' turnover was 40%. For all models the river segment turnover varied between 0 and 100%. Significant differences were only found between the models that only included hydrological variables and the remaining model types, both for species richness variation (Friedman chi-squared, p-value b 0.001; Post-hoc Friedman Nemenyi test, p-value b 0.001) and for species turnover (Friedman chi-squared, p-value b 0.001; Post-hoc Friedman Nemenyi test, p-value b 0.001). 4. Discussion Even though climatic variables are at the core of climate change scenarios and the primary variables at play when evaluating the impacts of climatic alterations on river fish, our results show that projected shifts in the habitat favourability of river fishes under global change scenarios will be impacted by hydrological stressors. This is true despite no significant differences were found between the accuracy of habitat models that included and excluded hydrological variables. In general, the incorporation of hydrological stressors within the habitat favourability models resulted in attenuated range shifts under future climatic change when compared to models that only included climatic variables and site descriptors; the incorporation of hydrological stressors in the models led to a decrease in habitat losses for species that were predicted to loose overall habitat favourability, whereas it resulted in a decrease in habitat gains for species that were predicted to gain overall habitat favourability. Hydrology is essentially a function of climate and therefore climatic change scenarios have predictably a strong impact on hydrological variables. However, the fact that hydrology is very often ignored in studies that forecast the future impacts of climate change on freshwater fish populations has possibly to do with the unbalanced number of studies that focus on cold water fish species (Keleher and Rahel, 1996; Nakano et al., 1996; Rahel et al., 1996). These species are usually restricted to mountainous headwater river habitats, where water availability is typically not the most important limiting factor. Although for this set of species the part played by hydrology can be considered to

be negligible, for other species with different ecologies, hydrology is central to their survival and might be limiting for the completion of their life cycle (Magoulick and Kobza, 2003). This is even truer in river systems that are heavily impacted by artificial network fragmentation, as is the case of Iberian rivers (Maceda-Veiga, 2013). This study aimed to demonstrate how important is the role of hydrological variables in explaining present and influencing future species distribution, and it did so by using a range of ecologically diverse fish species, from cold water to warmer water species, from diadromous to potamodromous to resident and non-native species, encompassing a representative set of functional guilds found in southern Europe. The climatic variable found to have the highest importance was total precipitation, which in fact might be argued to be a surrogate of hydrology. On the other hand, the variable which importance was found to be the lowest, was a strictly climatic variable – the temperature standard deviation – which was only retained in the best fitting model for G. lozanoi, a non-native species. When using the complete set of variables as candidate for inclusion in the models, hydrological variables were retained in all but two species' models (L. bocagei and A. anguilla) and showed a high degree of importance. Both L. bocagei and A. anguilla show a wide distribution along river networks which might reduce the importance of hydrological variables in explaining their distribution as the longitudinal gradient, strongly related with flow, might be lost. Because species evolved to be adapted to specific hydrological regimes of their native river system (Ficke et al., 2007), the relevance of hydrology as a habitat constraint would be expectable. When hydrological conditions suffer alterations, invasive species might gain competitive advantages (Baltz and Moyle, 1993; Ross et al., 2001; Annear et al., 2004) and native species were shown to experience a reduction in recruitment, which is particularly the case with specialist species (Cross and Moss, 1987; Mion et al., 1998; Modde et al., 2001) that are not able to adapt to the new hydrological conditions of the river system (Poff and Allan, 1995; Ficke et al., 2007). Due to the species-specific character of thermal ranges, communities will self-arrange themselves to attain new equilibria (Ficke et al., 2007). In this study, no clear relationship between the impacts of considering hydrological stressors in the habitat favourability models and the thermal dependency of species was found. For example, according to our results, the projected range shift of S. trutta fario, which has been most often modelled exclusively considering the loss of suitable thermal habitat (Almodóvar et al., 2012; Filipe et al., 2013), was strongly constrained by hydrological stressors. For other species that are not typically thermal dependent, such as A. anguilla and L. bocagei, our results showed an overall weak contribution of hydrological stressors in the projections. For the species that is most affected by climate change, S. trutta

Fig. 5. Boxplots showing the distribution of the predicted variation of species richness and the species turnover among the sites, for each model type (climate only, hydrology only and complete variable set).

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

fario, hydrological variables yielded an importance of 31% in the BRT models. The importance of hydrology to assess the impacts of climate change is reinforced by the results produced when hydrology was set aside to model favourability variations - the outputs predicted a loss of favourability at a much higher percentage of segments than when integrating hydrological variables in the models. The same pattern is found for S. carolitertii, a species more related with cold fast flowing waters, which was predicted to lose habitat, as a consequence of the increase in temperature, when hydrology was not considered in the modelling approach. In contrast, L. gibbosus, a non-native species of warmer waters, is predicted to experience a stronger expansion when climatic variables are included and hydrology is excluded from the models. This is likely because the increase in temperatures will approximate those found in its natural range, but when hydrological stressors are included in the models, the increased harshness of the hydrological regime will balance out the gains experienced when just considering climatic variability. A similar pattern occurs for G. lozanoi, a non-native species that has been translocated from Northern Spain, for which the amount of gained favourability within the river network is predicted to be strongly contracted when climatic variables are ignored and only hydrological constraints are at play. Climate change will have an impact not only in terms of species persistence and range shifts, but also on fish assemblages. In this study, the species richness was projected to show an overall positive variation when climate variables were set apart, but for all three types of models, the medium value among river segments sited around zero. Hence, globally, no strong variation in terms of species richness is expected to occur in the basin. Nonetheless, as gains and losses of species were predict to occur in several segments, the turnover of species composition was predicted to be high and still higher when hydrology was ignored. However, when allowing both variable types to be included in the models, the average predicted turnover was of about 40%, which is less than the 60% predicted turnover in freshwater fish species in French rivers (Conti et al., 2015), but in line with and earlier prediction of about 50% turnover of fish in French streams (Buisson et al., 2008). The predicted turnover in the present work finds parallel in other studies, for instance for the fauna in Mexico (Peterson et al., 2002), European plants (Thuiller et al., 2015), endemic flora in Southern African continent (Broennimann et al., 2006) and mammals in African national parks (Thuiller et al., 2006). Studies that assess the relative impact of climate change on native and no-native species are generally unanimous in showing a general trend towards stronger positive responses among non-native species, which is particularly more evident in aquatic systems (Bierwagen et al., 2008; Sorte et al., 2013). This study also predicts a stronger positive response of the two non-native species in comparison to native species in face of the projected changes in climate and hydrology. Our results show that climate and hydrological changes will facilitate the dispersal of the two non-native species into new river segments, especially when hydrological stressors are ignored. Furthermore, because non-native species are the main winners, a big part of the observed species turnover may be attributed to the spread of the non-native species and the loss of native species. According to our results and for our study area, when only climate was considered and hydrology was set apart, a harsher contraction of species distributions or stronger shifts in range were forecasted. It might be argued that this would be, in fact, a more conservative approach and that it is more advisable to use worse case scenarios when planning conservation actions and when trying to restrict the impacts of climate change on species and communities. However, this can result in feedback responses with unpredictable impacts on communities. It is the correct depiction, or prediction, of nature that serves conservation purposes best. The fact that significant differences, in turnover and species richness, only arose between models that only included hydrological stressors and the other two that included climate might indicate that the effects of hydrology were masked by the effects of climate and

443

overrun by the predicted impacts of climatic variables. A possible explanation is that, because all fish species are ectotherms (Moyle and Cech, 2004), i.e., can only regulate their temperature by changing their position along the river network, they will first of all favour sites with appropriate water temperature (Brett, 1971; Nevermann and Wurtsbaugh, 1994; Nielsen et al., 1994; Brio, 1998). Furthermore, in river systems, fish thermal regulation is limited by the temperature variations present in the network (Ficke et al., 2007). Nevertheless, when applying climatic scenarios, both variable types face important changes and are at play when predicting changes in species favourability. S. trutta fario, A. anguilla and S. alburnoides face the strongest predicted impacts of climate change by reducing their distributions. S. alburnoides is predicted to cope better with this distribution reduction by slightly shifting its range, which is not the case of S. trutta fario and A. anguilla that are not predicted to gain any new segments. Alternatively, S. alburnoides, and A. oligolepis, P. duriense, G. lozanoi and L. gibbosus will, by shifting their range, show a net increase of the number of segments in which they are predicted to be present, resulting in an effective distribution expansion. In conclusion, this work showed the importance of integrating hydrological stressor variables into the study of predicted impacts of climate change on freshwater fish species. It has done so by focusing on a pilot basin that climatic change scenarios predict to be even toughened in the future and forecasting geographical shifts in favourable habitats for a wide range of ecologically distinct fish species. These species will predictably differ both in their suffered impacts of climate change and on their ability to shift their distribution ranges to better be able to cope with future alterations. Richness will not suffer an overall alteration because of the expected turnover of species, due to species range shifts as a response to climate change. This work also contributed to identify the most appropriate species to be used as indicators to help monitoring the impacts of climate as well as hydrological changes. Both S. trutta and G. lozanoi can be considered as the most appropriate indicator species in the Tamega basin, even though their response to climate change shows opposing trends: while S. trutta will lose habitat favourability G. lozanoi will increase. However, because the distribution of G. lozanoi in the Douro river basin after its introduction from northern Spain probably did not reached an environmental equilibrium, the geographical range of the species might expand regardless of any climatic or hydrological change. Hence, the uncertainty regarding the direct causes of the observed range expansion of this species hampers its favourability as an indicator species. The results of this study are also relevant within a river management perspective, as they suggest that impacts of climate change may be affected by the interplay of hydrological stressors with climate and hence the implementation of appropriate management actions to control hydrological parameters in rivers may potentially be helpful to mitigate its negative effects on river fish populations and assemblages. In regulated rivers, which is the case of the Tamega River basin, where more large dams planned or under construction, these management actions would necessarily involve collaborative arrangements with dam managers to optimize reservoir release schedules (Palmer et al., 2008).

Acknowledgments This study was supported by EU funds (Preparatory Action on development of prevention activities to halt desertification in Europe) from the DURERO project “Duero river basin: water resources, water accounts and target sustainability indices” (Grant Agreement reference 07.0329/2013/671322/SUB/ENV.C1). Pedro Segurado is supported by the MARS project (Managing Aquatic ecosystems and water Resources under multiple Stress) funded under the 7th EU Framework Program, Theme 6 (Environment including Climate Change), Contract No: 603378. Paulo Branco is supported by a grant from the Fundação para a Ciência e Tecnologia (SFRH/BPD/94686/2013).

444

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2016.03.188. References Alcamo, J., Döll, P., Henrichs, T., 2003. Development and testing of the water GAP 2 global model of water use and availability. Hydrol. Sci. J. 48, 317–337. Allan, J.D., Erickson, D.L., Fay, J., 1997. The influence of catchment land use on stream integrity across multiple spatial scales. Freshw. Biol. 37, 149–161. Almodóvar, A., Nicola, G.G., Ayllon, D., Elvira, B., 2012. Global warming threatens the persistence of Mediterranean brown trout. Glob. Chang. Biol. 18, 1549–1560. Alofs, K.M., Jackson, D.A., Lester, N.P., 2014. Ontario freshwater fishes demonstrate differing range-boundary shifts in a warming climate. Diversity & Distribution 20, 123–136. Annear, T., Chisholm, I., Beecher, H., Locke, A., Aarrestad, P., Burkhart, P., Coomer, N., Estes, C., Hunt, J., Jacobson, R., Jobsis, G., Kauffman, J., Marshall, J., Mayes, K., Smith, G., Wentworth, R., Stalnaker, C., 2004. Instream Flows for Riverine Resource Stewardship. Instream Flow Council, Cheyenne, WY. Baltz, D.M., Moyle, P.B., 1993. Invasion resistance to introduced species by a native assemblage of California stream fishes. Ecol. Appl. 3, 246–255. Beatty, S.J., Morgan, D.L., Lymbery, A.J., 2014. Implications of climate change for potamodromous fishes. Glob. Chang. Biol. 20 (6), 1794–1807. Beaugrand, G., Reid, P.C., Ibanez, F., Lindley, J.A., Edwards, M., 2002. Reorganization of North Atlantic marine copepod biodiversity and climate. Science 296 (5573), 1692–1694. Beaugrand, G., Brander, K.M., Lindley, J.A., Souissi, S., Reid, P.C., 2003. Plankton effect on cod recruitment in the North Sea. Nature 426 (6967), 661–664. Bellard, C., Bertelsmeier, C., Leadley, P., Thuiller, W., Courchamp, F., 2012. Impacts of climate change on the future of biodiversity. Ecol. Lett. 15, 365–377. Bierwagen, B.G., Rahel, F.J., Thomas, R., 2008. Special section: a synthesis of climatechange effects on aquatic invasive species – introduction. Conservation Biology 22, 518–520. Branco, P., Segurado, P., Santos, J.M., Ferreira, M.T., 2014. Prioritizing barrier removal to improve functional connectivity of rivers. J. Appl. Ecol. 51 (5), 1197–1206. Braunschweig, F., Chambel, P., Fernandes, L., Pina, P., Neves, R., 2004. The object oriented design of the integrated modelling system MOHID. Proceedings of the XVth International Conference on Computational Methods in Water Resources (CMWR XV), June 13–17, Chapel Hill, NC, USA. Brett, J.R., 1971. Energetic responses of salmon to temperature. A study of some thermal relations in the physiology and freshwater ecology of sockeye salmon (Oncorhynchus nerka). Am. Zool. 11, 99–113. Brio, P.A., 1998. Staying cool: behavioral thermoregulation during summer by young-ofyear brook trout in a lake. Trans. Am. Fish. Soc. 127, 212–222. Broennimann, O., Thuiller, W., Hughes, G., Midgley, G.F., Alkemade, J.M.R., Guisan, A., 2006. Do geographic distribution, niche property and life form explain plants’ vulnerability to global change? Glob. Chang. Biol. 12, 1079–1093. Buisson, L., Thuiller, W., Lek, S., Lim, P., Grenouillet, G., 2008. Climate change hastens the turnover of stream fish assemblages. Glob. Chang. Biol. 14, 2232–2248. Cabecinha, E., Cortes, R., Pardal, M.A., Cabral, J.A., 2009. A stochastic dynamic methodology (StDM) for reservoir water quality management: validation of a multi-scale approach in a South Europe basin (Douro, Portugal). Ecol. Indic. 9, 329–345. Chessman, B.C., 2013. Identifying species at risk from climate change: traits predict the drought vulnerability of freshwater fishes. Biol. Conserv. 160, 40–49. Clavero, M., Hermoso, V., Levin, N., Kark, S., 2010. Geographical linkages between threats and imperilment in freshwater fish in the Mediterranean Basin. Divers. Distrib. 16, 744–754. Comte, L., Grenouillet, G., 2013. Do stream fish track climate change? Assessing distribution shifts in recent decades. Ecography 36, 1236–1246. Comte, L., Buisson, L., Daufresne, M., Grenouillet, G., 2013. Climate-induced changes in the distribution of freshwater fish: observed and predicted trends. Freshw. Biol. 58, 625–639. Conti, L., Comte, L., Hugueny, B., Grenouillet, G., 2015. Drivers of freshwater fi sh colonisations and extirpations under climate change. Ecography 38, 510–518. Cross, F.B., Moss, R.E., 1987. Historic changes in fish communities and aquatic habitats in plains streams of Kansas. In: Matthews, W.J., Heins, D.C. (Eds.), Community and Evolutionary Ecology of North American Stream Fishes. University of Oklahoma Press, Norman, OK, pp. 155–165. Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z., Knowler, D.J., Lèvêque, C., Naiman, R.J., Prieur-Richard, A.H., Soto, D., Stiassny, M.L., Sullivan, C.A., 2006. Freshwater biodiversity: importance, threats, status and conservation challenges. Biol. Rev. 81, 163–182. Easterling, D.R., Meehl, G.A., Parmesan, C., Changnon, S.A., Karl, T.R., Mearns, L.O., 2000. Climate extremes: observations, modeling, and impacts. Science 289 (5487), 2068–2074. Eaton, J.G., Sheller, R.M., 1996. Effects of climate warming on fish thermal habitat in streams of the United States. Limnol. Oceanogr. 41, 1109–1115. Elith, J., Leathwick, J.R., Hastie, T., 2008. A working guide to boosted regression trees. J. Anim. Ecol. 77, 802–813. European Environmental Agency (2010). b. (http://www.eea.europa.eu/data-and-maps/ data/corine-land-cover-2006-raster-3; accessed April 2015) Ficke, A.D., Myrick, C.A., Hansen, L.J., 2007. Potential impacts of global climate change on freshwater fisheries. Rev. Fish Biol. Fish. 17, 581–613. Fielding, A.H., Bell, J.F., 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ. Conserv. 24, 38–49.

Filipe, A.F., Markovic, D., Pletterbauer, F., Tisseuil, C., Wever, A., Schmutz, S., Bonada, N., Freyhof, J., 2013. Forecasting fish distribution along stream networks: brown trout (Salmo trutta) in Europe. Divers. Distrib. 19 (8), 1059–1071. Friedman, J.H., 2001. Greedy function Approximation: a gradient boosting machine. Ann. Stat. 29, 1189–1232. Gorman, O.T., Karr, J.R., 1978. Habitat structure and stream fish communities. Ecology 59, 507–515. Grant, E.H.C., Lowe, W.H., Fagan, W.F., 2007. Living in the branches: population dynamics and ecological processes in dendritic networks. Ecol. Lett. 10, 165–175. Guisan, A., Zimmermann, N.E., 2000. Predictive habitat distribution models in ecology. Ecol. Model. 135, 147–186. Hijmans RJ, Phillips S, Leathwick J and Elith J (2013) Species distribution modelling. Package “dismo”. – Available at: http://cran.r-project.org/web/packages/dismo/dismo.pdf. Hoffman, A.A., Parsons, P.A., 1997. Extreme Environmental Change and Evolution. Cambridge Univ Pres, Cambridge. Hughes, L., 2000. Biological consequences of global warming: is the signal already apparent? Trends in Ecology and Evolution 15, 56–61. IPCC (2013) Annex I: Atlas of Global and Regional Climate Projections Supplementary Material RCP8.5 [van Oldenborgh GJ, Collins M, Arblaster J, Christensen JH, Marotzke J, Power SB, Rummukainen M and Zhou T (eds.)]. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V and Midgley PM (eds.)]. Available from www.climatechange2013.org and www.ipcc.ch. Jarvis A, Reuter HI, Nelson A and Guevara E (2008) Hole-filled 415 SRTM for the Globe Version 4, Available From the CGIAR-CSI SRTM 416 90m Database (http://srtm.csi. cgiar.org; accessed in April 2015). Jenkins, M., 2003. Prospects for biodiversity. Science 302, 1175–1177. Keleher, C.J., Rahel, F.J., 1996. Thermal limits to salmonid distributions in the rocky mountain region and potential habitat loss due to global warming: a geographic information system (GIS) approach. Trans. Am. Fish. Soc. 125, 1–13. Maceda-Veiga, A., 2013. Towards the conservation of freshwater fish: Iberian Rivers as an example of threats and management practices. Rev. Fish Biol. Fish. 23, 1–22. Magoulick, D.D., Kobza, R.M., 2003. The role of refugia for fishes during drought: a review and synthesis. Freshw. Biol. 48, 1186–1198. McCarty, J.P., 2001. Ecological consequences of recent climate change. Conserv. Biol. 15 (2), 320–331. Mims, M.C., Olden, J.D., 2012. Life history theory predicts fish assemblage response to hydrologic regimes. Ecology 93, 35–45. Mion, J.B., Stein, R.A., Marschall, E.A., 1998. River discharge drives survival of larval walleye. Ecol. Appl. 8, 88–103. Modde, T., Muth, R.T., Haines, G.B., 2001. Floodplain wetland suitability, access, and potential use by juvenile razorback suckers in the middle Green River, Utah. Trans. Am. Fish. Soc. 130, 1095–1105. Mohseni, O., Stefan, H.G., Eaton, J.G., 2003. Global warming and potential changes in fish habitat in US streams. Clim. Chang. 59, 389–409. Moyle, P.B., Cech, J.J., 2004. Fishes: An Introduction to Ichthyology. fifth ed. Prentice Hall, Englewood Cliffs, NJ. Moyle, P.B., Kiernan, J.D., Crain, P.K., Quiñones, R.M., 2013. Climate change vulnerability of native and alien freshwater fishes of California: a systematic assessment approach. PLoS One 8 (5), e63883. Nakano, S., Kitano, F., Maekawa, K., 1996. Potential fragmentation and loss of thermal habitats for charrs in the Japanese archipelago due to climatic warming. Freshw. Biol. 36, 711–722. Nevermann, D., Wurtsbaugh, W.A., 1994. The thermoregulatory function of diel vertical migration for a juvenile fish, Cottus extensus. Oecologia 98, 247–256. Nielsen, J.L., Lisle, T.E., Ozaki, V., 1994. Thermally stratified pools and their use by steelhead in northern California streams. Trans. Am. Fish. Soc. 123, 613–626. Ninyerola, M., Pons, X., Roure, J.M., 2005. Atlas Climático Digital de la Península Ibérica. Metodología y aplicaciones en bioclimatología y geobotánica ISBN 932860-8-7 Universidad Autónoma de Barcelona, Bellaterra. Olden, J.D., Kennard, M.J., Leprieur, F., Tedesco, P.A., Winemiller, K.O., Garcãa-Berthou, E., 2010. Conservation biogeography of freshwater fishes: recent progress and future challenges. Divers. Distrib. 16, 496–513. Oliveira L, Martins T, Lobo-Ferreira JP, Oliveira MM, Novo ME, Leitão TE (2012) Água, Ecossistemas Aquáticos e Actividade Humana – Projecto PROWATERMAN. Quarto Relatório Temático – Contributos para o Desenvolvimento de Medidas para uma Gestão Sustentável dos Recursos Hídricos no Sul de Portugal. Relatório 153/2012DHA/NAS. LNEC. Lisboa. Palmer, M.A., Liermann, C.A.R., Nilsson, C., Florke, M., Alcamo, J., Lake, P.S., Bond, N., 2008. Climate change and the world's river basins: anticipating management options. Front. Ecol. Environ. 6, 81–89. Palmer, M.A., Lettenmaier, D.P., Poff, N.L., Postel, S.L., Richter, B., Warner, R., 2009. Climate change and river ecosystems: protection and adaptation options. Environ. Manag. 44, 1053–1068. Pearson, R.G., Dawson, T.P., 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Glob. Ecol. Biogeogr. 12 (5), 361–371. Peterson, A.T., Ortega-Huerta, M.A., Bartley, J., SanchezCordero, V., Soberon, J., Buddemeier, R.H., Stockwell, D.R.B., 2002. Future projections for Mexican faunas under global climate change scenarios. Nature 416, 626–629. Poff, N.L., Allan, J.D., 1995. Functional-organization of stream fish assemblages in relation to hydrological variability. Ecology 76, 606–627. R Core Team (2015) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. URL: http://www.R-project.org/ (accessed November 2015).

P. Segurado et al. / Science of the Total Environment 562 (2016) 435–445 Rahel, F.J., Keleher, C.J., Anderson, J.L., 1996. Potential habitat loss and population fragmentation for cold water fish in the North Platte River drainage of the Rocky Mountains: response to climate warming. Limnol. Oceanogr. 41, 1116–1123. Regier, H.A., Meisner, J.D., 1990. Anticipated effects of climate change on freshwater fishes and their habitat. Fisheries 15, 10–15. Riahi, K., Krey, V., Rao, S., Chirkov, V., Fischer, G., Kolp, P., Kindermann, G., Nakicenovic, N., Rafai, P., 2011. RCP-8.5: exploring the consequence of high emission trajectories. Climatic Change 109, 33–57. Ridgeway G (2007) Generalized Boosted Models: A Guide to the gbm Package. 12pp.– 490. http://cran.r-project.org/web/packages/gbm/vignettes/gbm.pdf (accessed in 20 September 2013). Ross, R.M., Lellis, W.A., Bennett, R.M., Johnson, C.S., 2001. Landscape determinants of nonindigenous fish invasions. Biol. Invasions 3, 347–361. Santos, R.M., Sanches Fernandes, L.F., Moura, J.P., Pereira, M.G., Pacheco, F.A., 2014. The impact of climate change, human interference, scale and modelling uncertainties on the estimation of aquifer properties and river flow components. J. Hydrol. 519, 1297–1314. Schindler, D.W., 2001. The cumulative effects of climate warming and other human stresses on Canadian freshwaters in the new millennium. Can. J. Fish. Aquat. Sci. 58, 18–29. Sharma, S., Jackson, D.A., Minns, C.K., Shuter, B.J., 2007. Will northern fish populations be in hot water because of climate change? Divers. Distrib. 13, 2052–2064. Sorte, C.J.B., Ibañez, I., Blumenthal, D.M., Molinari, N.A., Miller, L.P., Grosholz, E.D., Diez, J.M., D'Antonio, C.M., Olden, J.D., Jones, S.J., Dukes, J.S., 2013. Poised to prosper? A cross-system comparison of climate change effects on native and non-native species performance. Ecol. Lett. 16, 261–270. Sternberg, D., Kennard, M.J., 2013. Environmental, spatial and phylogenetic determinants of fish life-history traits and functional composition of Australian rivers. Freshw. Biol. 58, 1767–1778. Thuiller, W., Broennimann, O., Hughes, G., Alkemade, J.R.M., Midgley, G.F., Corsi, F., 2006. Vulnerability of African mammals to anthropogenic climate change under conservative land transformation assumptions. Glob. Chang. Biol. 12, 424–440.

445

Thuiller, W., Lavorel, S., Araujo, M., Sykes, M.T., Prentice, I.C., 2015. Climate change threats to plant diversity in Europe. Proc. Natl. Acad. Sci. 102, 8245–8250. Trancoso, A.R., Braunschweig, F., Leitão, P.C., Obermann, M., Neves, R., 2009. An advanced modelling tool for simulating complex river systems. Sci. Total Environ. 407, 3004–3016. Tzoraki, O., Nikolaidis, N.P., Trancoso, A.R., Braunschweig, F., Neves, R., 2009. A reach-scale biogeochemical model for temporary rivers. Hydrol. Process. 23, 272–283. Vogt, J., Soille, P., de Jager, A., Rimavičiūtė, E., Mehl, W., Foisneau, S., Bodis, K., Dusart, J., Paracchini, M.L., Haastrup, P., Bamps, C., 2007. A Pan-European River and Catchment Database. European Commission - JRC, Luxembourg (EUR 22920 EN). Vörösmarty, C., Green, P., Salisbury, J., Lammers, R.B., 2000. Global water resources: vulnerability from climate change and population growth. Science 289, 284–288. Walther, G.-R., Burga, C.A., Edwards, P.J. (Eds.), 2001. “Fingerprints” of Climate Change—Adapted Behaviour and Shifting Species Ranges. Kluwer Academic/Plenum, New York. Walther, G.R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T.J., Formentin, J.-M., Hoegh-Guldberg, O., Bairlein, F., 2002. Ecological responses to recent climate change. Nature 416 (6879), 389–395. Wenger, S.J., Isaak, D.J., Luce, C.H., Neville, H.M., Fausch, K.D., Dunham, J.B., Dauwalter, D.C., Young, M.K., Elsner, M.M., Rieman, B.E., Hamlet, A.F., Williams, J.E., 2011. Flow regime, temperature, and biotic interactions drive differential declines of trout species under climate change. Proc. Natl. Acad. Sci. U. S. A. 108, 14175–14180. Woodward, F.I., 1987. Climate and Plant Distribution. Cambridge Univ, Press, Cambridge. Woodward, G., Perkins, D.M., Brown, L.E., 2010. Climate change and freshwater ecosystems: impacts across multiple levels of organization. Philosophical Transactions of the Royal Society B, Biological Science 365, 2093–2106. Xenopoulos, M.A., Lodge, D.M., Alcamo, J., Märker, M., Schulze, K., Van Vuuren, D.P., 2005. Scenarios of freshwater fish extinctions from climate change and water withdrawal. Glob. Chang. Biol. 11, 1557–1564.