- Email: [email protected]
Interspecies variation in the risks of metals to bats
Environmental Pollution 206 (2015) 209e216
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Interspecies variation in the risks of metals to bats atrice V. Hernout a, b, *, Ste phane Pietravalle b, Kathryn E. Arnold a, Colin J. McClean a, Be b, c James Aegerter , Alistair B. A. Boxall a a
Environment Department, The University of York, Heslington, York, UK The Food and Environment Research Agency, Sand Hutton, York, UK c National Wildlife Management Centre, Animal Health and Veterinary Laboratories Agency, Sand Hutton, York, UK b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 March 2015 Received in revised form 31 May 2015 Accepted 2 June 2015 Available online xxx
A modeling framework was used to assess the risk of four metals to UK bat species. Eight species of bats were predicted to be “at risk” from one or more of the metals in over 5% of their ranges. Species differed significantly in their predicted risk. Contamination by Pb was found to pose the greatest risk, followed by Cu, Cd and Zn. A sensitivity analysis identified the proportion of invertebrates ingested as most important in determining the risk. We then compared the model predictions with a large dataset of metals concentrations in the tissues (liver, kidney) of Pipistrellus sp. from across England and Wales. Bats found in areas predicted to be the most “at risk” contained higher metal concentrations in their tissues than those found in areas predicted “not at risk” by the model. Our spatially explicit modeling framework provides a useful tool for further environmental risk assessment studies for wildlife species. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Insectivorous bats Spatially explicit model Trace metal elements Environmental risk assessment Model evaluation
1. Introduction Declines in many bat populations (e.g. Pipistrellus pipistrellus, Rhinolophus hipposideros, Rhinolophus ferrumequinum and Myotis myotis) have been observed across Europe (Jones et al., 2009; Dietz et al., 2009). Population declines can be the result of a number of factors including: environmental and climate change, changes in resources e.g. water, prey availability and quality, roost loss, disturbance, urbanization and industrialization, agricultural intensification, the increase in wind turbines, the pressure of disease and also exposure to chemicals in the environment (Frick et al., 2010; Jones et al., 2009; Walker et al., 2007; Wickramasinghe et al., 2003). Bats are long-lived mammals and consume a large amount of prey each night during their foraging period, and are thought to be particularly exposed to chemicals (Clark and Shore, 2001). Exposure to environmental contaminants, such as metals, may be considered as additional stressors to bats, although very few
* Corresponding author. Department of Environmental Toxicology The Institute of Environmental and Human Health (TIEHH), Texas Tech University, 1207 Gilbert Drive, Box 41163, TX 79409-3290, Lubbock, TX, USA. E-mail addresses: [email protected] (B.V. Hernout), Stephane. [email protected] (S. Pietravalle), [email protected] (K.E. Arnold), [email protected] (C.J. McClean), [email protected]. gov.uk (J. Aegerter), [email protected] (A.B.A. Boxall). http://dx.doi.org/10.1016/j.envpol.2015.06.016 0269-7491/© 2015 Elsevier Ltd. All rights reserved.
studies have considered the effects of metals on bat species (Clark and Shore, 2001). Environmental contamination by metal compounds has been widespread across Europe since the industrial revolution. In England and Wales, more than 80% of contaminated land sites have been reported to be contaminated by metals and metalloids (Environment Agency, 2009). In addition, as metals do not degrade, they are highly likely to accumulate in mammalian body tissues, especially for top predators and long-lived species such as bats (Dietz et al., 2009). Metals can elicit a range of toxic effects on wildlife, including induction of tremors, spasms, lethargy, lack of control in body movement, as well as sublethal effects at the biochemical, physiological and histological levels (e.g., oxidative stress, DNA damage, tissue damage including inclusion bodies), and, in some cases, can cause mortality (Clark and Shore, 2001; nchez-Chardi Hoffman et al., 2001; Hurley and Fenton, 1980; Sa et al., 2009). As flying mammals depend upon exceptional levels of motor-control and muscular activity, bats may show particular vulnerability to the physiological effects of exposure to metals. To explore the potential risks of chemical exposure to bats, we previously developed and applied a spatial modeling framework using a risk characterization approach, to assess the risks from soilassociated metals (cadmium, copper, lead and zinc) to the health of population of P. pipistrellus in England and Wales (Hernout et al., 2013). However, in our previous study, we only looked at one bat
210
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
species and the modeling framework was not evaluated against monitoring data. The exposure of different bat species to metals is likely to vary due to differences in factors such as their food intake, dietary composition and distribution. For example, bats specializing in consuming prey with a high metal accumulation capacity, which have a high food intake and a spatial range restricted to polluted areas, might be expected to have higher exposure than others. Studies monitoring metal residues in bats show that renal metal concentrations differ across bat species, which may reflect differences in dietary exposure (Walker et al., 2007). As observed for passerine birds, interspecific differences in metal exposure may be linked with their diet (Berglund et al., 2011). For example, it was shown that the pied flycatcher (Ficedula hypoleuca) accumulated more metals than great tits (Parus major) as the diet composition of pied flycatchers is composed of a large proportion of insects from higher trophic levels than the great tits (Berglund et al., 2011). When using modeling frameworks of the type described by Hernout et al. (2013), it is also important to understand the sensitivity of a framework to changes in model input parameters. This knowledge can be invaluable in informing the parameterisation process and guiding the model development. Sensitivity analyses are strongly recommended for use in Environmental risk assessment (ERA) (Schmolke et al., 2010), and emphasized by many institutions (e.g. EFSA Journal, 2009; Health Canada Contaminated Sites Division, 2005). However, the literature remains scarce. The analyses consist in examining how outputs vary as inputs are varied, to understand how the risk predictions are dependent on the variability and the uncertainty of the factors contributing to the risk (Grimm and Railsback, 2005; Risk assessment guidance for superfund, 2001). Complex approaches, as we have used, involve mathematical and statistical techniques and can include the effect of the combination of several factors having different statistical distributions (Risk Assessment Guidance for Superfund, 2001). Sensitivity analyses have been used previously in ecological modeling exercises such as an agent-based model, simulating skylark (Alauda arvensis) population response to landscape change (Parry et al., 2013). The most important parameters were identified for the model parameterization process or subsequent empirical studies. Model evaluation is also strongly recommended in modeling practise, although relatively scarce (Schmolke et al., 2010). In this study, to improve our knowledge of the potential threat of metal contamination to bats, we: (1) used the modeling framework which is based on a basic risk characterization approach (Hernout et al., 2013) to explore the risks of metal exposure for 14 bat species and identify the species most at risk from exposure to four metals; (2) used the modeling framework to determine the most important parameters affecting the predicted exposure, which are the main drivers in exposure risk, and to understand why certain species may be more vulnerable to metal exposure than others, and finally (3) compared levels of metals in different tissues (liver and kidney) of Pipistrellus sp. from across England and Wales (internal exposure) with our model predictions (based on oral exposure estimations) to evaluate our model. 2. Methods 2.1. Risk of UK bat species to metal exposure The modeling framework method described by Hernout et al. (2013) was applied to estimate the risks of four metals to 14 bat species present in the UK, namely: Barbastella barbastellus, Eptesicus serotinus, Myotis bechsteinii, Myotis daubentonii, Myotis mystacinus, Myotis nattereri, Nyctalus leisleri, Nyctalus noctula, Pipistrellus sp. (P. pipistrellus and Pipistrellus pygmaeus), Pipistrellus nathusii,
Plecotus auritus, Pl. austriacus, Rhinolophus ferrumequinum, R. hipposideros. The model used a risk characterization approach where the daily oral dose is compared with a ‘safe’ dose value to derive a ratio. The comparison of the ratio with a trigger value (of 1) indicates whether the risk is acceptable or not (using a resolution of 5 5 km2 cell) (Hernout et al., 2013). The modeling framework requires information on concentrations of metals in soils, soil-insect accumulation factors, bat diet, bat distribution and toxicity data on the metal studied. Concentrations of Cd, Cu, Pb and Zn in soil in England and Wales were obtained from NSRI (National Soil Resources Institute) at a 5 5 km2 resolution. Ecological data on bats (bat diet composition, foraging distance and weight) were gathered from the literature (Table S1, Table S2). Daily food intakes and no observed effect levels (NOAELs) were estimated based on the average bat weight for each species and were derived using the allometric relationships described in Nagy, 1987 and Sample et al., 1996, respectively (Table S2) (Hernout et al., 2013). The experimental studies used to derive the NOAELs considered a reproductive endpoint and chronic effects (Sample et al., 1996). Further details on the experimental studies are presented in Table S3. The NOAEL was divided by an uncertainty factor of five to calculate the “safe dose” (Hernout et al., 2013). Biota accumulation factor (BAF) data were obtained from the literature for each of the invertebrate orders listed in the bat diet for the four metals studied (Table S4). The bat distribution dataset (presence/absence data) was provided by the Bat Conservation Trust for each bat species (Data derived from the National Bat Monitoring Programme; NBMP). The spatial analysis was done using Geographic Information System (ArcGIS, ArcMap Version 9.3.1) (ESRI, Redlands Calif., USA). The final output was a risk characterization ratio (RCR) for each 5 5 km2 cell defined by the ratio between the daily dose of metal that a bat receives (mg/g body weight/d) and predicted safe daily dose for the metal (mg/g body weight/d), within the spatial distribution of the bat (Hernout et al., 2013). The percentage of areas at risk for each species and metal, as well as for the groups of metals combined were derived from the number of cells where a species was found to be at risk (i.e. with an RCR1) divided by the total number of cells in which the bat species is present (Hernout et al., 2013). 2.2. Identification of key drivers of risk A number of analyses were performed to identify the key factors that drive the risk of metals to bats as determined in the model. Distributions of selected model input parameters (Table 1) covering all species were used alongside the model to identify which of these were the most important in determining the risk values calculated. The Emulator GEM-SA 1.1 (Gaussian Emulation Machine for Sensitivity Analysis, Kennedy, 2005) was used to determine the effect of each individual input, or pairs of inputs on the output value. Further details on the emulation process are given is Text S2 (Supplemental data). The different input parameters and their respective ranges are shown in Table 1. The sensitivity analyses cannot integrate spatial components and therefore we did not include the spatial range in which the bat species reside. As each bat distribution is unique, the ranges of metal concentrations in soils will also vary for each bat species (Figure S1). Thus, the sensitivity analyses used a simplified generic distribution to represent all bats, representing the whole of England and Wales based on the totality of soil metal concentrations available for this area (see overall UK map soil concentrations in Hernout et al., 2013). The differences of RCRs across metals and bat species were tested using the non-parametric KruskaleWallis test. To explore
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216 Table 1 Input parameters and their ranges used in the sensitivity analysis to identify the key drivers of risks of soil-associated metals to bats. Parameters
Range used in the sensitivity analysis
Concentration of the metal in the soil (mg/g dry weight). Range limited to 95 and 99% of observations.
95% observations 99% observations Cd: 0e1.6 Cd: 0e3.2 Cu: 0e47.9 Cu: 0e100.3 Pb: 0e194.9 Pb: 0e514.6 Zn: 0e175.8 Zn: 0e350.8 Cd: 0.37-0.56 Cu: 5.86e8.85 Pb: 3.08e4.65 Zn: 61.59e93.03 Between bat species: 0.13-0.18
Safe daily dose for the metal (NOAEL/5) (mg/g body weight/d)
Amount of food eaten (g dry weight/g body weight/d) Proportion of the diet accounted for an individual prey item (% volume)
Coleoptera Diptera Lepidoptera Araneidea Dermaptera Hemiptera Hymenoptera Opiliones Orthoptera Trichoptera
0e1
0e0.25
the effects of species location (area in which the bat is living) and feeding range (area in which the bat is foraging), we compared the soil concentrations to which each species would be exposed using the non-parametric KruskaleWallis test. As multiple statistical tests were applied, the p-values were adjusted using the Holm method. The level of statistical significance was set to 0.05. We calculated the soil concentrations for the spatial distribution for each species using the method described by Hernout et al. (2013). This spatial application associating the bat distribution and its feeding range (based on the foraging distance of each species) does not take into account the diet or the food intake. Data analyses were performed with the software R version 2.12.1. 2.3. Model evaluation The model evaluation involved the comparison of our risk predictions (based on oral exposure estimations) with the measured internal concentrations of metal contained in bat tissues. Concentrations of metals in bat tissues (kidney and liver) were measured in 193 individuals of Pipistrellus sp. Bats were selected from a large archive of 3000 bats, using metal concentrations in soils from the locations at which the bats were found. We used the frequency distribution of soil metal concentrations across England and Wales to obtain our subsample of 193 bats reflecting that frequency distribution, which also includes bats from areas with high metal concentrations as well as areas wih low concentrations. The acquisition of the monitoring data on trace metal concentrations is detailed in SI (see Text S1: Sample collection and processing, Quantification of metal concentrations, Quality assurance and quality control, Data analyses). We compared tissue concentrations in two groups separated by their risk location status defined by the model (where bats are “at risk” and “not at risk”). The risk was defined for an area where bats have an RCR higher than one (Hernout et al., 2013). The hypothesis being that bat individuals found in areas predicted to be “at risk” would have higher internal concentrations of metals in their tissues than bats obtained from areas predicted to be “not at risk”. The Receiver-operating characteristic (ROC) was used to compare monitoring data with model predictions (Platts et al., 2008). The ROC curve provides a good visualization between high sensitivity and high specificity which can vary when discriminating two groups of data separated by a threshold value. Level of statistical significance was set to 0.05. Data analyses were performed with the software R
211
version 2.12.1. To compare our risk predictions across species against metal tissue concentrations in other bat species, we selected two publications studying several species for two areas (Austria and Britain, Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007; respectively). 3. Results 3.1. Risk of metal exposure to UK bat species RCRs for all species were calculated for all of the 5 5 km2 grid cells across which each species was distributed. Comparing across all species, median RCRs (range) were highest for Cu, followed by Pb, Cd and Zn: 0.30 (0.14e0.40), 0.23 (0.07e0.34), 0.16 (0.11e0.26) and 0.12 (0.03e0.20), respectively (Fig. 1). Within each bat species, the RCRs were significantly different across metals (KruskaleWallis c2 d.f ¼ 3 ¼ 1893, 3331, 251, 8790, 1092, 5841, 958, 5648, 193, 5383, 4582, 307, 510, and 2922, P < 0.05, for the bat species: B. barbastellus, E. serotinus, M. bechsteinii, M. daubentonii, M. mystacinus, M. nattereri, N. leisleri, N. noctula, P. nathusii, P. pipistrellus/pygmaeus, P. auritus, P. austriacus, R. ferrumequinum, and R. hipposideros, respectively) (Fig. 1). M. mystacinus and Pl. auritus appeared to be most exposed to Cd (Fig. 1a). M. daubentonii, M. mystacinus, Pipistrellus sp, P. Nathusii, N. leisleri, and N. noctula were most exposed to Pb (Fig. 1b). M. nattereri, N. noctula, M. daubentonii, P. nathusii, N. leisleri, E. serotinus, and M. mystacinus appeared to be most exposed to Cu (Fig. 1c). N. noctula, M. mystacinus, M. nattereri and M. daubentonii were predicted to be most exposed to Zn (Fig. 1d). The RCRs were significantly different across bat species for each metal (KruskaleWallis c2 h ¼ 3175, 5180, 7830, and 8273, d.f ¼ 14, 13, 13 and 13, P < 0.05, for Cd, Cu, Pb and Zn respectively). The overall risk (risk defined by the grid cell having an RCR higher than one) (Hernout et al., 2013) posed by any one or more of the four metals studied was calculated as the percentage of the overall distribution of each species identified as “at risk” (Fig. 2). It was found that 11.0 and 10.5% of the distribution of M. daubentonii and M. mystacinus respectively were predicted to be at risk (from Pb, Cu and Cd; and Pb, Cu, Zn and Cd, respectively) (Fig. 2). Regarding the percentage of species' distribution at risk, contamination by Pb posed the greatest risk to all bats species with between 0.3 and 8.5% (B. barbastellus e M. daubentonii) of species' distribution determined by the model to be at risk. The next most important metal was Cu (0e4.5% (M. bechsteinii e M. nattereri), followed by Cd (0e2.3% ((B. barbastellus, E. serotinus, N. noctula, P. nathusii) e M. mystacinus)) and Zn (0e0.8% ((B. barbastellus, E. serotinus, M. daubentonii, N. noctula, P. nathusii, Pl. austriacus) e M. mystacinus)) (Fig. 2). 3.2. Identification of key drivers of risk A first evaluation of the effects of non-spatial input values on the model outcome was performed. This evaluation included the range of soil concentrations covering 95% of the data. For all the metals, the sensitivity analysis of the framework identified that the proportion of Coleoptera and Diptera in a bat diet were particularly important in determining risk to a bat species' (Fig. 3). The key drivers determining the risk of bats to Pb exposure were the proportion of Diptera in the diet (contributing to 44% of the RCR), followed by the amount of food eaten (19%) and the predicted safe daily dose (14%) (Fig. 3). For Cu, the proportion of Coleoptera and Araneidae were found to be similar in terms of importance, contributing 28% and 24% to the total effect, respectively (Fig. 3). The amount of food eaten and the proportion of Diptera were also important, both of these parameters contributing 15% of the total effect (Fig. 3). For Cd, the proportion of
212
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
Fig. 1. Distributions of risk characterisation ratios for a) Cd, b) Pb, c) Cu and d) Zn for 14 bat species across their spatial distribution: B.b.: Barbastella barbastellus, E.s.: Eptesicus serotinus, M.b.: Myotis bechsteinii, M.d.: Myotis daubentonii, M.m.: Myotis mystacinus, M.n.: Myotis nattereri, N.l.: Nyctalus leisleri, N.n.: Nyctalus noctula, P.n.: Pipistrellus nathusii, P.p.: Pipistrellus Pipistrellus/pygmaeus, Pl.aur.: Plecotus auritus, Pl. aus.: Plecotus austriacus, R.f.: Rhinolophus ferrumequinum, R.h.: Rhinolophus hipposideros. Black lines represent the threshold RCR of 1. The boxes indicate the median RCR and upper and lower quartiles. The whiskers extend to 1.5 of the interquartile range outside of which RCR values are represented by points. The y axis has been square root transformed.
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
213
Fig. 2. Percentage of area determined at risk from metals: Cd, Cu, Pb, Zn and from all the metals combined for the 14 species studied. The bat species are the following: B.b.: Barbastella barbastellus, E.s.: Eptesicus serotinus, M.b.: Myotis bechsteinii, M.d.: Myotis daubentonii, M.m.: Myotis mystacinus, M.n.: Myotis nattereri, N.l.: Nyctalus leisleri, N.n.: Nyctalus noctula, P.n.: Pipistrellus nathusii, P.p.: Pipistrellus Pipistrellus/pygmaeus, Pl.aur.: Plecotus auritus, Pl. aus.: Plecotus austriacus, R.f.: Rhinolophus ferrumequinum, R.h.: Rhinolophus hipposideros.
Araneidae was the most important factor (33%), followed by the proportion of Diptera (15%), the amount of food eaten (15%), the predicted safe daily dose (14%) and the proportion of Lepidoptera (14%) (Fig. 3). For Zn, the most important factor was the proportion of Coleoptera (27%), followed by the proportion of Diptera (19%), Trichoptera (14%) and the amount of food eaten (14%) (Fig. 3). The second evaluation covered 99% of the soil data and, as expected, the soil had a greater contribution to the risk than when 95% of the soil data were covered. However, there were slight differences between the two analyses and the soil concentration was not the most important parameter in both analyses (Fig. 3). According to our analysis, the effect of the distribution of the different bat species on the risk predictions would be minor (further details in Text S3). 3.3. Model evaluation Comparison of measured concentrations of metals in tissues of Pipistrellus sp., collected across England and Wales, with the model predictions showed that concentrations of Pb (in kidneys and liver),
Cu (in liver) and Zn (in kidneys) were significantly higher for the bats found in areas predicted “at risk” than for the bats found in areas predicted “not at risk” by the model (Fig. 4; Table S5). In other words, for three metals studied, the bats living in areas predicted “at risk” contained higher metal tissue residues than bats predicted “not at risk” by the model. 4. Discussion Bat species varied significantly in their predicted risk of exposure to metal contaminants (Figs. 1 and 2). Diet composition appears to be the most important factor in determining the risk with the proportion of Coleoptera and Diptera being the main drivers (Fig. 3). Two species sharing similar foraging styles, M. mystacinus and P. pipistrellus/pygmaeus, appear to have the greatest risks of exposure (based on RCR values). Coleoptera, Diptera and Lepidoptera are the most important food items in bat diets and the proportion of these insect orders in the overall diet varies greatly across the different bat diets (Table 1). The sensitivity analyses highlighted that these prey items are also
Fig. 3. Cumulative effect for the different parameters given by the sensitivity analyses results (in percentage). The model parameters studied are: the proportion of invertebrates in the diet for each invertebrate type (Opiliones, Araneidae, Hymenoptera, Lepidoptera, Trichoptera, Diptera, Coleoptera, Hemiptera, Dermaptera, Orthoptera), the safe daily dose (mg/g body weight/d), the amount of food eaten (g dry weight/g body weight/d) and the concentration of metal in the soil (mg/g dry weight) (considering 95% and 99% of the soil distribution).
214
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
the most important parameters in determining the risk. The BAF values for different orders can explain some of the results of our sensitivity analyses. For example, the model is not as sensitive to the proportion of Lepidoptera in the diet compared to the proportions of Diptera and Coleoptera. Lepidoptera generally have a lower BAF (median (range): 0.89 (0.01e19.30), 0.45 (0.01e1.14), 0.04 (0.01e5.62) and 0.14 (0.09e0.82) for Cd, Cu, Pb and Zn, respectively) compared to Diptera (0.74 (0.02e32.64), 0.81 (0.02e25.80), 0.21 (0.01e5.30) and 0.77 (0.08e18.81), for Cd, Cu, Pb and Zn, respectively) and Coleoptera (0.53 (0.04e18.21), and 1.00 (0.02e35.4), 0.04 (0.01e5.86) and 1.01 (0.02e26.53), for Cd, Cu, Pb and Zn, respectively) (Fig. 3, Table S4). The exception was for Cd, where a higher percentage total effect was observed for Lepidoptera (14%, against 11% for Coleoptera) (Fig. 3). This can be explained by the higher BAF for Cd for Lepidoptera compared to Coleoptera (Table S4). Bat distribution and foraging distance were not included in the sensitivity analyses. Since the metal soil concentration does not appear to be the most important parameter in our study (Fig. 3), we believe that the analyses performed can provide satisfactory information on the model sensitivity. Our results show that the proportion of food items consumed per bat species is also important in determining risk. Therefore, research that aims to improve and refine our understanding of wildlife diet composition, e.g. by using molecular, OMICs or stable isotope methodologies, are to be encouraged (Bohmann et al., 2011; Razgour et al., 2011). Indeed, the contributions of small and soft body items to the diet may be underestimated using traditional
methods of diet analysis (Vaughan, 1997; Zeale et al., 2011). Metal bioaccumulation in invertebrates also needs to be better understood, since bioaccumulation studies are often based only on a few species that are cultured experimentally. Metal uptake can also be influenced by insect species traits (e.g., life history, ecological, physiological, and morphological traits, etc.) (Rubach et al., 2012). Information on the metal bioaccessibility from invertebrates may help to refine exposure predictions (Hernout et al., 2015). Other factors may explain interspecific variations in the wild. Our model used an allometric relationship to derive sensitivity data for each species, which only accounts for size differences among bats. However, specific differences in metal sensitivity may occur due to differences in detoxification processes. For example, the regulation of oxidative stress after metal exposure and the activity of the antioxidant enzymes have been shown to vary across other wildlife species such as passerine birds (Rainio et al., 2013). It may be that interspecific differences occur in metal accumulation, detoxification and regulation across bat species, as seen in other small terrestrial mammal species (Fritsch et al., 2010). The accumulation in these small mammals also varied within a group having similar diet, suggesting that the traditional differentiation based on trophic group would need to be verified (Fritsch et al., 2010). The authors indicate these interspecific differences may be related to the physiological (e.g. metabolic rate, digestive characteristics influencing metal bioaccessibility, excretion rate) and behavioral characteristics (diet composition, habitat preferences, etc.) of the organism. Comparison of measured concentrations of metals in bat tissues
Fig. 4. Differences in tissue metal (A: Pb, B: Cu, C: Zn and D: Cd) concentrations (mg/g dw) between bats found in areas predicted “at risk” (grey) and areas predicted “not be to risk” (white). The bat tissues analyzed are kidneys and liver (n ¼ 191 for both organs).* Indicates a significant difference (P < 0.05) (ROC analyses). The y axis has been transformed with a root square transformation. The upper and the lower whisker extend from the hinge to the highest and the lowest values that are within: 1.5 times the inter-quartile range.
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
of Pipistrellus sp. with the risk predictions, obtained using our model, indicate that the simple model is able to partly distinguish between bats at risk from metal exposure from those not at risk, especially for Pb, Cu and Zn (Fig. 4; Table S5). There are, however, some differences between our residue level results and the model predictions. It is important to highlight that our model outputs determine a risk based on the level of oral exposure while our tissue data considers internal exposure. The differences observed can be explained by a number of aspects of the modeling approach including the level of spatial and temporal resolution of the available metal soil concentration data, the effects of soil parameters on uptake into food items, the spatial availability of prey items and the bioavailability of metals from food items into the bats (Hernout et al., 2011, 2013). The sensitivity data (i.e. ratio NOAEL by uncertainty factor) used in our model are derived from experimental studies based on other organisms (i.e. rat and mink), which have different energetic requirements and life-histories than bats (Table S3). It has been emphasized that simplified ecotoxicological descriptors, such as NOAEL, present some limitations and do not describe the exposure-response curve fundamental in ecotoxicology (Landis and Chapman, 2011). Further modeling studies should investigate the curve-fitting exposure-response of metal for wildlife species (Landis and Chapman, 2011). In addition, the modeling framework and the tissue data used different toxicological endpoints. Our model evaluation was based on a targeted dataset of metal tissue concentrations obtained for a single genus: Pipistrellus sp. Monitoring data on metal tissue concentrations measured in various European bat species (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007) indicate that some of the bat species predicted to be at higher risk in this study do contain higher metal concentrations in their tissues than species predicted to be at lower risk. For example, M. mystacinus contained Pb renal concentrations around twice the median values for the other species studied (E. serotinus, N. noctula, P. Nathusii) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). For Cu, N. noctula contained the highest renal concentrations (34 mg/g dw) among the other species (E. serotinus, M. mystacinus, and R. hipposideros) and was predicted to be the second species most at risk from Cu (based on median rcr) (after M. nattereri for which the concentration has not been determined) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). From the bats predicted to be the most exposed to Zn, M. mystacinus contained Zn concentrations (76.5 mg/g dw) higher than the mean (74 mg/g dw) obtained from all the species studied (E. serotinus, N. noctula, Pipistrellus sp. and R. hipposideros) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). For Cd, M. mystacinus, P. auritus and N. noctula appeared to be most exposed and contained higher or similar Cd concentrations in their kidneys (1.2, 0.8, 0.8 for M. mystacinus, Pl. auritus and N. noctula, respectively) than the mean (0.8 mg/g dw) for all the species studied (Pipistrellus sp., Pl. auritus, E. serotinus, M. mystacinus, N. noctula, R. hipposideros) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007). Although, the metal concentrations in kidneys have not been determined for all of the bat species (Carravieri and Scheifler, 2013; Walker et al., 2007) investigated in our study, it was possible to distinguish similar patterns between our predicted risks and the measured metal concentrations across several bat species. In particular, M. mystacinus needs to be highlighted, as it was predicted to be the most exposed species to all metals in our study and in monitoring studies has been shown to globally contain higher metal concentrations in kidneys compared to the other species investigated in the Carravieri and Scheifler (2013) study. Further studies are encouraged to determine actual metal concentrations in bats to potentially validate our predictions.
215
For the species predicted to be most exposed (Figs. 1 and 2), declines in populations have been recorded in the UK. For example, for the Pipistrellus sp. species, significant declines have been recorded in the 1980's. For M. mystacinus and E. serotinus, the declines have been more localized (e.g. in the southeast of England for E. serotinus) and may be associated with roost loss (Dietz et al., 2009). However, there is a lack of good monitoring data on bat populations. The IUCN Red List, 2008 states that the population trends remain unknown for N. noctula and P. nathusius. Therefore, we are, at the moment, not able to determine whether our risk predictions are directly linked with recent population declines. Although some metal emissions have significantly reduced in the UK, metals are often strongly retained in the soil so will potentially occur for hundreds of years after emissions have ceased. The threat to the environment and wildlife species, such as bats, of current emissions and further deposition therefore remains (Review of Transboundary Air Pollution, 2009).
5. Conclusions This study identified bat species at risk from metal contamination in England and Wales and interspecies variation in risk to metal exposure. Particular conservation and management actions could be orientated on bat species and areas identified at risk by the model, either in monitoring populations of especially vulnerable species (e.g. M. mystacinus) in contaminated areas, or providing additional support for their populations in areas where reproduction may be affected by metal exposure. The sensitivity analyses performed highlighted the importance of the diet composition, amount of food ingested and toxicity data for the target species. Further studies employing sensitivity analysis in food chain modeling are therefore encouraged as they can help in prioritizing research needs on key factors driving contamination risk. While comparing our model predictions with monitoring data, the modeling framework provided a good estimation of the range of bats predicted at risk. The existing framework is therefore a useful tool to predict risk of metals to various wildlife species and may assist future decision making processes.
Acknowledgments BH was financially supported by the European Union under the 7th Framework Programme (project acronym CREAM, contract number PITN-GA-2009-238148) and KEA by a Royal Society University Research Fellowship. Thanks to the Animal Health and Veterinary Laboratory Agency and the National Museum Liverpool for supplying the bat individuals. The UK bat lyssavirus surveillance program is funded by Defra (grant SV3500). Thanks to the National Soil Resources Institute, UK, for providing data on soil metal concentrations in England and Wales; the Environment Risk Assessment Team and the Heavy metals in food Team at FERA, Sand Hutton; the Bat Conservation Trust for providing data on bat distribution in England and Wales; P. Moeschler and C. Mougin from um d'histoire Naturelle de Gene ve, Centre de Coordinathe Muse tude et la protection des chauves-souris for tion Ouest pour l’e providing a large number of publications on bat ecology; J. Lewis, M. Andreas and T. Kervyn for their constructive advice.
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.envpol.2015.06.016.
216
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
References Berglund, A.M.M., Koivula, M.J., Eeva, T., 2011. Species and age-related variation in metal exposure and accumulation of two passerine bird species. Environ. Pollut. 159, 2368e2374. Bohmann, K., Monadjem, A., Lehmkuhl Noer, C., Rasmussen, M., Zeale, M.R.K., Clare, E., Jones, G., Willerslev, E., Gilbert, M.T.P., 2011. Molecular diet analysis of two African free-tailed bats (Molossidae) using high throughput sequencing. PLoS One 6 (6), e21441. res: Carravieri, A., Scheifler, R., 2013. Effets des substances chimiques sur les chiropte se bibliographique. Le Rhinolophe 19, 1e46. synthe Clark, D.R., Shore, R.F., 2001. Chiroptera. In: Shore, R.F., Rattner, B.A. (Eds.), Ecotoxicology of Wild Mammals. John Wiley & Sons, New York, pp. 159e215. Dietz, C., Von Helversen, O., Nill, D., 2009. Bats of Britain, Europe and Northwest Africa. A&C Black Publishers, London, UK. EFSA Journal, 2009. Risk Assessment for Birds and Mammals. Guidance of European Food Safety Authority, EFSA, Parma, Italy. Environment Agency, 2009. Dealing with Contaminated Land in England and Wales. A Review of Progress from 2000-2007 with Part 2A of the Environmental Protection Act, Bristol, UK. Frick, W.F., Reynolds, D.S., Kunz, T.H., 2010. Influence of climate and reproductive timing on demography of little brown myotis Myotis lucifugus. J. Animal Ecol. 79, 128e136. Fritsch, C., Cosson, R.P., Coeurdassier, M., Raoul, F., Giraudoux, P., Crini, N., de Vaufleury, A., Scheifler, R., 2010. Responses of wild small mammals to a pollution gradient: host factors influence metal and metallothionein levels. Environ. Pollut. 158, 827e840. Grimm, V., Railsback, S.F., 2005. Individual Based Modelling and Ecology. In: Princeton series in Theoretical and Computional Biology (New Jersey, USA). Health Canada Contaminated Sites Division, 2005. Procedures for the Use of Risk Assessment under Part XV.1 of the Environmental Protection Act. Ontario ministry of the environment, Canada. Hernout, B.V., Arnold, K.E., McClean, C.J., Grimm, V., Boxall, A.B.A., 2011. Predicting the threats of chemicals to wildlife: what are the challenges? Integr. Environ. Assess. Manag. 7 (3), 499e506. Hernout, B.V., Somerwill, K.E., Arnold, K.E., McClean, C.J., Boxall, A.B.A., 2013. A spatially-based modeling framework for assessing the risks of soil-associated metals to bats. Environ. Pollut. 173, 110e116. Hernout, B.V., Bowman, S.R., Weaver, R., Jayasinghe, C.J., Boxall, A.B.A., 2015. Implications of in vitro bioaccessibility differences for the assessment of risks of metals to bats. Environ. Toxicol. Chem. 34 (4), 898e906. Hoffman, D.J., Rattner, B.A., Scheunert, I., Korte, F., 2001. Environmental contaminants. In: Shore, R.F., Rattner, B.A. (Eds.), Ecotoxicology of Wild Mammals. John Wiley & Sons, New York, pp. 159e215. Hurley, S., Fenton, M.B., 1980. Ineffectiveness of fenthion, zinc phosphide, DDT and two ultrasonic rodent repellers for control of populations of little brown bats (Myotis lucifugus). Bull. Environ. Contam. Toxicol. 25, 503e507. IUCN Red List, 2008. The IUCN Red List of Threatened Species. UK: IUCN Global Species Programme Red List Unit [cited 24.01.2014]. Available from: http:// www.iucnredlist.org/. Jones, G., Jacobs, D.S., Kunz, T.H., Willig, M.R., Racey, P.A., 2009. Carpe noctem: the importance of bats as bioindicators. Endanger. Species Res. 8, 93e115. Kennedy, M.C., 2005. GEM-sa, Version 1.1. Software: Gaussian Emulation Machine
for Sensitivity Analysis. Available from: http://www.tonyohagan.co.uk/ academic/GEM/index.html. Landis, W.G., Chapman, P.M., 2011. Well past time to stop using NOELs and LOELs. Integr. Environ. Assess. Manag. 7, vieviii. Nagy, K.A., 1987. Field metabolic rate and food requirement scaling in mammals and birds. Ecol. Monogr. 57, 111e128. Parry, H.R., Topping, C.J., Kennedy, M.C., Boatman, N.D., Murray, A.W.A., 2013. A Bayesian sensitivity analysis applied to an Agent-based model of bird population response to landscape change. Environ. Model. Softw. 45, 104e115. Platts, P.J., McClean, C.J., Lovett, J.C., Marchant, R., 2008. Predicting tree distributions in East African biodiversity hotspot: model selection, data bias and envelope uncertainty. Ecol. Model. 218, 121e134. Rainio, M.J., Kanerva, M., Salminen, J.-P., Nikinmaa, M., Eeva, T., 2013. Oxidative status in nestlings of three small passerine species exposed to metal pollution. Sci. Total Environ. 454e455, 466e473. Razgour, O., Clare, E.L., Zeale, M.R.K., Hanmer, J., Baerholm Schnell, I., Rasmussen, M., Gilbert, T.P., Jones, G., 2011. High-throughput sequencing offers insight into mechanisms of resource partitioning in cryptic bat species. Ecol. Evol. 1 (4), 556e570. Risk assessment guidance for superfund, 2001. Process for Conducting Probabilistic Risk Assessment, Appendix a. Volume 3 Part a. Washington: Office of Emergency and Remedial Response. US Environmental Protection Agency [cited 24 January 2014]. Available from: http://www.epa.gov/oswer/riskassessment/ rags3adt/pdf/appendixa.pdf. Review of Transboundary Air Pollution (Rotap), 2009. “Heavy Metals” in Review of Transboundary Air Pollution. In: Acidification, Eutrophication, Ground Level Ozone and Heavy Metals in the UK. Available online: http://www.rotap.ceh.ac. uk. Rubach, M.N., Ashauer, R., Buchwalter, D.B., De Lange, H.J., Hamer, M., Preuss, T.G., €pke, K., Maund, S.J., 2012. A framework for traits-based assessment in ecoTo toxicology. Integr. Environ. Assess. Manag. 7 (2), 172e186. DOI 10.1002/ ieam.105. Sample, B.E., Opresko, D.M., Suter, G.W., 1996. Toxicological Benchmarks for Wildlife: 1996 Revision. Oak Ridge National Laboratories: Health Sciences Research Division, Oak Ridge. nchez-Chardi, A., Alberto Oliviera Ribeiro, C., Nadal, J., 2009. Metals in liver and Sa kidneys and the effects of chronic exposure to pyrite mine pollution in the ~ ana. Chemoshrew Crocidura russula inhabiting the protected wetland of Don sphere 76, 387e394. Schmolke, A., Thorbek, P., DeAngelis, D.L., Grimm, V., 2010. Ecological modelling supporting environmental decision making: a strategy for the future. Trends Ecol. Evol. 25, 479. Vaughan, N., 1997. The diets of British bats (Chiroptera). Mammal. Rev. 27 (2), 77e94. Walker, L.A., Simpson, V.R., Rockett, L., Wienburg, C.L., Shore, R.F., 2007. Heavy metal contamination in bats in Britain. Environ. Pollut. 148, 483e490. Wickramasinghe, L.P., Harris, S., Jones, G., Vaughan, N., 2003. Bat activity and species richness on organic and conventional farms: impact of agricultural intensification. J. Appl. Ecol. 40, 984e993. Zeale, M.R.K., Butlin, R.K., Barker, G.L.A., Lees, D.C., Jones, G., 2011. Taxon-specific PCR for DNA barcoding arthropod prey in bat faeces. Mol. Ecol. Resour. 11 (2), 236e244.
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Interspecies variation in the risks of metals to bats atrice V. Hernout a, b, *, Ste phane Pietravalle b, Kathryn E. Arnold a, Colin J. McClean a, Be b, c James Aegerter , Alistair B. A. Boxall a a
Environment Department, The University of York, Heslington, York, UK The Food and Environment Research Agency, Sand Hutton, York, UK c National Wildlife Management Centre, Animal Health and Veterinary Laboratories Agency, Sand Hutton, York, UK b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 30 March 2015 Received in revised form 31 May 2015 Accepted 2 June 2015 Available online xxx
A modeling framework was used to assess the risk of four metals to UK bat species. Eight species of bats were predicted to be “at risk” from one or more of the metals in over 5% of their ranges. Species differed significantly in their predicted risk. Contamination by Pb was found to pose the greatest risk, followed by Cu, Cd and Zn. A sensitivity analysis identified the proportion of invertebrates ingested as most important in determining the risk. We then compared the model predictions with a large dataset of metals concentrations in the tissues (liver, kidney) of Pipistrellus sp. from across England and Wales. Bats found in areas predicted to be the most “at risk” contained higher metal concentrations in their tissues than those found in areas predicted “not at risk” by the model. Our spatially explicit modeling framework provides a useful tool for further environmental risk assessment studies for wildlife species. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Insectivorous bats Spatially explicit model Trace metal elements Environmental risk assessment Model evaluation
1. Introduction Declines in many bat populations (e.g. Pipistrellus pipistrellus, Rhinolophus hipposideros, Rhinolophus ferrumequinum and Myotis myotis) have been observed across Europe (Jones et al., 2009; Dietz et al., 2009). Population declines can be the result of a number of factors including: environmental and climate change, changes in resources e.g. water, prey availability and quality, roost loss, disturbance, urbanization and industrialization, agricultural intensification, the increase in wind turbines, the pressure of disease and also exposure to chemicals in the environment (Frick et al., 2010; Jones et al., 2009; Walker et al., 2007; Wickramasinghe et al., 2003). Bats are long-lived mammals and consume a large amount of prey each night during their foraging period, and are thought to be particularly exposed to chemicals (Clark and Shore, 2001). Exposure to environmental contaminants, such as metals, may be considered as additional stressors to bats, although very few
* Corresponding author. Department of Environmental Toxicology The Institute of Environmental and Human Health (TIEHH), Texas Tech University, 1207 Gilbert Drive, Box 41163, TX 79409-3290, Lubbock, TX, USA. E-mail addresses: [email protected] (B.V. Hernout), Stephane. [email protected] (S. Pietravalle), [email protected] (K.E. Arnold), [email protected] (C.J. McClean), [email protected]. gov.uk (J. Aegerter), [email protected] (A.B.A. Boxall). http://dx.doi.org/10.1016/j.envpol.2015.06.016 0269-7491/© 2015 Elsevier Ltd. All rights reserved.
studies have considered the effects of metals on bat species (Clark and Shore, 2001). Environmental contamination by metal compounds has been widespread across Europe since the industrial revolution. In England and Wales, more than 80% of contaminated land sites have been reported to be contaminated by metals and metalloids (Environment Agency, 2009). In addition, as metals do not degrade, they are highly likely to accumulate in mammalian body tissues, especially for top predators and long-lived species such as bats (Dietz et al., 2009). Metals can elicit a range of toxic effects on wildlife, including induction of tremors, spasms, lethargy, lack of control in body movement, as well as sublethal effects at the biochemical, physiological and histological levels (e.g., oxidative stress, DNA damage, tissue damage including inclusion bodies), and, in some cases, can cause mortality (Clark and Shore, 2001; nchez-Chardi Hoffman et al., 2001; Hurley and Fenton, 1980; Sa et al., 2009). As flying mammals depend upon exceptional levels of motor-control and muscular activity, bats may show particular vulnerability to the physiological effects of exposure to metals. To explore the potential risks of chemical exposure to bats, we previously developed and applied a spatial modeling framework using a risk characterization approach, to assess the risks from soilassociated metals (cadmium, copper, lead and zinc) to the health of population of P. pipistrellus in England and Wales (Hernout et al., 2013). However, in our previous study, we only looked at one bat
210
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
species and the modeling framework was not evaluated against monitoring data. The exposure of different bat species to metals is likely to vary due to differences in factors such as their food intake, dietary composition and distribution. For example, bats specializing in consuming prey with a high metal accumulation capacity, which have a high food intake and a spatial range restricted to polluted areas, might be expected to have higher exposure than others. Studies monitoring metal residues in bats show that renal metal concentrations differ across bat species, which may reflect differences in dietary exposure (Walker et al., 2007). As observed for passerine birds, interspecific differences in metal exposure may be linked with their diet (Berglund et al., 2011). For example, it was shown that the pied flycatcher (Ficedula hypoleuca) accumulated more metals than great tits (Parus major) as the diet composition of pied flycatchers is composed of a large proportion of insects from higher trophic levels than the great tits (Berglund et al., 2011). When using modeling frameworks of the type described by Hernout et al. (2013), it is also important to understand the sensitivity of a framework to changes in model input parameters. This knowledge can be invaluable in informing the parameterisation process and guiding the model development. Sensitivity analyses are strongly recommended for use in Environmental risk assessment (ERA) (Schmolke et al., 2010), and emphasized by many institutions (e.g. EFSA Journal, 2009; Health Canada Contaminated Sites Division, 2005). However, the literature remains scarce. The analyses consist in examining how outputs vary as inputs are varied, to understand how the risk predictions are dependent on the variability and the uncertainty of the factors contributing to the risk (Grimm and Railsback, 2005; Risk assessment guidance for superfund, 2001). Complex approaches, as we have used, involve mathematical and statistical techniques and can include the effect of the combination of several factors having different statistical distributions (Risk Assessment Guidance for Superfund, 2001). Sensitivity analyses have been used previously in ecological modeling exercises such as an agent-based model, simulating skylark (Alauda arvensis) population response to landscape change (Parry et al., 2013). The most important parameters were identified for the model parameterization process or subsequent empirical studies. Model evaluation is also strongly recommended in modeling practise, although relatively scarce (Schmolke et al., 2010). In this study, to improve our knowledge of the potential threat of metal contamination to bats, we: (1) used the modeling framework which is based on a basic risk characterization approach (Hernout et al., 2013) to explore the risks of metal exposure for 14 bat species and identify the species most at risk from exposure to four metals; (2) used the modeling framework to determine the most important parameters affecting the predicted exposure, which are the main drivers in exposure risk, and to understand why certain species may be more vulnerable to metal exposure than others, and finally (3) compared levels of metals in different tissues (liver and kidney) of Pipistrellus sp. from across England and Wales (internal exposure) with our model predictions (based on oral exposure estimations) to evaluate our model. 2. Methods 2.1. Risk of UK bat species to metal exposure The modeling framework method described by Hernout et al. (2013) was applied to estimate the risks of four metals to 14 bat species present in the UK, namely: Barbastella barbastellus, Eptesicus serotinus, Myotis bechsteinii, Myotis daubentonii, Myotis mystacinus, Myotis nattereri, Nyctalus leisleri, Nyctalus noctula, Pipistrellus sp. (P. pipistrellus and Pipistrellus pygmaeus), Pipistrellus nathusii,
Plecotus auritus, Pl. austriacus, Rhinolophus ferrumequinum, R. hipposideros. The model used a risk characterization approach where the daily oral dose is compared with a ‘safe’ dose value to derive a ratio. The comparison of the ratio with a trigger value (of 1) indicates whether the risk is acceptable or not (using a resolution of 5 5 km2 cell) (Hernout et al., 2013). The modeling framework requires information on concentrations of metals in soils, soil-insect accumulation factors, bat diet, bat distribution and toxicity data on the metal studied. Concentrations of Cd, Cu, Pb and Zn in soil in England and Wales were obtained from NSRI (National Soil Resources Institute) at a 5 5 km2 resolution. Ecological data on bats (bat diet composition, foraging distance and weight) were gathered from the literature (Table S1, Table S2). Daily food intakes and no observed effect levels (NOAELs) were estimated based on the average bat weight for each species and were derived using the allometric relationships described in Nagy, 1987 and Sample et al., 1996, respectively (Table S2) (Hernout et al., 2013). The experimental studies used to derive the NOAELs considered a reproductive endpoint and chronic effects (Sample et al., 1996). Further details on the experimental studies are presented in Table S3. The NOAEL was divided by an uncertainty factor of five to calculate the “safe dose” (Hernout et al., 2013). Biota accumulation factor (BAF) data were obtained from the literature for each of the invertebrate orders listed in the bat diet for the four metals studied (Table S4). The bat distribution dataset (presence/absence data) was provided by the Bat Conservation Trust for each bat species (Data derived from the National Bat Monitoring Programme; NBMP). The spatial analysis was done using Geographic Information System (ArcGIS, ArcMap Version 9.3.1) (ESRI, Redlands Calif., USA). The final output was a risk characterization ratio (RCR) for each 5 5 km2 cell defined by the ratio between the daily dose of metal that a bat receives (mg/g body weight/d) and predicted safe daily dose for the metal (mg/g body weight/d), within the spatial distribution of the bat (Hernout et al., 2013). The percentage of areas at risk for each species and metal, as well as for the groups of metals combined were derived from the number of cells where a species was found to be at risk (i.e. with an RCR1) divided by the total number of cells in which the bat species is present (Hernout et al., 2013). 2.2. Identification of key drivers of risk A number of analyses were performed to identify the key factors that drive the risk of metals to bats as determined in the model. Distributions of selected model input parameters (Table 1) covering all species were used alongside the model to identify which of these were the most important in determining the risk values calculated. The Emulator GEM-SA 1.1 (Gaussian Emulation Machine for Sensitivity Analysis, Kennedy, 2005) was used to determine the effect of each individual input, or pairs of inputs on the output value. Further details on the emulation process are given is Text S2 (Supplemental data). The different input parameters and their respective ranges are shown in Table 1. The sensitivity analyses cannot integrate spatial components and therefore we did not include the spatial range in which the bat species reside. As each bat distribution is unique, the ranges of metal concentrations in soils will also vary for each bat species (Figure S1). Thus, the sensitivity analyses used a simplified generic distribution to represent all bats, representing the whole of England and Wales based on the totality of soil metal concentrations available for this area (see overall UK map soil concentrations in Hernout et al., 2013). The differences of RCRs across metals and bat species were tested using the non-parametric KruskaleWallis test. To explore
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216 Table 1 Input parameters and their ranges used in the sensitivity analysis to identify the key drivers of risks of soil-associated metals to bats. Parameters
Range used in the sensitivity analysis
Concentration of the metal in the soil (mg/g dry weight). Range limited to 95 and 99% of observations.
95% observations 99% observations Cd: 0e1.6 Cd: 0e3.2 Cu: 0e47.9 Cu: 0e100.3 Pb: 0e194.9 Pb: 0e514.6 Zn: 0e175.8 Zn: 0e350.8 Cd: 0.37-0.56 Cu: 5.86e8.85 Pb: 3.08e4.65 Zn: 61.59e93.03 Between bat species: 0.13-0.18
Safe daily dose for the metal (NOAEL/5) (mg/g body weight/d)
Amount of food eaten (g dry weight/g body weight/d) Proportion of the diet accounted for an individual prey item (% volume)
Coleoptera Diptera Lepidoptera Araneidea Dermaptera Hemiptera Hymenoptera Opiliones Orthoptera Trichoptera
0e1
0e0.25
the effects of species location (area in which the bat is living) and feeding range (area in which the bat is foraging), we compared the soil concentrations to which each species would be exposed using the non-parametric KruskaleWallis test. As multiple statistical tests were applied, the p-values were adjusted using the Holm method. The level of statistical significance was set to 0.05. We calculated the soil concentrations for the spatial distribution for each species using the method described by Hernout et al. (2013). This spatial application associating the bat distribution and its feeding range (based on the foraging distance of each species) does not take into account the diet or the food intake. Data analyses were performed with the software R version 2.12.1. 2.3. Model evaluation The model evaluation involved the comparison of our risk predictions (based on oral exposure estimations) with the measured internal concentrations of metal contained in bat tissues. Concentrations of metals in bat tissues (kidney and liver) were measured in 193 individuals of Pipistrellus sp. Bats were selected from a large archive of 3000 bats, using metal concentrations in soils from the locations at which the bats were found. We used the frequency distribution of soil metal concentrations across England and Wales to obtain our subsample of 193 bats reflecting that frequency distribution, which also includes bats from areas with high metal concentrations as well as areas wih low concentrations. The acquisition of the monitoring data on trace metal concentrations is detailed in SI (see Text S1: Sample collection and processing, Quantification of metal concentrations, Quality assurance and quality control, Data analyses). We compared tissue concentrations in two groups separated by their risk location status defined by the model (where bats are “at risk” and “not at risk”). The risk was defined for an area where bats have an RCR higher than one (Hernout et al., 2013). The hypothesis being that bat individuals found in areas predicted to be “at risk” would have higher internal concentrations of metals in their tissues than bats obtained from areas predicted to be “not at risk”. The Receiver-operating characteristic (ROC) was used to compare monitoring data with model predictions (Platts et al., 2008). The ROC curve provides a good visualization between high sensitivity and high specificity which can vary when discriminating two groups of data separated by a threshold value. Level of statistical significance was set to 0.05. Data analyses were performed with the software R
211
version 2.12.1. To compare our risk predictions across species against metal tissue concentrations in other bat species, we selected two publications studying several species for two areas (Austria and Britain, Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007; respectively). 3. Results 3.1. Risk of metal exposure to UK bat species RCRs for all species were calculated for all of the 5 5 km2 grid cells across which each species was distributed. Comparing across all species, median RCRs (range) were highest for Cu, followed by Pb, Cd and Zn: 0.30 (0.14e0.40), 0.23 (0.07e0.34), 0.16 (0.11e0.26) and 0.12 (0.03e0.20), respectively (Fig. 1). Within each bat species, the RCRs were significantly different across metals (KruskaleWallis c2 d.f ¼ 3 ¼ 1893, 3331, 251, 8790, 1092, 5841, 958, 5648, 193, 5383, 4582, 307, 510, and 2922, P < 0.05, for the bat species: B. barbastellus, E. serotinus, M. bechsteinii, M. daubentonii, M. mystacinus, M. nattereri, N. leisleri, N. noctula, P. nathusii, P. pipistrellus/pygmaeus, P. auritus, P. austriacus, R. ferrumequinum, and R. hipposideros, respectively) (Fig. 1). M. mystacinus and Pl. auritus appeared to be most exposed to Cd (Fig. 1a). M. daubentonii, M. mystacinus, Pipistrellus sp, P. Nathusii, N. leisleri, and N. noctula were most exposed to Pb (Fig. 1b). M. nattereri, N. noctula, M. daubentonii, P. nathusii, N. leisleri, E. serotinus, and M. mystacinus appeared to be most exposed to Cu (Fig. 1c). N. noctula, M. mystacinus, M. nattereri and M. daubentonii were predicted to be most exposed to Zn (Fig. 1d). The RCRs were significantly different across bat species for each metal (KruskaleWallis c2 h ¼ 3175, 5180, 7830, and 8273, d.f ¼ 14, 13, 13 and 13, P < 0.05, for Cd, Cu, Pb and Zn respectively). The overall risk (risk defined by the grid cell having an RCR higher than one) (Hernout et al., 2013) posed by any one or more of the four metals studied was calculated as the percentage of the overall distribution of each species identified as “at risk” (Fig. 2). It was found that 11.0 and 10.5% of the distribution of M. daubentonii and M. mystacinus respectively were predicted to be at risk (from Pb, Cu and Cd; and Pb, Cu, Zn and Cd, respectively) (Fig. 2). Regarding the percentage of species' distribution at risk, contamination by Pb posed the greatest risk to all bats species with between 0.3 and 8.5% (B. barbastellus e M. daubentonii) of species' distribution determined by the model to be at risk. The next most important metal was Cu (0e4.5% (M. bechsteinii e M. nattereri), followed by Cd (0e2.3% ((B. barbastellus, E. serotinus, N. noctula, P. nathusii) e M. mystacinus)) and Zn (0e0.8% ((B. barbastellus, E. serotinus, M. daubentonii, N. noctula, P. nathusii, Pl. austriacus) e M. mystacinus)) (Fig. 2). 3.2. Identification of key drivers of risk A first evaluation of the effects of non-spatial input values on the model outcome was performed. This evaluation included the range of soil concentrations covering 95% of the data. For all the metals, the sensitivity analysis of the framework identified that the proportion of Coleoptera and Diptera in a bat diet were particularly important in determining risk to a bat species' (Fig. 3). The key drivers determining the risk of bats to Pb exposure were the proportion of Diptera in the diet (contributing to 44% of the RCR), followed by the amount of food eaten (19%) and the predicted safe daily dose (14%) (Fig. 3). For Cu, the proportion of Coleoptera and Araneidae were found to be similar in terms of importance, contributing 28% and 24% to the total effect, respectively (Fig. 3). The amount of food eaten and the proportion of Diptera were also important, both of these parameters contributing 15% of the total effect (Fig. 3). For Cd, the proportion of
212
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
Fig. 1. Distributions of risk characterisation ratios for a) Cd, b) Pb, c) Cu and d) Zn for 14 bat species across their spatial distribution: B.b.: Barbastella barbastellus, E.s.: Eptesicus serotinus, M.b.: Myotis bechsteinii, M.d.: Myotis daubentonii, M.m.: Myotis mystacinus, M.n.: Myotis nattereri, N.l.: Nyctalus leisleri, N.n.: Nyctalus noctula, P.n.: Pipistrellus nathusii, P.p.: Pipistrellus Pipistrellus/pygmaeus, Pl.aur.: Plecotus auritus, Pl. aus.: Plecotus austriacus, R.f.: Rhinolophus ferrumequinum, R.h.: Rhinolophus hipposideros. Black lines represent the threshold RCR of 1. The boxes indicate the median RCR and upper and lower quartiles. The whiskers extend to 1.5 of the interquartile range outside of which RCR values are represented by points. The y axis has been square root transformed.
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
213
Fig. 2. Percentage of area determined at risk from metals: Cd, Cu, Pb, Zn and from all the metals combined for the 14 species studied. The bat species are the following: B.b.: Barbastella barbastellus, E.s.: Eptesicus serotinus, M.b.: Myotis bechsteinii, M.d.: Myotis daubentonii, M.m.: Myotis mystacinus, M.n.: Myotis nattereri, N.l.: Nyctalus leisleri, N.n.: Nyctalus noctula, P.n.: Pipistrellus nathusii, P.p.: Pipistrellus Pipistrellus/pygmaeus, Pl.aur.: Plecotus auritus, Pl. aus.: Plecotus austriacus, R.f.: Rhinolophus ferrumequinum, R.h.: Rhinolophus hipposideros.
Araneidae was the most important factor (33%), followed by the proportion of Diptera (15%), the amount of food eaten (15%), the predicted safe daily dose (14%) and the proportion of Lepidoptera (14%) (Fig. 3). For Zn, the most important factor was the proportion of Coleoptera (27%), followed by the proportion of Diptera (19%), Trichoptera (14%) and the amount of food eaten (14%) (Fig. 3). The second evaluation covered 99% of the soil data and, as expected, the soil had a greater contribution to the risk than when 95% of the soil data were covered. However, there were slight differences between the two analyses and the soil concentration was not the most important parameter in both analyses (Fig. 3). According to our analysis, the effect of the distribution of the different bat species on the risk predictions would be minor (further details in Text S3). 3.3. Model evaluation Comparison of measured concentrations of metals in tissues of Pipistrellus sp., collected across England and Wales, with the model predictions showed that concentrations of Pb (in kidneys and liver),
Cu (in liver) and Zn (in kidneys) were significantly higher for the bats found in areas predicted “at risk” than for the bats found in areas predicted “not at risk” by the model (Fig. 4; Table S5). In other words, for three metals studied, the bats living in areas predicted “at risk” contained higher metal tissue residues than bats predicted “not at risk” by the model. 4. Discussion Bat species varied significantly in their predicted risk of exposure to metal contaminants (Figs. 1 and 2). Diet composition appears to be the most important factor in determining the risk with the proportion of Coleoptera and Diptera being the main drivers (Fig. 3). Two species sharing similar foraging styles, M. mystacinus and P. pipistrellus/pygmaeus, appear to have the greatest risks of exposure (based on RCR values). Coleoptera, Diptera and Lepidoptera are the most important food items in bat diets and the proportion of these insect orders in the overall diet varies greatly across the different bat diets (Table 1). The sensitivity analyses highlighted that these prey items are also
Fig. 3. Cumulative effect for the different parameters given by the sensitivity analyses results (in percentage). The model parameters studied are: the proportion of invertebrates in the diet for each invertebrate type (Opiliones, Araneidae, Hymenoptera, Lepidoptera, Trichoptera, Diptera, Coleoptera, Hemiptera, Dermaptera, Orthoptera), the safe daily dose (mg/g body weight/d), the amount of food eaten (g dry weight/g body weight/d) and the concentration of metal in the soil (mg/g dry weight) (considering 95% and 99% of the soil distribution).
214
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
the most important parameters in determining the risk. The BAF values for different orders can explain some of the results of our sensitivity analyses. For example, the model is not as sensitive to the proportion of Lepidoptera in the diet compared to the proportions of Diptera and Coleoptera. Lepidoptera generally have a lower BAF (median (range): 0.89 (0.01e19.30), 0.45 (0.01e1.14), 0.04 (0.01e5.62) and 0.14 (0.09e0.82) for Cd, Cu, Pb and Zn, respectively) compared to Diptera (0.74 (0.02e32.64), 0.81 (0.02e25.80), 0.21 (0.01e5.30) and 0.77 (0.08e18.81), for Cd, Cu, Pb and Zn, respectively) and Coleoptera (0.53 (0.04e18.21), and 1.00 (0.02e35.4), 0.04 (0.01e5.86) and 1.01 (0.02e26.53), for Cd, Cu, Pb and Zn, respectively) (Fig. 3, Table S4). The exception was for Cd, where a higher percentage total effect was observed for Lepidoptera (14%, against 11% for Coleoptera) (Fig. 3). This can be explained by the higher BAF for Cd for Lepidoptera compared to Coleoptera (Table S4). Bat distribution and foraging distance were not included in the sensitivity analyses. Since the metal soil concentration does not appear to be the most important parameter in our study (Fig. 3), we believe that the analyses performed can provide satisfactory information on the model sensitivity. Our results show that the proportion of food items consumed per bat species is also important in determining risk. Therefore, research that aims to improve and refine our understanding of wildlife diet composition, e.g. by using molecular, OMICs or stable isotope methodologies, are to be encouraged (Bohmann et al., 2011; Razgour et al., 2011). Indeed, the contributions of small and soft body items to the diet may be underestimated using traditional
methods of diet analysis (Vaughan, 1997; Zeale et al., 2011). Metal bioaccumulation in invertebrates also needs to be better understood, since bioaccumulation studies are often based only on a few species that are cultured experimentally. Metal uptake can also be influenced by insect species traits (e.g., life history, ecological, physiological, and morphological traits, etc.) (Rubach et al., 2012). Information on the metal bioaccessibility from invertebrates may help to refine exposure predictions (Hernout et al., 2015). Other factors may explain interspecific variations in the wild. Our model used an allometric relationship to derive sensitivity data for each species, which only accounts for size differences among bats. However, specific differences in metal sensitivity may occur due to differences in detoxification processes. For example, the regulation of oxidative stress after metal exposure and the activity of the antioxidant enzymes have been shown to vary across other wildlife species such as passerine birds (Rainio et al., 2013). It may be that interspecific differences occur in metal accumulation, detoxification and regulation across bat species, as seen in other small terrestrial mammal species (Fritsch et al., 2010). The accumulation in these small mammals also varied within a group having similar diet, suggesting that the traditional differentiation based on trophic group would need to be verified (Fritsch et al., 2010). The authors indicate these interspecific differences may be related to the physiological (e.g. metabolic rate, digestive characteristics influencing metal bioaccessibility, excretion rate) and behavioral characteristics (diet composition, habitat preferences, etc.) of the organism. Comparison of measured concentrations of metals in bat tissues
Fig. 4. Differences in tissue metal (A: Pb, B: Cu, C: Zn and D: Cd) concentrations (mg/g dw) between bats found in areas predicted “at risk” (grey) and areas predicted “not be to risk” (white). The bat tissues analyzed are kidneys and liver (n ¼ 191 for both organs).* Indicates a significant difference (P < 0.05) (ROC analyses). The y axis has been transformed with a root square transformation. The upper and the lower whisker extend from the hinge to the highest and the lowest values that are within: 1.5 times the inter-quartile range.
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
of Pipistrellus sp. with the risk predictions, obtained using our model, indicate that the simple model is able to partly distinguish between bats at risk from metal exposure from those not at risk, especially for Pb, Cu and Zn (Fig. 4; Table S5). There are, however, some differences between our residue level results and the model predictions. It is important to highlight that our model outputs determine a risk based on the level of oral exposure while our tissue data considers internal exposure. The differences observed can be explained by a number of aspects of the modeling approach including the level of spatial and temporal resolution of the available metal soil concentration data, the effects of soil parameters on uptake into food items, the spatial availability of prey items and the bioavailability of metals from food items into the bats (Hernout et al., 2011, 2013). The sensitivity data (i.e. ratio NOAEL by uncertainty factor) used in our model are derived from experimental studies based on other organisms (i.e. rat and mink), which have different energetic requirements and life-histories than bats (Table S3). It has been emphasized that simplified ecotoxicological descriptors, such as NOAEL, present some limitations and do not describe the exposure-response curve fundamental in ecotoxicology (Landis and Chapman, 2011). Further modeling studies should investigate the curve-fitting exposure-response of metal for wildlife species (Landis and Chapman, 2011). In addition, the modeling framework and the tissue data used different toxicological endpoints. Our model evaluation was based on a targeted dataset of metal tissue concentrations obtained for a single genus: Pipistrellus sp. Monitoring data on metal tissue concentrations measured in various European bat species (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007) indicate that some of the bat species predicted to be at higher risk in this study do contain higher metal concentrations in their tissues than species predicted to be at lower risk. For example, M. mystacinus contained Pb renal concentrations around twice the median values for the other species studied (E. serotinus, N. noctula, P. Nathusii) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). For Cu, N. noctula contained the highest renal concentrations (34 mg/g dw) among the other species (E. serotinus, M. mystacinus, and R. hipposideros) and was predicted to be the second species most at risk from Cu (based on median rcr) (after M. nattereri for which the concentration has not been determined) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). From the bats predicted to be the most exposed to Zn, M. mystacinus contained Zn concentrations (76.5 mg/g dw) higher than the mean (74 mg/g dw) obtained from all the species studied (E. serotinus, N. noctula, Pipistrellus sp. and R. hipposideros) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013). For Cd, M. mystacinus, P. auritus and N. noctula appeared to be most exposed and contained higher or similar Cd concentrations in their kidneys (1.2, 0.8, 0.8 for M. mystacinus, Pl. auritus and N. noctula, respectively) than the mean (0.8 mg/g dw) for all the species studied (Pipistrellus sp., Pl. auritus, E. serotinus, M. mystacinus, N. noctula, R. hipposideros) (Lüftl et al., 2003 cited by Carravieri and Scheifler, 2013 and Walker et al., 2007). Although, the metal concentrations in kidneys have not been determined for all of the bat species (Carravieri and Scheifler, 2013; Walker et al., 2007) investigated in our study, it was possible to distinguish similar patterns between our predicted risks and the measured metal concentrations across several bat species. In particular, M. mystacinus needs to be highlighted, as it was predicted to be the most exposed species to all metals in our study and in monitoring studies has been shown to globally contain higher metal concentrations in kidneys compared to the other species investigated in the Carravieri and Scheifler (2013) study. Further studies are encouraged to determine actual metal concentrations in bats to potentially validate our predictions.
215
For the species predicted to be most exposed (Figs. 1 and 2), declines in populations have been recorded in the UK. For example, for the Pipistrellus sp. species, significant declines have been recorded in the 1980's. For M. mystacinus and E. serotinus, the declines have been more localized (e.g. in the southeast of England for E. serotinus) and may be associated with roost loss (Dietz et al., 2009). However, there is a lack of good monitoring data on bat populations. The IUCN Red List, 2008 states that the population trends remain unknown for N. noctula and P. nathusius. Therefore, we are, at the moment, not able to determine whether our risk predictions are directly linked with recent population declines. Although some metal emissions have significantly reduced in the UK, metals are often strongly retained in the soil so will potentially occur for hundreds of years after emissions have ceased. The threat to the environment and wildlife species, such as bats, of current emissions and further deposition therefore remains (Review of Transboundary Air Pollution, 2009).
5. Conclusions This study identified bat species at risk from metal contamination in England and Wales and interspecies variation in risk to metal exposure. Particular conservation and management actions could be orientated on bat species and areas identified at risk by the model, either in monitoring populations of especially vulnerable species (e.g. M. mystacinus) in contaminated areas, or providing additional support for their populations in areas where reproduction may be affected by metal exposure. The sensitivity analyses performed highlighted the importance of the diet composition, amount of food ingested and toxicity data for the target species. Further studies employing sensitivity analysis in food chain modeling are therefore encouraged as they can help in prioritizing research needs on key factors driving contamination risk. While comparing our model predictions with monitoring data, the modeling framework provided a good estimation of the range of bats predicted at risk. The existing framework is therefore a useful tool to predict risk of metals to various wildlife species and may assist future decision making processes.
Acknowledgments BH was financially supported by the European Union under the 7th Framework Programme (project acronym CREAM, contract number PITN-GA-2009-238148) and KEA by a Royal Society University Research Fellowship. Thanks to the Animal Health and Veterinary Laboratory Agency and the National Museum Liverpool for supplying the bat individuals. The UK bat lyssavirus surveillance program is funded by Defra (grant SV3500). Thanks to the National Soil Resources Institute, UK, for providing data on soil metal concentrations in England and Wales; the Environment Risk Assessment Team and the Heavy metals in food Team at FERA, Sand Hutton; the Bat Conservation Trust for providing data on bat distribution in England and Wales; P. Moeschler and C. Mougin from um d'histoire Naturelle de Gene ve, Centre de Coordinathe Muse tude et la protection des chauves-souris for tion Ouest pour l’e providing a large number of publications on bat ecology; J. Lewis, M. Andreas and T. Kervyn for their constructive advice.
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.envpol.2015.06.016.
216
B.V. Hernout et al. / Environmental Pollution 206 (2015) 209e216
References Berglund, A.M.M., Koivula, M.J., Eeva, T., 2011. Species and age-related variation in metal exposure and accumulation of two passerine bird species. Environ. Pollut. 159, 2368e2374. Bohmann, K., Monadjem, A., Lehmkuhl Noer, C., Rasmussen, M., Zeale, M.R.K., Clare, E., Jones, G., Willerslev, E., Gilbert, M.T.P., 2011. Molecular diet analysis of two African free-tailed bats (Molossidae) using high throughput sequencing. PLoS One 6 (6), e21441. res: Carravieri, A., Scheifler, R., 2013. Effets des substances chimiques sur les chiropte se bibliographique. Le Rhinolophe 19, 1e46. synthe Clark, D.R., Shore, R.F., 2001. Chiroptera. In: Shore, R.F., Rattner, B.A. (Eds.), Ecotoxicology of Wild Mammals. John Wiley & Sons, New York, pp. 159e215. Dietz, C., Von Helversen, O., Nill, D., 2009. Bats of Britain, Europe and Northwest Africa. A&C Black Publishers, London, UK. EFSA Journal, 2009. Risk Assessment for Birds and Mammals. Guidance of European Food Safety Authority, EFSA, Parma, Italy. Environment Agency, 2009. Dealing with Contaminated Land in England and Wales. A Review of Progress from 2000-2007 with Part 2A of the Environmental Protection Act, Bristol, UK. Frick, W.F., Reynolds, D.S., Kunz, T.H., 2010. Influence of climate and reproductive timing on demography of little brown myotis Myotis lucifugus. J. Animal Ecol. 79, 128e136. Fritsch, C., Cosson, R.P., Coeurdassier, M., Raoul, F., Giraudoux, P., Crini, N., de Vaufleury, A., Scheifler, R., 2010. Responses of wild small mammals to a pollution gradient: host factors influence metal and metallothionein levels. Environ. Pollut. 158, 827e840. Grimm, V., Railsback, S.F., 2005. Individual Based Modelling and Ecology. In: Princeton series in Theoretical and Computional Biology (New Jersey, USA). Health Canada Contaminated Sites Division, 2005. Procedures for the Use of Risk Assessment under Part XV.1 of the Environmental Protection Act. Ontario ministry of the environment, Canada. Hernout, B.V., Arnold, K.E., McClean, C.J., Grimm, V., Boxall, A.B.A., 2011. Predicting the threats of chemicals to wildlife: what are the challenges? Integr. Environ. Assess. Manag. 7 (3), 499e506. Hernout, B.V., Somerwill, K.E., Arnold, K.E., McClean, C.J., Boxall, A.B.A., 2013. A spatially-based modeling framework for assessing the risks of soil-associated metals to bats. Environ. Pollut. 173, 110e116. Hernout, B.V., Bowman, S.R., Weaver, R., Jayasinghe, C.J., Boxall, A.B.A., 2015. Implications of in vitro bioaccessibility differences for the assessment of risks of metals to bats. Environ. Toxicol. Chem. 34 (4), 898e906. Hoffman, D.J., Rattner, B.A., Scheunert, I., Korte, F., 2001. Environmental contaminants. In: Shore, R.F., Rattner, B.A. (Eds.), Ecotoxicology of Wild Mammals. John Wiley & Sons, New York, pp. 159e215. Hurley, S., Fenton, M.B., 1980. Ineffectiveness of fenthion, zinc phosphide, DDT and two ultrasonic rodent repellers for control of populations of little brown bats (Myotis lucifugus). Bull. Environ. Contam. Toxicol. 25, 503e507. IUCN Red List, 2008. The IUCN Red List of Threatened Species. UK: IUCN Global Species Programme Red List Unit [cited 24.01.2014]. Available from: http:// www.iucnredlist.org/. Jones, G., Jacobs, D.S., Kunz, T.H., Willig, M.R., Racey, P.A., 2009. Carpe noctem: the importance of bats as bioindicators. Endanger. Species Res. 8, 93e115. Kennedy, M.C., 2005. GEM-sa, Version 1.1. Software: Gaussian Emulation Machine
for Sensitivity Analysis. Available from: http://www.tonyohagan.co.uk/ academic/GEM/index.html. Landis, W.G., Chapman, P.M., 2011. Well past time to stop using NOELs and LOELs. Integr. Environ. Assess. Manag. 7, vieviii. Nagy, K.A., 1987. Field metabolic rate and food requirement scaling in mammals and birds. Ecol. Monogr. 57, 111e128. Parry, H.R., Topping, C.J., Kennedy, M.C., Boatman, N.D., Murray, A.W.A., 2013. A Bayesian sensitivity analysis applied to an Agent-based model of bird population response to landscape change. Environ. Model. Softw. 45, 104e115. Platts, P.J., McClean, C.J., Lovett, J.C., Marchant, R., 2008. Predicting tree distributions in East African biodiversity hotspot: model selection, data bias and envelope uncertainty. Ecol. Model. 218, 121e134. Rainio, M.J., Kanerva, M., Salminen, J.-P., Nikinmaa, M., Eeva, T., 2013. Oxidative status in nestlings of three small passerine species exposed to metal pollution. Sci. Total Environ. 454e455, 466e473. Razgour, O., Clare, E.L., Zeale, M.R.K., Hanmer, J., Baerholm Schnell, I., Rasmussen, M., Gilbert, T.P., Jones, G., 2011. High-throughput sequencing offers insight into mechanisms of resource partitioning in cryptic bat species. Ecol. Evol. 1 (4), 556e570. Risk assessment guidance for superfund, 2001. Process for Conducting Probabilistic Risk Assessment, Appendix a. Volume 3 Part a. Washington: Office of Emergency and Remedial Response. US Environmental Protection Agency [cited 24 January 2014]. Available from: http://www.epa.gov/oswer/riskassessment/ rags3adt/pdf/appendixa.pdf. Review of Transboundary Air Pollution (Rotap), 2009. “Heavy Metals” in Review of Transboundary Air Pollution. In: Acidification, Eutrophication, Ground Level Ozone and Heavy Metals in the UK. Available online: http://www.rotap.ceh.ac. uk. Rubach, M.N., Ashauer, R., Buchwalter, D.B., De Lange, H.J., Hamer, M., Preuss, T.G., €pke, K., Maund, S.J., 2012. A framework for traits-based assessment in ecoTo toxicology. Integr. Environ. Assess. Manag. 7 (2), 172e186. DOI 10.1002/ ieam.105. Sample, B.E., Opresko, D.M., Suter, G.W., 1996. Toxicological Benchmarks for Wildlife: 1996 Revision. Oak Ridge National Laboratories: Health Sciences Research Division, Oak Ridge. nchez-Chardi, A., Alberto Oliviera Ribeiro, C., Nadal, J., 2009. Metals in liver and Sa kidneys and the effects of chronic exposure to pyrite mine pollution in the ~ ana. Chemoshrew Crocidura russula inhabiting the protected wetland of Don sphere 76, 387e394. Schmolke, A., Thorbek, P., DeAngelis, D.L., Grimm, V., 2010. Ecological modelling supporting environmental decision making: a strategy for the future. Trends Ecol. Evol. 25, 479. Vaughan, N., 1997. The diets of British bats (Chiroptera). Mammal. Rev. 27 (2), 77e94. Walker, L.A., Simpson, V.R., Rockett, L., Wienburg, C.L., Shore, R.F., 2007. Heavy metal contamination in bats in Britain. Environ. Pollut. 148, 483e490. Wickramasinghe, L.P., Harris, S., Jones, G., Vaughan, N., 2003. Bat activity and species richness on organic and conventional farms: impact of agricultural intensification. J. Appl. Ecol. 40, 984e993. Zeale, M.R.K., Butlin, R.K., Barker, G.L.A., Lees, D.C., Jones, G., 2011. Taxon-specific PCR for DNA barcoding arthropod prey in bat faeces. Mol. Ecol. Resour. 11 (2), 236e244.