SEARCH

SEARCH BY CITATION

Keywords:

  • Amphibians;
  • climate change;
  • conservation biogeography;
  • conservation policy;
  • distribution forecast;
  • species distribution modelling;
  • taxonomy;
  • Triturus pygmaeus ;
  • Triturus marmoratus ;
  • uncertainty

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

Aim

To analyse a source of uncertainty when forecasting the possible effects of climate change on species distributions, which may appear when the species studied have doubtful taxonomy or are subdivided into subspecies with different environmental requirements.

Location

Mainland Spain.

Methods

Iberian newts (Triturus pygmaeus and Triturus marmoratus, a recently identified species) were used as examples. Environmental favourability models were performed on the occurrence of the newts according to two taxonomic options: (1) the two species separately; and (2) the two species together, as they were considered a single species by taxonomists until recently. The models were projected to three time periods between 2011 and 2100 within a context of climate change, using two different general circulation models and two emission scenarios. We calculated the discrepancy between forecasts produced with the different taxonomic options and their consistency under the same climate change scenario.

Results

The model based on the two species together did not distinguish between particular environmental requirements of either of the two species. Discrepancy values between taxonomic options increased over time. A reduction in areas favourable to T. pygmaeus and its north-eastward displacement were only predicted when this species was analysed separately. Nevertheless, the uncertainty derived from taxonomic ambiguity barely affected the predictions for T. marmoratus.

Main conclusions

Qualitatively and quantitatively different distribution forecasts for two newt species in mainland Spain were obtained depending on the taxonomic option considered. Taxonomic uncertainty also affected other sources of uncertainty. Some guidelines are suggested to aid in similar cases.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

Studies on the impact of climate change on species distributions are numerous (Skov & Svenning, 2004; Brooker et al., 2007; and references therein) and are affected by several sources of uncertainty (Carvalho et al., 2011; Beale & Lennon, 2012). For example, Real et al. (2010) described the uncertainty in distribution models associated with using different atmosphere–ocean general circulation models and Intergovernmental Panel on Climate Change (IPCC) SRES emission scenarios. Beale & Lennon (2012) identified observed distribution data as a relevant source of uncertainty in distribution modelling. Distribution data can be severely affected by taxonomic ambiguity, because the denomination and classification of organisms is subject to continuous change. Current taxonomic modifications are often associated with improvements in molecular and genetic techniques (Carranza & Amat, 2005), which is increasing the number of new species through the subdivision of former species (e.g. Carranza & Amat, 2005; Miralles et al., 2010). Furthermore, these changes involve modifications to species distributions, conservation status and, therefore, to conservation management plans (e.g. Real et al., 2005; Pearman et al., 2010). Taxonomic changes that result in the recognition of morphologically cryptic species are frequent among amphibians in Europe (e.g. Real et al., 2005; Carretero et al., 2011).

The pygmy newt, Triturus pygmaeus (Wolterstorff, 1905), and the marbled newt, Triturus marmoratus (Latreille, 1800), are taxa now treated as species (Dorda & Esteban, 1986; García-París et al., 1993) but were formerly treated as subspecies, T. marmoratus pygmaeus and T. m. marmoratus. Currently, T. pygmaeus is considered ‘Near Threatened’ by the IUCN, although Arntzen et al. (2006a) affirmed that, because of its current population decline, it could soon be placed in the ‘Vulnerable’ category. In fact, the pygmy newt is considered ‘Vulnerable’ in Spain (Pleguezuelos et al., 2004). However, T. marmoratus is of ‘Low Concern’ according to the IUCN, both at the global level and in our study area, with declining population trend (Arntzen et al., 2006b).

Amphibians are thought to be the most threatened vertebrate group (Houlahan et al., 2000; Hoffmann et al., 2010), and many authors consider that they are declining globally (e.g. Pounds et al., 2006). Climate warming has been suggested as a possible direct and indirect cause of amphibian decline (Stuart et al., 2004; Pounds et al., 2006; Alford et al., 2007) and some effort has been made to predict its effect on the distribution of amphibians (e.g. Parra-Olea et al., 2005), including in the Iberian Peninsula (e.g. Teixeira & Arntzen, 2002; Araújo et al., 2006; Carvalho et al., 2010).

Ecological modelling is a useful tool for predicting potential species distribution changes in the context of global warming and is helpful for conservation management (Hoffmann & Sgrò, 2011). This paper analyses the effects on distribution forecasting of using different taxonomic criteria, and proposes guidelines to deal with this kind of uncertainty. With this aim, we compare alternative distribution forecasts derived from climate change modelling according to two taxonomic options: one considering the two newts as different parapatric species, T. marmoratus and T. pygmaeus, and another considering them as the former T. marmoratus species (hereinafter Triturus spp.). We also discuss the implications of taxonomic-dependent differences in distribution forecasts in relation to future management plans.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

Study area and species selected

We studied the distribution of T. marmoratus and T. pygmaeus in mainland Spain. The two species overlap slightly in the Tagus River basin (Pleguezuelos et al., 2004; Espregueira Themudo et al., 2012), where hybridization events are believed to have occurred (García-París et al., 2001; Espregueira Themudo et al., 2012). Distribution records of the Portugal atlas of amphibians and reptiles (Loureiro et al., 2008) were excluded because they do not distinguish between T. marmoratus and T. pygmaeus. Also, Portugal and France were not included as part of our study area because of incompatibilities between the high-quality climatic predictions available for the different countries (see ‘Distribution modelling’ below). Thus, our distribution models locate the favourable environmental conditions for Triturus populations in mainland Spain.

Spain has a complex climate because of its orography and geographical situation between two continents (Europe and Africa) and two water masses – the Atlantic Ocean and Mediterranean Sea. The annual distribution of precipitation is highly heterogeneous (Font, 2000; Ninyerola et al., 2007). The Mediterranean part of the Iberian Peninsula is part of a biodiversity hotspot (Myers et al., 2000); 41% of the amphibian species are endemic and about 35% are threatened according to the IUCN (Pleguezuelos et al., 2004).

The distribution data for both species were taken from Pleguezuelos et al. (2004). Species distributions are represented on a 10 km × 10 km UTM grid (Fig. 1). The total number of cells studied was 5161; T. pygmaeus occurs in 468 cells and T. marmoratus in 1046 cells. The distribution of both species in Spain occupies 1505 cells, about 30% of the study area. The absence data set corresponded to all grid cells where the modelling taxa were not observed.

image

Figure 1. The study area in the European context. Current distributions of Triturus pygmaeus and T. marmoratus represented in 10 km × 10 km UTM cells within the study area of mainland Spain (data from Pleguezuelos et al., 2004).

Download figure to PowerPoint

Distribution modelling

Models forecasting future species distributions should include the impact of climate change and other factors that are not expected to change or whose change is not actually foreseeable (Márquez et al., 2011). We modelled the distribution of taxa using variables related to four explanatory factors: climate, spatial distribution, topography and human activity (see Appendix S1 in Supporting Information). We modelled the distribution of taxa considering two taxonomic options: one considering T. pygmaeus (Tp) and T. marmoratus (Tm) as parapatric species, and another considering them together as two subspecies of the former T. marmoratus species (T).

We took into account uncertainty in the climate predictions, and so climatic variables were used according to different socioeconomic scenarios and atmosphere–ocean general circulation models. We considered two IPCC SRES emission scenarios (Nakićenović et al., 2000) – A2 and B2 – which represent different socioeconomic scenarios but avoid extremes; and two atmosphere–ocean general circulation models – CGCM2 (Canadian Centre for Climate Modelling and Analysis) and ECHAM4 (Max Planck Institut für Meteorologie), which were regionalized to mainland Spain by the Spanish Meteorological Agency (AEMET) (Brunet et al., 2007). This improves quality in the climatic predictions with respect to the forecasts available from WorldClim (Hijmans et al., 2005), which used only 134 stations in Spain, whereas 373 stations for temperature and 2326 stations for rainfall were used for mainland Spain by the AEMET.

For each period 1961–1990, 2011–2040, 2041–2070 and 2071–2100, 12 climate variables (Appendix S1) were obtained. The variables regarding climate, spatial distribution, topography and human activity were obtained following Márquez et al. (2011). We included the square of elevation and slope in the topographical factor to evaluate a possible unimodal response of the species distribution to these variables. In addition, variables related to a polynomial trend-surface analysis (Legendre & Legendre, 1998) were considered in order to include the effect of factors that may involve purely spatial trends, such as history or spatial ecological dynamics (Legendre, 1993).

Present distribution models for each taxon were constructed using forward–backward stepwise logistic regression. These models were performed using separately each set of variables related to one of the following explanatory factors: spatial distribution, topography and human activity. For the climate factor, three models were constructed according to the climate variable values estimated for the period 1961–1990 using all possible combinations of the general circulation models and emission scenarios. ECHAM4 applied to the 1961–1990 period proposes the same precipitation and temperature values for both A2 and B2 scenarios, but these values differ in predictions for the 21st century (Márquez et al., 2011). Thus, we obtained three climate models: CGCM2-A2, CGCM2-B2 and ECHAM4-A2/B2. To avoid type I errors due to multiple tests (Benjamini & Hochberg, 1995), we controlled the false discovery rate (FDR) in every model using the procedure proposed by Benjamini & Hochberg (1995), entering in the model only variables that were significant under an FDR of < 0.05.

Other models that combined the four explanatory factors (hereinafter combined models) were built by performing a stepwise regression using the set of variables entered in the models based on a single factor. Thus, three combined models were obtained for each taxon (Tp, Tm and T) for the period 1961–1990, each one considering a different climate model. Finally, the favourability function (Real et al., 2006) was applied to the combined models. This function allows direct comparison of favourability values for species differing in their prevalence (Acevedo & Real, 2012).

Five criteria were used to assess the fit of the models: sensitivity; specificity; correct classification rate (CCR), based on the 0.5-favourability threshold – which, in the favourability function, makes probability be equal to overall prevalence; the area under the curve of the receiver operating characteristic (AUC), which is independent of any favourability threshold (Hosmer & Lemeshow, 2000); and the parsimony test based on the Akaike information criterion (AIC; Akaike, 1973). Finally, a model calibration test was applied (Landis & Koch, 1977) to test the extent to which the observed presences fitted those predicted by each model.

The combined models were projected into the future to obtain forecasts about variations in favourability for each taxonomic option. With this aim, the values of the climate variables in the models corrected by the favourability function were replaced by their corresponding future values.

For every favourability forecast (Ff) considering the nine combinations of general circulation models, emission scenarios and time periods for each of the three taxonomic groups, some fuzzy logic parameters were used to assess to what extent the initial favourability (F0, period 1961–1990) was modified because of climate change (Kuncheva, 2001):

  • display math

where I represents increment and M represents maintenance. In these equations, c(FX) is the cardinality of X favourability – where favourability is treated as a fuzzy set (Estrada et al., 2008) – that is, the sum of all the cells' favourability values (these are treated as degrees of membership in the fuzzy set); and the intersection between future and present favourability values is defined as follows:

  • display math

Positive values of increment (I) indicate a net increase in favourability for the taxon, that is, a gain in favourable areas, whereas negative values of I mean a net loss of favourable areas. Maintenance values (M) indicate the degree to which the favourable areas in F0 overlap with the favourable forecasted areas.

Assessment of uncertainty derived from taxonomic ambiguity

The uncertainty in forecasts regarding the taxonomic options was analysed by comparing the predicted distribution areas and the fit of the models based on Triturus spp. with the union of the predicted areas of occurrence for T. marmoratus and T. pygmaeus as separate species. The differences between the favourability forecasts for Triturus spp. (FT) and for the combination of both species (FTp ∪ FTm) were visualized using ArcGIS 9.3 (ESRI, Redlands, CA, USA) by mapping the degree of discrepancy (D) attributable to the taxonomic criterion:

  • display math

where inline image; = 0 indicates no uncertainty, D = +1 indicates maximum discrepancy when the predicted favourability for Triturus spp. is the highest, and D = −1 indicates maximum discrepancy when the predicted favourability for the united model for separate species is the highest. A total discrepancy (TD) value was quantified by adding all absolute discrepancies throughout the study area:

  • display math

where n = 5161, that is, the total number of cells in which the presence/absence of taxa has been recorded in the study area.

The uncertainty analysis regarding climate models and scenarios was performed to improve assessment of the importance of the uncertainty derived from taxonomic ambiguity. The uncertainty of models that forecast species distributions according to different emission scenarios should be reasonably low to be useful for policy planning. The consistency of forecasts derived from applying different methodological options to the same emission scenario should be higher than the coincidence between forecasts derived from applying the same methodological options to different scenarios (Real et al., 2010). Consistency, when using different circulation models, was formulated as follows:

  • display math

where c(FX) is the cardinality of X favourability (see above in the previous subsection) and FC and FE are the predicted favourability for taxa according to the circulation models CGCM2 or ECHAM4, respectively.

When using different taxonomic criteria, consistency was formulated as:

  • display math

Coincidence was mathematically calculated as:

  • display math

where, for a given taxonomic criterion and circulation model, FA2 and FB2 are the favourability for taxa predicted using emission scenarios A2 or B2, respectively. Consistencies and coincidences were compared using the Student's t-test.

Assessment of the relative importance of the explanatory factors

Some of the influence of the environment on species distributions could be simultaneously attributed to climate and other factors as a result of correlations between variables (Márquez et al., 2011). A variation partitioning procedure (Muñoz et al., 2005) was used to determine which proportion (p) of the variation in the favourability models was explained by the pure effect of climate (pPureCl), by non-climate factors (pPureNonCl), and by the intersection of climate with the other factors (pInt). All statistical analysis were performed using IBM SPSS statistics 19.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

The variables entered in each of the three combined models are presented in Appendix S2. Variables related to the four environmental factors were entered in the models of T. pygmaeus (Tp) and of the Triturus spp. (T), whereas only topographic variables did not enter in the model of T. marmoratus (Tm). Precipitation variables were entered in all the models and temperature variables were included in all the models except in the models for T. pygmaeus based on CGCM2. Both elevation and the degree of southern exposure were entered in all the models of Tp while slope was entered in all the models of T. The distance to urban centres was entered in all the models of Tm and T, and both the distance to highways and population density were entered in the models of T. With regard to the spatial factor, every model displayed a different and complex combination of variables. The highest proportion of explained variation in most of the models is attributable to the intersection between climate and non-climate factors (Table 1). The pure effect of climate explained hardly any spatial variation of favourability in the models of Tm and of T. In contrast, the areas favourable to Tp seem to be conditioned by the interaction between climatic and non-climatic factors.

Table 1. Results of the variation partitioning of the favourability models for Triturus pygmaeus (Tp), Triturus marmoratus (Tm) and Triturus spp. (T) in mainland Spain, based on climatic variables according to the combination of different atmosphere–ocean general circulation models (CGCM2, ECHAM4) and IPCC CO2 emission scenarios (A2, B2). Values indicate which percentage of the variation in the favourability models was explained by the pure effect of climate (pPureCl), by the pure effect of non-climatic factors (pPureNonCl), and by the intersection of climate with other factors (pInt)
Climate modelsPartitioningTpTmT
CGCM2-A2 p PureCl 31.01.40.0
p PureNonCl 11.937.819.3
p Int 57.160.880.7
CGCM2-B2 p PureCl 31.21.68.2
p PureNonCl 12.035.256.5
p Int 56.863.235.3
ECHAM4-A2/B2 p PureCl 2.18.90.0
p PureNonCl 37.721.440.5
p Int 60.269.759.5

The predicted favourability values are shown, respectively, in Fig. 2 for Tp, Tm and the T in mainland Spain according to the different climate models, and for each of the time periods considered. The forecasts show that, over time, the current latitudinal ranges for Tp and Tm are approximately maintained. Increment (I) and maintenance (M) values are shown in the lower right corner of each map (Fig. 2). Both I and M experienced a decreasing trend in the models of Tp, and an increasing trend in the models of Tm and T. The forecasts based on CGCM2 for Tp show an increase in favourability linked to an easterly displacement of the favourable areas, whereas the forecasts based on ECHAM4 show a decrease in favourable areas for the species (Fig. 2a). However, the forecasts for Tm show varied spatial trends (Fig. 2b). For T, a general increase in favourability was predicted, which did not involve geographical displacement of the current favourable areas (Fig. 2c).

image

Figure 2. Favourability in mainland Spain predicted for (a) Triturus pygmaeus and (b) T. marmoratus, as separate species, and (c) for both Triturus taxa as belonging to the same species, respectively, according to each climate model and for each time period – ranging from 0 (low favourability) to 1 (high favourability). Increment (I) and maintenance (M) values are shown at the lower right corner of each map.

Download figure to PowerPoint

The model-fit assessment parameters provided acceptable scores for T and for the species separately (Appendix S3). Ranging from 0 to 1, sensitivity was always higher than 0.8, specificity was higher than 0.5, the CCR was always higher than 0.6. The AUC (ranging from 0.5 to 1) was always higher than 0.8 (i.e. between excellent and outstanding according to Hosmer & Lemeshow, 2000). The AIC values ranged from 1876.04 to 4923.28, and the calibration ranged from 16.46 to 482.52. Tp and Tm showed similar values with regard to sensibility, specificity, CCR and AUC. The AIC showed higher values in the models for Tp, and in contrast, the model calibration test was better for Tp according to CGCM2 but was better for Tm according to ECHAM4. Comparing the united models of the two separate species (Tp ∪ Tm) with the model of the taxa as belonging to the same species (T), sensitivity was higher whereas specificity and CCR were lower (Appendix S3); the AUC and AIC reached similar values; and the calibration tests were either better or worse depending on the circulation model and emission scenario considered.

The models run with T showed a higher average sensitivity in relation to the distribution that had the highest geographical prevalence: the average sensitivity of these models was 0.613 with regard to the distribution of Tp (prevalence = 0.091), whereas it was 0.926 with regard to that of Tm (prevalence = 0.202).

Total discrepancies between forecasts generally increase over time for all climate models (Fig. 3). The geographical distribution of discrepancies is variable depending on the climate model considered.

image

Figure 3. Discrepancy values (difference between favourability predicted for both Triturus taxa as belonging to the same species and favourability predicted for the combination of T. pygmaeus and T. marmoratus as separate species) in mainland Spain according to each climate model and for each time period. When favourability was maximum for the Triturus spp. and minimum for the combination of the species, discrepancy was +1 (shown in red); in the opposite case, discrepancy was −1 (shown in blue). Total discrepancies (sum of all absolute discrepancy values, TD) are shown at the lower right corner of each map.

Download figure to PowerPoint

The forecasts for a single general circulation model based on different taxonomic criteria provided consistency values in the same range as the coincidence values (t = 0.119, n = 12, P > 0.05; see Table 2a). Coincidences when both taxa were modelled considering them as being a single species (ranging from almost 0.5 to about 0.9) were lower than coincidences when the two taxa were treated separately (around 0.9; t = 2.586, n = 12, < 0.05; see Table 2a). If consistencies based on taxonomic options (Table 2a) are compared with consistencies based on climatic models (Table 2b), both consistencies showed similar values (t = 1.094, n = 12, P > 0.05). Finally, the forecasts for a single taxonomic option based on different general circulation models also provided similar consistency and coincidence values (t = 0.928, n = 12, P > 0.05; see Table 2b), consistencies being lower than coincidences only when the two species were modelled separately (t = 2.425, n = 12, P < 0.05; see Table 2b).

Table 2. Assessment of uncertainty derived from taxonomic ambiguity (a) and from the availability of different climate models (b) in the favourability models for Triturus pygmaeus (Tp), Triturus marmoratus (Tm) and Triturus spp. (T) in mainland Spain. Consistencies measure the agreement between forecasts derived from different criteria (either taxonomic or climatic) assuming the same IPCC CO2 emission scenario; coincidences measure the agreement between forecasts derived from different emission scenarios using the same circulation model and taxonomic option
MethodPeriodConsistencyCoincidence
(a)A2B2Tp ∪ TmT
 CGCM22011–20400.890.590.990.66
2041–20700.940.690.970.67
2071–21000.860.810.700.49
Average0.900.700.880.61
 ECHAM42011–20400.910.930.870.89
2041–20700.540.710.910.83
2071–21000.960.810.890.68
Average0.800.810.890.80
Total average0.850.760.890.70
 (b)A2B2CGCM2ECHAM4
 Tp ∪ Tm2011–20400.740.860.990.87
2041–20700.570.600.970.92
2071–21000.700.880.700.89
Average0.700.780.880.89
 T2011–20400.910.640.660.89
2041–20700.990.810.670.83
2071–21000.570.580.490.68
Average0.820.680.610.80
Total average0.750.730.740.84

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

Qualitative model comparison based on environmental variables

Differences between the sets of variables that were included in the separate models for T. pygmaeus and T. marmoratus (Appendix S2) support the fact that, despite both species being morphologically similar, they are well differentiated regarding their environmental requirements in Spain (Pleguezuelos et al., 2004) and also along their entire distribution ranges (Vieites et al., 2009). This is consistent with the existence of genetic divergence between the species (García-París et al., 2001; Espregueira Themudo & Arntzen, 2007) and with their long history as separate populations since 3.2 Ma, according to García-París et al. (2001) or since 1.7 Ma, according to Carranza & Amat (2005). Thus, our results are consistent with the ecological diversification hypothesis (Pfenninger et al., 2003), which states that environmental requirements diverge even in parapatry. This seems to contradict the niche conservatism hypothesis (Kozak & Wiens, 2006), according to which sister species do not diverge much in environmental requirements. The different responses to the environment shown by the two species eventually led us to find differences in the forecasts derived from separate models compared with the forecasts derived from a common model.

Precipitation variables were more frequently important than temperature variables in the combined models of T. pygmaeus, whereas the opposite was observed in the case of T. marmoratus. This might be explained by the fact that T. pygmaeus is found in the southern half of the Iberian Peninsula, where low precipitation could be a limiting factor for amphibians (Pleguezuelos et al., 2004). Atmospheric humidity was recognized by Rodríguez-Jiménez (1988) as an important variable for T. pygmaeus. Notwithstanding the unequal geographical distribution of precipitation throughout the Iberian Peninsula, the spatial factor itself was shown to have some explanatory power for the biased distributions of these two species, probably indicating historical reasons behind the current ranges (Carranza & Amat, 2005). The topographic variables (elevation, slope and amount of southern exposure) entered in the model of T. pygmaeus, but not in the model of T. marmoratus. Thus, high elevations, slopes and exposure to the south were able to explain the presence of T. pygmaeus in some zones where neither climate nor space was able to distinguish the favourable areas for this species. The models for T. marmoratus did not find support in topographic variables, probably because of the wide elevational range shown by this species (between 0 and 2100 m; see García-París et al., 2004). Finally, human variables were not highly important, but indicated, for both species, a positive relationship with closeness to medium-sized towns. This might result from possible biases caused by higher sampling efforts in the most accessible sites for herpetologists (Sutherland, 2000). Highly populated cities were, on the contrary, negatively related to the distribution of T. marmoratus. The variables that were recurrently entered in our models for the two species include those detected by Carvalho et al. (2011) as important predictors for Triturus spp.

The environmental variables that were entered in the model based on Triturus spp. were a combination of those included in the models of the separate species. Thus, the information on the environmental requirements of newts in the studied area obtained by both taxonomic options was coherent. However, the model based on Triturus spp. was logically unable to distinguish between particular environmental requirements of either of the two species.

Forecast comparison based on the geographical trends of favourability

Several authors have predicted a general decrease in the distributions of amphibians as a result of climate change (Stuart et al., 2004; Pounds et al., 2006; Alford et al., 2007). According to our results, most models predicted a general decrease in areas favourable to T. pygmaeus in Spain – at the end of the analysed period, the increase in parameter I was negative in the four forecasts, and lower than −0.3 in three of them (see Fig. 2a). All the forecasts based on CGCM2 showed a north-eastward displacement of the favourable areas – the maintenance parameter M decreased even in those cases in which I remained close to 0 (see Fig. 2a). This does not imply that changes in the distribution areas will occur within the corresponding time periods, because species could respond to changes by adapting to the new conditions (Hoffmann & Sgrò, 2011). If species move to new favourable areas instead, their displacement could also be limited by the effect of physical barriers (Rey Benayas et al., 2006) and by the lower dispersal speed of species compared with the speed of climate change (Early & Sax, 2011).

The forecasts for T. marmoratus showed slight fluctuations of the favourable areas in interim periods but showed a net increase – I was positive and higher than 0.4 in three out of the four forecasts performed for the end of the analysed period, and it was only −0.1 in the unique forecast with negative I (see Fig. 2b). Some authors, such as Stuart et al. (2004), have described only negative effects of future climate changes on the suitable areas for amphibians; our results, as well as those of Araújo et al. (2006), also detected potentially positive and neutral effects.

Our predictions are different from those of Carvalho et al. (2010) and Pearman et al. (2010), as these authors forecasted a range expansion for T. pygmaeus and a range contraction for T. marmoratus as a consequence of climate change. The entire range of T. marmoratus was not taken into account in our models, and this might cause a bias in the results with regard to the importance of variables. Thus, the discrepancies between our results and those of the cited authors could be due a priori to differences in the study area, but also to differences in the climatic predictions or in the set of variables considered. Nevertheless, the study area is probably not the cause of these differences because Carvalho et al. (2010) considered the same area as we did. Climatic uncertainty caused by the use of different climatic predictions is a recognized source of differences between forecasts (Real et al., 2010). We used the same emission scenarios as Carvalho et al. (2010) and Pearman et al. (2010), but not the same circulation models, which has crucial effects on predictions (Real et al., 2010). Another reason for the differences is the set of variables considered. We used variables that, compared with WorldClim (which was the source used by the cited authors), were based on 3 and 17 times more meteorological stations for temperature and rainfall, respectively (Brunet et al., 2007). In addition, our models were the only ones including factors other than climate. The contribution of climate ranged from only 2.1% – the pure effect of climate in the ECHAM4 models for T. pygmaeus – to 88.1% – the sum of the pure effect and the undistinguished effect of climate in the CGCM2-A2 model for T. pygmaeus (see Table 1). Thus, not all the distribution of the species was considered to be driven by climate. This should be taken into account for attenuating the forecasted response of species to climate change (Real et al., 2013). The importance of using factors other than climate for predicting the consequences of climate change on distributions is widely supported in the literature (e.g. Thuiller et al., 2004; Márquez et al., 2011; Real et al., 2013).

Assessment of the uncertainty derived from taxonomic ambiguity

The prediction trends for Triturus spp. were generally similar to those of T. marmoratus within the distribution area of this species and so discrepancy between both models was low (Fig. 2c, Table 2). However, the decreasing spatial trend of favourability observed in the distribution area of T. pygmaeus was not found in the forecasts for Triturus spp., just the opposite, and so discrepancy was high in its distribution area (Fig. 3). Discrepancy values between taxonomic options were rarely higher than 1000 in the models of the initial period (1961–1990), but they increased to around 1300 over time. Thus, in our case study, the uncertainty derived from taxonomic ambiguity clearly affected the predictions for T. pygmaeus, whereas it hardly affected those for T. marmoratus.

Some degree of uncertainty caused by taxonomic ambiguity is also suggested by comparing the consistency of forecasts in which different taxonomic options are considered for the same scenario and the coincidence between forecasts for different emission scenarios. It is reasonable to expect consistency to be greater than coincidence in the absence of uncertainty (Real et al., 2010). However, consistencies in our results are similar to coincidences. Nevertheless, these differences were lower than those detected by Real et al. (2010) in an analysis of the uncertainty caused by differences in circulation models, where the consistencies were lower than coincidences. The uncertainty caused by taxonomic ambiguity also had an unexpected effect: it affected the degree of uncertainty caused by the general circulation models. This was detected in the fact that forecasts based on different circulation models showed higher coincidences with one taxonomic option than with the other one, and in the fact that the consistencies in forecasts based on different circulation models were lower than coincidences only in one of the taxonomic options.

Relevance to conservation of taxonomic uncertainty

To summarize, taxonomic uncertainty did not simply affect other sources of uncertainty, but Iberian newts obtained qualitatively different forecasts about their future distribution depending on the taxonomic option considered. These differences were relevant, at least in the case of T. pygmaeus. The reduction in areas favourable to this species and its north-eastward displacement forecasted by CGCM2 models, predicted when the species was analysed separately, were not reflected in the forecasts derived from the analysis of Triturus spp. This could be a consequence of the greater relevance given to the climate factor by the models for this species separately, compared with the models for the Triturus spp. (see Table 1 and Fig. 2a,c). Therefore, the consequences of choosing one taxonomic option over the other are important for conservation. The probable higher vulnerability of T. pygmaeus to climate change in the studied area became apparent only when the species was analysed separately, which unveils an additional impact of taxonomic ambiguity on conservation with respect to the impacts described by Morrison et al. (2009). Conversely, the forecasts made for both Triturus spp. and T. marmoratus showed higher coherence. This might be due to the fact that T. marmoratus represents a higher spatial proportion of Triturus spp.; that is, the prevalence of T. marmoratus is higher compared with that of T. pygmaeus. Actually, the models run with Triturus spp. showed a higher average sensitivity in relation to the distribution of T. marmoratus than to the distribution of T. pygmaeus.

In our study case, the uncertainty caused by taxonomy was derived from possible doubts about the taxonomic category in which different animal populations could be included. Although this issue was resolved for Triturus by García-París et al. (1993), this question remains open today with regard to numerous amphibian species, for example Salamandra salamandra/longirostris (Escoriza et al., 2006), Calotriton asper/arnoldi (Carranza & Amat, 2005), and Discoglossus galganoi/jeanneae (Velo-Antón et al., 2008). However, the type of uncertainty studied in this paper is not only relevant when analysing groups of doubtful taxonomy, but it may also have consequences when the studied species contains subspecies whose environmental requirements differ, for example Capra pyrenaica victoriae and C. p. hispanica (Acevedo & Real, 2011), and Icterus spurius subspecies (Martin & Omland, 2011). A single common model may not be able to reflect these types of differences over their entire distribution range. Subspecies with different environmental requirements are sometimes classified into different threat categories, which are often higher than those for the entire species, for example Salamandra salamandra longirostris and Triturus helveticus punctillatus (Pleguezuelos et al., 2004). According to our results, a reasonable way to proceed could be to construct a set of models taking into consideration all possible taxonomic options and to analyse the differences that may appear between forecasts. The first step of the analysis could be to determine whether any uncertainty exists arising from taxonomy. If so, then it would be necessary to assume a range of possible variation in the forecasts that should be taken into account in conservation policies. Even so, some of the methodological approaches applied in the present study could be used to decide which taxonomic options generated the most reliable forecast. The taxonomic option that generated the models with the best values according to a series of evaluation indices could be selected. If, as occurred in the present study, different indices positively valued the alternative models similarly (or if they did not point towards a best model on the basis of a consensus), there are still ways to decide which is the best model to retain as the basis for forecasts, for example: (1) to choose the taxonomic option that minimizes other sources of uncertainty, such as the one derived from the existing variety of global climate models; and/or (2) to choose the one that maximizes the degree of sensitivity regarding the distribution of each of the population units analysed.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

D. Romero was supported by a grant from the Ministerio de Educación: AP2007-03633. This study was also supported by project CGL2009-11316 (Ministerio de Ciencia e Innovación, Spain, and FEDER). We thank S. Coxon for his help in the English revision of the manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information
  • Acevedo, P. & Real, R. (2011) Biogeographical differences between the two Capra pyrenaica subspecies, C. p. victoriae and C. p. hispanica, inhabiting the Iberian Peninsula: implications for conservation. Ecological Modelling, 222, 814823.
  • Acevedo, P. & Real, R. (2012) Favourability: concept, distinctive characteristics and potential usefulness. Naturwissenschaften, 99, 515522.
  • Akaike, H. (1973) Information theory and an extension of the maximum likelihood principle. Proceedings of the Second International Symposium on Information Theory (ed. by B.N. Petrov and F. Csaki), pp. 267281. Akadémia Kiadó, Budapest.
  • Alford, R.A., Bradfield, K.S. & Richards, S.J. (2007) Global warming and amphibian losses. Nature, 447, E3E4.
  • Araújo, M.B., Thuiller, W. & Pearson, R.G. (2006) Climate warming and the decline of amphibians and reptiles in Europe. Journal of Biogeography, 33, 17121728.
  • Arntzen, J.W., Beja, P., Bosch, J., Tejedo, M., Lizana, M., Jehle, R., Martínez-Solano, I., Salvador, A., García-París, M., Gil, E.R., Marquez, R., Sá-Sousa, P. & Diaz Paniagua, C. (2006a) Triturus pygmaeus. IUCN 2008. 2008 IUCN Red List of Threatened Species. Available at: http://www.iucnredlist.org (accessed 1 July 2012).
  • Arntzen, J.W., Jehle, R., Bosch, J., Miaud, C., Tejedo, M., Lizana, M., Martínez-Solano, I., Salvador, A., García-París, M., Gil, E.R., Sá-Sousa, P. & Marquez, R. (2006b) Triturus marmoratus. IUCN 2008. 2008 IUCN Red List of Threatened Species. Available at: http://www.iucnredlist.org (accessed 1 July 2012).
  • Beale, C.M. & Lennon, J.J. (2012) Incorporating uncertainty in predictive species distribution modelling. Philosophical Transactions of the Royal Society B: Biological Sciences, 367, 247258.
  • Benjamini, Y. & Hochberg, Y. (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological), 57, 289300.
  • Brooker, R.W., Travis, J.M., Clark, E.J. & Dytham, C. (2007) Modelling species' range shifts in a changing climate: the impacts of biotic interactions, dispersal distance and the rate of climate change. Journal of Theoretical Biology, 245, 5965.
  • Brunet, M., Jones, P.D., Sigró, J., Saladié, O., Aguilar, E., Moberg, A., Della-Marta, P.M., Lister, D., Walther, A. & López, D. (2007) Temporal and spatial temperature variability and change over Spain during 1850–2005. Journal of Geophysical Research Atmospheres, 112, 19842012.
  • Carranza, S. & Amat, F. (2005) Taxonomy, biogeography and evolution of Euproctus (Amphibian: Salamandridae), with the resurrection of the genus Calotriton and the description of a new endemic species from the Iberian Peninsula. Zoological Journal of the Linnean Society, 145, 555582.
  • Carretero, M.A., Ayllón, E. & Llorente, G. (2011) Lista patrón de los anfibios y reptiles de España (actualizada a julio de 2011). Asociación Herpetológica Española, Barcelona.
  • Carvalho, S.B., Brito, J.C., Crespo, E.J. & Possingham, H.P. (2010) From climate change predictions to actions – conserving vulnerable animal groups in hotspots at a regional scale. Global Change Biology, 16, 32573270.
  • Carvalho, S.B., Brito, J.C., Crespo, E.J., Watts, M.E. & Possingham, H.P. (2011) Conservation planning under climate change: toward accounting for uncertainty in predicted species distributions to increase confidence in conservation investments in space and time. Biological Conservation, 144, 20202030.
  • Dorda, J. & Esteban, M. (1986) Evolution and subspeciation of Iberian Triturus marmoratus. Studia Herpetologica (ed. by Z. Rocek), pp. 159164. Charles University, Prague.
  • Early, R. & Sax, D.F. (2011) Analysis of climate paths reveals potential limitations on species range shifts. Ecology Letters, 14, 11251133.
  • Escoriza, D., Comas, M.M., Donaire, D. & Carranza, S. (2006) Rediscovery of Salamandra algira Bedriaga, 1833 from the Beni Snassen massif (Morocco) and phylogenetic relationships of North African Salamandra. Amphibia-Reptilia, 27, 448455.
  • Espregueira Themudo, G. & Arntzen, J.W. (2007) Molecular identification of marbled newts and a justification of species status for Triturus marmoratus and T. pygmaeus. Herpetological Journal, 17, 2430.
  • Espregueira Themudo, G., Nieman, A.M. & Arntzen, J.W. (2012) Is dispersal guided by the environment? A comparison of interspecific gene flow estimates among differentiated regions of a newt hybrid zone. Molecular Ecology, 21, 53245335.
  • Estrada, A., Real, R. & Vargas, J.M. (2008) Using crisp and fuzzy modelling to identify favourability hotspots useful to perform gap analysis. Biodiversity and Conservation, 17, 857871.
  • Font, I. (2000) Climatología de España y Portugal, 2nd edn. Ediciones Universidad de Salamanca, Salamanca, Spain.
  • García-París, M., Herrero, P., Martin, C., Dorda, J., Esteban, M. & Arano, B. (1993) Morphological characterization, cytogenetic analysis, and geographical distribution of the pygmy marbled newt Triturus marmoratus pygmaeus (Wolt., 1905) (Caudata, Salamandridae). Bijdragen tot de Dierkunde, 63, 314.
  • García-París, M., Arano, B. & Herrero, P. (2001) Molecular characterization of the contact zone between Triturus pygmaeus and T. marmoratus (Caudata: Salamandridae) in central Spain and their taxonomic assessment. Revista Española de Herpetología, 15, 115126.
  • García-París, M., Montori, A. & Herrero, P. (2004) Amphibia. Lissamphibia. Fauna Ibérica, Vol. 24 (ed. by M.A. Ramos). Museo Nacional de Ciencias Naturales, pp. 190232. CSIC, Madrid.
  • Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, A. (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 19651978.
  • Hoffmann, A.A. & Sgrò, C.M. (2011) Climate change and evolutionary adaptation. Nature, 470, 479485.
  • Hoffmann, M., Hilton-Taylor, C., Angulo, A. et al. (2010) The impact of conservation on the status of the world's vertebrates. Science, 330, 15031509.
  • Hosmer, D.W. & Lemeshow, S. (2000) Applied logistic regression, 2nd edn. John Wiley & Sons, New York.
  • Houlahan, J.E., Findley, C.S., Schmidy, B.R., Meyer, A.H. & Kuzmin, S.L. (2000) Quantitative evidence for global amphibian declines. Nature, 404, 752755.
  • Kozak, K.H. & Wiens, J.J. (2006) Does niche conservatism promote speciation? A case study in North American salamanders. Evolution, 60, 26042621.
  • Kuncheva, L.I. (2001) Using measures of similarity and inclusion for multiple classifer fusion by decision templates. Fuzzy Sets and Systems, 122, 401407.
  • Landis, J.R. & Koch, G.C. (1977) The measurement of observer agreement for categorical data. Biometrics, 33, 159174.
  • Legendre, P. (1993) Spatial autocorrelation: trouble or new paradigm? Ecology, 74, 16591673.
  • Legendre, P. & Legendre, L. (1998) Numerical ecology, 2nd English edn. Elsevier Science, Amsterdam.
  • Loureiro, A., Ferrand de Almeida, N., Carretero, M.A. & Paulo, O.S. (2008) Atlas dos anfíbios e répteis de Portugal. Instituto da Conservação da Natureza e da Biodiversidade, Lisbon.
  • Márquez, A.L., Real, R., Olivero, J. & Estrada, A. (2011) Combining climate with other influencial factors for modelling the impact of climate change on species distribution. Climate Change, 108, 135157.
  • Martin, M.D. & Omland, K.E. (2011) Environmental niche modeling reveals climatic differences among breeding ranges of Orchard Oriole subspecies. American Midland Naturalist, 166, 404414.
  • Miralles, A., Vasconcelos, R., Perera, A., Harris, D.J. & Carranza, S. (2010) An integrative taxonomic revision of the Cape Verdean skinks (Squamata, Scincidae). Zoologica Scripta, 40, 1644.
  • Morrison, W.R., Lohr, J.L., Duchen, P., Wilches, R., Trijillo, D., Mair, M. & Renner, S.S. (2009) The impact of taxonomic change on conservation: does it kill, can it save, or is it just irrelevant? Biological Conservation, 142, 32013206.
  • Muñoz, A.R., Real, R., Barbosa, A.M. & Vargas, J.M. (2005) Modelling the distribution of Bonelli's eagle in Spain: implications for conservation planning. Diversity and Distributions, 11, 477486.
  • Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B. & Kent, J. (2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853858.
  • Nakićenović, N., Alcamo, G., Davis, B. et al. (2000) Emissions scenarios. A special report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge.
  • Ninyerola, M., Pons, X. & Roure, J.M. (2007) Monthly precipitation mapping of the Iberian Peninsula using spatial interpolation tools implemented in a Geographic Information System. Theoretical and Applied Climatology, 89, 195209.
  • Parra-Olea, G., Martínez-Meyer, E. & Pérez-Ponce de León, G. (2005) Forecasting climate change effects of salamander distribution in the highlands of central Mexico. Biotropica, 37, 202208.
  • Pearman, P.B., D'Amen, M., Graham, C.H., Thuiller, W. & Zimmermann, N.E. (2010) Within-taxon niches structure: niche conservatism, divergence and predicted effects of climate change. Ecography, 33, 9901003.
  • Pfenninger, M., Eppenstein, A. & Magnin, F. (2003) Evidence for ecological speciation in the sister species Candidula unifasciata (Poiret, 1801) and C. rugosiuscula (Michaud, 1831) (Helicellinae, Gastropoda). Biological Journal of the Linnean Society, 79, 611628.
  • Pleguezuelos, J.M., Márquez, R. & Lizana, M. (2004) Atlas y libro rojo de los anfibios y reptiles de España, 3rd edn. Dirección General de Conservación de la Naturaleza–Asociación Herpetológica Española, Madrid.
  • Pounds, A.J., Bustamante, M.R., Coloma, L.A., Consuegra, J.A., Fogden, M.P.L., Foster, P.N., La Marca, E., Masters, K.L., Merino-Viteri, A., Puschendorf, R., Ron, S.R., Sánchez-Azofeifa, G.A., Still, C.J. & Young, B.E. (2006) Widespread amphibian extinctions from epidemic disease driven by global warming. Nature, 439, 161167.
  • Real, R., Barbosa, A.M., Martínez-Solano, I. & García-París, M. (2005) Distinguishing the distributions of two cryptic frogs (Anura: Discoglossidae) using molecular data and environmental modeling. Canadian Journal of Zoology, 83, 536545.
  • Real, R., Barbosa, A.M. & Vargas, J.M. (2006) Obtaining environmental favourability functions from logistic regression. Environmental and Ecological Statistics, 13, 237245.
  • Real, R., Márquez, A.L., Olivero, J. & Estrada, A. (2010) Species distribution models in climate change scenarios are still not useful for informing policy planning: an uncertainty assessment using fuzzy logic. Ecography, 33, 304314.
  • Real, R., Romero, D., Olivero, J., Estrada, A. & Márquez, A.L. (2013) Estimating how inflated or obscured effects of climate affect forecasted species distribution. PLoS ONE, 8, e53646.
  • Rey Benayas, J.M., de la Montaña, E., Belliure, J. & Eekhout, X.R. (2006) Identifying areas of high herpetofauna diversity that are threatened by planned infrastructure projects in Spain. Journal of Environmental Management, 79, 279289.
  • Rodríguez-Jiménez, A.J. (1988) Fenología de una comunidad de anfibios asociada a cursos fluviales temporales. Doñana, Acta Vertebrata, 15, 2943.
  • Skov, F. & Svenning, J.-C. (2004) Potential impact of climatic change on the distribution of forest herbs in Europe. Ecography, 27, 366380.
  • Stuart, S.N., Chanson, J.S., Cox, N.A., Young, B.E., Rodrigues, A.S.L., Fischman, D.L. & Waller, R.W. (2004) Status and trends of amphibian declines and extinctions worldwide. Science, 306, 17831786.
  • Sutherland, W.J. (2000) The conservation handbook: research, management and policy. Blackwell Science Ltd., Oxford.
  • Teixeira, J. & Arntzen, J.W. (2002) Potential impact of climate warming on the distribution of the golden-striped salamander, Chioglossa lusitanica, on the Iberian Peninsula. Biodiversity and Conservation, 11, 21672176.
  • Thuiller, W., Araújo, M.B. & Lavorel, S. (2004) Do we need land-cover data to model species distribution in Europe? Journal of Biogeography, 31, 353361.
  • Velo-Antón, G., Martínez-Solano, I. & García-París, M. (2008) β-fibrinogen intron 7 variation in Discoglossus (Anura: Discoglossidae): implications for the taxonomic assessment of morphologically cryptic species. Amphibia-Reptilia, 29, 523533.
  • Vieites, D.R., Nieto-Román, S. & Wale, D.B. (2009) Reconstruction of the climate envelopes of salamanders and their evolution through time. Proceedings of the National Academy of Sciences USA, 106, 1971519722.

Biosketch

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information

The authors of this paper are part of the Biogeography, Diversity and Conservation Research Team of the Universidad de Málaga. This team is interested in biogeography, macroecology, distribution modelling, biodiversity patterns and conservation, mainly of vertebrates but also of invertebrates and plants, with a special focus on the impacts of global change (see http://www.biogeografia.uma.es).

Author contributions: D.R., J.O. and R.R. conceived the ideas and analysed the data; D.R. and A.M. collected and digitalized the variables of climate change and performed the distribution models; D.R. and J.B. developed the initial structure of the text; and J.O. and R.R. revised the manuscript.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Biosketch
  10. Supporting Information
FilenameFormatSizeDescription
jbi12189-sup-0001-AppendixS1-S3.docWord document81K

Appendix S1 Explanatory factors and associated variables used to perform the favourability models.

Appendix S2 Variables entered in the favourability models.

Appendix S3 Values of different indices for the assessment of the favourability models.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.