A procedure to assess the spatial variability in the importance of abiotic factors affecting distributions: the case of world freshwater fishes

A procedure to assess the spatial variability in the importance of abiotic factors affecting... Understanding the factors shaping species’ distributions is a key longstanding topic in ecology with unresolved issues. The aims were to test whether the relative contribution of abiotic factors that set the geographical range of freshwater fish species may vary spatially and/or may depend on the geographical extent that is being considered. The relative contribution of factors, to dis- criminate between the conditions prevailing in the area where the species is present and those existing in the considered extent, was estimated with the instability index included in the R pack- age SPEDInstabR. We used 3 different extent sizes: 1) each river basin where the species is present(local);2)all riverbasinswhere thespecies is present (regional); and 3) the whole Earth (global). We used a data set of 16,543 freshwater fish species with a total of 845,764 geographical records, together with bioclimatic and topographic variables. Factors associated with tempera- ture and altitude show the highest relative contribution to explain the distribution of freshwater fishes at the smaller considered extent. Altitude and a mix of factors associated with temperature and precipitation were more important when using the regional extent. Factors associated with precipitation show the highest contribution when using the global extent. There was also spatial variability in the importance of factors, both between species and within species and from region to region. Factors associated with precipitation show a clear latitudinal trend of decreasing in importance toward the equator. Key words: anisotropic predictors, environmental data selection, geographical background, non-stationary predictors V C The Author (2017). Published by Oxford University Press. 549 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 550 Current Zoology, 2018, Vol. 64, No. 5 The geographical area where a species is distributed is a complex and unique study that we know using a logistic GWR in the context expression of its ecology and evolutionary history (Brown 1995), of a use-availability scheme. being determined by several factors that interact dynamically and An instability index that does not require normalized data has with different strengths at different scales (Gaston 2003; Pearson been proposed recently to discriminate the variables with a higher and Dawson 2003). Four main classes of factors determine areas in likelihood of being relevant for presence/background absence data which a species is found: abiotic and biotic factors, the regions that (Guisande 2016; Guisande et al. 2016). Here, this instability index are accessible to dispersal by the species from some source area, and is used to examine and visualize the spatial variation in the compa- the evolutionary capacity of populations of the species to adapt to rative importance of a high number of continuous predictors, thus new conditions (Sobero ´ n and Peterson 2005). In this context, the showing the anisotropic and non-stationary character of predictors. estimation of the factors correlated with the distribution of organ- Worldwide freshwater fish species data are used to exemplify the isms is the necessary first step to infer the possible mechanisms able capacity of the proposed procedure to describe the comparative rele- to explain why the location, shape, extent, and boundaries of spe- vance of temperature and precipitation variables in explaining spe- cies’ ranges are as they are (Van de Pol et al. 2016), as well as to cies distributions at different scales as well as showing their project the probable distribution of species in absence of exhaustive contrasting non-stationary character. distribution information (Peterson et al. 2011). The explanatory var- iables for these purposes are generally established by examining the Material and Methods relationships between their values and the abundance, density, cover or presence/absence data in the case of individual species. In these Estimation of the relative contribution of factors to studies both the information of the predictors and of the response species’ distribution variable come from different localities, which are managed in a The relative contribution of the selected explanatory variables on “global” way independently of their spatial location. Following the distribution of each one of the species was estimated with the Ramo ´n Margalef (1968) we could say that species “do not dance on recently proposed instability index implemented in the R package the head of a pin” and those simple stationary and isotropic global SPEDInstabR (Guisande 2016; Guisande et al. 2016; see help man- models cannot represent appropriately the complex nature of the ual in Guisande 2016, https://cran.r-project.org/web/packages/ relations between environmental conditions and species occurrences, SPEDInstabR/SPEDInstabR.pdf), which is based on the fluctuation which may differ spatially. index of Dubois (1973) modified by Guisande et al. (2006, 2011). The spatial variation in the weights of a predictor has been esti- This instability index has been designed to improve the correct iden- mated, for example by examining the change in the explanatory tification of the variables controlling species distributions, when capacity of a variable across distance classes or resolutions (Steffan- these are continuous and difficult to normalize. To do that, the Dewenter et al. 2002; Chust et al. 2004), a particularly important SPEDInstabR application of RWizard (Guisande et al. 2014) allows issue in the case of freshwater fishes (Radinger et al. 2015), due to the reliable identification of the environmental factors that better the linear dependence of local characteristics. However, it is the stat- discriminate between the conditions prevailing in presence locations, istical technique geographically weighted regression (GWR) that has against those in the geographical background or area over which a been used in ecology to estimate the local influence of a predictor study was carried out. Once selected, these variables can be used to variable when its relevance varies spatially (Osborne and Suarez- estimate the probable and potential occurrence of the species from Seoane 2002; Foody 2004; Bickford and Laffan 2006; Holloway fragmentary data (Pelayo-Villamil et al. 2012; Lobo 2016) but also and Miller 2015). GWR can calculate local regression coefficients of for other purposes as estimating the variables with a higher proba- each explanatory variable considering subsamples of neighbors, bility of being relevant to explain the distribution of species. which are weighted differently by their distance to the focal point The variables selected are divided into quantiles or bins decided (see Fotheringham et al. 2002). Although having many advantages, by the user (the default is 30), which divide the values of each varia- GWR is computer-intensive, difficult to apply in the case of binary ble into groups with the same number of observations. Both the response variables (presence–absence), and subjected to the same number of observations (cells) in the selected geographical back- requirements of ordinary least squares regressions. ground belonging to each bin for the considered variable and the The selection of the most relevant environmental variables is key number of presence observations in each bin is calculated. For each in the so-called species distribution models (SDMs) or ecological of the considered bins, the relative frequency of the environmental niche models (ENMs), in which environmental characteristics of the variable data as well as that of presence observation are then used to localities in which a species is observed are used as predictors to calculate an index of instability (I). I is a modification of the fluctua- account for species occurrences (Peterson et al. 2011). As reliable tion index (Guisande et al. 2006) as: absence information about species is generally lacking and difficult n i to obtain, the selection of the variables with a higher probability of zj I ¼ R R p log zj z¼1 j¼1 p being influential in predicting the distribution of species can be car- zj ried out by comparing the environmental values in the occurrence localities (use or presences) against those existing in the selected territory (availability or background absences). This is the classic I  I zje zjp p ¼ ; zj procedure used in the resource selection functions (Johnson 1980)to R I  I zje zjp recognize the environmental predictors that contribute most to explain the preferences of the species. These use/availability or where n is the number of environmental variables, i is the number of presence/background absence comparisons are rarely carried out intervals or bins, p is the relative proportion, considering all varia- zj considering the non-stationary nature of the species occurrence– bles and intervals, of the absolute difference between the interval j environment relationships; that is, the lack of constancy of these of the variable z obtained in the cells where the species is present relationships across the geography. Mcnew et al. (2013) is the first (I ) and the cells of the whole GB (I ), and p is the reference state zjp zje zj Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 551 that is calculated as the mean of all the p values. If I and I are distribution information (with available geographical records), in zj zjp zje the same, as zero values cannot be included in the algorithm, the the extent E1 only those species with more than 30 records were minimum instability value is assigned that is obtained for this considered (1,124 species with a total of 732,604 geographical records). In the case of the extents E2 and E3, we used in our study species. only those species with more than 100 records (846 freshwater fish For each environmental variable, a peak of instability is observed species with a total of 723,874 geographical records; see Online for a bin when there are important differences in the relative fre- Appendix 2 for detailed description of the species included in the quency of the cells with presence data compared with those of the analysis). geographical background, thus suggesting that some values of this variable seem to be preferred by the species. Once the instability index is calculated for each species, the values are standardized to a Used environmental variables range between 0 and 1, being thus possible to estimate the percent- From the 19 bioclimatic variables of the WorldClim data set for all age of contribution of each environmental variable to the complete the Earth’s terrestrial area (Hijmans et al. [2005], see the acronyms index value. Those environmental variables with a higher percentage in the web site http://www.worldclim.org/bioclim), we firstly select those with a variance inflation factor (VIF) lower than 20. VIF quan- of contribution to the instability index are those with the highest tify the multicollinearity of predictors (Dormann et al. 2013) and in capacity of discrimination between areas of presence and the geo- our case this value was selected in order to eliminate the variables graphical background. showing the most severe multicollinearity. We used this unusual high VIF threshold to only eliminate those highly correlated environ- Geographical extent mental variables in order to maximize the differences in the selected Due to the importance of the selected geographical extent both in predictors when the considered extent varies. the results and in the discrimination capacity of SDMs (Barve et al. As consequence, the following 10 bioclimatic variables were 2011; Acevedo et al. 2012; Niamir et al. 2016), 3 progressively wid- finally selected: annual mean temperature (BIO1), mean diurnal ening extents were used to determine the comparative importance of range (BIO2), isothermality (BIO3), temperature seasonality (BIO4), environmental predictors. The more restricted used extent (E1) is mean temperature of wettest quarter (BIO8), annual precipitation delimited as each one of the river basins of level 2 (Gonza ´ lez-Vilas (BIO12), precipitation of driest month (BIO14), precipitation sea- et al. 2016), where there are observations of each species. Thus, if a sonality (BIO15), precipitation of warmest quarter (BIO18), and species has presence data in 3 different basins of level 2, the contri- precipitation of coldest quarter (BIO19). We also used other varia- bution of predictors is individually estimated for each basin. The bles (see Pelayo-Villamil et al. 2015), such as human population rationale to use this geographical extent is that basins would encom- density (number of people per km in year 2000), slope (topographic pass the set of accessible localities for each species, and that the com- slope in degrees), slope-aspect (which is defined as the compass parative relevance of each environmental predictor will be better direction to which a slope faces measured in degrees), altitude determined when the effect of dispersal limitations are minimized (meters), vegetation index (VI), terrestrial primary production (TPP, (Peterson et al. 2011; Acevedo et al. 2012). The geographical extent 2 1 gCm d ), and TH24 (topographic heterogeneity calculated for that follows includes all those river basins of level 2 where the spe- the 24 surroundings cells, see Pelayo-Villamil et al. 2015). We there- cies is present (E2); that is, assuming that all the presence localities fore used both direct and indirect variables (Austin 2007) probably are accessible and connected, even if they belong to different basins. related with unconsidered ones, such as solar radiation, river flow, Finally, the whole Earth (E3) was also used as a geographical extent. etc. The number of intervals in which each factor is divided to compare Human population density comes from a globally consistent, the conditions in presence localities versus the selected geographical spatially explicit map based on the Gridded Population of the World extent was the default option in SPEDInstabR. dataset, Version 3 (GPWv3). To develop the global data set, national population data are transformed from their native spatial Origin of freshwater fish species data units, which are usually administrative (such as state or county- The data set of geographical records for freshwater fishes developed level) and of varying resolutions to a global grid of quadrilateral, lat- by Pelayo-Villamil et al. (2015) was updated to reflect the taxo- itude–longitude cells at a resolution of 2.5 arc min, and then down- nomic changes and new species described until the end of April scaled to 6 arc min. A proportional allocation gridding algorithm, 2016. Online Appendix 1 shows a detailed description of all sources utilizing more than 300,000 national and sub-national administra- obtained from the Global Biodiversity Information Facility (GBIF) tive units, is used to assign population values to the 1-degree grid used in this data set, and other sources used such as web pages, cells. Population densities show the number of humans per square museums, etc. are described in Pelayo-Villamil et al. (2015). kilometer, based on census data available in 2000 and with esti- Records were downloaded and filtered using the data cleaning facili- mates when necessary to fill in missing or incomplete data. ties available in the ModestR software (Garcı ´a-Rosello ´ et al. 2013, The source for slope and aspect comes from combining data 2015): 1) records with the same latitude and longitude were not from NASA’s Shuttle Radar Topography Mission covering the land included; 2) records with the latitude and longitude 0 were not surface from 60 south to 60 north. The data for the rest of the included; 3) duplicated records were not included; and 4) habitat Northern Hemisphere (60–90 north) come from digital elevation data cleaning (see Garcı ´a-Rosello ´ et al. [2014] for details). At the models (digital versions of paper-based topographic maps) produced end of April 2016, 16,543 species of freshwater fishes were recog- by the US Geological Survey. The data for the remainder of the nized as valid by systematists and are available in IPez (http://www. Southern Hemisphere (60–90 south) come from the “RAMP II” ipez.es, Guisande et al. 2010), so this taxonomic list of species was project of the Radarsat Antarctic Mapping Project Digital Elevation used in our study. Of these, 16,479 species (99.6% of the total) have Model, Version 2. 2 1 associated geographical information for a total of 845,764 geo- VI and TPP (in g C m d ) come from the Moderate graphical records (without duplication). From species with Resolution Imaging Spectroradiometer (MODIS) instrument aboard Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 552 Current Zoology, 2018, Vol. 64, No. 5 E1 BIO8 BIO1 Altitude BIO3 BIO4 BIO12 BIO14 BIO18 BIO15 BIO19 BIO2 VI TPP Slope TH24 Aspect Pop Variables E2 Altitude BIO19 BIO1 BIO4 BIO12 BIO14 BIO18 BIO8 BIO3 BIO15 TPP BIO2 VI Slope TH24 Pop Aspect Variables E3 BIO19 BIO14 BIO12 BIO18 BIO4 BIO3 BIO1 BIO8 TPP BIO15 Altitude Slope BIO2 VI TH24 Aspect Pop Variables Figure 1. Boxplots representing the median contribution of each one of the considered variables comparing the conditions of each freshwater fish species in their 0 0 presence cells of 5  5 against the conditions prevailing in the cells i) included in each river basin of level 2 where there are observations of each species (E1); ii) all river basins where the species is present (E2); and iii) the complete world as extent (E3). The limit of the bars indicates the minimum and maximum, the limit of the box indicates the first and third quartile, and the points are the outliers. If the notches of 2 groups do not overlap it seems to be an evidence that the 2 medians differ. BIO1, annual mean temperature; BIO2, mean diurnal range; BIO3, isothermality; BIO4, temperature seasonality; BIO8, mean temperature of wet- test quarter; BIO12, annual precipitation; BIO14, precipitation of driest month; BIO15, precipitation seasonality; BIO18, precipitation of warmest quarter; BIO19, precipitation of coldest quarter; number of people per km , Pop, population density; topographic slope in degrees, slope; slope-aspect (which is defined as the 2 1 compass direction to which a slope faces measured in degrees); altitude (meters); VI, vegetation index; TPP, terrestrial primary production in g C m d ; and TH24, topographic heterogeneity calculated for the 24 surroundings cells. Percentage of contribution to instability index Percentage of contribution to instability index Percentage of contribution to instability index 0 5 10 15 20 0 5 10 15 20 25 0 102030 Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 553 Figure 2. Mean6 SD frequencies (lines6 shaded areas) for each variable interval for the cells with presence records and the cells of the E2 geographical extent (all river basins of level 2 where the species is present). BIO1, annual mean temperature; BIO4, temperature seasonality; BIO12, annual precipitation; BIO14, pre- cipitation of driest month; and BIO19, precipitation of the coldest quarter. NASA’s Terra satellite. Specifically, monthly data of terrestrial net different extents may be influencing the provided results but con- primary productivity and VI from 2001 to 2010 were obtained by versely using different resolutions may complicate the estimation of averaging available information for each pixel of selected variables the effect that we try to measure (the extent) as showed by other using the statistical software RWizard (Guisande et al. 2014). The authors (Gillingham et al. 2012). net primary productivity indicates how much carbon dioxide is taken up by vegetation during photosynthesis minus how much car- bon dioxide is released when plants respire. The values indicate how Results fast carbon was taken in, or released, for every square meter of land 2 1 At the more restricted extent (E1), variables associated to tempera- over the indicated time span. Values range from 1.0 g C m d 2 1 ture (BIO8, BIO1, BIO3, and BIO4) and the altitude were the fac- to 6.5 g m d . A negative value means decomposition or respira- tors with the highest relative contribution to explain the tion exceeded carbon absorption; in other words, more carbon was distributions of each individual freshwater fish species (Figure 1 released into the atmosphere than was absorbed by the plants. We upper panel). Altitude and a mix of factors both associated with also include the VI as a productivity variable. This variable repre- temperature (BIO1 and BIO4) and precipitation (BIO19, BIO12, sents a measure of the greenness of Earth’s landscapes. and BIO14) become important factors at E2 (regional extent; see All these variables were included in the analyses at a resolution 0 0 of 5  5 (100 km ). The use of the same resolution for the medium panel in Figure 1). Lastly, some variables associated with Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 554 Current Zoology, 2018, Vol. 64, No. 5 until 13 C, approximately, but more frequent at temperatures from 13 Cto22 C(Figure 2). The frequencies of presence cells were higher at intermediate values of temperature seasonality (BIO4), but freshwater species seem to avoid areas with high temperature sea- sonality (Figure 2). In the case of BIO12 (annual precipitation), the frequencies of presence cells were lower than those of the geographi- cal extent in places with <1,200 mm, approximately, but higher at greater precipitation values. Both the precipitation of the driest month (BIO14) and the precipitation of the coldest quarter (BIO19) showed a similar pattern (Figure 2), with the frequencies of the pres- ence cells being lower at lower values and higher at intermediate val- ues than the frequencies of cells at the considered extent. The spatial variability in the relative contribution of each factor, both for all species (mean contribution) and single species, show that the same variable may exercise a different influence depending on the geographical location and the considered extent (Figure 3 upper panel). This spatial variability can be observed in all the con- sidered environmental variables (not shown). In fact, a clear latitudi- nal gradient can be observed in the variation of the percentage of contribution of some variables, which may vary depending on the considered extent (i.e., BIO14 and BIO19 see Figure 4). The contri- bution of an indirect variable such as altitude seems to be highly dependent on the considered extent, although is generally lower at the higher latitudes of the southern hemisphere. The contribution of annual mean temperature was rather homogeneous and erratic (Figure 4), while the variation in the contribution of precipitation variables (i.e., BIO19 and BIO14 in Figure 4) show a clear pattern. Tropical and subtropical basins were significantly less influenced by precipitation variables than northernmost and southernmost high- latitude basins. Interestingly, in the northern hemisphere the contri- bution of precipitation variables seems to be higher when the consid- ered extent increases. Figure 3. Map representing the geographical variation in the mean percent- age of contribution of precipitation of the coldest quarter (BIO19, upper panel) Discussion for all freshwater fish species in the cells of 1 , as measured by the instability index. These contribution scores were calculated using the SPEDInstabR In this study, we were able to demonstrate that the identification of algorithm comparing the values in the presence cells against those present in the factors accounting for the distribution range of the species all the cells of the E2 geographical extent (all river basins of level 2 where the clearly depend on the extent to which these relationships are exam- species is present). Percentage of contribution of altitude on the distribution ined, a topic which has long been hypothesized (MacArthur 1972; Esox lucius (medium panel) using the E1 geographical extent, and contribu- tion of annual precipitation (BIO12) for the species Hoplias malabaricus Wiens 2015) and that only recently has been considered fundamen- (lower panel) at the same extent. tal to correctly estimate the predictor functions relating species occurrences with environmental variables (Barve et al. 2011; precipitation (BIO19, BIO14, BIO12, and BIO18) were those that Acevedo et al. 2012). Our study also shows that the influence of had the highest contributions when using the world as extent (E3; explanatory variables changes spatially in a non-stationary way, as lower panel in Figure 1). Therefore, the contribution of the factors demonstrated in other studies (Osborne and Suarez-Seoane 2002; associated with precipitation seems to increase with the size of the Foody 2004; Bickford and Laffan 2006; Hortal et al. 2011; Mcnew considered extent, while the contribution of the factors associated et al. 2013; Holloway and Miller 2015). In our study, altitude is an with temperature variables increased as the size of the extent important factor that influences the smaller considered extent distri- decreases. bution patterns, but not at global extents. Factors associated to tem- The differences in the frequency of presence cells along each vari- perature are also important at local extent, whereas factors able gradient versus the frequency of the cells at the selected extent associated with precipitation seem to be more important at global can be geographically represented (Figure 2). The results for the E2 extents. Moreover, the relative contribution of factors to explain the situation were described, although the patterns were similar for the distribution of freshwater fish species varies from region to region, E1 and E3 extents (shown in Online Appendix 3, Figure A3.5 for E1 due partially to the different species present in each area, but mainly and Figure A3.6 for E3). Thus, presence cells seem to be compara- due to the spatial variation in the importance of the factors account- tively more frequent at lower altitudes (<250 m) and less frequent, ing for freshwater species’ distributions. This spatial variability in at altitudes from 300 m to 2,000 m, than the frequencies of the cells the importance of environmental factors has been formerly docu- of the E2 geographical extent (Figure 2). At higher altitudes mented in the case of fishes (Windle et al. 2010; Radinger et al. (>2,000 m) both frequencies seem to be similar. In the case of 2015). Here, a latitudinal and geographical gradient in the impor- annual mean temperatures (BIO1), presence cells were less frequent tance of abiotic factors has been shown. Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 555 Local Regional Global Local Regional Global -40 -20 0 20 40 -40 -20 0 20 40 Latitude Latitude Local Regional Global Local Regional Global -40 -20 0 20 40 -40 -20 0 20 40 Latitude Latitude Figure 4. Mean latitudinal contribution to the values of the instability index (%) of altitude, annual mean temperature (BIO1), precipitation of driest month (BIO14), and precipitation of coldest quarter (BIO19), depending on the considered geographical extent. Higher latitudes (90–45 both north and south) were not included in the plots due to the low number of species in those regions. The frequency distribution of presence data compared with those fishes should be managed with caution. The provided results are not existing in the occupied basins (Figure 2) is in agreement with classic only limited by potential errors in the number and characteristics of environmental gradients (Griffiths et al. 2014): the frequency of the used environmental variables (Van Neil et al. 2004), the exis- occurrence of freshwater fishes diminishes at temperatures below tence of false presence records (Tyre et al. 2003) or survey biases 13 C and precipitations lower than 1,200 mm, approximately. (Pelayo-Villamil et al. 2015), but also by the lack of consideration of However, freshwater fishes in tropical and subtropical basins are the historical biogeography of the species (Wiens and Donoghue comparatively less influenced in their distributions by precipitation 2004) or species interactions (Gonza ´ lez-Salazara et al. 2013). variables than those located in northernmost and southernmost However, in spite of this, our findings clearly suggest that different high-latitude basins. Interestingly, in the northern hemisphere the predictors can be detected as relevant at different scales. This fact contribution of precipitation variables seems to be higher when the has important implications for studies about SDMs and ENMs, for considered extent increases, so that precipitation variables increase which predictor selection is an important issue affecting models’ pre- in importance at higher latitudes, both south and north, although dictive ability (Austin 2007). We suggest when modeling species dis- this latitudinal trend is more or less pronounced depending on the tribution, that careful attention should be paid to the selection of the extent considered. Thus, when the variability in the climatic condi- extent used to estimate the SDMs and ENMs as previously suggested tions is minimized by considering only a geographical background (Barve et al. 2011; Acevedo et al. 2012), and that a single set of pre- limited to the basins in which a species occurs, the comparative rele- dictors for each species should not be used, but instead different pre- vance of precipitation variables in explaining occurrence and distri- dictors and predictor functions for each region where the species is bution seems to be higher than that of temperature under temperate present (Mcnew et al. 2013). These are not new issues and we are conditions. We may hypothesize that the higher levels of annual pre- aware of how its consideration complicates the building of models cipitation in tropical and subtropical basins suggest that precipita- capable of providing reliable estimations of species distributions. tion is not a limiting factor in these areas, so its effect on fish How to select the most appropriate extent for each species, and how distribution is not as important as in higher latitudes, where hydro- to manage the variation in the parameters of the environmental pre- logical variations related to precipitation may be a more limiting dictors obtained at different extents to generate reliable regional pre- factor when compared with temperature. dictions are key issues to address (see, e.g., Sua ´ rez-Seoane et al. Of course, the procedure proposed to identify the most impor- 2014). Be that as it may, modelers should justify how they have tant factors in determining the occurrence of individual freshwater managed these 2 questions or why it has not done so. BIO14 Altitude 2 4 6 8 10 12 14 16 24 68 10 12 BIO19 BIO1 46 8 10 12 14 567 8 9 10 Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 556 Current Zoology, 2018, Vol. 64, No. 5 Guisande C, Garcı´a-Rosello ´ E, Heine J, Gonza ´ lez-Dacosta J, Goza ´ lez Vilas L Supplementary Material et al., 2016. SPEDInstabR: an algorithm based on a fluctuation index Supplementary material can be found at https://academic.oup.com/cz. for selecting predictors in species distribution modeling. Ecol Inform 37: 18–23. Guisande C, Manjarre ´ s-Herna ´ ndez A, Pelayo-Villamil P, Granado-Lorencio References C, Riveiro I et al., 2010. IPez: an expert system for the taxonomic identifica- Acevedo P, Jimenez-Valverde A, Lobo JM, Real R, 2012. Delimiting the geo- tion of fishes based on machine learning techniques. Fish Res 102:240–247. graphical background in species distribution modelling. J Biogeogr 39: Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A, 2005. Very high reso- 1383–1390. lution interpolated climate surfaces for global land areas. Int J Climatol 25: Austin M, 2007. Species distribution models and ecological theory: a critical 1965–1978. assessment and some possible approaches. Ecol Model 200:1–19. Holloway P, Miller JA, 2015. Exploring spatial scale, autocorrelation and Barve N, Barve V, Jime ´ nez-Valverde A, Lira-Noriega A, Maher SP et al., nonstationarity of bird species richness patterns. ISPRS Int J Geo-Inform 4: 2011. Crucial role of the accessibility area in ecological niche modeling and 783–798. species distribution modeling. Ecol Model 222:1810–1819. Hortal J, Diniz-Filho JAF, Bini LM, Rodriguez MA, Baselga A et al., 2011. Ice Bickford S, Laffan S, 2006. Multi-extent analysis of the relationship between age climate, evolutionary constraints and diversity patterns of European pteridophyte species richness and climate. Glob Ecol Biogeogr 15:588–601. dung beetles. Ecol Lett 14:741–748. Brown JH, 1995. Macroecology. Chicago (IL): University of Chicago Press. Johnson DH, 1980. The comparison of usage and availability measurements Chust G, Petrus JL, Ducrot D, Ventura D, 2004. Scale dependence of insect for evaluating resource preference. Ecology 61:65–71. assemblages in response to landscape pattern. Landsc Ecol 18:41–57. Lobo JM, 2016. The use of occurrence data to predict the effects of climate Dormann CF, Elith J, Bacher S, Buchmann C, Carl G et al., 2013. Collinearity: change on insects. Curr Opin Insect Sci 17:62–68. a review of methods to deal with it and a simulation study evaluating their MacArthur RH, 1972. Geographic Ecology. New York (NY): Harper and performance. Ecography 36:27–46. Row. Dubois DM, 1973. An index of fluctuations, Do, connected with diversity and Margalef R, 1968. Perspectives in Ecological Theory. Chicago (IL): The stability of ecosystems: applications in the Lotka-Volterra model and in an University of Chicago Press. experimental distribution of species. Rapport de sythe ` se III, Programme Mcnew LB, Gregory AJ, Sandercock BK, 2013. Spatial heterogeneity in habi- National sur l’environment Physique et Biologique, Project Mer. tat selection: nest site selection by greater prairie-chickens. J Wildl Manage Commision Interministe ´ rielle de la Politique Scientifique. Lie ` ge. 77:791–801. Foody GM, 2004. Spatial nonstationarity and scale-dependency in the rela- Niamir A, Skidmore AK, Toxopeus AG, Real R, 2016. Use of taxonomy to tionship between species richness and environmental determinants for the delineate spatial extent of atlas data for species distribution models. Glob sub-Saharan endemic avifauna. Glob Ecol Biogeogr 13:315–320. Ecol Biogeogr 25:227–237. Fotheringham AS, Brundson C, Charlton M, 2002. Geographically Weighted Osborne PE, Suarez-Seoane S, 2002. Should data be partitioned spatially Regression: The Analysis of Spatially Varying Relationships. West Sussex, before building large-scale distribution models? Ecol Model 157:249–259. UK: John Wiley and Sons, Ltd. Pearson RG, Dawson TP, 2003. Predicting the impacts of climate change on Garcı´a-Rosello ´ E, Guisande C, Gonza ´ lez-Dacosta J, Heine J, Pelayo-Villamil P the distribution of species: are bioclimate envelope models useful? Glob et al., 2013. ModestR: a software tool for managing and analyzing species Ecol Biogeogr 12:361–371. distribution map databases. Ecography 36:1202–1207. Pelayo-Villamil P, Guisande C, Gonza ´ lez-Vilas L, Carvajal-Quintero JD, Garcı´a-Rosello ´ E, Guisande C, Heine J, Pelayo-Villamil P, Manjarre ´s- Jime ´ nez-Segura LF et al., 2012. ModestR: Una herramienta informa ´ tica Herna ´ ndez A et al., 2014. Using ModestR to download, import and clean para el estudio de los ecosistemas acua ´ ticos de Colombia. Actual Biol 34: species distribution records. Methods Ecol Evol 5:703–713. 225–239. Garcı´a-Rosello ´ E, Guisande C, Manjarre ´ s-Herna ´ ndez A, Gonza ´ lez-Dacosta J, Pelayo-Villamil P, Guisande C, Vari RP, Manjarre ´ s-Herna ´ ndez A, Garcı´a- Heine J et al., 2015. Can we derive macroecological patterns from primary Rosello ´ E et al., 2015. Global diversity patterns of freshwater fishes: poten- GBIF data? Glob Ecol Biogeogr 24:335–347. tial victims of their own success. Divers Distrib 21:345–356. Gillingham PK, Huntley B, Kunin WE, Thomas CD, 2012. The effect of spatial Peterson AT, Sobero ´ n J, Pearson RG, Anderson R, Martı´nez-Meyer E et al., resolution on projected responses to climate warming. Divers Distrib 18: 2011. Ecological Niches and Geographic Distributions. Princeton (NJ): 990–1000. Princeton University Press. Gaston K, 2003. The Structure and Dynamics of Geographic Ranges. Oxford: Radinger J, Wolter C, Kail J, 2015. Spatial scaling of environmental variables Oxford University Press. improves species–habitat models of fishes in a small, sand-bed lowland river. Gonza ´ lez-Salazara C, Stephensb CR, Marquet PA, 2013. Comparing the rela- PLoS One 10:e0142813. tive contributions of biotic and abiotic factors as mediators of species’ distri- Sobero ´ n J, Peterson AT, 2005. Interpretation of models of fundamental eco- butions. Ecol Model 248:57–70. logical niches and species’ distributional areas. Biodivers Inform 2:1–10. Gonza ´ lez-Vilas L, Guisande C, Vari RP, Pelayo-Villamil P, Manjarre ´s- Steffan-Dewenter I, Mu ¨ nzenberg U, Burger C, Thies C, Tscharntke T, 2002. Herna ´ ndez A et al., 2016. Geospatial data of freshwater habitats for macro- Scale-dependent effects of landscape context on three pollinator guilds. ecological studies: an example with freshwater fishes. Int J Geogr Inf Sci 30: Ecology 83:1421–1432. 126–141. Sua ´ rez-Seoane S, Virgo ´ s E, Terroba O, Pardavila X, Barea-Azco ´ n JM, 2014. Griffiths D, McGonigle C, Quinn R, 2014. Climate and species richness pat- Scaling of species distribution models across spatial resolutions and extents terns of freshwater fish in North America and Europe. J Biogeogr 41: along a biogeographical gradient: the case of the Iberian mole Talpa occi- 452–463. dentalis. Ecography 37:279–292. Guisande C, 2016. Estimation of the relative importance of factors affecting Tyre AJ, Tenhumberg B, Field SA, Niejalke D, Parris K et al., 2003. Improving species distribution based on stability concept. R package version 1.3. precision and reducing bias in biological surveys: estimating false-negative Available from: http://CRAN.R-project.org/package¼SPEDInstabR error rates. Ecol Appl 13:1790–1801. (accessed 6 November 2016). Van de Pol M, Bailey LD, McLean N, Rijsdijk L, Lawson CR et al., 2016. Guisande C, Heine J, Gomza ´ lez-DaCosta J, Garcı´a-Rosello ´ E, 2014. Vigo, Identifying the best climatic predictors in ecology and evolution. Methods Spain: RWizard Software. University of Vigo. Ecol Evol 7:1246–1257. Guisande C, Barreiro A, Maneiro I, Riveiro I, Vergara-Castano ~ AR et al., Van Neil KP, Laffan SW, Lees BG, 2004. Effect of error in the DEM on envi- 2006. Tratamiento de Datos. Madrid: Ediciones Dı ´az de Santos. Guisande C, Barreiro A, Vaamonde A, 2011. Tratamiento de Datos con R, ronmental variables for predictive vegetation modelling. J Veg Sci 15: Statistica y SPSS. Madrid: Ediciones Dı´az de Santos. 747–756. Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 557 Wiens JJ, 2015. The niche, biogeography and species interactions. Philos Windle MJS, Rose GA, Devillers R, Fortin MJ, 2010. Exploring spatial Trans R Soc B 366:2336–2350. non-stationarity of fisheries survey data using geographically weighted Wiens JJ, Donoghue MJ, 2004. Historical biogeography, ecology and species regression (GWR): an example from the Northwest Atlantic. ICES J Mar Sci richness. Trends Ecol Evol 19:639–644. 67:145–154. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Current Zoology Oxford University Press

A procedure to assess the spatial variability in the importance of abiotic factors affecting distributions: the case of world freshwater fishes

Free
9 pages

Loading next page...
 
/lp/ou_press/a-procedure-to-assess-the-spatial-variability-in-the-importance-of-extOC0J6eL
Publisher
Oxford University Press
Copyright
© The Author (2017). Published by Oxford University Press.
ISSN
1674-5507
eISSN
2396-9814
D.O.I.
10.1093/cz/zox063
Publisher site
See Article on Publisher Site

Abstract

Understanding the factors shaping species’ distributions is a key longstanding topic in ecology with unresolved issues. The aims were to test whether the relative contribution of abiotic factors that set the geographical range of freshwater fish species may vary spatially and/or may depend on the geographical extent that is being considered. The relative contribution of factors, to dis- criminate between the conditions prevailing in the area where the species is present and those existing in the considered extent, was estimated with the instability index included in the R pack- age SPEDInstabR. We used 3 different extent sizes: 1) each river basin where the species is present(local);2)all riverbasinswhere thespecies is present (regional); and 3) the whole Earth (global). We used a data set of 16,543 freshwater fish species with a total of 845,764 geographical records, together with bioclimatic and topographic variables. Factors associated with tempera- ture and altitude show the highest relative contribution to explain the distribution of freshwater fishes at the smaller considered extent. Altitude and a mix of factors associated with temperature and precipitation were more important when using the regional extent. Factors associated with precipitation show the highest contribution when using the global extent. There was also spatial variability in the importance of factors, both between species and within species and from region to region. Factors associated with precipitation show a clear latitudinal trend of decreasing in importance toward the equator. Key words: anisotropic predictors, environmental data selection, geographical background, non-stationary predictors V C The Author (2017). Published by Oxford University Press. 549 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 550 Current Zoology, 2018, Vol. 64, No. 5 The geographical area where a species is distributed is a complex and unique study that we know using a logistic GWR in the context expression of its ecology and evolutionary history (Brown 1995), of a use-availability scheme. being determined by several factors that interact dynamically and An instability index that does not require normalized data has with different strengths at different scales (Gaston 2003; Pearson been proposed recently to discriminate the variables with a higher and Dawson 2003). Four main classes of factors determine areas in likelihood of being relevant for presence/background absence data which a species is found: abiotic and biotic factors, the regions that (Guisande 2016; Guisande et al. 2016). Here, this instability index are accessible to dispersal by the species from some source area, and is used to examine and visualize the spatial variation in the compa- the evolutionary capacity of populations of the species to adapt to rative importance of a high number of continuous predictors, thus new conditions (Sobero ´ n and Peterson 2005). In this context, the showing the anisotropic and non-stationary character of predictors. estimation of the factors correlated with the distribution of organ- Worldwide freshwater fish species data are used to exemplify the isms is the necessary first step to infer the possible mechanisms able capacity of the proposed procedure to describe the comparative rele- to explain why the location, shape, extent, and boundaries of spe- vance of temperature and precipitation variables in explaining spe- cies’ ranges are as they are (Van de Pol et al. 2016), as well as to cies distributions at different scales as well as showing their project the probable distribution of species in absence of exhaustive contrasting non-stationary character. distribution information (Peterson et al. 2011). The explanatory var- iables for these purposes are generally established by examining the Material and Methods relationships between their values and the abundance, density, cover or presence/absence data in the case of individual species. In these Estimation of the relative contribution of factors to studies both the information of the predictors and of the response species’ distribution variable come from different localities, which are managed in a The relative contribution of the selected explanatory variables on “global” way independently of their spatial location. Following the distribution of each one of the species was estimated with the Ramo ´n Margalef (1968) we could say that species “do not dance on recently proposed instability index implemented in the R package the head of a pin” and those simple stationary and isotropic global SPEDInstabR (Guisande 2016; Guisande et al. 2016; see help man- models cannot represent appropriately the complex nature of the ual in Guisande 2016, https://cran.r-project.org/web/packages/ relations between environmental conditions and species occurrences, SPEDInstabR/SPEDInstabR.pdf), which is based on the fluctuation which may differ spatially. index of Dubois (1973) modified by Guisande et al. (2006, 2011). The spatial variation in the weights of a predictor has been esti- This instability index has been designed to improve the correct iden- mated, for example by examining the change in the explanatory tification of the variables controlling species distributions, when capacity of a variable across distance classes or resolutions (Steffan- these are continuous and difficult to normalize. To do that, the Dewenter et al. 2002; Chust et al. 2004), a particularly important SPEDInstabR application of RWizard (Guisande et al. 2014) allows issue in the case of freshwater fishes (Radinger et al. 2015), due to the reliable identification of the environmental factors that better the linear dependence of local characteristics. However, it is the stat- discriminate between the conditions prevailing in presence locations, istical technique geographically weighted regression (GWR) that has against those in the geographical background or area over which a been used in ecology to estimate the local influence of a predictor study was carried out. Once selected, these variables can be used to variable when its relevance varies spatially (Osborne and Suarez- estimate the probable and potential occurrence of the species from Seoane 2002; Foody 2004; Bickford and Laffan 2006; Holloway fragmentary data (Pelayo-Villamil et al. 2012; Lobo 2016) but also and Miller 2015). GWR can calculate local regression coefficients of for other purposes as estimating the variables with a higher proba- each explanatory variable considering subsamples of neighbors, bility of being relevant to explain the distribution of species. which are weighted differently by their distance to the focal point The variables selected are divided into quantiles or bins decided (see Fotheringham et al. 2002). Although having many advantages, by the user (the default is 30), which divide the values of each varia- GWR is computer-intensive, difficult to apply in the case of binary ble into groups with the same number of observations. Both the response variables (presence–absence), and subjected to the same number of observations (cells) in the selected geographical back- requirements of ordinary least squares regressions. ground belonging to each bin for the considered variable and the The selection of the most relevant environmental variables is key number of presence observations in each bin is calculated. For each in the so-called species distribution models (SDMs) or ecological of the considered bins, the relative frequency of the environmental niche models (ENMs), in which environmental characteristics of the variable data as well as that of presence observation are then used to localities in which a species is observed are used as predictors to calculate an index of instability (I). I is a modification of the fluctua- account for species occurrences (Peterson et al. 2011). As reliable tion index (Guisande et al. 2006) as: absence information about species is generally lacking and difficult n i to obtain, the selection of the variables with a higher probability of zj I ¼ R R p log zj z¼1 j¼1 p being influential in predicting the distribution of species can be car- zj ried out by comparing the environmental values in the occurrence localities (use or presences) against those existing in the selected territory (availability or background absences). This is the classic I  I zje zjp p ¼ ; zj procedure used in the resource selection functions (Johnson 1980)to R I  I zje zjp recognize the environmental predictors that contribute most to explain the preferences of the species. These use/availability or where n is the number of environmental variables, i is the number of presence/background absence comparisons are rarely carried out intervals or bins, p is the relative proportion, considering all varia- zj considering the non-stationary nature of the species occurrence– bles and intervals, of the absolute difference between the interval j environment relationships; that is, the lack of constancy of these of the variable z obtained in the cells where the species is present relationships across the geography. Mcnew et al. (2013) is the first (I ) and the cells of the whole GB (I ), and p is the reference state zjp zje zj Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 551 that is calculated as the mean of all the p values. If I and I are distribution information (with available geographical records), in zj zjp zje the same, as zero values cannot be included in the algorithm, the the extent E1 only those species with more than 30 records were minimum instability value is assigned that is obtained for this considered (1,124 species with a total of 732,604 geographical records). In the case of the extents E2 and E3, we used in our study species. only those species with more than 100 records (846 freshwater fish For each environmental variable, a peak of instability is observed species with a total of 723,874 geographical records; see Online for a bin when there are important differences in the relative fre- Appendix 2 for detailed description of the species included in the quency of the cells with presence data compared with those of the analysis). geographical background, thus suggesting that some values of this variable seem to be preferred by the species. Once the instability index is calculated for each species, the values are standardized to a Used environmental variables range between 0 and 1, being thus possible to estimate the percent- From the 19 bioclimatic variables of the WorldClim data set for all age of contribution of each environmental variable to the complete the Earth’s terrestrial area (Hijmans et al. [2005], see the acronyms index value. Those environmental variables with a higher percentage in the web site http://www.worldclim.org/bioclim), we firstly select those with a variance inflation factor (VIF) lower than 20. VIF quan- of contribution to the instability index are those with the highest tify the multicollinearity of predictors (Dormann et al. 2013) and in capacity of discrimination between areas of presence and the geo- our case this value was selected in order to eliminate the variables graphical background. showing the most severe multicollinearity. We used this unusual high VIF threshold to only eliminate those highly correlated environ- Geographical extent mental variables in order to maximize the differences in the selected Due to the importance of the selected geographical extent both in predictors when the considered extent varies. the results and in the discrimination capacity of SDMs (Barve et al. As consequence, the following 10 bioclimatic variables were 2011; Acevedo et al. 2012; Niamir et al. 2016), 3 progressively wid- finally selected: annual mean temperature (BIO1), mean diurnal ening extents were used to determine the comparative importance of range (BIO2), isothermality (BIO3), temperature seasonality (BIO4), environmental predictors. The more restricted used extent (E1) is mean temperature of wettest quarter (BIO8), annual precipitation delimited as each one of the river basins of level 2 (Gonza ´ lez-Vilas (BIO12), precipitation of driest month (BIO14), precipitation sea- et al. 2016), where there are observations of each species. Thus, if a sonality (BIO15), precipitation of warmest quarter (BIO18), and species has presence data in 3 different basins of level 2, the contri- precipitation of coldest quarter (BIO19). We also used other varia- bution of predictors is individually estimated for each basin. The bles (see Pelayo-Villamil et al. 2015), such as human population rationale to use this geographical extent is that basins would encom- density (number of people per km in year 2000), slope (topographic pass the set of accessible localities for each species, and that the com- slope in degrees), slope-aspect (which is defined as the compass parative relevance of each environmental predictor will be better direction to which a slope faces measured in degrees), altitude determined when the effect of dispersal limitations are minimized (meters), vegetation index (VI), terrestrial primary production (TPP, (Peterson et al. 2011; Acevedo et al. 2012). The geographical extent 2 1 gCm d ), and TH24 (topographic heterogeneity calculated for that follows includes all those river basins of level 2 where the spe- the 24 surroundings cells, see Pelayo-Villamil et al. 2015). We there- cies is present (E2); that is, assuming that all the presence localities fore used both direct and indirect variables (Austin 2007) probably are accessible and connected, even if they belong to different basins. related with unconsidered ones, such as solar radiation, river flow, Finally, the whole Earth (E3) was also used as a geographical extent. etc. The number of intervals in which each factor is divided to compare Human population density comes from a globally consistent, the conditions in presence localities versus the selected geographical spatially explicit map based on the Gridded Population of the World extent was the default option in SPEDInstabR. dataset, Version 3 (GPWv3). To develop the global data set, national population data are transformed from their native spatial Origin of freshwater fish species data units, which are usually administrative (such as state or county- The data set of geographical records for freshwater fishes developed level) and of varying resolutions to a global grid of quadrilateral, lat- by Pelayo-Villamil et al. (2015) was updated to reflect the taxo- itude–longitude cells at a resolution of 2.5 arc min, and then down- nomic changes and new species described until the end of April scaled to 6 arc min. A proportional allocation gridding algorithm, 2016. Online Appendix 1 shows a detailed description of all sources utilizing more than 300,000 national and sub-national administra- obtained from the Global Biodiversity Information Facility (GBIF) tive units, is used to assign population values to the 1-degree grid used in this data set, and other sources used such as web pages, cells. Population densities show the number of humans per square museums, etc. are described in Pelayo-Villamil et al. (2015). kilometer, based on census data available in 2000 and with esti- Records were downloaded and filtered using the data cleaning facili- mates when necessary to fill in missing or incomplete data. ties available in the ModestR software (Garcı ´a-Rosello ´ et al. 2013, The source for slope and aspect comes from combining data 2015): 1) records with the same latitude and longitude were not from NASA’s Shuttle Radar Topography Mission covering the land included; 2) records with the latitude and longitude 0 were not surface from 60 south to 60 north. The data for the rest of the included; 3) duplicated records were not included; and 4) habitat Northern Hemisphere (60–90 north) come from digital elevation data cleaning (see Garcı ´a-Rosello ´ et al. [2014] for details). At the models (digital versions of paper-based topographic maps) produced end of April 2016, 16,543 species of freshwater fishes were recog- by the US Geological Survey. The data for the remainder of the nized as valid by systematists and are available in IPez (http://www. Southern Hemisphere (60–90 south) come from the “RAMP II” ipez.es, Guisande et al. 2010), so this taxonomic list of species was project of the Radarsat Antarctic Mapping Project Digital Elevation used in our study. Of these, 16,479 species (99.6% of the total) have Model, Version 2. 2 1 associated geographical information for a total of 845,764 geo- VI and TPP (in g C m d ) come from the Moderate graphical records (without duplication). From species with Resolution Imaging Spectroradiometer (MODIS) instrument aboard Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 552 Current Zoology, 2018, Vol. 64, No. 5 E1 BIO8 BIO1 Altitude BIO3 BIO4 BIO12 BIO14 BIO18 BIO15 BIO19 BIO2 VI TPP Slope TH24 Aspect Pop Variables E2 Altitude BIO19 BIO1 BIO4 BIO12 BIO14 BIO18 BIO8 BIO3 BIO15 TPP BIO2 VI Slope TH24 Pop Aspect Variables E3 BIO19 BIO14 BIO12 BIO18 BIO4 BIO3 BIO1 BIO8 TPP BIO15 Altitude Slope BIO2 VI TH24 Aspect Pop Variables Figure 1. Boxplots representing the median contribution of each one of the considered variables comparing the conditions of each freshwater fish species in their 0 0 presence cells of 5  5 against the conditions prevailing in the cells i) included in each river basin of level 2 where there are observations of each species (E1); ii) all river basins where the species is present (E2); and iii) the complete world as extent (E3). The limit of the bars indicates the minimum and maximum, the limit of the box indicates the first and third quartile, and the points are the outliers. If the notches of 2 groups do not overlap it seems to be an evidence that the 2 medians differ. BIO1, annual mean temperature; BIO2, mean diurnal range; BIO3, isothermality; BIO4, temperature seasonality; BIO8, mean temperature of wet- test quarter; BIO12, annual precipitation; BIO14, precipitation of driest month; BIO15, precipitation seasonality; BIO18, precipitation of warmest quarter; BIO19, precipitation of coldest quarter; number of people per km , Pop, population density; topographic slope in degrees, slope; slope-aspect (which is defined as the 2 1 compass direction to which a slope faces measured in degrees); altitude (meters); VI, vegetation index; TPP, terrestrial primary production in g C m d ; and TH24, topographic heterogeneity calculated for the 24 surroundings cells. Percentage of contribution to instability index Percentage of contribution to instability index Percentage of contribution to instability index 0 5 10 15 20 0 5 10 15 20 25 0 102030 Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 553 Figure 2. Mean6 SD frequencies (lines6 shaded areas) for each variable interval for the cells with presence records and the cells of the E2 geographical extent (all river basins of level 2 where the species is present). BIO1, annual mean temperature; BIO4, temperature seasonality; BIO12, annual precipitation; BIO14, pre- cipitation of driest month; and BIO19, precipitation of the coldest quarter. NASA’s Terra satellite. Specifically, monthly data of terrestrial net different extents may be influencing the provided results but con- primary productivity and VI from 2001 to 2010 were obtained by versely using different resolutions may complicate the estimation of averaging available information for each pixel of selected variables the effect that we try to measure (the extent) as showed by other using the statistical software RWizard (Guisande et al. 2014). The authors (Gillingham et al. 2012). net primary productivity indicates how much carbon dioxide is taken up by vegetation during photosynthesis minus how much car- bon dioxide is released when plants respire. The values indicate how Results fast carbon was taken in, or released, for every square meter of land 2 1 At the more restricted extent (E1), variables associated to tempera- over the indicated time span. Values range from 1.0 g C m d 2 1 ture (BIO8, BIO1, BIO3, and BIO4) and the altitude were the fac- to 6.5 g m d . A negative value means decomposition or respira- tors with the highest relative contribution to explain the tion exceeded carbon absorption; in other words, more carbon was distributions of each individual freshwater fish species (Figure 1 released into the atmosphere than was absorbed by the plants. We upper panel). Altitude and a mix of factors both associated with also include the VI as a productivity variable. This variable repre- temperature (BIO1 and BIO4) and precipitation (BIO19, BIO12, sents a measure of the greenness of Earth’s landscapes. and BIO14) become important factors at E2 (regional extent; see All these variables were included in the analyses at a resolution 0 0 of 5  5 (100 km ). The use of the same resolution for the medium panel in Figure 1). Lastly, some variables associated with Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 554 Current Zoology, 2018, Vol. 64, No. 5 until 13 C, approximately, but more frequent at temperatures from 13 Cto22 C(Figure 2). The frequencies of presence cells were higher at intermediate values of temperature seasonality (BIO4), but freshwater species seem to avoid areas with high temperature sea- sonality (Figure 2). In the case of BIO12 (annual precipitation), the frequencies of presence cells were lower than those of the geographi- cal extent in places with <1,200 mm, approximately, but higher at greater precipitation values. Both the precipitation of the driest month (BIO14) and the precipitation of the coldest quarter (BIO19) showed a similar pattern (Figure 2), with the frequencies of the pres- ence cells being lower at lower values and higher at intermediate val- ues than the frequencies of cells at the considered extent. The spatial variability in the relative contribution of each factor, both for all species (mean contribution) and single species, show that the same variable may exercise a different influence depending on the geographical location and the considered extent (Figure 3 upper panel). This spatial variability can be observed in all the con- sidered environmental variables (not shown). In fact, a clear latitudi- nal gradient can be observed in the variation of the percentage of contribution of some variables, which may vary depending on the considered extent (i.e., BIO14 and BIO19 see Figure 4). The contri- bution of an indirect variable such as altitude seems to be highly dependent on the considered extent, although is generally lower at the higher latitudes of the southern hemisphere. The contribution of annual mean temperature was rather homogeneous and erratic (Figure 4), while the variation in the contribution of precipitation variables (i.e., BIO19 and BIO14 in Figure 4) show a clear pattern. Tropical and subtropical basins were significantly less influenced by precipitation variables than northernmost and southernmost high- latitude basins. Interestingly, in the northern hemisphere the contri- bution of precipitation variables seems to be higher when the consid- ered extent increases. Figure 3. Map representing the geographical variation in the mean percent- age of contribution of precipitation of the coldest quarter (BIO19, upper panel) Discussion for all freshwater fish species in the cells of 1 , as measured by the instability index. These contribution scores were calculated using the SPEDInstabR In this study, we were able to demonstrate that the identification of algorithm comparing the values in the presence cells against those present in the factors accounting for the distribution range of the species all the cells of the E2 geographical extent (all river basins of level 2 where the clearly depend on the extent to which these relationships are exam- species is present). Percentage of contribution of altitude on the distribution ined, a topic which has long been hypothesized (MacArthur 1972; Esox lucius (medium panel) using the E1 geographical extent, and contribu- tion of annual precipitation (BIO12) for the species Hoplias malabaricus Wiens 2015) and that only recently has been considered fundamen- (lower panel) at the same extent. tal to correctly estimate the predictor functions relating species occurrences with environmental variables (Barve et al. 2011; precipitation (BIO19, BIO14, BIO12, and BIO18) were those that Acevedo et al. 2012). Our study also shows that the influence of had the highest contributions when using the world as extent (E3; explanatory variables changes spatially in a non-stationary way, as lower panel in Figure 1). Therefore, the contribution of the factors demonstrated in other studies (Osborne and Suarez-Seoane 2002; associated with precipitation seems to increase with the size of the Foody 2004; Bickford and Laffan 2006; Hortal et al. 2011; Mcnew considered extent, while the contribution of the factors associated et al. 2013; Holloway and Miller 2015). In our study, altitude is an with temperature variables increased as the size of the extent important factor that influences the smaller considered extent distri- decreases. bution patterns, but not at global extents. Factors associated to tem- The differences in the frequency of presence cells along each vari- perature are also important at local extent, whereas factors able gradient versus the frequency of the cells at the selected extent associated with precipitation seem to be more important at global can be geographically represented (Figure 2). The results for the E2 extents. Moreover, the relative contribution of factors to explain the situation were described, although the patterns were similar for the distribution of freshwater fish species varies from region to region, E1 and E3 extents (shown in Online Appendix 3, Figure A3.5 for E1 due partially to the different species present in each area, but mainly and Figure A3.6 for E3). Thus, presence cells seem to be compara- due to the spatial variation in the importance of the factors account- tively more frequent at lower altitudes (<250 m) and less frequent, ing for freshwater species’ distributions. This spatial variability in at altitudes from 300 m to 2,000 m, than the frequencies of the cells the importance of environmental factors has been formerly docu- of the E2 geographical extent (Figure 2). At higher altitudes mented in the case of fishes (Windle et al. 2010; Radinger et al. (>2,000 m) both frequencies seem to be similar. In the case of 2015). Here, a latitudinal and geographical gradient in the impor- annual mean temperatures (BIO1), presence cells were less frequent tance of abiotic factors has been shown. Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 555 Local Regional Global Local Regional Global -40 -20 0 20 40 -40 -20 0 20 40 Latitude Latitude Local Regional Global Local Regional Global -40 -20 0 20 40 -40 -20 0 20 40 Latitude Latitude Figure 4. Mean latitudinal contribution to the values of the instability index (%) of altitude, annual mean temperature (BIO1), precipitation of driest month (BIO14), and precipitation of coldest quarter (BIO19), depending on the considered geographical extent. Higher latitudes (90–45 both north and south) were not included in the plots due to the low number of species in those regions. The frequency distribution of presence data compared with those fishes should be managed with caution. The provided results are not existing in the occupied basins (Figure 2) is in agreement with classic only limited by potential errors in the number and characteristics of environmental gradients (Griffiths et al. 2014): the frequency of the used environmental variables (Van Neil et al. 2004), the exis- occurrence of freshwater fishes diminishes at temperatures below tence of false presence records (Tyre et al. 2003) or survey biases 13 C and precipitations lower than 1,200 mm, approximately. (Pelayo-Villamil et al. 2015), but also by the lack of consideration of However, freshwater fishes in tropical and subtropical basins are the historical biogeography of the species (Wiens and Donoghue comparatively less influenced in their distributions by precipitation 2004) or species interactions (Gonza ´ lez-Salazara et al. 2013). variables than those located in northernmost and southernmost However, in spite of this, our findings clearly suggest that different high-latitude basins. Interestingly, in the northern hemisphere the predictors can be detected as relevant at different scales. This fact contribution of precipitation variables seems to be higher when the has important implications for studies about SDMs and ENMs, for considered extent increases, so that precipitation variables increase which predictor selection is an important issue affecting models’ pre- in importance at higher latitudes, both south and north, although dictive ability (Austin 2007). We suggest when modeling species dis- this latitudinal trend is more or less pronounced depending on the tribution, that careful attention should be paid to the selection of the extent considered. Thus, when the variability in the climatic condi- extent used to estimate the SDMs and ENMs as previously suggested tions is minimized by considering only a geographical background (Barve et al. 2011; Acevedo et al. 2012), and that a single set of pre- limited to the basins in which a species occurs, the comparative rele- dictors for each species should not be used, but instead different pre- vance of precipitation variables in explaining occurrence and distri- dictors and predictor functions for each region where the species is bution seems to be higher than that of temperature under temperate present (Mcnew et al. 2013). These are not new issues and we are conditions. We may hypothesize that the higher levels of annual pre- aware of how its consideration complicates the building of models cipitation in tropical and subtropical basins suggest that precipita- capable of providing reliable estimations of species distributions. tion is not a limiting factor in these areas, so its effect on fish How to select the most appropriate extent for each species, and how distribution is not as important as in higher latitudes, where hydro- to manage the variation in the parameters of the environmental pre- logical variations related to precipitation may be a more limiting dictors obtained at different extents to generate reliable regional pre- factor when compared with temperature. dictions are key issues to address (see, e.g., Sua ´ rez-Seoane et al. Of course, the procedure proposed to identify the most impor- 2014). Be that as it may, modelers should justify how they have tant factors in determining the occurrence of individual freshwater managed these 2 questions or why it has not done so. BIO14 Altitude 2 4 6 8 10 12 14 16 24 68 10 12 BIO19 BIO1 46 8 10 12 14 567 8 9 10 Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 556 Current Zoology, 2018, Vol. 64, No. 5 Guisande C, Garcı´a-Rosello ´ E, Heine J, Gonza ´ lez-Dacosta J, Goza ´ lez Vilas L Supplementary Material et al., 2016. SPEDInstabR: an algorithm based on a fluctuation index Supplementary material can be found at https://academic.oup.com/cz. for selecting predictors in species distribution modeling. Ecol Inform 37: 18–23. Guisande C, Manjarre ´ s-Herna ´ ndez A, Pelayo-Villamil P, Granado-Lorencio References C, Riveiro I et al., 2010. IPez: an expert system for the taxonomic identifica- Acevedo P, Jimenez-Valverde A, Lobo JM, Real R, 2012. Delimiting the geo- tion of fishes based on machine learning techniques. Fish Res 102:240–247. graphical background in species distribution modelling. J Biogeogr 39: Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A, 2005. Very high reso- 1383–1390. lution interpolated climate surfaces for global land areas. Int J Climatol 25: Austin M, 2007. Species distribution models and ecological theory: a critical 1965–1978. assessment and some possible approaches. Ecol Model 200:1–19. Holloway P, Miller JA, 2015. Exploring spatial scale, autocorrelation and Barve N, Barve V, Jime ´ nez-Valverde A, Lira-Noriega A, Maher SP et al., nonstationarity of bird species richness patterns. ISPRS Int J Geo-Inform 4: 2011. Crucial role of the accessibility area in ecological niche modeling and 783–798. species distribution modeling. Ecol Model 222:1810–1819. Hortal J, Diniz-Filho JAF, Bini LM, Rodriguez MA, Baselga A et al., 2011. Ice Bickford S, Laffan S, 2006. Multi-extent analysis of the relationship between age climate, evolutionary constraints and diversity patterns of European pteridophyte species richness and climate. Glob Ecol Biogeogr 15:588–601. dung beetles. Ecol Lett 14:741–748. Brown JH, 1995. Macroecology. Chicago (IL): University of Chicago Press. Johnson DH, 1980. The comparison of usage and availability measurements Chust G, Petrus JL, Ducrot D, Ventura D, 2004. Scale dependence of insect for evaluating resource preference. Ecology 61:65–71. assemblages in response to landscape pattern. Landsc Ecol 18:41–57. Lobo JM, 2016. The use of occurrence data to predict the effects of climate Dormann CF, Elith J, Bacher S, Buchmann C, Carl G et al., 2013. Collinearity: change on insects. Curr Opin Insect Sci 17:62–68. a review of methods to deal with it and a simulation study evaluating their MacArthur RH, 1972. Geographic Ecology. New York (NY): Harper and performance. Ecography 36:27–46. Row. Dubois DM, 1973. An index of fluctuations, Do, connected with diversity and Margalef R, 1968. Perspectives in Ecological Theory. Chicago (IL): The stability of ecosystems: applications in the Lotka-Volterra model and in an University of Chicago Press. experimental distribution of species. Rapport de sythe ` se III, Programme Mcnew LB, Gregory AJ, Sandercock BK, 2013. Spatial heterogeneity in habi- National sur l’environment Physique et Biologique, Project Mer. tat selection: nest site selection by greater prairie-chickens. J Wildl Manage Commision Interministe ´ rielle de la Politique Scientifique. Lie ` ge. 77:791–801. Foody GM, 2004. Spatial nonstationarity and scale-dependency in the rela- Niamir A, Skidmore AK, Toxopeus AG, Real R, 2016. Use of taxonomy to tionship between species richness and environmental determinants for the delineate spatial extent of atlas data for species distribution models. Glob sub-Saharan endemic avifauna. Glob Ecol Biogeogr 13:315–320. Ecol Biogeogr 25:227–237. Fotheringham AS, Brundson C, Charlton M, 2002. Geographically Weighted Osborne PE, Suarez-Seoane S, 2002. Should data be partitioned spatially Regression: The Analysis of Spatially Varying Relationships. West Sussex, before building large-scale distribution models? Ecol Model 157:249–259. UK: John Wiley and Sons, Ltd. Pearson RG, Dawson TP, 2003. Predicting the impacts of climate change on Garcı´a-Rosello ´ E, Guisande C, Gonza ´ lez-Dacosta J, Heine J, Pelayo-Villamil P the distribution of species: are bioclimate envelope models useful? Glob et al., 2013. ModestR: a software tool for managing and analyzing species Ecol Biogeogr 12:361–371. distribution map databases. Ecography 36:1202–1207. Pelayo-Villamil P, Guisande C, Gonza ´ lez-Vilas L, Carvajal-Quintero JD, Garcı´a-Rosello ´ E, Guisande C, Heine J, Pelayo-Villamil P, Manjarre ´s- Jime ´ nez-Segura LF et al., 2012. ModestR: Una herramienta informa ´ tica Herna ´ ndez A et al., 2014. Using ModestR to download, import and clean para el estudio de los ecosistemas acua ´ ticos de Colombia. Actual Biol 34: species distribution records. Methods Ecol Evol 5:703–713. 225–239. Garcı´a-Rosello ´ E, Guisande C, Manjarre ´ s-Herna ´ ndez A, Gonza ´ lez-Dacosta J, Pelayo-Villamil P, Guisande C, Vari RP, Manjarre ´ s-Herna ´ ndez A, Garcı´a- Heine J et al., 2015. Can we derive macroecological patterns from primary Rosello ´ E et al., 2015. Global diversity patterns of freshwater fishes: poten- GBIF data? Glob Ecol Biogeogr 24:335–347. tial victims of their own success. Divers Distrib 21:345–356. Gillingham PK, Huntley B, Kunin WE, Thomas CD, 2012. The effect of spatial Peterson AT, Sobero ´ n J, Pearson RG, Anderson R, Martı´nez-Meyer E et al., resolution on projected responses to climate warming. Divers Distrib 18: 2011. Ecological Niches and Geographic Distributions. Princeton (NJ): 990–1000. Princeton University Press. Gaston K, 2003. The Structure and Dynamics of Geographic Ranges. Oxford: Radinger J, Wolter C, Kail J, 2015. Spatial scaling of environmental variables Oxford University Press. improves species–habitat models of fishes in a small, sand-bed lowland river. Gonza ´ lez-Salazara C, Stephensb CR, Marquet PA, 2013. Comparing the rela- PLoS One 10:e0142813. tive contributions of biotic and abiotic factors as mediators of species’ distri- Sobero ´ n J, Peterson AT, 2005. Interpretation of models of fundamental eco- butions. Ecol Model 248:57–70. logical niches and species’ distributional areas. Biodivers Inform 2:1–10. Gonza ´ lez-Vilas L, Guisande C, Vari RP, Pelayo-Villamil P, Manjarre ´s- Steffan-Dewenter I, Mu ¨ nzenberg U, Burger C, Thies C, Tscharntke T, 2002. Herna ´ ndez A et al., 2016. Geospatial data of freshwater habitats for macro- Scale-dependent effects of landscape context on three pollinator guilds. ecological studies: an example with freshwater fishes. Int J Geogr Inf Sci 30: Ecology 83:1421–1432. 126–141. Sua ´ rez-Seoane S, Virgo ´ s E, Terroba O, Pardavila X, Barea-Azco ´ n JM, 2014. Griffiths D, McGonigle C, Quinn R, 2014. Climate and species richness pat- Scaling of species distribution models across spatial resolutions and extents terns of freshwater fish in North America and Europe. J Biogeogr 41: along a biogeographical gradient: the case of the Iberian mole Talpa occi- 452–463. dentalis. Ecography 37:279–292. Guisande C, 2016. Estimation of the relative importance of factors affecting Tyre AJ, Tenhumberg B, Field SA, Niejalke D, Parris K et al., 2003. Improving species distribution based on stability concept. R package version 1.3. precision and reducing bias in biological surveys: estimating false-negative Available from: http://CRAN.R-project.org/package¼SPEDInstabR error rates. Ecol Appl 13:1790–1801. (accessed 6 November 2016). Van de Pol M, Bailey LD, McLean N, Rijsdijk L, Lawson CR et al., 2016. Guisande C, Heine J, Gomza ´ lez-DaCosta J, Garcı´a-Rosello ´ E, 2014. Vigo, Identifying the best climatic predictors in ecology and evolution. Methods Spain: RWizard Software. University of Vigo. Ecol Evol 7:1246–1257. Guisande C, Barreiro A, Maneiro I, Riveiro I, Vergara-Castano ~ AR et al., Van Neil KP, Laffan SW, Lees BG, 2004. Effect of error in the DEM on envi- 2006. Tratamiento de Datos. Madrid: Ediciones Dı ´az de Santos. Guisande C, Barreiro A, Vaamonde A, 2011. Tratamiento de Datos con R, ronmental variables for predictive vegetation modelling. J Veg Sci 15: Statistica y SPSS. Madrid: Ediciones Dı´az de Santos. 747–756. Downloaded from https://academic.oup.com/cz/article-abstract/64/5/549/4596537 by Ed 'DeepDyve' Gillespie user on 18 October 2018 Manjarre ´ s-Herna ´ ndez et al.  Non-stationary abiotic predictors 557 Wiens JJ, 2015. The niche, biogeography and species interactions. Philos Windle MJS, Rose GA, Devillers R, Fortin MJ, 2010. Exploring spatial Trans R Soc B 366:2336–2350. non-stationarity of fisheries survey data using geographically weighted Wiens JJ, Donoghue MJ, 2004. Historical biogeography, ecology and species regression (GWR): an example from the Northwest Atlantic. ICES J Mar Sci richness. Trends Ecol Evol 19:639–644. 67:145–154.

Journal

Current ZoologyOxford University Press

Published: Oct 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off