COMMUNITY AND ECOSYSTEM ECOLOGY

Diversity Patterns in Iberian Calathus (Coleoptera, Carabidae: Harpalinae):Species Turnover Shows a Story Overlooked by Species Richness

ISRAEL GANAN,1 ANDRES BASELGA,2 AND FRANCISCO NOVOA1

Environ. Entomol. 37(6): 1488Ð1497 (2008)

ABSTRACT We assessed the relationships between diversity patterns of Iberian Calathus andcurrent environmental gradients or broad-scale spatial constraints, using 50-km grid cells as samplingunits. We assessed the completeness of the inventories using nonparametric estimators to avoidspurious results based on sampling biases. We modeled species richness and beta diversity, using spatialposition, and 23 topographical, climatic, and geological variables as predictors in regression andconstrained analysis of principal coordinates modeling. Geographical situation does not seem to affectCalathus species richness, because no spatial pattern was detected. The environmental variables onlyexplained 23% of the variation in richness. Spatial and environmental predictors explained a large partof the variation in species composition (58%). The fraction shared by both groups of variables wasrelatively large, but the pure effect of each model was still important. Our results show that it isnecessary to assess the completeness of inventories to avoid drawing false conclusions. Also, IberianCalathus represent a clear example of the need for combined analyses of species richness and betadiversity patterns, because the lack of patterns in the former does not imply the invariance of bioticcommunities.

KEY WORDS Calathus, Iberian Peninsula, completeness of inventories, species richness, betadiversity

Most studies on patterns of species richness and betadiversity have been conducted on well-studied ver-tebrate and plant groups (Baquero and Tellerõa 2001,Lobo et al. 2001, Rodrõguez and Arita 2004, Rodrõguezet al. 2005, Gaston et al. 2007, Montoya et al. 2007,Svenning and Skov 2007b, Whittaker et al. 2007), eventhough the highest levels of biodiversity are foundamong invertebrates (Erwin 1991, Odegaard 2000). Inthe Palaearctic region, despite existing work on Eu-ropean Lepidoptera (Dennis et al. 1998, Hawkins andPorter 2003) or Coleoptera groups as dung beetles(Lobo et al. 2002, Lobo and Martõn-Piera 2002, Loboet al. 2004, Cabrero-Sanudo and Lobo 2006) and long-horn beetles (Baselga 2008), more invertebrate stud-ies of this type are still needed. However, inverte-brates are rarely selected as study organisms inbiogeographical research because of the scarcity ofpublished faunistic information. The lack of informa-tion leads to geographical biases in biodiversity pat-terns derived from raw databases (Hortal et al. 2007),and it is obvious thatbiasedbiodiversitypatternscouldproduce spurious correlations with environmentaland spatial variables. It is thus important to determine

to what extent the inventories are complete, a meth-odology that has been widely explored (Soberon andLlorente 1993, Soberon et al. 2000, Hortal et al. 2001,Martõn-Piera and Lobo 2003, Petersen et al. 2003,Meier and Dikow 2004, Baselga and Novoa 2006, Hor-tal et al. 2006, Lobo 2008).

In those regions for which the sampling effort couldbe considered adequate, the assessment of speciesrichness (alpha diversity) and the variation in speciescomposition (beta diversity) can be addressed as ameans of determining which factors are responsible ofdiversity patterns. The relative importance of currentand past environmental conditions, neutral processesand the evolutionary history of taxa as causal driversof diversity patterns have not yet been clariÞed (Rick-lefs 2006). Strong correlations have been found be-tween species richness patterns and current environ-mental conditions (Hawkins et al. 2003, Whittaker etal. 2007). However, the causal link is not clear and,although richnessÐenvironment correlations arestrong, they could simply result because both variablesare similarly structured in space (Bahn and McGill2007, Currie 2007). In the case of beta diversity, muchless research has been devoted to its large-scale pat-terns (but see Rodrõguez and Arita 2004, Qian et al.2005, Gaston et al. 2007, McKnight et al. 2007, Qian andRicklefs 2007, Baselga 2008), but variation in speciescomposition is usually also related to environmentalgradients. However, purely spatial patterns indepen-dentof climaticconditionsare ingeneralmoremarked

1 Departamento de Zoologõa, Facultad de Biologõa, Universidad deSantiago de Compostela, c/ Lope Gomez de Marzoa s/n, 15706 San-tiago de Compostela, Spain.

2 Corresponding author: Andres Baselga, Departamento de Biodi-versidad y Biologõa Evolutiva, Museo Nacional de Ciencias Naturales-CSIC, c/ Jose Gutierrez Abascal, 2, 28006 Madrid, Spain (e-mail:baselga@mncn.csic.es).

0046-225X/08/1488Ð1497$04.00/0 � 2008 Entomological Society of America

in composition (Qian et al. 2005, Baselga and Jimenez-Valverde 2007, Baselga 2008) than in richness patterns,suggesting a greater importance of neutral dispersionand historical effects on beta diversity. Also, historicalfactors such as past climate (Svenning and Skov 2007b,Araujo et al. 2008), climatic stability (Hawkins et al.2006), and postglacial recolonization processes (He-witt 2004, Svenning and Skov 2007a) have been shownto be correlated with the observed patterns of speciesrichness and beta diversity (Graham 2006). Therefore,macroecological analyses such as this one need toconsider the inßuence of environmental factors (bothpresent and past ones) and historical factors as neutraldispersal processes or evolutionary patterns in biodi-versity patterns (Ricklefs 2006).

The following macroecological analysis concen-trates on the species of a large ground beetle genusthat are present in the Iberian Peninsula and theBalearic Islands (Fig. 1A). The genusCalathusBonelliconsists of 177 species arranged in 10 subgenera. It isrepresented throughout Holarctica and in the moun-tains of the northern portion of both the Oriental andNeotropical Regions (Ball and Bousquet 2001). Al-though recorded from many different types of habi-tats, most of the members of this genus, including,above all, those with the most restricted distributions,are associated with upland areas. In the Iberian Pen-insula, there are 23 species, 14 (70%) of which areendemic to the peninsula (Ganan and Novoa 2005).The presence of so many endemic species is closelylinked to the role of the Iberian Peninsula as glacialrefuge, because glaciated areas are obviously affectedby episodes of extinction (Willis and Niklas 2004) andare less prone to witness quite as many speciationprocesses (Ribera and Vogler 2004). The Iberian Pen-insula was able to conserve much of its fauna duringthis period, but once the ice retreated, the Pyreneesand other barriers halted the spread of many Iberianelements into the rest of Europe (Hewitt 1999, 2004).IberianCalathus assemblages are thus a good model toassess the relevance of different factors on richnessgradients and turnover processes for taxa with a greatproportion of southern endemics. This kind of taxarepresents the most peculiar faunas of Europe and isthe main source of European biodiversity as well.

In sum, this study aims to assess the interrelationbetween diversity patterns and current environmentalgradients or broad-scale spatial constraints that couldreßect historical diversiÞcation processes, dispersallimitation, or stochastic colonization events. There-fore, a spatially explicit analysis was performed,searching for the pure effects of environment or spa-tial position and the shared effects of the spatiallystructured environmental gradients (Borcard et al.1992). The rationale is that spatial variables submittedafter environmental ones can aid in detection of ef-fects that have not been accounted for, such as geo-graphical or historical effects (Ricklefs 2006), as wellas to measure the spatial structure of the environmen-tal variables. The latter is important because the bioticvariation explained by the shared fraction of environ-ment and space cannot be unambiguously attributed

to any of the predictors (Currie 2007). This analysisaims to test (1) the relevance of data quality (i.e.,inventory completeness) on the modeling of biodi-versityÐenvironment relationships and (2) the twofollowing hypotheses: richness gradients of IberianCalathus are related with contemporary water andenergy variables, as has been found in other taxonomicgroups (Hawkins et al. 2003); and species turnoverpatterns in Iberian Calathus are independent of rich-ness gradients, being determined by both environ-mental factors and spatial constraints that reßectdispersal limitation of assemblages.

Materials and Methods

Analysis of the Completeness of Existing Invento-ries. To describe patterns of diversity and analyze itscauses, good-quality data are essential. Biological in-ventories must be complete enough to reßect realpatterns of biodiversity, because the use of incompleteinventories can lead to erroneous conclusions (Gotelliand Colwell 2001). As a Þrst step, a database contain-ing all recordsÑfrom collections and from the existingbibliographyÑfor each species was built. Each recordcontained the following Þelds: species, locality, Uni-versal Transversal Mercator (UTM) coordinates, date,and collector or bibliographical source. Any change ina Þeld creates a new record. Greater sampling effortproduces more records (Martõn-Piera and Lobo 2003,Lobo 2008); thus, the number of records was used asa surrogateorproxy for the samplingeffort carriedout.In total, 5,475 recordsÑ2,864 from studied material(10,864 individuals) and 2,607 from bibliographicalsourcesÑwere included. All records with erroneousor doubtful data were excluded.

Once the database was complete, each record wassituated on a map of the Iberian Peninsula and theBalearic Islands divided up into 50-km UTM grid cells(275 grid cells of 2,500 km2) using IDRISI (CLARKLABS 2000) to assess the presence and the number ofrecords for each species in each grid cell. The ob-served richness in each grid cell is provided in Fig. 1B.A total of 80 grid cells with less than eight records orwith a land surface area of �1,390 km2 were omittedfrom the analysis. To quantify the degree of complete-ness of the inventories from each grid cell, the fournonparametric estimators recommended by Hortal etal. (2006) were calculated: ICE, Chao 2, Jackknife 1,and Jackknife 2. These estimators were computed us-ing EstimateS 7.5 (Colwell 2005). The mean of thesefour estimators was taken as a good estimate of the realspecies richness of each grid cell. An arbitrary cut-offpoint was established, whereby those grid cells inwhich the observed species richness was equal to or�85% of the estimated species richness were consid-ered to be well sampled: the inventories for these gridcells were thus used in the subsequent analyses.Species Richness Analysis.Taking into account only

well sampled grid cells and their estimated speciesrichness, a total of 25 spatial (UTM X and Y coordi-nates), topoclimatic (factors related to altitude, tem-perature, and precipitation), and geological (percent-

December 2008 GANAN ET AL.: DIVERSITY PATTERNS IN IBERIAN Calathus 1489

age of acidic and basic rocks) variables (Table 1) wereanalyzed to test their relative importance in deter-mining species richness in the genus Calathus. Theclimatic variables were obtained from WorldClim 1.4layers (Hijmans et al. 2005), and mean values werecalculated for each grid cell (with a resolution of 8.3

km) using IDRISI (CLARK LABS 2000). Geologicalvariables were obtained by digitizing an Iberian map(Instituto GeograÞco Nacional 1995). As spatial vari-ables, the nine terms of a third degree polynomial ofthe X and Y coordinates were used (trend surfaceanalysis, see Legendre and Legendre 1998).

Fig. 1. Study area and geographical patterns of species richness and species composition using 50-km UTM grid cells assampling units. (A) Study area and main physiographic features cited in the text. (B) Scores of observed species richness(gray tones). (C) Scores of estimated completeness (gray tones) and grid cells lacking enough data (white). (D) Scores ofestimated species richness in well sampled grid cells (gray tones) and grid cells discarded for analyses (white). (E) Mainfaunal groups yielded by the cluster analysis based on the NMDS performed with Simpson similarity measure. (F) Secondaryfaunal groups yielded by the former analysis using a more restricted cut-off point.

1490 ENVIRONMENTAL ENTOMOLOGY Vol. 37, no. 6

The relationships between estimated species rich-ness and the 25 predictor variables were performed bycalculating multiple regression models (Legendre andLegendre 1998) with Statistica 6.0 (StatSoft 2001).First of all, simple regression models were constructedfor the estimated richness against linear, quadratic,and cubic equations of each variable to determineindependently signiÞcant relationships for each one.Using only those variables that were signiÞcant in thesimple regressions, a new model for each group ofvariables (space, topography, climate, and soils) wasconstructed with variables selected for each of theÞnal models by means of a backward stepwise proce-dure. Finally, a partition of the variance explained bya complete model (space � topography � climate �soils) was performed to determine both the pure andcombined effects of each group of variables (Leg-endre and Legendre 1998). Such an approach allowsus to calculate how much of the total variance isexplained, as well as identify the portions of explainedvariability that are shared by different sets of variablesand those that are independent (Borcard et al. 1992).

As an assessment of the effect of sampling biases onthe observed richness patterns and how these biasescould affect the picture of spatial and environmentalcorrelates, we have also performed the same simpleregression models mentioned above, but using the 233gridcellswithat leastone record.Therefore,we testedfor the relationship between observed richness withlinear, quadratic, and cubic functions of each spatialand environmental variable. Just for comparison, be-cause we consider these relationships to be spuriousbecause of sampling biases, a multiple regressionmodel was constructed using a backward stepwiseprocedure, but variation was not partitioned.

Species TurnoverAnalysis.The variation in the spe-cies composition (beta diversity) between well-sam-pled grid cells was analyzed. On the basis of a pres-ence-absence matrix of the species found in each gridcell, a distance matrix using SimpsonÕs index of betadiversity (�sim) was calculated using PAST v. 1.77(Hammer et al. 2001). This index has the property ofbeing independent of richness gradients between gridcells and thus can be used to search for differences inspecies composition that are not caused by differencesin species richness (Koleff et al. 2003). This is a wayof guaranteeing that richness patterns (alpha diver-sity) do not bias the observed patterns of speciesturnover (beta diversity) (Baselga 2007, 2008).

An ordination of the grid cells was based on thefaunal distance matrix, which was submitted to a non-metric multidimensional scaling analysis (NMDS)(Legendre and Legendre 1998) performed with Sta-tistica 6.0 (StatSoft 2001). The number of dimensionswas selected using a scree-plot procedure: stress valuedecreased from 0.227 for two dimensions to 0.092 forsix dimensions. Finally, Þve dimensions were ex-tracted (stress � 0.109). The position of each grid cellin all Þve dimensions was submitted to a cluster anal-ysis to identify the main faunal groups. A dendrogramwas drawn with Statistica 6.0, taking the Euclideandistance in that Þve-dimensional space as the mea-surement of resemblance between the grid cells andan unweighted pair-group average as the linkage rule.The signiÞcance of the faunal groups yielded by thecluster analysis was assessed by means of the ANOSIMtests computed with R (R Development Core Team2006) using the vegan package (Oksanen et al. 2007).

Next, the spatial and environmental variables inßu-encing patterns of beta diversity were searched for. Aconstrained analysis of principal coordinates (CAP)was performed in R using the “capscale” command oftheveganpackage(Oksanenetal. 2007) toanalyze therelationships between variation in species composi-tion in the studied grid cells and two sets-spatial andenvironmental of predictor variables. CAP was se-lected as the-ordination procedure because it allowsany dissimilarity index to be used as a measurement ofdistance. Thus, �sim, which is not inßuenced by rich-ness variation, can be selected.

The same predictor variables were used as in therichness analysis, considering all the environmentalvariables as a single set (climate�altitude�geology).As spatial variables, the nine terms of a third degreepolynomial of the X and Y coordinates were used(trend surface analysis, see Legendre and Legendre1998). These variables were used as predictors to ex-plain the ordination and yielded a spatial and an en-vironmental model. Given that the order of inclusionof the variables in the model affects the signiÞcanceobtained from the permutation tests (vegan command“permutest”; Oksanen et al. 2007), the amount bywhich the explained variation was reduced because ofthe elimination of a single variable (compared withthe complete model) was tested before the Þnal anal-ysis. This allowed the variables to be ranked in orderof their independent contribution to the total variance

Table 1. Description of predictor variables used in the analyses

Code Description

X Mean UTM X coordinateY Mean UTM Y coordinateAlt Mean altitudeAlt_Range Altitudinal rangeBIO1 Annual mean temperatureBIO2 Mean diurnal range (mean of monthly maximum

temperature � minimum temperature)BIO3 Isothermality (BIO2/BIO7) (�100)BIO4 Temperature seasonality (SD � 100)BIO5 Maximum temperature of the warmest monthBIO6 Minimum temperature of the coldest monthBIO7 Temperature annual range (BIO5-BIO6)BIO8 Mean temperature of the wettest quarterBIO9 Mean temperature of the driest quarterBIO10 Mean temperature of the warmest quarterBIO11 Mean temperature of the coldest quarterBIO12 Annual precipitationBIO13 Precipitation of the wettest monthBIO14 Precipitation of the driest monthBIO15 Precipitation seasonality (coefÞcient of variation)BIO16 Precipitation of the wettest quarterBIO17 Precipitation of the driest quarterBIO18 Precipitation of the warmest quarterBIO19 Precipitation of the coldest quarterAcid Percentage of surface occupied by acid rocksBasic Percentage of surface occupied by basic rocks

December 2008 GANAN ET AL.: DIVERSITY PATTERNS IN IBERIAN Calathus 1491

in the response variable (from greatest to least). Thevariables were introduced into the model in this orderand their signiÞcance calculated using permutationtests. Only signiÞcant variables were retained (P �0.05) to avoid overÞtting caused by the inclusion ofnonsigniÞcant terms. Finally, to calculate the fractionsof variation explained by each model a partial CAP wasperformed, eliminating the effects of each model (co-variable) on the other to determine the pure effect ofeach one (Borcard et al. 1992).

Results

Completeness of Inventories. The inventories of 72grid cells reached a degree of completeness �85%(Fig. 1C) and thus were considered to be well sam-pled. It is clear that the majority of well-sampledsquares coincide with the mountainous regions of theGalaico-Duriense massif, the Basque Mountains, thePyrenees, the Spanish Central System, the CatalonianMountains, and both Iberian systems. In addition, aclear difference between the north and the south ofthe Peninsula in terms of the sampling effort is shown:in the south, only the Sierra Nevada and the northeastof the Betic Mountains can be said to have been wellsampled. Elsewhere, it is clear that the Balearic Islandshave not been well prospected and that only the Sierrade Monchique has been well sampled in the southwestof the Peninsula.

These results show that there are many areas of theIberian Peninsula and Balearic Islands that to datehave been insufÞciently prospected and whose biodi-versity is still relatively poorly known. On the basis ofthe analysis of the completeness of the inventories, wecan elicit the fact that the situation of the best sampledareas is slightly biased to mountainous regions (meanaltitude of all grid cells � 663 m asl, mean altitude ofwell sampled grid cells � 833 m), a difference beingsigniÞcant (t � 3.30, P � 0.001). Likewise, samplingeffort is biased to the northern half of the IberianPeninsula (mean Y of all grid cells � 4,503,673 m, meanY of well sampled grid cells � 4,567,361 m, t� 2.21,P�0.03).Patterns of Species Richness. Figure 1D shows the

estimated richness of each of those well-sampled gridcells. The analysis of the well-sampled grid cellsshowed that only altitude, altitudinal range, annualmean temperature (Bio1), daily temperature range(Bio2), mean temperature of the wettest season(Bio8), mean temperature of the coldest season(Bio11), and the percentage of acidic rocks (Acid)were signiÞcant when they were regressed indepen-dently against estimated species richness (Table 2).With the exception of the model for Bio2, which es-tablished a quadratic relationship with estimated rich-ness, all the other relationships with richness werelinear, with positive correlations for Alt and Acid andnegative for Bio1, Bio8, and Bio11.

Nevertheless, geographical situation did not seemto affect Calathus species richness, because no spatialpattern correlated to latitude or longitude was de-tected. Environmental models for topography (with

Alt andAlt_Range), geology (Acid), and climate (Bio1,Bio8,andBio11) were constructed; the backward step-wise procedure only retained Alt_Range in the topo-graphic model and Bio8 in the climatic model.

A complete environmental model built with allthree variables only explained 23% of the variation inrichness, leaving 77% unexplained (Fig. 2A). The pureeffect of topography independent of climate and ge-ology explained 5.8% of the variation in richness andother relatively important fraction were the sharedeffects between climate with geology (5.2%) and cli-mate and topography (4.1%). The remaining fractionsexplained smaller amounts of the variation in richness.As the lack of spatial structure in the dependent vari-able suggests, the residuals of the environmentalmodel also lack any spatial structure and were notsigniÞcantly related to any of the nine terms of a trendsurface analysis (all R2 � 0.01, F1,70 � 0.65, P� 0.42).Also, residuals were not spatially autocorrelated. Mo-ranÕs I test computed in SAM (Rangel et al. 2006) foreight distance classes yielded values of MoransÕs Ibetween �0.095 and 0.063 (all P � 0.11).

Table 2. Relationships between predictors and estimated spe-cies richness and models for each group of variables, consideringonly well-sampled grid cells (n � 72)

Variable Function (sign) R2 (%) F df P

X NS 3.7 2.684 1,70 0.106Y NS 1.0 0.708 1,70 0.403XY NS 3.0 2.18 1,70 0.145X2 NS 4.1 2.99 1,70 0.088Y2 NS 1.0 0.67 1,70 0.415X2Y NS 3.7 2.70 1,70 0.105XY2 NS 2.4 1.70 1,70 0.196X3 NS 4.2 3.10 1,70 0.083Y3 NS 0.9 0.64 1,70 0.427Alt Linear (�) 7.2 5.405 1,70 0.023Alt_range Linear (�) 12.3 9.80 1,70 0.003Bio1 Linear (�) 6.5 4.848 1,70 0.031Bio2 Quadratic (�, �) 9.5 3.604 2,69 0.032Bio3 NS 0.4 0.308 1,70 0.581Bio4 NS 0.5 0.350 1,70 0.556Bio5 NS 1.3 0.890 1,70 0.349Bio6 NS 5.2 3.825 1,70 0.054Bio7 NS 0.3 0.244 1,70 0.623Bio8 Linear (�) 15.7 13.054 1,70 0.001Bio9 NS 0.0 0.001 1,70 0.976Bio10 NS 4.2 3.075 1,70 0.084Bio11 Linear (�) 6.7 5.020 1,70 0.028Bio12 NS 2.3 1.665 1,70 0.201Bio13 NS 2.4 1.738 1,70 0.192Bio14 NS 0.1 0.104 1,70 0.748Bio15 NS 0.0 0.001 1,70 0.970Bio16 NS 2.4 1.725 1,70 0.193Bio17 NS 0.2 0.146 1,70 0.704Bio18 NS 0.0 0.002 1,70 0.966Bio19 NS 0.0 0.002 1,70 0.966Acid Linear (�) 9.2 7.125 1,70 0.009Basic NS 0.3 0.176 1,70 0.676Model for T Alt_range 12.3 9.80 1,70 0.003Model for C Bio8 15.7 13.054 1,70 0.001Model for G Acid 9.2 7.125 1,70 0.009Model for T

� C � GAlt_range � Bio8

� acid22.8 6.711 3,68 �0.001

T, C, and G are the topographic, climatic, and geological models,respectively.

NS, not sigiÞcant.

1492 ENVIRONMENTAL ENTOMOLOGY Vol. 37, no. 6

Richness PatternsDerived fromIncompletely Sam-pled Grid Cells. The analysis of all grid cells with atleast one record showed a different picture (Table 3).Considering this data, the variation in species richnesswas signiÞcantly related with almost all the variablesconsidered as potential predictors, with exception oflongitude (X), its quadratic and cubic terms and itsinteraction with latitude (Y), daily temperature range(Bio2), temperature seasonality (Bio4), temperature

annual range (Bio7), and percentage of basic rocks(Basic). Therefore, the use of all available data with-out considering the completeness of inventories yieldssigniÞcant relationship between the observed speciesrichness and latitude, altitude, temperature, precipi-tation, and the percentage of acid rocks. A completemodel with variables selected using a backward step-wise procedure explained 41.4% of the variation inrichness (F � 14.18; df � 11,221; P � 0.001).Patterns of Beta Diversity. The dendrogram built

after performing a Þve-dimensional NMDS (stress �0.109) yielded Þve main faunal groups present in the72 grid cells in the Iberian Peninsula that were ana-lyzed (Fig. 1E; ANOSIM R � 0.707, P � 0.001): (1)Pyrenees, Basque Mountains, and central part of theCantabrian mountains; (2) Galaico-Duriense massif;(3) western part of the northern Iberian Mountains,eastern part of the Cantabrian Mountains and Sierrade Monchique; (4) a less well-deÞned group that en-compasses mountainous areas in the Betic and Peni-betic Mountains, the Sistema Central, the Catalonianmountains, the southern part of the northern IberianMountains, eastern part of the Galaico-Duriense mas-sif, and parts of the northern Meseta; and (5) southernpart of the Iberian Mountains and some parts of theEbro Valley.

Nevertheless, if we restricted the cut-off point, theÞrst four faunal groups were subdivided (Fig. 1F;ANOSIM R� 0.799, P� 0.001) and yielded a total of10 groups: (1) central and eastern Pyrenees; (2) west-ern Pyrenees, Basque Mountains, and central part ofthe Cantabrian Mountains; (3) mountainous areas ofthe Galaico-Duriense massif; (4) coastal areas in thenortheast of the Peninsula; (5) eastern part of theCantabrian Mountains; (6) western part of the north-ern Iberian Mountains and Sierra de Monchique; (7)western part of the Betic Mountains, the Sistema Cen-tral, part of the northern Iberian Mountains, someareas of the Galaico-Duriense massif, and a small partof the Cantabrian Mountains; (8) mountainous areasof the Mediterranean coastline; (9) eastern part ofBetic and Penibetic Mountains and some areas of thenorthern Meseta; and (10) the southern IberianMountains and parts of the Ebro Valley.

After describing the patterns in the variation ofspecies composition (beta diversity), the factors thatdetermine this variation were analyzed. Faunal com-position was not signiÞcantly related with richness(pseudo-F� 0.35, P� 0.359), and as such, the patternsof beta diversity are not correlated to the patterns ofalpha diversity (species richness). The analysisshowed that the two groups of variables used (spatialand environmental) yielded signiÞcant contributionsto the explanation of the variation in species compo-sition (Table 4). Regarding spatial variables, all theterms of the third grade polynomial of X and Y coor-dinates (with the exception of Y and X2) were signif-icant, whereas for the environmental variables, Alt,Bio1, Bio2, Bio5, Bio8, Bio10, Bio11, Bio12, Bio14, Bio15,Bio16, Bio17, and Acid were signiÞcant. Nonetheless,to avoid spurious interpretations caused by comparingmodels with different numbers of predictor variables,

Fig. 2. (A) Partitioning of the variation (%) in speciesrichness among groups of explanatory variables: T, C, and Gare the topographic, climatic, and geological models, respec-tively. (B) Partitioning of the variation (%) in faunal com-position among groups of explanatory variables: E and S arethe environmental and spatial models, respectively. In bothcases, U is the unexplained fraction and gray tones are pro-portional to the amount of explained variation.

Table 3. Relationships between predictors and observed spe-cies richness, considering all grid cells with at least one record (n �232)

Variable Function (sign) R2 (%) F df P

X NS 0.5 1.20 1,231 0.274Y Linear (�) 7.7 19.17 1,231 �0.001XY NS 1.5 3.55 1,231 0.061X2 NS 0.5 1.12 1,231 0.291Y2 Linear (�) 7.6 19.00 1,231 �0.001X2Y NS 0.9 2.02 1,231 0.156XY2 Linear (�) 2.5 5.99 1,231 0.015X3 NS 0.4 0.85 1,231 0.358Y3 Linear (�) 7.5 18.78 1,231 �0.001Alt Linear (�) 19.1 54.63 1,231 �0.001Alt_range Quadratic (�, �) 13.4 17.82 2,230 �0.001Bio1 Cubic (�, �, �) 27.5 28.97 3,229 �0.001Bio2 NS 0.0 0.01 1,231 0.905Bio3 Linear (�) 2.0 4.66 1,231 0.032Bio4 NS 0.1 0.22 1,231 0.637Bio5 Linear (�) 11.2 29.03 1,231 �0.001Bio6 Linear (�) 18.9 53.68 1,231 �0.001Bio7 NS 0.0 0.08 1,231 0.775Bio8 Cubic (�, �, �) 22.1 21.66 3,229 �0.001Bio9 Cubic (�, �, �) 17.2 15.86 3,229 �0.001Bio10 Linear (�) 20.1 58.05 1,231 �0.001Bio11 Linear (�) 22.0 65.09 1,231 �0.001Bio12 Linear (�) 4.9 11.99 1,231 0.001Bio13 Linear (�) 1.9 4.46 1,231 0.036Bio14 Quadratic (�, �) 15.5 21.06 2,230 �0.001Bio15 Linear (�) 10.1 25.87 1,231 �0.001Bio16 Linear (�) 1.7 3.98 1,231 0.047Bio17 Quadratic (�, �) 14.8 19.97 2,230 �0.001Bio18 Quadratic (�, �) 11.1 14.34 2,230 �0.001Bio19 Quadratic (�, �) 11.1 14.34 2,230 �0.001Acid Linear (�) 2.1 4.85 1,231 0.029Basic NS 0.1 0.20 1,231 0.657

NS, not signiÞcant.

December 2008 GANAN ET AL.: DIVERSITY PATTERNS IN IBERIAN Calathus 1493

only the Þrst seven variables ranked in terms of theamount of variation explained (Alt, Bio1, Bio2,Bio11, Bio15, Bio17, and Basic) were included in themodel. In the spatial model the variables X, XY, Y2,X2Y, XY2, X3, and Y3 were included. Both modelswere signiÞcant (spatial model, pseudo-F � 3.34,P � 0.001; environmental model, pseudo-F � 2.83,P � 0.001; Table 4).

The overall model explained a large part of thevariation (58%) and, although the fraction shared byboth groups of variables (environment and space) isrelatively large (27%; Fig. 2B), the pure effect of eachmodel is still fairly important (17% explained by thespatial model and 13% by the environmental model).The variables that remained signiÞcant in the partialmodels and, as such, were responsible for the pureeffects of each model, were Bio11 and Basic in the en-vironmental model, whereas in the spatial model, allthe variables remained signiÞcant.

Discussion

The analysis of biodiversity patterns in Iberian Ca-lathus yielded two main Þndings: (1) our view ofmacroecological patterns may be compromised by thequality of data and (2) the conjoint analysis of species

richness andbetadiversity improveourunderstandingof biotic patterns because the variation in speciesrichness and composition could have different struc-tures.

The need for assessment of the quality of bioticinventories has been widely recognized, given that, asthe number of individuals sampled increases, so doesthe number of species detected (Bunge and Fitz-patrick 1993, Colwell and Coddington 1994). More-over, when we compare inventories from differentareas, we must take into account the fact that neitherthe sampling effort, nor the techniques used in eachcase, will necessarily be the same. Thus, to be able tocarry out comparisons, richness data from the studiedinventories need to be standardized (Palmer 1990,Gotelli and Colwell 2001, Walther and Moore 2005),for if we fail to estimate the completeness of inven-tories before carrying out analyses such as those per-formed here, we run the risk of drawing false conclu-sions. A clear example of this danger can be seen in therelationship that exists between species richness ofCalathus and abiotic predictors. On the basis of thecrude data (Fig. 1B), species richness seems to bestrongly structured by latitudinal, altitudinal, temper-ature, and precipitation factors (Table 3). However,after performing the completeness analysis and con-sidering only well-sampled grid cells, it becomes clearthat, in fact, the effects of these variables are not thatimportant (Table 2). Therefore, it seems that latitu-dinal and especially altitudinal sampling biases haveproduced a spurious pattern in the observed speciesrichness, which seems to be correlated with abioticfactors. However, when this effect was removed byconsidering only well-sampled grid cells, we failed toÞnd clear geographical or environmental pattern inthe variation of species richness. Geographical sam-pling biases, showed by our completeness analysis,have probably been caused by the oversampling pro-duced in hotspots and areas close to recordersÕ bases(Dennis and Thomas 2000) and reßect the historicalstructure in the way that species distributions areshown along time (Lobo et al. 2007, Hortal et al. 2008).Altitudinal biases may be explained because uplandareas have always held a greater fascination for cara-bidologists because of the possibilities of discoveringnew endemic species. However, latitudinal biasesmight be because of the fact that most of the materialin the main Iberian collections comes from Þeld workcarried out in the Spanish Central System mountains(Museo Nacional de Ciencias Naturales, Madrid,Spain) and the Catalonian Mountains and thePyrenees (Museu de Zoologia, Barcelona, Spain). Itshould be noted that some of the cells discarded fromour analyses could actually be poor areas instead ofrich but incompletely surveyed cells. This possibilityis based in that taxonomists usually do not publish theabsence of species, so the lack of information could bereßecting real poor regions. In this case, differencesbetween patterns derived from raw and good-qualitydata would be less marked. However, this uncertaintywill remain until further studies discriminate poorareas from poorly surveyed ones.

Table 4. Relationships between predictors and beta diversityand models for each group of variables, considering only well-sampled grid cells

Variable InertiaVariation

(%)Pseudo-F P

X 141.5 14.32 7.88 �0.001Y 14.3 1.45 0.8 0.083XY 138.1 13.98 7.69 �0.001X2 11.5 1.16 0.64 0.19Y2 27.5 2.78 1.52 0.006X2Y 27.9 2.82 1.55 �0.001XY2 32.7 3.31 1.82 0.001X3 20.6 2.09 1.7 0.001Y3 43.1 4.36 2.4 �0.001Alt 31.9 3.23 2.41 0.001Bio1 47.7 4.83 3.61 �0.001Bio2 109.4 11.07 8.27 �0.001Bio3 2.71 0.27 0.21 0.937Bio4 0 0.00 0 �0.999Bio5 46.1 4.67 3.49 �0.001Bio6 6.1 0.62 0.46 0.562Bio7 0 0.00 0 �0.999Bio8 29.3 2.97 1.53 0.01Bio9 5.8 0.59 0.44 0.603Bio10 29.7 3.01 2.24 �0.001Bio11 93.8 9.49 7.1 �0.001Bio12 20 2.02 1.51 0.015Bio13 8.2 0.83 0.62 0.318Bio14 18.8 1.90 1.42 0.017Bio15 73.8 7.47 5.58 �0.001Bio16 27 2.73 2.04 0.001Bio17 21.1 2.14 1.6 0.005Bio18 10.8 1.09 0.82 0.173Bio19 0 0.00 0 �0.999Acid 7.2 0.73 0.55 0.433Basic 24.5 2.48 1.85 0.002Model for S 441.2 44.66 3.34 �0.001Model for E 401.3 40.62 2.83 �0.001Model for S � E 572.7 57.97 2.86 �0.001

S and E are the spatial and environmental models, respectively.

1494 ENVIRONMENTAL ENTOMOLOGY Vol. 37, no. 6

Thus, if we are to discover the true diversity pat-terns of the genus Calathus in the Iberian Peninsulaand use this information to carry out more conclusivestudies fromconservationist, ecological, andbiologicalpoints of view, it is vital that we increase samplingeffort in the least studied areas (e.g., both the northernand southern Mesetas, the Ebro Valley, the BeticMountains, the southern half of Portugal, and theBalearic Islands). Until then, by selecting only well-sampled regions, thereby using only good-quality data(Hortal et al. 2004, Hortal and Lobo 2005), and dis-carding incomplete inventories to avoid reaching spu-rious conclusions, it is still possible to analyze diversitypatterns and their relationship with certain environ-mental factors and spatial limitations. From the studyof richness patterns we can deduce that, for IberianCalathus, species richness is not strongly environmen-tally nor spatially structured throughout the areasstudied, given that the model obtained explains only asmall part of the variance (23%) despite the highnumber of potential predictors tested in our analysis.In fact, richness patterns are completely unrelated tothe geographical situation, in part because of the factthat in those areas in which there are more endemicspecies with restricted distributions, there are alsofewer wide-ranging species (and vice versa). Never-theless, in mountain areas with a larger percentage ofacid rocks and lower temperatures during the wettestseason of the year, the number of Calathus species isslightly greater. The effect of topography may be be-cause of the fact that the number of endemicCalathusspecies in the Iberian Peninsula is high (70%) and thatthe majority are associated with upland areas. As well,the wettest months of the year tend to correspondto the spring and autumn, critical periods that coincidewith the appearance of the imagos and the breedingseason in many Calathus species. Although Thiele(1977), who only studied central Europe, where thereare fewer species and local endemisms, afÞrmed thatCalathus species are thermophilous, the Iberian spe-cies of the genus tend to prefer cooler temperaturesduring these seasons. Although our models explaineda small amount of the variance in species richness(23%), the greatest part of the explained variation canbe attributed to the pure effect of topography and thefractions shared by the models for topography andtemperature and for temperature and soil type. In theformer, the covariation between altitude and temper-ature is clear (at greater altitude temperatures arelower). Likewise, areas with low temperatures duringthe wettest season of the year and a high percentageof acid rocks seem to favor diversity in the genusCalathus.

In contrast to the small inßuence of environmentaland spatial variables in species richness patterns, whenwe analyze beta diversity, the model explains 58% ofthe variation. Most of this explained variation is causedby the fraction shared by the spatial and environmen-tal models (27%), a product of the fact that geograph-ical position is linked to temperature and rainfall:along a northwest to southeast axis, temperatures in-crease and rainfall decreases. Areas of higher temper-

atures and less precipitation seem to harbor fewerlocalized endemisms and a greater number of specieswith wide-ranging distributions. Nevertheless, the ex-istence of two relatively important pure fractions in-dicates that (1) environmentally similar cells tend tohave similar species compositions, but (2) spatiallyseparated cells may in fact have different species com-positions despite having similar environmental condi-tions. Although it may not inßuence richness patterns,it is interesting to note that geographical situationgreatly affects patterns of species composition. Con-sidering the high number of environmental factorsincluded in our analyses, we assume that the purespatial fraction reßects other biological processes asspeciation, dispersal limitation, and stochastic coloni-zation events. Therefore, the importance of the spatialfraction (17%) seems to indicate that historically therehave been limitations on species dispersal that pre-vented the colonization by species of environmentallyfavorable areas, a situation that has been observedin Iberian Chrysomelidae (Baselga and Jimenez-Valverde 2007). The pure effect of environmentalvariables is also relatively important (13%) and is in-ßuenced by the mean temperature of the coldest sea-son of the year and by the percentage of basic rocks:the species with the broadest distributions seem to bemost frequent in areas with the highest average tem-peratures during the coldest season of the year andlargest percentages of basic rocks. It is worth notingthat the percentage of basic rocks does not followeither a latitudinal or a longitudinal gradient.

To conclude, we should emphasize that the statis-tical techniques that enable this type of biogeograph-ical analysis to be performed would be of no use if theywere not supported by complete faunal and taxonomicstudies providing good-quality data. A lack of suchdata is a recurrentproblemin invertebrate studies, andsampling efforts must be increased in areas whosebiodiversity is relatively unknown if we are to under-take ambitious biogeographical studies on speciesrichness and beta diversity. The lack of taxonomic andgeographical data on species (Linnean and Wallaceanshortfalls, according to Whittaker et al. 2005) is one ofthe greatest obstacles toward the advancement ofknowledge of biodiversity and its origins. Thus, takinginto account the fact that we Þnd problems of com-pleteness even in one of the most studied of all beetlefamilies, the Carabidae, it is vital that more taxonomicand faunal studies of invertebrate groups be under-taken. Finally, Iberian Calathus represent a clear ex-ample of the need for combined analyses of speciesrichness and beta diversity patterns, because speciesrichness does not reßect species composition, and thelack of patterns in the former does not imply theinvariance of biotic communities.

Acknowledgments

We thank I. Izquierdo (Museo Nacional de CienciasNaturales, Madrid) and O. Escola (Museu de Zoologia,Barcelona) for loan of the material of Calathus. This workwas supported by Xunta de Galicia with project

December 2008 GANAN ET AL.: DIVERSITY PATTERNS IN IBERIAN Calathus 1495

PGIDIT05RF020002PR. A predoctoral grant to I.G. was Þ-nanced by Xunta de Galicia and Universidad de Santiago deCompostela.

References Cited

Araujo, M. B., D. Nogues-Bravo, J.A.F. Diniz-Filho, A. M.Haywood, P. J. Valdes, andC. Rahbek. 2008. Quaternaryclimate changes explain diversity among reptiles and am-phibians. Ecography 31: 8Ð15.

Bahn, V., and B. J. McGill. 2007. Can niche-based distribu-tion models outperform spatial interpolation?. Global.Ecol. Biogeogr. 16: 733Ð742.

Ball, G. E., and I. Bousquet. 2001. Carabidae Latreille, 1810,pp. 33Ð132. In R. H. Arnett, Jr. and M. C. Thomas [eds.]American Beetles. vol. 1. Archostemata, Myxophaga,Adephaga, Polyphaga: Staphyliniformia. CRC, Boca Ra-ton, FL.

Baquero, R. A., and J. L. Tellerıa. 2001. Species richness,rarity and endemicity of European mammals: a biogeo-graphical approach. Biodivers. Conserv. 10: 29Ð44.

Baselga, A. 2007. Disentangling distance decay of similarityfrom richness gradients: response to Soininen et al. Ecog-raphy 30: 838Ð841, 2007.

Baselga, A. 2008. Determinants of species richness, ende-mism and turnover in European longhorn beetles. Ecog-raphy 31: 263Ð271.

Baselga, A., and F. Novoa. 2006. Diversity of Chrysomelidae(Coleoptera) in Galicia, Northwest Spain: estimating thecompleteness of the regional inventory. Biodivers. Con-serv. 15: 205Ð230.

Baselga, A., and A. Jimenez-Valverde. 2007. Environmentaland geographical determinants of beta diversity of leafbeetles (Coleoptera: Chrysomelidae) in the Iberian Pen-insula. Ecol. Entomol. 32: 312Ð318.

Borcard, D., P. Legendre, and P. Drapeau. 1992. Partiallingout the spatial component of ecological variation. Ecol-ogy 73: 1045Ð1055.

Bunge, J., and M. Fitzpatrick. 1993. Estimating the numberof species-a review. J. Am. Stat. Assoc. 88: 364Ð373.

Cabrero-Sanudo, F. J., and J. M. Lobo. 2006. Determinantvariables of Iberian peninsula Aphodiinae diversity (Co-leoptera, Scarabaeoidea, Aphodiidae). J. Biogeogr. 33:1021Ð1043.

CLARK LABS. 2000. Idrisi 32.02. GIS software packagecomputer program. CLARK LABS, Worcester, MA.

Colwell, R. K. 2005. EstimateS: statistical estimation of spe-cies richness and shared species from samples. Computerprogram, version 7.5. Connecticut. (http://purl.oclc.org/estimates).

Colwell, R. K., and J. A. Coddington. 1994. Estimating ter-restrial biodiversity through extrapolation. Phil. Trans. R.Soc. Lond. B. 345: 101Ð118.

Currie, D. J. 2007. Disentangling the roles of environmentand space in ecology. J. Biogeogr. 34: 2009Ð2011.

Dennis, R.L.H., and C. D. Thomas. 2000. Bias in butterßydistribution maps: the inßuence of hot spots and record-erÕs home range. J. Insect. Conserv. 4: 73Ð77.

Dennis, R.L.H., W. R. Williams, and T. G. Shreeve. 1998.Faunal structures among European butterßies: evolution-ary implications of bias for geography, endemism andtaxonomic afÞliation. Ecography 21: 181Ð203.

Erwin, T. L. 1991. How many species are there-revisited.Conserv. Biol. 5: 330Ð333.

Ganan, I., and F. Novoa. 2005. El genero Calathus Bonelli,1810 en la Penõnsula Iberica y Baleares. (Coleoptera:Carabidae: Harpalinae). Elytron 19: 5Ð36.

Gaston, K. J., R. G. Davies, C.D.L. Orme, V. A. Olson, G. H.Thomas, T.-S. Ding, P. C. Rasmussen, J. J. Lennon, P. M.Bennett, I.P.F. Owens, and T. M. Blackburn. 2007. Spa-tial turnover in the global avifauna. Proc. R. Soc. Lond. B.274: 1567Ð1574.

Gotelli, N. J., and R. K. Colwell. 2001. Quantifying biodi-versity: procedures and pitfalls in the measurement andcomparison od species richness. Ecol. Lett. 4: 379Ð391.

Graham, C. H. 2006. Habitat history improves prediction ofbiodiversity in rainforest fauna. Proc. Natl. Acad. Sci.U.S.A. 103: 632Ð636.

Hammer, Ø., D.A.T. Harper, and P. D. Ryan. 2001. PAST:paleontological statistics software package for educationand data analysis. Palaeontol. Electron. 4: 1Ð9.

Hawkins, B. A., and E. E. Porter. 2003. Water-energy bal-ance and the geographic pattern of species richness ofwestern Palearctic butterßies. Ecol. Entomol. 28: 678Ð686.

Hawkins, B. A., J.A.F. Diniz-Filho, C. A. Jaramillo, and S. A.Soeller. 2006. Post-Eocene climate change, niche con-servatism, and the latitudinal diversity gradient of NewWorld birds. J. Biogeogr. 33: 770Ð780.

Hawkins, B. A., R. Field, H. V. Cornell, D. J. Currie, J. F.Guegan, D. M. Kaufman, J. T. Kerr, G. G. Mittelbach, T.Oberdorff,E.M.O’Brien,E.E.Porter, and J.R.G.Turner.2003. Energy, water, and broad-scale geographic pat-terns of species richness. Ecology 84: 3105Ð3117.

Hewitt, G. M. 1999. Post-glacial re-colonization of Euro-pean biota. Biol. J. Linn. Soc. 68: 87Ð112.

Hewitt, G. M. 2004. Genetic consequences of climatic os-cillations in the Quaternary. Phil. Trans. R. Soc. Lond. B.359: 183Ð195.

Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A.Jarvis. 2005. Very high resolution interpolated climatesurfaces for global land areas. Int. J. Climatol. 25: 1965Ð1978.

Hortal, J., and J. M. Lobo. 2005. An ED-based protocol foroptimal sampling of biodiversity. Biodivers. Conserv. 14:2913Ð2947.

Hortal, J., J. M. Lobo, and F. Martın-Piera. 2001. Forecast-ing insect species richness scores in poorly surveyed ter-ritories: the case of the Portuguese dung beetles (Col.Scarabaeinae). Biodivers. Conserv. 10: 1343Ð1367.

Hortal, J., P. Garcıa-Pereira, and E. Garcıa-Barros. 2004.Butterßy species richness in mainland Portugal: predic-tive models of geographic distribution patterns. Ecogra-phy 27: 68Ð82.

Hortal, J., P.A.V. Borges, and C. Gaspar. 2006. Evaluatingthe performance of species richness estimators: sensitiv-ity to sample grain size. J. Anim. Ecol. 75: 274Ð287.

Hortal, J., J. M. Lobo, and A. Jimenez-Valverde. 2007. Lim-itations of biodiversity databases: case study onseed-plantdiversity in Tenerife, Canary Islands. Conserv. Biol. 21:853Ð863.

Hortal, J., A. Jimenez-Valverde, J. F. Gomez, J. M. Lobo, andA. Baselga. 2008. Historical bias in biodiversity invento-ries affects the observed environmental niche of the spe-cies. Oikos 117: 847Ð858.

Instituto Geografico Nacional. 1995. Atlas nacional de Es-pana. Volumes 1 and 2. Centro Nacional de Informacion,Madrid, Spain.

Koleff, P., K. J. Gaston, and J. K. Lennon. 2003. Measuringbeta diversity for presence-absence data. J. Anim. Ecol.72: 367Ð382.

Legendre, P., and L. Legendre. 1998. Numerical ecology,2nd ed. Elsevier, Amsterdam, The Netherlands.

1496 ENVIRONMENTAL ENTOMOLOGY Vol. 37, no. 6

Lobo, J. M. 2008. Database records as a surrogate for sam-pling effort provide higher species richness estimations.Biodivers. Conserv. 17: 873Ð881.

Lobo, J. M., and F. Martın-Piera. 2002. Searching for a pre-dictive model for species richness of Iberian dung beetlebased on spatial and environmental variables. Conserv.Biol. 16: 158Ð173.

Lobo, J. M., I. Castro, and J. C. Moreno. 2001. Spatial andenvironmental determinants of vascular plant speciesrichness distribution in the Iberian Peninsula andBalearic Islands. Biol. J. Linn. Soc. 73: 233Ð253.

Lobo, J. M., J. P. Lumaret, and P. Jay-Robert. 2002. Model-ling the species richness distribution of French dungbeetles (Coleoptera, Scarabaeidae) and delimiting thepredictive capacity of different groups of explanatoryvariables. Global Ecol. Biogeogr. 11: 265Ð277.

Lobo, J. M., P. Jay-Robert, and J. P. Lumaret. 2004. Model-ling the species richness distribution for French Apho-diidae (Coleoptera, Scarabaeoidea). Ecography 27: 145Ð156.

Lobo, J. M., A. Baselga, J. Hortal, A. Jimenez-Valverde, andJ. F. Gomez. 2007. How does knowledge of the spatialdistribution of species accumulate?. Divers. Distrib. 13:772Ð780.

Martın-Piera, F., and J. M. Lobo. 2003. Database records asa sampling effort surrogate to predict spatial distributionof insects in either poorly or unevenly surveyed areas.Acta. Entomol. Iber. Macar. 1: 23Ð35.

McKnight, M. W., P. S. White, R. I. McDonald, J. F. Lam-oreux, W. Sechrest, R. S. Ridgely, and S. N. Stuart. 2007.Putting beta-diversity on the map: Broad-scale congru-ence and coincidence in the extremes. Plos. Biol. 5: 2424Ð2432.

Meier, R., and T. Dikow. 2004. SigniÞcance of specimendatabases from taxonomic revisions for estimating andmapping the global species diversity of invertebrates andrepatriating reliable specimen data. Conserv. Biol. 18:478Ð488.

Montoya, D., M. A. Rodrıguez, M. A. Zavala, and B. A.Hawkins. 2007. Contemporary richness of holarctictrees and the historical pattern of glacial retreat. Ecog-raphy 30: 173Ð182.

Odegaard, F. 2000. How many species of arthropods? Er-winÕs estimate revised. Biol. J. Linn. Soc. 71: 583Ð597.

Oksanen, J., R. Kindt, P. Legendre, and B. O’Hara. 2007.vegan: Community Ecology Package. R package version1.8Ð5. (http://cran.r-project.org/).

Palmer, M. W. 1990. The estimation of species richness byextrapolation. Ecology 71: 1195Ð1198.

Petersen, F. T., R. Meier, and M. N. Larsen. 2003. Testingspecies richness estimation methods using museum la-bel data on the Danish Asilidae. Biodivers. Conserv. 12:687Ð701.

Qian, H., and R. E. Ricklefs. 2007. A latitudinal gradient inlarge-scale beta diversity for vascular plants in NorthAmerica. Ecol. Lett. 10: 737Ð744.

Qian,H.,R.E.Ricklefs, andP. S.White. 2005. Beta diversityof angiosperms in temperate ßoras of eastern Asia andeastern North America. Ecol. Lett. 8: 15Ð22.

R Development Core Team. 2006. R: a language and en-vironment for statistical computing. (http://www.r-project.org).

Rangel, T.F.L.V.B., J.A.F. Diniz-Filho, and L. M. Bini. 2006.Towardsan integratedcomputational tool for spatial anal-ysis in macroecology and biogeography. Global. Ecol.Biogeogr. 15: 321Ð327.

Ribera, I., and A. P. Vogler. 2004. Speciation of Iberian div-ing beetles in Pleistocene refugia (Coleoptera, Dytisci-dae). Mol. Ecol. 13: 179Ð193.

Ricklefs, R. E. 2006. Evolutionary diversiÞcation and theorigin of the diversity-environment relationship. Ecology87: S3ÐS13.

Rodrıguez, M. A., J. A. Belmontes, and B. A. Hawkins. 2005.Energy, water and large-scale patterns of reptile andamphibian species richness in Europe. Acta Oecol.-Int. J.Ecol. 28: 65Ð70.

Rodrıguez, P., and H. T. Arita. 2004. Beta diversity and lat-itude in North American mammals: testing the hypothesisof covariation. Ecography 27: 547Ð556.

Soberon, J. M., and J. B. Llorente. 1993. The use of speciesaccumulation functions for the prediction of species rich-ness. Conserv. Biol. 7: 480Ð488.

Soberon, J. M., J. B. Llorente, and L. Onate. 2000. The useof specimen-label databases for conservation purposes:an example using Mexican Papilionid and Pierid butter-ßies. Biodivers. Conserv. 9: 1441Ð1466.

StatSoft 2001. STATISTICA computer program, version 6.0.StatSoft, Tulsa, OK.

Svenning, J. C., and F. Skov. 2007a. Could the tree diversitypattern in Europe be generated by postglacial dispersallimitation?. Ecol. Lett. 10: 453Ð460.

Svenning, J. C., and F. Skov. 2007b. Ice age legacies in thegeographical distribution of tree species richness in Eu-rope. Global. Ecol. Biogeogr. 16: 234Ð245.

Thiele,H.U. 1977. Carabid beetles in their environments. Astudy on habitat selection by adaptations in physiologyand behaviour. Springer, Berlin, Germany.

Walther, B. A., and J. L. Moore. 2005. The concepts of bias,precision and accuracy, and their use in testing the per-formance of species richness estimators, with a literaturereview of estimator performance. Ecography 28: 815Ð829.

Whittaker, R. J., M. B. Araujo, P. Jepson, R. J. Ladle, J.E.M.Watson, and K. J. Willis. 2005. Conservation biogeogra-phy: assessment and prospect. Divers. Distrib. 11: 3Ð23.

Whittaker, R. J., D. Nogues-Bravo, and M. B. Araujo. 2007.Geographical gradients of species richness: a test of thewater-energy conjecture of Hawkins et al. 2003 usingEuropean data for Þve taxa. Global. Ecol. Biogeogr. 16:76Ð89.

Willis, K. J., and K. J. Niklas. 2004. The role of quaternaryenvironmental change in plant macroevolution: the ex-ception or the rule?. Phil. Trans. R. Soc. Lond. B. 359:159Ð172.

Received 16 July 2008; accepted 2 September 2008.

December 2008 GANAN ET AL.: DIVERSITY PATTERNS IN IBERIAN Calathus 1497