American Arachnological Society

Distribution of Geraeocormobius sylvarum (Opiliones, Gonyleptidae): Range Modeling Based on Bioclimatic Variables Author(s): Luis E. Acosta Source: Journal of Arachnology, Vol. 36, No. 3 (2008), pp. 574-582 Published by: American Arachnological Society Stable URL: http://www.jstor.org/stable/25434330 . Accessed: 10/07/2011 08:55
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=aas. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Arachnological Society is collaborating with JSTOR to digitize, preserve and extend access to Journal of Arachnology.

http://www.jstor.org

2008.

The

Journal

of Arachnology

36:574-582

Distribution of Geraeocormobius sylvarum (Opiliones, Gonyleptidae): Range modeling based on bioclimatic variables
Luis E. Acosta: Universidad -C?tedra de Diversidad Animal CONICET I, Facultad de C?rdoba, Av. V?lez Sarsfield 299, X5000JJC Nacional de Ciencias C?rdoba, Exactas, F?sicas y Naturales, E-mail:

Argentina.

lacosta@com.uncor.edu 1887 Holmberg (Opiliones, the presence-only, GIS-based using The 19 derived bioclimatic variables. layers using limiting factors in each part of the range are discussed. sylvarum is modeled

Abstract.

The

potential

distribution

of

the

harvestman

Geraeocormobius

Gonyleptidae, Gonyleptinae) Bioclim. The model method bioclimatic Modeled

and Paraguay, Brazil from Argentina, run on 2.5 min resolution was climate

is described, and presumable profile of the species to the Alto Paran? Atlantic forest ecoregion, with of G. sylvarum shows a remarkable distribution correspondence Results forests towards the Southwest. a marginal forests and in gallery/flood the Araucaria around support the presence in northwestern as previously and reveal that localities 650 km yungas-Mesopotamia Argentina proposed, disjunction, extreme values concerning that the disjunct pattern may parameters with seasonality have been derived by antropic potential distribution, rainfall in winter. remarkably decreased is briefly discussed. introduction niche modeling, environmental Evidence suggesting

have

Keywords:

Neotropics,

bioclim,

ecological

envelope

Harvestmen (Arachnida, Opiliones) are generally regarded as a well suited taxon for biogeographic studies (Ringuelet 1959; Giribet & Kury 2007). Two key features make them useful for those purposes: their low vagility and a close dependence on environmental conditions, mainly humidity et al. et al. 2005; Machado (Acosta 2002; Pinto-da-Rocha Distribution of most species is thus dependent on the 2007), geographic continuity of suitable environments (Acosta 2002). In addition, harvestman endemicity may be particularly remarkable in some areas (Machado et al. 2007). Small-area endemicity is striking in forested ranges like the Brazilian et al. 2005), and is also Serra do Mar (Pinto-da-Rocha important in themontane forests of northwestern Argentina, the so called "yungas" region (Acosta 2002). rule for harvestman distribution. Under
many species are known to spread over However, small-ranged endemics should not be seen as the

1887 (Opiliones, Geraeocormobius sylvarum Holmberg is a large and conspicuous Gonyleptidae, Gonyleptinae) harvestman known to inhabit subtropical forests in north eastern Argentina, southern Brazil and southeastern Paraguay (Ringuelet 1959; Kury 2003; Acosta et al. 2007), being thus characteristic of theMisiones sub-area (Acosta 2002). Acosta et al. (2007) provided several new records for this species; among them three localities in the province of Tucum?n, Argentina, revealing a disjunct presence inmontane forests in northwestern Argentina (NWA). These separate populations lie about 650 km away from the westernmost record in the core area, with the sub-xeric Chaco in between. As stressed by Acosta et al. (2007), most records of G sylvarum concentrate in theAlto Paran? Atlantic forests ecoregion (referred to as
"Paranense the Araucaria Olson et al. forests" moist 2001). below) forest and in some ecoregions of sectors adjacent after (nomenclature the men outside captures

certain conditions,
thousands of km2

(Curtis & Machado 2007) as long as the suitable environment is not restricted by any geographic or ecological barrier. For
example, while many central European species appear geo

graphically restricted,no less than 30 harvestmen in that region extend over quite large ranges (Martens 1978). Broad ranged species are also typical formost of theArgentinean Mesopo tamia, which is the humid and sub-humid region between the Paran? and Uruguay rivers (Acosta 2002). Its northernmost of portion, roughly matching the administrative province forest physiognomy actually a Misiones, bears subtropical of the "Paranense Biogeographic Province" (Cabrera & part Willink 1973) that covers adjacent areas inParaguay and Brazil Cabrera grasslands and gallery forests (Hueck & Seibert 1972; Willink 1973). In accordance with these differences, the & opiliofauna has been split into two different, Mesopotamian Misiones sub though overlapping, sub-areas (Acosta 2002): the
area as well. The rest of Mesopotamia is a mosaic of shrubs, swamps,

tioned ecoregions (in provinces of Corrientes and Chaco) seemingly are associated with gallery forests and/or seasonal and Southern Cone inundation sites (the Humid Chaco savanna ecoregions of Olson et al. (2001)). Mesopotamian In turn, findings of G sylvarum inNWA correspond to the
"tucumano-boliviano"

Argentinean

the Southern Andean Yungas 1972; Acosta 2002), et al. 2001). The available localities show ecoregion (Olson some spatial bias, since they are concentrated in and around the province of Misiones (a traditionally well sampled area; are left Ringuelet 1959), while extensive areas in Paraguay almost undocumented (cf. Fig. 3). harvestmen, Dis Mesopotamian Interestingly, two further dilatatus Sorensen 1884 and D. prospicuus (Holmberg cocyrtus 1876) (Gonyleptidae, Pachylinae) have Mesopotamia-yungas disjunct ranges as well (Acosta 1995, 2002). A very basic
question

forests,

or

or

"yungas"

(Hueck

&

Seibert

West, as precipitation decreases, these humid and sub-humid realms give way to the semiarid Chaco, an effective distribu tional limit for Mesopotamian harvestmen (Acosta 2002). 574

and

the Mesopotamian

sensu

stricto

sub-area.

Towards

the

tions really inhospitable to Mesopotamian species or is our record incomplete, leading us to wrongly assume this region to is known about the be hostile for harvestmen? Nothing climatic tolerances of D. dilatatus, D. prospicuus or G

remains

unanswered,

however:

Are

Chacoan

condi

ACOSTA?DISTRIBUTION

MODELING

OF GERAEOCORMOBIUS

(OPILIONES)

575

sylvarum as very little is known about harvestmen physiology and tolerance to physical factors in general (Santos 2007). It is generally accepted that harvestman distribution is governed by climatic constraints, mainly humidity and temperature, but papers addressing how actual climatic conditions affect a 2007; given species are almost lacking (Curtis & Machado et al. 2007). Machado
The records occurrence are the records very basis contain for some useful a clues. species Point range, discovering

road maps and digital gazetteers (mainly NGA GEOnet Names Server [GNS], United States Board on Geographic Names; Google Earth ?), all cross-checked for final accuracy.
The label data or collector's information was used

to

though the vast majority ofNeotropical harvestmen have been scarcely recorded, so that species with more than 15 records are rare (cf.Kury 2003). In fact,most distributional patterns are then intuitively extrapolated from the few records available, emphasizing the need for gathering more data and filling in the gaps. Alternatively, recent developments aimed to model potential species ranges, using available records and several types of environmental predictors, offer innovative methods to help detect areas where presence of a given species may be expected though still not documented (Guisan & Zimmermann 2000; Guisan & Thuiller 2005; Hern?ndez et al. 2006; Peterson 2006). Using a Geographic Information System (GIS) and appropriate climate information, the envelope approach infers the species' bioclimatic profile, a species specific feature (actually, a subset of the fundamental ecological niche) that gives us a rough insight into their presumed tolerance limits,at leastwith respect to the variables used to build themodel. It is then possible to project this profile onto the geographic space, to identify all areas with
similar climatic conditions, i.e., where

determine location as precisely as possible. Published records typically refer to a locality, which may represent not more than the nearest reference to the actual collecting site; such an imprecision in coordinates (impossible tomeasure) is deemed not to affect the results considering both the regional scale and
coarse

1.4. layers.?The model was run on Worldclim et al. 2005a), a set of global climate layers, (Hijmans containing extrapolated monthly data for the 1950-2000 Climate precipitation. The 2.5 min resolution (i.e., approximately 4.5 x 4.5 km gridcell) was selected. Information contained in Worldclim (climate data, together with a digital elevation model) is used by the software to derive the 19 bioclimatic listed in Table 1. The variables available for modeling, abbreviation "be" followed by a number is used below to identifyeach of these bioclimatic variables. was performed through the Modeling method.?Modeling 5.4 (Hijmans et al. 2005b). Bioclim is a frequency distribution based algorithm, which calculates the envelope that bounds the bioclimatic preferences of the species (Fischer et al. 2001; Walther et al. 2004; Hern?ndez et al. 2006). Values of each derived bioclimatic variable are extracted from all localities and arranged in a cumulative frequency distribution (cf. Fig. 2). The envelope is defined as a multi-dimensional hyper box (each variable representing a dimension) delimiting, at a localities (Guisan & Zimmermann 2000). This set of values constitutes the bioclimatic profile of the species (Fischer et al. 2001), as summarized for the target species inTable 2. In the potential distribution maps (Figs. 1, 4-5), gridcells are scored as suitable (if within the envelope; i.e., the presence of the species can be expected) or unsuitable (outside the envelope). User-defined percentiles were set to define the extent of the envelope (as cut-off) or to rank the gridcells suitability (Fischer et al. 2001; Hijmans et al. 2005b; Hern?ndez et al. 2006).
Evaluation.?The given percentile, the climatic conditions in the occurrence presence-only method Bioclim, as implemented in Diva-Gis period on maximum, minimum and mean temperature, and

approach

used.

potentially suitable environment (Guisan & Zimmermann 2000; Hijmans & Graham 2006; Hern?ndez et al. 2006; Pearce & Boyce 2006). This paper is intended to characterize the bioclimatic profile of G. sylvarum and tomodel its potential distribution using the bioclim algorithm, taking advantage of the number of records now available for this species (close to the minimum sample size for the method to attain an acceptable accuracy; Hern?ndez et al. 2006). Thereby it is also intended to test if the disjunct Mesopotamian-yungas pattern is supported by the model and to identify areas in which future sampling efforts would be productive in filling in
distribution

the species

would

meet

types understanding,
requirements man. and

habitat

gaps.

Results

are matched

in order to get a more comprehensive though still preliminary, of the ecological

potential range of this wide-ranged harvest

to major

biomes

and

METHODS
Data acquisition.?All published references for G. sylvarum were taken into account (Holmberg 1887; Ringuelet 1959; Soares & Soares 1985; Kury 2003; Acosta et al. 2007). Historical records were inspected for taxonomic reliability (cf. Acosta et al. 2007 for localities excluded), while three localities from the province ofMisiones (published by Ringuelet 1959) remained unrecognizable and were set aside: "Campamento
Yacu-Poi, near Puerto

model was assessed by calculating theAUC (area under by the curve) in a receiver operating characteristic (ROC) plot, and themaximum Kappa (max-K) value, both analyses made in Diva-Gis et al. 2005b). In this study, 70% of the (Hijmans original points were randomly resampled as training data to perform 20 repetitions of the model. Test data included ground. ROC/max-K were calculated against the grids stack (0-100 percentile) of the 20 range polygons thatwere modeled using the training points. AUC values over 0.8 are deemed to reflecta "good" model performance; above 0.9 the accuracy is considered "high" (Luoto et al. 2005). In turn, max-K over 0.4 are deemed to be "good" and "excellent" if above 0.75 (Randin et al. 2006). fromNWA removed to verify if the species is stillpredicted in that area. To identify the factors limiting the distribution, the
Input variants.?A separate run was performed with records pseudo-absence points selected at random from the back

accuracy

of the predictive

range

generated

(which direction?), and "Pasarela R?o Uruguay." The full dataset consisted of 48 unique point localities; during the analysis, duplicate records from the same gridcell were removed by the software, resulting in 46 effective records. Localities were identified and geo-referenced using printed

Bemberg";

"60

km

Puerto

Iguazu"

576
Table Absolute Table

THE JOURNAL OF ARACHNOLOGY

minimum and maximum variables in the envelope values, and range for all 19 bioclimatic inmm. Numbers each variable temperature values are in degrees Celsius (? C), precipitation preceding 2. Main differences between profiles with and without in bold. yungas records are emphasized 1.?Median, of Geraeocormobius are referred

sylvarum. to in the text and

All records (n = 46)

Bioclimatic (1) Annual mean (2) Mean monthly variables temperature T? range Median 20.39 12.43 56.08 Min-max 16.46-23.12 10.65-13.83 47.25-65.00 (range) (6.66) (3.18) (17.75) 12.43 56.21 Median 20.62

Yungas

removed Min-max

(n

44) (range)

(5) Max

(3) Isothermality(2/7x 100) (4) T? seasonality (STD x 100)

T? of warmest month

16.46-23.12(6.66) 10.65-13.83 (3.18) 49.48-65.00 27.20-34.20

368.87
31.65 9.25 22.50 21.52 16.84 24.76 16.01

252.92^161.59 (208.67)
27.20-34.20(7.00) 4.80-12.50 (7.70) 19.00-25.40 16.20-25.23 13.40-19.22 20.33-27.58 (6.40) (9.03) (5.82) (7.25) 9.40

366.90
31.70 22.45 21.36 16.92 24.89 16.03

252.92-^430.76 (177.84)
(7.00) (7.70) (5.30) (9.03) 4.80-12.50 19.00-24.30 16.20-25.23

(15.52)

(6)Min T? of coldestmonth
range (7) T? annual T? wettest (8) Mean (5-6) quarter T?

(9) Mean (10) Mean (11) Mean

driest quarter T? warmest quarter T? coldest quarter

14.03-19.22(5.18) 20.33-27.58 (7.25) 12.50-18.50

(12) Annual precipitation

(13) Precipitation (14) Precipitation (16) Precipitation (18) Precipitation (19) Precipitation

wettest month driest month wettest quarter

1723.5 188.5 99.5

20.34

12.35-18.50(6.15)

(15) Precipitation seasonality (CV) (17) Precipitationdriestquarter

warmest coldest quarter quarter

9.42-83.74

934-2235 (1301) 157-245 (88) 186 11-150 (139) 101 495.5 (193) (456) (271) 455 (500) 362
(74.32)

1730.5

19.94

1250-2235 (985) 157-245 (88) 42-150 (108)

9.42-^41.88

(6.00)

495.5 344.5

432-625 40-496 457 354-625 360 40-540

349.5

432-625 140-496 354-625 140-540

(32.46)

(193) (356) (271) (400)

model

(full dataset) was alternatively run for each variable separately and combining some of them so as to visually inspect the effects on the resulting predicted range.

bidimensional envelope). With the default percentile of 0.025, localities within the species envelope are reduced to 21 out of 46 observations (45.7%).
Potential

RESULTS
Bioclimatic

the bioclimatic tolerance range of G sylvarum. Separate profiles with and without the Tucum?n localities are given fromwhich it is clear that, for some variables, conditions in the yungas suggest differences from the core area (cf. Table 2 also). Sites from Tucum?n represent both the westernmost records and the highest elevation. In those localities precip itation (be 12) is the lowest (Fig. 2B), with rainfall decreasing substantially during thewinter (Fig. 2D). There, seasonality is in higher than in sites of the core area (Figs. 2A, C); particular, precipitation seasonality (be 15) is strongly skewed to low values, with both Tucum?n records clearly separated by a decided gap from the rest (Fig. 2C). When these localities of are set aside, Rio Tragadero and 10 km Puerto NWA Antequera (close to each other) become the species' western most edge and come to hold many of the bioclimatic extremes related to seasonality and winter decrease of precipitation geographically extreme (Table 2 and Fig. 3). However, localities did not necessarily hold the highest or lowest bioclimatic values in all cases (Table 2). Other localities with many bioclimatic variables showing extreme values are Clevel?ndia (Para (Santa Catarina, Brazil) and Asunci?n Clevel?ndia has nine end values indicating, in general, a guay). cooler and more humid climate than the rest; in turn,
Asunci?n stands as the warmest site of the

profile.?Table

1 summarizes

values

relevant

to

themodel parameters is displayed inFig. 1.To a great extent, the predicted core area of G sylvarum roughly matches the
Paranense forests

range.?The

predicted

range

of G.

sylvarum

under

where records are almost lacking, the prediction partially redraws the boundaries of this ecoregion with the contiguous Humid Chaco (Fig. 3). On the Brazilian side, the predicted
range i.e., in enters the Araucaria forests only marginally, seemingly areas Paranense eastward surrounded by complex

ecoregion

(Fig.

3).

In

eastern

Paraguay,

projections
Araucaria

that follow large rivers. A

is scored

be noted that Clevel?ndia, one of the localities with many bioclimatic extreme values (Table 2), is placed near this presumable distributional limit (Fig. 3). In Brazil themodeled range reaches up to the southern states ofMato Grosso do Sul and S?o Paulo in theNorth (though with no records so far), and weakly up to thewestern slopes of the Serra do Mar in the East (Fig. 1). Geraeocormobius sylvarum was not hitherto collected in the eastern slopes of the Serra do Mar; probably replaced there by its congener Geraeocormobius rohri (Mello Leit?o 1933) (A.B. Kury, pers. comm.). On the other side,
occurrences of G.

ecoregion

as unsuitable;

large portion of the

moreover, it is to

with highest values for six variables, all temperature-related. With respect to the resulting envelope, a permissive percentile cut-off of 0.005 allowed 33 out of 46 observations (71.7%) to be included within all possible 171 bidimensional envelopes (the remaining 13 observations being outsiders in at least one

species

range,

Corrientes and eastern Chaco give support to the potential areas southwest of the paranense forests. Subtropical vegeta tion partially extends into northern Corrientes, as well as along flood and gallery forests along theParan? and Paraguay rivers (Hueck & Seibert 1972), and this seems to provide suitable conditions for G. sylvarum some hundreds of kilometers away from themain range. With the full dataset, the bioclimatic model predicts the presence of G. sylvarum in the yungas but not across the sub xeric Chaco (Fig. 1); thus, the presumed disjunction is

sylvarum

in

the

northern

province

of

> o o H>

C/J 3 H 2

c H O

o u mr Z O o TI

o SI S

Soares 1945 & Source 2007 Acosta al. al. 2007 Acosta 2007etAcosta al. 2007et Acosta2007 et Acosta 2007 Acosta al. et 2007 Acosta Acosta al. Acosta 2007 Acosta2007et 2007 et 2007 Acosta al. al. al. al. al. al. et et et et et 1959 1887 1959 1959 1959 Tavares 1980 Tavares 1980 1887 Ringuelet Ringuelet Holmberg 2003 Kury Ringuelet Ringuelet Holmberg

15 15

7 4,

(15)

(4)

(7)

10,13, 1, 16, 5,6,8, 2(7)11 19 18, 12,

17 14,

Latitude (S)

be highest lowest bevariables, variables, 11 11 9, 19 19) 19) 17, 14, 17, 17, (3, (14, 18 (12) 16 3, 3, 14, 12,

10(11) 1, 13 5,6,

7 4,

5(9)

15

7 2,

Table 2.?Localities Geraeocormobius of (be) bioclimatic (underlined) that (numbers sylvarum geographical for variables coordinates either extreme represent values, or as in Latitude Table 1). identified and longitude first in treated Diva-Gis degrees. given the localities be and are longitude point brackets, of second In single values extreme record. as a Longitude (W) Irai -53.2500 -27.2000 Gaurama -52.0867 -27.5867 -27.3667 Mini Asunci?n Ana Santa -55.6000 -26.3667 Piray -25.3000 -54.6667 -57.6333 Paranava? -23.0833 -52.4667 Guara?na -25.3333 -50.3667 Puerto Rico -55.0333 -26.8000 SaoValentim-27.5500 Nova Teut?nia -27.0500 (-58.8667) R?o Tragadero -27.4333 -52.5333 -52.4000 Encantado Salto -27.0667 -54.8333

San Cerro Road -26.8000 700 Javier, -65.3417 to Cristo -26.8083 Villa Nougu?s Between -65.3583 andm km Puerto Antequera 10 -27.3833 -58.8000 (500-700 Corte El -26.8083 -65.3333 m) (Ipor?) Ibera -57.1833 -28.5167 Laguna Locality

(nowadays Cachoeirinha -24.1533 Arapoti) -49.8217 (Pt. -25.9167 -54.6167 Libertad) Bemberg Puerto Cavi?na, Fazenda Tocantins -23.3000 -51.3667 Clevel?ndia, -26.4000 Fazenda Tunas -52.3667

State Province or if variables records from Tucum?n are removed.

do Rio Sul Sul Rio Sul Santa do Rio Catarina do GrandeGrande Grande Corrientes Central Misiones Misiones Misiones Tucum?n Paran? Paran? Paran? Tucum?n Chaco Paran? Paran? Misiones Misiones Tucum?n Chaco

Country Argentina Argentina Argentina Argentina Argentina Argentina Argentina

Argentina Argentina Argentina

Paraguay Argentina Brazil Brazil Brazil Brazil Brazil Brazil Brazil Brazil Brazil

578

THE JOURNAL OF ARACHNOLOGY

1.?Predicted Figure 19 bioclimatic variables. M, Misiones. States

range (0-100 percentile) inArgentina: Provinces RS, Rio Grande

of Geraeocormobius T, Tucum?n do Sul, SC,

in Brazil:

(detailed map Santa Catarina,

as modeled with Bioclim, using the full data set and all sylvarum Holmberg, in Fig. 4); Fo, Formosa; Ch, Chaco; SF, Santa Fe, Co, Corrientes; Matto Grosso do Sul. SP, S?o Paulo, MGS, Pr, Paran?,

model for supported. However, the probability assigned by the its occurrence in Tucum?n is very low. Even in a 0-100 percentile envelope, suitable gridcells are few (just four), and if ranked, theyfit in the lowest percentile levels (Fig. 4A). With a

cut-off, no suitable gridcell remains in NWA. As expected by these results, modeling with Tucum?n records removed did not predict the species there either. Otherwise, modeling the range without theseNWA records had minimal 0.025

0.0

240

280

320

360

400

440

480 B

800

1120

1440

1760

2080

2400

bc4: temperature seasonality

bc12: annual precipitation

a> 0.6

10

20

30

40

50

60

70

80

90 D selected variables: (NWA). (coefficient

100

200

300

400

bd 5: precipitation seasonality Figure sylvarum, profile indicate the position of the two localities from province of Tucum?n bel 2 annual precipitation (mm); C: bel 5 precipitation seasonality 2.?Bioclimatic of Geraeocormobius

bc17: precipitation driest quarter full data set plotted for cumulative relative frequency; arrows x 100); B: bc4 deviation temperature (standard seasonality of variation); D: bel 7 precipitation of the driest quarter (mm).

A:

ACOSTA?DISTRIBUTION

MODELING

OF GERAEOCORMOBIUS

(OPILIONES)

579

Figure 3.?Correspondence forests ecoregion (squared line), where G R?o Tragadero;

and the Alto Paran? Atlantic range of Geraeocormobius sylvarum (grey area, 0-100 percentiles) Araucaria is shown as a lighter grey squared area. AMF: moist forests (bordered by a thin fill); the overlap to occur marginally. Dots: 1:Asunci?n; 2: all locality records for the species. References of selected localities: sylvarum is predicted 4: Cavi?na; 5: Paranava?. 3: Clevel?ndia; pattern

of the modeled

effects in the core area: just a small clipping occurs, reducing the range on marginal sides, like the Corrientes-Chaco portions, and its northernmost extension in Brazil. of G sylvarum run separately, Limiting factors.?Models with either temperature (bel?11) or precipitation (bcl2-19) variables, uncover the relative contribution of these variables to the predicted range shape. Prediction based on temperature variables bcl-bcll (Figs. 4B, 5A) more closely recovers the model obtained by all 19 variables together, with a slight "permissiveness" in the southernmost border (province of Corrientes) and especially in NWA (patchy suitable areas the yungas in Tucum?n, Jujuy, Salta, and appear along southern Bolivia). Precipitation variables bcl2-bcl9 (Figs. 4C,

5B) clipmost of these areas away from themodel, but, in turn, extend suitable conditions considerably eastwards and north wards. In fact,many precipitation variables (when run alone) predict G. sylvarum in the yungas, but almost all prevent it entering the Chaco, some being more rigorous (bcl5, bcl6, bcl8), othersmore "permissive" (bcl2, bcl4, bcl7, bcl9); only
be 13 enables Chaco a somewhat continuous variables

few temperature variables

gap. Temperature

(bc5 and bc8) contribute to the

are more related to range

range.

In contrast,

only

restrictions in the North and the East. The northernmost boundaries inBrazil, where the rangemeets theCerrado and is limited by high temperatures, are fairlywell shaped by bel, bc4, bc6, bc9, and bell. As noted above, themodeled range

yyi;ii/yy/

in province 4.?Detail of the predictive modeling of Tucum?n, Figure indicate the occurrence records), B: model with only temperature variables dark grey: above 0.25. Light grey: 0-25 percentile;

i VyA y

y i VyB

Argentina. (bcl-bcll),

1 VVC y
A: model C: model

with

all

19 bioclimatic

variables variables

(small

dots

with

only precipitation

(bcl2-bcl9).

580

THE JOURNAL OF ARACHNOLOGY

5.?Potential distribution Figure dark grey: above 0.25). A: modeled

of Geraeocormobius range with

temperature

variables (light grey: 0-25 percentile; sylvarum modeled only with temperature or precipitation variables variables range with precipitation (bcl-bcl 1), B: modeled (bcl2-bcl9).

seems

to

circumvent

the

Araucaria

ecoregion only marginally (Fig. 3). Several temperature variables (bel, bc5, bc6, bc8, bc9, bclO, bell) leave a defined gap there.The unsuitability appears to be related to the high elevation, which in turn causes the seemingly unsuitable cold climate: Clevel?ndia, elev. 890 m, is the highest record of G. sylvarum in Brazil and bears several extreme cold climate values (Table 2), but many negative sites in theAraucaria area are above 1100m.

forest,

entering

this

with the training samples (30% of the Accuracy.?Modeling original points randomly removed) consistently recovered the main portions of the range predicted with all points. More than half of the 20 repetitionsmatched in most parts of the core area, and just marginal sectors overlapped in only 10 repetitions or less. AUC values ranged from 0.8505 to 0.9402 (mean = = 0.8935), and max-K from 0.7082 to 0.8832 (mean 0.7900); individual and average AUC and max-K values resulted in thus, good to high model performances. These values are consistent

ACOSTA?DISTRIBUTION

MODELING

OF GERAEOCORMOBIUS

(OPILIONES)

581

with performances of Bioclim obtained by Hern?ndez et al. (2006) for similar sample sizes, and with comparative evalua tionsmade by Sangermano & Eastman (2007).

DISCUSSION
As
species

stressed by Peterson
nothing more

(2006), for the vast majority of

than a few "dots on maps" -

quite limited distribution, and the bioclimatic model accord ingly predicts this species only weakly for the region. The yungas is one of the best sampled areas in Argentina for harvestmen (Acosta 2002), so it seems unlikely to find G sylvarum elsewhere inNWA. People from the species' core
range are

is known

and certainly, this applies formost Neotropical harvestmen, too. Ecological niche (or habitat suitability)modeling offers a first step towards inferring the basic ecological dimensions that are relevant to limit the species' distribution. While previous knowledge just imprecisely related G sylvarum to Paranense forests (and thereby assumed itwas dependent on humid conditions), the bioclimatic analysis provided, for the first time, defined values to describe and discuss the species profile. The envelope method proved to be well suited as an initial approach when species records and biological knowl edge are still scarce (Pearce & Boyce 2006). Results were convincing when matching the predicted range polygon to the Paranense forests ecoregion (Fig. 3), a biogeographical area that was previously associated with G sylvarum based on a
"non was

and smelly harvestmen are also locallywell known inEl Corte, although people that have lived there for the last 30 years suggest that the species "appeared" only in recent years (A.M.
Fr?as,

its large

size,

aware in general of G. sylvarum, probably abundance and These strong odor.

because conspicuous

of

apparent tolerance of G. sylvarum to disturbed areas (Acosta et al. 2007), may suggest that transportation by humans in historic times might best account for the presence of this
Mesopotamian harvestman in Tucum?n.

pers.

comm.).

These

observations,

together

with

the

ACKNOWLEDGMENTS
I am indebted to Maria Laura Ju?rez, Gustavo Flores, Dardo Mart?, Adriano Kury, and Miryam Damborsky for providing coordinates or details of the collecting sites of G sylvarum. Comments and suggestions given by Raquel M. Gleiser, Soren Toft, Paula Cushing and an anonymous referee greatly improved themanuscript. Financial support was given by Consejo Nacional de Investigaciones Cient?ficas y T?cnicas P.I.P. N? 6319) and SECyT-UNC. The author is (CONICET: a researcher of CONICET. LITERATURE
Acosta, L.E. 1995. Nuevos notas Argentina, (Opiliones, 207-217. Acosta, L.E. argentinos 6:69-84. Acosta, L.E. Argentinean bioclimatic con

2007). The model accuracy as measured by AUC One

acceptable as well.

analytical,

expert-based"

assessment

(Acosta

et

al.

and max-K

of

the main
at the

Mesopotamia-yungas
model; but

same

issues tackled in this paper, the disjunction, received support from this
time, some new questions arose

concerning the presence of G sylvarum inNWA. Despite the availability of presence records, the species is only weakly predicted there. This contrasts with other Mesopotamian
harvestmen

CITED

in preliminary models, were predicted in the yungas, with or without positive records in NWA (Acosta 2007). Both D. dilatatus and D. prospicuus belong to the sensu Mesopotamian stricto sub-area (Acosta 2002), so their climatic requirements are not expected to be as humidity-dependent as for G sylvarum. Since Bioclim is deemed to over-predict a species' range (Peterson 2001), these facts, together with the profile values for some variables, suggest an important disparity of the bioclimatic conditions of NW and NE Argentina not lose sight of the importance of the resolution of the climatic data. In general the resolution used (2.5 min) is
acceptable adequately at a regional scale, but its coarseness in montane may not areas, reflect small-distance variations concerning preferences of G sylvarum. However, one should

{D.

dilatatus

and D.

prospicuus

among

them)

that,

en de Discocyrtus dilatatus hallazgos sobre taxonom?a, sinonimia y distribuci?n Revue Arachnologique 10: Pachylinae). Gonyleptidae, 2002. Patrones de los Opiliones zoogeogr?ficos Revista Ib?rica de Aracnolog?a

(Arachnida: 2007.

Opiliones).

Sao Pedro, SP, Brazil. Abstracts, p. 49. A.B. Ju?rez. 2007. New records of Acosta, L.E., Kury & M.L. Geraeocormobius sylvarum (Arachnida, Opiliones, Gonyleptidae), with a remarkable in northwestern Argentina. Bolet?n disjunction de la Sociedad 41:303-306. Entomol?gica Aragonesa A.L. & A. Willink. 1973. Biogeograf?a Cabrera, Colecci?n Cient?ficas O.E.A., Monograf?as 1-117. pp. I-VI, Machado. Biology de Am?rica serie Biolog?a, Latina. n?

Distribution of harvestmen in the (Opiliones) a modeling on based Mesopotamia: approach variables. 17th International of Arachnology, Congress

13,

like inTucum?n (records range from 500 to 1250m elevation within a distance of less than 3 km; i.e., less than the gridcell As for the causes of the disjunction, a satisfactory explanation remains an open question. Acosta (1995, 2002) suggested that the disjunct ranges of the two Discocyrtus species may be a consequence of paleoclimatic cycles, with associated expansion/retraction events affecting the humid forests.A hypothetical vegetational "bridge" (as proposed by Nores 1992 for birds) would have acted as a corridor, enabling harvestmen to expand their ranges up to the Mesopotamian yungas, to leave isolated populations there as climate turned
rigorous however, and the forests elements retracted. may In the case this of G several render explanation sylvarum, less size).

DJ. & G. Curtis, Harvestmen: The Machado

& G. Giribet, bridge, Massachusetts. Fischer, J., D.B. Lindenmayer, 2001. Climate and animal marsupial phy 28:293-304. G. & A.B. Giribet, Kury. In Harvestmen: Pp. 62-87. da-Rocha, Press, Guisan, G. Machado Cambridge, A. & W. Australian

2007. 280-308. In Ecology. Pp. of Opiliones. G. (R. Pinto-da-Rocha, eds.). Harvard Press, Cam University

H.A. J.L. Stein & J.A. Stein. Nix, a climatic distribution: analysis of the Trichosurus caninus. Journal of Biogeogra 2007. The & G. and biogeography. Phylogeny of Opiliones. (R. Pinto Giribet, eds.). Harvard University Biology

Massachusetts.

offering more 1009.

Thuiller. 2005. Predicting species distribution: than simple habitat models. Letters 8:993 Ecology

likely. Records

in the yungas are actually few, indicating a

A. & N.E. Zimmermann. 2000. Predictive Guisan, habitat distribu tion models in ecology. Ecological 135:147-186. Modelling

582
Hern?ndez, P.A., The effect of mance graphy Hijmans, 2005a. 2006. L.L. Master & D.L. Albert. C.H. Graham, on perfor size and species characteristics sample Eco methods. of different species distribution modeling 29:773-785. S.E. Peterson, based Peterson, models 3:59-72. Pinto-da-Rocha, R., M.B. A.T. 2001.

THE JOURNAL OF ARACHNOLOGY

distributions Predicting species' geographic niche modeling. Condor 103:599-605. Uses and of ecological niche requirements distributional models. Biodiversity Informatics

on ecological 2006. A.T. and related

J.L. Parra, P.G. Jones & A. Jarvis. Cameron, interpolated climate surfaces for global Very high resolution 25:1965-1978. International Journal of Climatology land areas. R.J., R.J. & C.H. to predict Global Graham. the 2006. The effect of ability of climate envelope on climate change species 12:1-10.

Hijmans, models Hijmans,

similarity and historic biogeography Atlantic Rain and southeastern Arachnology Randin, Zappa models 1703. Ringuelet, Revista Zool?gicas R.A. 1959. Los 5:127-439. ar?cnidos del Museo Argentino 33:290-299.

da Silva & C. Bragagnolo. 2005. Faunistic of southern of the harvestmen Forest of Brazil. Journal of

Holmberg, Nacional

? Misiones. de Bolet?n 1887. Viaje 10:5-391. de Ciencias, C?rdoba von 1972. Vegetationskarte K. & P. Seibert. Hueck, der einzelnen Gro?r?umen, Vegetationsmonographien E.L. G.

Biology Change P. Mathur, C. A. Jarvis, R. O'Brien, R.J., L. Guarino, & E. Rojas. 2005b. DIVA-GIS, I. Barrantes Cruz, Bussink, M. at http://www.diva-gis.org/ version 5.4.0.1. Online la Academia

distributions.

N.E. T. Dirnb?ck, S. Dullinger, CF., 2006. Are niche-based & A. Guisan. transferable in space? Journal

species

M. Zimmermann, distribution 33:1689 Opiliones. Ciencias

of Biogeography del Orden Naturales,

de 2007.

argentinos Ciencias

S?damerika. 2a: 1-71 +

Fischer, map. Stuttgart. of the New 2003. Annotated of the Laniatores catalogue Kury, A.B. Ib?rica de Aracnolog?a, Revista World Opiliones). (Arachnida, 1:5-337. Volumen especial monogr?fico Luoto, M., Uncertainty J. P?yry, R.K. of bioclimate Heikkinen & K. Saarinen.

Sangermano, use of Mahalanobis 1-13. Pp. de Sistemas Aires. Santos, F.H.

F. & R. Eastman.

The and Ecology Linking GIS to model distribution. species typicalities XI de Iberoamericana In Memorias Conferencia Informaci?n Geogr?fica. (G.D. Buzai, ed.). Buenos

2005.

distribution graphical phy 14:575-584. G, Machado, harvestmen? R. Pp. 1-13.

on the geo models based envelope and Biogeogra of species. Global Ecology & G. Giribet. The 2007. What are

Biology Giribet, setts.

In Harvestmen: The 2007. Ecophysiology. Pp. 473-488. & G. G. Machado of Opiliones. Pinto-da-Rocha, (R. Massachu Press, Cambridge, University eds.). Harvard

Pinto-da-Rocha

In Harvestmen:

Biology

G. Machado & G. Giribet, (R. Pinto-da-Rocha, Massachusetts. Press, Cambridge, University Die Tierwelt J. 1978. Weberknechte, Martens, Opiliones. Jena Gustav lands 64:1^164. Fischer, M. ?ores, relation

of Opiliones. eds.). Harvard Deutsch

do Museu 1945. Alguns Soares. & H.E.M. Soares, B.A.M. opili?es de Zoolog?a Nacional do Rio de Janeiro. Pap?is do Departamento 5:221-226. de S?o Paulo do Estado Soares, H.E.M. & B.A.M. Soares. 1985. Contribution ? l'?tude des du Cosmetidae, Gonyleptidae) Phalangodidae, (Opiliones: 92:3-18. Revue Suisse de Zoologie Paraguay. no Rio Grande de opili?es 1980. Novas ocorr?ncias Tavares, M.L.R. opilions do Sul & e descric?o da f?mea de Melloleitaniana riodariensis Soares 1945 (Arachnida-Opiliones-Gonyleptidae). Soares, 55:155-159. S?rie Zoolog?a Rahbek. and 2004. Known use and of habitat griseldis) eineracea). Iheringia, and predicted the endangered the near-threatened of Ornithology

in South America in subtropical 1992. Bird speciation 109:346-357. to forest expansion and retraction. Auk N.D. E.D. Wikramanayake, E. Dinerstein, Burgess, D.M., Olson, H.E. JA. DAmico, G.V.N. Strand, J.C. Powell, E.C. Underwood, Morrison, Ricketts, Kassem. I. C.J. Loucks, T.F. Itoua, W.W. 2001. Terrestrial BioScience M.S. & Wettengel, Allnutt, Y. T.H. J.F. Lamoreux, P. Hedao & K. Kura, a new map of of the world: distribution of Applied and

Walther, B.A., M.S. Wisz & C winter distributions African Basra reed warbler bunting cinereous 145:287-299. Manuscript received (Emberiza

life on Earth. J.L. Pearce, abundance 43:405^12.

ecoregions 51:933-938. 2006.

(Acrocephalus

Journal

Boyce.

with

presence-only

Modelling Journal data.

Ecology

1 June 2007,

revised

18 February

2008.