".: ' Rev. Biol. Trop., 46(3):515-524, Ü)98

Spatial heterogeneity and woody speeies distribution in a �alansae (Anacardiaceae) forest of the Southern Chaco,

I. M. Barberis, �.F p,4x andJ. P. Lewis1 . '. I � . . . .�. FlJ.Cultadde Cie¡tci� Agraiíl\S; Un�versidad Nacional de Rosario, Casilla de Correo 14, 2123 Zavalla, Arg'tntina. FAX 54- ,. '041"257164. E-mail: bá[email protected]

30-111-1998. 14-IV-1998 .. Received 24-VI-199,7L�9�ed Aet;epted

<'AbStr�ct:Sp�� h eterogeneity of a Schin/¡J?;t�balansae forest ("Quebra,chal'i) near Vera (Santa Fe, Ari� �tina)' Was ¡ :F\;.lltU9i�.q fqtlliji:t;orrebltionbetweén wbQdy�iell distributíon aJÍdenvironmental Cactors:·This type of'Cotestis diemost ,j � imp��ílt;&l!QÍl;Ií¡.Pityofthe.santaFeF0rest;We'dg�¡which is the southernmost portionefthe Easteril Chaco.. ' lO x 10111i'ÍllPtswere laid along twIJ W'U'ansel$. ,OOdy vegetation andmicrozones determined CoqIiiororelief, ' is." . . �dtge:�\,nceoC bromeUads wer;e1Íi8P�d; 54% oí the ground is level, abouthalfofit is muddy o�j�¡:ntly.:; , ijl)oded¿:and.llttlemoreth¡m n� i� concave.a4lÍostalways tlooded. Qonvex soils are well drainedand...... , 'ered by Iiº�U1atibJÍsqC spiny bronIél¡J¡¡srM9�hvoQé:i:y individt¡!i1s high and w�lld í:iiined wÍñ �hly:'.. w�ki:ll1nÍpedbn 'soií� " Geolfroea' dé�orticiins, Pro�ojJi;�pp .imdS. �ílln%é Jh:1�r.Y,humid micro�o9J:�¡,These �últs stfu giy .. ·· we�pr;e�i1t . � "', , sugges� !,bats6il héterogeneity (microrelieCIIÍ,Id soil moisture), is �:illbst iíhportant factor't\'IÍIttdétenfiíneS tbe distii� ¡ ¡ ·'.ufioJ,lÓfwOI),dy¡sPecJes. . ¡;¡ . ;" :.

,"::. ,',t ., ,f-":' :\i� j,';Ji':, .�;,': ,>,}:'" " '\;r. Natural s�t�ms ¡h�y�J)é�n¡ recogn1z�� !l$ of �p�cies ¡ t e � �� h pr���nt9rit>�O)iQtic· �dabi­ tibally' h��rogeQeoos · & otlC abio in . rt�� ¡ruido tf�e, C9ndltIons, (Oleª�on'. ln. "<>1 Whl��ker '. > . . . .é'. ana ¡ :v�al>le�. in orgamsm' abutÍdance (Sousa 19(7).:" .. , ' �' . /'�'/ '!f'�Y'.""'':' '. ,c' 0,. 'Ó1F,:':".' "".' : , ; :�'� . 19&,4r Sim:e"�espatial lj��F¡()ge�e1ty()f thé ¡Th:�Chaco regiop{ÍSo .10 3,SóS) isavery , en:Vp'0nw�ntMiects man); 4,�p�tso� the popu� largese4imentary piáiri ih��c()v i!s�aqQU� 1000 . '. latio�t' lJio. () X oLplants; �pc!fi"Jls populaíon of eastem B ivia, te 'Pafa�uay, , , � . � ú90tqn2 ?� : ��� �. ... ; 'dyn�cs� s��dis ¡�rsar alld. recruihIlent northem Arge�tina and a. sm!lll'pat;&,.Qf south­

. {HaIP,�r, 197.:t;J1ru15� 1977),it �s therefore a eastem Br¡:¡,sil'(Prado 1993); Th� wpble r:egion major factor ré latirigilie disfrÍlJ4tion of in4k is not homogeneous,thete i& a gil ' ilqttrid Ch�cbat :vidu�ls and species (Úr1)an J7�tdl, 1987) . . The the eru¡tand � c4i �ne aft6e W�§t\Ragg�e��ian9 ch��cteristicsofthe .habjgt� determine which Castiglioni i9,70), During .th.e.lIó10é�nr·�Ífer� jl}dlyif:1\l�s fro� those th�twri�� lita phlc,e, ential mov�ID�nts of bedrock. blocks oí ,the

. .. Eastem ¡<�ct[l!llli&:cupy tpat ,piptis41ár jl�se (l3azzaz c:h�có, ptoduced � �bnt,a�jting., ·,iand­ a . t1orl�tic<�,?: osi- . sCllpes that' were further m'Odifie by fJ.uvial 1� ll�.*? �()�unity� tftf( we � , i gpn · i's:de��rnUried by}he... di.ffereJ)ti�l��ii�nse activity (Popolizio el al. 19SP); At th� loc�l, 'í·, 516 REVISTA DE BIOLOGIA TROPICAL level topography produces strong ecological MATERIALSAND METHODS differences, that results in the presence of dif­ ferent plant formations (Morello and Adámoli The study area: It is located in the experi­ 1974, Lewis 1991). mental center of the Provincial Ministry of The southernmost segment of the Eastern Agriculture at Las Gamas, near Vera (29°28'S - Chaco is the Santa Fe Forest Wedge ("Cuña 60028'W), Province of Santa Fe. The forest was boscosa"), which is located between the heavily lumbered in the first half of this centu­ Submeridional Lowlands (Spartina argentinen­ ry for tannín industry and subsequent1y for fuel sis Parodigrasslands with almost no trees) at the extraction and charcoal production. This stand, west and the Paraná riverat the east from paral­ although is partialIy recovered, is not complete­ lel 30030' S to 28°S. The climate is temperate­ ly mature or at a steady state, and cattle grazing, warm humid with a mean annual precipitation mild wood extraction and sporadic fires are the slíghtly aboye 1 000mm, which occurs mainly in main disturban ces. summer, and with a strong winter drought of Data collection and aualysis: On a transect variable length (Burgos 1970). Thevegetation is 22 contiguous 10 x 10 m plots were laid (Lewis a mosaic of forests, savanna grasslands and 1991) and another ten plots of the same size hygrophilous communities ordered along mois­ were laid on a second discontinuous transect ture and salinity gradients correlated with topo­ parallel to the fírst one. On March 1989, all graphic elevation (Lewis and Pire 1981). The trees taller than 1.6 m and shrubs taller than 0.6 Schinopsis balansae Engl. forest ("Quebrachal") m present in the plots, were counted and appears on halo-hydromorphic soils (Espino et mapped and at the same time, soil surface was al. 1983), in the middle of the elevation gradient divided in microzones that were also mapped. between Prosopisnigra (Gris.) Hier. varo ragone­ Microzones were determined according to sei Burk. forest ("Algarrobal") at the lower microrelief (D: concave, P: level, E: convex; places and the mixed dense forest ("Bosque margins of these areas were characterized as: m: chaqueño") at the higher places (Lewis and Pire stepped n: sloped), soil moisture (1: well 1981, Lewis 1991), and it is the most important drained, 2: dry, 3: muddy, 4: swampy and 5: community of the region, quite different from flooded) and the presence of bromeliads and other forests where Schinopsis species may be small mounds in ponds. Relative topographic present (Lewis and Pire 1981, Prado 1993). elevatíon of 40 points on different microzones Several aspects of this forest have been studied along one of the transects, was measured with by Ragonese and Covas (1940) and Lewis and an optical level. With these data, cartographic Pire (1981). Lewis (1991) suggested that the diagrams of vegetation and microzones were clumped distribution of tree individuals in this drawn on a grid of quadrats at a scale of 1: 100. forest is the result of the spatial heterogeneity Microzones were grouped in the following and that microzones are colonized differentially categories: convex with bromeliads (EH), con­ by woody species. However, a detailed analysis vex without bromeliads (E), level most dry (PS, of its spatial heterogeneity and the development that includes PI, P2 and P3), level most humid of its pattern have not been carried on so faro The (PH, that includes P4 and P5) and concave (D). object of this paper is to analyse the spatial het­ Microzones are not continuous over the transect erogeneity of one stand of a S. balansae forest plots, but divided in severa} sectors each, and near Vera (Santa Fe, Argentina) and to establish each sector can be spread in one, two or more whetherthere is any correlation between the dis­ plots. tribution of the woody species and environmen­ Results were analysed following two differ­ tal factors. The underlying hypothesis is that the ent approaches (Fig. 1). First, to analyse the pattern of the woody layers is correlated to sorne spatial heterogeneity two cluster analyses of soil characteristics, mainly microrelief and plots with complementarydata matrix were per­ edaphic humidity. formed. Plots were classified according to the BARBERIS el al.: SpatiaI heterogeneity and woody species distribution in the Chaco 517 number of trees taller than 1.6 m and shrubs To analyse woody species distribution on taller than 0.6 m, with Mulva's (Wildi and different microzones an ordination method was Orlóci 1990) Mínimal Variance Clustering tech­ performed. Microzone diagrams were com­ nique using van der Maarel's coefficient to con­ pared by superimposing them on the carto­ struct the resemblance matrix. This technique graphic diagrams of vegetation, and the number and similarity measure were chosen because of individuals of woody species in each micro­ they proved to be the most robust technique zone sector was calculated, and results were available for these data (Torres et al. 1995). Jater corroborated in the fieId. The area of the Plots were alsoClassified according tothesur­ microzonesectors was estimatedwithan area" face proportion of different microzones with meter. The area of different sectors of micro­ farthest neighbour technique, using Sorensen zones is very variable ranging from 20 to 315 coefficient to construct the resemblance matrix m2, therefore, in order to compare woody (McCune and Mefford 1995). In this case this species distribution on microzones sectors, den­ technique was used because it is space-dilating, sity provides a more appropiate measure than so group separation is more evident (Greig­ the number of individuals. Woody species and Smith 1983). microzone sectors were ordered according to

WOODY INDIVIDUALS MICROZONE SEClDRS

CARTOGRAPHIC DIAGRAMS

SUPERPOSITION OF BOTII CARlDGRAPHIC DIAGRAMS

SAMPLlNG UNITS

PLOTS MICROZONE SEClDRS ¡¿ � W MICROZONE PLOTS PLOTS SEClDRS WOODY %OF WOODY MATRIX SPECIES D MICRO O SPEClES D ABUND. ZONE DENSITY .J; .J; CLUSTER � CLUSTER CORRESPONDENCE ANALYSIS ANAL YSIS -7 ANAL YSIS ANALYSIS.

DENDROGRAM COMPARISON "" .J; WOODY SPECIES D1STRIBUTION SPATlAL HETEROGENEITY ON DIFFERENT MICROZONES

Fig. 1. A simplified representation of steps in data anaIysis. 518 REVISTA DE BIOLOGIA TROPICAL the density of individuals in each sector by one, two plots without any individuals at aH, are means of Correspondence Analysis and grouped together (7), appart from most other Decorana from Pc-ord (McCune and Mefford plots which are clustered in four groups (3, 4,5 1995). Default option of Decorana was used. and 6) f10ristically quite different from each Only those sectors with an area higher than 20 other as shown in the dendrogram and Table. m2 were analysed, depressed sectors (D) and Zizyphus mistol Gris., Sideroxylon obtusifolium species with less than three individuals were not (Roem. et Schult.) Pennington, Schinusfascicu­ taken into account to avoid noíse and distortion latus (Gris.) Johnst. and Aspidospemza quebra­ of results. cho-blanco Schlecht. are characteristic species of the first main group, in contrast to decorticans (Hook. et Arn.) Burk. and to a less­ RESULTS er degree, spp. (P. alba Gris., P. nigra (Gris.) Hieron. and their hybrids), which are Indíviduals of woody species are clumped, characteristic species of the second one (Fig. 3). therefore tree canopy is very dense in sorne The distribution of microzones defined by places, but nevertheless discontinuous, so that the microtopography shows a gradual transition all along the transects there are gaps that can be in sorne places, such as plots 6 and 7, while in rather large (Fig. 2). The large gaps are covered others there is an abrupt change as in the most by a dense understory of grasses, whíle the part of the boundary between plots 7 and 8 (Fig. more shaded parts of the forest are ínva.ded by 2). There are also very small ponds in the high­ two spiny bromeliads (Aechmea distichantha er microzones líke the one near the limit Lemaíre and Bromelía serra Gris.). between plot 8 and 9 or larger ones like one in the southern part of plots 10 and 11 (not shown

Plot 6 Plol 8 Plol 9

Pn3

West East

o 5 m

Fig. 2. Cartographic diagram of microzones and woody vegetation of plots 6 10 9. Species codes: Ap: Acacia praecox, Aq: Aspidosperma quebracho-blanco.Bo: Sideroxylon obtusifolium, Ca: Cocc%ba argentinensis, Cr: Capparis retusa, Cs: Celtis pal­ lida, Ec: Myrcianthes cisplatensis, Gd: Geoffroea decorticans, Mv: Maytenus vitis-idaea, Pr: Prosopis spp., Sb: Schinopsis bal­ ansae, Sp: Schinus fasciculatus, Tn: Achatocarpus praecox, Zy: Zizyphus mist%�� Microzones codes: E: convex, P: level, D: con­ cave; m: slepped, n: sloped; 1: well drained, 2: dry, 3: muddy, 4: swampy, 5: flooded; b: bromeliads.Arrows indicate water runof f.

Plots along the transects show differences in in Fíg. 2). The bromeJiads occur only in the their f10ristic compositíon and when they are highest and better draíned microzones, but not classified, they forrotwo main groups, one with all higher microzones necessarily have bromeli­ dense woody vegetation and the other with ads. In relatively deep ponds, small hummocks more sparse woody vegetation (Fíg. 3). The first near the edges may be common. group is divided in two subgroups (1 and 2) The most humid microzones, or the best according to the relative abundance of Acacia drained ones are in concave areas or conversely praecox Gris. and S. balansae. In the second in the convex parts of the ground, but this does BARB.ERIS et al.: Spatial heterogeneity and woody species distriblltion inthe Chaco 519

2.25

2.00

1.75

Distance

1.50

1.25

1.00 __--L. __ Group 2 3 4. 5 6 7

Plots 13 12 9 8 28 29 11 26 27 19 1720 18 10 15 23 22 7 14 6 3 31 1 4 30 25 J2 24 16.21 Me.e. B A A A A A A A A D D D e A e D e D A B D e B B e B B e B D D D Species códe Sb 4 4 11 4 ·2 2 4 Cs 2 2 4 4 3 6 12 2 2 1 Al' 17 17 10 10 10 6 6 6 6 2 3 2 6. 4 Ee 2 2 6 4 Cr 1 4 6 4 1 Tn 2 6 4 4 Mv ·1 3 1 Zy 2 1 Bo Sp Aq p¡' Gd Gw Ca Ci Ly Ce Af La Al Ac

Fig. 3. Dendrogram afthepermanent plots cIas�ificationaccording to the �umbec ofil1divid�a1srendered by MULVA'sMinirnal Vanance Clilstering technique using van derMaarel's coefficient foc constructingthe resemblance matrix (above) and asSOCÍ¡l­ tionTable of the 32 plots in the Schinopsis ba/ansae foresr'ríear Vera.Seores are numher of individualsper plot (below). Species codes:.Af: Acanthosydsfalcata, Ac: Á.cacia caven,AI: Aloysia gratissima, Ap: Acacia praeco�, Aq: Aspidosperm(l.quebracho­ blpnco, Bo: Sideroxy/on obtuSijolium, Ca: Coccolob(l argentinensis, Ce: Erythroxylum microphyllum, Ci: Ce/tis iguanea, Cr: La: Cápparis retusa, Cs: Celtispa/lida, Ec:Myrcianthes cisp/atensis, Gd:.Geoffroea decortÍ(:ans, Gw: Gra,b9w:vkia tfuplic(lta, Tabebuia nodosa, Ly:Lyciumcuneaíum, Mv: Maytenus vitis-idaea, Pr: Prosopisspp., Sb: Schinops�s ba/ansae; Sp:Schinusfas­ cicu/atus, Tn: Achatocarpus praecox, Zy: Ziz)phusmisto/. M.C.C.: Plot'sclusters aeeording to the proportion·.ofJJÜcrozones, as in Fig. 4. 520 REVISTADE BIOLOGIA TROPICAL

not mean that they are the lowest or highest the right of the first axis and all other species to parts otthe ground, but only relatively so. Most the left in two clusters, one (H) with Acacia of the area, (little more than 54 %) is flat, half praecox,A. quebracho-blanco and S. obtusifoli­ of it swampy or frequently flooded,and only lit­ um in the upper side of axis 2, and another (III) tle more than 11 % is concave most of which is with Myrcianthes cisplatensis (Camb.) Berg., often flooded. Fig. 2 is a fairly representative Achatocarpus praecox Gris., Cocc% ba argen­ sample of the distribution of microzones, tinensis Speg., Ce/tis pallida Torrey, Capparis although as a whole. the analysed area has a retusa Gris., Maytenus vitis-idaea Gris., and S. lower proportion of well drained soils. fasciculatus to the bottom of the same axis, thus The distribution of microzones among plots leaving Z. mistol between both clusters and S. is very variable, so when plots are classified balansae in the uppermost part of the diagram according to microzone proportion, they form (Fig. 5). If the same analysis is done with two main clusters. One of them, groups plots Detrended Correspondence Analysis which are mostly with a convex microrelief (A) (Decorana) (Hill and Gauch 1980), the results and the other one, groups plots which have a (not shown here) are similar, but with all lower proportion of convex microrelief. The species of the dry sites at the left side of axis 1, second main cluster is subdivided in a group apart from S. balansae, in a verycompact clus­ with plots mainly dry level (B), and another ter, that cannot be segregated on axis 2. which is also divided in two groups, the fina with plots mainly humid level (D) and the sec­ DISCUSSION ond one with plots with a mixture of micro­ zones (C) (Fig. 4). The tree layer of this stand is spatially very Most individual s of woody species are dis­ heterogeneous (Lewis et al. 1997) and plots are tributed mainly on well drained convex and grouped in high resolution clusters. While plots level soils. Very few individuals occur on the form well defined clusters, floristic groups are concave and very wet soils (Fig. 2); while well not so well defined, and plot clusters are char­ drained soils account for only 35 % of ground acterized more by therelative abundance of the surface, 75% of the woody species individuals species, rather than by their floristic composi­ appear on them. If the classifications of plots tion. If the classification is done with the same according to their floristic composition and the method, but using as resemblance matrix the proportion of microzones are compared, at first cross product of non center data matrix, results sight appears a rather narrow correspondence are almost the same. Moreover, the dendro­ between them. For example, plots of groups 1 grams were quite similar to that published by and 2 which have more individuals of woody Lewis (1991) constructed using only 22 of these species, as a whole belong to plot group A with plots. larger proportion of higher soils (EB and E), The microzone rnaps (Fig. 2) show that the while most plots of groups 5 and 6 belong to soils have irregular rnicrotopography, which in plot group B and the empty plots (group 7) are certain places is rnarkedly contrasting. Water in plot group D with large proportion of humid accurnulates in the concave parts of the soil soils (Fig. 3 and 4). which are not necessarily the lowest parts. Correspondence analysís of the mícrozones Therefore the wettest soils are in the deepest sectors segregates to theright of the fírst and to parts of depressions while convex soils are the apper part of the second axes the most always well drained. Hurnmocks in the edges of humid ones and to the left and the bottom of the sorneponds seem to be the result of erosion pro­ same axes, the best drained microzones (Fig. 5). duced by livestock trampling. Brorneliadscolo­ The same anaIysis of the species, groups them nize the relatively higher rnicrozones, but as . in an open cluster (1) with Prosopis spp., G. sorne of the convex and well drained soils have decorticans and Acacia caven (Mol.) Mol. on no brorneliads at aH, sorne other factor, perhaps . BARBERIS et aL: Spatial heterogeneity and woody species distribution in Ihe Chaco 521

2,79_

2,44

2,09_

O

1,74_

1,39 A B e D

a

n 1,04_

e 0,69_

0,35_

0,01-

Plots 292827 9 8 12 26 11 10 7 14 31 32 30 13 25 1 23 18 15 4 3 2 6 17 24 1920 5 22 16 21 100"10

80% C,) '" � 60% '"S ... 40% .9 ¡:¡., 20%

0%

Fig. 4. Above: Dendrogram of plots classification according lo Ihe proportion of microzones, using Sorensen similarity coeffi­ cient and farthest neighbour c1ustering technique; A, B, C and D are the main plot's groups. Below: Microzone proportion of each plot. Microzone codes: EB: convex with bromeliads (black), E: convex without bromeliads (inclined hatched), PS: leve! most dry (horizontal hatched), PH: leve! mos! humid (dotted), and D: concave (white). light intensity, may condition their presence in is a correspondence between the plots which therelatively high soils (Brokaw 1983, Lee et have a high number of trees and higher and bet­ al. 1989). Also the two species of bromeliads ter drained microzones. tend to be segregated from each other. Aggregated distribution of woody individuals The spatial heterogeneity of this stand can have been linked to different factors such as: be indicated by tree distribution within the seedfall patterns (Hubbell 1979), edaphic condi­ plots, which as shown aboye, is clumped (Fig. tions (Tumer and Franz 1985) or canopy struc­ 2). As pointed out by Palmer and Dixon (1990), ture (Augspurger 1984). In thisforest the analy­ many plots may not represent a single point on sis of thedistribution of woody species in differ­ any of the environmental gradients, but rather en! microzones, strongly supportsthe hypothesis occur over a segment of certain length. The thatthe clumped distribution of treesis the result analysis of plots in this forest shows that there of edaphic or topographic heterogeneity as it 522 REVISTA DE BIOLOGIA TROPICAL

Nevertheless, distribution of species on differ­ o Sb ent microzones is not uniform, Acacia praecox, 150 M. cisplatensis, C. pallida, C. relusa and most species occur on dry soils and cannot be on 100 lower and very wet soils, while G. decorlicans, Prosopis spp. and A. caven grow on wetter 50 soils. S. balansae can grow on more humid soils than the first group, and actualIy in sorne places it appears next to species of the second group, -50 but grows on well drained soils as well.

-100 Prosopis spp., G. decorticans and A. caven, which are present mainly in the mos! humid

_ ___� -150 L_-----"2::±===---_ _ soils in this forest, are the dominant woody -100 -50 O 50 100 150 200 250 300 species of the Peristeppic Thornforest which is at the south of the Santa Fe Forest Wedge Fig. 5. Scatter diagram on the plane ofaxes 1 and 2 of cor­ (Lewis and Collantes 1973), whereas most respondence analysis of microzone sectors and species. Microzone codes: empty squares (EB): convex with other species which thrive in well drained soils bromeliads, empty triangles (E): convex wíthout bromeli­ of this forest are not present at aH in the ads, filled circles (PS): level most dry, filled diamonds Peristeppic Thornforest. Moreover the (PH): level most humid. Species codes: Ac: Acacia caven, Amazonian Iineage species (sensu Prado 1991) Ap: Acacia praecox, Aq: Aspidosperma quebracho-blanco, are present neither in the most humid soils of Bo: Sideroxylon obtusifolillm, Ca: Coccoloba argentilJelJ­ sis, Cr: Capparis retllsa, Cs: Celtis pallida, Ee: this forest nor in the Peristeppic Thornforest. Myrcianthes cisplatelJsis, Gd: Geoffroea decorticalJs, Mv: Soil heterogeneity, mainly microrelief and MaytelJus vitis-idaea, Pr: Prosopisspp., Sb: Schinopsis bal­ moisture, is the most important factor that deter­ alJsae, Sp: Schinusfasciculatus, Tn: Achatocarpus praecox, mine fue distribution of species on the ground. Zy: Zizyphus mis¡ol. I,n and III: Species clusters. However, how this heterogeneity develops is in need of further research, and severa! hypotheses was previously suggested by Lewis (1991). can be putforward. The irregularity of the sur­ AIso as in many other Chaquenian forests; there face may be the result of the physiographic his­ are islands of trees interwoven with grass patch­ tory of the site, pseudokarstic processes and wind es, the latter usually found in relatively lower deflation as welJ as hydraulic accumulation topographic position (Morello and Adámoli processes, but also plant-soil and animal-soil 1974). Similar causes have been suggested for interactions such as tramplíng and the building the cJumped distribution of woody species in actívity of ants and other small animals should be Chaco-related and non-related communites of considered (Bucher 1980, Popolizio el al. 1980, the Pantanal in Mato Grosso (Prado el al. 1992, Lewis 1991). Adámoli 1995) The abo ve discussion on microsites consid­ The slight differences between both ordina­ ered only microzones defined by microrelief tion analyses to determine the distribution of and soil characteristics associated with it. �oody individuals on microzones, may be the However, each one of these microzones are fur­ result of an artifactof Decorana (Wartenberg el ther divided by the aerial environment, funda­ al. 1987). The analysis shows that most woody mentally by the Iíght c1imate; there are places species do not grow on often flooded soils, exposed to full irradiation, and others on full though sorne individuals of trees and shrubs shadow, with all intermediates. AIso, it should apparently grow on depressed mícrozones; be taken into account that there are more than however, field inspection showed that most of one layer of woody species which are not con­ them were on relatively high ground or liule tinuous, lea ves are of different types and sorne elevatíons out of scale within these mícrozones. species are deciduous and others evergreen BARBERIS el al.: Spatial heterogeneity and woody species distribution in the Chaco 523

(Lewis el al. 1997). SoJar, only a very eoarse micrositios bien drenados mientras ,que en suelos ,muy húmedos solamente crecen decorticans, Prosopis qualitative eharaeterization' of the forest micro­ Geoffroea spp. y S. balansae, Los resultados obtenidos sugierenque la climate can be done.1llerefore, this aspect is in heterogeneidad del suelo, (microrrelieve y humedad) es el need offuítherresearch in order to havea quan­ factor más importante que determina la distribuc:ión de las titativecharacterization of rnicrohabitats. In especiesleñosas sobre el terreno. addition to these static aspects of the spatial het­ erogeneity of the rnicrosites, there are also dynarnic processes at work, sucl) �s changes in REFERENCES herbaceous/woody'species caused by overgraz­ Adámoli,J.1995. Diagnósticodo Pantanal: (Características ing (MorelIo 1970, Adámoli et al. 1990), ecológicas e problemas ambientais). Programa Nacional gallery forest dynamics resulting from intensive do Meio Ambiente,.Brasilia. 50 p. river-bed migratíon processes (Adámoli et al. 1990, Sennhauser 1991) and bther processes as Adámoli, 1., E. Sennhauser, J.M. Acero & A. Rescia. 1990, well, that also need further research. Stress and disturbance: vegetarion dynamics in the dry Chaco regíon of Argentina, J. Biogeogr. 17: 491-500.

Augspurger, CX. 1984. Seedling survivaJ of tropical tree ACKNOWLEDGMENTS species: interactions of dispersal distance, light gaps, and pathogens. Ecology 65: 1075-1712. Financial qid from theCONICET (Council Bazzaz. F.A. 1991. Habitat selection in .Am. Na!. 137 for Scientific and Technical Research, (suppl.): 116-130. Argentina) is gratefuIly acknowledged. The authors are indebted to fhe Santa FeMinistry' of Brokaw, N.Y.L. 1983, Gro1.lndlayer dominance and appar­ Agriculturefor 16gishc �tlpportand to the staff ent inhibition of [ree regeneration py Aechmea, mag­ of Las Gamasex.perjmentaI ,Genter for the,ir dalenae (I3romeliaceae) in a tropical foresto Trop, Ecol. 24: 194-200. invaluable help. J,L. Vespdni and ,E. Bmich field work provided invaluáble help during and Bucher, E.H. 1980. Ec610gía de la fauna chaqueñ�. Una ' P.S. Tones and Fernand�z ' ""ith" the eom­ revisión, Ecosur7: 111'221. }.R" , .' putin� work. D.E. P�adollridE:A. Fnuieesehj EI"limade .laregión nores!edela read QritiFally themanuscript and made" yery Burgqs,J.J.1970. RepúblicaArgentina.j3ol. Soco Arg.Bo!. 11 (supL): 37- useful suggestions, . . lOL

Espino, L.�., MASeveso & M.A. Sabatier. .1983. Mapa 'RESUMEN" de suelos de la Prov. de SantaFe. Tomo n. MAG Santa Fe-INTA EERA Rafaela,Argentina, �20 p. El objeto de esle trabfljo es analizarla ,heterogeneidad espacial de un bosque de Schinopsis balansae Gleason, HA 1926. The individualistic concepto of lheplant ("Quebrachal") cerca de Vera (Santa Fe, Argentina) y assQciation. l3ulLTorrey Bol.Club 53: 7-26 .. establecer sí ,hay correlaCión entre la distribución de las especies leñosas y los factores ambient,ales.Este tipo de Greig-Smith, P. 198'3. Quantitative plantecology, Studies in bosque es la comunidad vegetal más importante de la Cuña Ecology Vol. 9. Universityof California, Betkeley. 359 p. Boscosa. de Sant¡¡ Fe, que es la. porción más austral del Chaco Oriental. En un stand,parcialmente recuperado se Grubb, Pi..1 977. The maintenance, ofspecies rídll}€ss in instalar\ln 32 parcelas contiguas de 10 x 10 m a lo largo de plant communitíes: the importante,'ofthé regeneration dos transectas. Se cartografiaron todos los individuos de las niche. Biol. Rev. 52: 107-145. distintas especies de leñosas, así como los distintos micro­ sitios caracterizados por el microrelieve, humedad del suelo Harper, J.L. 1977. Population biology of plants. Academic, y presencia de bromeliáceas terrestres. El 54% del suelo es London. 492p. plano, la mitad del cual es barroso o está. frecuentemente inundado, y algo más del JI % es cóncavo y está casi siem­ Hill, M.O. & H.G. Gauch, Jr. 1980. Detrendedcorrespon­ pre inundado, La mayor parte de las leñosas crecen en dence analysis: an improved ordination technique. Vegetatio 42: 47-58. 524 REVISTA DE BIOLOGIA TROPICAL

Hubbell. S.P. 1979. Tree dispersion, abundance, and diver­ America. Ph. D. thesis, University of S1. Andrews, S1. sity in a tropical dry fores!. Science 203: 1299-1309. Andrews, Scotland. 420 p.

Lee. H.SJ., U. Lüttge, E. Medina, JAC. Smith, W.J. Crarn, Prado,D.E. 1993. What is theGran Chaco vegetation in South M. Díaz, H. Griffiths, M. Popp, C. Schafer, K.H. Stirnmel Arnerica? l. A review. Contribution to the study of flora & B. Thonke. 1989. Ecophysiology of xerophytic and and vegetation of the Chaco. V. Candollea 48: 145-172. halophytic vegetation of a coastal alluvial plain in north­ em Venezuela. III. Bromelia humilis Jacq., a terrestrial Prado, D.E., P.E. Gibbs, A. Pott & V.J. Pott. 1992. The CAM bromeliad. New Phytol. 111: 253-271. Chaco-Pantanal transition in southem Mato Grosso,

Brazil, 451-470. In: Furley, P.A .• J. Proctor & J.A. Lewis, J.P. 1991. Three levels of floristicaJ variation in the Ralter (eds.). Nature and dynamics of forest-savanna forests of Chaco, Argentina. J. Veg. Sci. 2: 125-130. boundaries. Chapman & Hall, London.

Lewis, 1.P. & M.B. Collantes.1973. El espinal periestépico. Ragonese, A.E.& G. Covas. 1940. La distribución geográfi­ Ciencia Investigación 29: 360-377. ca de los quebrachales en la provincia de Santa Fe. Rev. Arg. Agron. 8: 176-184. Lewis, J.P. & E.F. Pire. 1981. Reseña sobre la vegetación del Chaco santafesino. Serie Fitogeográfica 18. INTA, Ragonese, A.E. & J.e. Castiglioni. 1970. La vegetación del Buenos Aires. 42 p. parque chaqueño. Bol. Soco Arg. Bo!. 11 (supl.): 133- 160. Lewis, J.P., E.F. Pire & I.M. Barberis. 1997. Structure, physiognomy and floristic composition of a Schinopsis Sennhauser, E. 1991. The concept of stability in connection balansae (Anacardiaceae) forest in the Southem Chaco, with the gallery forests of the Chaco region. Vegetatio Argentina. Rev. Biol. Trop. 45: 1013-1020. 94: 1-13.

McCune, B. & MJ. Mefford. 1995. PC- ORD. Multivariate Sousa, W.P. 1984. The role of disturbance in natural com­ analysis of ecological data, Version 2.0. M.J.M. munities. Annu. Rev. Ecol. Sys1. 15: 353-391. Software Design,Gleneden Beach, Oregon. 126 p. Torres, P. S., I.M. Barberis & J.P. Lewis. 1995. Robustness Morello, J. 1970. Modelos de relaciones entre pastizales y of numerical methods for vegetation cJassification. leñosas colonizadoras en el Chaco Argentino. IDIA 276: Coenoses 10: 11-16. 31-52. Tumer, D.P. & E.H. Franz. 1985. Size cJass structure and Morello, J. & J. Adámoli. 1974. Las grandes unidades de tree dispersion pattems in old-growth cedar-hemlock vegetación y ambiente del Chaco argentino. Segunda forests of the northern Rocky Mountains (USA). parte: vegetación y ambiente de la provincia del Chaco. Oecologia 68: 52-56. Serie Fitogeográfica13. I.N.T.A., Buenos Aires. 130 p. Urban, D.L., R.V. O'Neill & H.H. Shugart. 1987. Palmer, M.V. & P.M. Dixon. 1990. Small-scale environ­ Landscape ecology. Bioscience 37: 119-127. mental heterogeneity and the analysis of species distrib­ utions along gradients. lVeg. Sci. 1: 57-65. Wartenberg, D., S. Ferson.& F.J. Rohlf. 1987. Putting things in order: a critique of detrended correspondence analy­ Popolizio, E., P.Y. Serra & G.O. Hortt. 1980. Bajos siso Am. Na!. 129: 434-448. Submeridionales. Grandes unidades taxonómicas de Chaco. Centro de Geociencias Aplicadas. Serie C - Whittaker, R.H. 1967. Gradient analysis of vegetation. Biol. Investigación. Resistencia, Chaco, Argentina. 205 p. Rev. 42: 207-264.

Prado, D.E. 1991. A critical evaluation of the floristic links Wildi, O & L. Orlóci. 1990. Numerical exploration of com­ between Chaco and Caatingas vegetation in South munity patterns. SPB Academic, The Hague. 124 p.