Habitat structure and vegetation relationships in central Argentina salt marsh landscapes

Plant Ecology 137: 79–100, 1998.© 1998Kluwer Academic Publishers. Printed in Belgium. 79

Habitat structure and vegetation relationships in central Argentina saltmarsh landscapes

Juan Jose Cantero1, Rolando Leon2, Jose Manuel Cisneros1 & Alberto Cantero11Universidad Nacional de R´ıo Cuarto, Facultad de Agronom´ıa y Veterinaria , Agencia Postal N◦3, 5800 R´ıoCuarto, Argentina.2Universidad Nacional de Buenos Aires, Facultad de Agronom´ıa, Departamento de Ecolog´ıa,Avenida San Martin 4453, 1417 Buenos Aires, Argentina

Received 20 March 1997; accepted in revised form 26 March 1998

Key words:Gradient analysis, Hydrohalomorphic landscapes, Vegetation-environment relationships

Abstract

Relationships between habitat structure and spatial variations in vegetation composition were determined in catenasof central Argentina salt marsh landscapes. Vegetation was classified following a multi-technique strategy. Ananalysis of species distributions along an environmental gradient was made and a redundancy analysis was used torelate the environmental variables to vegetation data. The spatial covariation was evaluated through fractal analysis.The vegetation can be divided into four discrete noda that correspond to different topographic positions (summit,backslope, footslope and toeslope, respectively): nodum ofStipa trichotoma+ St. tenuissima, nodum ofDistichlisscoparia, nodum ofDistichlis spicataand nodum ofSpartina densiflora. Each of these nodum is characterizedby a definite combination of floristic groups. Symmetric aggregation of vegetation borders was observed in allthree sites. The existence of vegetation discontinuities along the catenas depended largely on water table depth andsalinity which in turn controlled the edaphic salinity. The arrangement of sites by multivariate analysis reflected theinfluence of a complex gradient representing halomorphic and hydromorphic factors.β-diversity was associatedwith abrupt changes in the physical structure of the catena for a reduced spatial dimension (300-m scale). Absolutediversity and evenness were higher on the summit and declined progressively toward the toeslope. The rate ofchange was higher on the backslope and the dominant species have different ecological amplitudes overlappingalong the gradient. The main operational factors associated with the floristic differences are: (i) the variations inthe chemical composition and the seasonal dynamics of the soil solution in the aerated layer of soils, (ii) the salinityand dynamics of groundwater, and (iii) the length of time that the soil is flooded during the rainy season (summer).The fractal dimension was close to 2, implying weak spatial dependence. Fractal dimension varied as a functionof scale. The fractogram only revealed a significant spatial dependence on the summit. This spatial dependencewas associated with short distances of gradient showing that the organizational pattern ofDistichlis spp.- Spartinawas related to combinations of underlying environmental factors rather than to a specific position in the catena.The catenas are highly structured spatially, floristic compositions are inextricably linked to this structure. Habitatcomplexity may directly affect associated vegetation by regulating the hydrohalomorphic conditions in the aeratedlayer of the soils. The relationships of the habitat-floristic groups are not simple, hydromorphism interacts in acomplex way with halomorphism.

Introduction

Many ecologists call the construction of the naturalenvironmenthabitat structure(Downing 1991). Thehabitat is spatially structured by various energy in-puts, resulting in patchy structures or gradients. These

discontinuities and heterogeneity are the result ofgeomorphologic processes, which through a spatio-temporal structuring of the physical environment in-duces a similar organization of living beings and ofbiological processes (Legendre & Fortin 1989). Thesalt marsh habitats are highly structured environmen-

80

tally and they can provide a gradient of environmen-tal conditions from extremely inundated and salineto relatively mesic. This monotonic gradient is thedominating environmental influence on the vegeta-tion patterns and it usually causes obvious zonation(Pielou & Routledge 1976). The soil moisture-salinityinteraction has been widely recognized as the mostimportant factor in the distribution of the salt-tolerantplants (Ungar 1967; 1972, 1974; Waisel 1972; Flow-ers 1975; Benayas & Scheiner 1993). The chemicaland hydrophysical characteristics of soils also affectthe diversity and structure of the vegetation (Chapman1960; Pomeroy & Weigert 1981; Butler et al. 1986).Flooding is the other active enviromental factor influ-encing the vegetation pattern (Ranwell 1972; Chap-man 1974; Sykora et al. 1987; Linthurst & Senaca1981). When no obvious habitat heterogeneity is ob-served the spatial aggregation of vascular plants canbe the result of many different causes including inter-specific competition (Ungar 1972; Pielou & Routledge1976; Brotherson 1987; Brewer & Grace 1990).

In central Argentina, the processes of saliniza-tion, sodification, flooding and sedimentation of soilsinvolve 1.5–2.2×106 ha (INTA 1990). In these hydro-halomorphic plains, some geosystems locally called‘catenas’, show a typical vectorial structure (Solntsiev1974; Ruhe 1975). Those catenas are ideal examplesof geochemical gradients because salinity composi-tion and groundwater depth changes along the flowpaths, and soils influenced by groundwater followthis gradient (Cantero 1993; Cisneros et al. 1996).There are no quantitative studies which identify themajor environmental factors correlated with compo-sitional gradient of vegetation in these catenas. Thepurpose of this study was to examine the effects of saltmarsh environmental heterogeneity (Li & Reynolds1995) on the spatial distribution of floristic groups andvegetation noda. The techniques of gradient analysishave been used (ter Braak & Prentice 1988) to in-dicate which characteristics of structured habitat areecologically important. The specific objectives are:(i) describe the structure and ecosystem functioning ofthese topolithosequences (ii) analyze and describe thecompositional variation of the vegetation and (iii) ex-plore the relationships between the vegetation systemsand the environmental factors.

Methods

Data collection

The study area is located in central Argentina (Fig-ure 1). It is a great plain occupying two millionhectares, wich is characterized by fluvial paleoactiv-ity and interconnected lagoons. The geomorphologi-cal surfaces vary from catenas of variable slope andlength, to concave cells, hydrologically connected, toconvex or flat ones that are associated with differentvegetational systems. The entire area shows hydro-halomorphic characteristics. It belongs to a semi-drydomain with high mesothermal uniformity and with-out an excess of water receiving monsoon type rain-falls. Water table oscillation results from a rechargeduring the warm period (October–March) and a dis-charge during the cold period (April–September).These endogenous rhythms of the ecosystem show animportant internal heterogeneity related to evapotran-spiration and rainfall in each period; the mobilizingvehicle of salts being water in the warm period andwind in the cold period (Cantero 1993; Cisneros1994).

Three transects which involved similar environ-mental situations in distant geographic points withinthe region, La Chanchera (T1), Pacheco de Melo (T2)and Assunta (T3) were studied. Sample plots of 20 m2

divided into 100 equal subunits, were located in ap-proximately 12 sites for each transect (total composedsamples= 35), at variable intervals in accordancewith the different combinations of floristic groups.For each sample subunit, the presence of all observedspecies was recorded. The final expresion of speciesabundance was a frequency value, ranging from 1 to9 (Acosta et al. 1991). The total number of identi-fied species was 119 using the taxa nomenclature ofCantero & Bianco (1986). Monthly from September1986 to December, we evaluated the groundwater level(wt) and chemistry in each sampling site, through aseries of observation holes. Titration with acids wasused to determine carbonates and bicarbonates (COH)(Richards 1973), titration with Ag NO3 to determinechlorides (CL) (Richards 1973). The difference withthe sum of the remainder anions provided the sul-phate (SO) concentration (Bresler et al. 1982), titrationwith EDTA determined calcium (CA) and magnesium(MG) (Richards 1973), and flame photometry wasused to determine sodium (NA) and potassium (K).The soil properties of different sites were determinedas follows:soil texture, by means of the pipette method

81

82

for the 2, 20 and 50µ fractions; the dry sieve was usedfor sands, separating the 44–88, 88–210, 210–500 and500–1000µ fractions; exchangeable Na, due to thehigh soil salinity, the exchangeable sodium percent-age (ESP) was estimated by the RAS-ESP relationship(Richards 1973);soil salinity, was estimated by elec-trical conductivity (EC) of soil saturation extracts andanions and cations from each horizon were analyzed.

Environmental data were standardized for theirmaximum value:y = 100x/x maximum. Two nom-inal variables were considered: topographic position(top) and natural drainage, with four classes in eachone (Etchevehere 1976). The quantitative variableswere: water table depth (wtd), summation of anions+ cations of thewt (suac), low permeability horizondepth (hord), wt electrical conductivity/wt depth ra-tio (wtecd), electrical conductivity weighted by soilprofile (soilec) (weighted average of the electrical con-ductivity obtained from the saturation extracts of eachhorizon in dS m−1), weighted average geometric di-ameter by profile (geomd) (weighted average of themean average geometric diameters of particles in mm,according to Campbell 1978), cationic exchange ca-pacity (cic), soil moisture content at 0.3 b (wt3) andsoil moisture content at 15 b (wt15).

Data analysis

Classifying vegetation dataA community classification was based on a multi-technique strategy applying Wildi’s proposals (1989).The description of algorithms for the different meth-ods can be found in Wildi & Orloci (1990). Weconducted multivariate numerical procedures with theMULVA-4 package (Wildi & Orloci 1990). The ter-minology proposed by Orloci and Stanek (1979) wasused to identify the vegetation types or noda.

Analyzing vegetation-environment relationshipsAs the length of the gradient was 1.88 SD (accord-ing to a previous Detrended Correspondence analy-sis), it was supposed that the greater part of theresponses of the species were monotonic (ter Braak& Prentice 1988), and advantages could be achievedwith respect to the quantitative information of thevegetation-environment relationships, employing ananalysis technique that assumed an underlying modelof linear responses of the species (Jogman et al. 1987;ter Braak 1987; ter Braak & Prentice 1988). We con-sidered the Redundancy Analysis (RDA) to be onetype of direct gradient analysis which fulfills this

requirement we’ve chosen and we selected it to deter-mine dominant compositional gradients. The programCANOCO, version 3.10 (ter Braak 1990) was usedfor all ordinations. For the examination of abundanceof the dominant species in the ordination space, weplotted the multiple species responses (as fitted byGeneralized Linear Modelling -GLM- and the bino-mial distribution assumed for the species responses)against the values of the position of the correspondingsamples on an ordination axis. Plots were made withthe CanoDraw program (Smilauer 1992).

Measuring compositional change along gradientThe analysis of species distributions along an en-vironmental gradient was made using Wilson &Mohler’s procedure and the GRADBETA program(Wilson & Mohler 1983, 1986). The method cal-culated: (a) length of the gradient (β-diversity inhalf changes, HC) in terms of species compositionalturnover, (b) the relative rate of compositional changeat points along the gradient, and (c) an ecologicallymeaningful spacing of samples along the gradient. Wealso interpreted the pattern of changes in diversity inthe ordination plane in terms of the environmental fac-tors, using point-kriging plots based on the sphericalmodel (Smilauer 1992).

Evaluation of the spatial covariation of thevegetation-environment relationshipsThe possibility that the data in the catena were struc-tured in the space (autocorrelated) did not guaran-tee correct conclusions with respect to the study ofthe coenocline, unless a quantitative measure ex-isted of the spatial complexity which would serve asconceptual support for evaluating their dependencelevel (Burrough 1983a, b; Phillips 1985). Mandelbrot(1962) introduced the concept of fractal, a geomet-ric form which exhibits structure at all spatial scales.Fractal geometry is a useful tool in summarizing thespatial variation of the vegetation, because composi-tional patterns cannot be closely described by Euclid-ean forms (Palmer 1988). Fractals are difficult tomeasure consistently in real habitats, but their appar-ent variability may well be related to some importantaspects of the variability of communities (Williamson& Lawton 1991). Fractal geometry allows dimen-sions between 1 and 2, and most graphs of ecologicaldata are likely to fall between the extremes (Palmer1988). In this study, the spatial scales of covaria-tion between the environmental descriptors and thevegetation was evaluated employing results from an

83

RDA ordination following the procedural suggestionsof Palmer (1988) and using the catena ‘La Chanchera’as a reference. First RDA axis scores were chosen forthe construction of the semivariogram and calculationof the fractal dimension. The environmental variationpattern coordinated with the vegetation was studied inthe fractogram generated from the representation ofthe fractal dimension D as a function of position alongthe catena. The semivariogram was calculated withthe statistical program written by Robertson (1987)and the fractal dimension was obtained according toBurrough’s method (1983a, b). Only distances lessthan half of the total transect were represented sincethe larger distances are more affected by low samplesizes and spurious properties of the data (Journel &Huijbregts 1988; Palmer 1988).

Results

Compositional variation of the vegetation

Twenty-four floristic groups were identified, socio-logically structured in four relévé groups or noda.The arrangement of sites by the multivariate analysisreflected the influence of a complex gradient rep-resenting halomorphic and hydromorphic factors incompositional patterns, from group 1 in the moreelevated topographical positions to group 24 in thelower topographical positions (Table 1). The sequencebegins with nodum A (Stipa trichotomaNees+ St.tenuissimaTrin.) on the summit, B (Distichlis sco-paria (Kunth) Arechav.) in the backslope, C (Dis-tichlis spicata (L.) Greene) in the footslope and D(Spartina densifloraBrongn.) in the toeslope. NodumA or Stipa trichotoma+ Stipa tenuissimais a tallgrassland, made up of the floristic groups 1, 2, 3,4, 5, 7, 8 and 9, the most frequent biological formsare caespitose and scapose hemicryptophytes as wellas scapose and caespitose therophytes. Nodum B orDistichlis scopariais a dense, short grassland, inte-grated by florisitic groups 6, 10, 11, 12, 13, 14, and21, the dominant biological forms are geophytes withrhizomes, geophytes with sprouting roots and cespi-tose hemicryptophytes. Nodum C orDistichlis spicatais a short grassland with predominance of rhizoma-tous and sprouting root geophytes, integrated by thefloristic groups 13, 15, 16, 17, 18 and 22. Nodum Dor Spartina densiflorais a tall grassland integrated byfloristic groups 19, 20, 23, and 24. It is characterizedby hemicryptophytes and caespitose therophytes.

Geosystemic structure

The soil genesis in these ecosystems is a complexprocess which involves variations in the origin ofparental material and relief-drainage conditions. Allof the region has a fluvial origin but its original ma-terials were deposited in different time periods. Morerecently, over the old fluvial material, eolic episodesoccurred that totally or partially covered the plains,originating poligenetic soils, frequently presentinglithological discontinuities. This edaphic complexityis common for all the studied catenas (Figure 2). InTable 2 some soil and hydrologic characteristics of thestudied noda are shown.

In the summit (Hall 1983) the soils are developedover sandy loam eolic material and taxonomicallyare classified asTypic Hapludollsand Entic Haplu-dolls (USDA 1992), characterized by poor horizontaldifferentiation, and some excesive internal drainage.The wt oscillation is always below the critical level(Wosten et al. 1988). For this reason the soil profileis non saline in the first 150 cm. In the backslope,the soil (Thapto natric Entic Haplustolls)shows poli-genesis (contrasting aeolian-fluvial soil materials arefrequently found) and evidence of halomorphism (Bnahorizons). In this topographic position, thewt os-cillation is near the critical level, and soil surfacesalinization is frequent. This salinization intensity inthe backslope is related to grassland management andto the capillary conductivity –wt level relationships(Cisneros 1994). The soils associated with the foots-lope show clear evidence of hydrohalomorphicgenesis(Bna, duripan and fragipan horizons), and taxonomi-cally areTypic Natraquoll, Typic NatraqualfandTypicDuraqualf. Thewt oscillation is always over the criti-cal level, and its EC define the average profile salinity.The toeslope is characterized by extreme external andinternal drainage conditions. The soil genesis is gov-erned by a severe reducing environment, althoughwith highly variablewt salinity. In these sites the soilranges betweenTypic DuraqualfandTypic Natraqualf(USDA 1992).

The three principal water and salt flow processesare the capillary rise, the infiltration-percolation andsurface runoff. Their balance in each elementarylandscape cell defines the intensity of hydrohalomor-phism (Koslovski 1972). In the summit, infiltration-percolation prevails. Due to its coarse soil texture,surface runoff (of fresh water) occurs only when therain intensity is high. Under these conditions thewt isprobably a source for the vegetation and the capillary

84

Table 1 Vegetation table documenting the communities of catenas from central Argentina saltmarsh landscapes; species scores arefrequency values ranging from 1 to 9. Numbers between brackets are floristic groups.

85

Table 1 Continued.

86

Figure 2. Topographic sequence of soil and vegetation along a landscape gradient in the Central Argentina salt marshes.

rise rarely reach the soil surface. In the backslope, theflow equilibria is more complex, since these landscapeelements behave as both a surface and subsurface flow,receiving and emitting water. The surface runoff salin-ity is related to vegetation coverage (Cisneros 1994),with coverage the runoff is smaller and more diluted,conversely, without plant cover, the topsoil salt contentis greater and the infiltration is lower thus increasingthe runoff portion and salinity. In this position thecapillary rise is important and defines, as a positivefeedback loop, the runoff salinity. The footslope is alsoa receiving-emitting landscape cell, but with a highercapillary rise intensity. The soil infiltration is veryslow because of thewt proximity and the unfavourabletopsoil hydrophysical conditions. With very low infil-tration and intensive surface salinization, this cell is animportant water and salt source by surface runoff. Thetoeslope constitutes the most unstable position of thelandscape due to the nature of its construction: a water,salt and sediment receiving area. The flooding periodis always over 60 days per year, although its salinity isvariable, ranging between 1 to 20 dS m−1 as a func-tion of the capillary rise-infiltration-runoff-vegetationrelationships of the other associated landscape cells.

Compositional and environmental gradient

RDA found colinearity when trying to fit the nominalvariabledrainage class, therefore it was automaticallyeliminated. The values of the correlation coefficientsbetween the environmental variables and the axis forRDA ordination (Table 3 and Figure 3) suggest thatthe first axis was associated withwtd and the secondwith a relationship betweengeomdand the water hold-ing capacity. In the RDA ordination plane (Figure 4)we could clearly see the importance of some variablessuch assoilec, suac, wtecdandwtd. The direction ofthe greatest abundance of groups 15 (Distichlis spi-cata) and 20 (Spartina densiflora) coincided with thevariableswt3, wt15, cic, top, soilec and suac. Thisreflected a close relationship between those groups(noda C and D) and the environments with a highsoil salinity, regulated by its connection to the salineand shallowwt. These combinations are frequent inthe lowest topographical positions of the catenas. Incontrast, for the same variables, the most importantmembers of groups 6 (Hordeum stenostachysGodr.)and 12 (Cynodon dactylon(L.) Pers.) were negativelycorrelated. On the other hand, the most representativespecies of group 1 (nodum A)Stipa trichotoma, Eu-

87

Table 2. Some edaphic and hydrological conditions for hydrohalomorphic noda in the catenas of Central Argentina, Vaqlues are means± standard deviation of the sampling period.

Table 3. Correlation coefficients of abiotic variables with 1st and2nd axis of RDA analysis.

stachys retusa(Lag.) Kunth were positively correlatedto hord.

Wtd, wtecdandgeomd, were positively correlatedwith group 1 (Stipa spp.) and 12 (Cynodon dactylon),indicating a close association with the sites from thetop, related to coarse textural classes which impliedless specific surfaces, less water holding capacity andless capacity of cationic exchange, and finally, witha highly deep level ofwt oscillation. The members ofgroups 13 (Spergula ramosa(Cambess.) D. Dietr.) and14 (Kochia scoparia(L.) Schrader), of nodum B, werepositioned almost orthogonally with respect to all en-vironmental variables, indicating either a very lowcorrelation with them, or an association with their in-termediate values. The member of groups 5 (Cenchrus

pauciflorus Benth.) and 3 (Cenchrus myosuroidesH.B.K.) were located in the center of the diagram,so its relationships with the environmental variableswere not very precise.The succulent species of group15 differed in their ecological behavior, bothSesuviumportulacastrum(L.) L., Heliotropium curasavicumL.and Sarcocornia perennis(Miller) A. J. Scott werenegatively correlated withgeomd, wtecd, wtd andhord, while Distichlis spicataand Cressa truxillen-sis H. B. K. presented low correlations. These lasttwo species appear to be restricted more exclusivelyto stands of the noda C and D affected by prolongedand periodic flooding. In contrast, the extreme andsucculent halophytes of the group had a greater eco-logical amplitude and reached important values ofabundance-cover not only within those noda but alsothey were found integrating nodum B in intermediatetopographical positions.

Becausetop andwtd apparently governed the pat-tern of floristic variation, a partial RDA was per-formed, consideringtop and wtd as covariables andthe rest as the variables of interest. Examination ofFigure 5 suggests that the first axis was associatedwith the water holding capacity and the second axiswith soil salinity.Distichlis scoparia, was clearly re-lated towtecdand was negatively correlated with thevariables relating to water holding capacity. This wasconsistent with the environmental situations which itoccupied in the catenas, wherewtd and wtec haveintermediate values compared to the extremes, andwhere middle to largegeomddominate, with a smallerwater holding capacity. This site contrasts with thelower topographical levels, which have thinner geo-

88

Figure 3. Redundancy analysis, RDA, triplot with (1) sampling plots: (2) environmental variables: topographic position (TOP), water tabledepth (WTD), summation of anions+ cations of thewt (SUAC), low permeability horizon depth (HORD), electrical conductivity of watertable/depth of thewt ratio (WTECD), electrical conductivity weighted by profile (SOILEC), weighted average geometric diameter by pro-file (GEOMD), cationic interchange capacity (CIC), percentage of retained water at 0.3 b (WT3 and, percentage of retained water at 15 b(WT15), and (3) the most abundant species. Species abbreviations: Ambrtenu= Ambrosia tenuifoliaAmmimaju+ Ammi majus, Apiucomm=apium commersonii, Astesqua= Aster squamatus, Baccping= Baccharis pingraea, Baccsten= Baccharis stemophylla, Boopanth= Boopisanthemoides, Brizsuba= Briza subaristata, Bromunio= Bromus unioloides, Cencpauc= Cenchrus pauciflorus, Cencmyos= Cenchrusmyosuroides, Chenmacr= Chenopodium macrospermum, Chloberr= Chloris berroi, Chlohalo+ Chloris halophila, Chloretu= Eustachysretusa, Centpulc= Centaurium pulchellum, Cirsvulg= Cirsium vulgare, Conybona= Conyza bonariensis, Cresstrux= Cressa truxillensis,Cypecory= Cyperus corymbosus, Cynodact=Cynodon dactylon, Daucpusi=Daucus pusillus, Desmdepr=Desmodium depressus, Descarge= Descurainia argentina, Disisang= Digitaria sanguinalis, Digicali = Digitaria californica, Diplunni = Diplachne uninervia, Distscop=Distichlis scoparia, Distspic= Distichlis spicata, Euphport= Euphorbia portulacoides, Franpulv= Frankenia pulverulenta, Gamospic=Gamochaeta spicata, Gamofila= Gamochaeta filaginea, Glanplat= Glandularia platensis, Halimont= Halimolobos montanus, Helicura=Heliotropium curassavicum, Hordsten= Hordeum stenostachys, Hypochil= Hypochoeris chillensis, Kochscop= Kochia scoparia, Lepispic= Lepidium spicatum, Limobras= Limonium brasiliense, Lolimult = Lolium multiflorum, Medilupu= Medicago lupulina, Muhlaspe=Muhlenbergia asperifolia, Oxypsola= Oxypetalum solanoides, Paniberg= Panicum bergii, Paspvagi= Paspalum vaginatum, Pappmucr=Pappophorum philipianum, Plantome= Plantago tomentosa, Pfafgnap= Pffafia gnaphalilioides, Phylcann= Phyla canescens, Polyelon=Chaetotropis elongata, Physvisc= Physalis viscosa, Relbrich= Relbumium richardianum, Rumecris= Rumex crispus, Sesuport= Sesuviumportulacastrum, Spardens= Spartina densiflora, Sciramer= Scirpus americanus, Salivirg= Sarcocornia perennis, Setageni= Setaria parv-iflora, Sperramo= Spergularia ramosa, Sporindu= Sporobolus indicus, Stiperio= Stipa eriostachya, Stiptenu= Stipa tenuissima, Stipnees= Stipa neesiana, Stiptric= Stipa trichotoma, Stippapp= Stipa papposa, Suaepata= Suaeda patagonica, Turnpinn= Turnera pinnatifida,Verbbona= Verbena bonariensis, Verbgrac= Verbena gracilescens, Verbvirg= Verbascum virgatum. Eigenvalue axis 1, 0.35; axis 2, 0.20 rspecies-environment: axis 1, 0.97; axis 2, 0.98; 87.9% variance accounted for by two axes. Monte Carlo test significant (P<0.01). Sum of allcanonical eigenvalues, 0.740.

89

Figure 4. Environmental variables in the RDA ordination plane a)soilec, b) suac, c) wtecdand d)wt. The centres of the circles are placed atthe positions of the individual sites in the ordination space. The size of the circles corresponds to the value of the variable as observed at theparticular site.

logical fluvial materials. Group 15 (Distichlis spicata)had a high correlation with the water holding capac-ity and total soil salinity. In contrast, the membergroups of nodum A (Stipa spp.) were located in theextreme lower part of the third and fourth quadrant,associated with the maximum expression ofhord andgeomd. This combination coincides with the highesttopographical positions of the catenas.

The fractal dimension calculated from the slopem of the double logarithm plot (semivariance as afunction of distance) was estimated to be 1.47 (Di=(4–1.16)/2), in a 1 to 2 scale, thus indicating a com-plex pattern of variability. The change in the fractaldimension as a function of the distance of the catena(Figure 6) showed D (Hausdorff-Besicovich dimen-sion) much closer to 2 than to 1, which indicated thatalthough a certain spatial dependence existed in thesample values, this was generally fairly weak. With the

exception of some of the samples corresponding to thestand of nodum A, all the rest had a fractal dimensionbetween 1 and 2.

Diversity analysis

The coenocline has aβ-diversity of 3138+ 0420 HC,which suggested an important complexity in the gra-dient. When the partition of the diversity of the wholesystem was evaluated it was revealed thatβ-diversitywas one of its most important properties in relation tothe species richness of the stands (Figure 7a). The re-lationship between number of species in a sample (S)and absolute diversity (H) was associated (Figure 7b)to the hypothetical geometric model (May 1975). Theabsolute diversity and evenness were higher on thesummit (nodum A,Stipa trichotoma+ Stipa tenuis-sima) and progressively declined toward the stands

90

Figure 5. Partial redundancy analysis, RDA, triplot with (1) sampling plots:#; (2) environmental variables, and (3) the most abundant speciesas arrows. Species and environmental variables abbreviations as in figure 2. Eigenvalue axis 1, 0.09; axis 2, 0.06. r species-environment; axis1, 0.93; axis 2, 0.77; 51.10% variance accounted for by two axes. Monte Carlo test significant (P<0.01). Sum of all canonical eigenvalues afterfitting covariables, 0.25.

dominated bySpartina densiflora(toeslope: nodumD). This tendency reflected the environmental dif-ferences of the catena, from edaphic sites texturallycoarser, withwt oscillation depths below or at thecritical level (summit), to the dependent ecosystems(footslope-toeslope), with significant periods of per-manent submersion and linking to the highly salinewt. The relationship between the relative abundanceof the species and the absolute diversity is shown inFigure 6c. The configuration fits into a hypotheticallognormal distribution, which marked the importanceof the species that showed intermediate abundance.

The rate of change (Figure 8a) showed that thehighest rates occurred in the backslope–footslope(nodas B–C), while it was very insignificant in thelower positions. The highest replacement rate, mea-

sured in an infinitesimal length of the gradient, oc-cured in theDistichlis scoparianodum. This couldindicate that the environmental changes associatedwith the topographical level were biologically moresignificant in the high positions than in the low posi-tions. When the length of the gradient was readjustedbetween 0 and 100 (Figure 8b) the replacement ratevaried significantly. The highest values, with respect tothe initial positions evaluated in the field, occurred inthe stands ofStipa trichotomaandDistichlis scoparia,indicating that the most significant replacements occurmore frequently in the high-intermediate topograph-ical positions than in the low-intermediate positions.GRADBETA, readjusted the positions of the relevésin such a way that the rate was almost constant,‘smoothing the differences’ and considerably displac-

91

Figure 6. Fractogram of vegetation variation.

ing the composed samples towards the more humidextreme of the coenocline. These tendencies suggestthat the growing isotropism in the soil environment,from the highest positions to the lowest positions, as-sociated with a strong linking to the highly salinewtand flooding of prolonged duration, resulted in a pro-gressively smaller environmental heterogeneity, witha consistent simplification of the structure and floristiccomposition of the vegetational systems.

When the distribution of the four dominant specieswas examined in relation to the readjusted positions ofthe gradient, a differential niche breadth was observedfor each one of them (Figure 9). Using the partialβ values, as weight factors for the different gradientpositions in the modified version (Hulbert 1978) ofthe Levins equation (1968), the niche amplitude was1.65 HC forStipa trichotoma, 0.49 HC forDistichlisscoparia, 2.04 HC forDistichlis spicataand 3.18 HCfor Spartina densiflora. This indicated thatSpartinadensiflorapossessed the widest ecological distributionfor the ecosystems studied, an aspect which was con-sistent with the observations in the field, since thestands of this species are found in environmental sit-uations which differ greatly in soil salinity (from 4-5dS m−1 in Assunta up to dS m−1 in Pacheco de Melo)and are only comparable for the position on the slopeand their time of submersion. The wide range of tol-erance of the genusSpartinahas also been cited byLongstreth & Strain (1977). On the other hand,Stipatrichotomawas restricted exclusively to the highest

topographic positions, never entering the surroundingenvironments.

Figure 10 displayed the postulated unimodal rela-tionships of the abundances (or probabilities of pres-ence) of these species related to four environmentalvariables in the RDA ordination analysis. The species’responses had a definite pattern for the variables:Spartina densifloraandStipa trichotomamarked theextremes, and invariably, for the intermediate situa-tions Distichlis spicatawas associated withSpartinadensifloraandDistichlis scopariato Stipa trichotoma.Species diversity in the ordination plane is plotted inFigure 11. The summit, associated with the maxi-mum expression ofgeomd, wtecdandhord, possessedthe highest diversity values. The lowest diversity val-ues, richness and evenness were related to the themaximum expression of the environmental variableindicators of one of the extremes of hydric-osmoticstress (wt3,wt15, soilec, suac, top, cic).

Discussion

Although discerning how many noda exist along anenvironmental gradient, or where their boundaries are,often relies on a somewhat subjective interpretation ofthe data (Austin 1985; Shipley & Keddy 1987), theformal numerical analysis used in the present work(Wildi 1989), showed an evident sociological structurethat coincided with definite combinations of environ-mental factors. If plant communities are discrete enti-

92

Figure 7. Diversity relationships of the coenocline:a) alfa diversity-beta diversity,b) species number(S)-absolute diversity (H) and,c) relativeabundance-absolute diversity.

93

Figure 8. Rates of ecological change:a) corrected rates of change (HC/gradient unit) through the composite samples in the catenas and,b) newranges of change (HC/gradient unit) along the catenas rescaled by GRADBETA. Sample positions are relative to 0–100 scale.

94

Figure 9. Distribution of the four most abundant species of the catena,Stipa trichotoma, Distichlis scoparia, Distichlis spicataandSpartinadensiflora. Sample positions heve been rescaled by GRADBETA and are expressed in half-changes along the gradient from the summit to thetoeslope.

tities, there should be a high degree of concordanceamong the ranges of species in a particular nodum(Auerbach & Shmida 1993). Along an environmen-tal gradient, end-points of species ranges should beaggregated at noda borders (Whittaker 1975). In ourstudy, at the 300-m scale, there are discrete transitionsamong summit, backslope, footslope and toeslope,with abrupt floristic transitions; discontinuities thatcoincide with discrete noda.

The fractal dimension did not appear as a con-stant function of the scale, thus the global structureof the system studied could not be obviously synthe-sized alone by means of a unique value of D (Palmer1988). The fractal dimension closest to 1 was asso-ciated with the highest topographical positions of thecatena, which indicated that the topographical controlmay be subordinated to events of a large scale such asthe geological structure.. The value of D increased to-ward the lowest topographical positions, reaching 1.89at the 120 point, coinciding with stands in nodum Cand indicating complex processes where the variationsof local manifestation (combination of underlying en-vironmental factors) have a greater importance withrespect to those of a large scale. The form of thefractogram only revealed an important spatial depen-dence at the highest topographical positions which wasassociated with short distances of the gradient. Theorganizational pattern of the groups belonging to noda

B, C and D seem more likely to be related to the com-binations of hidden environmental factors than to adefined position in the catena. The fractal dimensionfor the different positions along the catena reached val-ues close to 2, thus indicating the spatial independenceof the dependent variable. These high values of D aswell as the narrow range of spatial dependence con-firm that it is not possible to establish a relationshipbetween the floristic variation of the vegetation onlyby considering the position which the relevés occupywithin the landscape.

The floristic groups reflect the changes in the dif-ferent combinations of the main environmental factorsand they show a functional coherence to the phytosoci-ological criteria with which they were established. Themain operational factors associated with the floristicdifferences are: (i) the variations in the chemical com-position and seasonal dynamics of the soil solution inthe aerated layer of soils, (ii) salinity and dynamics ofgroundwater, and (iii) length of soil flooding duringthe rainy season (summer). Both the upper and lowerboundary of plant species along a catena may also de-pend on their capacity to tolerate abiotic stresses andto compete with the neighbouring species. Since theshallow wt and fine particles in middle and downs-lope favor the capillary rise of soil solution, mostrhizospheric zones received a good supply of watereven during winter, despite the strong tendency to rise

95

Figure 10. Generalized linear modeling probabilities of occurrences of most abundant species,Stipa trichotoma, Distichlis scoparia, DistichlisspicataandSpartina densiflorarelated to selected environmental variables: a) electrical conductivity weighted by soil profile, b) water tableelectrical conductivity/water table depth ratio, c) summation of anion+ cations of the watertable and d) water table depth. See text for theestimation of the responses surfaces.

(by evaporation and evapotranspiration). The hydricconstraint on these halophytes may be more related tothe osmotic potential of soil solution than to capillarypotential. The differences in the salinity observed inthe upper centimeters of soil during winter are notlarge enough for B, C and D noda to explain theirspatial distribution, but the important differences inwtsalinity and the use of thiswt by their deep root sys-tems, associated with a differential summer floodingresistance can provide a better understanding of thepattern.

Possibly, different adaptive responses exist in thefloristic groups for the above mentioned interactingcomplex of environmental factors, which act as reg-ulating mechanisms for the abundance of species andas organizers for the groups and for the coenocline.Effectively, the elevation gradient and the differentsoils along the catena together with the redistribution

of the water and solutes interact as a whole, generat-ing a complex spatial-temporal gradient of availabil-ity/excess of water. Although we did not either do atest pattern of species boundaries or an experimentaltest of causal mechanism, the exploratory multivariatedata analysis (James & McCulloch 1990) permits usto suggest causes and formulate an empirical descrip-tive model of vegetation organization. The iterativeand historic effect of horizontal and vertical hydro-geochemical flows, along with the topographic charac-teristics, the different original soil materials, physicalsurface conditions and vegetation cover have defineddifferent functional sectors in the catenas. At the sametime, the exposure of these habitats to those inter-actions, following circulatory flows, have led to theexpression of the current vectorial structure (Solntsiev1974).

96

Figure 11. Kriging plots: general trends of diversity, richness and evenness in the RDA ordination plane.

97

Our study of compositional changes found a clearevidence of symmetric aggregation of vegetation bor-ders on the three catenas. The existence of vegetationdiscontinuities depended largely on the water tabledepth and salinity which, simultaneously, control theedaphic salinity. Upslope and downslope boundariescould be determined by these dominant combina-tions of factors and consequently, show similar spatialpatterns. These patterns suggest that physiologicalconstraints should predominate along the gradient, es-pecially in backslope, footslope and toeslope, whilecompetition may play a more important role at thesummit. In salt marshes, Pielou & Routledge (1977)found aggregation of upslope boundaries but notdownslope ones, and concluded that competition wasthe primary determining factor of upslope borders andsalinity stress was the principal cause of downslopetermini. Butler et al. (1986) found that landscape po-sition was correlated with vegetation composition andsoil characteristics and that some plant species wererestricted to specific landscape components. Brother-son (1987) stated that the distribution of the speciesalong the slope gradient cannot easily be accountedfor by one or two variables, but that composition, soilmoisture, soil chemistry and texture, soil minerals andvertebrate and invertebrate relations all play a role.Ungar et al. (1969) presented a theory of tolerance toextreme environments. Vascular plant taxa growing atextreme ends of ecological gradients are often capableof growing under intermediate conditions, but are lim-ited to extreme environments due to their inability forcompeting with other species at moderate sites.

We could assume that the studied halophytes allthrough their evolution have developed an especialability for surviving high salinities and they may havesurvived due to this capacity. However, this markedspecialization they achieved have enabled them tocompete with their upslope competitors. At the sum-mit, the chief factors preventing halophytic speciesfrom invading this area are a very low soil moisturepercentage and the inability of halophytes for compet-ing with grassland species under these conditions. Yet,competition could also be playing an additional role indetermining species abundance here. The distinguish-ing characteristic of halophytes may not be their needfor high salinity, but their ability to endure it (Ungar1967). A general trade-off between salt tolerance andcompetitive ability have been recognized (Brewer &Grace 1990).

Redman (1972) has pointed out that distributionof species in saline areas depends entirely on the

steepness of the environmental gradient. Ungar (1974)reported that competitive exclusion occurred along asteep environmental gradient and was characterizedby dominance. However, dominance was found to beless important and coexistence of species was possi-ble when more gradual changes occured, indicatinga strong environmental selection at the extreme sitesand a marked competition for the dominant species.Scholten et al. (1987) confirmed the idea that thegrowth and distribution of plant species in salt marshesis determined by competitive interactions particularlyamong the young emerging sprouts and seedlings dur-ing springtime. This is evident with the sharp changein the distributions ofDistichlis spicata-Spartina den-sifloraand the very low representation of other speciesat the wet, highly saline footslope-toeslope.

The three catenas share the same identity in theirgeoforms and differ partly in the typology of their as-sociatedwt, however, the floristic gradient was similarfor the three situations. The most important environ-mental differences are related to the linking of thetoeslopes to highly mineralizedwt in Pacheco de Melo(T2) and La Chanchera (T1) and weakly saline inAssunta (T3). The submersion of the soils for pro-longed periods and the anoxia are both characteristicswhich functionally link the three situations, thus hid-ing the differences in the total salinity that exist amongthe sites. The zonation pattern is also associated witha distribution of the life forms: at the summit themesothermic caespitose hemicryptophytes and at therest of the positions the meso-megathermic rhizoma-tous geophytes. The recurrent summer flooding isthe most important environmental filter on the toes-lope, associated with morphological and anatomicalchanges in plants as a response to soil waterlogging.On the footslope and backslope the complex interac-tion between salinity associated with the recharges ofthe wt in springtime could have regulated life formsbetter adapted to the efficient use of water and toler-ance for the metabolic effect of the specific ions. Inthe site of greatest stability, the summit, it is probablethat coarse grain factors, like the climate, have actedprimarily in the selection of the groups of taxa.

Some authors (Ungar 1972; Chapman 1972, 1974)assert that evident zonation in salt marsh indicates thedynamic successional stages which may occur with adecrease in soil moisture and salinity concentrations.Adam (1990), Roozen & Westhoff (1985) and deLeeuw (1992) remarked that rapid succession of saltmarsh vegetation has been recorded, but in most stud-ies, community boundaries remained relatively stable

98

for decades. It is not possible to link in the samefeedback loop, functional groups which are so differ-ent, whether it would be in cyclical or unidirectionalsequences.

There is usually a sharp decrease in species di-versity even if low increments of soil salinity occur.Then diversity drops to a point where only one ortwo species remain because of their tolerance to theextreme salinity (Ungar 1972; Russell et al. 1985).Here, there were consistent relationships among thediversity of the species, their relative abundance, andheterogeneity of the physical space of the catenas.The dominant species have different ecological ampli-tudes which overlap along the length of the gradient.The absolute diversity increases as the species abun-dances becomes more similar and the greatest valuesof evenness occur in those situations. The general re-duction of number of species and absolute diversityin the noda, from the highest topographical positionsto the lowest probably reflect the stress imposed bythe conditions of salinization and anoxia in the associ-ated soils. The low diversity of J and L noda may bedue to the extremely high osmotic stress developing inthese sites, eliminating all but the most salt tolerantspecies. The value found inβ-diversity was associ-ated with the abrupt changes in the physical structureof the catena for a reduced spatial dimension. Thetopographical position and the mineralization of theprofile, governed almost exclusively by thewt, wereshown to be the most important factors associated withβ-diversity. The coenocline is related to the proper-ties of its physical structure. It is possible that thesevectors, extrinsic to the community, play a prepon-derant role in the dimensions of the niches and inthe organization of the floristic groups, with differenttolerances, in complexes of higher hierarchical order(noda). Probably, such extrinsic factors hid others ofthat category, like grazing pressure and also intrinsicfactors of the community, like inter or intraspecíficcompetition.

The catenas are highly structured spatially, andfloristic compositions are inextricably linked to thisstructure. Habitat complexity may affect associatedvegetation directly by regulating the hydrohalomor-phic conditions in the aerated layer of the soils. Therelationships of the habitat-floristic groups are notsimple: hydromorphism interacts in a complex waywith halomorphism; different life forms such as geo-phytes, hemicriptophytes and therophytes may be sen-sitive to different components or scales of physicalstructure; vegetation may alter the physical habitat

structure; and finally, plants may be affected primar-ily by other organisms and not by the physical habitatstructure directly.

References

Acosta, A., Díaz, S. & Cabido, M. 1991. Patch structure in nat-ural grasslands of Córdoba mountains (Argentine) in relation todifferent rock substrates. Coenoses 6(1): 21–27.

Adam, P. 1990. Saltmarsh ecology. Cambridge University Press,Cambridge. Auerbach, M. and Schmida, A. 1993. Vegetationchange along an altitudinal gradient on Mt Hermon, Israel-noevidence for discrete communities. J. Ecol. 81: 25–33.

Austin, M. P. 1985. Continuum concept, ordination methods, andniche theory. Ann. Rev. Ecol. Syst. 16: 39–61.

Black, C. A. 1965. Methods of soil analysis. Part I and II. Agronomy9, American Society of Agronomy, USA.

Bresler, E., McNeal, B. L. & Carter, D. L.1982. Saline and SodicSoils. Principles-Dynamics-Modelling, Springer-Verlag, Berlin.

Brewer, S. J. & Grace, J. B. 1990. Plant community structure in anoligohaline tidal marsh. Vegetatio 90: 93–107.

Brotherson, J. D. 1987. Plant community zonation in response tosoil gradients in a saline meadow near Utah Lake, Utah County,Utah. Great Basin Nat. 47(2): 322–333.

Burroughs, P. A.1983a. Multiscale sources of spatial variation insoil. I. Application of fractal concepts to nested levels of soilvariation. J. Soil Sci. 34: 577–597.

Burrough, P. A.1983b. Multiscales sources of spatial variation insoil. II. A non-Brownian fractal model and its application to soilsurvey. J. Soil Sci. 34: 599–620.

Butler, J., Goetz, H. & Richardson, J. 1986. Vegetation andsoil-landscape relationships in the North Dakota badlands. Am.Midland Nat. 116(2): 372–385.

Cantero, J. J. & Bianco, C. 1986. Las plantas vasculares del sur-oeste de la Provincia de Córdoba. III. Catálogo preliminar de lasespecies. Rev. Universidad Nacional Río Cuarto 6: 5–52.

Cantero, J. J. 1993. Vegetation and its relationships with envi-ronmental factors in hydrohalomorphic landscapes. MagisterScientiae Thesis, Universidad Nacional de Buenos Aires.

Cisneros, J. M. 1994. Characterization of hydrohalomorphism intypical environments in the centre-south area of Córdoba, Ar-gentina. Magister Scientiae Thesis, Universidad Nacional deBuenos Aires.

Cisneros, J. M., Cantero, J. J. & A. Cantero. 1996. Effects ofvegetation on the hydrophysical properties of central Argentinesaline-sodic soils. Catena (submitted).

Chapman, V. J. 1960. Salt marshes and salt deserts of the world.Leonard Hill, London.

Chapman, V. J. 1974. Salt marshes and salt desert of the world.pp: 3–19. In: Reinold, R. J. & Queen W. H. (eds), Ecology ofHalophytes. Academic Press, London.

Chapman, V. J. 1974. Introduction. pp. 1–29. In: Chapman, V. J.(ed.), Wet coastal ecosystems. Elsevier, Amsterdam.

Campbell, E. 1985. Soil physics with basic. Transport models forsoil-plant Systems. Elsevier, The Netherlands.

de Leeuw, J. 1992. Dynamics of salt marsh vegetation. ITC Publi-cation No. 13. The Netherlands.

Downing, J. A. 1991. The effect of habitat structure on the spa-tial distribution of freshwater invetevrate populations. pp. 69–86.In: Bell, S., McCoy, E. D. & H. R. Mushinsky (eds), HabitatStructure. Chapman & Hall, London.

99

Etchevehere, P. 1976. Normas de Reconocimiento de Suelos. INTA,152.

Flowers, T. J. 1975. Halophytes. pp. 309–334. In: Baker, D. A.& Hall, J. L. (eds), Ion transport in Cells and Tissues. NorthHolland Publushing Co., Amsterdam.

Hall, G. F. 1983. Pedololgy and Geomorphology. pp. 117–140. In:Wilding, L. P., Smeck, N. E. & Hall, G. F. (eds), Pedogenesisand soil taxonomy. I. Concepts and Interactions. Elsevier, TheNetherlands.

Hurlbert, S. H. 1978. The measurement of niche overlap and somerelatives. Ecology 59: 67–77. INTA- Instituto Nacional de Tec-nologia Agropecuaria. 1990. Atlas de Suelos de la RepublicaArgentina. Provincia de Cordoba. Escala 1:500.000.

James, F. C. & McCulloch, C. E. 1990. Multivariate Analysis inEcology and Systematics: Panacea or Pandora’s Box. Ann. Rev.Ecol. Syst. 21: 129–166.

Jongman, R. H. G., ter Braak, C. J. F. & van Tongeren, O. F. R.1987. Data analysis in community and landscape ecology. Pudoc,Wageningen.

Journel, A. G. & C. Huijbregts. 1988. Mining geostatistics. Aca-demic Press, London. Keddy, P.A. 1983. Shoreline vegetationin Axe Lake, Ontario: effects of exposure on zonation patterns.Ecology 64: 331–344.

Koslovski, F. 1972. Structural-functional and matematical model ofmigrational landscape geochemical process. Soviet Soil Sci. 4(2): 228–243.

Legendre, P. & Fortin, M. J. 1989. Spatial pattern and ecologicalanalysis. Vegetatio 80: 107–138. Levins, R. 1968. Evolution inchanging environments. Princenton Monographs in populationbiology, No. 2.

Li, H. & Reynolds, J. F. 1995. On definition and quantification ofheterogeneity. Oikos 73 (2): 280–284.

Linthurst, R. A. & Seneca, E. D. 1981. Aeration, nitrogen andsalinity as determinants ofSpartina alternifloragrowth response.Estuaries 4: 53–63.

Longstreth, D. J. & Strain, B. R. 1977. Effects of salinity andillumination on photosynthesis and water balance ofSpartinadensifloraLois. Oecologia 31: 191–199.

Mandelbrot, B. B. 1982. The fractal geometry of nature. Freeman,San Francisco.

May, R. M. 1975. Patterns of species abundance and diversity.pp. 81–120. In: Cody, M. L. & Diamond, J. M. (eds), Ecol-ogy and evolution of communities. Harvard University Press,Cambridge.

Orlóci, L. & Stanek, W. 1979. Vegetation survey of the AlaskaHighway, Yukon Territory: types and gradients. Vegetatio 41:1–56.

Palmer, M. W. 1988. Fractal geometry: a tool for describing spatialpatterns of plant communities. Vegetatio 75: 91–102.

Phillips, J. D. 1985. Measuring complexity of environmental gradi-ents. Vegetatio 64: 95–102.

Pielou, E. C. & Routledge, R. D. 1976. Salt marsh vegetation: lati-tudinal gradients in the zontion patterns. Oecologia 24: 311–321.

Pizarro, F. 1978. Drenaje agrícola y recuperación de suelos salinos.Editorial Agrícola Española.

Pomeroy, L. R. & Wiegert, R. G. 1981. The ecology of a salt marsh.Springer-Verlag, New York. Ranwell, S. 1972. Ecology of thesalt marshes and salt dunes. Chapman and Hall, London.

Redman, R. E. 1975. Production ecology of grasslands communitiesin western North dakota. Ecol. Monog. 45: 83–106.

Rey Benayas, J. M. & Scheiner, S. M. 1993. Diversity patterns ofwet meadow along geochemical gradients in central Spain. J.Veg. Sci. 4: 103–108.

Richards, L. A. 1973. Diagnóstico y Rehabilitación de SuelosSalinos y Sódicos. Ed. Limusa.

Robertson, G. P. 1987. Geostatistics in ecology: interpolating withknow variance. Ecology 68: 744–748.

Roozen , A. J. M. and Westhoff, V. 1985. A study of long-term salt-marsh succession using permanent plots. Vegetatio 61: 23–32.

Ruhe, R. 1975. Geomorphology. Houghton Mifflin Co., New YorkRussell, P. J., Flowers, F. J. & Hutchings, M. J. 1985. Comparison

of niche breadths and overlaps of halophytes on salt marshes ofdiffering diversity. Vegetatio 61: 171–178.

Shipley, B. and Keddy, P. A. 1987. The individualistic andcommunity-unit concepts as falsifiable hypotheses. Vegetatio 69:47–55.

Scholten, M., Blaauw, P., Stroetenga, M., & Rozema, J. 1987. Theimpact of competitive interactions on the growth and distributionof plant species in salt marshes. pp. 270–281. In: Huiskes, A. H.L., Blom, C. W. P. M. & Rozema, J. (eds), Vegetation betweenland and sea. Dr W. Junk Publishers, Dordrecht.

Smilauer, P. 1992. CanoDraw. User’s guide . v 3.0. MicrocomputerPower. Ithaca, USA.

Solntsiev, N. A. 1974. O niekotorykn fundamentalnycn svoistakhgheosistemnoi strukturv. Metody kompleksnykh issliedovaniigheosistem. Akademiya Nauk SSSR, Irkutsk.

Sýkora, K. V., Van Katwijk, M. & Meier, R. 1987. Synecologicalrelations in the moist grasslands of Ballyteige Innish, Ireland.Dr. W. Junk Publishers, The Netherlands.

ter Braak, C. J. F. 1987. The analysis of vegetation-environment re-lationships by canonical correspondence analysis. Vegetatio 69:69–77.

ter Braak, C. J. F. 1990. CANOCO a FORTRAN Pro-gram for Canonical Community Ordination by (Partial) (De-trended)(Canonical) Correspondence analysis, Principal compo-nent analysis and redundancy analysis (version 3.10.) Agricul-ture Mathematics Group, Wageningen.

ter Braak, C. J. F. & Prentice, I. C. 1988. A theory of gradientanalysis. Adv. Ecol. Res. 18: 271–317.

Ungar, I. A. 1967. Vegetation-soil relationships on saline on salinesoils in northern Kansas. Am. Midland Nat. 78: 98–120.

Ungar, I. A. 1972. The vegetation of inland salines marshes of northAmerica, north of México. On Grundfragen und methoden in derpflagensoziologie. Dr. Junk Publishers, Den Haag.

Ungar, I. A. 1974. Halophyte communities of Park County, Col-orado. Bull. Torrey Bot. Club 101: 145–152.

Ungar, I. A., Hogan, W. & McClelland, M. 1969. Plant communitiesof saline soils at Lincoln, Nebraska. Am. Midland Nat. 82: 564–577.

USDA (United States Departament of Agriculture. 1975. Soil Tax-onomy: A basic system of soil classification for making and in-terpreting soil surveys. – Soil Conservation Service. AgriculturalHandbook 436. U.S. Government Printing Office, Washington,D.C., USA.

Waisel, Y. 1972. Biology of halophytes. Academic Press, New York.Whittaker, R. H. 1975. Communities and ecosystems. MacMillan,

New York.Wildi, O. 1989. A new numerical solution to traditional phytosoci-

ological tabular classification. Vegetatio 81: 95–106.Wildi, O, & Orlóci, L. 1990. Numerical exploration of community

patterns. SPB. Academic Publishing, The Hague.Wilson, M. V. & Mohler, C. L. 1983. Measuring compositional

change along gradients.Vegetatio 54: 129–141.Wilson, M. V. & Mohler C. L. 1986. Gradbeta – a fortran program

for measuring compositional change along gradients. CornellUniversity, New York.

100

Williamson, M. H. & Lawton, J. H. 1991. Fractal geometry ofecological habitats. pp. 69–86. In: Bell, S., McCoy, E. D. &Mushinsky, H. R. (eds), Habitat Structure. Chapman & Hall,London.

Wosten, J. H. M. & van Genutchen, M. Th. 1988. Using textureand other soil properties to predict the unsatured soil hydraulicfunctions. Soil Sci. Soc. Am. J. 52: 1762–1770.