Levantamentos Fitossociológicos: Ferramentas para a ...Levantamentos Fitossociológicos: Ferramentas para a Ciência das Plantas Daninhas? CONCENÇO, G. 2, TOMAZI, M. 3, CORREIA,

Planta Daninha, Viçosa-MG, v. 31, n. 2, p. 469-482, 2013

469Phytosociological surveys: tools for weed science?

1 Recebido para publicação em 22.11.2012 e aprovado em 9.3.2013.2 Agronomist, D.Sc., Weed Science researcher at Embrapa Western Agriculture, Dourados, MS, Brazil, <[email protected]>;3 Agronomist, D.Sc., Climate Change researcher at Embrapa Western Agriculture, Dourados, MS, Brazil, <[email protected]>;4 Undergraduate student in Agronomy, University Anhanguera, trainee in Weed Science at Embrapa Western Agriculture, Dourados,MS, Brazil, <[email protected]>; 5 Undergraduate student in Biology, University Unigran, trainee in Weed Science at EmbrapaWestern Agriculture, Dourados, MS, Brazil, <[email protected]>; 6 Agronomist, D.Sc., Weed Science professor at theUniversidade Federal da Fronteira Sul, Erechim, RS, Brazil, <[email protected]>.

PHYTOSOCIOLOGICAL SURVEYS: TOOLS FOR WEED SCIENCE?1

Levantamentos Fitossociológicos: Ferramentas para a Ciência das Plantas Daninhas?

CONCENÇO, G.2, TOMAZI, M.3, CORREIA, I.V.T.4, SANTOS, S.A.5, and GALON, L.6

ABSTRACT - In simple terms, a phytosociological survey is a group of ecological evaluationmethods whose aim is to provide a comprehensive overview of both the composition anddistribution of plant species in a given plant community. To understand the applicability ofphytosociological surveys for weed science, as well as their validity, their ecological basisshould be understood and the most suitable ones need to be chosen, because cultivatedfields present a relatively distinct group of selecting factors when compared to natural plantcommunities. For weed science, the following sequence of steps is proposed as the mostsuitable: (1) overall infestation; (2) phytosociological tables/graphs; (3) intra-characterizationby diversity; (4) inter-characterization and grouping by cluster analysis. A summary of methodsis established in order to assist Weed Science researchers through their steps into therealm of phytosociology.

Keywords: weed community; density; frequency; dominance; diversity; similarity.

RESUMO - Levantamento fitossociológico, em termos simples, é um grupo de métodos de avaliaçãoecológica com o objetivo de fornecer uma visão compreensiva tanto da composição como da distribuiçãode espécies vegetais em uma certa comunidade. Para compreender a aplicabilidade desseslevantamentos para a ciência das plantas daninhas, bem como sua validade, devem-se escolher osmétodos mais adequados e com base ecológica, uma vez que áreas cultivadas apresentam um gruporelativamente distinto de fatores de seleção, em comparação com os ambientes naturais. Para estudosfitossociológicos de plantas daninhas, a seguinte sequência de passos é proposta como a maisadequada: (1) infestação geral; (2) tabelas ou gráficos fitossociológicos; (3) intracaracterização pordiversidade; e (4) intercaracterização e agrupamento por similaridade. Um apanhado dos métodos éapresentado, visando apoiar pesquisadores e estudantes da área de Plantas Daninhas em seuspassos no reino da fitossociologia.

Palavras-chave: comunidade infestante; densidade; frequência; dominância; diversidade; similaridade.

INTRODUCTION

A phytosociological survey, in simpleterms, is a group of ecological evaluationmethods whose aim is to provide acomprehensive overview of both thecomposition and distribution of plant speciesin a given plant community. These methodswere originally developed for describing

relatively stable and solid plant communities,such as forests and prairies, with little to nohuman intervention (Pandeya et al., 1968),but they are widely used in other areas ofknowledge.

In recent years, this group of methods hasbeen vastly applied in studies of agriculturalsystems and arable fields (Adegas et al., 2010;

CONCENÇO, G. et al.


470

Guglieri-Caporal et al., 2010; Fialho et al.,2011), actually assuming an important role forweed science. The term “phytosociology”,however, is directly associated with the“structure of an association of plant species”.Associations among plant species, althoughtrue in nature, are controversial in someaspects because they depend greatly on theeffect of biotic and abiotic factors which act onthe community (Greig-Smith, 1980). Thus, agiven association may be valid only undercertain conditions.

To understand the applicability ofphytosociological surveys for weed science,their ecological basis need to be understoodand the most suitable ones have to be chosen,because it is considered that arable fields havea relatively distinct group of selecting factorswhen compared to natural plant communities.A plowing operation or a herbicide applicationis a more powerful and instantaneousselection factor than most of the factors foundin a natural, undisturbed forest (Frenedoso-Soave, 2003; Malik et al., 2007). This reviewwill also address ecological concepts, in asmuch detail as required, to justify the aimsand methods suggested for use in weedscience. For more specific data, pleaserefer to Pandeya et al. (1968), Barbour et al.(1998), Gurevitch et al. (2009) and Stohlgren(2007).

Synecology and autecology

Ecology may be roughly divided in two sub-sections: synecology and autecology (Barbouret al., 1998). These areas have distinct aimsand one should be aware that methods suitablefor one of these groups may not be fullyapplicable to the other because there is therisk of obtaining inaccurate data, as the onlycommon point between them is EvolutionaryEcology (Figure 1).

Autecology deals with the adaptation andbehavior of individual species or populationsin relation to their environment (Barbouret al., 1998) and it encompasses seed germin-ation (including soil seed banks), reproductivecapacity of the species, behavior under distinctlight intensities, tolerance to water deficit,plant identification (herbariums) and severalother aspects (Pandeya et al., 1968).

Synecology deals with the community asa whole, including the full set of speciespresent, with all the interactions surroundingit (Barbour et al., 1998). This set is called a“basic unit of vegetation” (Pandeya et al., 1968).Within synecology, plant sociology holds thedescription and mapping of vegetation typesand communities (Barbour et al., 1998).

Two contrasting theories

Due to the nature of the Basic Science ofecology, there is a range of controversiesregarding current concepts (Gurevitch et al.,2009). The main discussion surroundingsynecological methods is focused on thenature of community. “Community” is definedas a group of populations which co-exist inspace and time, interacting with one anotherdirectly or indirectly (Gurevitch et al., 2009).

The concept of community is based on theprinciple of “Associations”, which are differentclusters of plant species, found generallytogether in sites with similar environmentalconditions. The nature of the relationshipamong species inside a cluster is thebasis for the most controversial point ofphytosociological studies.

The discrete view

The first theory regarding plant associationand the interdependence of species within the

Figure 1 - Theoretical sub-divisions of plant ecology, whichguides the nature of the methods adopted for ecologicalanalyses. Embrapa Western Agriculture, Dourados-MS,2012.



same community was proposed by Clements(1916). Such theory states that plantcommunities are very organized entitiescomposed by mutually inter-dependent species– the so called “super-organisms”. Thus, theemergence and disappearance of a given plantcommunity could be easily and preciselyestimated, because it was considered a soleorganism (Clements, 1936).

Two of Clements’s most remarkableaffirmations were the occurrence of severalnarrow connections between two or morespecies, and the cooperation among species forsurvival (Ludwig & Reynolds, 1988). Clements’stheories were widely accepted and rarelychallenged while he was alive, mainly becauseof his energetic and dominant personality(Gurevitch et al., 2009). Clements’s theorypredicted that the optimal and the amplitudeof species were expected to present distinctclusters; hence, changes in vegetation wereexpected to be abrupt.

The continuum view

In contrast to Clements’s ideas, Gleason(1926) believed that communities were aresult of interactions both among species andbetween species and the environment,combined to casual historically extremeclimatic events. Gleason defended the ideathat each species had its own tolerance togiven selection factors; thus, they answeredto environmental stresses in particular ways.

According to Gleason, within the range ofstress which a species is able to tolerate,casual events determine when a species isactually found in a given place (Gurevitch,2009). Gleason’s theories were confirmed byCurtis & McIntosh (1951). Gleason’s theorypredicted that the optimal and the amplitudeof species were expected to be independent,creating a gradient of occurrence as theenvironment (e.g. temperature, rain, soilfertility, altitude) changed.

Unifying points and limitations

Currently, both theories contribute witha share to the concept most widely acceptedamong modern plant ecologists: plantassociation exists to a certain degree; the

gradient of plant composition of clusters isdefined by the environment (or managementin arable areas); and abrupt changes areobserved in the composition of species insideclusters when abrupt selection factors areapplied (Pandeya et al., 1968; Barbour et al.,1998; Stohlgren, 2007; Gurevitch et al., 2009).For example, in frequently plowed andharrowed areas, plant species are expected todiffer greatly from those in a nearby areagrown for several years under no-tillagesystem, as found by Concenço et al. (2011).

Aims and methods

The aim of phytosociological studies forweed science is similar to that of ecologicalstudies. Weed science researchers should,however, take into account that the nature ofagricultural experiments usually implies (1)plots with much smaller size than the oneexpected for phytosociological samplings; and(2) much stronger selection factors than thoseacting in the natural environment. Moreover,selection factors are usually momentaneousas the treatments are applied, e.g. distinctcrop planting densities, row spacings or cropcanopy structure; previous residual or frequentpost-emergence herbicides applications, andsometimes the unknown use history of thearea.

In this context, the use of phytosociologicalmethods for weed science should be directlyassociated with the nature of the treatmentsapplied, considering as mandatory a commonhistory for all the area where the wholeexperiment will be installed, with nodifferential selection factors other than thosecomprised by the treatments.

In long-term field trials, phytosociologicalsurveys may be more comprehensivelyinterpreted because of the larger size of theplots and the consolidation of a “system” ineach one of the treatments. In addition, thesoil seed bank of plant species will tend tobe equalized and to reflect more reliablythe effects of management. In other words,plant communities in long-term, consolidatedtrials are usually more closely in conformitywith Gleason’s theory of gradient occurrenceof species as the selection factors arechanged.



472

The methods used in plant sociology relyon two key points: (1) sampling the areasaccurately and (2) describing the plantcommunity as clearly as possible so that thedata can reflect the real plant community.

Methods for sampling the community

Synecologists seek to understand thedegree of species interdependence withincommunities, how the distribution ofcommunities depends upon past and presentenvironmental factors (in long-term trials), andthe role played by communities in suchecosystem or agricultural system (Barbouret al., 1998).

Pandeya et al. (1968) and Barbour et al.(1998) point out several sampling methods, butconsidering the limitations imposed byexperiments in agriculture, only two of themwill be addressed in this article: the relevé andthe random quadrats methods.

Relevé

The relevé method was improved, if notdeveloped, by Josias Braun-Blanquet, a Swissecologist who helped classify much of Europe’svegetation. His sampling method is alsoreferred to as SIGMA, Braun-Blanquet, andZurich-Montpellier (Barbour et al., 1998). Boththis method and the theories involving itsapplication are controversial, because its

theoretical background is mainly associatedwith Clements’s theory; thus, it may lead toinaccurate samplings in constantly disturbedenvironments, e.g., arable fields.

The first step of the relevé method is todraw an species-area curve (Figure 2). For areliable sampling with the Relevé method,researchers should essentially consider aminimum size of sampled area from thetotal area, where all species present in thefield are represented. This means that thearea sampled needs to be calibrated to theenvironment it will represent. The researchershould start sampling a small area (0.25 m2 inFigure 2) and count the number of species foundin the single quadrat. The size of the quadratshould be progressively increased until thenumber of species has been stabilized (about8 m2 in Figure 2). When the number of speciesis stabilized, the size of the single quadratwhich should be evaluated is defined for thatgiven area.

The main limitation of the relevé(Braun-Blanquet) method is that it assumesalmost no variation all along the sampledecosystem (“superorganism” of Clements). Asa consequence, in Braun-Blanquet’s view,there is no need to sample more than one pointinside the community, since the minimumsize of the quadrat is properly calibrated. Someauthors, however, adopt the relevé while“splitting” its size in several pieces after

Figure 2 - Calibration of the relevé method – determination of the minimal area of the single quadrat to be sampled for fidelity in terms

of number of plant species. Embrapa Western Agriculture, Dourados-MS, 2012.



calibration. This will not, on the one hand,overcome the limitations of this methodand may, on the other hand, even eliminatethe representative community as well, bypositioning each piece of sampling in locationswhich differ greatly from the originaltheoretical whole quadrat.

Although Mueller-Dombois & Ellenberg(1974) defend the use of this samplingmethod, the relevé method should be avoidedin agroecosystems because of differencesregarding soil type, fertility and naturalrandom dispersion of weeds in the area. Infact, one of the main criticisms to this methodis that there will always be a differential biasbetween two different sampled quadrats withthe same area, and thus the calibration basedon number of species would not be enough toprovide precision to community description(Barbour et al., 1998). This method also makesit difficult to obtain data of frequency for thespecies (which will be addressed later). Inaddition, it is difficult to obtain statistical datasuch as standard error of the samples.

For each type of plant community, theminimum average size of areas to be sampledwith the revelé method can be found inMueller-Dombois & Ellenberg (1974), and itranges from 0.1 m2 for lichen communitiesto 50,000 m2 for tropical rain forests. Thus,this method can be very human labor-intensive depending on the type and size ofthe community to be represented by thesample.

Random quadrats

Sampling by random quadrats is widelyadopted by North-American ecologists, who arenot satisfied with the European vision of simplyunderstanding the structure of a community(Barbour et al., 1998). This method consistsin subjectively finding patterns inside thecommunity to be sampled, and sampling insuch a way not to favor a particular pattern(Pandeya et al., 1968; Barbour et al., 1998). Itmeans that for data to be as reliable aspossible, sampling should be accomplished asrandomly as possible. For arable fields, these“patterns” may consist of regions of the fieldwith distinct traits (wet soil opposite to dry soil)or weeds distribution, e.g. a region with

predominance of a given species because itis the point from where that species startedto spread in the field, or a region with apredominance of species which survived to thelast application of herbicides.

In addition to the correct identification ofthe patterns, the empirical accuracy of thismethod also relies on the number and size ofthe individual quadrats (Pandeya et al., 1968).Several considerations are made by Barbouret al. (1998) regarding these aspects, butonly the most significant ones for arablefields are highlighted: (1) quadrat shape shouldpreferably be square, with equal side lengths.This will make the sampling less likely tofollow a particular pattern (e.g. crop interrows).Rectangular forms result in higher perimeterof the quadrat thus increasing the Edge Effect;(2) the square quadrat size should be as largeas possible to dilute the Edge Effect. The EdgeEffect is associated with the mistake ofthe evaluator while deciding if plants close tothe border of the sampled area are inside oroutside the quadrat.

In addition, for agricultural ecosystems,there are some additional observations: (1)there is no need for statistical replications(experimental design) to allow the use ofphytosociological methods for evaluating weedoccurrence, once variation comes from thedifferences among quadrats (descriptivestatistics). A minimum representative areaof each treatment/community, however,should be sampled, and the community shouldbe as large as possible to dilute externalinfluences; and (2) statistical blocking ofexperiments in areas with low uniformity (e.g.half area constantly plowed and half areawith no tillage) will not give higher reliabilityfor the comparison. The areas to be chosenshould be as homogeneous as possible, whenexperimental design trials are installed toevaluate weed occurrence by phytosociologicalmethods. The only source of variation forstatistical data should be the quadrats.

Barbour et al. (1998) cite three rules ofthumb, from distinct authors, to be adoptedwhen decisions are made about the size of thequadrat. The most appropriate rule for arablefields is the one proposed by Greig-Smith(1964): the size of the quadrat should be at leasttwice as large as the average canopy spread of



474

the largest species. Considering traditionalarable fields, it is hypothesized that a quadratwith 0.5m of side length will be suitable formost of the situations.

Although the Accuracy of the samplingcannot be calculated – because it requiresknowledge of the true mean of the community,Precision (Pr) is a good indicator of theefficiency of the sampling procedure (Barbouret al., 1998). In the Random Quadrats method,Pr can be calculated with the formula:

means sampleof variance

1 = Pr

Pandeya et al. (1968) describes twomethods for sampling the area when using therandom quadrats method: “random” and“transects”. As transects are unlikely to besuitable for arable fields, only “random” willbe discussed. As regards random samplings,there are three sub-types: “even spaced”,“chance distribution”, and “zoned random”(Figure 3).

The even spaced method requires previousknowledge of the area and previous planningof the locations to be sampled, which is usuallynot a problem for arable fields, and the quadratsare equally distributed in the area (Figure 3A).Chance distribution comprises a completelyrandomized choice of the locations to besampled, thus increasing the possibility of notdetecting abundant species which are notfrequent (this issue will be addressed later),

as well as increasing the chance of leavingbig gaps of areas of unknown evenness (grayzones in Figure 3B) with no sampling.

The zoned random consists in previouslydefined sub-areas with distinct traits, andrandomly choosing the locations to be sampledinside each zone. For this method, the numberof quadrats to be sampled in each zone willdepend on the proportion of the total areait represents (Figure 3C). For example, inpastures under grazing, animals tend toconcentrate for overnight in specific locationswhere most of the feces (and seeds of someweeds) tend to concentrate. If the overnightarea represents about 15% of the total, only15% of the quadrats should be sampled insidethat area.

Methods for describing the community

After data are collected in the field,they need to be translated into easilyunderstandable tables and graphs. The mostdiverse methods are reported in the literature,and researchers are actually encouraged tofind their own ways to show their data to thescientific community. Most of the authors,however, agree that both tables and graphicsshould be used in the same manuscript (withno repeated data), avoiding excessively longtables or excessively summarized graphs.

It is suggested that researchers should usethe following sequence for data presentation:(1) explorative graphs, with % of area covered(if available), number of plants and dry mass;(2) importance components, in tables or graphs;(3) intra-population inferences; and (4) inter-population inferences.

A simple exploratory graph showing thenumber of individuals and dry massaccumulated per area (no need for speciesdistinction) in each treatment, will providereaders with a comprehensive overview of thedata to be further explored. Figure 4 shows theresults from a long-term trial with fourtreatments, provided as an example (adaptedfrom Concenço et al., 2011).

Importance components

Importance components are associatedwith plant traits which turn a given species

(A) (B) (C)

Figure 3 - Distribution of samplings for the random quadratsmethod. (A) even spaced distribution, (B) chancedistribution, and (C) zoned random distribution of quadrats.

Embrapa Western Agriculture, Dourados-MS, 2012.



into a weed inside the community. Severalsynecological parameters may be considered forthe importance of each species in the system(Pandeya et al., 1968; Barbour et al., 1998),namely: abundance, density, cover, frequency,homogeneity, dominance, sociability, vitality,periodicity, constance and fidelity. Differencesin application of these terms are observed ifthe researcher chooses either the relevé or therandom quadrats method (Barbour et al., 1998).A summarized survey about these parametersis found in Pandeya et al. (1968). These

parameters, however, were developed forecological studies of natural, undisturbedenvironments.

Although authors are relatively free todecide which parameters they are going toconsider in a particular analysis, theparameters chosen should be as independentas possible. Three of these parameters aresuggested as the most significant ones fordescribing weeds dynamics in arable fields:density, frequency and dominance. Abundanceis a rather nebulous term, but it is often usedas a synonym for density (Barbour et al., 1998).The clearest definitions for these parametersare found in Barbour et al. (1998), summarizedbelow.

Density (or for instance abundance) is thenumber of plants rooted within each quadrat.The average density per quadrat of each speciescan be extrapolated to any convenient unitarea. Frequency is the proportion of totalquadrats which contains at least one rootedindividual of a given species. A Dominantspecies of a community is the overstory specieswhich contributes the most cover or basal area(in case of large trees) to the community,compared to other overstory species. Theseparameters are graphically shown in Figure 5.According to Barbour et al. (1998), frequency inthe random quadrats method is roughlyequivalent to sociability in the relevé method,because frequency itself cannot be obtained inthe latter because of its unique sampling point.

CT = conventional tillage; NT = no-tillage; CLI = crop-livestock

integration; PP = permanent pasture. Error bars are presented

above each column. Adapted from Concenço et al. (2011).

Figure 4 - Illustration of the overall weed infestation of a long-term trial, by treatment.

Figure 5 - (A) Density or Abundance (DE), associated with the number of plants of a given species found in all quadrats; (B)

Frequency (FR), associated with the number of quadrats where a given species was found, regardless of the number ofindividuals; (C) dominance (DO), associated with the amount of space in the canopy attributed to a given species, inarable fields measured usually by dry mass accumulation. Embrapa Western Agriculture, Dourados-MS, 2012.

(A) (B) (C)



476

Mueller-Dombois & Ellenberg (1974)consider Density and Abundance as differentparameters, but in fact they are both based onthe same raw data: number of individuals. Thus,authors who adopt this reference are advisednot to use the two parameters in the samestudy, under the risk of giving more importanceto number of individuals compared to speciesdistribution and dry mass accumulation,which would imbalance the Importance Value(IV).

As for the application of managementpractices to control weed species, although notwidely accepted and still not well consolidated,we propose to plan the control of abundantspecies preferably in pre-emergence; the lessfrequent species by localized applications ormanagement practices, and the mostdominant in post-emergence, preventing themfrom accumulating mass and dominating thefield.

Abundant species are widely distributedin the area; hence the application of pre-emergence herbicides will play an importantrole in reducing their occurrence. As lessfrequent species occur in specific locations ofthe field, in many cases there should be noneed to apply the control all over the area inorder to eliminate these species. Dominantspecies, which are not frequent, may presentjust a few individuals randomly distributed inthe field; thus, it will be difficult to locate thembefore emergence. As a result, locating themin the area in early post-emergence may bethe correct time to apply control practices.

Based on the three parameters (density,frequency, dominance), the Importance Valueof each species in the community can beeasily estimated. The most important weedspecies will be those with a higher number ofindividuals (density), widely distributed in thearea (frequency), and capable of suppressingthe other species as a result of faster growthand mass accumulation (dominance). Thus,the Importance Value for each weed speciescan be obtained with the formula:

( )( ) ( ) ( )( )

3

%%%%IV

dominance+frequency+density=

In addition, none of these parametersneed to be presented in the absolute form, andit is advisable to present only the relativescores. In Table 1, provided as an example, greycolumns can be suppressed in the finalpresentation. Authors are also free to decidewhether or not they should present the entirelist of species found in a given treatment, oronly the main weed species, grouping theremaining ones as “others”. In the examplein Table 1, only the four main species arepresented.

For the IV, authors are advised to obtainthe mean of the three parameters instead ofthe simple sum, because in this way theimportance value will be associated with a“100%” of importance, e.g., if the IV of Bidens

pilosa (Table 1) is 32.9, this means that 32.9%of the overall importance for infestation isattributed to that species. This option, however,is more viable when data are presented in

Table 1 - Phytosociological parameters which comprise the Importance Value of infestation of weed species. Embrapa WesternAgriculture, Dourados-MS, 2012

Species DE DE (%) FR FR (%) DO DO (%) IV (%)

Bidens pilosa 15 20.8 8 26.7 325.8 47.2 32.9

Commelina benghalensis 25 34.7 3 10.0 125.7 18.2 21.5

Richardia brasiliensis 12 16.7 5 16.6 36.4 5.3 13.7

Conyza bonariensis 8 11.1 2 6.7 82.7 12.0 10.3

Other species 12 16.7 12 40.0 120.1 17.4 21.6

Total 72 100 30 100 690.7 100 100

NOTE: DE = density; FR = frequency; DO = dominance. This example comprises a theoretical sampling of 8 quadrats per treatment. Thus,

the maximum “frequency” to be observed for each species is 8, except for “other species”. When simplifying the table from the

“complete” to the “main species” model, authors should sum the frequencies of the species involved, as should be done with DE and DO.



tables (Table 1).The graphical representationof the same data (in stacked bar graphs) isshown in Figure 6, with the limitation that inthe graphical form, the IV will be by defaultpresented in the basis of 300 instead of 100,otherwise DE (or AB), FR and DO would have tobe each divided by three. Barbour et al. (1998)present other optional graphs (pictograms)which are suitable for representing datagraphically, mainly when a few areas and a fewweed species are being evaluated.

thus allowing inferences about a particularplant community in terms of both the numberof species found and the balancing in thenumber of individuals per species (Barbouret al., 1998). These indexes allow the intra-characterization of each area, supplyingadditional support for researchers’ inferencesabout a given community. In general, intra-population inferences do not properly receivethe deserved attention, and authors areencouraged to explore this aspect.

The most widely used diversity indexes areMargalef (α), Menhinick (Dm), Simpson (D) andmodified Shannon-Weiner (H’), in addition todensity of species itself (Gurevitch et al., 2009).Simpson’s D relates the likelihood that tworandomly selected individuals from aninfinitely large community will belong todifferent species (Simpson, 1949). The Simpsonindex considers the abundance of species inthe sample while being less sensitive tospecies richness (Simpson, 1949; Barbouret al., 1998). Giavelli et al. (1986) state that Dis less prone to errors because of factorsassociated with sampling problems, and itshould be chosen instead of H’, and α wasreported for its notable statistical imprecision.In addition, Simpson’s D gives very little weightto rare species, and it is most sensitive to thenumbers of abundant species; Shannon-Weiner’s H’, in contrast, is more sensitive torare species; this is where sampling errors maybe most pronounced (Barbour et al., 1998).

As D and H’ are theoretically distinct andaffected in different ways by rare or abundantspecies, authors are advised to use bothindexes in order to make inferences aboutthe diversity of a given plant community.There are optional formulas with differentparameters to calculate both D and H’, but theeasiest ones to use and, hence, less prone toerrors, are the ones supplied by Barbour et al.(1998):

D = 1− Σ (pi)2

H' = − Σ (pi)(log2

pi)

where pi = proportion of all individuals in the

sample which belong to species i. Thus, by

using the formulas supplied by Barbour et al.(1998), only the relative abundance (divided by

B.

pilo

sa

C.

benghale

nsis

R. bra

sili

ensis

C.

bonariensis

Oth

ers

Figure 6 - Graphical representation of survey data of AB (orDE), FR, DO and IV in relative terms. Embrapa WesternAgriculture, Dourados-MS, 2012.

After sampling is finished, the precisionof sampling (Pr) can be calculated as previouslystated. For this purpose, researchers are free

to decide whether or not they should useabundance (number of individuals per quadrat)or dominance (total dry mass per quadrat) forthat. It is not defined in the literature whichof them is the most appropriate, but authorsare advised to use abundance to represent theprecision of sampling.

Diversity indexes

There are three types of diversity: (1)diversity of differentiation, (2) diversity ofstandard and (3) diversity of inventory(Gurevitch et al., 2009). We will focus only ondiversity of inventory. A diversity index is astatistic which is intended to understand thevariety of individuals of a given population,



478

100) is necessary for obtaining D and H’. Morerecently, the Natural Log (ln) is being usedinstead of Log base 2 for obtaining H’. Althoughit makes less sense, it also makes no realdifference for the final value. Authors aresuggested to use log base 2 for H’, but ln is alsovalid. Data of relative abundance from Table 1will be used to illustrate both, adopting log

2 for

H’:

D = 1 - [(0.208)2+(0.347)2

+(0.167)2+ (0.111)2+(0.167)2] = 0.77

H’= - [(0.208*-2.26)+(0.347*-1.53)+(0.167*

-2.58)+(0.111*-3.17)+(0.167*-2.58)=2.22

As both coefficients are affected differentlyby abundant or rare species, H’ will usually bemore appropriate for areas recently submittedto big changes in management (e.g. shiftedfrom conventional tillage to sod seeding),where different species start to appear as aconsequence of the new environment. In thissituation, higher H’ would usually tend torepresent higher environmental sustainabilityof the cropping system.

Simpson’s D, in contrast, is moreappropriate for well consolidated areas whereno abrupt changes were recently implemented.For example, long-term RoundupReady®

soybean fields tend to present high infestationof Commelina benghalensis, Conyza spp. andIpomoea spp., which are also the mostimportant weeds; thus, D will more accuratelyreflect the diversity in this area as it is mostlyweighted by abundant species.

As an example of differences in theapplication of such coefficients, Table 2 showsboth D and H’ from areas submitted to distinctmanagements for 16 years (Concenço et al.,2011).

For the situation shown in Table 2,Simpson’s D showed a higher diversity for theCT area compared to NT while H’ indicatedhigher diversity for NT compared to CT.This means that the most important weeds inCT are the most abundant – probably the oneslargely selected by management. In contrast,H’ indicated that diversity may be on theincrease at the NT area because of theemergence of some new plant species notpresent at CT.

Stohlgren (2007) reports that lowproductivity (high stress) areas usually presentlow diversity, but this is also true for veryproductive sites, as a result of competitiveexclusion (a link with autecological studies);high diversity is usually observed in sites withintermediate productivity. As a consequence,long-term fertilization tends to decrease plantdiversity because it will select those specieswith higher ability to use a given fertilizer. Thishelps to explain why stressed areas usuallyincrease their diversity as they are recoveredfrom stress, thus highlighting the importanceof diversity indexes for inferences in long-termfield trials.

Another widely used coefficient is theEvenness (E’) one, based on Shannon-Weiner’sdiversity. This index reflects the degree ofdominance of species in a given community(Magurran, 2003). McManus & Pauly (1990),however, propose the use of a derivativecoefficient: Shannon-Weiner EvennessProportion, which is able to evaluate trendsof stress in a given environment overtime. This coefficient seems to be applicablefor phytosociological studies, with threeadvantages: (1) it considers both Density andDominance, creating a new link of synecologywith competition studies (autecology); (2) itallows inferences about ecological stress fromstatic data; and (3) it allows inferences aboutstressing factors in long-term trials over time– in fact, it was developed with this aim.

densityH'

dominanceH'=SEP

Table 2 - Diversity coefficients for weed occurrence in areassubmitted to distinct types of management for 16 years.Embrapa Western Agriculture, Dourados-MS, Brazil, 2011

Treatment1/ D2/ H' 2/

CT 0.64 0.43

NT 0.59 0.47

CLI 0.02 0.02

PP 0.63 0.50

1/ CT = conventional tillage; NT = no-tillage; CLI = crop-livestock

integration; PP = permanent pasture. Adapted from Concenço et al.

(2011). 2/ D = Simpson; H’ = Shannon-Weiner (based on density).



where SEP = Shannon-Weiner Evenness

Proportion; H’ dominance = Shannon-Weiner

based on biomass; H’ density = Shannon-Weiner

based on number of individuals. Authors areencouraged to read the original study (McManus& Pauly, 1990) for further information about SEP.

Multivariate analysis

Because the diversity indexes allowinferences about each given studied area withno comparison across areas, there is the needto adopt statistical methods which allowresearchers to infer which areas are similarin terms of weed infestation. This can beaccomplished in two ways: univariate ormultivariate analysis (Gurevitch et al., 2009).

Univariate analysis consists in studyingindividually each one of the traits evaluated forthe group of communities, e.g. if the dry massof weeds or the number of weed individuals perarea is equal among communities. For thispurpose, usual tests such as ANOVA and thesubsequent multiple mean comparison can beused (Thornley, 1976), and each quadrat isconsidered as a “replication” if the area understudy is a trial with no experimental design(like observation units).

Multivariate analysis, however, focuses ona group of traits evaluated for all communitieswhich, when put together, allow the estimationof the differences among communities througha complex way which yields a “distance”between each pair of communities (Barbouret al., 1998). For this purpose, the Euclideandistance is usually used (Danielsson, 1980).Authors can refer to Podani (2000) for furtherinformation about multivariate analysis inbiological systems.

Clustering by similarity

For phytosociological studies, the distancebased on a set of community characters isusually not the most suitable technique forverifying the level of resemblance of a givenpair of areas or communities. Gurevitch et al.(2009) reports that the abundance of species,the main trait used for comparing plantcommunities, is not usually a simple relationas assumed for a usual multivariate clusteranalysis. Based on this, communities in

phytosociological studies should be groupedbased on binary data, e.g. presence or absenceof each weed species in each community. Themost frequently used binary similaritycoefficients (all based on number ofindividuals) are Jaccard, Sørensen, Sørensen-Dice, simple combination, Ochioi, andasymmetric similarity (Barbour et al., 1998;Gurevitch et al., 2009). Among them, the mostaccurate for most situations is Jaccard (J):

cb+a

c

−

= J

where J = Jaccard similarity index; a = total

number of plant species in area “a”; b = total

number of plant species in area “b”; c = total

number of plant species common to areas “a”

and “b”. J may also be presented in a slightly

different way, by adding “c” instead of

subtracting it. In fact, as this index aims toattribute higher similarity to areas with thehighest numbers of plant species in common,

it is nonsense to add “c” to the denominator of

J, because it will decrease the similarity forareas which are actually more similar; thus,authors are strongly advised to adopt theformula presented by Barbour et al. (1998). Formore detailed studies, the same author alsopresents an additional formula: Jaccardweighted by cover (Jc).

These data will be used for generating thesimilarity matrix. For this purpose, first theresearcher needs to analyze the list of speciesfor each area. The following data were extractedfrom the raw tables of Concenço et al. (2012),supplied as an example (Table 3).

Table 3 - Number of plant species required for generatingthe similarity matrix. Embrapa Western Agriculture,Dourados-MS, 2012

Area # of species Area # of species

1 13 1x3 7

2 6 1x4 7

3 11 2x3 5

4 9 2x4 5

1x2 5 3x4 7

Example from raw data of Concenço et al. (2012). NOTE: area

crossings indicate number of species common to both areas.



480

Table 4 shows the similarity matrix,obtained from the data in Table 3 by using theJaccard similarity coefficient (J). Someauthors prefer to present only the data inTable 4 and not proceed to the cluster analysis.This is correct and can be done, and for theseauthors, it is advised to consider two areas as“similar” if the Jaccard coefficient betweenareas is equal to or higher than 0.25 (Mueller-Dombois & Ellenberg, 1974), or whenSørensen’s coefficient is equal to or higherthan 0.50 (Felfili & Venturoli, 2000). Authors,however, are advised to proceed and presentthe cluster analysis by similarity (Figure 7).In this case, Tables 3, 4 and 5 should not bepresented as they are intermediary steps forcluster analysis.

Most software products, however, are notable to generate a cluster analysis fromsimilarity data, and stating that areas areeither “equal” or “not different” actuallymeans distinct things in statistical terms(Thornley, 1976; Ludwig & Reynolds, 1988).The dissimilarity matrix should be generatedfrom the similarity matrix by “1-J” (Table 5).

The dissimilarity matrix should be suppliedto an appropriate statistical software productfor cluster analysis – the software must beinformed it is a dissimilarity matrix. Thestatistical environment R (R-Development,2012) is highly recommended for this task, butseveral other software products will also besuitable. Areas should be grouped by clusteranalysis considering the qualitative trait only(presence or absence of the species), accordingto the dissimilarities obtained from the inverseof Jaccard’s similarity matrix. Hierarchicalgrouping should be obtained from the distancematrix (dissimilarities) (Barbour et al., 1998)

by using the Unweighted Pair Group Method with

Arithmethic Mean (UPGMA) (Sneath & Sokal,1973). The final expected result for the clusteranalysis is shown in Figure 7.

After cluster analysis, grouping validationshould be accomplished by the copheneticcorrelation coefficient, obtained by the Pearsonlinear correlation between the copheneticmatrix and the original matrix of distances(Sokal & Rohlf, 1962). The copheneticmatrix is easily obtained under statisticalenvironments like R (R-Development, 2012).The cophenetic coefficient should be equal orabove 0.85, which indicates that the groupingproperly reflects the original data (Sokal &Rohlf, 1962). In the example given, thecophenetic coefficient was equal to 0.98 (areasand/or treatments were reliably grouped bythe cluster analysis).

Additional data are needed for a completecluster analysis in order to determine the

Table 4 - Similarity matrix based on Jaccard’s similaritycoefficient. Embrapa Western Agriculture, Dourados-MS,

2012

Areas A1 A2 A3 A4

A1 1 0.36 0.41 0.47

A2 0.36 1 0.42 0.50

A3 0.41 0.42 1 0.54

A4 0.47 0.50 0.54 1

Table 5 - Dissimilarity matrix based on Jaccard’s similaritycoefficient. Embrapa Western Agriculture, Dourados-

MS, 2012

Areas A1 A2 A3 A4

A1 0 0.64 0.59 0.53

A2 0.64 0 0.58 0.50

A3 0.59 0.58 0 0.46

A4 0.53 0.50 0.46 0

Figure 7 - Areas and/or treatments grouped by cluster analysisfor the raw data from Concenço et al. (2012), based on theUPGMA method.



threshold level (either by similarity ordissimilarity) to be used for determining thenumber of homogeneous groups. This is usuallyan empirical task to be chosen among distinctcriteria available in specialized bibliography,but authors are encouraged to define thethreshold level by the simple mean of thematrix (either similarity or dissimilarity,according to the scale at the graph). This mean,however, should not consider matchingareas (“1s” at the similarity and “0s” at thedissimilarity matrix). Thus, the proposedthreshold level for Figure 7 would be 0.45(for the similarity scale), or 0.55 (for thedissimilarity scale). As a consequence, onlyarea 1 is considered as distinct from the othersin terms of composition of infestation, at 45%similarity (Figure 7).

Phytosociological surveys are useful astools to shed light on the dynamics of weedspecies and their interactions in arable fields.The methods, however, are the most diverseas several indexes and coefficients areavailable, depending on the literature used asa reference by a given author. Basic careshould be taken, however, when sampling anddescribing the plant community as well.

For weed science, the following sequenceof steps is proposed as the most suitable fora phytosociological survey: (1) overallinfestation; (2) phytosociological tables/graphs; (3) intra-characterization by diversity;(4) inter-characterization and grouping bycluster analysis. Any other set of data or wayof presentation, however, may still be adequatedepending on the nature of the environmentthat is being studied.

The literature is definitely not clear aboutmethods for phytosociological surveys, and theauthors were not able to find all the set ofinformation in the same source. Even classicalreferences miss some important aspects ofphytosociological studies. In this review, asummary of methods was made in order to toassist Weed Science researchers throughtheir steps into the realm of phytosociology.

LITERATURE CITED

ADEGAS, F. S. et al. Levantamento fitossociológico deplantas daninhas na cultura do girassol. Planta Daninha,v. 28, n. 4, p. 705-716, 2010.

BARBOUR, M. G. et al. Terrestrial plant ecology. MenloPark: Benjamin/Cummings, 1998. 688 p.

CLEMENTS, F. E. Nature and structure of the climax.J. Ecol., v. 24, n. 2, p. 252-284, 1936.

CLEMENTS, F. E. Plant succession: an analysis of thedevelopment of vegetation. Washington, DC: Carnegie, 1916.(Publication, 242).

CONCENÇO, G. et al. Phytosociology in agricultural areassubmitted to distinct wintercropping management.Planta Daninha, v. 30, n. 2, p. 297-304, 2012.

CONCENÇO, G.; SALTON, J. C.; CECCON, G. Dinâmicade plantas infestantes em sistemas de cultivo. Dourados:Embrapa Agropecuária Oeste, 2011. 49 p. (EmbrapaAgropecuária Oeste. Documentos, 114).

CURTIS, J. T.; McINTOSH, R. P. An upland forestcontinuum in the prairie-forest border region of Wisconsin.Ecology, v. 32, n. 3, p. 476-498, 1951.

DANIELSSON, P. Euclidean distance mapping. Comput.Graphics Image Process., v. 14, n. 1, p. 227-248, 1980.

FELFILI, J. M.; VENTUROLI, F. Tópicos em análise devegetação. Comum. Técn. Flor., v. 2, n. 2, p. 34-44, 2000.

FIALHO, C. M. T. et al. Fitossociologia da comunidade deplantas daninhas na cultura da soja transgênica sob doissistemas de preparo do solo. Sci. Agr., v. 12, n. 1, p. 9-17,2011.

FRENEDOSO-SOAVE, R. C. Phytosociological studies onnatural establishment of vegetation in an unreclaimedlimestone mining. Braz. Arch. Biol. Technol., v. 46, n. 2,p. 259-269, 2003.

GIAVELLI, G.; ROSSI, O.; SARTORE, F. Comparativeevaluation of four species diversity indices related to twospecific ecological situations. Field Stud., v. 6, n. 3,p. 429-438, 1986.

GLEASON, H. A. The individualistic concept of the plantassociation. B. Torrey Bot. Club, v. 53, n. 1, p. 7-26, 1926.

GREIG-SMITH, P. The development of numerical

classification and ordination. Vegetatio, v. 42, n. 1, p. 1-9,1980.

GREIG-SMITH, P. Quantitative plant ecology. London:Butterworths, 1964. 256 p.

GUGLIERI-CAPORAL, A.; CAPORAL, F. J. M.; POTT, A.Phytosociology of sown pasture weeds under two levels of

degradation in Brazilian savanna areas, Mato Grosso do SulState, Brazil. Pesq. Agropec. Trop., v. 40, n. 3, p. 312-321,2010.



482

GUREVITCH, J.; SCHEINER, S. M.; FOX, G. A. Ecologiavegetal. Porto Alegre: Artmed, 2009. 574 p.

LUDWIG, J. A.; REYNOLDS, J. F. Statistical ecology. NewYork: J. Wiley, 1988. 337 p.

McMANUS, J. W.; PAULY, D. Measuring ecological stress:variations on a theme by R. M. Warwick. Marine Biol.,v. 106, n. 2, p. 305-308, 1990.

MAGURRAN, A. E. Measuring biological diversity.London: Blackwell Publishing, 2003. 260 p.

MALIK, N. Z.; ARSHAD, M.; MIRZA, S. N.Phytosociological attributes of different plant communities ofPir Chinasi hills of Azad Jammu and Kashmir. Intern. J.Agric. Biol., v. 9, n. 4, p. 569-574, 2007.

MUELLER-DOMBOIS, D.; ELLENBERG, H. Aims andmethods of vegetation ecology. New York: Wiley, 1974.547 p.

PANDEYA, S. C.; PURI, G. S.; SINGH, J. S. Researchmethods in plant ecology. New York: Asia PublishingHouse, 1968. 272 p.

PODANI, J. Introduction to the exploration ofmultivariate biological data. Leiden: Backhuys Publishers,2000. 407 p.

R DEVELOPMENT. R: a language and environment forstatistical computing. Vienna: R Foundation for StatisticalComputing, 2012. Available at: <http://www.R-project.org/>.Access in: 10, Aug, 2012.

SIMPSON, E. H. Measurement of diversity. Nature, v. 163,n. 4148, p. 688, 1949.

SNEATH, P. H.; SOKAL, R. R. Numerical taxonomy: theprinciples and practice of numerical classification. SanFrancisco: W. H. Freeman, 1973. 573 p.

SOKAL, R. R.; ROHLF, F. J. The comparison of dendrogramsby objective methods. Taxon, v. 11, n. 1, p. 33-40, 1962.

STOHLGREN, T. J. Measuring plant diversity: lessonsfrom the field. New York: Oxford University Press, 2007.390 p.

THORNLEY, J. H. M. Mathematical models in plantphysiology. London: Academic Press, 1976. 318 p.

Levantamentos Fitossociológicos: Ferramentas para a ...Levantamentos Fitossociológicos: Ferramentas para a Ciência das Plantas Daninhas? CONCENÇO, G. 2, TOMAZI, M. 3, CORREIA,

Documents