Floristic , frequency , and vegetation life-form spectra of a cerrado site

We used Raunkiaer's system to classify in life-forms the vascular plants present in 12 random 25 m2 quadrats of a cerrado site. The study area is covered by cerrado sensu stricto and is located in the Valério fragment, at about 22 degrees 13'S and 47 degrees 51'W, 760 m above sea level, in the Itirapina Ecological and Experimental Station, São Paulo State, southeastern Brazil. The floristic spectrum considers the life-form of each species, while in the frequency spectrum, each species is weighted by its frequency. The vegetation spectrum does not consider the species at all, but only the individuals in each life-form class. In the floristic spectrum, the most represented life-forms were the phanerophytes and the hemicryptophytes, as in other cerrado sites. This spectrum differed significantly from Raunkiaer's normal spectrum, mainly due to under-representation of therophytes and over-representation of phanerophytes. The floristic and frequency spectra were similar, but both differed from the vegetation spectrum. We recommend the floristic spectrum when working at larger scales and a description of the phytoclimate is wanted. The vegetation spectrum is preferable when working at smaller scales and wanting a quantitative description of the physiognomy. The frequency spectrum is not recommended at all.

ABSTRACT -(Floristic, frequency, and vegetation life-form spectra of a cerrado site).We used Raunkiaer's system to classify in life-forms the vascular plants present in 12 random 25-m 2 quadrats of a cerrado site.The study area has a physiognomy of cerrado sensu stricto, and is located in the Valério fragment, at about 22º13'S and 47º51'W, 760-m altitude, in the Estação Experimental e Ecológica de Itirapina, São Paulo State.The floristic spectrum considers the lifeform of each species; in the frequency spectrum each species is weighted by its frequency; and the vegetation spectrum does not consider the species at all, but only the individuals in each life form class.In the floristic spectrum, the most represented life-forms were the phanerophytes and the hemicryptophytes, as in other cerrado sites.It differed significantly from Raunkiaer's normal spectrum, due mainly to the under-representation of therophytes and over-representation of phanerophytes.The floristic and the frequency spectra were similar, but both differed from the vegetation spectrum.We thus recommend the floristic spectrum when one is working at larger scales and wants a description of the phytoclimate; and the vegetation spectrum when one is working at smaller scales and wants a quantitative description of the physiognomy.The frequency spectrum is not recommended at all.RESUMO -(Espectros biológicos florísticos, de freqüência e vegetacional de uma área de cerrado).Usamos o sistema de Raunkiaer para classificar em formas de vida as plantas vasculares de 12 parcelas aleatórias de 25 m 2 , uma área de cerrado.A área estudada tem fisionomia de cerrado sensu stricto e localiza-se no fragmento Valério, ao redor de 22º13'S e 47º51'W e 760 m de altitude, na Estação Experimental e Ecológica de Itirapina, estado de São Paulo.O espectro florístico considera a forma de vida de cada espécie; no espectro de freqüência, cada espécie é ponderada pela sua freqüência; no espectro vegetacional apenas os indivíduos são considerados, sem qualquer referência às suas espécies.No espectro florístico, as formas de vida mais bem representadas foram os fanerófitos e os hemicriptófitos, como em outras áreas de cerrado.O espectro florístico diferiu significativamente do espectro normal de Raunkiaer, devido

Introduction
Plants can be grouped into life-form classes on the basis of their similarities in structure and function (Mueller-Dombois & Ellenberg 1974).A life-form is characterized by the adaptation of the plants to certain ecological conditions (Mera et al. 1999).Life-form study is an important part of vegetation description, ranking next to floristic composition (Cain 1950).Raunkiaer (1934) proposed a system to classify plant life-forms based on the position and degree of protection to the renewing buds, which are responsible for the renewal of the plant's aerial body when the favorable season comes.In his system, the more pronounced the unfavorable season, the more protected the renewing buds.There are, in Raunkiaer's (1934) classification, five major classes, arranged according to increased protection of the renewing buds: phanerophytes, chamaephytes, hemicryptophytes, cryptophytes, and therophytes.This life-form classification has been modified (see Mueller-Dombois & Ellenberg (1974), yet according to Begon et al. (1996), Raunkiaer's system is still the simplest and, in many ways, the most satisfactory classification of plant lifeforms.Raunkiaer (1934) proposed the "biological spectrum" to express both the life-form distribution in a flora and the phytoclimate under which the prevailing life-forms have evolved.The biological spectrum is the percent representation of the number of species belonging to each life-form in a given flora.Raunkiaer (1934) constructed a "normal spectrum", which could act as a null model against which different life-form spectra could be compared.Differences in the life-form distribution between the normal spectrum and a biological spectrum would point out which lifeform characterizes the phytoclimate and/or vegetation in study.
When working with a species list, every species has the same weight in the biological spectrum, and this is called floristic biological spectrum (Godron et al. 1969).However, when the number of individuals, instead of species, of each life-form is counted, each class can be weighted by its abundance, giving rise to the vegetation biological spectrum, which indicates the phenomena relative to the vegetation rather than to the flora (Godron et al. 1969).Also, the vegetation biological spectrum is more readily comparable to biological spectra similarly constructed for other sites.Nevertheless, Raunkiaer (1934) stated that counting all plant individuals in a survey is too problematic, because it is not always possible to distinguish what an individual is.To avoid this, Raunkiaer (1934) proposed a frequency spectrum, where the number of sampling units in which the species is present is used to weight the species.
In the Brazilian cerrados, some studies used Raunkiaer's system to classify the sampled species in life-forms.For instance, Mantovani (1983) classified a cerrado site flora in life-forms, constructed its biological spectrum, and compared it with other cerrado spectra obtained from Warming (1892) and Ratter (1980).Batalha (1997) and Batalha et al. (1997) carried out floristic surveys in two cerrado sites and also classified the species in life-form classes.In all these sites, higher proportions of hemicryptophytes and phanerophytes were found.
Our aim is to answer the following questions, considering the life-forms of vascular species found in a cerrado site in the municipality of Itirapina, SE Brazil. 1) Are the hemicryptophytes and phanerophytes the most represented life-forms also in our survey, as found in other cerrado sites (Warming 1892, Ratter 1980, Mantovani 1983, Batalha 1997, Batalha et al. 1997)?2) Is the floristic biological spectrum in the cerrado of Itirapina significantly different from Raunkiaer's normal spectrum?3) If so, which classes characterize the Itirapina life-form spectrum?4) Do the floristic, frequency, and vegetation spectra differ significantly among them and between them and the normal spectrum?

Material and methods
We carried out our study in a cerrado site, in Itirapina Municipality, São Paulo State, southeastern Brazil, approximately at 22º13'S and 47º51'W, 760-m altitude.The site is classified as cerrado sensu stricto following Coutinho's (1978) classification, a woodland sensu Sarmiento (1984).The climate is Koeppen's Cwa: macrothermic temperate with rainy summer and not severely dry winter.The area belongs to the São Paulo State Forestry Institute and is surrounded by pine plantation.
In this site, there is a permanent grid of 64 quadrats (each one with 25 m 2 ), where other researchers have developed studies on population dynamics and community structure.We randomly surveyed 12 quadrats in this grid at the middle of the rainy season (February 2001), and classified all vascular plants in life-forms, following Raunkiaer's (1934) system, adapted by Mueller-Dombois & Ellenberg (1974).
The life-form of each species was assigned with the aid of an identification key (Mueller-Dombois & Ellenberg 1974).Since we did not follow the individuals through the year, we used, in some cases, information from other studies that did so and classified the species in life-forms (Mantovani 1983, Batalha 1997, Batalha et al. 1997).We included an individual in our sample if its leaves were morphologically similar to the adult ones.For caespitose plants, an individual was defined as the very caespit; for non-caespitose plants, an isolated axis at soil level was considered as an individual.
In the construction of the floristic and frequency spectra, each species was assigned only to a single life-form class.When the species could be assigned to more than one life-form class, we always considered the one in which the renewing buds were less protected.In the vegetation spectrum, the abundance of each life-form was counted independently of the species to which the individual belonged.For example, a species was classified as "phanerophyte" in the floristic and frequency spectra, even though some of their individuals were smaller than 50 cm, provided there was at least one individual of the species that fulfilled the characteristics of a phanerophyte.On the contrary, in the vegetation spectrum, the individuals of a phanerophytic species were classified as "chamaephytes" or "phanerophytes", depending on their height.
The data regarding the life-form of each species were used to construct the floristic, frequency, and vegetation spectra (Raunkiaer 1934).The floristic spectrum was compared to Raunkiaer's normal spectrum.In this case, to allow comparison, lianas and epiphytes were included in the "phanerophyte" class, and geophyte and saprophyte, in the "cryptophyte" class.
To verify whether the floristic life-form spectrum was significantly different from the expected according to the normal spectrum, we used the chi-square test (Zar 1999).If there were significant difference, we calculated the contribution percentage of each class in the chi-square value.In this case, the higher the difference between the expected and the observed in the class, the higher was the percentage of its contribution.
In the frequency life-form spectrum, each species was weighted by the number of quadrats in which it was present, while, in the vegetation spectrum, each life-form was weighted by its number of individuals.The floristic, frequency, and vegetation spectra were compared pairwisely by homogeneity analysis (Zar 1999), to test whether the life-form proportions were the same in both spectra.In this analysis, only Raunkiaer's major classes (phanerophytes, chamaephytes, hemicryptophytes, cryptophytes, and therophytes) were considered.Since there was a low number of cryptophytes, they were included in the "hemicryptophyte" class, and the analysis was done with four classes only.
We calculated Shannon-Wiener diversity index (Shannon & Weaver 1963) for the two classes with higher proportions in the vegetation life-form spectrum and tested for differences between them with the t-test proposed by Hutcheson (1970).NÃO ENTENDI POR QUE ISSO FOI FEITO (FRM).
In the frequency spectrum, we recorded 516 occurrences and, in the vegetation spectrum, 2834 individuals.The most represented classes in the floristic spectrum, phanerophyte and hemicryptophyte, were also the most represented ones in the other two spectra, but their pattern of variation was different.The proportion of hemicryptophytes increased from the floristic to the vegetation spectrum, while the proportion of phanerophytes decreased (table 1).The diversity indices of these classes were, respectively, 3.15 ± 0.10 nats/ind for the phanerophytes, and 1.73 ± 0.01 nats/ind for the hemicryptophytes, values considered different by the t-test (t = 24.88,υ = 533, p < 0.
When compared to the normal spectrum, the class with higher contribution to the chi-square value was the therophytes in Itirapina.Although our sample could underestimate annual plants, the proportion of therophytes is also low in life-form spectra of other cerrado sites (Mantovani 1983, Batalha 1997, Batalha et al. 1997).
From the floristic to the vegetation life-form spectra, the classes with higher proportions, phanerophytes and hemicryptophytes, varied differently.While the proportion of phanerophytes decreased toward the vegetation spectrum, the proportion of hemicryptophytes increased.This could be explained by the inclusion of young individuals of phanerophytes as chamaephytes in the vegetation spectrum and by the significantly higher diversity of phanerophytes in relation to the hemicryptophytes.Since there were more individuals of few species in the hemicryptophyte class, its relative importance increased towards the vegetation spectrum.
The floristic life-form spectrum differed significantly from the vegetation spectrum, but not from the frequency one.The frequency and vegetation spectra were significantly different.According to these results, frequency count provided no additional information in relation to the species list used to construct the life-form spectrum.On the contrary, the vegetation spectrum was quite distinct from the floristic one and provided a more accurate description of the vegetation physiognomy.
Another advantage of the vegetation spectrum is that there is no need of recognizing each species, since the individuals are counted by its present life-form.The vegetation spectrum, as its name implies, shall be applied when one wants to know on the vegetation of a given site and not on its flora.Buell & Wilbur (1948) compared life-form spectra based on species lists and frequency counts in a hardwood forest site, and found a profound shift toward the less protected life-forms in the frequency spectra.They suggested frequency data as a basis for comparing life-form spectra of similar communities in different regions.However, our results indicated that this procedure would be not valid for cerrado sites.Qadir & Omar (1986) studied four Lybian communities and compared the qualitative spectrum, based solely on floristics, and the quantitative one, based on importance value.When the life-form spectra were expressed on quantitative basis, each community appeared quite distinct from each other (Qadir & Shetvy 1986).The index they used is probably a better descriptor of the vegetation physiognomy than the count of individuals we used, but using their index poses much more difficulties on collecting data in the field.Cain (1950) suggested that some measure of the relative dominance of each species in the community would provide the most significant data.We recommend the number of individuals as a measure of life-form abundance in the vegetation, since this procedure is relatively easy in the field and avoids the need of taxonomic identification of the species.
Frequency as a descriptor was proposed by Raunkiaer (1934) himself to avoid the problems of recognizing plant individuals in the field and counting all individuals of each species.Frequency is considered a measure of abundance, but it is influenced by the spatial distribution of the individuals (Mueller-Dombois & Ellenberg 1974) and has a logarithmic, non-linear relationship with density (Greig-Smith 1983).Density may be obtained from frequency counts only if the spatial distribution of individuals is random, which is rarely found for plant species (Greig-Smith 1983).According to Greig-Smith (1983), frequency, instead of giving a measure of the bulk of material contributed by each species, is an uncertain assessment of several different characteristics and thus its principal value lies simply in the rapidity with which it is obtained.Sarmiento & Monasterio (1983) criticized the applicability of Raunkiaer's system to tropical communities, because it classifies life-forms on the supposition that the limiting factor for plant growth is low winter temperatures, which obviously is not an important ecological factor in such communities.However, in the cerrado vegetation other factors can be analogous to low winter temperatures, as, for example, winter droughts and periodic burning.The application of Raunkiaer's system, in this case, would be possible and recommended if one intends to investigate which factors, besides climate, define the vegetation physiognomy in question.

Conclusions
When a biological spectrum is constructed with data collected in a single period, all life-forms with renewing buds not exposed to the air, i.e., hemicryptophytes, cryptophytes, and therophytes, are underestimated.Hence, care must be taken when comparing spectra constructed at different seasons.Nevertheless, the life-form spectra of the cerrado vegetation so far constructed seem to be quite consistent, with phanerophytes and hemicryptophytes being always the most represented classes.
Though restricted to a small cerrado site, our study did not support Raunkiaer's idea of the frequency spectrum as a good descriptor of the life-forms distribution in a certain plant community.
The floristic life-form spectrum is recommended if one is working in sufficiently large areas, when it could give an indication of the prevailing phytoclimate.We recommend the vegetation spectrum, when one is working in smaller areas, and a more detailed description of the physiognomy is desired or local ecological factors are being studied.The frequency spectrum is not recommended at all, since it was not significantly different from the floristic one and frequency is not a good estimator of abundance.
Aknowledgements -We are greatly indebted to São Paulo Forestry Institute, for research permission, and to Flavio Antônio Maës dos Santos, for useful suggestions.