«Biodiversity and Conservation 8: 1281–1294, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands. Alien grasses in Brazilian ...»
Biodiversity and Conservation 8: 1281–1294, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
Alien grasses in Brazilian savannas:
a threat to the biodiversity
VÂNIA REGINA PIVELLO∗, CLÁUDIA NAGAKO SHIDA and
SÉRGIO TADEU MEIRELLES
Departamento de Ecologia, Instituto de Biociências, Universidade de São Paulo, Rua do Matão,
Travessa 14, Cidade Universitária Armando Salles de Oliveira, São Paulo, Brazil, CEP 05508-900;
∗ Author for correspondence (fax: (55) (11) 813-4151; e-mail: firstname.lastname@example.org) Received 14 September 1998; accepted in revised form 25 January 1999 Abstract. African grasses used as forage are spreading fast in cerrado (Brazilian savanna) patches, probably displacing native species. An analysis of the graminoid species abundance was performed in Cerrado Pé-de-Gigante Reserve (São Paulo State, Brazil), where their relative frequency, density, dominance and the value of importance were assessed in two cerrado forms: cerrado sensu stricto (denser) and campo cerrado (open). Thirty-six transects were determined, along which 324 0.5 m × 0.5 m herbaceous samples were taken. Ordination by CCA analysis was performed to detect gradients in the graminoid species distribution, according to shading, distance from the reserve border and aspect. Interspeciﬁc associations among the species were tested. A total of 93 species were sampled, predominantly Poaceae and Myrtaceae families. Two alien grasses were found, Melinis minutiﬂora and Brachiaria decumbens, with very high values of importance. Light availability proved to be the most important analyzed environmental factor related to graminoid distribution, strongly correlated with the abundance of M. minutiﬂora. Both alien grasses were negatively associated with most native graminoids, suggesting they exert a strong competitive pressure on the native herbaceous community. Attention must be taken to the introduction of alien species in the country.
Key words: African grasses, biological invasion, Brazilian savanna, cerrado, edge effect Introduction The Brazilian cerrados – savanna vegetation type – comprise a gradient from the grassland form (named ‘campo limpo’) to a sclerophylous woodland form (named ‘cerradão’), where basically the herbaceous layer gives place to arboreal elements, and the most apparent variation is in tree density and height. The intermediate ecotonal scrub forms are: ‘campo sujo’, ‘campo cerrado’ and ‘cerrado sensu stricto’, in an increasing density of trees. In cerradão, the canopy cover is around 30–60%; in cerrado sensu stricto, 30–40%; in campo cerrado, around 10%; in campo sujo, up to 1%, and there is no tree cover in campo limpo (Eiten 1972, 1983; Coutinho 1978, 1982a) (Figure 1). According to local conditions of soil, topography, ground water and ﬁre history, a combination of these forms may appear in a cerrado patch, as a mosaic.
Figure 1. A gradient of cerrado (Brazilian savanna) physiognomies, from the grassland (‘campo limpo’) to the woodland (‘cerradão’) form (modiﬁed from Coutinho, 1982a).
The herbaceous layer holds a great deal of Asteraceae, Fabaceae, Rubiaceae and Cyperaceae species, but the bulk of it is formed by Poaceae, mainly C4 grasses.
Among the woody elements, the dominant families are Myrtaceae, Fabaceae, Ceasalpiniaceae, Melastomataceae, Mimosaceae (Batalha 1997; Goodland and Ferri 1979;
Klink and Joly 1989; Mantovani 1983).
In the more open forms, African grasses once brought to the country as forage, have spread in such a magnitude that they are present today in practically every cerrado fragment, dominating patches of the environment and outcompeting native herbs, as already stressed by some authors (Coutinho and Hashimoto 1971; Filgueiras 1990; Klink 1996a,b; Pivello et al. 1999), and therefore, representing a threat to the natural biodiversity. As stressed by D’Antonio and Vitousek (1992), grasses are especially threatening invaders, as they can spread very easily, they are very competitive against native plants in many circumstances, most of them tolerate ﬁre and they are able to modify the environment severely.
The most common grasses invading cerrados are Melinis minutiﬂora Beauv., Brachiaria decumbens Stapf, Hyparrhenia rufa (Nees) Stapf, Andropogon gayanus Kunth and Panicum maximum Jacq., and they seem to enter cerrados through the borders, ﬁrst establishing in disturbed spots and then spreading all over. It has been observed that termite or ant mounds may function as toeholds for the spread of invader species (Coutinho 1982a,b). The intense fragmentation of cerrado environments, transforming the natural vegetation into patches surrounded by pastures and crop cultures and creating borders, favors the dissemination of invader grasses even more.
Although they have a very aggressive behaviour against native grasses, concerning spreading and establishment, the planting of such exotic grasses continues to be encouraged by agricultural agencies because of their high productivity as forage, and there is no control, concerning the bad effects they can bring. Several negative ecological effects of some introduced species on native communities are known nowadays (Williamson 1996; Cronk and Fuller 1995; Pysek et al. 1995) and it is time to establish a control scheme for their introduction and dissemination in the country. Also in this sense, Hobbs and Humphries’ proposal for management actions centered in the ecosystem, and not only focusing the invader species (Hobbs and Humphries 1995), is very pertinent and stresses the important role of landscape ecology as a working approach.
Regarding the great problem that alien invasive grasses are posing to cerrado reserves and parks in Brazil, the present study was designed to: (a) verify the level of invasion by alien graminoid species in the Cerrado Pé-de-Gigante Reserve; (b) verify changes in the density of invasive plants according to distance from the edge; (c) explore relationships between graminoid species and some physical environmental conditions; (d) investigate association patterns among alien and native graminoid species.
This study was carried out in the Cerrado Pé-de-Gigante Reserve (21◦37 30 S;
47◦ 37 30 W), which is part of the Vassununga State Park, in Santa Rita do Passa Quatro municipality, São Paulo State, Brazil. The Reserve comprises 1225 ha, in altitudes ranging from 590 to 740 m (Mesquita Jr. 1998).
Regional climate is tropical with wet summer and dry winter, corresponding to Walter and Lieth’s type II (Walter 1986); annual rainfall is around 1300 mm. The relief is gently rolling, formed by extensive and ﬂat topped hills. A big depression in the Central part of the Reserve, in the form of a big footprint, caused by ﬂuvial erosion in the sandy soils, gave it the name, since ‘Pé-de-Gigante’ means giant’s foot. Soils are predominantly acid and ferruginous, chemically poor, sandy and well drained, prevailing the Red-Yellow Latosol (Oliveira et al. 1982).
Although the Reserve is covered by assorted cerrado physiognomies, from the more open forms – dominated by grasses – to the woodland cerrado type, the regional landscape also includes semideciduous mesophyllous forest, where soil is richer.
Today, only fragments of the original vegetation remain in a silvicultural–agricultural matrix.
The present study was carried out in cerrado sensu stricto and campo cerrado patches inside the Cerrado Pé-de-Gigante Reserve, where intense invasion by alien grasses was identiﬁed.
The herbaceous layer (including plants up to 1 m tall) was sampled through
324 0.5 m×0.5 m quadrats (sample units), where every plant was identiﬁed at species level, the number of individuals of each species was counted, and the area covered by each species was estimated. Quadrats were placed along transects starting at Reserve borders delimited by dirt roads, or marginal to a grassland which follows a drainage canal inwards the Reserve, distant 0, 10, 20, 30, 40, 50, 60, 80 and 100 m from the edge (dirt road or grassland). Six groups of six transects each (as replicates), were arranged following the directions NE, N, NW, SW, W and E, corresponding respectively to groups A, B, C, D, E and F, in Figure 2. Groups A to D were placed in cerrado sensu stricto patches, while groups E and F were placed in campo cerrado patches. Inside the group, transects were distant about 100 m from each other (Figure 2). Sampling effort was proportional to the area occupied by the cerrado form in the Reserve.
As the primary concern of this study was the graminoid species, which constituted the great majority of alien species in the area, more detailed analyses were carried out for Poaceae and Cyperaceae families.
To detect gradients in the distribution of the most frequent graminoid species associated with meaningful environmental variables in both cerrado sensu stricto and campo cerrado, ordination by canonical correspondence analysis (CCA) was performed with the CANOCO package, version 3.12 (Ter Braak 1991). The analysis was done using a matrix of species abundance in the sample units with information on shading, distance from the Reserve edges, and aspect (represented in degrees, being the highest values in the South). Shading was estimated with an arbitrary scale, from 1 to 9, which considered the height of arboreal elements and the projection of their Figure 2. Experimental design adopted in this study. Six groups of six transects each are indicated at different aspects in the reserve: A, B, C, D, E and F (respective directions were: NE, N, NW, SW, W and E).
canopies on the quadrat (1 = no trees in the quadrat and no shading;... 9 = tall trees in the quadrat and maximum shading).
Only graminoid species occurring in at least 4 samples were included in the analysis, comprising one Cyperaceae and 6 Poaceae species. Brachiaria decumbens Stapf. (Poaceae) was excluded from the CCA because this species showed a very typical distribution pattern, occurring only in pure stands and very close to the road margins. In such case, the inclusion of this species would bias the analysis.
Signiﬁcance of species-environment canonical axes was assessed via unrestricted Monte-Carlo permutations test, also using the CANOCO package (Ter Braak 1991). Results are presented in a triplot diagram of species, sites and environmental variables.
A phytosociological analysis was carried out in order to analyze the behavior of graminoids and the other species, separately in cerrado sensu stricto and in campo cerrado, considering: the relative species density, relative dominance, and relative frequency, as well as the percentual value of importance, as follows (Mueller-Dombois
and Ellenberg 1974):
– DR (relative density) = 100ni /N (%), where ni = number of individuals of species i and N = number of all individuals – DoR (relative dominance) = 100ci /C (%), where ci = cover area of species i and C = cover area of all species – FR (relative frequency) = 100Oci /Oc (%), where Oci = number of occurrence of species i and Oc = number of occurrence of all species VI (value of importance) = (DR + DoR + FR)/3 (%) To explore interspeciﬁc associations among the graminoids in cerrado sensu stricto and in campo cerrado, χ 2 contingency tables (following Ludwig and Reynolds 1988) were prepared based on a binary matrix of presence–absence data. To remove the inﬂuence of the most obvious community heterogeneity, the analysis was conducted in separate subsets of campo cerrado and cerrado sensu stricto. An overall measure of community association tendencies was also obtained through the variance ratio measure. The signiﬁcance of χ 2 association tests was assessed at the 0.05 level.
A further interspeciﬁc analysis focused on relationships among species abundances, applying Spearman Rank Correlation to species cover values on sample units.
The test was performed using the SPSS package (SPSS 1996). Signiﬁcance of correlation coefﬁcients was assessed at the 0.05 level.
Results Phytosociological analysis The dominant families in cerrado sensu stricto were Poaceae, Myrtaceae, Asteraceae and Malpighiaceae and, in campo-cerrado, Poaceae and Myrtaceae. Eighty-ﬁve species were sampled in cerrado sensu stricto and 36 in campo cerrado, 28 of them being common to both cerrado forms. Considering the graminoid species (Cyperaceae and Poaceae), we sampled 15 and 11 species, respectively in cerrado sensu stricto and in campo cerrado. Two alien grasses were sampled: Melinis minutiﬂora Beauv. and Brachiaria decumbens Stapf., both of African origin (Tables 1 and 2).
In cerrado sensu stricto, the graminoid species were present in 53.69% of the samples and accounted for 55.78%, 74.05%, 53.68% and 61.18% of the relative density, relative dominance, relative frequency, and value of importance, respectively (Table 1). In campo cerrado, graminoid species were present in 77.42% of the samples, representing 78.82%, 90.30% and 77.41% of the relative density, relative dominance and relative frequency, respectively, accounting for 82.17% of the value of importance (Table 2).
Brachiaria decumbens was sampled only in cerrado sensu stricto and Melinis minutiﬂora only in campo cerrado. These alien species were very abundant where they occurred, being responsible for, respectively, the second and the ﬁrst values of importance in the communities (Tables 1 and 2). B. decumbens occurred in every sample unit at the 0 m distance and, in most samples, it covered 100% of the quadrat; in only one sample it was collected at 10 m from the road. On the other hand, M. minutiﬂora was more disseminated in the central part of the Reserve, following the drainage and tracks.
Apart from these two species and Tristachya leiostachya, the same graminoid species appeared as the most frequent in both cerrado forms, sometimes changing positions in rank value of importance. Tristachya leiostachya did occur in both forms, but it was much less important in cerrado sensu stricto.
The results of the CCA, considering both cerrado forms, are summarized in Table 3 and Figure 3. The dispersion along the ﬁrst axis reﬂects changes in species composition along a gradient, while the second and third axes reﬂect only a change in species abundance. The three environmental variables analyzed were poorly correlated to each other.