Skip to main content

Low-intensity cattle grazing is better than cattle exclusion to drive secondary savannas towards the features of native Cerrado vegetation

Cite this dataset

Durigan, Giselda (2022). Low-intensity cattle grazing is better than cattle exclusion to drive secondary savannas towards the features of native Cerrado vegetation [Dataset]. Dryad.


Although livestock have been historically associated with land conversion and biodiversity loss, well-managed cattle grazing has been reported to contribute to conservation of open ecosystems. Knowing the balance between positive and negative effects of livestock (presence or exclusion) on different ecosystems is, therefore, crucial to support management decisions. We conducted an experiment in a secondary savanna with exotic grasses, used as pasture, to assess the effect of cattle presence in low density and cattle exclusion (in paired plots) on the trajectory of these ecosystems. Richness, composition and structure of the woody community, and exotic grass cover and biomass were compared between treatments in the beginning of the experiment and after seven years. At the end of the experiment, we also compared composition, richness, and density of the native ground layer. We verified that (a) cattle exclusion accelerates the undesirable woody encroachment, changes the species composition and leads to huge grass fuel accumulation, while (b) cattle grazing/browsing hinders changes in savanna structure and composition and reduces the exotic grass cover and biomass, thus favoring native herbaceous plants. By decreasing the grass biomass, cattle grazing also reduces the system flammability and, therefore, the risk and intensity of wildfires. Together, the positive effects of cattle presence and the negative effects of cattle exclusion lead to the conclusion that cattle should be maintained in these systems. Low-intensity cattle grazing limits woody and exotic grass invasion, improves native forb biodiversity, and help maintain composition and structural features of secondary savannas of the Cerrado.


In the year 2005, we set up 10 pairs of 10 x 10 m permanent plots within a 6-ha fenced area used as pasture. Each pair comprised one open plot (grazed) and one exclusion plot (ungrazed). The ten pairs were haphazardly distributed in the area, keeping at least 50 m distance from each other. Around the exclusion plots, a barbed-wire fence was installed (12 x 12 m), keeping 1-m distance from the 10 x 10 m sampling plot inside. The paired plots were diagonally positioned, with a distance of 2 m between their closest corners to avoid excessive trampling in the grazed plot. We sampled the woody species and the exotic grasses within each plot immediately after fencing and once a year at the end of the dry season (September), up to 2012, comprising a 7-yr period of observations. In this study, we compared only data collected in 2005 (initial) and 2012 (after seven years). In each plot, we counted (stems and individuals), tagged, identified and measured all woody plants (height ≥ 50 cm), and recorded stem diameter (from 3 cm minimum) at 50 cm above ground to estimate basal area. Using the line-intercept method (Canfield, 1941), we estimated tree canopy cover and exotic grass cover percentage by sampling three 10-m parallel lines inside each plot. Biomass of the exotic grass (Urochloa decumbens) was sampled in two 50 x 50 cm randomized subplots within each 10 x 10 m plot, by clipping at 10 cm above ground. The grass biomass was dried at 70°C for 48 h and weighed. To support the interpretation of our results, using the same method for comparison, we collected 10 biomass samples of the ground layer (all plants below 50 cm in height) in uninvaded cerrado vegetation with similar canopy cover, as a reference ecosystem (at Santa Barbara Ecological Station, 22°4641′′S and 49°1610W”). In the last sampling occasion, after seven years (2012), we also sampled (counting and identifying) all native plants (herbaceous as well as young woody plants) with height < 50 cm in the whole area of each plot (100 m2), to verify the effects of each treatment on native plants of the ground layer. Native species sampled were functionally categorized as woody (shrubs, lianas and trees) or herb (graminoids and forbs). Native forbs and graminoids were virtually absent in the beginning of the experiment (see first-year pictures at Figure 1) and their spontaneous recovery was not expected. Considering the great importance of their return, we decided to include the native ground layer in the analyses, even without the initial information. In all samplings of native plants, multiple stems connected to the same underground structure were counted as a single individual. Rhizomatous and stoloniferous native plants were absent and exotic grasses were not counted, because it would be technically impossible.