Water infiltration in the Brazilian tropical savanna: the case of Cerrado típico

Keyworks natural vegetation soil hydrology ecohydrology hydrological processes plants The Brazilian savanna, regionally known as Cerrado is an important water source due to its spatial position in the highlands known as Planalto Central of Brazil. Yet, the hydrological functioning of such complex ecosystems and its many variations is poorly understood. In this paper, we characterized the topsoil permeability to water under in one of the most common vegetation types: the Cerrado típico which is a type of savanna. In doing so, we measured infiltration capacity using the Mini-Disk infiltrometer (n = 15) and soil resistance to penetration using the Stolf penetrometer (n = 7) in a Latossolo (oxisol) under Cerrado típico located in Brasília (Planaltina), Distrito Federal, Brazil. Our results indicated a high infiltration capacity under Cerrado típico which is potentially associated with a low penetration resistance. This fact might highlight the importance of the natural undisturbed ecosystems, in this case Cerrado típico, as a land-cover which favors hydrological processes which, in turn, promotes soil-groundwater recharge that are inportant to maintain water bodies and reservoirs in the region.

The Brazilian savanna, regionally known as Cerrado is an important water source due to its spatial position in the highlands known as Planalto Central of Brazil. Yet, the hydrological functioning of such complex ecosystems and its many variations is poorly understood. In this paper, we characterized the topsoil permeability to water under in one of the most common vegetation types: the Cerrado típico which is a type of savanna. In doing so, we measured infiltration capacity using the Mini-Disk infiltrometer (n = 15) and soil resistance to penetration using the Stolf penetrometer (n = 7) in a Latossolo (oxisol) under Cerrado típico located in Brasília (Planaltina), Distrito Federal, Brazil. Our results indicated a high infiltration capacity under Cerrado típico which is potentially associated with a low penetration resistance. This fact might highlight the importance of the natural undisturbed ecosystems, in this case Cerrado típico, as a land-cover which favors hydrological processes which, in turn, promotes soil-groundwater recharge that are inportant to maintain water bodies and reservoirs in the region. Este fato destaca a importância dos ecossistemas naturais sem distúrbio, no presente caso o Cerrado típico, como uma cobertura da terra que favorce processos hidrológicos que promovem a recarga de água no solo e de água subterrânea que mantém a água nos corpos d´água e reservatórios da região.

INTRODUCTION
The Brazilian savanna, regionally known as Cerrado, is composed by a variety of forests, grasslands and savannas (Castro e Kauffmann, 1998). However, as its name suggests, the main vegetation type is savanna. These savannas are not similar in terms of woody components density (Furley, 1999). For example, Cerrado, which is also the term used to describe the most abundant savanna type, comprises at least three variation with regards to woody components density (Ribeiro e Walter, 2008), that is, (i) Cerrado ralo (savanna with low density of woody components), (ii) Cerrado denso (savanna with high density of woody components) and (iii) Cerrado típico (savanna with moderate density of woody components).
Cerrado is widely distributed within Brazil. Though massive land-use change has been taken place within this biome (Strassburg et al., 2017), original area is estimated to be in the order 2 million km 2 (Ratter et al., 1997). This means that Cerrado covers a large extent of important river-basins. For example, Tocantins-Araguaia, São Francisco and Paraná have their springs within the Cerrado biome (Lima e Silva, 2005;Lima, 2011). Although, natural ecosystems are considered to play an important role in protecting the water of such basins, few precise basic information is available with respect to soil permeability of the Cerrado. From the available data, few have precisely described the exact type of vegetation under which measurements have been taken. As shown earlier, this description in insufficient if the hydrological functioning of the different types of vegetation (frequently called "phytophysiognomies") is required to better understand the types of Cerrado. Furthermore, the few publications regarding soil permeability used a number of samples sufficient to properly describe hydraulic properties variability (Brito, 2019). Generally, a high number of samples (n ≥ 15) is necessary to characterize such properties that vary in space (Bonell et al., 2010;Salemi et al., 2020).
Understading soil permeability to water is crucial to comprehend the hydrological functioning of the Cerrado and, as a consequence, the ecosystem (hydrological) services they provide. In this respect, infiltration is one of the most relevant hydrological processes in order to allow soilgroundwater recharge which is necessary to maintain the dry season flow in rivers. The objective of the present paper was to characterize the soil surface permeability to water under a savanna type vegetation (Cerrado típico).

Study area
We used an area of Cerrado típico (savanna with moderate density of woody components) located within the campus Planaltina of the University of Brasília (Figure 1). The site is located in Brasília (Planaltina region), Distrito Federal, Brazil in the following coordinates: 15°35´59.48"S and 47°39´28.90"O. Regarding pedology, the soil type is a Latossolo Vermelho-Amarelo (Brazilian classification system) or Oxisol (US Soil taxonomy) or Ferralsol (FAO). Texture is classified as clayey (J.V. Bernardi, personal communication).
The climate of the region is markedly seasonal. Annual rainfall is around 1393 mm (Malaquias et al., 2010). Wet season (October to April) generally presents nearly 100% of rainfall (Malaquias et al., 2010).
The vegetation is described as Cerrado típico which, as shown earlier, is a type of savanna where trees and shrubs are present at moderate density when compared to other forms of Cerrado (Ribeiro e Walter, 2008). Cerrado típico (in english "Typical Cerrado") means that it is the most abundant type of savanna of the Cerrado biome.

Variables and sampling design
Infiltration capacity have been measured using Mini disk infiltrometer (Decagon Devices). Such device is a type of tension-infiltrometer which operates from 0 to -7 cm suction rate and uses the mathematical solution proposed by Zhang (1997) to calculate the hydraulic conductivity. In the present paper, aiming to estimate infiltration capacity, 0 cm suction rate has been adopted. In order to improve the contact of the infiltrometer with the soil, we used a thin layer of sand. Measurements have been taken at constant time intervals until a steady infiltration rate had been found for at least 3 times consecutively. We sampled 15 points in two linear transects following contour lines (Figure 2). The mean distance between each sampling point was 2 m.
In order to clarify indirectly soil porosity and bulk density, soil penetration resistance was measured using the Stolf impact penetrometer (Kamaq). Such device operates using 2 kg weight which drives a metal tap into to the soil (Stolf et al., 2012). We arbitrarily selected four impacts to measure soil resistance. Seven sampling units have been chosen randomly within the area of infiltration linear transects ( Figure 2). Aiming to avoid interference of soil moisture content on soil penetration resistance, measurements have been performed during the dry season.
Figure 2 -Sampling design used to sample infiltration capacity (white circles) and soil resistance to penetration (gray circles). Not scaled for simplicity.

Analysis
Descriptive statistics have been used to present the observations. In order to evaluate if the number of samples were sufficient to properly characterize soil infiltration capacity variability, we carried out the sufficiency analysis provided by Salemi et al. (2020) in which a sufficient number of samples is considered met when variability statistics reached a relatively stable (i.e. ≤ 5%) value for, at least, three times consecutively. As our infiltration data followed the normal distribution, we used standard deviation as the variability statistics. Our data was tested for normality using PAST software at α < 0.05 (Hammer et al., 2001).

RESULTS AND DISCUSSION
Infiltration capacity was estimated as 716 ± 222 mm.hr -1 (mean ± standard deviation). A high variability could be detected (Figure 3). The number of samples collected suffice following the criterion established by Salemi et el. (2020). That is, varability statistics stabilised as the number of samples increased (Figure 4). Soil presented a low resistance to penetration along all impacts ( Figure  5).

Figure 3 -Box-plot of infiltration capacity in soil under Brazilian tropical savanna (Cerrado típico).
The length of the box represents the sample interquartile range; the bar in the box is the median whereas the x represents the mean. The extremes of the vertical lines outside the box represents the minimum and maximum observed values.
The infiltration capacity of the soil under Cerrado típico was considered rapid (≥ 254 mm.hr -1 ) based on Soil Survey Staff (1993) standard. When comparing among natural ecosystems, it is within the order of magnitude of those observations carried out in the Brazilian tropical savanna documented by Souza and Alves (2003) and an order of magnitude lower compared to those by Hunke et al. (2015) (Table 1). Compared to forest ecosystems such as those of the Amazon, our infiltration estimate is also one order of magnitude lower (Zimmermann et al., 2006;Scheffler et al., 2011) (Table 1).
We hypothesize that differences in woody components density between forests and savannas and even within the savanna group (Cerrado denso vs. cerrado típico) may explain such differences. Thus, though no information on the type of "natural Cerrado" is available in Hunke et al. (2015), we suggest it was probably a Cerrado denso. In ecosystems with higher woody density, a higher biomass below and aboveground (Castro e Kauffmann, 1998) would promote higher biological activity (root growth and decay, higher litterfall, decomposition). These process may contribute to a high porosity already documented in Latossolos (Araújo et al., 2004) and observed here through our penetrometer assays. In addition, a less marked dry season in the Amazon forest would also enhance such differences through the promotion of the continuity of those biological processes since under marked dry season, as the one under our Cerrado típico region (Malaquias et al., 2010), those process may diminish with water limitation (Nardoto e Bustamante, 2003;Valenti et al., 2008).  Finally, the infiltration capacity observed here is likely to exceed the maximum rainfall intensities of the region. For example, the highest 60-minute rainfall intensity recorded at Estação Ecológica das Águas Emendas Metereological Station in 2019 (located 5 km distant from our Cerrado típico site) was 56 mm.hr -1 . Though there was no finer resolution record to estimate higher rainfall intensities, assuming the heaviest convective events available in Salemi et al. (2013) and Moraes et al. (2006) which were 105 mm.hr -1 and 80 mm.hr -1 respectively, even our lowest infiltration capacity observation (340 mm.hr -1 see Figure 1) was clearly higher than those rainfall intensities. This fact might highlight the importance of the natural undisturbed ecosystems, in this case Cerrado típico, as a land-cover which favors hydrological processes which, in turn, promotes soil-groundwater recharge that are inportant to maintain water bodies and reservoirs in the region. That is not the case of other land-uses such as soy and pasture which, demonstrably decrease infiltration capacity (Araújo et al., 2007;Bono et al., 2012;Hunke et al., 2015) favoring erosion and, as a consequence, land degradation (Montgomery, 2007 In order to predict the flowpath of the water that infiltrates the soil, observations of percolation are needed at higher soil depth. The only study which reported percolation capacity was by Hunke et al. (2015) who measured such property at 0.14, 0.2 and 0.4 m. Given that Latossolos (Oxisols) are deep soils (>>> 2 m), there is a clear gap to be bridged in future research. Likewise, though Latossolos are the most abundant soil type within the Cerrado biome (Reatto et al., 2003;Reatto e Martins, 2005), the savannas like Cerrado típico also occur in other different types of soils such as Entisols, Inceptisols, and Plinthic soils (Ribeiro e Walter, 2008). This may provide interesting opportunity to understand differences in near-surface hydrological processes in the same type of savannas under different soils.

CONCLUSION
The present study carried out an assessment of topsoil permeability under Cerrado típico, the most abundant type of savanna of the Cerrado biome (Brazil). We showed a remarkably high infiltration which probably accomodates the highest rainfall intensities in the region. This finding demonstrates the hydrological importance of the Cerrado típico in terms of promoting ecosystem services related to soil, groundwater and also biological conservation.