1. Introduction

Knickzones are defined as locally steep riverbed segments, such as a convex reach in a concave-up longitudinal profile. These particular river stretches are often characterized by changes in bed cover, channel geometry and erosion processes that are difficult to incorporate within a stream-power framework (^{Dibiase et al., 2014}; ^{Hayakawa and Oguchi, 2009}). The formative causes of knickzones include hydraulic forces, hydrological changes, climatic changes and volcanic activities. Knickzones are composed by different substrates, such as sedimentary, volcanic, plutonic and metamorphic rocks (^{Hayakawa and Oguchi, 2009}).

In case of knickzones composed by volcanic rocks it is remarkable the representativeness of the basaltic composition. In South America, they are commonly found in the Paraná River Basin. This is a geographically vast intracratonic basin, fully developed on continental crust and filled with sedimentary and volcanic rocks, whose ages range between the Silurian and the Cretaceous (^{Milani, 1997}; ^{Souza Filho and Stevaux, 2004}).

These natural rocky platforms are propitious sites for the positioning of hydroelectric dam’s axes, in terms of engineering strategy. Worldwide, electricity production increased by 72% between 1993 and 2010 and is expected to rise by an additional 56% by 2040 (^{The World Bank, 2014a}). Thirty two countries including Brazil, Mozambique, Nepal, and Norway use hydropower to produce more than 80% of their electricity requirements (^{The World Bank, 2014b}). In Brazil, there is the perspective of construction of hundreds of new hydropower plants (1871 projects in distinct stages of implementation – ^{ANEEL, 2013}), representing an imminent regional threat for this kind of habitat (locally named as “pedrais”).

Despite the fact that knickzones are worldwide distributed and geologically well studied (^{Crosby and Whipple, 2006}; ^{Hayakawa and Oguchi, 2009}; ^{Dibiase et al., 2014}; ^{Lima and Binda, 2013}), they are practically unknown environments in terms of ecological structure and limnological functioning. This lack of knowledge can be explained by the inherent difficulties to access this kind of environment, including total impossibility of safe navigation. The fact that is a mosaic of distinct habitats, including lotic and lentic ones, result in an additional methodological problem. The purpose of this research is to provide a physical description, in terms of habitats distinctiveness and a limnological characterization of a basaltic knickzone.

2. Material and methods

2.1. Study area

The studied knickzone (20° 34' 34.1” S 47° 47' 06.5” W) is located in Sapucaí-Mirim River, a tributary of Grande River basin, border of São Paulo and Minas Gerais States, Brazil. Based on recent satellite images at least eight basaltic knickzones can be still be recognized in this middle-sized river.

In the selected kind of knickzone the immersed or exposed condition of the substrate is highly variable in terms of volume and level. Differences are mainly seasonal (summer-dry and winter-rainy periods) or can also occur in stochastic short-term periods (storms and dry spells events). When the river water level is high the substrate of knickzone is fully immersed (Figure 1b). When the rainfall stops the water level begins to decrease and consequently the degree of substrate exposition increases (Figure 1a). In the study it was considered two contrasting conditions: highly immersed (rainy condition) and highly exposed (dry condition).

Figure 1 Study area – knickzone of the Sapucaí-Mirim River, Brazil, during dry (a) (July/2010) and flood condition (b) (May/2013). 

3. Results

The studied knickzone is physically characterized by the massive presence of a large rock outcrop, which originates a complex of habitats composed by riffles, runs and pools with distinct magnitudes (Figure 2a). Riffles are distributed discontinuously along the knickzone macrohabitat and are characterized by relatively narrow channels, with steep slopes and high turbulence, velocity and water flow (Figure 2b). Runs are continuous habitats, deeper than riffles, with lower turbulence and velocity (Figure 2c). Finally, the pools, such as the riffles, are heterogeneously distributed, with distinct degrees of connectivity with river flow. The pools exhibit low or no turbulence, the water exchanges tend to zero and deepness is highly variable (Figure 2d).

Figure 2 General view of a knickzone (a) formed by riffle (b), run (c) and pool (d) habitats in Sapucaí-Mirim River, Brazil. 

The fluctuation of rain precipitation and river flow during the studied period is showed in Figure 3. Both variables were positively correlated (rs = 0.618; p < 0.05). The precipitation exhibited a well-marked seasonal pattern, with high volumes in summer and maximum peaks in December and January. Annual mean precipitation, for the three year period, was 1.454 mm. Simultaneous daily values of outflow and rain precipitation along the rainy season - October 2014 to March 2015 are shown in Figure 4. A comparison between visits, satellite images and river flow suggest that this particular knickzone is totally under water when river flow values reach 90 m³ s^-1. Between October 2014 and March 2015 the knickzone exhibited this condition during 15 days.

Figure 3 Precipitation in nearest meteorological station (Franca municipality - 47 km) and outflow Bellow Palmeiras dam in Sapucaí-Mirim River, Brazil. 

Figure 4 Daily values of rain precipitation (Alta Mogiana sugar cane plant station) and outflow (downstream Palmeiras dam) in Sapucaí-Mirim River, Brazil, along the rainy season (October 2014 to April 2015). 

The amplitude of variation for all limnological measurements was higher in pools compared to riffles and runs, except for total phosphorus. Conversely, the runs were more conservative habitats, with lower variability. Comparison among habitats data obtained in dry conditions (riffles were not accessed in rain conditions as previously mentioned) is presented in Table 2.

Most limnological measurements were statistically different among habitats (p < 0.05), except for total solids and total phosphorus (Table 3). Higher water temperature occurred in pools during the rainy condition and lower values were observed in the riffles. Connected pools during dry condition exhibited higher pH values opposite to runs in the rainy condition. Lower ORP values occurred in the connected pools during dry condition and higher in pools (connected and isolated) and runs during the rainy condition. Lower values of total dissolved solids occurred in pools and runs during the rainy condition. For total nitrogen higher values were observed in the dry condition for connected pools, riffles and runs. In terms of electric conductivity higher values were measured in runs and isolated pools in dry condition contrasting to pools (isolated and connected) and runs in rainy condition.

Isolated pools in dry condition were the least turbid habitat different from connected pools and runs in the rainy condition. Runs and connected pools in dry condition, as well as riffles (only sampled in dry condition), were the most oxygenated habitats. Higher concentrations of total suspended solids in the runs during rainy condition contrasted with low values in isolated pools, riffles and runs during dry period. Remarkable higher values of inorganic suspended solids were observed for pools and runs in the rainy condition. For organic suspended solids lower mean values occurred in riffles and runs and higher in connected pools, both during dry conditions. Chlorophyll-a higher concentration was measured in isolated pools during dry condition and the lowest value in runs during the same period.

The PCA (Figure 5) explained 69.9% of data variability, considering the PC1 (54.4%) and PC2 axis (15.5%). The seasonal influence was clearly evidenced by the PC1 axis. Rainy condition was plotted on the negative side associated with high values of turbidity, total suspend solids, oxide reduction potential and temperature. The dry condition, on the positive side was positively associated with dissolved oxygen, pH, conductivity, total dissolved solids and total nitrogen.

Figure 5 Principal components analysis showing the distribution of the sampling sites in the rainy and dry condition, according to the limnological variables. 

In the rainy condition the habitats distribution is more aggregated than in the dry period. Pools were grouped on the positive side of PC2 axis associated with high values of temperature and oxide reduction potential and runs were grouped on the negative side associated with high values of turbidity and total suspend solids. During the dry condition runs were grouped on the positive side of PC2 axis associated with high values of conductivity and total dissolved solids, riffles were grouped on the negative side associated to dissolved oxygen, pH and total nitrogen and pools are distributed randomly.

4. Discussion

The basaltic knickzones exhibit a unique complex structure due to the occurrence of distinctive kind of habitats: pools, riffles and runs of different magnitudes and degree of connectivity, distributed in relatively short river stretches. These environments are different when compared with certain habitats such as streams and temporary pools. Riffles, runs and pools in stream ecosystems have a sequential distribution, in a continuous longitudinal organization (^{Allan, 1995}). In case of inland temporary pools the location is generally apart from main water bodies, like river channels. Therefore, the possibilities for comparisons of our results are limited, basically restricted to the literature related to lateral river habitats and their interactions with the central channel.

Pools, runs and rapids distributed in mosaic have a distinct limnological dynamic among them. In terms of limnological variation, pools exhibit higher amplitude when compared to runs and riffles. This is explained by their particularities, such as small volume and low water renovation, but also by the considerable physical and biotic variability features (e.g. surface area, volume, degree of connection with adjacent running channels, presence of marginal vegetation, quantity of periphyton and fish biomass). Particularly in case of freshwater rock pools the highly variable environmental conditions in such small restricted water volumes, in addition to exposition of seasonal an erratic weather, require high stress tolerance of the biota, with adaptations for surviving the dry phase such as the production of resistant stages and active emigration followed by quick re-colonization (^{Jocque et al., 2010}).

The seasonality also influence limnological characteristics in basaltic knickzones among the different microhabitats – especially the alternation between dry and rainy periods, considering that river flow is promptly influenced by the rain precipitation. In the rainy period the similarity between habitats is expected to be higher because of considerable exchanges of water, sediment, nutrients, and organisms. The opposite occurs in the dry period when the homogenization process is less effective with a higher differentiation among habitats. This dynamics is evidenced when compared river channels and lateral ecosystems such as floodplains and lagoons (^{Domitrovic, 2003}; ^{Thomaz et al., 2007}; ^{Ferrareze et al., 2014}).

In addition to the predictable seasonal changes, the basaltic knickzone habitats are exposed to erratic weather variation – random short-term events of intense precipitation and dry spell. Therefore, this environment exhibits a high disturb frequency. For more confined habitats (pools), certainly the hourly variation (nictemeral), especially in warmer periods plenty of sunshine, is also significant and would justify an additional investigation.

For characterization purposes, it is important to note that each limnological factor can vary in different ways for each particular habitat and time period. Higher temperatures were attained in pools during the rainy condition, coincident with the warmer period (late spring and summer), due to the smaller volume, low exchange rate with the main river flow and high influence of atmospheric conditions. The lowest temperature occurred in riffles, habitat with higher water flow/velocity, in the dry condition (autumn and late winter).

Pools, under dry condition, exhibited higher values of pH, organic suspended solids and oxygen values. This is probably associated to the intense periphyton colonization, resulting in high photosynthesis rate and organic detritus accumulation. In some lateral river ecosystem and lagoons, the oxygen reduction in the rainy condition is associated due to the decomposition process of organic matter coming from the adjacent plain (^{Granado and Henry, 2012}). This process may also occur in knickzone pools.

In general ORP is higher in the rainy season (pools and runs), probably due to the entrance of allochthonous material (high turbidity) from catchment, as already mentioned. An opposite tendency was observed for oxygen, which reached higher values during the dry condition in connected pools and riffles.

The lowest electrical conductivity of water (pools and runs), TDS (pools and runs) and total nitrogen (runs) in the rainy condition can be explained by the dilution effect. This pattern also occurs regionally in river lateral lagoons (^{Martins and Henry, 2004}; ^{Granado and Henry, 2012}; ^{Ferrareze et al., 2014}).

Regarding turbidity, total and inorganic suspend solids, higher values occurred in rainy condition (mainly in connected pools and runs), while lower values (except for riffles due excessive turbulence) were seen in dry period (isolated pools). High loads of external material from the drainage basin are introduced into the river system during rains, especially in habitats with more intensive water exchange. Similar findings are reported for floodplains (^{Oliveira and Calheiros, 2000}; ^{Taniguchi et al., 2004}, ²⁰⁰⁵).

The chlorophyll-a distribution was contrasting between pools and runs. Higher values occurred in isolated pools (phytoplankton colonization) and lower in runs (washout effect), both during the dry condition. The increase of chlorophyll-a during the dry period is a common pattern for large rivers and lakes of tropical/subtropical floodplains (^{Welcome, 1986}; ^{Garcia de Emiliani, 1990}; ^{Neiff, 1990}), but this seems not to be applied for all knickzone habitats.

Our research shows that the studied basaltic knickzone exhibits a clear spatial limnological heterogeneity due to the presence of distinct habitats. This condition is even more accentuated during the dry season. Probably such structural and functional features are common traits for other kinds of knickzones, which could be demonstrated through further limnological studies.