**The Role of Landscape Configuration, Season, and Distance from Contaminant Sources on the Degradation of Stream Water Quality in Urban Catchments**

#### **António Carlos Pinheiro Fernandes 1, Luís Filipe Sanches Fernandes 1, Rui Manuel Vitor Cortes <sup>1</sup> and Fernando António Leal Pacheco 2,\***


Received: 8 July 2019; Accepted: 25 September 2019; Published: 28 September 2019

**Abstract:** Water resources are threatened by many pollution sources. The harmful effects of pollution can be evaluated through biological indicators capable of tracing problems in life forms caused by the contaminants discharged into the streams. In the present study, the effects on stream water quality of landscape configuration, season, and distance from contaminant emissions of diffuse and point sources were accessed through the evaluation of a Portuguese macroinvertebrate index (IPtIN) in 12 observation points distributed within the studied area (Ave River Basin, Portugal). Partial least-squares path models (PLS-PMs) were used to set up cause–effect relationships between this index, various metrics adapted to forest, agriculture, and artificial areas, and the aforementioned emissions, considering 13 distances from the contaminant sources ranging from 100 m to 56 km. The PLS-PM models were applied to summer and winter data to explore seasonality effects. The results of PLS-PM exposed significant scale and seasonal effects. The harmful effects of artificial areas were visible for distances larger than 10 km. The impact of agriculture was also distance related, but in summer this influence was more evident. The forested areas could hold onto contamination mainly in the winter periods. The impact of diffuse contaminant emissions was stronger during summer, when accessed on a short distance. The impact of effluent discharges was small, compared to the influence of landscape metrics, and had a limited statistical significance. Overall, the PLS-PM results evidenced significant cause–effect relationships between land use metrics and stream water quality at 10 km or larger scales, regardless of the season. This result is valid for the studied catchment, but transposition to other similar catchments needs to be carefully verified given the limited, though available, number of observation points.

**Keywords:** water quality; landscape metrics; PLS-SEM; scale; season; distance from pollution sources

#### **1. Introduction**

The growing population and demographic expansion threaten hydric resources, not only by inducing stressful water demands but also because of the continuous surge of pollution sources. The response to anthropogenic pressures relies on proper management that should always stand on environmental research. The risks to water quality are well known by experts, but continuous research should be applied since the world is in constant change [1,2]. Effluent discharges are an undeniable threat. The potential contamination by wastewaters is frequently reduced in urban and industrial

areas where treatment stations are efficient, but in many regions proper treatment is not applied [3]. In those situations, surface waters are directly contaminated by bacteria [4], nutrient loads [5–7], heavy metals [8,9] and even microplastics [10]. The runoff transports herbicides and pesticides from agriculture [11] and high organic loads from livestock [7] to surface waters. The presence of forested areas or riparian vegetation can create a barrier that retains such flow of contaminants [12,13]. Wildfires are another threat to water quality [14,15], not only because they can destroy the aforementioned barriers [16], but also because soil erosion increases [17], and ash-derived contaminants are leached towards the streams [18]. Another factor that can affect water quality is land occupation planning. When land use is not conformed to land capability (natural use), land use conflict is generated [19], which amplifies soil losses [20] and accelerates other phenomena that cause water deterioration [21]. Land use type and configuration are other key aspects that have been studied by many authors in the context of water quality changes [22]. Relevant conclusions achieved in these studies are that landscapes retain nutrients [23]; a high edge density is an indicator of high anthropogenic activity [24]; as Shannon's diversity index (SHDI) increases, the water quality decreases [25]; and aggregated urban land uses are more suited to preserve surface water quality [24], among others.

The effects of landscape metrics on water quality are commonly accessed by Spearman or Pearson correlation coefficients and, when predictions are involved, through multiple linear regression analyses [26]. When these matters are studied, the authors are aware that the spatial resolution of land cover maps can affect the results [23,27]. But another critical aspect that is questioned by many authors is the spatial extent for statistical sampling [28]. This can vary from circular buffers, riparian extents, or catchments [29,30]. For proper management of river basins or even urban planning, it is essential to use an appropriate scale. By comparing different studies, some inconsistencies can be detected regarding the option for a suitable scale. Some authors infer that the use of entire watersheds provides better results [28,31–34], but other working groups argue that a riparian scale is more suitable [35–37]. These inconsistent results can be attributed to differences between study designs and study areas [38], but other factors such as stream order [28], season [39], and topology [34,39] can also play prominent roles.

Water quality research requires the use of statistical or process-based models [40]. The first type has the advantage to access the relationships between the pertinent variables, while in mechanistic models, the interactions are already determined by chemical, biological, and physical processes, which makes them preferable for prediction purposes [41].

An example of a statistical method is the multivariate method called partial least-squares path modeling (PLS-PM). The first steps of PLS-PM have been given in social sciences [42]. Nowadays, this method is being practiced in many studies in diverse research areas, namely the environment [43–46], geology [47], flood effects [48], and ecological conservation [49], among others. The authors have adopted this technique because it can exhibit cause–effect relations straightforwardly using a graphical interface. The present work continues a sequence of studies developed by this research team, who has been studying the quality of water in the Ave River Basin (Portugal) through multivariate statistics. In the first study, three PLS regression models were tested in a row to explain the pollution of surface waters and the resulting impacts on ecological integrity [50]. In a second study, an identical dataset was used in PLS-PM [51] to trace the difference of cause–effect relations between an anthropogenic (Ave River) and a rural (Sabor River) basin. Since the results were promising in both studies, in a third study they were used to predict the ecological status of the Ave River Basin in the near future [52]. The aim in the present work was to take another step forward and explore the influence of landscape metrics and contaminant emissions on ecological integrity, as well as the impact of season and scale on the results.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The Ave River Basin is located in the northern region of Portugal (Figure 1A), occupying an area of approximately 1322 km2. The main water course extends for 100 km, the most important tributary catchments are the Este (247 km2) and Vizela (323 km2) rivers. The altitude ranges from 0 m along the Atlantic coast to 1254 m at the Cabreira mountains, where the catchment headwaters are located. This river basin is surrounded to the west the by the Atlantic Ocean, to the south by the Leça River Basin, to the east by the Douro River Basin, and to the north by the Cávado River Basin. The group of three river basins, Ave, Leça and Cávado, belong to the same management unit, namely hydrographic region number 2 [53].

**Figure 1.** (**A**) Map of Portugal with the distribution of hydrographic regions. (**B**) Ave River Basin and sampling sites. (**C**) Drainage area of sampling site 111 and intersection of buffer limits with the drainage area.

In the second half of the 20th century, the Ave River Basin was heavily contaminated by untreated domestic and industrial effluents, being tagged as "Europe's Great Sewer". Following the construction of public wastewater treatment plants in the 90s, water quality increased, but some microbial contamination persisted related to improper functioning of some domestic plants. The heavy pollution of Ave comprised high concentrations of heavy metals in sediments and freshwater, especially in the Este, Selho, and Vizela rivers [9,54]. This condition improved after public investment in the wastewater treatment plants [55,56]. The abundant and persistent nutrient and metal contamination deteriorated the river's ecological status at the central area and lowlands of the Ave River Basin [57,58] where industrial areas have been settled on the river banks for years. Besides the effluents from domestic and industrial origins, contributions from agriculture and livestock production have also been reported as significant causes of water quality and ecological deterioration [59–61].

#### *2.2. Workflow*

The purpose of the present study was to show how the cause–effect relationships between ecological integrity and pollution sources/indicators changed with the season and distance from contaminant sources. Ecological integrity was assessed through the measurement of a macroinvertebrate index (IPtIN; see Equation (1) below) in 12 sampling sites along the Ave River Basin (Figure 1B) during the winter and summer seasons of 2017. For these sites, the entire upstream drainage area was delineated and then sectioned at predefined distances from the sampling site (Figure 1C). Subsequent to catchment delineation and sectioning, land use and contaminant emission data were prepared for each section to be used in PLS-PM models. Two separate models were defined based on IPtIN values determined in winter and summer, respectively. The purpose was to explore how the effects of anthropogenic pressures could change as a function of season and scale. Figure 2 displays the adopted workflow, summarized in 6 steps.

**Figure 2.** Workflow distributed in six steps.

For the delineation of drainage areas and drainage networks (Step 1), ArcMap [62] and its embedded module ArcHydro [63] were used. These computer packages are the key technical elements used in environmental studies with a strong spatial incidence [64–71]. As an input, a digital elevation model with a pixel resolution of 25 m was used. To design each drainage area, the tool "Batch Watershed Delineation" was used, followed by previous procedures [72]. For each sampling site, the "Buffer" ArcMap tool was used to create a circular area, while the created buffers extended for 100, 250, 500, 1000, 2000, 3000, 4000, 5000, 7000, 10,000, 15,000, 20,000, and 56,000 m (Step 2). Each drainage area intersected with the respective buffer in order to create drainage sections (Step 3). The collected spatial data (Table 1) were processed and applied in the PLS-PM models (Step 4). The land uses and discharge emissions were collected using the ArcMap "Intersect" tool. A total of 26 PLS-PM models were created, one for each season and distance combination. The algorithm was executed in SmartPLS software [73]. PLS-PM analysis was the chosen method because it could establish cause–effect relationships for the studied latent variables, which were termed "Land Use", "Contaminant Emissions", and "Ecological Integrity". Formative models were chosen because these have prediction capabilities and, at the same time, establish the wanted cause–effect relationships [74]. The 26 model outputs were measured with

weights, path coefficients, and R-squared values. The outputs were compiled and compared using graphics designed in Excel [75] (Step 6).



#### *2.3. Dataset*

In this study, a group of variables was gathered (Table 1) that could be connected to the variation of IPtIN. The actual data are provided as Supplementary Material. The elevation model was used for the delineation of drainage areas. The effluent discharge points values were provided by the Agência Portuguesa do Ambiente (APA; in English, Portuguese Environmental Agency) in the form of shapefiles. Each point contained was attached to information on the total discharge of nitrogen, phosphorous, and chemical and biological oxygen demands, expressed in released kilograms during the year of 2016. The discharge of nitrogen and phosphorous from livestock production, agriculture, and forestry were also provided by the APA in the form of shapefiles. Each polygon was a catchment containing the released kilograms of nutrients during the year of 2016. The most recent Portuguese land use map refers to 2015 (COS 2015) and is available at the Portuguese Territory Planning website. This map was obtained in the form of a shapefile, containing the land use or occupation from each zone. The land use types were assembled into 4 categories: agriculture, artificial surfaces, forest, and seminatural areas and water bodies. For the calculation of IPtIN, samples were collected in situ during the summer and winter of 2017. After laboratory analyses, the index was calculated for the twelve locations.

For each drainage section, the discharging values of BOD (biochemical oxygen demand), COD (chemical oxygen demand), N, and P were summed and then divided by the drainage area, creating four variables representing the discharge of each type of nutrient and oxygen demand. For diffuse discharges, the total release discharge of N and P in the drainage sections was calculated and then divided by the respective area, resulting in 4 variables: the releases of N and P from livestock and forest/agriculture. For each section, landscape metrics were calculated using a python toolbox embedded in ArcMap [76]. A total of 17 metrics were calculated for all the drainage sections (Table 2).


**Table 2.** IPtIN values and classification for the twelve measurement sites for winter and summer of 2017. An identification color was linked to each class that shades the corresponding cells: red—Very Poor, orange—Poor, yellow—Moderate, green—Excellent. The shades were added to the table for illustration and prompt interpretation of the quality classes.

The north invertebrate Portuguese index (IPtIN) is widely used to evaluate the ecological status of stream waters in northern Portugal [77]. The IPtIN index reflects the abundance and diversity of benthic invertebrates that are sensitive to all forms of pollution [77–79].

For the measurement of this indicator, organism samples were collected from 12 surface water locations, illustrated in Figure 1C. For each site, the organisms were classified and counted, and then Equation (1) was used to calculate the IPtIN score. The equation is complex since it uses a variety of parameters, namely, the number of taxonomic groups present in the sample (N◦ taxa); the number of families that belong to Ephemeroptera, Plecoptera, and Trichoptera orders (EPT); Pioleu index or evenness [80,81]; biological monitoring working party index divided by the number of families included in this index (IASPT) [82]; and the sum of individuals belonging to Heptageniidae, Ephemeridae, Brachycentridae, Goeridae, Odontoceridae, Limnephilidae, Polycentropodidae, Athericidae, Dixidae, Dolichopodidae, Empididae, and Stratiomyidae families (Sel.ETD):

$$\text{IPtI}\_{\text{N}} = \text{N}^{\circ} \text{ Taxa} \times 0.25 + \text{EPT} \times 0.15 + \text{Evenness} \times 0.1 + \text{(IASC } -2) \times 0.3 + \text{Log (Sel.ETD + 1)} \times 0.2. \tag{1}$$

Among all possible variables that could be used in this study, only 8 were chosen for the PLS-PM models. The purpose was to reach low variance inflation factors (VIFs), and hence statistical significance, making a note that variables of the same domain can be strongly correlated in raising the VIFs. To represent the effluent discharges, the released annual flow divided by the drainage area was used, naming this variable as "Point Source". For diffuse contamination, the discharges of nitrogen from livestock, forestry, and agriculture were used, naming these variables as "Livestock" and "Forest and Agriculture", respectively. The chosen landscape metrics were Shannon's diversity, the edge density of forest and seminatural areas, the number of patches of artificial surfaces that were connected at a distance of 500 m, and the percentage of "agricultural areas" that were connected at a distance of 500 m. The land use variables were named as "Diversity", "Forest", "Artificial" and "Agriculture", respectively. The Portuguese index of macroinvertebrates was also used, named as "IPtIN" for the PLS-PM models, as an evaluator of biodiversity as well as water quality.

#### **3. Results**

#### *3.1. Spatial Data*

Figure 3 illustrates the spatial distribution of pressures in the Ave River Basin. Figure 3A depicts the land use map of 2015. It can be noted that 50% was occupied by the dominant land use, forest and seminatural areas. Agricultural areas occupied 30% of the river basin, 20% was artificial surfaces, and less than 0.5% was water bodies. The values of discharged nitrogen from livestock, forest, and agricultural areas in the river basin catchments can be seen in Figure 3B,D. The scattered effluent discharge sites are represented in Figure 3C. Among a total of 60 locations, 24 were discharge sites of industrial treatment plants, while 36 were from domestic sewage treatment plants.

**Figure 3.** Spatial distribution of pressure data used in the partial least-squares path models (PLS-PMs). (**A**) Land Uses. (**B**) Forest and agriculture discharges of nitrogen. (**C**) Effluent discharge locations. (**D**) Livestock discharges of nitrogen.

The twelve river sites where the IPtIN was measured are represented in Figure 1, numbered from 101 to 112. Table 2 depicts the IPtIN values and respective classification. During the winter of 2017, the ecological status was classified as "Excellent" in two sites, "Moderate" in 4, "Poor" for another 4 sites, and "Very Poor" in 2. Overall, the values increased from winter to summer, as none of the sites was classified as "Very Poor" in summer, 6 were classified as "Poor", 3 as "Moderate", and 3 as "Excellent". In the locations 103, 105, 111, 112, 107, and 108, the classification changes were minimal. Besides, in site 111 the classification changed from "Moderate" to "Poor" because the IPtIN value was very close to the class threshold. In site 106 the classification changed from "Poor" to "Moderate", but in sites 109, 110, 103, and 104 the decrease in ecological status was startling because there was a significant decrease in IPtIN and subsequent classification of a less-rich class. For site 101 there was a significant increase, from "Poor" to the maximum class "Excellent". These changes of values were dependent upon seasonal effects but also on the pressures in surface waters.

#### *3.2. Interpretation of a PLS-PM Example Model*

The output models of SmartPLS were all similar to the one represented, as an example, in Figure 4. Each measured variable (MV) is represented as a yellow rectangle, and latent variables (LVs) as blue circles. Inside the LVs preceding other LVs, the R-squared value is portrayed. In the example model, only "Ecological Integrity" has an R-squared value, since this is the only variable that has a measured score (calculated by the sum of the product of MVs with the own weight) and a predicted score (calculated by the sum of the product between the LVs). In a PLS-PM model, weights and path coefficients are determined through an iterative process, termed the path algorithm [83], with the purpose to maximize the R-squared value.

**Figure 4.** PLS-PM example model.

Several PLS-PM models were built and analyzed in this study. In order to exemplify how these models can be interpreted, a PLS-PM model is demonstrated in Figure 4, where the data were gathered for the drainage sections within a distance of 4 km and IPtIN values measured during the winter of 2017.

Each LV was formed by one or more MVs. For the present case study, an LV "Land Use" was created and composed of 4 MVs, namely "Diversity", "Forest", "Agriculture", and "Artificial". The LV "Contaminant Emissions" was composed of three MVs pertaining to different types of contaminant flows: "Point Source", "Livestock", and "Forest and Agriculture". "Ecological Integrity" was formed by a single MV, which is the IPtIN. This LV accumulated the effects of the other LVs that were pressures in surface waters, which is why "Contaminant Emissions" and "Land Use" were connected to "Ecological Integrity". The PLS-PM model was divided into two sub-models, inner and outer. The equations of the measured scores of each LV were calculated according to Equations (2)–(4) for "Land Use", "Contaminant Emissions", and "Ecological Integrity", respectively, and are the equations that composed the outer models. The inner model was composed of relations between latent variables, which, in this case, is solely expressed by Equation (5).

$$\begin{array}{l}\text{Land Use}\_{\text{Measured Score}} =\\\text{Arithmetic} \times (-0.033) + \text{Agriculture} \times (0.516) + \text{Fores} \times (-0.569) + \text{Diversity} \times (0.208);\end{array} \tag{2}$$

Contaminant EmissionsMeasured Score = Point Source × (−0.264) + Forest and Agriculture × (−1.060) + Livestock × (1.222); (3)

$$\text{Ecolological Intensity}\_{\text{Measured Score}} = \text{PitI}\_{\text{N}} \times (1.000) \tag{4}$$

Ecological IntegrityPredicted Score = Land UseMeasured Score × (−1.085)+ Contaminant EmissionsMeasured Score × (0.266); (5)

Ecological IntegrityPredicted Score = Artificial × (0.035) + Agriculture × (−0.560) + Forest × (0.627) + Diversity × (−0.226); + (6)

Point Source × (0.070) + Forest and Agriculture × (−0.282) + Livestock × (0.325).

To interpret the example model, the weights and path coefficients should be analyzed simultaneously, which can be viewed in Equation (6), where the combination of Equations (2)–(5) is made. For example, the MV "Diversity" has a positive weight (0.208), so it increases the LV "Land Use", while the same applies to "Agriculture" (0.561). Conversely, "Forest" and "Artificial" have negative weights, namely −0.569 and −0.033, and therefore decrease "Land Use". But since the path coefficient of "Land Use" in "Ecological Integrity" is negative (−1.085), "Diversity" and "Agriculture" are variables that decrease "Ecological Integrity" because the product of the weight and path coefficient is negative: −0.226 and −0.560, respectively. On the other hand, "Forest" and "Artificial Areas" increase "Ecological Integrity", since the product between the path coefficient and weight is positive, respectively 0.617 and 0.036. Equation (6) expresses the total effect of each pressure in "Ecological Integrity". The results of this study were based on the analysis of the product between the path coefficients and weights (termed pcw) for the studied 26 models.

#### *3.3. Results of All PLS-PM Models*

As shown in Table 2, the IPtIN values were collected during two seasons, winter and summer. For this reason, the 26 PLS-PM models were divided into two groups, winter (2017) and summer (2017), and traced as two dot arrays colored as blue and red, respectively, in Figure 5. For each model in the respective group, the pressure values were used as input data, gathered from the 13 drainage sections and the IPtIN values collected in winter or summer. Figure 5 portrays the results of the PLS-PM models. Each graphic describes the pcw of a measured variable in all models (*y* axis). The *x* axis represents the logarithm of the buffer distance for the respective model. For the distances of 100, 250, 500, 1000, 2000, 3000, 4000, 5000, 7000, 10,000, 15,000, 20,000, and 56,000 meters, the log10 scores were 2, 2.4, 2.7, 3, 3.3, 3.5, 3.6, 3.7, 3.8, 4, 4.2, 4.3, and 4.7, respectively. The purpose of the plots was to illustrate the effects of the pressures in "Ecological Integrity" ("IPtIN").

The effect of "Artificial" was independent of the season since the variations with distance were practically identical for both winter and summer. For distances shorter than 10 km, the effect was positive, but for longer distances the effect became negative. The strongest positive effects were detected for a distance of 100 m in summer (pcw = 0.386) and for 1000 m in winter (pcw = 0.310). The strongest negative effects were detected for the maximum distance (56 km) (i.e., for the entire drainage areas) either in winter (pcw= −0.247) or summer (pcw= −0.201).

For "Agriculture" it was seen that for both winter and summer periods, the effect was practically identical, but the summer line was below the winter line for a majority of buffer distances (only between 3 km and 5 km is the red line above the blue). The effect of agriculture was practically null for a distance of 100 m in summer (pcw = 0.008). For the same distance, it was positive during winter (pcw = 0.276) and practically null for a distance of 250 m (pcw = 0.01). For longer distances, the effect was negative, which indicated that agriculture decreased water quality. Peak values were found for distances of 4 km in winter (pcw = −0.560) and 10 km in summer (pcw = −0.648), but for distances larger than 10 km, the changes were minimal. The results lead to the conclusion that, for the Ave River basin, agriculture is a threat to water quality, while the impact seems to be stronger during the summer period.

**Figure 5.** PLS-PM results.

Peak values of "Diversity" were detected for the minimum distance, 100 m, either in winter (pcw = −0.856) or in summer (pcw = −0.456). The effect was always negative for winter periods, and stronger in this season for almost all distances, except between 4 and 5 km. For the summer period, the effect was close to zero, but still negative; only distances of 500 m, 10, and 15 km were close to zero, since the pcw values were −0.024, 0.059, and 0.026, respectively. For the longest distance (56 km), the effect was practically the same for winter (pcw = −0.290) and summer (pcw = −0.298). The results provide evidence that the impact of "Diversity" is a threat in winter.

The effect of "Forest" was essentially positive. For the winter period, the effect was positive for all distances, and always higher than in summer, where the effect was negative but close to zero for the distances 500 m (pcw = −0.080) and 56 km (pcw = −0.049). Peak values occurred on the shortest scale, 100 m. In winter the pcw was 0.853, and in summer it was 0.419. As the buffer distances increased, the effects changed irregularly (drop and rise), but from an overall view, values were always between 0.853 and 0.374 during winter. It can be said that globally (for winter and summer), "Forest" favors water quality.

The variable that had less effects along all distances and seasons was "Point Source". For this variable and the models that comprehended distances between 100 to 500 m, the attributed weight was 0, since for short distances there were no discharge points. Even so, compared to all the other variables, it had less impact because the pcw values were contained in a short range that varied from −0.198 to 0.226. For winter, negative values were found in 1 km (pcw = −0.198) and 7 km (pcw = −0.006), while in summer the negative values were observed for distances longer than 7 km. This indicates that the effect of effluent discharges only decreased IPtIN values during the summer period when long distances were analyzed, but with minimal impact.

The discharges of nitrogen from diffuse pressures were represented by the variables "Forest and Agriculture" and "Livestock". When both graphics were compared, it was seen that there was an inverse relationship between these two variables for all models in both seasons. When the effect "Forest and Agriculture" increased, "Livestock" decreases. For the summer period, the effect of "Livestock" was always negative, while the effect of "Forest and Agriculture" was always positive. These effects were stronger over shorter distances, since maximum values for "Forest and Agriculture" and minimum values for "Livestock" appeared over the short distances. But as the distance increased, both effects approached zero. The variations of both variables were minimal for short distances (≤1 km), positive for "Forest and Agriculture", and negative for "Livestock". At the distances 2, 3, and 4 km, the effect became positive for "Livestock" and negative for "Forest and Agriculture", with a peak at 4 km. For distances longer than 4 km, the effect tended to zero.

The analysis of the pcw variations for the 7 measured variables for all the models is crucial to comprehend the cause–effect relationship changes as function of season and distance. But the analysis of the R-squared values (Figure 6) reveals the models' capacity to explain IPtIN variations.

**Figure 6.** PLS-PM R-squared values.

The calculated R-squared values of summer varied less than the winter counterparts. The range of values varied from 0.75 (1 km) to 0.91 (56 km) in summer, while in winter they ranged from 0.58 (500 m) to 0.93 (7 km). For the winter period, for distances comprehended between 250 m and 3 km, the model's explicability was below 0.75, but the in winter models that comprehended distances between 4 to 20 km, the values were higher than in the corresponding summer models.

In order to assure that the models had no multicollinearity, it was ensured that all the VIF values were below 5 (please see Supplementary Material). The significance of weights and path coefficients was accessed through bootstrapping. By approaching the traditional threshold for statistical significance, *p* values larger than 0.05 were achieved for the weights (please see Supplementary Material). On the other hand, it was verified that the path coefficients of "Land Use" were significant, characterized by *p* values lower than 0.05 for long distances (Figure 7).

**Figure 7.** PLS-PM significance of path coefficients.

The significance of land uses seemed to follow a sigmoid pattern. For the summer period, statistical significance (*p* < 0.05) was achieved for distances larger than 5 km, but for the winter period, significance was achieved for distances larger than 3 km. When it comes to "Contaminant emissions" none of the models achieved statistical significance. For the summer period, *p* values increased with distance. For winter, *p* values seemed not to change for distances below 2 km, increased to 0.124 at 3 km, and then dropped consistently until a distance of 15 km. For 20 and 56 km, there was a notable loss of statistical significance.

#### **4. Discussion**

Before analyzing the results, some expectations regarding the effect of the variables are outlined. It was expected that all pressures had a negative impact on ecological integrity, while "Forest" was expected to have a positive effect. This was expected because in catchments with a high presence of forested areas, good water quality can be found [84]. It was also thought that the measured variables that belonged to the latent variable "Contaminant Emissions" would have a higher effect than "Land Use". This was because the discharges of COD from point sources and nitrogen from livestock, agriculture, and forestry represented the mass flow of contaminants that were transported to surface water, while land use metrics were only indicators of possible pollution. In terms of season and scale, no expectations were anticipated because authors already recognized in different studies that different conclusions can be achieved [34].

The positive effect of "Artificial" for distances below 10 km is hard to explain. By studying the impact of land uses on biological integrity, a positive response was found in urbanized areas [31] by accessing a riparian scale. Another author [85] compared the effects of land uses in the same Index, of Biotic Integrity (IBI) and noted that, in a riparian range, the impacts of urban activity were positively correlated to this index, but for the watershed scale, the outcome was negative, which is in concordance with the present study. This might happen because, at a short range, the impacts of urban presence

may be hard to capture, compared to an extended scale. Possibly, urbanized areas only affect water quality when their predominance occurs over a long range.

The agricultural land uses are revealed as a threat to Ave River Basin. For distances larger than 200 m, the effect was negative and increased in scale for both winter and summer seasons. This can reflect that agriculture only affects water quality when the predominance is on a long scale, as there is a high accumulation of contaminants. Likewise, other studies revealed a negative impact over a long scale [24,34,39,86].

The edge density of forested areas "Forest" was the variable that had an expected impact on winter and summer (except at the distances of 56 km and 500 m, where the effects were null and negative, respectively). Many authors have concluded that the effect of forestry improves water quality. Positive impacts are found with biotic indexes [31,32], and negative correlations or effects with contaminant concentrations are found [23–25,28,39,86,87] independently of scale, season, and study area, or even accessed metric.

In light of the previously mentioned reasons, it was not expected that the effect of "Land use" would be greater than "Contaminant Emissions". It was noticed that in the winter period, the measured variables that belonged to "Land Use" were the ones with stronger effects (except in the model of the scale 500 m, where livestock had the strongest impact). But, in the summer period, the contaminant emissions from "Forest and Agriculture" and "Livestock" were the variables with the highest pcw. Only at a scale higher than 7 km did the effect of agricultural land use overlap contaminant emissions.

When it comes to significance, the path coefficient of "Land Use" achieved statistical significance in both seasons at a long scale, while "Contaminant emissions" did not. This leads to the conclusion that, in concordance with other authors, the effect of land use should be accessed on a long scale, also called a complete watershed. Both diffuse and point-source discharges have temporal changes [7,88–90]. To access the seasonal effects of contaminant emissions, it is important to trace the temporal changes in the contaminant flow, just like it was done in other studies [91]. In the present study, the point source flow of COD and diffuse discharge resulting from livestock production, forested, and agricultural activities was in the form of annual flow. In fact, it is quite hard to access the released contamination on shorter temporal scales for a whole river basin. But, if the data of point and diffuse sources from APA were monthly, or even seasonal, it is believed that the significance of "Contaminant Emissions" would be higher in the model and could possibly reveal higher effects than landscape metrics. In previous studies where the Ave River Basin water quality was assessed, it was noted that point-source pressures [51] and livestock production [52] were major threats to water quality. But, in those studies, landscape metrics were not used in such a detailed form, only the percentage of catchments occupied by agricultural and artificial areas were used. During summer, the contribution of underground water to river discharge increased due to the lack of rainfall [92]. In the presented models this is shown because contaminant emissions from agriculture, forest, and livestock effects were indeed larger in summer periods than in winter. During winter, the runoff effect was higher due to strong rainfalls, which explains why variables related to land use metrics had a stronger impact in winter rather than in summer.

Another important aspect is that when formative PLS-PM models are adopted, the number of explaining variables cannot be large because of the shortcomings of variance inflation. One technique to reduce variance inflation is by restraining the number of variables [93]. For this reason, the effects of other land use metrics were not accessed in this study, only the edge density of forested areas, Shannon's diversity index, and connectance metrics of agricultural and artificial surfaces. However, this study succeeded in demonstrating the seasonal and scale effects in the interaction between various pollution sources and ecological integrity. The low sample size (n = 12) was a limitation. As a consequence, the weights were not significant. Despite this, it was still possible to achieve significance for the path coefficient "Land Use" latent variable. So, it is recommended to use larger sample sizes in similar and future studies. Since the weights were not significant, the tested models may not be suited for prediction purposes. At any rate, the presented study indicates that, in terms of prediction in the Ave

River Basin, longer scales should be adopted, considering the consistent high R-squared values and significance of land use variables observed at these scales.

For proper river basin management, landscape metric variables should always be assessed, as it is quite easy to calculate them using computer packages such as FRAGSTATS [94] or the ArcGIS toolbox [76]. In terms of landscape management, when the consequences of anthropogenic pressures for water quality are to be assessed, the exercise should always be applied to various scales. This is recommended because this study concluded that the sense (positive or negative) of an impact can change with scale.

The most limiting factor in this study was probably the small number of sampling points used to assess the IPtIN, just 12. However, these few samples allowed us to provisionally expose the significant role (*p* < 0.05) of land use metrics for a satisfactory (*R*<sup>2</sup> <sup>≈</sup> 0.85) explanation of water quality (IPtIN) in the long range (>4 km from the contaminant sources). This result is noteworthy. It can be (and was) argued that more samples would render the possibility to reveal the influence of other anthropogenic pressures, eventually hidden in this study by the sample's coarse resolution. Nevertheless, it would not be a surprise if the results obtained in this study were replicated with a finer resolution, as contaminant emissions are subject to larger inter annual variations than are land uses, and, hence, they could eventually be inefficient in the studied period. A larger sample would probably capture fine-resolution effects, for example, related to point-source contaminant emissions, but is not certain that would change the general outcomes and conclusions taken from this study. The main goals were achieved, which were to explore the influence of anthropogenic pressures on water quality as function of scale and season using a novel statistical method. The 26 PLS-PM models implemented in a predefined sequence were capable of identifying the most important variables and distances from contaminant sources that controlled water quality in the Ave River Basin in the studied period. The model results may not be directly used in management initiatives without prior verification using a larger sample, but they suggested how scale and season can affect the conclusions about cause–effect relationships involving anthropogenic pressures and water quality. In that context, the outcomes from this study provided interesting clues for managers of water quality at catchment scale, which are inherently an important scientific result.

#### **5. Conclusions**

This study has shown to be effective in demonstrating seasonal and scale impacts in the interplay between the effect of landscape metrics and contamination sources on water quality. This analysis plays an important role for decision makers to take into account that territory planning is intrinsically linked to water quality. As it was found in this study, the effect of metrics can be greater than contamination sources. From a statistical point of view, this study showed that the use of a long scale is preferable, since it obtained higher coefficients of determination in both winter and summer, but also high statistical significance for the latent variable "Land Use". Even so, it is advised that when water quality studies are carried out, effects should always be analyzed not at a long scale but also at a short scale. This is because each river basin is unique and reveals natural and anthropogenic interactions that are different in each river basin. So, in other locations, stronger effects can be found at shorter scales. In terms of water quality improvement, besides the constant monitoring and reduction of pollution sources, it is pointed out that the presence of forested areas improves the ecological integrity, not in total area but in terms of edge density. As far as agricultural areas are concerned for ecological integrity, strategic relocations can also be a key strategy in order to decrease connectance. While in urban areas, these hardly can be changed, given the enormous costs that such processes can entail.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4441/11/10/2025/s1, The supplementary materials comprise an Excel file with base data and results of PLS-PM models.

**Author Contributions:** Conceptualization, A.C.P.F.; methodology, A.C.P.F. and F.A.L.P.; software, A.C.P.F.; validation, F.A.L.P.; formal analysis, L.F.S.F.; investigation, A.C.P.F.; resources, L.F.S.F. and R.M.V.C.; data curation, A.C.P.F., F.A.L.P., L.F.S.F. and R.M.V.C.; writing—original draft preparation, A.C.P.F.; writing—review and editing, F.A.L.P.; visualization, A.C.P.F.; supervision, F.A.L.P. and L.F.S.F.; project administration, R.M.V.C.; funding acquisition, R.M.V.C.

**Funding:** This research was funded by the INTERACT project "Integrated Research in Environment, Agro-Chain and Technology", no. NORTE-01-0145-FEDER-000017, in the line of research entitled BEST "Bio-economy and Sustainability", and co-financed by the European Regional Development Fund (ERDF) through NORTE 2020 (the North Regional Operational Program 2014/2020). For authors at the CITAB research centre, this research was further financed by the FEDER/COMPETE/POCI—Operational Competitiveness and Internationalization Programme under Project POCI-01-0145-FEDER-006958, and by National Funds of FCT—Portuguese Foundation for Science and Technology under the project UID/AGR/04033/2019. For the author in the CQVR, this research was additionally supported by National Funds of FCT—Portuguese Foundation for Science and Technology under the project UID/QUI/00616/2019.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **The Buffer Capacity of Riparian Vegetation to Control Water Quality in Anthropogenic Catchments from a Legally Protected Area: A Critical View over the Brazilian New Forest Code**

**Carlos Alberto Valera 1,2,3, Teresa Cristina Tarlé Pissarra 2,3, Marcílio Vieira Martins Filho 2,3, Renato Farias do Valle Júnior 3,4, Caroline Fávaro Oliveira 3,4, João Paulo Moura 5and Fernando António Leal Pacheco 3,6,\***


Received: 10 February 2019; Accepted: 12 March 2019; Published: 16 March 2019

**Abstract:** The riparian buffer width on watersheds has been modified over the last decades. The human settlements heavily used and have significantly altered those areas, for farming, urbanization, recreation and other functions. In order to protect freshwater ecosystems, riparian areas have recently assumed world recognition and considered valuable areas for the conservation of nature and biodiversity, protected by forest laws and policies as permanent preservation areas. The objective of this work was to compare parameters from riparian areas related to a natural watercourse less than 10 m wide, for specific purposes in Law No. 4761/65, now revoked and replaced by Law No. 12651/12, known as the New Forest Code. The effects of 15, 30 and 50 m wide riparian forest in water and soil of three headwater catchments used for sugar cane production were analyzed. The catchments are located in the Environmental Protection Area of Uberaba River Basin (state of Minas Gerais, Brazil), legally protected for conservation of water resources and native vegetation. A field survey was carried out in the catchments for verification of land uses, while periodical campaigns were conducted for monthly water sampling and seasonal soil sampling within the studied riparian buffers. The physico-chemical parameters of water were handled by ANOVA (Tukey's mean test) for recognition of differences among catchments, while thematic maps were elaborated in a geographic information system for illustration purposes. The results suggested that the 10, 30 or even 50 m wide riparian buffers are not able to fulfill the environmental function of preserving water resources, and therefore are incapable to ensure the well-being of human populations. Therefore, the limits imposed by the actual Brazilian Forest Code should be enlarged substantially.

**Keywords:** water pollution; riparian forest; environmental Law; anthropogenic catchment; watershed management; land use policy

#### **1. Introduction**

Riparian forests are woodlands in association with streams, rivers and lakes. The location of riparian forests adjacent to water courses ensures that they can exert a strong influence on the quality of freshwater and help to protect the whole ecosystem from anthropogenic activities taking place upwards in the watershed [1–4]. Besides protection, riparian forests provide multiple services such as habitat for aquatic species, soil biodiversity, sediment filtering, flood control, stream channel stability and aquifer recharge [5–13].

The water, soil and vegetation of riparian forests are state indicators of conservation and preservation of land and stream suitability [14]. The biotic community components act as integrator of ecological conditions [12,15,16] and form the transition between the aquatic environment and the anthropogenic pressure. From a different standpoint, the biotic community components express the different spatial and temporal scales of anthropogenic pressures, and therefore support the environmental assessment of watersheds [3,17,18]. For this reason, efforts should be made to understand the theory and metrics of soil attributes and water quality in riparian buffer ecosystems and their link to specific or aggregated types of anthropogenic disturbance [19].

Studies on the width of riparian forests are abundant and relevant [20–22], but only a few works had the main purpose to contribute, from scientific grounds, to the evaluation of environmental laws. In Brazil, riparian forests are called permanent preservation areas (PPA) under the terms of articles 4th, 5th and 6th of Law No. 12651/12 (the so-called New Forest Code), being defined as: "*protected area, covered or not by native vegetation, with the environmental function of preserving water resources, the landscape, the geological stability and the biodiversity, facilitating the gene flow of fauna and flora, protecting the soil and ensuring the well-being of human populations*" [23].

The technical concept of PPA was introduced in the first Brazilian Forest Code, published on the 23rd January 1934 (Federal Decree No. 23793/34), which has categorized the national forest into four types: protected forest, remaining forest, model forest and income forest. Among other roles, the protected forests were meant to preserve the water flow, minimize the erosion process and ensure public health conditions, and therefore fall into the current concept of permanent preservation area. It is worth to mention that the legal concept of PPA, already used in the revoked Federal Law No. 4771/65 and reproduced in the current Federal Law No. 12651/12, was not created by the legislator. Instead, the legislator has appropriated the existing technical and scientific knowledge for the normative definition of that ecosystem. On 1965, when the Forest Law was published, there was no Ministry of Environment, and the environmental terms were themselves incipient, resulting that Federal Law No. 4771/65 was created and managed by the Ministry of Agriculture, Livestock and Supply.

The 1965 and 2012 forest laws were mostly based on the concept of preservation. Other concepts and definitions equally relevant for the role of riparian forests as preservation, such as ecological function or ecosystem service, were not emphasized in these laws. The Ecological function is "*the operation by which the biotic and abiotic elements that are part of a given environment contribute, in their interaction, to the maintenance of the ecological balance and to the sustainability of the evolutionary processes*". By fulfilling this function the PPA would provide ecosystem services through ecological and evolutionary processes, including gene flow, disturbance and nutrient cycling, besides the preservation issue. The ecosystem service concept and the practical assessment of ecosystem services [24] in watersheds [25] should be more explicitly applied to the PPAs of anthropogenic catchments.

The New Forest Code has also reduced the overall protection of riparian forests. The width of riparian buffers has not been altered in the new Law, but the location criteria used to measure it have changed. This has led to a smaller area of protected forests, besides the implications for the renting of such protected spaces as well as for transition rules (article No. 59 of Federal Law No. 12651/12). The reduction of riparian vegetation reduces the environmental protection of streams provided by these "green filters", and therefore the likelihood of ecological disasters is expected to increase compromising the sustainability of aquatic systems. A coherent forest code should look upon catchments as spaces where man and nature coexist and self-sustain. A different look inevitably opens the space for radically opposing goals based on the same concept [14]. Therefore, a scientifically based assessment of forest laws represents an environmental policy topic worthy of investigation.

This study aims to take that step forward, namely to compare riparian buffer widths as defined by the revoked (Law No. 4771/65; [26]) and current (Law No. 12651/12; [23]) forest laws for the marginal areas of streams less than 10 m wide, and verify their effects on water and soil resources. The specific goals are: (1) to study riparian buffer soils and water quality along watercourses of anthropogenic watersheds, namely watersheds used for sugar cane production. Watercourses in these catchments may be affected by a diversity of pollutants, including nitrogen and phosphorus from fertilizers or fine sediments from soil erosion. In this study, water quality was assessed by an index that involves the measurement of dissolved oxygen, turbidity, total dissolved solids, which means parameters that can be interpreted as proxies to those pollutants. The index is called the *IWQ*—Index for Water Quality and was proposed by the Environmental Company of São Paulo State—CETESB (https://cetesb.sp.gov.br) to be used in water quality assessments. The study was replicated in watercourses with 15, 30 and 50 m wide riparian forests; (2) to look upon the riparian buffer width defined by the New Forest Code and attempt to understand the underlying environmental function of conserving water resources; (3) to define metrics for the evaluation of water and soil resources within riparian buffers. Due to its regional and national importance, this research was carried out in the Uberaba River Basin, namely at the Municipal Environmental Protection Area.

#### **2. Materials and Methods**

#### *2.1. Study Area*

The study area comprises the Municipal Environmental Protection Area of Uberaba River Basin (EPA-URB), which is located in the Triângulo Mineiro Region, State of Minas Gerais, Brazil (Figure 1). The EPA-URB occupies an area of approximately 525 km<sup>2</sup> between the Meridian coordinates 188–220 km East and Parallel coordinates 7815–7840 km North of Universal Transverse Mercator coordinate system, 23K. The EPA-URB was acknowledged as a Sustainable Land Use Conservation Unit, which is a portion of Minas Gerais State territorial waters subject to a special regime of administration. The demarcation of the EPA-URB involved the recognition of important natural characteristics besides water resources, namely native vegetation (Cerrado Biome), worth of state protection by the Municipal Law No. 9892 of 28 December 2005.

According to Köppen's climate classification, the region is classified as Aw, tropical, and the climatic domain is classified as semi-humid with 4 to 5 dry months, with a relative humidity of 70–75%. The average annual temperature varies between 20 and 24 ◦C. The warmest months are October to February, with temperatures ranging between 21 and 25 ◦C. The month of July is the coldest month with temperatures ranging from 16 to 18 ◦C. The long-term (sixty two year record) mean annual precipitation in Uberaba municipality is 1584.2 mm. On a monthly basis, average rainfall varies between 42.8 and 541 mm (www.inmet.gov.br/).

The EPA-URB is located in the Central Brazil Plateau and northeast portion of Paraná Basin. Topography is characterized by undulated landscapes. Geology is dominated by a sedimentary sequence comprising two major geologic groups and associated formations: the Sao Bento Group and Serra Geral Formation; the Bauru Group and the Marilia and Uberaba formations. The São Bento Group is composed of basalts cropping out at lower altitudes. The Uberaba Formation is made up of Cenozoic sediments, with a predominance of sedimentary rocks with volcaniclastic contribution, and overlays the Serra Geral Formation along an erosive contact. The upper contact with the Marília

Formation is also considered abrupt, being marked by a silexite level and a conglomerate rich in quartz grains cemented by calcite [27].

**Figure 1.** Location of the Environmental Protection Area of Uberaba River Basin ( EPA-URB) in the Uberaba Municipality, State of Minas Gerais, and Brazil.

The main soil units are latosols (predominant) and argisols (small areas), according to the Brazilian system of soil classification (https://www.embrapa.br/solos/sibcs). These soil units correspond to ferralsols in the World Reference Base (http://www.fao.org/soils-portal) and oxisols in the classification scheme of Natural Resources Conservation Service (https://www.nrcs.usda.gov/wps/ portal/nrcs/site/soils). The latosols are characterized by clayey texture whereas the argisols are characterized by sandy texture.

#### *2.2. Experimental Sites (Sub-Catchments)*

The experimental sites comprised three sub-basins selected within the EPA-URB, termed Mangabeira 1 (area: 373.09 ha), Mangabeira 2 (426.6 ha) and Lanhoso (1243.64 ha) (Figure 2). In all cases the catchments were mostly used for sugar cane plantations, which occupy 49.4, 39.5 and 34.3% of the area, respectively, and therefore could be considered anthropogenic basins. Besides this use, the catchments were substantially occupied by native forests (36.1, 30.9 and 53.1%). However, the riparian buffers marginal to the watercourses were characterized by quite different widths: on average, 15 m in Mangabeira 1, 30 m in Mangabeira 2 and 50 m in Lanhoso. The samples of soil and water were collected at the sub-basin outlet.

**Figure 2.** Experimental sites (sub-basins Mangabeira 1, Mangabeira 2 and Lanhoso).

Land uses in the three sub-basins are illustrated in Figure 3a–c and can be summarized as follows: *pasture*—natural or managed pastures (used for the grazing of domestic livestock), sometimes composed of grasses and forbs, in other cases including native vegetation; *sugar cane*—sugar cane plantations; *rural dwelling*—space occupied by the people who work on farms and related activities; *native forest*—area occupied by spontaneous native vegetation, sometimes deforested; *managed forest*—mainly eucalyptus stands; *water bodies*—lakes and reservoirs; *roads*—paved roads; *other land uses*—include orchards, and areas used for rain fed or irrigated corps.

**Figure 3.** *Cont*.

**Figure 3.** (**a**) Land use in Mangabeira 1 sub-basin. Buffer strip width: 15 m; (**b**) Land use in Mangabeira 2 sub-basin. Buffer strip width: 30 m.; (**c**) Land use in Lanhoso sub-basin. Buffer strip width: 50 m.

#### *2.3. Sampling and Analysis*

#### 2.3.1. Soils

The sampling of soils was carried out in the areas occupied by riparian vegetation. Depending on the sub-basin, this represented a buffer extending 15, 30 or 50 m from the watercourse upwards. The sampling procedure followed the guidelines of São Paulo State Environmental Agency [28],

and took place in April and November of 2015 at least 10 m away from the stream within the buffer zone. The campaigns involved the collection of undisturbed as well as disturbed samples. The undisturbed samples were collected at the 0–20 and 20–40 cm depth layers to determine soil density and other physical attributes. The number of sites per sub-basin was five, and therefore the total number of undisturbed samples was 60 spanning the two sampling seasons. The disturbed soils were collected within a 10 m square grid with 3 columns and 6 rows, in a total of 18 sites (repetitions). Considering the number of seasons (2), the number of sub-basins (3), and the number of sites per sub-basin (18), the amount of disturbed soil samples was 108.

Following collection, the soil samples were air dried, stripped and passed through a 2 mm mesh screen for chemical analyses. The analyses followed the methods of [29] and involved determination of: pH; Al3+(cmol*<sup>c</sup>* dm−3); Ca—exchangeable calcium (cmol*<sup>c</sup>* dm−3); Mg—exchangeable magnesium (cmol*<sup>c</sup>* dm−3); H+Al—potential acidity (cmol*<sup>c</sup>* dm−3); SB—sum of bases (cmol*<sup>c</sup>* dm−3); t = SB + Al3+—Cation exchange capacity (cmol*<sup>c</sup>* dm−3); T—Cation exchange capacity at pH 7.0, calculated as a function of (SB) + (H+Al), expressed as cmol*<sup>c</sup>* dm−3; K—exchangeable potassium (mg dm−3); P—available phosphorus (mg dm<sup>−</sup>3); V = SB/CEC—Base Saturation (%); m—aluminum saturation (%); SOM—soil organic matter (dag kg<sup>−</sup>1); OC—organic carbon (dag kg−1); Sand (%); Silte (%); Clay (%).

#### 2.3.2. Water

The stream water samples were collected immediately downstream from the soil sampling sites, in sectors of the stream that were adjacent to the riparian buffer. The sampling took place approximately 60 cm far from the stream margin, every month during 13 months (January 2016–January 2017). The annual rainfall in 2016 was 1214.4 mm. This value is smaller than the long-term average (1584.2 mm), meaning that 2016 was a dry year. Each month, the samples were collected between calendar days 15 and 20. The weather conditions in the sampling day as well as during the three antecedent days are summarized in Table 1. In the sampling day, rainfall was always <5 mm with the exception of February 2016 and January 2017 campaigns, when rainfall reached 5.9 and 10.9 mm, respectively. In the antecedent days, average rainfall was also small (3.5–5.8 mm), with few exceptions represented in bold in Table 1. The antecedent days with a substantial rainfall were 13 November 2016, and 16 January 2017, with precipitation >25 mm. Therefore, the average analytical results should reflect long-term effects of land use and buffer strip width on the quality of stream water rather than short term effects related with storm events. In the sampling site of a catchment, each campaign involved the measurement of water quality parameters in 10 samples (repetitions), according to CONAMA Resolution No. 357/2005. The parameters were measured using a Horiba U-50 Series multi-parameter probe, and comprised: T—water temperature (◦C), pH, ORP—Oxidation Reduction Potential (mV), Ec—Electrical conductivity (μS cm<sup>−</sup>1), Turbidity, measured in Nephelometric Turbidity Units (NTU), DO—Dissolved oxygen (mg L<sup>−</sup>1), PDO—Percentage of Dissolved Oxygen (%), and TDS—total dissolved solids (mg L−1).

**Table 1.** Weather conditions (rainfall) in the water sampling day and the three antecedent days (day-1 until day-3). Values larger than 10 mm day−<sup>1</sup> are represented in boldface.


A subset of parameters was used to calculate the Index for Water Quality (*IWQ*) proposed by the Environmental Company of São Paulo State—CETESB (https://cetesb.sp.gov.br):

$$I\mathcal{W}Q = \prod\_{i=1}^{n} q\_i^{w\_i} \tag{1}$$

where 0 ≤ *IWQ* ≤ 100, *qi* is the quality of *i*th parameter obtained from standardization of the measured values into a 0–100 range, *wi* is the weight of *i*th parameter, which varies in the 0 ≤ *wi* ≤ 1 interval as function of its importance to the overall quality, and *n* is the total number of parameters. According to CETESB, *n* = 9 and comprises water temperature, pH, dissolved oxygen, turbidity, total dissolved solids, biochemical oxygen demand, fecal coliforms, total nitrogen and total phosphorus. When data is lacking for some of these parameters, the index can still be calculated using a different set of weights as proposed by [30]. The calculus of *IWQ* in the present study was based on the first five parameters from the list (*n* = 5) and on the following weights: 0.10 (water temperature); 0.21 (pH); 0.17 (turbidity); 0.2 (dissolved oxygen); 0.17 (total dissolved solids). The standardization curves for these parameters, which transform the measured parameters into *q* scores (Equation (1)), are portrayed in Figure 4.

**Figure 4.** Standardization curves used to transform the water quality parameters into *q* scores (Equation (1)). Source: https://cetesb.sp.gov.br.

According to the *IWQ*, the quality of stream water is graded as follows: extremely poor (*IWQ* ≤ 19), poor (19 < *IWQ* ≤ 36), regular (36 < *IWQ* ≤ 51), good (51 < *IWQ* ≤ 79), excellent (79 < *IWQ* ≤ 100). It is worth to note that the *IWQ* is rather sensitive to small changes in the bearing parameters, given the multiplicative formulation of Equation (1). As corollary of this conception, a good water quality (*IWQ* > 51) requires that all *q* values are high while an excellent quality (*IWQ* > 79) implies that all *q* scores are very high.

#### *2.4. Thematic Maps and Statistical Treatment of Soil and Water Data*

The thematic maps (e.g., Figures 1–3) were prepared in ArcMap software of ESRI [31], a common tool in spatial analysis of hydrologic and environmental data widely used in many recent studies [32,33]. The base information was compiled from various spatial databases, namely the maps published by the Brazilian Institute for Geography and Statistics (https://ww2.ibge.gov.br) on the 1:100,000 scale, and the digital terrain model obtained from the ASTER GDEN V2 satellite image with spatial resolution of 30 m. The statistical treatment of water data was based on the analysis of variance (ANOVA) and Tukey's mean test (*p* < 0.05). The data were processed in the *R* computer program (https://www.r-project.org/).

#### **3. Results**

The analytical results are depicted in Table 2 (soils) and Table 3 (water). The water quality index (*IWQ*) is depicted in Table 4.


**Table 2.** Analytical results for the soil samples. The symbols were defined in the text (Section 2.3.1).

**Table 3.** Analytical results for the water samples: average value, standard deviation, Tukey's mean test result (ANOVA; *p* < 0.05). The symbols were defined in the text (Section 2.3.2). Values with different label (lowercase letters *a*, *b*, *c* or *d*) are considered significantly different from each other by the Tukey's test, and therefore can be used to differentiate the sub-basins.



**Table 4.** Water quality index (*IWQ*) and potentially related environmental variables.

The relationship between soil parameters and the riparian buffer width (Table 2) was not detected for most parameters. However, the percentage of sand and the content of aluminum increased as function of width, while the silt content decreased. The results for sand and aluminum seem to expose the capacity of buffer strips to retain mineral aggregates, especially the more coarse grained. The results for silt may be apparent because sand, silt and clay in a texture analysis sum 100% and therefore when a fraction increases the other tend to decrease regardless of their abundance in the sample.

The analytical results for water (Table 3) indicate a statistical difference between the Mangabeira catchments (CM1 and CM2) and the Lanhoso catchment (CLn), in the case of pH, oxidation-reduction potential, conductivity, dissolved oxygen (in mg L−<sup>1</sup> or %) and total dissolved solids, which means the majority of parameters. This is strong indication that the catchment with a wider riparian buffer is different from the catchments with a narrower buffer, as regards water quality. The results of the *IWQ* calculation showed reduced values in all basins. These results qualified the stream waters as poor.

When the *IWQ* parameter was plotted as a function of buffer width (Figure 5a) and a trend line was fitted to the scatter points, the fitting equation was parabolic:

$$MWQ = 0.003 \text{ BW}^2 - 0.1238 \text{BW} + 31.971 \tag{2}$$

where *BW* means buffer width. A similar plot, but of *IWQ* as function of *BW* combined with land use/occupation (*NF*/*SC* = native forest/sugar cane ratio; Figure 5b), could be fitted to the following linear equation:

$$I\%Q = 0.0391BW\frac{NF}{SC} + 30.26\tag{3}$$

**Figure 5.** Plot of water quality index (*IWQ*) as function of: (**a**) riparian buffer width (*BW*); (**b**) buffer strip width combined with native forest/sugar cane ratio (*BW* × *NF/SC*). The points represent the values of *IWQ* versus *BW* or *IWQ versus BW* × *NF/SC* in the Mangabeira 1 (CM1), Mangabeira 2 (CM2) and Lanhoso (CLn) sub-basins.

Equation (2) attempts to describe the independent influence of buffer strip width on water quality, while Equation (3) analyzes this influence but coupled the potential interference of land use. In fact, the two effects are barely separable because an increase of buffer strip width tends to increase the NF/SC ratio.

The parabolic trend in Figure 5a may reflect the fact that water quality in CM1 is affected by water quality of CM2, besides the influence of local land use and buffer strip width. Because the *IWQ* in both catchments are similar, regardless the differences between buffer strip widths, the contribution from CM2 is probably large. Put another way, if water quality in CM1 reflected solely land use and buffer strip width, then a lower *IWQ* would be expected in this catchment. In the figure, the CM1 point would drop along the *y*-axis and the general trend would shift from the parabolic trend towards a (more likely) linear trend.

Using Equation (3) with constant *NF/SC* values allows relating water quality (*IWQ*) and buffer strip width for specific land occupations. Table 5 describes these relationships for three *NF/SC* ratios: 0.7 (the lowest ratio in the studied catchments), 1.6 (highest ratio) and 3.2 (twice the highest ratio, forecasting implementation of conservation practices through expansion of native forest). In the first case (*NF/SC* = 0.7), it is expected that a regular water quality is attained for *BW* = 205 m, while this value reduced to *BW* = 90 m for *NF/SC* = 1.6, and to *BW* = 45 m for *NF/SC* = 3.2. If the buffer strip width would duplicate in the Lanhoso catchment through implementation of conservation practices, the water quality would become good for *BW* = 155 m. Therefore, the 30 m threshold foreseen in Federal Law No. 12651/2012 may not satisfy the water quality requirements in the studied basins, while it is worth remembering that these catchments are located in a protected area for water resources. It should be admitted, however, that this study was based on a small number of catchments and that more general conclusions would require a more exhaustive analysis based on a larger sample. Besides, the interpretations so far hold for anthropogenic catchments used for sugar cane production and cannot be extended directly to other land uses. Finally, the weather reference for this study is a dry year.


**Table 5.** Riparian buffer width (*BW*), native forest/sugar cane ratio (*NF/SC*) and related water quality (*IWQ*), as predicted by Equation (3) for three constant values of *NF/SC*.

It is also worth recalling that water quality expose the aggregate effects of all natural processes and anthropogenic inputs [34,35] that can occur along the flow paths [36], namely chemical weathering [37–40], uptake/release from/to biota [41], leachates from fertilizers that also affect chemical weathering [42–44], discharge of domestic sewage [45,46], among others. The trend depicted in Figure 5b exposes the impact of sugar cane on *IWQ* combined with the buffering capacity of native forest. It is expected that in absence of native forest (*NF* = 0) the water quality index would drop to *IWQ* = 30.26 because *NF* = 0 implies *BW* × *NF/SC* = 0. This rather low level of *IWQ* would represent the impact on water quality exclusively attributed to sugar cane production (*IWQ*0), including the effects related to fertilizing and management (e.g., erosion control). According to Equation (3), if fertilizing and management conditions are kept unaltered in the future the regular water quality can only be achieved if the proportion of native forest over sugar cane plantations rises substantially. This may not be economically feasible, the reason why the route to follow is to improve management practices to raise the value of *IWQ*0.

#### **4. Discussion**

The Environmental Protection Area of Uberaba River Basin (EPA-URB) and other similar conservation units exist to reconcile human occupation with the sustainable use of their natural resources, not to expel human populations. However, the activities and uses developed in these areas are subject to specific rules. This work exposed the need to manage properly the permanent preservation areas of three small catchments located in the EPA-URB to accomplish environmental sustainability.

The capacity of riparian buffers to retain particles and dissolved compounds from catchment uplands depend on the buffer width (50, 30 and 15 m) as well as on the types of land use or occupation and their management practices. This study exposed a stronger buffer capacity in the catchment where the riparian forests extend 50 m upwards from the stream margin, but even in this catchment the water quality is generally poor. A regular or good quality would require much wider strips and larger *NF/SC* ratios. In complement, better management practices could be implemented to prevent or at least reduce substantially the exports of sediment and nutrients towards the streams.

The Brazilian law defined the buffer width limits based on two scenarios: the Federal Law No. 4771 and the New Forest Code. In the first case, for watercourses up to 10 m wide the permanent preservation areas need to extend at least 30 m upwards from the stream margin considering the widest seasonal riverbank. In the second case, there are two rules: The transition rule takes into account the size of land property calculated as fiscal modules and creates a distance from the stream margin that goes from a minimum of 5 m to a maximum of 20 m, considering the regular river bank; the permanent rule defines 30 m as unique distance. By changing the reference from the largest riverbank (wet season) to the regular riverbank, the New Forest Code has decreased the legal riparian buffer width. A huge amount of scientific literature has reported the importance of riparian buffer width for water quality and ecological functions [3,11–14,18]. In the present study, it was suggested for a range of native forest/sugar cane occupational ratios (0.7–3) that the legal width should be at least 45 m, but preferably more, corroborating the studies of GAEMA [47]. Besides, efforts should be made to better understand the theory and the metrics of soil attributes and water quality in riparian forest ecosystems to develop ecological functions for these areas based on buffer width [19].

In general, the owner of a rural property has the legal right to use, enjoy, possess and dispose of it. However, this legal right is not applicable to permanent preservation areas included in the property. The permanent preservation area is a legal area. It is not an area for the socioeconomic use of a land owner. The permanent preservation areas are subject to a restriction of use imposed by the Brazilian Constitution, which aims to ensure a provisional ecosystem function, namely the provision of soil and water as resource. In this context, the management of permanent preservation areas is allowed solely when there is no local option of public and social interest, and the interventions are to be done with a low impact. The New Forest Code recommends a riparian buffer width that can keep fundamental ecosystem functions. This study suggested that this width should be increased to 45 m, at least. Besides the correct dimensioning of the riparian buffer width, a number of mitigation measures are ought of implementation to increase the *IWQ*<sup>0</sup> far above its current value (*IWQ*<sup>0</sup> = 30.26 = poor water quality). To become effective, the causes and paths of pollution should be assessed [48–51] and then the measures should be modeled in spatial decision support systems focused on water resources planning and management [52–56], evidenced and discussed by government agencies and public and private companies, and integrate public policies and environmental management plans [57]. From a broader standpoint, the specific widths along the drainage network should be reviewed, being defined as function of basin area, watercourse and the catchments' social and economic importance for public water supply.

The results illustrated in Figure 5b raised a striking question: What should be the area released to the agro system, in replacement of forest areas that have the aim of protecting the environment? The criticism we make to the New Forest Code in this case, is that the ratio of permanent preservation area over area used for agriculture or other anthropogenic activities, should be defined technically and not on the basis of political or socioeconomic convenience. This rationale is also valid for riparian buffer widths. The technical work of Kageyama, Cordeiro and Metzer [47] strongly suggested a minimum buffer strip width of 50 m to protect small streams from anthropogenic activities located

upward of the catchment hillsides. In the present study, the data collected on water are in favor of a buffer strip width even larger than that threshold if good quality water is aimed at the studied catchments. If anthropogenic activities are practiced in this protected area, the erosion and transport processes, followed by the silting and eutrophication of stream water, will be accentuated, a situation that will be defined as environmental damage according to Art. No. 3 of the Federal Law No. 6938/81 and handled through the "polluter-pays" principle [58].

The present study is corroborated in the literature, and was founded on principles of Environmental Law, namely the principles of precaution and prevention. The New Forest Code should take this and other scientific studies as example, and always be interpreted as *pro-nature*: in favor of Environment. Thus, if field data suggest that environmental vulnerability occurs within 50 m from the stream margin at the widest riverbank, it is clear that one should opt for riparian buffer solutions that result in greater environmental protection.

#### **5. Conclusions**

The role of riparian vegetation and forest cover in the control of stream water quality in anthropogenic catchments was investigated in this study. The analysis involved three headwater catchments characterized by increasing buffer strip widths, namely 15, 30 and 50 m widths, as well as increasing native forest to sugar cane ratios (*NF*/*SC*). The studied basins are located in the Environmental Protection Area of Uberaba River Basin (EPA-URB; state of Minas Gerais, Brazil). The water quality analysis aimed to evaluate a recent forest law (Law No. 12651/12) in this very important water resources and native forest (Cerrado Biome) conservation unit. A linear trend was defined between a specific water quality index (*IWQ*) and the combined protective effects of buffer with (*BW*) and *NF*/*SC* (*BW* × *NF*/*SC*). Presently, the quality of stream water in the three catchments is poor (*IWQ* < 36%). The linear trend allows estimating a regular water quality (36 ≤ *IWQ* ≤ 51%) if buffer widths were larger than 45 m, but only if the coverage by native forest increased substantially (e.g., duplicated) in the studied basins. Under the current land use (0.7 ≤ *NF*/*SC* ≤ 1.6) the regular water quality would be reached for buffer strip widths in the 90–205 m range. While keeping the current *BW* an *NF*/*SC* values, water quality could be improved if conservation practices were implemented in the sugar cane fields to reduce the export of sediments and nutrients towards the aquatic media. Overall, it was suggested in this study that the 30 m buffer strip width proposed in the New Forest Code, is barely capable of protecting water quality in the EPA-URB.

**Author Contributions:** Conceptualization, C.A.V. and T.C.T.P.; methodology, C.A.V. and T.C.T.P.; software, R.F.d.V.J. and J.P.M.; validation, C.A.V., L.F.S.F. and F.A.L.P.; formal analysis, C.A.V., F.A.L.P., T.C.T.P. and M.V.M.F.; investigation, C.A.V.; resources, T.C.T.P.; data curation, C.A.V., C.F.O. and T.C.T.P.; writing—original draft preparation, C.A.V.; writing—review and editing, F.A.L.P.; visualization, R.F.d.V.J.; supervision, T.C.T.P. and M.V.M.F.; project administration, T.C.T.P. and M.V.M.F.; funding acquisition, T.C.T.P., M.V.M.F. and R.F.d.V.J.

**Funding:** The present study was carried out within the framework of the Post Graduation Research Programme of Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES); Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq); Agência do Ministério da Ciência, Tecnologia, Inovações e Comunicações (MCTIC); and Land Use Policy Brazilian Group (POLUS). The author is affiliated with IFTM Renato Farias do Valle Júnior wishes to acknowledge the funding through the CNPq research scholarship Proc. 307921/2018-2. The authors integrated in the CITAB research center were financed by National Funds of the FCT–Portuguese Foundation for Science and Technology POCI-01-0145-FEDER-006958, under the project UID/AGR/04033/2019. The author integrated in the CQVR was funded by National Funds of the FCT–Portuguese Foundation for Science and Technology POCI-01-0145-FEDER-006958, under the project UID/QUI/00616/2019.

**Acknowledgments:** Hygor Evangelista Siqueira, Mauro Ferreira Machado and Renata Cristina Araújo Costa are acknowledged for fruitful discussions, sharing of information and mapping of the area.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
