Water quality changes in floodplain lakes due to the Amazon River flood pulse : Lago Grande de Curuaí ( Pará

Assurance of water quality for human consumption is essential for public health policies. In the Amazon floodplain, the seasonal water level variation causes periodic flooding of marginal areas that are usually used for settlements, agriculture and livestock. Therefore, the exchange of materials between the terrestrial and aquatic ecosystem affects the proportion of suspended and dissolved components in water and its physical-chemical characteristics, and consequently the quality of the water used by local people. Following this approach, the aim of this study is to evaluate changes in water quality in Lago Grande de Curuaí floodplain, Óbidos, Pará in response to the flood pulse, during one hydrological year from 2003 to 2004, based on water use classes (according to National Water Agency 357/2005 resolution) using chlorophyll-a and dissolved oxygen concentration as parameters and the eutrophication index. Ordinary kriging was applied to interpolate chlorophyll-a and dissolved oxygen and to predict values at non sampled locations. Each location was then classified according to water use acceptable parameters and to Carlson Trophic State Index modified by Toledo to map lake water classes and trophic status. The result showed that Lago Grande de Curuaí floodplain is a supereutrophic system, with levels of dissolved oxygen and chlorophyll-a not suitable for human supply during the receding water phase. These areas are located near the riverine communities, which can cause health problems due to the presence of potentially toxic algae. Therefore, monitoring water quality in Amazon lakes is essential to ensure the availability has appropriate quality for human and animal supplies.


Introduction
Assurance of water quality for human consumption is essential for public health policies.The increasing world population requires more water and therefore its quality and quantity provision is at risk due to constant degradation caused by irrigation, sewage disposal, industrial pollution, among other factors (WHO, 1996).Using aquatic environments for water supply, power generation, irrigation, navigation, aquaculture and recreation requires proper maintenance and monitoring in order to "ensure the availability of good quality water to current and future generations for their purposes" according to Law 9433 of January 8, 1997 of the Brazilian National Policy on Water Resources (ANA, 2008).
Water quality monitoring for public supply is based on measuring microbiological, physical and chemical constituents potentially harmful to human health according to standards established by relevant authorities.In Brazil, the CONAMA 357/2005 resolution defined five classes of freshwater quality standard required for their predominant use (Table 1) (ANA, 2008).For each class, the maximum concentration of organic and inorganic material allowed in water and also its physical and chemical properties are defined.
In spite of the clear definition of the parameters used to assess water quality, this quality is easily affected by weather, solar radiation, hydrology, and human activities (Hooda et al., 2000;Tundisi et al., 2004;Whitehead et al., 2009).The dynamic nature of the aquatic systems makes it difficult to distinguish between natural variability and anthropogenic influences (Chipps et al., 2006).This distinction is even more difficult in the Amazon floodplain lakes where the flood pulse controls the ecosystem dynamics.
The Amazon River floodplain, also known as "várzea" is one of the richest wetland systems in the Brazilian Amazon (Goulding et al., 1996;Junk et al., 2000) and has a population of approximately 1.5 million inhabitants, corresponding to 54% of the rural population of Amazonas and Pará states (IBGE, 2000).Public policies carried out in the 1970s aiming to integrate the Amazon region into southeastern Brazil attracted large number of immigrants who were engaged in huge development projects (hydroelectric power plants and highway constructions, mineral exploitation and cattle ranching).These policies changed the use of floodplain resources, originally related to subsistence fishing and small-scale agriculture.In 2008, the total livestock in the lower Amazon basin (from Parintins to Almeirim municipalities) was estimated at 1,390,740 heads, 60% of which were confined to the floodplain and in deforested areas near the Amazon River bank (Barbarisi, 2010).
The seasonal water level variation controls the life in the "várzea".The annual and monomodal variation of the Amazon River water level, is mainly caused by the high abundance and seasonal rain in the Andin region.This leads to an annual average water level amplitude that can reach up to 10 m, which corresponds to 230 days of flooding period (Junk, 1997).This flood pulse causes periodic flooding of marginal areas that are usually used for settlements, agriculture and livestock.The conversion of forest areas into pasture and agricultural areas either in the floodplain or in the Terra Firme forest (upland forest) affects the water flow and the exchange of materials between the terrestrial and aquatic ecosystem, changing the proportion of suspended and dissolved components in water and their physical-chemical characteristics (Melack and Forsberg, 2001;Tundisi et al., 2002).Novo et al. (2006) demonstrated that there is a six-month delay between the maximum water level and the maximum average of chlorophyll-a concentration in floodplain lakes between Parintins and Santarém, i.e. the phytoplankton peak production is achieved when the Amazon pulse recedes and the lakes are enriched by dissolved nutrients but less turbid water due to the sedimentation process.High chlorophyll-a concentrations, above normal levels, were also observed (Barbosa, 2005) suggesting that these lakes might be under severe eutrophication related to cattle ranching and agriculture (McGrath et al., 2007).
Aquatic system eutrophication is caused by increases in nutrient concentration (mainly phosphorus and nitrogen compounds) and/or organic matter in an aquatic ecosystem.This increase can be both natural and human induced (Moss 1988;Wetzel 2001).Various studies have already addressed the problems related to aquatic system eutrophication which impairs its ability to supply both human needs and maintenance of aquatic life (Espíndola et al., 2000;Sampaio et al., 2002;Sotero-Santos et al., 2008;Garcia et al., 2010).Moreover, this phenomenon can cause major functional and compositional shifts in ecosystems (Scheffer et al., 2001;Schindler, 2006) and the increase in diseases in humans and wildlife (Johnson et al., 2010).Furthermore, eutrophication can cause excessive growth of algae which can result in red tide blooms in oceans, and cyanobacterial blooms in freshwater (Landsberg 2002;GEOHAB 2005;Heisler et al., 2008;Richlen et al., 2010).The occurrence of algae and cyanobacteria in water can alter its color, odor and taste and also be a potential source of contamination because of the toxins derived from cyanobacterias.Therefore, the water may become unfeasible for human and animal supply (Pitois et al., 2000;Mankiewicz et al., 2003).
Water quality monitoring of Amazon floodplain lakes is vital for rural inhabitants, locally called "ribeirinhos", who use the same water as flood cattle and buffalo pasturelands.In spite of the huge water supply in the Amazon, 55.7% of households are not supplied with indoor plumbing, 32.4% of the total water volume distributed is untreated and 92.9% of the municipalities do not have a sewage system (IBGE, 2002).The only prevention provided by the local authorities is the use of a chlorine compound added to the drinking water (Moura, 2007;Renó, 2010).
Taking those facts into account, the aim of this paper is to characterize the water quality of the Lago Grande de Curuaí floodplain-lake system located at Óbidos, Pará over one hydrological year.The parameters used in the characterization were chlorophyll-a and dissolved oxygen concentration, the eutrophication index and the water use classes (according to CONAMA 357/2005 resolution) (MMA, 2005).

Study area
The study area includes Lago Grande de Curuaí floodplain which is located along the Amazon River near Óbidos city (Brazil), 900 km upstream from the Atlantic Ocean (Figure 1).The Curuaí floodplain is formed by quaternary sandy sediments with more than 30 lakes interconnected with each other and permanently connected to the Amazon River by channels (Moreira-Turcq et al., 2004;Dosseto et al., 2006).Flooding extension can vary from 1,340 to 2,000 km² depending on the water level.The seasonal amplitude can vary from 5 to 7 m and interannual variability can be up to 2 m (Figure 2) (Barbosa, 2005).A census conducted in 1996 by Iara Project counted 11,372 residents divided into 33 communities and 1,600 houses (Isaac et al., 2003).Barbosa (2005) noted in 2003 that there were 20,000 inhabitants in 96 communities.Curuaí has a strong fishing (450t of fish was caught in 2003) and ranch tradition (Pinto et al., 2007) with 234,678 cattle heads according to the 2006 census (IBGE, 2006)..

Field sampling
Chlorophyll-a and dissolved oxygen data were collected at different hydrograph states (receding, low, rising and high water) in the Lago Grande de Curuaí floodplain (Figure 1).Dissolved oxygen was measured using the HORIBA-U10, which has a 0,01 mg/L resolution.Surface water samples were collected, filtered (Whatmann GF/C fiberglass filter (0.5 -0.7 mm)) and stored in silica gel at 0 °C until further analysis in a laboratory for chlorophyll-a determination using a spectrophotometer (Wetzel and Likens, 1991).
Information on water sampling and analyses protocols can be found in Barbosa (2005).The number of samples varied in different states (Table 2) as the open water extent changed from one state to other.There were two types of sampling station: 1) Complete sample: where both chlorophyll-a and dissolved oxygen were measured (magenta dots in Figure 1), and 2) Incomplete sample: whereas only dissolved oxygen was measured (yellow dots in Figure 1).

Spatial analysis
In order to characterize the spatial dynamic of water properties, spatial modeling methods were applied to in situ data so as to infer variable values at positions which were not sampled.The first step in the process was to carry out the variogram analysis which describes the spatial structure among samples for each variable in each state of the hydrograph.Based on the variogram parameters, in situ data was then spatially interpolated using the ordinary kriging method (Burrough and McDonnell, 1998).

Water quality characterization
Water quality was characterized in the different states of the hydrograph according to two approaches.In the first one, dissolved oxygen and chlorophyll-a spatialized data were classified according to CONAMA´s resolution standards (Table 3).
The dissolved oxygen and chlorophyll-a spatialized data were used to generate a water class map according to Table 3 taking into consideration the highest restriction i.e. if the oxygen concentration was higher than 6 mg/L (Class 1) but the chlorophyll-a was between 10 mg/L to 30 mg/L, then that region was classified as Class 2.
The second approach consisted of using the spatial distribution of chlorophyll-a to compute Carlson's Trophic State Index (TSI) modified by Toledo-Junior et al. (1983) as follows: Where, ln (chl) is neperian logarithm of chlorophyll-a concentration.
The TSI equation ( 1) was applied to the chlorophyll-a spatialized data and subsequently sliced (the spatialized data was divided into a number of intervals which reflected the range of TSI values) according to trophic level classification system described in Table 4.

Results
Chlorophyll-a concentrations varied throughout the hydrological year with the lowest average concentration during the low water period (8.16 mg/L) and the highest average during the receding water period (63.9 mg/L), and the minimum value recorded was 0.21 mg/L (flooding season) and the maximum was 337.5 mg/L (receding water period).In relation to dissolved oxygen concentrations, the higher average value was during the high water phase (7.30 mg/L) and the lower occurred at low water phase (4.68 mg/L), and the minimum concentration measured was 2.54 mg/L (high water period) and the maximum was 13.2 mg/L (receding water period).
Water quality for Curuaí floodplain lakes varied through the hydrograph states (Figure 3).During the flood period (rising water) Curuaí floodplain lakes could be used for human supply after simplified treatment (adding a chlorine compound into the water) (Classes 1 and 2) (Table 5).At the high water phase, the lake water can only be used for human supply after a conventional (flocculation, filtration, pH adjustment, disinfection (chlorination) and fluoridation) or advanced treatment (removal and / or inactivation of residual constituents after conventional treatment) (Classes 2 and 3).
However, the receding state is the most critical for water supply especially in the south communities.Because of the high chlorophyll-a concentration (the highest range  from 4,15 mg/L to 337,5 mg/L), the water is inappropriate for human, animal supply and even for fishing, recreation and any other activity, for southern communities.During this time of the year, the water can only be used for navigation and landscape design.Finally, in the low water phase the water can be used for human supply only after conventional or advanced treatment (Class 3).We can also observe some patches of Class 4 in the high and low water phase, probably derived from an algae bloom in the region which is used for buffaloes ranching, and therefore not appropriate for human or animal supply.Figure 4 presents the changes in the trophic state of Curuaí floodplain lakes during the hydrological year.A seasonal variation in the trophic level of the water body can be observed in response to the flood pulse.During the flood (rising water) the lake ranges from oligotrophic to mesotrophic, with some ultraoligotrophic regions.In the high water phase, the large amount of nutrients from the terrestrial environment causes an increase in primary productivity and the lake becomes supereutrophic and even hypereutrophic in some regions.During receding water, the lake becomes hypereutrophic, with high concentrations of chlorophyll-a all over the lake.In low water, the lake becomes supereutrophic in response to the high primary productivity.

Discussion
In the hydrological year of 2003 to 2004, Lago Grande de Curuaí floodplain was a supereutrophic system, with levels of dissolved oxygen and chlorophyll-a not suitable for human and animal supply during the receding water period, according to standards established by relevant authorities (MMA, 2005).
Chlorophyll-a and dissolved oxygen concentration have similar dynamics during the hydrological year.At the beginning of the flooding phase, the oxygenated waters from the Amazon River entered the floodplain enriching lake waters.At the same time, a large amount of nutrient input provides an environment for algae bloom in regions far from the turbidity plumes caused by large amounts of suspended inorganic particles.This dynamic can change or restrict the use of water for certain purposes throughout the hydrological year, and therefore affect local communities.The linkage between eutrophication of a water body and harmful algae bloom has been described in detail in the literature (Pitois et al., 2001;Anderson et al., 2002;Granéli et al., 2008;Heisler et al., 2008).The nutrient increase can lead to high levels of phytoplankton biomass, which in turn might decrease its assemblage diversity when the water body becomes more eutrophic, and ultimately will be   (Dokulil and Teubner, 2000).Cyanobacteria blooms can cause not only an unpleasant taste and smell in the water, but also may undergo stress releasing toxins (hepatic, neuro and dermatotoxins) (Mankiewwicz et al., 2003).Furthermore these toxins have been responsible for lethal, acute and chronic poisonings of wild/domestic animals and humans in different water bodies all over the world (Carmichael, 2001).Although it had never been described in the literature concerning the phytoplankton communities in Lago Grande de Curuai, a recent characterization in that region detected a 38% relative abundance of cyanobacteria, 38% of Bacillariophyceae and 24% of Chlorophyceae Moreover, Vieira et al. (2005) reported the first bloom of cyanobacteria in a water supply reservoir in the Amazon region in Belém, Pará State.The increase in these algae concentration in Amazon natural water bodies might be a problem for local communities that use this water for human and animal supply, fisheries, and recreation.Silva (2006) showed the relationship between water source for human supply (primarily from rivers and streams) and the prevalence of multiple intestinal parasites in Amazon floodplain communities.Toxic algae blooms are regular and recurrent in the Brazilian water bodies outside the Amazon region (Yunes et al., 2003;Silva et al., 2007;Sotero-Santos et al., 2008;Chaves et al., 2009;Costa et al., 2009;Yunes, 2009), and had already caused the death of 70 chronic renal patients in a hemodialysis clinic in the Caruaru municipality in the State of Pernambuco in 1996 (Azevedo et al., 2002).The incidence of these toxins in reservoirs responsible for the production and distribution of potable water made water agencies around the world adopt improved purification for providing safe drinking water (Hitzfeld et al., 2000;Schmidt et al., 2008).
Notwithstanding the limited (one hydrological year) and the large inter-annual variability of the hydrology forcing functions in the region, the results suggest that Lago Grande de Curuaí floodplain water properties are under severe eutrophication and may pose a threat not only to human health, but also to aquatic fauna and flora.It also points out the urgency of a temporal and spatial analysis of phytoplankton communities and water quality monitoring in the Amazon region due to the enormous dependence of the local communities to the natural water supply.

Figure 2 .
Figure 2. Water level variation of Lago Grande de Curuaí throughout 2003 and 2004 (studied hydrological year) and the minimum, maximum and average water level amplitude from 1993 to 2009.Black dotted boxes indicate the field sampling period where: A, rising water; B, high water; C, receding water; and D, low water.

Figure 3 .
Figure 3. Water classes during the four states of hydrograph and the local communities (black stars).
, during the receding water phase in 2009 (Casali et al., in press).Even after the second highest high water period over the last 50 years, in which various municipalities declared an emergency state, because they had already reached the maximum water level quota (CPRM, 2009), Lago Grande de Curuai presented high levels of chlorophyll-a concentrations (ranging from 47 mg/L to 133 mg/L) (Barbarisi, 2010) and a high relative abundance of cyanobacterias (Casali et al., in press).Besides that, Sá et al. (2010) noted an intense proliferation of the genera Anabaena and Microcystis in the left margin of the Tapajos River during a rising water period in 2007.

Figure 4 .
Figure 4. Trophic states during the four states of hydrograph and the local communities (black stars).

Table 1 .
Freshwater classes and acceptable uses.

Table 2 .
Number of sample points collected per parameter in different phases of the hydrograph.

Table 3 .
Water quality appropriate standards (maximum and minimum acceptable concentration values) according to water use classes by CONAMA resolution nº 357/2005.

Table 4 .
Trophic State Index classification according to Carlson's Index modified by Toledo-Junior et al. (1983).

Table 5 .
Water classes during the hydrograph and its acceptable uses according to CONAMA 357/2005 resolution.