Temporal and spatial variation of limnological variables and biomass of different macrophyte species in a Neotropical reservoir ( São Paulo – Brazil )

Aim: This study reports an investigation of limnological characteristics and aquatic macrophyte occurrence in a neotropical reservoir in order to assess the spatio-temporal variation of water and sediment variables and their influence on plant distribution. Methods: Macrophytes, water and sediment samples were collected from a Brazilian reservoir in different seasons from four main arms of the reservoir. In total sixteen water-sediment variables were analyzed including N:P ratio and Trophic State Index. The plants were collected using a quadrat sampling procedure and the dry weight per sample was measured. MANOVA was performed to evaluate spatial and temporal variation of environmental variables as well as seasonal biomass differences. To assess the relationship among environmental variables and macrophytes an ordination analysis (using Canonical Correspondence Analysis: CCA) was carried out. Results: The spatial and temporal variation of limnological variables generated a heterogeneous system which supports the presence of different species of macrophyte. pH, dissolved oxygen and sediment composition were important predictors of Polygonum lapathifolium occurrence while nutrients were associated with Eichhornia crassipes and Pistia stratiotes. Inorganic substances were related to biomass variation of Eichhornia azurea and Myriophyllum aquaticum. Conclusions: The spatial variation of the environmental variables has caused heterogeneity in the reservoir and it may support the occurrence of different species of macrophyte. Limnological variables highlighted in CCA are important to predict the species occurrence and their control in the study area.

State are eutrophic, and as a result an increase of the macrophyte biomass of mainly free-floating species has been reported (Cavenaghi et al., 2005;Martins et al., 2008).Although the benefits of macrophytes for aquatic systems are recognized (Kouki et al., 2009;Hassan et al., 2010;Vymazal, 2011), when growing in high abundance these plants interfere with the utilization of water resources (Pieterse and Murphy, 1993), blocking water flow, depleting oxygen in the water, and causing problems for hydropower generation and navigation.Due to these issues (both benefits and nuisance) the study of aquatic macrophyte communities and the factors which drive their distribution are important for water resource management of reservoir systems.This study aimed to assess the seasonal variation of water and sediment variables and their influence on macrophyte distribution in a Neotropical reservoir.

Introduction
Prediction of the occurrence and abundance of macrophytes species is a major issue in ecological studies of these plants, so establishing what factors determine their growth and distribution is essential.Macrophyte distribution is related to several environmental and anthropogenic factors (eg.climate, hydrology, geomorphology, nutrient availability, biological interactions, and human activities) (Lacoul and Freedman, 2006;Feldmann and Nõges, 2007;Hrivnák et al., 2009).
In neotropical regions abundant aquatic macrophyte biomass is a usual feature of reservoirs due to slow or near-static water velocity, sediment stability, water level control, shoreline development, high temperatures and often high nutrient availability (Bini et al., 2005).Most Brazilian reservoirs are shallow, with a dendritic morphology, which supports the development of coastal areas (Thomaz and Bini, 1998).The presence of these shallow areas, coupled with the generally favorable growing conditions of the tropics, as well as human activities (i.e.domestic and industrial effluent input, fertilizer use, dam construction) which promote macrophyte growth, all tend to favour abundant macrophyte production in neotropical reservoirs.
For macrophyte live-biomass analyses, the main plant species (i.e.most abundant species) in the littoral zone were collected within quadrat samples (area 0.25 m 2 ) (Westlake, 1965).The quadrat was randomly-located (at sites with healthy and mature plants; in triplicate) and aerial and submerged parts of the macrophytes were sampled (including roots in the emerged species) using sickle and hook.The collected biomass was washed with tap water to remove coarse material and dried (60 °C) to constant weight.The plants were weighed to obtain values of mean biomass (calculated per 1 m 2 ).The main species in ITU 1 were Eichhornia crassipes (Mart) Solms and Pistia stratiotes L., while in ITU 2 and ITU 3 Polygonum lapathifolium L. was predominant and in ITU 4 Eichhornia azurea (Sw.)Kunth and Myriophyllum aquaticum (Vell.)Verdc.
Temporal and spatial variations of water variables as well as time variation of the biomass were tested using MANOVA (significance level p<0.05) (Statistica® Software Version 10.0).The macrophyte environment dataset was analyzed using CCA ordination (CANOCO 4.5) to determine the relative importance of individual environmental factors as predictors of macrophyte distribution in Itupararanga Reservoir.

Results
Table 2 shows mean environmental data recorded for the four sampling stations and seasons.MANOVA results suggested a significant difference among sampling stations (spatial variation; p < 0.05) and maximum air temperatures in the area are between 15.0 °C and 24.0 °C and the mean wind speeds range from 1.5 m s -1 to 3.0 m s -1 (INMET, 2009(INMET, , 2010)).The main uses of the reservoir are public water supply and power generation for an aluminium company.Despite the existence of a permanent preservation area around the reservoir, agriculture (mainly strawberries, onions, potatoes, lettuce and tomato crops) is the predominant land use in the watershed, occupying 42% of the reservoir area (Garcia et al., 1999).
The sampling stations were located in four main arms of Itupararanga Reservoir (Figure 1) which are influenced by rivers running through different parts of Ibiuna city.These arms were coded as ITU 1 (23° 37' 28.3" S; 47° 13' 52.24" W) located near the headwater of the reservoir, and surrounded by small riparian forest, road and building in its sides; ITU 2 (23° 39' 24.65" S; 47° 18' 36.5"W) which has agriculture (lettuce crop) as the main land use in the sub-catchment and Campo Verde Stream is the main tributary of this arm; ITU 3 (23° 37' 42.53" S; 47° 17' 04.62" W) which is influenced by Ressaca Stream and has buildings and roads on its shore lines; and ITU 4 (23° 40' 10.72" S; 47° 21' 21.02" W) which is surrounded by buildings and fishing area, whilst its tributary (Paruru Stream) receives effluent input from a sewage treatment plant.

Sampling regime
Sampling was carried out in the winter, spring, summer and autumn periods, respectively in August and October 2009 and February and April 2010.Table 1 and Figure 2 show the main differences among the seasons and the hydrological pattern in Itupararanga Reservoir, respectively.In situ measurements for water temperature (T), pH, electrical conductivity (EC) and dissolved oxygen (DO) were obtained with a multiparameter probe (Yellow Spring Instruments YSI 556®).Total suspended solids (TSS), organic suspended solids (SSO), inorganic suspended solids (SSI) total phosphorus (TP), nitrite (NO 2 ) and nitrate (NO 3 ) were analyzed using standard methods predominance of SSI (maximum value: 77.1% in ITU 4 ) while the SSO ranged from 51% to 87.1% in other seasons.In general the TP and NO 2 concentrations were higher in spring except for TN.ITU 4 showed the highest nutrient concentration.
The N:P ratio was generally higher in autumn and the lowest values ranged between winter and spring and the Trophic Status Index (TSI) was on average mesotrophic for the four arms.
The CCA results (Figure 4) showed that ISS has significant relation with pH, TSS, NO 2 , IC and TP, whilst NO 2 is related with TSS, IC and TP; TP with IC, and TSS with IC and TP (p < 0.05).
The sediment analyses indicated significant spatial and temporal variation of PTS (p = 0.01) with higher values in the spring (maximum concentration: 3.0 µg.g -1 in ITU 3 ).The highest as well as among periods (temporal variation; p < 0.01).In general the main differences among the arms (spatial variation) were recorded for pH, DO, SSI, TP, T, OMS and TPS; the variables responsible for seasonal differences (temporal variation) were pH, T, NO 3 , TOC, OMS, TPS, TNS (Table 3).
In relation to the thermal structure ITU 3 was stratified in the summer and micro-stratifications occurred in the other periods (Figure 3).The highest water temperature (27-29 °C) and pH (7.67-9.8)occurred in the summer.DO concentrations were generally high in all sampling stations mainly in the winter (6 mg.L -1 to 8 mg.L -1 ).Electrical conductivity ranged from 54 µS.cm -1 to 81 µS.cm -1 and the highest values were recorded mainly in spring, as well as the SST concentrations (9.7-57 mg.L -1 ).Moreover in the same period there was a percentages of OMS were recorded in both ITU 2 and ITU 3 while TN showed high concentration in ITU 2 (Table 2).From the CCA we found significant correlation (p < 0.05) between OMS and DO; OMS and TP; OMS and TPS; OMS and TNS as well as between TNS and NO 2 and TNS and TPS. Figure 5 shows the results of sediment granulometric analyses.Silt was the predominant fraction in all sampling stations and seasons (90% to 60.8%), while gravel occurred mainly after rainfall but in small proportion.The variation in Table 2. Mean data of environmental variables in four arms of Itupararanga Reservoir.EC: Electrical Conductivity (µS.cm -1 ); DO: Dissolved Oxygen (mg.L -1 ); TP: Total Phosphorus (mg.L -1 ); TN; Total nitrogen (mg.L -1 ); NO 2 : Nitrite (µg.L -1 ); TSS: Total Dissolved Solids (mg.L -1 ); ISS: Inorganic Solids (mg.L -1 ); TOC: Total Organic Carbon (mg.L -1 ); IC: Inorganic Carbon (mg.L -1 ); TN:TP: Total Nitrogen and Total Phosphorus molecular ratio; TSI: Trophic Status Index; MOS: Organic Matter in Sediment (%); TPS: Total Phosphorus in Sediment (µg.g -1 ); TNS: Total Nitrogen in Sediment (µg.g -1 ).substrate fractions did not show seasonal significant differences.
Macrophyte biomass varied seasonally as showed by MANOVA (Table 4).The highest biomass of Polygonum lapathifolium was recorded in ITU 2 in the winter and in the autumn (966 gDW.m 2 and 917 gDW.m 2 respectively) while Eichhornia crassipes and Pistia stratiotes showed high biomass values in ITU 1 in the spring and summer, respectively.In ITU 4 the stands of Eichhornia azurea and Myriophyllum aquaticum were senescent in both summer and autumn.
The CCA analysis (total variance: 90%) suggests that ITU 1 sampling station was influenced by TP and TOC in all periods while DO, pH, OMS, TPS and TNS were related with ITU 2 and ITU 3 .TOC was associated with ITU 4 in winter and with NO 2 , TSS and ISS in the spring.So the species distribution was correlated with these variables (Figure 4) which explained almost 63% of data variability on the first axis of the CCA (eigenvalue: 0.89).The most important variables for Factor 1 were TP, DO, pH, MOS, TPS, NTS, P. lapathipholium, E. crassipes and P. stratiotes.

Discussion
Water temperature followed the normal seasonal pattern and DO concentrations were highest in the summer due probably to phytoplankton activity.CE, TSS, ISS, IC and TP are likely to be related to the rainfall in spring (76 mm) while high pH values may be associated with high photosynthesis rates in summer, which is usual in tropical reservoirs.Both TOC and IC showed lowest concentrations when the pH values were relatively neutral.Probably the tributaries and runoff provide a constant TN load to the reservoir since this variable did not show temporal variation in either water or sediment.Our TSI results are in agreement with Pedrazzi et al. (2013) although the differences between methods used.Previous research concluded the seasons and rainfall pattern are important to predict the heterogeneity of Itupararanga Reservoir (Pedrazzi et al., 2013).
Granulometric analysis showed the prevalence of silt in sediment and in rainy periods (mainly in the spring) there was increased gravel occurrence which may be from tributary drainage, due to high flows (see Table 1).Sediment variables (mainly OMS and TPS) were the main responsible for spatio-temporal heterogeneity in Itupararanga Reservoir suggesting the influence of different land covers in the arms (see item 2.1).
The patterns of aquatic macrophyte cover are strongly dependent on the spatial and temporal heterogeneity of aquatic ecosystems (Jeffries, 2008).This statement corroborates with the results given spatial variation of limnological variables for aquatic macrophytes through the CCA analysis.The results of the Canonical Correspondence highlighted the relationship of variables responsible for spatial heterogeneity and macrophyte species.
E. crassipes and P. stratiotes were influenced by TP and TOC in Itupararanga Reservoir.It is known these species are associated with high nutrient concentration (Henry-Silva et al., 2008) and are abundant in eutrophic ecosystems: ITU 1 was the sampling station with eutrophic conditions for most of the study period.Although not clear in CCA analysis, the maximum P. stratiotes biomass coincided with the highest TN concentration and this species growth is influenced by nitrogen concentration (Henry-Silva et al., 2008).
The senescence of P. stratiotes in the spring may be related to the competition process with E.
here, which suggested a substantial amount of spatial and seasonal variation of water and sediment variables, and macrophyte occurrence.So we may point out the macrophyte cover is related mainly to the spatial habitat heterogeneity of Itupararanga Reservoir.We suppose the influence of different land covers and streams (tributaries) running through different areas were responsible to generate the limnological differences among sampling stations (Table 3).In this regard different macrophyte species grew on the arms studied.
Similarly Pozo et al. (2011) have concluded that emergent and floating macrophytes composition is best explained by land use variables within the catchment.Due to the proximity of stream mouths to the sites with macrophyte biomass we propose too that the fluvial activity has an important role on biomass development.This fact is relevant when comparing the occurrence of the same macrophyte species in the whole river basin (Tietê River Basin: e.g.Cavenaghi et al., 2005;Martins et al., 2008).In summary, variables like pH, DO, TOC, IC, NO 2 , ISS, TSS, TPS, TNS and OMS were more important to predict the macrophyte occurrence and may be used as a tool for water management.The variables highlighted here may be useful to control macrophyte excessive growth since the nutrients increasing is a problem in Brazilians reservoirs mainly in Tietê River Basin.In this regard we suggest that P. lapathifolium generates different redox conditions influencing phosphorus release from the sediment.E. crassipes may be considered as indicating high trophic status while E. azurea and M. aquaticum are indicative of inorganic compounds in the water.Moreover P. lapathifolium and E. crassipes presence may be also associated with different land uses.The heterogeneity of the ecosystem due to spatial variation of environmental variables may be more important than seasonal variation to predict the occurrence of different macrophyte species.The composition of the species did not range with the temporal variation of the environmental variables.However water and sediment variables were important to predict biomass development and although land use data were not available we supposed they have an important role in macrophyte distribution.
This study has demonstrated in general way the factors driving the macrophytes distribution in Itupararanga Reservoir.Our results may support future researches in similar environments and to contribute to the water resources management.
crassipes during this period (P.stratiotes is known to be a fairly weak competitor with other tropical free-floating macrophytes: e.g.Agami and Reddy, 1990).P. stratiotes was senescent when the E. crassipes biomass increased 56% approximately.E. crassipes specimens normally have advantages due to their intensive growth, more efficient use of solar radiation and higher capacity of absorbing nutrients when they are abundant (Agami and Reddy, 1990;Henry-Silva et al., 2008).
The CCA species-environmental analyses showed a positive relationship between P. lapathifolium (in both ITU 2 and ITU 3 ) and pH and DO.Higher pH values were recorded in sampling stations with P. lapathifolium presence.In addition TP was inversely correlated with that species while the results suggest that the sediment is an important factor for P. lapathifolium development and probably drives the distribution of this plant.According to Takamura et al. (2003) Polygonaceae are important to restrain the P loadings from the sediment due to high DO concentration in the stands with these plants, even in the bottom as in Itupararanga Reservoir (from 6.5 to 8.94).Both ITU 2 and ITU 3 have shown mesotrophic condition in almost all seasons and had relatively high TPS concentration when the biomass of P. lapathifolium was high.
In ITU 4 M. aquaticum was correlated with solids and NO 2 while E. azurea and TOC were associated with the presence of this plant.Interestingly E. azurea was absent when the pH and SST concentrations were high, as also noted by Bini et al. (1999) in another Brazilian reservoir (Itaipu).On the other hand M. aquaticum occurrence coincided with the highest concentrations of solids and with the SSI predominance period.Both inorganic compounds and probably electrical conductivity appear to be predictors of Myriophyllum presence (see also Lacoul and Freedman, 2006).
The importance of anthropogenic activities for predicting macrophyte presence should be highlighted.For example, Cheruvelil and Soranno (2008) have recorded a link between agricultural activities and emergent macrophyte presence, as well as between road density and floating macrophyte cover.These findings support the results presented here since P. lapathifolium was present in sampling stations surrounded by rural activities while E. crassipes and P. stratiotes were in the headwater of the reservoir, near to the city and roads.Many species of Polygonaceae are recognized as ruderal species (R-strategists: Grime, 1979), which establish soon after disturbance and they are tolerant of

Figure 1 .
Figure 1.Location of the study area in Brazil, São Paulo State, including a map of the Itupararanga Reservoir with the sampling stations arrowed.

Figure 5 .
Figure 5. Granulometric fractions in different arms and seasons in Itupararanga Reservoir.
*Data provided by the dam operator; **meters Above Mean Sea Level; ***Seven days before the sampling.

Table 3 .
Main variables responsible for spatio-temporal variation in Itupararanga Reservoir.Results from MANOVA (significance level p < 0.05).
*The macrophyte stand was senescent.