Watershed ecosystem modeling of land-use impacts on water quality

Page 1

W

Aa

b

a

ARR1A

KLWLNSB

1

wugtnsssqitbeeiaT

r

0h

Ecological Modelling 270 (2013) 54– 63

Contents lists available at ScienceDirect

Ecological Modelling

jo ur nal home p ag e: www.elsev ier .com/ locate /eco lmodel

atershed ecosystem modeling of land-use impacts on water quality

yten Erola,1, Timothy O. Randhirb,∗

Department of Watershed Management, Süleyman Demirel University, Faculty of Forestry, Isparta, 32260 TurkeyDepartment of Environmental Conservation, University of Massachusetts, 160 Holdsworth Way, Holdsworth Hall, Amherst, MA, USA

r t i c l e i n f o

rticle history:eceived 31 May 2013eceived in revised form2 September 2013ccepted 13 September 2013

eywords:ake watershed

a b s t r a c t

Sustaining freshwater systems in the face of rapid changes in land use continue to be a major challenge inlake watersheds. Lake Egirdir is a major freshwater lake in Turkey and is significant in supporting severalecosystem services of the region. The nutrient loading from both point and nonpoint sources has reducedthe quality of the tributaries and the lake ecosystem. There is a need for comprehensive modeling ofstrategies to protect the lake and watershed ecosystems. This study uses a watershed systems modelingto assess hydrologic flows and nutrient loads in order to quantify effectiveness of management practices inreducing runoff and nonpoint source pollution. The study uses GIS to process spatial data and simulation

ater qualityand useutrient loadingimulation modelingMPs

modeling to assess hydrologic and contaminant processes at a watershed scale. Management strategiesare evaluated by quantifying the effect of targeted best management practices (BMPs) in urban, forestand agricultural land uses. The results show that urban and forest loss can affect the watershed andlake ecosystems. There is a need for spatially targeted policies to sustain the lake watershed. Land usemanagement through education and incentives through conservation practices can be used to reducenutrient and sediment loads in sensitive areas of the watershed.

. Introduction

As increasing human and climatic stressors impact fragile fresh-ater resources like lakes, evaluation of conservation strategiessing a systems-based framework is needed. Of the estimatedlobal water volume of 1360 m km3 (Cain and Gleick, 2005), lesshan 3% is fresh and of this only 0.014% is readily available. Theatural lakes of the world constitute more than 50% of this totalupply, and hundreds of millions of people depend on these waterources for drinking water and sanitation. Vast amounts of watertored in lakes and groundwater aquifers provide reliable, high-uality supplies of water in many countries. In spite of their critical

mportance, many lake systems around the world are approachingheir limits or reaching thresholds where their sustainability andeneficial uses are becoming severely diminished or lost (Dinart al., 1995). In addition, ecosystem services provided by thesecosystems are increasingly being threatened by land use changen watershed systems. There is a critical need to study the link-

ge between land uses and the water quality in lake watersheds.his study aims at studying land use impacts on lake ecosystems to

∗ Corresponding author. Tel.: +1 413 545 3969; fax: +1 413 545 4358.E-mail addresses: [email protected] (A. Erol), [email protected],

[email protected] (T.O. Randhir).1 Tel.: +90 246 211 3988; fax: +90 246 237 1810.

304-3800/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.ecolmodel.2013.09.005

© 2013 Elsevier B.V. All rights reserved.

identify opportunities for managing water quality at a watershedscale.

1.1. Background

Lake watersheds supply essential goods and services to sustainecosystems and livelihood of people of the region; for example, pri-mary productivity and inputs from watersheds support food websthat yield fish for commercial and recreational purposes; decom-position and biological uptake purifies water supplies by removingorganic materials and nutrients (Naiman et al., 1995; Stenseth et al.,2002). Lakes are sensitive to a wide array of factors that includechanges in the climate (Hauer et al., 1997; Meyer et al., 1999; Gibsonet al., 2000; Stenseth et al., 2002) and watershed land use. Landuse (Gibson et al., 2000) can increase nutrient loading (Boyntonet al., 1995; Harding and Perry, 1997), lower dissolved oxygen,and degrade water quality. Increases in aquatic primary produc-tivity through high nutrient input can accelerate eutrophicationin lakes (Mulholland et al., 1997; Murdoch et al., 2000; Stensethet al., 2002). Significant increases in nitrogen loading are observedas the watershed gets urbanized (Valiela et al., 1997; Stenseth et al.,2002), specifically with increase in impervious cover and popula-tion densities (CWP, 2003). Over-allocation and overuse of water

(Gleick, 2003) and land use changes result in loss of these ecosystemservices.

Future freshwater management depend on developing new sup-plies, implementing conservation options, foregoing certain water

Page 2

A. Erol, T.O. Randhir / Ecological Modelling 270 (2013) 54– 63 55

Runoff ET Infilt rati on Groundwa ter

Water quant ity Landus e

Watershed

Water quality

Sedim ent

Weathe r

Temperatur e Nitrogen Pho sph oru s

Precipita tion

Sc5a-5c

Sc1 Sc2

Sc3 Sc4

uacamwtlt2a

Gtqtsm

otaostwctac

2

tcwatIastraiS

Table 1Scenario used in simulation.

Scenario Description

Baseline (Sc0) Calibrated model with current state of the watershedSc1 Runoff reduction: 10% reduction urban, crop, and

pasture runoffSc2 %10 reduction in ET loss through species changeSc3 Nitrogen reduction: 10% reduction in point and

nonpoint sourcesSc4 Phosphorus reduction: 10% reduction in point and

nonpoint sourcesSc5a Land use urban: 10% increase (from forest land)Sc5b Land use forest loss: 10% decrease (to urban,

agriculture, pasture)

Fig. 1. Conceptual model.

ses, meeting water quality standards, and protecting naturalquatic ecosystems. Lake watersheds contain a multitude of spe-ialized subsystems such as wetlands, floodplains and groundwaterquifers that are linked together, thereby needing a systems-basedanagement (Hauer et al., 1997). There is a significant need foratershed conservation strategies to reduce nutrient loading and

o prevent degradation of ecological integrity of the lake. There areimited studies in evaluating watershed-wide processes that affecthe lake’s water quality (Atilgan et al., 2009; Gunes and Tuncsiper,009; Sener, 2010). There is also a need for an integrated systemspproach for assessment and policy.

This study fills these gaps in lake watershed management usingIS, simulation modeling and policy simulation to quantify effec-

iveness of best management practices (BMPs) to protect wateruality and sustain water flows. The results of the study can be usedo identify management principles and watershed-scale protectiontrategies, especially in identifying vulnerable areas and manage-ent practices for targeting education and incentive policies.The general objective of the study is to model land use impacts

n the watershed processes and to identify targeted strategieshrough the use of best management practices. Specific objectivesre: (i) to assess hydrologic flows, point, and nonpoint sourcesf nutrients in the watershed; (ii) to model impacts of water-hed changes on nutrient loading into the lake system, and (iii)o simulate watershed-wide strategies to manage nutrients in theatershed system. Hypotheses tested include: (i) watershed pro-

esses are significantly affected by land use of the watershed; (ii)he watershed land use has significant effect on lake nutrient loads;nd (iii) watershed-level, land use policies using BMPs can signifi-antly reduce nutrient levels in the lake.

. Materials and methods

The conceptual model (Fig. 1) presents the general informa-ion flow in the research and is represented by submodels -fouromponents of water quantity (runoff, ET, infiltration and ground-ater), three components of water quality (nitrogen, phosphorus

nd sediment), other characteristics of watershed (land use, soils,opography, flows) and weather (temperature and precipitation).n the model, specific policy variables (Table 1) target the man-gement of water flows and water quality in the lake watershedystem. Runoff and Evapotranspiration (ET) that affect water quan-ity are targeted using policy scenarios Sc1 and Sc2. Reduction in

unoff is tested through a reduction in runoff rates (Sc1) from urban,gricultural and pasture land uses. Scenario Sc2 aims at a reductionn ET loss through changes in plant species composition. Scenarioc3 targets a reduction in nitrogen loads from both point and

Sc5c Land use agriculture: 10% decrease (to urban,agriculture, pasture)

nonpoint sources, and Sc4 aims at phosphorus reduction frompoint and nonpoint sources. Scenarios Sc5a–Sc5c aim at simulat-ing increase in urban land use, a decrease in forestland and loss ofagricultural land to urban land use, respectively.

2.1. Study area

Lake Egirdir is located in the Republic of Turkey, in the Eurasianregion (Fig. 2). The lake is the fourth largest in Turkey, and the sec-ond largest freshwater lake in the country. The lake is situated in theprovince of Isparta, in the southwestern part of the Mediterraneanregion. The climate of the Lake Egirdir watershed constitutes atransition between the Mediterranean and Central Anatolian Con-tinental climates. As a result, winters are harsh and rainy, whereassummers are hot and partially dry. Northern and southern windsdominate the region for most of the year (Ugurlu et al., 1999). The62-year average precipitation is 581 mm, the average temperatureis 12 ◦C, and the average humidity is 61% in the region.

The surface area of Lake Egirdir is 46,800 ha, and the volumeof the lake is 327.6 km3, with an average depth of 7 m. About70–100 km3 of lake water is taken annually for different pur-poses through six pumps and one regulator operated by the StateHydraulic Works (Kesici and Kesici, 2006). Major inflows into thelake are from small tributaries and through lake interception. Thedischarge from Lake Egirdir flows into more than 20 chasms. Someof the water is lost through evaporation (Gunes, 2008). The annualamount of evaporation in the lake is around 0.572 km3 (Gunes,2008). The lake is mainly fed by precipitation that drains through37 streams and rivers.

The lake watershed includes five districts: one of them is thedistrict of Egirdir located downstream at lower elevations includ-ing the lake. Total watershed area is 302,000 ha (Municipality ofEgirdir, 2010; Basayigit and Dinc, 2010), which is fed primarily byprecipitation. The watershed consists of forest, agriculture, pasture,water, and urban land uses that cover 98,975 ha (32.8%), 92,330 ha(30.06%), 7529 (2.5%), 46,800 ha (15.5%), 56,366 ha (18.6%), respec-tively (Gunes, 2008). About 39.30% of the watershed is relatively flatin terms of elevation and is at an altitude of 950–1050 m (Basayigit,2002) located at the lake level. Soils of the watershed consist of LandCapability Classes (Kumbur and Kocak, 1998) that cover 37.18%(Classes I–IV) and 62.82% (classes VI–VIII) of the watershed land.Soils of the watershed are generally from coarse and medium coarsetextures, except for the rocky areas. Soil groups are called for-est soils and chestnut-brown soils in the soil classification system.Considering the soil taxonomy, soils of the lake watershed have axeric moisture regime in which soil moisture is susceptible to leach-

ing. Agricultural land in this soil type leaches nutrients dependingon the weather. The watershed’s average soil N is estimated at2520 mg/kg (Carreira et al., 1994) and soil P at 6 mg/kg (Henkinet al., 1994).

Page 3

56 A. Erol, T.O. Randhir / Ecological Modelling 270 (2013) 54– 63

girdir

wHad8Tis(w

lii1iiiorrrteu

Fig. 2. Lake E

The annual flow of surface and groundwater in Lake Egirdiratershed is 27.2 km3. According to the General Directorate of Stateydraulic Works (Turkey), the residence time changes between 2.5nd 3.3 years. The lake watershed’s outflow is used for irrigation,rinking water, and hydropower supplies. The average outflow is4.84 km3 while the total inflow is at 87.19 km3 (Gunes, 2008).he watershed has an extensive stream network and multiplenflows to the Lake, but the main streams (Pupa stream, Degirmentream and Akcay stream) drain major residential developmentsthe districts of Uluborlu, Senirkent, Yalvac and Gelendost) of theatershed.

The lake’s water quality is classified as a mesotrophic-eutrophicake (Gulle et al., 2009), which is categorized as poor and need-ng protection strategies. Historically, the water quality of the lakes good for drinking purposes (Timur et al., 1988; Bayrak et al.,991; Diler et al., 1997; Kazanci, 1999) In recent years, the lake

s increasingly being polluted because of land use changes andncreases in nutrient loads (Gulle et al., 2009). Shortage in drink-ng water is an increasing threat (Gunes, 2008). Esendal (2007)bserves that irrigation water returning to the lake is polluted andequires preventive measures. During recent years, the lake is expe-iencing decreasing water levels through siltation, and increasing

ates of agricultural pollution. Drinking water quality is at risk dueo eutrophication. In general, change in land use is of high concern,specially the conversion of forest or open land to agricultural andrban land uses.

Pre cip itation Evapotranspiratio n Sedim ent,

Nutrients

(N, P, C)

Forest and agric ult ural land,

urban / subu rban, and oth ers

Unsa turat ed zone

Shallow satura ted zone

Deep seep age

Runoff

Grou ndw

Fig. 3. Conceptual framework

Watershed.

2.2. Empirical modeling

We simulate the lake watershed processes using the BasinSimModel (Dai et al., 2000), which is a continuous time simulationmodel based on the Generalized Watershed Loading Functions(GWLF) (Haith and Shoemaker, 1987). The GWLF model providesthe ability to simulate runoff, sediment, and nutrient loadings (Nand P) from a watershed with variable-size source areas (e.g.,agricultural, forested, and developed land) (Dai et al., 2000). TheGeneralized Watershed Loading Functions (GWLF) model simu-lates dissolved and solid-phase nitrogen and phosphorus in streamflow (Fig. 3). Both surface runoff and groundwater sources are usedas inputs, as well as nutrient loads from point sources and on-site wastewater disposal (septic) systems. In addition, the modelsimulates stream flow, soil erosion, and sediment yield values.

The GWLF model simulates nonpoint sources using a distributedmodel of runoff, erosion and urban runoff, and a lumped parame-ter linear reservoir groundwater model (Fig. 2). Point sources areadded as constant mass loads. Water balances are computed usingdaily weather data. Daily values are summed to provide monthlyestimates of stream flow, sediment and nutrient fluxes. Nutrientloads are estimated using the following Eq. (1):

LDm = DPm + DRm + DGm + DSm and LSm = SPm + SRm + SUm (1)

where LDm is dissolved nutrient load, LSm is solid-phase nutri-ent load, DPm, DRm, DGm and DSm are point source, rural runoff,

Dissol ved nu trient s (N, P,

C etc ., including nutrients

from septic sy stems)

Point sou rce s

(N, P, C etc .)

ate r

Streamflo w

Outp ut: Water qu anti ty;

Water quali ty

of the watershed model.

Page 4

A. Erol, T.O. Randhir / Ecological M

Table 2Data sources used in the analysis.

Database Information sources

Climate Turkish State Meteorological Service (2010)ET cover coefficient Basayigit and Dinc (2010)Population Gunes (2008), Municipality of Egirdir District (2010)Population growth CIA (2011)Land use Municipality of Egirdir (2010), Basayigit and Dinc (2010)Lake evaporation Gunes (2008)Lake flow Gunes (2008)Nitrogen Gunes (2008), Carreira et al. (1994)Phosphorus Henkin et al. (1994)

Table 3Calibrated parameters in the simulation model.

Landuse CN Soil loss (KLSCP)

Forest 72.33 0.0371085Agriculture 81.95 0.040701Pasture 79.5 0.0001357Water 100 0Urban 87.45 0.019024

ParametersET-cover coefficient 0.85 Basayigit and Dinc (2010);

0.6375-OffseasonErosivity coefficient 0.3 – Season; 0.12 – Nonseason

gtrm(tda

uttnsiDtl(ns

Sener (2010).Water quantity is calibrated using lake level data in the water-

TB

Septic service 350,000 – Population of the watershed

roundwater and septic system dissolved nutrient loads, respec-ively, and SPm, SRm and SUm and are solid-phase point source,ural runoff and urban runoff nutrient loads, respectively, in month

(m = 1, 2, . . ., 12). The equations assume that the point sourcesgroundwater and septic system loads) are entirely dissolved, whilehe urban nutrient loads are all in the solid phase. The model pre-icts stream flow from precipitation, evapotranspiration, land uses,nd soil characteristics (Fig. 2).

In this study, hydrological information, pollution data, and landse in the watershed are used to develop the simulation model ando determine conservation strategies for reducing nutrient load intohe lake. Specific nutrients that are simulated include sediment,itrogen and phosphorus in surface runoff, groundwater, pointources and septic systems. Various inputs on watershed character-stics are derived using GIS and past studies in the region (Table 2).ata on watershed characteristics, climate and land use is input into

he simulation to quantify runoff, nutrients (N and P), and sedimentoads under baseline and policy scenarios. The calibrated modelbaseline) is used in evaluating the effects of land use changes on

utrient and sediment loading into the lake. The outcome of eachcenario (Table 1) is presented as changes in hydrologic process and

able 4aseline flows in the study watershed.

Precipitation (cm) ET (cm)

April 5.42 0.07

May 2.76 7.11

June 5.03 8.63

July 0.25 1.06

August 0.17 0.15

September 0.83 0.70

October 8.46 2.24

November 4.86 1.78

December 15.96 1.30

January 12.3 1.14

February 26.19 1.33

March 3.34 2.35

Year 85.57 27.87

odelling 270 (2013) 54– 63 57

nutrient loading compared to the baseline. These outputs are usedto compare scenarios for reducing nutrient loadings into the lake.

2.3. Data and calibration

The watershed has time series and spatial data, which is usedin model calibration and validation. The data for the study areaon land use, soil features, population, and pollution are obtainedfrom remote sensed data and spatial information from Global GIS(U.S. Geological Survey (USGS, 2007), previous studies in the water-shed (Basayigit, 2002; Gunes, 2008; Cimen and Kisi, 2009; Basayigitand Dinc, 2010), and government publications (Altunbas, 2008;Municipality of Egirdir, 2010). Meteorological data for the areais obtained from local authorities and the Turkish State Meteoro-logical Service Turkish State Meteorological Service (TSMS, 2010)(Table 2). The flows in the Kovada channel and lake fluctuations(Cimen and Kisi, 2009) were used to calibrate and validate the GWLFmodel.

Calibration: The data on the characteristics of the lake water-shed obtained from recent studies (Gunes, 2008; Cimen and Kisi,2009; Basayigit and Dinc, 2010), along with data from local author-ities are used to calibrate the model (Tables 3–5). The total area ofthe lake watershed is modeled for 302,000 ha, with a baseline pop-ulation of 350,000. The lake discharge into the Kovada channel isat 264.16 km3 and the ET (Evapotranspiration) cover coefficient is0.85 (offseason reduced by 25%). Erosivity is calibrated at 0.3 in thegrowing season and 0.12 for the non-season with evaporation fromthe lake at 572 km3. The average soil N is calibrated to 2520 mg/kgand soil P to 6 mg/kg. In addition, evapotranspiration is calibratedusing values in literature (Basayigit and Dinc, 2010). The popula-tion growth estimate of 1.235 (CIA, 2011) is used in modeling thetrend. The population estimate is used to estimate the amount ofnutrients from septic effluents in the GWLF model. The model cal-ibration uses observed data from monitoring water quantity andquality data (Table 2). Meteorological data from the Turkish StateMeteorological Service (TSMS, 2010) is used in creating weatherinputs. Data from previous studies in the region on the amountsof nitrogen (Gunes, 2008; Carreira et al., 1994) and phosphorus(Henkin et al., 1994) transported to Lake Egirdir is used in cali-brating the nutrient output of the model. Nutrient loading fromhuman activities, nutrients from animals in the lake watershed,and the amounts of nitrogen and phosphorus transported to thelake are used from Gunes (2008), soil characteristics of the lakewatershed and pollution assessment are from Basayigit (2002) and

shed (Gunes, 2008; Cimen and Kisi, 2009). Due to lack of gaugingdata in the watershed for the study period (April 2010–March

Groundwater (cm) Runoff (cm) Streamflow (cm)

1.61 0.88 2.492.05 0.43 2.480.57 0.78 1.350.17 0.04 0.210.05 0.02 0.070.01 0.13 0.140.00 1.36 1.370.00 0.95 0.952.18 6.43 8.614.83 2.81 7.639.29 12.18 21.477.47 0.52 7.99

28.24 26.52 54.75

Page 5

58 A. Erol, T.O. Randhir / Ecological Modelling 270 (2013) 54– 63

Table 5Changes in water budget under baseline and scenarios.

Scenario ET (cm) Groundwater (cm) Runoff (cm) Yield (cm)

Baseline 27.87 28.24 26.52 54.75Sc1 27.87 31.27 (+10.73) 23.18 (−12.59) 54.45 (−0.55)Sc2 26.88 (−3.55) 29.00 (+2.69) 26.52 55.52 (+1.41)Sc3 27.87 28.24 26.52 54.75Sc4 27.87 28.24 26.52 54.75Sc5a 27.87 28.03 (−0.74) 26.74 (+0.83) 54.77 (+0.03)Sc5b 27.87 28.04 (−0.57) 26.74 (+0.83) 54.77 (+0.03)Sc5c 27.87 28.07 (−0.60) 26.70 (+0.68) 54.77 (+0.03)

Significant numbers are highlighted. Figures in parenthesis are percent change from baseline levels.

Table 6Water quality impacts under baseline and scenarios.

Year Erosion (kt) Sediment (kt) Dissolved N (t) Total N (t) Dissolved P (t) Total P (t)

Baseline 1423.36 46.97 1377.79 1515.06 151.00 170.19Sc1 1423.36 46.97 1255.50 (−8.88) 1385.65 (−8.54) 138.49 (−8.29) 150.55 (11.5)Sc2 1423.36 46.97 1385.65 (+0.57) 1522.92 (+0.52) 151.30 (+0.20) 170.49 (+0.18)Sc3 1423.36 46.97 1269.00 (−7.90) 1406.28 (−7.18) 151.00 170.18 (−0.01)Sc4 1423.36 46.97 1377.79 1515.06 141.40 (−6.36) 160.59 (−5.64)Sc5a 1383.30 (−2.81) 45.65 (−2.81) 1374.10 (−0.27) 1509.93 (−0.34) 150.86 (−0.09) 171.94 (+1.02)Sc5b 1378.83 (−3.13) 45.50 (−3.13) 1401.61 (1.73) 1536.29 (+1.40) 153.31 (+1.53) 173.60 (+2.00)Sc5c 1351.40 (−5.06) 44.50 (−5.06) 1330.87 (−3.41) 1465.25 (−3.29) 146.88 (−2.73) 169.16 (−0.61)

S base

2sd

Dsw1dnrceaap

3

tagP

TW

S

ignificant numbers are highlighted. Figures in parenthesis are percent change from

011), the model is calibrated using aggregate values reported intudies in the watershed and past observation in Kovada channeliversions.

Validation: The total runoff estimate of 26.45 cm (Basayigit andinc, 2010) is used in validating our model and the final runoff of the

imulation model was at 26.52 cm (0.0026 percent difference). Forater quality, the model is validated with estimates of 1514 and

50.5 t/year of nitrogen and phosphorus (Gunes, 2008). The pre-icted estimates of the validated model are 1515.06 and 151.003 foritrogen and phosphorus (0.0007 and 0.0033 percent differences,espectively). Sediment data is not available for the watershed foralibration and model estimates are validated for the average lev-ls of sediment loss in the watershed. The erosivity coefficient wasssumed to vary with seasons. The calibrated and validated resultsre used as the baseline scenario, and are used to study differentolicy scenarios in the lake watershed.

. Results and discussions

The results of the baseline and policy scenarios are presented in

erms of their impacts on water quantity (ET, Groundwater, Runoffnd Water Yield) and quality (Erosion, Sediment, Dissolved Nitro-en (DN), Total Nitrogen (TN), Dissolved Phosphorus (DP), Totalhosphorus (TP)).

able 7ater quality changes in specific components (Sc1 to Sc4).

Parameter Baseline Sc1

GroundwaterDissolved N 289.94 321.07 (+10.74)

Total N 289.94 321.07 (+10.74)

Dissolved P 11.09 12.28 (+10.73)

Total P 11.09 12.28 (+10.73)

Point sourceDissolved N 579.6 579.6

Total N 579.6 579.6

Dissolved P 96 96

Total P 96 96

ignificant numbers are highlighted. Figures in parenthesis are percent change from base

line levels.

3.1. Baseline scenario (Sc0)

The baseline results represent the output of the calibratedmodel. The results of the baseline scenario (Table 4) show an aver-age annual precipitation, evapotranspiration, groundwater, runoffand stream flow of 85.57 cm, 27.87 cm, 28.24 cm, 26.52 cm and54.75 cm, respectively. The summer months (July and August) andthe beginning of autumn (September) have ET and precipitationlevels that are relatively low. From May to June, ET increaseswith precipitation. This indicates that water loss is higher duringearly spring, with high transpiration from vegetative cover. Higherrates of precipitation, stream flow, runoff, groundwater rates arein February, the highest among all months. Groundwater levelsare high in February, when precipitation is also high. The annualerosion rate is at 1423.36 kt and sediment yield is 46.97 kt. Dis-solved N is estimated at 1377.79 t, Total N at 1515.06 t, dissolved Pat 151.00 t, and Total P at 170.19 t. Runoff is at 26.52 cm with a lowrunoff-rainfall ratio of 0.31, representing a higher pervious coverin the watershed.

This baseline scenario (Sc0) is compared to the outcomes of

each policy scenario (Sc1–Sc4 and Sc5a–Sc5c) and is presented inTables 6–11. This is evaluated in terms of the impacts on waterquantity and quality, and specific water quality and land usechanges.

Sc2 Sc3 Sc4

297.81 (+2.71) 289.94 289.94297.81 (+2.71) 289.94 289.94

11.39 (+2.71) 11.09 11.0911.39 (+2.71) 11.09 11.09

579.6 521.64 (−10) 579.9579.6 521.64 (−10) 579.6

96 96 86.4 (−10)96 96 86.4 (−10)

line levels.

Page 6

A. Erol, T.O. Randhir / Ecological Modelling 270 (2013) 54– 63 59

Table 8Land use specific changes under scenarios (Sc1–Sc4).

Parameter Baseline Sc1 Sc2 Sc3 Sc4

ForestRunoff 10.75 10.75 10.75 10.75 10.75Erosion 7.11 7.11 7.11 7.11 7.11Dissolved N 27.67 27.67 27.67 24.91(−9.97) 27.67Total N 86.17 86.17 86.17 83.40 (−3.21) 86.17Dissolved P 1.01 1.01 1.01 1.01 1.01Total P 1.15 1.15 1.15 1.15 1.15

AgricultureRunoff 16.88 11.46 16.88 16.88 16.88Erosion 7.80 7.80 7.80 7.80 7.80Dissolved N 452.07 306.81 −32.13) 452.07 406.86 (−10) 452.07Total N 511.92 366.66 (−28.38) 511.92 466.71 (−8.83) 511.92Dissolved P 40.53 27.51 (−32.12) 40.53 40.53 40.53Total P 40.67 27.47 (−32.46) 40.67 40.67 40.67

PastureRunoff 12.62 9.01 (−28.61) 12.62 12.62 12.62Erosion 0.026 0.026 0.026 0.026 0.026Dissolved N 28.50 20.35 (−28.60) 28.50 25.65 (−10) 28.50Total N 28.52 20.36 (−28.61) 28.52 25.70 (−9.89) 28.52Dissolved P 2.38 1.70 (−28.57) 2.38 2.38 2.38Total P 2.38 1.70 (−28.57) 2.38 2.38 2.38

UrbanRunoff 22.87 14.38 (−37.12) 22.87 22.87 22.87Erosion 0 0 0 0 0Dissolved N 0 0 0 0 0Total N 18.91 11.78 (−37.70) 18.91 18.91 18.91Dissolved P 0 0 0 0 0

S base

3

3

i

TL

(

Total P 18.91 11.78 (−37.70)

ignificant numbers are highlighted. Figures in parenthesis are percent change from

.2. Policy scenarios

.2.1. Runoff policy (Sc1)This scenario evaluates the impacts of policy that reduces runoff

n a watershed. A 10% decrease in runoff rates is simulated for urban,

able 9and use specific changes (Sca–Sc5c).

Parameter Baseline Sc5a

ForestRunoff 10.75 10.75

Erosion 7.11 7.11

Dissolved N 27.67 26.10 (−5Total N 86.17 82.26 (−4Dissolved P 1.01 0.95 (−5Total P 1.15 1.08 (−6

AgricultureRunoff 16.88 16.88

Erosion 7.80 7.80

Dissolved N 452.07 452.07

Total N 511.92 511.92

Dissolved P 40.53 40.53

Total P 40.67 40.67

PastureRunoff 12.62 12.62

Erosion 0.026 0.026

Dissolved N 28.50 28.50

Total N 28.52 28.52

Dissolved P 2.38 2.38

Total P 2.38 2.38

UrbanRunoff 22.87 22.87

Erosion 0 0

Dissolved N 0 0

Total N 18.91 20.80 (+9.Dissolved P 0 0

Total P 18.91 20.80 (+9.

Significant numbers are highlighted. Figures in parenthesis are percent change from bas

18.91 18.91 18.91

line levels.

agriculture and pasture land uses. Sc1 decreased total runoff by

12.59% from the baseline, increased groundwater by 10.73%, anddecreased water yield by 0.55% (Table 6). On the water qualityfront, it decreased dissolved N by 8.88%, total N by 8.54%, dissolvedP by 8.29% and total P by 11.54%. A runoff reduction impacted

Sc5b Sc5c

10.75 10.757.11 1.11

.67) 24.91 (−9.97) 27.67

.54) 77.55 (−10) 86.17

.94) 0.91 (−9.99) 1.01

.09) 1.04 (−9.57) 1.15

16.88 16.887.80 7.80

468.22 (+3.57) 406.86 (−10)530.21 (+3.57) 460.73 (−10)

41.98 (+3.58) 36.48 (−9.99)42.13 (+3.59) 36.61 (−9.98)

12.62 12.620.03 0.03

40.99 (+43.82) 28.5041.02 (+43.83) 28.52

3.42 (+43.70) 2.383.42 (+43.70) 2.38

22.87 22.870 00 0

99) 20.01 (+5.82) 22.00 (+16.34)0 0

99) 20.01 (+5.82) 22.00 (+16.34)

eline levels)

Page 7

60 A. Erol, T.O. Randhir / Ecological Modelling 270 (2013) 54– 63

Table 10Water quality changes under land use scenarios (Sca–Sc5c).

Parameter Baseline Sc5a Sc5b Sc5c

GroundwaterDissolved N 289.94 287.83 (−0.73) 287.89 (−0.71) 288.23 −0.59)Total N 289.94 287.83 (−0.73) 287.89 (−0.71) 288.23 (−0.59)Dissolved P 11.09 11.01 (v0.72) 11.01 (−0.72) 11.02 (−0.63)Total P 11.09 11.01 (−0.72) 11.01 (−0.72) 11.02 (−0.63)

Point sourceDissolved N 579.6 579.6 579.6 579.6Total N 579.6 579.6 579.6 579.6Dissolved P 96 96 96 96Total P 96 96 96 96

S base

gbPwtflNPr2brbtq

TB

ignificant numbers are highlighted. Figures in parenthesis are percent change from

roundwater quality by increasing dissolved N in groundwatery 10.74%, dissolved P by 10.73%, total N by 10.74%, and total

by 10.74% of the baseline. Runoff reduction improved surfaceater quality, but had the potential to increase groundwater con-

amination (Table 7). A reduction in runoff rate impacted loadingrom specific land uses by decreasing agricultural nonpoint pol-ution (Table 8). In agricultural land, runoff decreased dissolved

by 32.13%, total N by 28.38%, dissolved P by 32.12%, and total by 32.46% of the baseline. In pasture land, a 10% reduction inunoff rate decreased pasture runoff by 28.61%, dissolved N by8.60%, and total N by 28.61%, dissolved P by 28.57% and total Py 28.57% of the baseline. In urban land, this scenario decreased

unoff by 37.12%, total N by 37.70% and total P by 37.70% fromaseline. The runoff rate scenario (Sc1) shows that reducing runoffhrough stormwater BMPs can improve surface water flows anduality.

able 11est management practices (BMPs) and their removal efficiencies.

BMPs by land use Function

ForestSkid trail erosion control Erosion control and water quantity an

Access road erosion control Safely logging road systems

Erosion control on landings Erosion control

Stream crossing erosion control Erosion and sedimentation

Filter strip sediment control Sediment controlRunoff reduction 50%

AgricultureAgricultural management practice Erosion reduction and reduction 50%N

Vegetative and tillage practices Erosion reduction 40–60%Sediment reduction 30–50%

Structural practices Pollutants reduction 50–75%

UrbanDetention ponds for urban runoff Pollutants reduction by up to 90%

Infiltration trenches for urban runoff Groundwater recharge by diverting 60annual runoff back into the soil

Porous pavement for urban runoff Groundwater recharge by diverting 60annual runoff back into the soilRunoff reduction 75%

Vegetative filter strips The velocity of storm water runoff red

Constructed wetlands Settlement of particulate pollutants

Road salt storage facility Keeping chemicals

Urban forestry The volume of surface water runoff reRunoff reduction 30–50%

line levels.

3.2.2. ET scenario (Sc2)In this scenario, ET is reduced by 10% to reflect changes

in plant composition and reduction in evaporation rates.Major plant species observed in the watershed are conifers(Juniperus exelsa) and broad leaved (Quercus aucheri). Othercommon species distributed along with them in the water-shed are Rosa canina L., Berberis crataegina, Euphorbia spp.,Juniperus oxycedrus, Prunus divaricata, Astragalus spp., Cratae-gus monogyna, Rhamnus rhodopeus, Quercus coccifera, Verbascumspp. (Karatepe, 2007). The ET rate in the lake watersheddepends on plant species composition and soil moisture. Monthlyactual evapotranspiration value of deciduous species ranged

from 9.2 mm to 159.4 mm and from 10.0 mm to 155.7 mmof the coppice stand. Mean annual evapotranspiration wasfound to be highest in deciduous species (773.3 mm) and itwas followed by coppice (720.8 mm) species (Özhan, 1982). By

Remarks

d quality Prevent erosion and provide to divert surface water runoffKittredge and Parker (1989).Non-erosive transfer of surface water runoff Kittredge andParker (1989).Prevent erosion and sediment Winer (2000).Prevent erosion and sedimentation due to the disturbanceof stabilized stream banks Kittredge and Parker (1989).Reduction environmental damage to water bodiesKittredge and Parker (1989), Winer (2000).

, 90%P Current pesticide use could be reduced by 40% that Preventcontamination of surface or groundwater Winer (2000).Prevent soil erosion and leaching of nutrients of pesticidesWiner (2000).Control sediment and pollutant runoff Winer (2000).

Settlement of urban sediment by 60–70%, total phosphorusby 40–50% 40–50% and total organic matter Schueler(1987), NJDEP (1981), MADEP (1985)

–90% of Soluble and particulate pollutants Schueler (1987), NJDEP(1981), MADEP (1985)

–90% of Soluble and fine particulate pollutants Schueler (1987),NJDEP (1981), MADEP (1985), Winer (2000)

uction Sediments and pollutants Schueler (1987), NJDEP (1981),MADEP (1985)Nutrients Schueler (1987), NJDEP (1981), MADEP (1985)Potential nonpoint source pollutant from entering surfacewater or from leaching down to groundwater Schueler(1987), NJDEP (1981), MADEP (1985)

duction Infiltration of stormwater runoff Schueler (1987), NJDEP(1981), MADEP (1985)

Page 8

gical M

cstiiaidbiaEq

3

brNgNdaei

3

swips

3

a

3iwiibtiimda(bt

3dt0(mldsa9d

A. Erol, T.O. Randhir / Ecolo

hanging composition, ET levels can be targeted as a policycenario to alter water budget. Juniper enhances evapotranspira-ion by canopy interception (Hester, 1996) and its compositionn the watershed could be altered to achieve change in ET. Sc2mpacted water flows by increasing groundwater levels by 2.69%nd water yield by 1.41% of baseline (Table 5). A decrease in ETncreased to a small extent dissolved N (0.57%), total N (0.52%),issolved P (0.20%), and total P (0.18%) in runoff compared to theaseline (Table 6). Scenario Sc2 impacted groundwater quality by

ncreasing dissolved N (2.71%), total N (2.71%), dissolved P (2.71%)nd total P (2.71%) from the baseline (Table 7). The reduction inT through landscape change can affect groundwater quantity anduality through changes in the throughfall and infiltration process.

.2.3. Nitrogen policy (Sc3)In this policy scenario, nitrogen loading is reduced by 10% from

oth point and nonpoint sources through appropriate BMPs. Aeduction in nitrogen loading reduced dissolved N (7.90%), total

(7.18%), and total P (0.01%) in the watershed (Table 6). A nitro-en load reduction improved water quality by decreasing dissolved

(10%) and total N (10%) from point sources (Table 7). A 9.97%ecrease in dissolved N and in total N by 3.21% of the baselinere observed in forest land under Sc3. Dissolved and total N lev-ls decreased by 10%, and 8.83% in agricultural and 10%, and 9.89%n pasture lands (Table 8).

.2.4. Phosphorus policy (Sc4)A phosphorus reduction policy scenario (Sc4) decreases dis-

olved P and total P by 6.36%, and 5.64%, respectively in theatershed through appropriate land use-specific BMPs. Changes

n phosphorus, ET, groundwater, runoff and water yield for Sc4 areresented in Tables 5 and 6. Phosphorus reduction did not haveignificant impact on land use specific loadings (Table 8).

.2.5. Land use policies (Sc5a–Sc5c)In these scenarios, land uses are changed by 10% for urban, forest

nd agricultural sectors.

.2.5.1. Urbanization policy (Sc5a). In this scenario, urban landncreases by 10% from forest land. Urbanization decreased ground-

ater by 0.74 from the baseline, increased runoff (0.83), andncreased water yield (0.037%) in the watershed (Table 5). Urban-zation decreased erosion by 2.81%, sediment by 2.81%, dissolved Ny 0.27%, total N by 0.34%, and dissolved P by 0.09%, but increasedotal P by 1.02% of the baseline (Table 6). Such a reduction in load-ng is not significant but can be a result of lower erosion rates inmpervious areas that have lower nutrient loads attached to sedi-

ent. Urbanization decreased dissolved N (5.67%), total N (4.54%),issolved P (5.94%), total P (6.09%) in forest land of the watershed,nd increased total N (9.99%) and total P (9.99%) in urbanized areasTable 9). The Sc5a scenario decreased dissolved N in groundwatery 0.73% of the baseline, total N by 0.73%, dissolved P by 0.72%, andotal P by 0.72% (Table 10).

.2.5.2. Forest loss reduction policy (Sc5b). This scenario simulates aecrease in forest land by 10% which is converted equally to agricul-ural, pasture and urban land. Forest loss decreased groundwater by.57%, increased runoff (0.83%), and increased water yield (0.037%)Table 5). This scenario decreased soil erosion by 3.13% and sedi-

ent by 3.13%. This decrease is because of urban transition. Forestoss resulted in increases in dissolved N (1.73%), total N (1.40%),issolved P (1.53%), and total P (2%) of the baseline in the water-

hed (Table 6). Scenario Sc5b decreased dissolved N and total Nnd dissolved P and total P in forest land by 9.97%, 10%, 9.99%, and.57%, respectively. Deforestation increased dissolved N, total N,issolved P, and total P in agriculture by 3.57%, 3.57%, 3.58%, and

odelling 270 (2013) 54– 63 61

3.59%, respectively, and 43.82%, 43.83%, 43.70%, and 43.70%, respec-tively, in pasture land. A Decrease in forestland increased total N(5.82%) and total P (5.82%) from urban sources (Table 9). Forest lossdecreased dissolved N (0.71%), total N (0.71%), dissolved P (0.72%),and total P (0.72%) in groundwater (Table 10).

3.2.5.3. Agricultural loss policy (Sc5c). In this scenario, agriculturalland decreased by 10% for urban use. Agricultural loss to urbanuse decreased groundwater (0.60%), increased runoff (0.68%) andincreased water yield (0.041%) (Table 5). Agricultural loss affectedwater quality through decreases in erosion (5.06%) and sediment(5.06%), dissolved N (3.41%), dissolved P (2.73), total N (3.29%)and total P (0.61%) of baseline levels (Table 6). Agricultural lossdecreased dissolved N (10%) and total N (10%), dissolved P (9.99%)and total P (and 9.98%) in agricultural land use, but increased totalN (16.34%) and total P (16.34%) in urban land use (Table 9). Agricul-tural loss impacted water quality through decreases in dissolved N(0.59%), total N (0.59%), dissolved P (0.63%) and total P (0.63%) ingroundwater (Table 10).

3.3. Comparison between policies

3.3.1. Water quantityReduction in runoff increases recharge into groundwater and

results in higher groundwater discharge into the lake. Similarly,the groundwater of the watershed increases with a reduction in ET,and resulted in a decrease in water yield. Changes in land use canimpact water flows in all scenarios, especially through a decrease ingroundwater, increase in runoff, and increase in water yield. Urban-ization and forest loss impacted water quantity through a decreasein groundwater and an increase in runoff.

3.3.2. Water qualityRunoff reduction decreased dissolved N, total N, dissolved P,

and total P in the runoff, but impacted groundwater water qualitythrough an increase in dissolved N, total N, dissolved P, and totalP. Change in the runoff decreased agricultural loading (dissolvedN, total N, dissolved P, and total P). Runoff reduction had a betterimpact than reducing ET. Runoff reduction decreased runoff, dis-solved N, total N, dissolved P, and total P in pasture land use, whileit decreased total N and total P in urban areas. Changes in ET lossthrough plant composition increased dissolved N, total N, dissolvedP, and total P, while increasing dissolved N, total N, dissolved P, andtotal P in groundwater. Nitrogen reduction impacted water qualitythrough decreases in both point and nonpoint sources of dissolvedN, total N, and total P. Nitrogen reduction policy decreased dis-solved N and total N from point sources. The nitrogen reductionscenario decreased dissolved N and total N in forest, agriculturaland pasture. Urbanization increased total P in surface water, butdecreased dissolved N, total N and dissolved P and total P in ground-water. The impacts of urban and agriculture are higher comparedto the impact of forest loss. Changes in agricultural land affectederosion, sediment, dissolved N, total N, dissolved P, and total P. Agri-cultural loss decreased dissolved N, total N, dissolved P, and totalP from agricultural land, but increased total N and total P in urbanland. Agricultural loss decreased dissolved N, total N, dissolved P,and total P in groundwater. Forest loss decreased erosion and sedi-ment, decreased dissolved N, total N, and dissolved P but increasedtotal P. Forest loss decreased dissolved N, total N, dissolved P, and

total P in forest, but increased dissolved N, total N, dissolved P, andtotal P in agriculture and pasture land, and increased total N andtotal P in the urban land. Forest loss decreased dissolved N, total N,dissolved P, and total P in groundwater.

Page 9

6 gical M

3

stiIt(ciapaibap

wlpuic

3

crpiasctebgstocipVrpIs

4

elaLaasuiao

2 A. Erol, T.O. Randhir / Ecolo

.4. Practical implications

The lake watershed is sensitive to land use change. The resultshow that urban land use affects groundwater that discharges intohe lake and increases stream flow in the watershed. This can alsompact lake water quality, especially sediment and nutrient loads.ncreases in sediment and nutrients in the lake can be reducedhrough BMPs and other incentives to reduce runoff in urban areasTable 11). A reduction in ET through the type of forest or crop coveran increase groundwater and water yield into the lake. Changesn forest compositions can be achieved through forest incentivesnd biodiversity planning at a watershed scale. Urbanization canlace severe stress on the lake watershed, which can be man-ged through land policies that encourage the adoption of BMPsn impervious areas. In the lake watershed, nutrient policies cane used to encourage use and loss reduction through educationnd incentives to adopt BMPs. In addition, zoning and mitigationlanning can be used along with economic policies.

In the watershed, forest loss can affect groundwater, runoff, andater quality. Given this risk, forest loss can be minimized through

and use policies, conservation education, and regulatory/incentiveolicies. In general, loss of agricultural land to support urban landse decreased groundwater and increased runoff. To minimize loss

n agricultural land, land protection, conservation incentives, andost sharing can be used.

.4.1. Best management policy recommendationsThe installation and use of best management practices (BMPs)

an be used to implement various scenarios of the study. Theesults of watershed analysis can determine the most effectiveractices in the watershed for runoff reduction, erosion and sed-

mentation control, and pollutant reduction in forest, agriculturalnd urban land areas (Table 11). BMPs for forest land include ero-ion control in vulnerable areas like skid trails, access roads, streamrossing, and landings. Filter strips can be used in riparian areaso reduce runoff and sediment movement. In agricultural areas,rosion control practices can reduce nutrient loading into waterodies. Reductions in pesticide use can be achieved through inte-rated pest management techniques to reduce contamination inurface and groundwater. Vegetation management and changes inillage practices can be used to reduce soil loss. Implementationf structures for manure storage, livestock fences, and cover cropsan be used to reduce the nutrient contamination of water bod-es. Urban BMPs include detention ponds, infiltration trenches, andorous pavements to reduce excess runoff in impervious areas.egetative filter strips and constructed wetlands are effective ineducing pollutants generated upstream. Urban forests can reduceeak flows and provide wildlife habitat functions in urban areas.

n the areas with winter road conditions, road salt storage andelective use can reduce the contamination of water bodies.

. Summary and conclusions

Human impacts on lake watersheds are increasing worldwide,specially in regions with intense pressure for water resourcesike the Lake Egirdir watershed. Lakes are a major source of waternd are very sensitive to changes in their contributing watershed.and use changes have a significant influence on water qualitynd hydrologic processes in the lake watershed system. There is

need for a watershed-wide assessment of runoff and non-pointource pollution into the lake ecosystem. This study assesses land

se impacts and identifies watershed-scale strategies to develop

ntegrated strategies for conservation. The watershed system isnalyzed using GIS and simulation modeling to quantify the effectsf conservation strategies to reduce nutrient loads, runoff, and

odelling 270 (2013) 54– 63

sediment yield into the lake. In the lake watershed, the major prob-lems resulting from land use are erosion, sedimentation, runoff, andpollutants. We quantify the usefulness of BMPs to protect waterflows and quality at the watershed scale.

Changes in land use have impacted water quantity in all sce-narios, especially in terms of decreasing groundwater, increasingrunoff, and increasing water yield. Urbanization and forest lossimpacted water flows through a decrease in groundwater and anincrease in runoff. The results show that urban land can have aneffect on the groundwater and stream flow. The increases in sedi-ment and nutrients in the lake can be reduced through BMPs andother incentives to reduce runoff in the urban areas of the water-shed. A reduction in ET increased groundwater and yields in thelake. Changes in forest compositions to reduce ET can be achievedthrough forest incentives and biodiversity planning.

Nutrient policies can be used to encourage reductions in nutri-ent loading and lower runoff through education and incentivesin BMPs. In addition, zoning and mitigation planning can be usedalong with economic policies. Forest loss can result in a decreasein groundwater and an increase in runoff, which can be minimizedthrough education and regulatory/incentive policies.

In summary, comprehensive assessment and policies need awatershed-wide systems assessment to protect the lake system.The impacts of urbanization can be mitigated through incen-tives and landscape planning. Nutrient management for specificland uses is important to prevent eutrophication of the lakesystem. Runoff can be minimized through strategic placementof BMPs in the watershed system. To enable future researchand assessment, there is a need for long-term monitoring andmulti-disciplinary research in the watershed. Community partic-ipation, interdisciplinary research, and conservation policy at thewatershed scale are important for the sustainability of the lakesystem.

References

Altunbas, S., 2008. TC Egirdir Golu Yonetim Plani 2008–2012, ISBN 978-975-585-956-9. Isparta Valiligi Il Cevre ve Orman Mudurlugu Gol Yonetim Plani Dizisi 1,Isparta.

Atilgan, A., Coskan, A., Isler, E., Oz, H., 2009. Definition of the amounts of nitrogenand phosphorus related to agricultural pollution elements in Egirdir lake. AsianJournal of Chemistry 21 (4), 3107–3116.

Basayigit, L., 2002. Egirdir golu havzasinda erozyon riskinin saptanmasi uzerinearastirmalar. Cukurova University, Adana-Turkey (Doctoral dissertation).

Basayigit, L., Dinc, U., 2010. Prediction of soil loss in lake watershed using GIS: a casestudy of Egirdir Lake, Turkey. Journal of Natural and Environmental Sciences 1(1), 1–11.

Bayrak, M., Karakoyun, S., Menengic, M., Vatansever, H., 1991. Egirdir golu ekolojisive ekonomik su urunlerinin incelenmesi projesi (Project report). In: TubitakDEB-CAG 51. Egirdir Su Urunleri Arastirma Enst, Isparta, pp. 44.

Boynton, W.R., Garger, J.H., Summers, R., Kemp, W.M., 1995. Inputs, transformations,and transport of N and P in Chesapeake Bay and selected tributaries. Estuaries18, 285–314.

Cain, N.L., Gleick, P.H., 2005. Real numbers: the global water crisis. Issues in Scienceand Technology. Summer 1 (4).

Carreira, J.A., Niell, F.X., Lajtha, K., 1994. Soil nitrogen availability and nitrificationin Mediterranean shrublands of varying fire history and successional stage. Bio-chemistry 26, 189–209.

CIA (Central Intelligence Agency), 2011. World Factbook. Turkey population,http://www.cia.gov

Cimen, M., Kisi, O., 2009. Comparison of two different data-driven techniquesin modeling lake level fluctuations in Turkey. Journal of Hydrology 378,253–262.

CWP (Center for Watershed Protection), 2003. Impacts of impervious cover onaquatic systems. Water Protection Research Monograph 1, http://www.cwp.org

Dai, T., Richard, L.W., Christensen, T.R.L., Lewis, E.A., 2000. BasinSim 1.0: A windows-based watershed modeling package (Report No. 362). Special report in AppliedMarine Science and Ocean Engineering.

Diler, O., Altun, S., Atay, R., 1997. Egirdir golu su kalitesi fiziksel, kimyasal, mikro-biyolojik parametreleri. SDU Egirdir Su Urun. Fa. Dergisi 5, 1–34.

Dinar, A., Seidl, P., Olem, H., Jorden, V., Dada, A., Johnson, R., 1995. Restoring andprotecting the World’s lakes and reservoirs. World Bank Technical Paper N. 289.The World Bank: 85–90, Washington DC.

Esendal H., Modelling of seasonal lake level fluctuations of lake egirdir using fuzzylogic. SDU Fen Bilimleri Enstitusu, Isparta, MSc Thesis, 2007, 107 pp.

Page 10

gical M

G

G

G

G

G

H

H

H

H

H

K

K

K

K

K

M

A. Erol, T.O. Randhir / Ecolo

ibson, G., Carlson, R., Simpson, J., Smeltzer, E., Gerritson, J., Chapra, S., Heiskary, S.,Jones, J., Kennedy, R., 2000. Nutrient criteria-Technical Guidance Manual: Lakesand Reservoirs, http://www.epa.gov

leick, P.H., 2003. Global freshwater resources: soft-path solutions for the 21stcentury. Science 302, 1524–1528.

ulle, I., Zeki Yildirim, M., Kucuk, F., 2009. Limnological history of lake Egirdir(Turkey): from 1950s to the present. Natura Montenegrina, Podgorica 7 (2),115–128.

unes, K., 2008. Point and nonpoint sources of nutrients to lakes- ecotechnologicalmeasures and mitigation methodologies-case study. Ecological Engineering 34,116–126.

unes, K., Tuncsiper, B., 2009. A serially connected sand filtration and constructedwetland system for small community wastewater treatment. Ecological Engi-neering 35, 1208–1215.

aith, D.A., Shoemaker, L.L., 1987. Generalized watershed loading functions forstream flow nutrients. Water Resources Bulletin 23 (3), 471478.

arding, L.W., Perry, E., 1997. Long-term increase of phytoplankton biomass inChesapeake Bay. Marine Ecological Progress Series 157, 39–52.

auer, F.R., Baron, J.S., Campbell, D.H., Fausch, K.D., Hostetler, S.W., Leavesley, G.H.,Leavitt, P.R., McKnight, D.M., Stanford, J.A., 1997. Assessment of climate changeand freshwater ecosystems of the Rocky Mountains, USA and Canada. Hydro-logical Processes 11 (8), 903–924.

enkin, Z., Noy-Meir, I., Kafkafi, U., Seligman, N., Gutman, M., 1994. Soil phosporusin a managed Mediterranean woodland ecosystem: herbage response and cattlegrazing effects. Agriculture, Ecosystems & Environment 47 (4), 299–311.

ester, K.W., 1996. Influence of woody dominated rangelands on site hydrology andherbaceous production. A&M University, Edwards Plateau, Texas, M.S. Thesis.

aratepe, Y., 2007. Egirdir Golu Havzasindaki Ormanlarin EkolojikDegerlendirilmesi, Goller Kongresi (Goller Yoresi, Ic Anadolu Golleri veSorunlari) Bildiriler Kitabi , 09–10 Haziran 2007, Isparta. syf, pp. 171–190.

azanci, N., 1999. Koycegiz, Beysehir, Egirdir, Aksehir, Eber, Corak, Kovada, Yarisli,Bafa, Salda, Karatas, Cavuscu golleri, Kucuk Menderes deltasi, Gulluk sazligi,Karamuk batakliginin limnolojisi, cevre kalitesi ve biyolojik cesitliligi. TurkiyeIc Suları Arastirma Dizisi 4, 372.

esici, E., Kesici, C., 2006. The effects of interferences in natural structure of LakeEgirdir (Isparta) to ecological disposition of the lake. E.U. Journal of Fisheries &Aquatic Sciences 23, 99–103.

ittredge, D., Parker, M., 1989. Massachusetts Best Management Practices: Tim-ber Harvesting Water Quality Handbook. Massachusetts Cooperative ExtensionService, Amherst.

umbur, H., Kocak, S., 1998. Saglikli arazi kullanma kosullarinin tesbiti ile fiziki veyasal kaliciliginin saglanmasi. CEV-KOR Journal 7 (27), 6–10.

assachusetts Department of Environmental Protection (MADEP), 1985. Road Saltsand Water Supplies-Best Management Practices. Massachusetts Department ofEnvironmental Protection (MADEP), Boston, MA.

odelling 270 (2013) 54– 63 63

Meyer, J.L., Sale, M.J., Mulholland, P.J., Poff, N.L., 1999. Impacts of climate changeon aquatic ecosystem functioning and health. Journal of the American WaterResources Association 35, 1373–1386.

Mulholland, P.J., Marzolf, E.R., Webster, J.R., Hart, D.R., Hendricks, S.P., 1997.Evidence that hyporheic zones increase heterotrophic metabolism andphosphorus uptake in forest streams. Limnology and Oceanography 42,443–451.

Municipality of Egirdir, 2010. Project of Preparation to Watershed Protection Plan.Tubitak, Turkey.

Murdoch, P.S., Baron, J.S., Miller, T.L., 2000. Potential effects of climate changeon surface-water quality in North America. Journal of the American WaterResources Association 36 (2), 347–366.

Naiman, R.J., Magnuson, J.J., McKnight, D.M., Stanford, J.A., 1995. The Fresh-water Imperative: A Research Agenda. Island Press, Washington DC,pp. 21–49.

New Jersey Department of Environmental Protection (NJDEP), 1981. New JerseyStormwater Quantity/Quality Management Manual. New Jersey Department ofEnvironmental Protection (NJDEP), USA.

Özhan, S., 1982. Determination Of Evapotranspiration From Various Stands in Bel-grad Forest and Comparison of the Results With Those of Calculated By EmpiricalFormulas. I.U No. 2906. I.U Faculty of Forestry, Istanbul.

Schueler, T.R., 1987. Controlling Urban Runoff: A Practical Manual For Planningand Designing Urban BMPs. Metropolitan Washington Council of Government,Washington D.C.

Sener, S., 2010. Egirdir gol suyu ve dip sedimanlarının hidrojeokimyasal incelemesi(Doctoral dissertation). Department of Geological Engineering, SuleymanDemirel University, Isparta, Turkey.

Stenseth, N.C., Mysterud, A., Ottesen, J.G., Hurrell, W., Chan, K.S., Lima, M., 2002.Ecological effects of climate fluctuations. Science 297, 1292–1296.

Timur, M., Timur, G., Ozkan, G., 1988. Egirdir golunun verimliliginde biyolojikve kimyasal faktorlerin etkinlik derecelerinin incelenerek golun dogal verimduzeyinin artirilmasinda alinmasi gereken onlemlerin arastirilmasi. AkdenizUnv. Su Ur. Muh. Dergisi 1, 1–10.

Turkish State Meteorological Service, 2010. Meteorological Data,http://www.dmi.gov.tr

Ugurlu, A., Latifoglu, A., Akin, B., Onocak, T., 1999. Conservation of Lake Egirdir asa Potable Water Source. Hacettepe University/Environment Applications andResearch Center, Ankara.

U.S. Geological Survey, 2007. Global GIS: Astrogeology Research Program,http://webgis.wr.usgs.gov/globalgis/index.html

Valiela, I., McClelland, J., Hauxwell, J., Behr, P., Hersh, D., Foreman, K., 1997. Macroal-gal blooms in shallow estuaries: controls and ecophysiological and ecosystemconsequences. Limonology and Oceanography 42 (5/2), 1105–1118.

Winer, R., 2000. National Pollutant Removal Performance Database for StormwaterTreatment Practices, 2nd ed. Center for Watershed Protection, Elliot City, MD.