CROSS-BORDER CONSEQUENCES and CONFLICTS of INTEREST in RIVER BASIN MANAGEMENT PLANNING the Case of the Tisza River (Ukraine, Romania, Hungary, Serbia)

CROSS-BORDER CONSEQUENCES AND CONFLICTS OF INTEREST IN RIVER BASIN MANAGEMENT PLANNING The case of the Tisza River (Ukraine, Romania, Hungary, Serbia)

Mark Honti & Vera Istvánovics MTA-BME Water Research Group, Hungarian Academy of Sciences Stream eutrophication

• Eutrophication management relies on nutrient control • P control successful in lakes • Less obvious in streams • Interfering factors: • Hydromorphology (bedform) • Hydrodynamics (turbulence, WRT) • Stream network topology (reaches, reservoirs)

• Algal development may occur 100’s of kms downstream, management requires basin-scale approach • WFD focuses on domestic water bodies Basin. doi: 10.1007/s10750-012-0999-y Istvánovics

Number of sites nutrients Stream eutrophication: fuzzyrelation with 15 30 45 60 0 & Honti (2012)Efficiency ofnutrient management controlling eutrophication ofrunning waters intheMiddle Danube

P or N down * * P or N up

no trend

other Chl up no trend Chl down

Number of sites 10 15 20 25 30 0 5

P down

N down *

P and N down * no trend * P and N up Five countries share the Tisza catchment (UA,RO,SK,HU,SRB)

EU member state Non-member state UKRAINE

Tisza / Tisa Szamos / / Theiß Someș / Тиса Somesch HUNGARY

Maros / Mure ș ROMANIA

Danube 100 km SERBIA Hydrobiologia a 12 b 1

2500 200 60 20 400 20 ) )

2000 -3

150 -3 -3 -3 ) )

40 -1 -1 1500 s s 3 3 100 10 200 10

1000 Chl m

20 g Q (m Q (m

50 m mg Chl m

500 m (mg P TP Eutrophication status in the Tisza River m (mg P TP 0 0 0 0 0 0 1000 1100 1200 1300 1400 1500 0 50 100 150 200 250 • Tisza receives algae from 2 large tributaries1 • Tisza is too deep (up toL 10(km) m) to support meroplanktonic algal growth2 L (km) c 123Chl 45 d TP Maros 750 200 Szamos 200 30 3 1200 30 ) ) ] -3 -3 -3 150 150 -3 -3 ) ) -1 -1 500 20 2 800 20 s s 3 3 100 100 Chl m

250 10 1 400 10 g Q (m TP [mg P m Q (m mg Chl m 50 50 m TP (mg P m (mg P TP TP (mg P m (mg P TP

0 0 0 0 0 0 0 00 250250 500500 750750 1000 1000 0 50 100 150 200 L (km)L [km] L (km) 1: Istvánovics & Honti (2012) doi: 10.1007/s10750-012-0999-y 2: Honti et al. (2008) Assessing phytoplankton growth in River Tisza (Hungary). Verh. Internat. Verein. Limnol. 30 (1): 87-89. Fig. 8 Longitudinal profiles of median discharge (Q), TP and situated on the riverbank. Numbered arrows indicate large Chl in the period 1997–2006 in four rivers that exemplify four tributaries. a 1 Va´g (Vah) River, 2 Garam (Hron) River. different Chl response classes. a Danube; b Sajo´ River, c Tisza b 1 Herna´d (Horna´d) River. c 1 Szamos (Somes¸) River, 2 River, d Zagyva River; see Fig. 1. Dashed line Q, thick line TP, Bodrog River, 3 Sajo´ (Slana´) River, 4 Ko¨ro¨s (Cris¸) River, 5 closed symbol Chl. Arrows with circles indicate major cities Maros (Mures¸) River, dashed arrow indicates a major dam river. Longitudinal patterns in decadal (1997–2006) leach algae into the river. Class 2 response was median data in a few rivers of this study revealed four characteristic of several medium-sized and large rivers characteristic classes of trophic response to hydraulic of this study. properties and nutrient loads (Fig. 8). The Tisza River, the largest tributary of the Danube The Danube represented Class 1 response (Fig. 8a). represented Class 3 response (Fig. 8c). The highly In this large river the median biomass of phytoplankton discontinuous Chl profile contrasted the relatively was P-determined (cf. Figs. 5, 7), and thus, nutrient smooth downstream increase in Q and TP. Most of the criteria could be applied to further decrease Chl. Chl originated from the hypertrophic Szamos (Somes¸ The Sajo´ (Slana´ in Slovakian) River, a highly in Romanian) and Maros (Mures¸ in Romanian) Rivers. polluted tributary of the Tisza River (Somlyo´dy et al., Algae exported by the shallow Szamos were lost while 1999) exemplified Class 2 response (Fig. 8b). Q and advecting along the deep channel of the Tisza River. Chl increased relatively smoothly along the flow in This loss was further enhanced by the rapid sedimen- spite of elevated nutrient emission from a large tation of diatoms upstream of a large dam (Fig. 8c; industrial town along the riverbank (170,000 inhabit- Honti et al., 2008; Istvańovics et al., 2010). Notice- ants). Neither tributaries nor reservoirs caused dis- ably, the Szamos River was not significantly richer in continuity in nutrients or Chl that could be detected by nutrients than the Upper Tisza River, but its shallow the methods of this study. A reasonable nutrient channel favored the growth of meroplanktonic dia- criterion might aim at reducing the longitudinal toms (Istvańovics & Honti, 2011). The trophic status increase in TP (and DIN, not shown) along the Sajo´ of the Tisza River depended primarily on the status of River to eliminate the elevated nutrient concentrations its large, shallow tributaries (the Szamos and the in the recipient Tisza River (Fig. 8c). However, to Maros Rivers), so water quality management should reduce local Chl concentrations one should manage focus on these rivers to decrease Chl in the main the numerous upstream ponds and reservoirs that may channel.

123 SZAMOS / SOMEŞ Szamos & Maros Catchment area 18 000 km2 Population 1 200 000 Drinking water supply 74% Sewerage (& WWTP*) 65% (45%) Romania Hungary SzSzB

SM MM

SJ BN CJ Békés MS HR Csongrád AR AB SB HD MAROS / MUREŞ Catchment area 27 000 km2 Population 2 300 000 Drinking water supply 63% Sewerage (& WWTP*) 48% (30%) *tertiary treatment Conflicting development objectives along these international rivers

• Downstream: improve • Upstream: improve water quality, incl. drinking water and trophic and sanitation infrastructure toxicological status

• Downstream has only indirect influence on incoming water quality Approach

• Objectives • Model eutrophication in the Szamos and Maros • Assess improvement strategies

• Methods • Detailed modeling for the Szamos • Identify conflicts of interest • Propose compromise solution • Simplified modeling for the Maros (method testing) • Describe current status • Assess sensitivity / vulnerability Szamos: Methods

Meteorology Hydrology Nutrient budget Algal growth

• Nutrient budget on municipality-level • Point and diffuse sources • Unified catchment and water quality model • Embedded in a GIS environment • Modelled discharge, nutrient fluxes and algal growth in the entire stream network • Scenario analysis • Realistic and hypothetical states Districts Administrative & institutional differences

Statistical data 2850000 Administrative boundaries ‣ ‣ Data collection on NUTS 5 level (RO: municipality, HU: settlement) 2800000

Y ‣ Diﬀerent land use and crop categories N ‣ Institutions 2750000 ‣ RO: The Environmental Agency (Agenţia pentru Protecţia Mediului) doesn’t do routine

method: Voronoi polygon interpolation water quality monitoring Vectorized catchment boundary: 1 km 2700000 accuracy ‣ RO: The Water Agency (Apele Române) 5250000 5300000 5350000 Land5400000 use 5450000 focuses on water quantity data X ‣ HU: United Environmental, Water and Nature Protection Agency (until 2012), now under Ministry of Internal Aﬀairs N ‣ Water quality monitoring network ‣ HU: high spatial resolution, monthly data ‣ RO: minimum requirements from EU WFD,

Data source: CORINE mostly NO3 49

NUTS_3 data, we calculated the area of land-use categories situated within the Somes catchment assuming a homogeneous land-use distribution across each county.

Figure 13. Land-use map based on re-classified CORINE land-use categories.

Picture 2. Examples of re-classified CORINE land-use types. (2.4.3. Agriculture + natural vegetation re-classified as arable land; 3.2.4. Transitional scrub re- classified as pasture.) 52 Low population density, extensive agriculture

within the settlements. If so, the difference between the CORINE-based estimate and 2.4.3. 3.2.4. statistical data would represent a potentially intensely cultivated agricultural land (Picture 3) that remains unaccounted in this study in the lack of data. This area was equivalent to a mean 13.5% (range 8.8-22.7%) of the accounted arable land or 13.2% (range 7.0-19.3%) of the accounted grassland.

Picture 3. Examples of rural settlement structure.

Agricultural land in a village Marisel, centrum

To calculate surface runoff and soil erosion, we further modified land-use categories that were used to assess nutrient status of agricultural soils (Table 7). The consideration behind this distinction was that runoff and erosion depend as much on vegetation phenology (canopy height, surface cover) as on land-use. Thus, the broad category grassland included dominantly grass-covered urban areas, meadows and pastures when modeling erosion. Simultaneously, shrubland was treated as a separate category.

3.8. Lakes and reservoirs

We compiled information on the location and volume of lakes and reservoirs. Apele Romane (2009) listed 2 natural lakes and 11 reservoirs with surface area over 50 ha from the Somes catchment. Data for several other lakes and fishponds were found in various reports (Table 9). When only surface area and mean depth were reported, volume was taken equal to their product. We identified the coordinates of

lakes and reservoirs that did not appear in the CORINE land-use map from Google Earth.

Szamos: catchment modeling ‣ Agricultural ‣ Soil loss, diﬀuse ‣ Point sources nutrient budget nutrient loads ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● P ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ●

‣ Stream network ‣ Seasonal hydrology ‣ Algal biomass Szamos

Lapus Somesul Mare

Sieu Somes Fizes

Somesul Mic Szamos: catchment modeling

Modelled vs. observed Q Q Error Annual load (Csenger)

● ● ●● Satu Mare ● Csenger 6000 Cavnic 100.0 ● ● ValeaSomes Vinului VV fölött Sasar ● Circalau ● Sasar ● ● ●●Aciua ● ●LapusLapus● observed ●● BailorSomesu Mare ● Barsau ● ● ● Cavnic modelled Ilva ●● Lapus ] ● Lapus Salaj A ● ●Somesu Mare

1 ● ● ●Ulmeni ● ●

− ● 20.0 ● Salauta ● ● ● RebraNepos 4500 Bargau s ● ● ●● ● Rastoci Tibles ●Bistrita 3 ●● ● ● ● ●● SomesuSieu Mare ● ● Ilisua Beclean [m ●● Bistrita Meles ● ●● ● ● Almas−torkolat Somesu Mare s ● 5.0 Somes ● Dipsa Sieu é ● ● ● r ● ● ●Somes Mic−Salatiu ●● é ● ● ● 3000 Somes Mic−Gherla ● Fizes

m 0% Agrij ● ● ● ● Almas−Hida ● Q ● ●● ● ● ● ● Lonea ● Borsa●Gadalin

1.0 ● ●● ● ● ●● ● Nadas−Aghiresu Somes Mic−Apahida 1500 ● Nadas−Mera ●●Cluj ● ● 100% Capus Somesu●SomesuSomesuSomesul Cald Cald Cald Rece ● ●●●

Somesu Cald 0.2 Somesul● Cald ●●Belis 0 0.2 1.0 5.0 20.0 100.0 SS TP DIN 3 −1 [103 t év-1] -1 Qmodell [m s ] [t év ]

Algal profile in 2012 Water Quality Sampling Points Observed & modelled algal biomass in Somes

Olcsvaapáti 200 Mica Pomi Ulmeni Caseiu Galgau Ardusat Caraseu Csenger Dabiceni Ciocmani Satu Mare Satu Mare

Caraseu Olcsvaapati

Pomi Somesul Mic Tunyogmatolcs

2850000 Tunyogmatolcs Somesul Odorhei

Csenger Ardusat 150 ] 3 ● Ulmeni m model (2012) ● Galgau ● Ciocmani ● Somesul Odorhei ●

Mica ● 100 a [mg ● ● − ● observed N ●

Dabiceni (2012) Caseiu Chl ● ●

Somesul Mic poldf$y 50 model (long term mean) ● ● 2750000 ● ● 0

30 km 0 50 100 150 200 250 300 distance downstream of Dej Upstream length [km]

2650000 5250000 5350000 5450000

poldf$x Szamos: hydromorphology & algal growth

mean multiannual biomass

4x in 1 day 2012 ‣ Generous P supply ‣ Shallow, braided channels increase apparent growth rate of algae long-term mean ‣ Natural hydromorphology implies sensitivity to eutrophication Szamos: nutrient loading scenarios

P N

Present

+5% RBMP +5%

-36% BAT -25%

0 1500 3000 4500 6000 mean annual load [t] ‣ RBMP: current river basin management plan (Apele Române: Planul de Management al Spatiului Hidrograﬁc Someş-Tisa) ‣ BAT-BMP scenario: upgrade of 9 major WWTPs to enhance P removal + agricultural BMPs on erosion hot-spots Szamos: eutrophication scenarios

‣ BAT-BMP scenario: upgrade of 9 major WWTPs to enhance P removal + agricultural BMPs on erosion hot-spots ‣ Societal background: present landuse + no point sources ‣ Biogeochemical background: no inhabitants, natural vegetation everywhere Lessons learnt from Szamos

• Management • Compromise solution exists, requires extra resources to improve status in Romania • Science • Network topology is crucial • Rapid development of meroplanktonic algae in shallow, diverse streambeds • Free growth length from closest obstacle (e.g. large reservoir) • 66% of annual P load available for algal growth Maros: Methods

• Discharge is estimated from catchment area • Simplified nutrient emission is calculated at county (județ) level • Point and diffuse sources from population and WWTP data, agricultural statistics (inorganic fertilizers & manure, large animal farms) • Transfer efficiencies from Szamos • Stream topological model • Simulation of present status • Assessment of vulnerability 123 Maros: Topological map Distance from Tisza Major reservoir/cascade Major settlement

Szeged 0

HUNGARY Nadlac 74 ROMANIA

Arad 148 Mureș

AR 278 HD

Strei Deva 335

HD 385 Sebeș AB Alba Iulia 420 620 442 AB

AB 581 CJ CJ AB Arieș 463 Blaj Turda 529 MS CJ AB AB 496 MS 497 504 SB MS Târnaveni SB 507 Mureș 549 MS 568 Sighișoara 573 MS MS Târgu Mureș 580 580 581 HR HR 622 Odorheiu Secuiesc MS 681 640 HR Târnava MareTârnava Mica Târnava Maros: Subcatchments

Szeged

HUNGARY Nadlac ROMANIA Lower Mureș

Arad Mureș

Strei Deva

Sebeș Arieș Alba Iulia

Arieș Blaj Turda

Târnaveni Târnave Mureș Upper Mureș

Târgu Mureș Târnava MareTârnava Mica Târnava Maros: Results 1000

100 Mean discharge [m3 s-1] 10 Calculated

1 1 10 100 1000 Observed

10 Turda

188 166

132 44

19 ș Mure Târgu 9 Arad Deva + Hunedoara Alba Iulia Szeged

33 Secuiesc 12 Odorheiu HUNGARY ROMANIA Maros: Results

-1 964 Mean total P load [t yr ]

5.2 Turda 730

580 310

! ș Mure Târgu 107 18.7 Arad Deva + Hunedoara Alba Iulia Szeged Secuiesc 36.5 31.9 Odorheiu HUNGARY ROMANIA

• Observed TP load at the border: 900-1100 [t yr-1] (~200-250 mg P m-3)

• City of Târgu Mureș adds ~200 t P/yr quite upstream Maros: Results

Mean summer algal -3 188 biomass [mg Chl-a m ] 4 Turda

137 20 117 117 100

ș Mure Târgu 5 136 Arad Alba Iulia Szeged Secuiesc 29 Odorheiu 1 Deva + Hunedoara ROMANIA HUNGARY

• Observed mean concentration at the border: 128 [mg Chl-a m-3] • Algal growth explodes downstream of Târgu Mureș • Algae exhaust P capacity in the last 500 km • Suﬃcient diluting capacity for the large city loads in lower reaches Maros: Vulnerability

• Full P exploitation of algae means that any additional P load will directly converted into Chl • Reduction of P load is necessary to improve water quality along the river • Infrastructural development without increasing WWTP efficiencies will increase P load • Heavy morphological changes would not change outflowing biomass Issues with the RBMP practice

• Both Hungarian and Romanian RBMPs concentrate on local issues & solutions • Most large river sections are classified as “heavily modified” because of flood defence infrastructure • No real attempt is seen to improve ecological status • Discrepancy of the “Water body” concept: a middle- sized creek counts as much as a section of a large river • Virtual statistical improvement can be produced without touching the root of problems Conclusions

• Controlling eutrophication in large tributaries would improve water quality 100s of kms downstream • Harmonisation between domestic RBMPs is needed to • achieve improvement downstream • prevent worsening by pursuing alternative development objectives • Meaningless to elaborate local RBMPs for downstream sections of large rivers • except improving state of local tributaries • RBMP in SRB, HU should “target” upstream catchments, but how? The missing link?

Domestic International Entire water body tributary river basin (Danube)

Local management Cooperative ICPDR body management

• International tributaries are sources of conflicts, which can’t be resolved locally • Typically not critical on the scale of the entire Danube Basin

• RBMP for such large tributaries should be done by international panels instead of glueing local RBMPs together Summary

• Szamos & Maros are heavily eutrophicated • P load from point sources (infrastructural deficit) • Natural hydromorphology boosts algal growth • Droughts (climate change) increase algal growth • Management can reduce algal concentrations to about half • Infrastructural development without considering river properties will worsen status • Water quality in Tisza is determined by tributaries END