Ó Springer 2006
Hydrobiologia (2006) 553:89–109
DOI 10.1007/s10750-005-6427-9
Primary Research Paper
Is Lake Prespa jeopardizing the ecosystem of ancient Lake Ohrid?
A. Matzinger1,*, M. Jordanoski2, E. Veljanoska-Sarafiloska2, M. Sturm1, B. Müller1 & A. Wüest1
1
Swiss Federal Institute of Environmental Science and Technology (EAWAG), Limnological Research Center,
CH-6047 Kastanienbaum, Switzerland
2
Hydrobiological Institute, Naum Ohridski 50, MK-6000 Ohrid, Macedonia
(*Author for correspondence: E-mail: andreas.matzinger@eawag.ch)
Received 16 July 2004; in revised form 30 March 2005; accepted 25 April 2005
Key words: Lake Ohrid, Lake Prespa, karst, phosphorus, eutrophication, water level fluctuation
Abstract
Lake Prespa and Lake Ohrid, located in south-eastern Europe, are two lakes of extraordinary ecological
value. Although the upstream Lake Prespa has no surface outflow, its waters reach the 160 m lower Lake
Ohrid through underground hydraulic connections. Substantial conservation efforts concentrate on
oligotrophic downstream Lake Ohrid, which is famous for its large number of endemic and relict species. In
this paper, we present a system analytical approach to assess the role of the mesotrophic upstream Lake
Prespa in the ongoing eutrophication of Lake Ohrid. Almost the entire outflow from Lake Prespa is found
to flow into Lake Ohrid through karst channels. However, 65% of the transported phosphorus is retained
within the aquifer. Thanks to this natural filter, Lake Prespa does not pose an immediate threat to Lake
Ohrid. However, a potential future four-fold increase of the current phosphorus load from Lake Prespa
would lead to a 20% increase (+0.9 mg P m)3) in the current phosphorus content of Lake Ohrid, which
could jeopardize its fragile ecosystem. While being a potential future danger to Lake Ohrid, Lake Prespa
itself is substantially endangered by water losses to irrigation, which have been shown to amplify its
eutrophication.
Introduction
Lake Ohrid and Lake Prespa form a very unusual
lake system. Situated in south-eastern Europe
between Albania, Macedonia and Greece (Fig. 1),
they are both of tectonic origin with an estimated
age between 2 and 35 million years (Jakovljevic,
1935; Stankovic, 1960; Meybeck, 1995). According
to Stankovic (1960), the two lakes formed one lake
at their earliest stages of existence. Today, Lake
Prespa lies about 160 m higher than Lake Ohrid
(Table 1), separated by the Mali i Thate/Galicica
mountain range, which consists of karst rock
structures (Eftimi et al., 2001). Stable isotope
measurements, as well as recent tracer experiments
have revealed that water from Lake Prespa is
flowing to Lake Ohrid through karst channels
(Anovski et al., 1980; Eftimi & Zoto, 1997; Zoto,
pers. comm., 2003).
Lake Ohrid harbors a large number of endemic
species, which have been studied extensively since
the beginning of the 20th century. Most of the
endemic species are benthic forms (e.g., Pyrgula
macedonica (freshwater snail; Stankovic, 1960),
Ochridaspongia rotunda (freshwater sponge;
Gilbert & Hadzisce, 1984)), but there is also a
number of endemic plankton species (e.g., Cyclops
ohridanus (zooplankton), Cyclotella fottii (phytoplankton), both described by Stankovic, 1960) as
well as fish (e.g., Salmo letnica (Ohrid trout; Sell &
Spirkovski, 2004)). Through rising international
interest, the research institution ‘Station Hydrobiologique – Ohrid’ was founded in Ohrid (Fig. 1)
in 1935, dedicated to the study of Lake Ohrid and
90
Figure 1. Overview of study area indicating the different sampling sites. The grey arrows indicate flow direction (the broken arrow
signifies underground flow). Crossed circles indicate main lake sampling sites, empty circles are sediment core locations and filled
squares are sampled karst springs. The inset map shows the location of the study area within Europe.
Table 1. Characteristics of Lake Prespa and Lake Ohrid
Property
Unit
Lake Prespa
Lake Ohrid
Altitude
m asl
849 (854)a
693
1300b
2610c
Catchment area
a
2
km
2
Lake surface area
Maximal depth
km
m
254 (282)
48 (54)a
358
288
Mean depth
m
14 (19)a
155
3
Volume
km
3.6 (4.8)a
55.4
Hydraulic residence time
yr
11 (17)a
70
Average phosphorus concentration TP (2003)
mg P m)3
31
4.5
Number of endemic species
–
10?
>150
Number of inhabitants in catchment area
–
24,000
174,000c
Number of tourists per year
–
<1000?
50,000
Value in parentheses: in the 1980s before recent water level decline of Lake Prespa.
Including Small Lake Prespa and its catchment.
c
Including Lake Prespa and its catchment.
b
a
91
its extraordinary species (Serafimova-Hadzisce,
1985). The importance of Lake Ohrid was further
emphasized by UNESCO, when the region was
declared a World Heritage site (UNESCO, 1979).
With increasing public attention concerns
about the conservation of Lake Ohrid are also
growing. Particular focus is the preservation of its
oligotrophic state, which seems to be in jeopardy
due to rising population and tourism. Indeed a
slow eutrophication can be detected in sediment
cores of Lake Ohrid over the past 100 years
(Matzinger et al., 2004). Moreover, changes in
biological communities have been observed over
the past decades (Watzin et al., 2002). In 1997, a
project was launched by the Global Environment
Facility to tackle future problems with (i) the
improvement of the urban waste water treatment
system covering the major settlements in Macedonia, (ii) consolidation and extension of water
quality monitoring activities and (iii) the establishment of bilateral lake management (Ernst
Basler & Partners, 1995). Apart from direct
pollution from riparian towns, villages and agricultural fields, it is important to understand the
influence of the underground inflow from Lake
Prespa. Given that Lake Prespa contributes 50%
of the total catchment of Lake Ohrid and that the
total phosphorus (TP) concentration (Table 1) of
Lake Prespa is seven times higher than in Lake
Ohrid, the development of Lake Prespa is a
worrying concern for the eutrophication of
downstream Lake Ohrid.
Although similar in surface area, Lake Prespa
is much shallower than Lake Ohrid (Table 1).
Archaeological findings indicate that its water level
was subjected to large variations in the past
(Sibinovik, 1987; Milevski et al., 1997). Compared
to Lake Ohrid, much fewer endemic species have
been described (Karaman, 1971; Crivelli et al.,
1997; Shapkarev, 1997). With its nearly untouched
shoreline and large reed-belts, Lake Prespa is an
important breeding ground for various water
birds, such as the rare Dalmatian Pelican Pelecanus crispus (Crivelli, 1996; Nastov, 1997). As a
result Lake Prespa was declared a Ramsar site in
2000 (Ramsar, 2000). However, serious concerns
have been expressed about the recent water level
decline, as well as potential eutrophication of the
lake (Naumoski et al., 1997; Löffler et al., 1998;
Goltermann, 2001).
The goal of this paper is to quantify the role of
these anthropogenic changes in Lake Prespa in
regard to the ongoing eutrophication of Lake
Ohrid. Such an assessment is important and urgent
for defining potential management options and for
making optimal use of the limited financial
resources to protect the two lakes. Furthermore, it
is crucial that mitigation measures are taken in
time, given the sluggish dynamics of Lake Ohrid
with its hydraulic residence time s 70 year
(Table 1). In our analysis, we focused on phosphorus, as it is clearly the growth-limiting nutrient
(N:P > 25:1).
A system analytical approach is presented,
which begins with the assessment of the anthropogenic changes in upstream Lake Prespa, quantification of the underground transfer of water and
phosphorus and finally evaluation of the effects on
downstream Lake Ohrid in a linear phosphorus
model.
Study sites
Lake Prespa
Lake Prespa, situated at an altitude of 849 m asl,
is surrounded to the east and west by up to
2000 m high mountains. Symptomatic of the
sparse knowledge of Lake Prespa is the broad
range of values given by various authors for the
areas of the lake (254–285 km2) and its catchment
(822–1775 km2). In the following, we use the
values from a recent compilation by Anovski
(2001) given in Table 1, to which scientists of all
three riparian countries contributed. The
bathymetry shows that the lake is mostly shallower than 30 m with a few local deeper holes
(Fig. 2). A topographic map from the 1940s
(provided by the Hydrobiological Institute Ohrid
(unpublished data, ca. 1940)) indicates that the
present water level is 5–6 m lower than in the
1940s. Significant lake level decreases have also
been observed since the late 1980s (Fig. 3). To the
south, Lake Prespa is connected to Small Lake
Prespa (Fig. 1) by a controllable man-made
channel with a current hydraulic head of 3 m
(Hollis & Stevenson, 1997). As Lake Prespa is
relatively shallow compared to its large surface
area, wind and convective mixing lead to complete
92
3
Lake volume (
0
) [km ]
2
4
6
0
~ current water level
10
Depth [m]
20
settlements from
th
th
11 / 12 century ad
30
40
50
0
50
100
150
Cross-sectional area (
200
250
2
) [km ]
Figure 2. Hypsograph of Lake Prespa: cross-sectional area and
enclosed volume as a function of depth relative to the ‘‘historic’’
water level (6 m above current level). Data are from a map
from the 1940s (Hydrobiological Institute, unpublished data).
The horizontal lines indicate known water levels of Lake
Prespa.
856
Lake level [m asl]
854
852
850
848
1950
1960
1970
1980
1990
and one of the most voluminous lakes
(55 km3) in Europe (Table 1). Apart from its
unique flora and fauna, a peculiarity is the long
hydraulic residence time, which results from the
relatively dry, Mediterranean climate and the
small drainage basin. The water balance is
dominated by inflow from karst aquifers (55%)
with slightly smaller shares from river runoff and
direct precipitation (Matzinger et al., 2004). The
fraction of river runoff was even below 10%
before 1962, when River Sateska was deliberately
diverted into Lake Ohrid (Fig. 1). In contrast to
Lake Prespa, Lake Ohrid is an oligomictic lake
with complete mixing occurring roughly once per
decade (Hadzisce, 1966).
Karst springs
Along the western side of the Galicica/Mali
i Thate Mountain Range (Fig. 1), which separates the two lakes, numerous karst springs arise
and often seep away shortly after their appearance or flow directly into Lake Ohrid. The
springs, originating at an altitude higher than
Lake Prespa, are charged from mountain range
precipitation. Recent tracer experiments on the
lower altitude springs have revealed that two
particularly large spring areas, which flow into
Lake Ohrid on its south-eastern shore, are partly
fed from Lake Prespa. Each of the two areas,
Tushemishti in Albania and St. Naum in Macedonia, consist of dozens of spring holes
(Fig. 1).
2000
Figure 3. Development of Lake Prespa water level over the past
five decades. The fine line shows seasonal fluctuations according
to Greek measurements presented in Hollis and Stevenson
(1997). The bold, grey line represents annual means from
Albanian measurements, adapted from Anovski et al. (2001).
destratification of the entire water column from
September to April/May and consequently all
dissolved substances are homogenized annually.
Lake Ohrid
Lake Ohrid, situated between mountain ranges
to the east and the west, is oligotrophic, deep
(max. depth 288 m), large (surface area 358 km2)
Materials and methods
In order to assess the effects of a potential deterioration of Lake Prespa on Lake Ohrid, a sampling
and measurement program was established in the
area, as the basis of our system analytical
approach. The sampling program, analytical
methods, as well as calculations are presented in
the following. The declaration of materials and
methods is arranged in the same order and under
the same subtitles as the results in the ‘Results and
discussion’ section. Table 2 gives an overview of
all samples and analytical parameters within our
program.
93
Table 2. Overview of sampling and measurement program
Site
Sampling
Number of
period
samples
Parameters
Lake Prespa (0, 5, 10, 15, 20, 25, 30 m)
12/2002–9/2003
44
36
SRP, TP, DO
NO)2 , NH+
4 , TN
Lake Prespa sediment core
5/2002
43
TC, TIC, TN, Water content
Phosphorus load from Lake Prespa a
Transfer to Lake Ohrid
15
TP
18
137
Cs,
210
Pb
a
Lake Prespa (0, 5, 15 m)
4/2001–9/2003
3
d18O, dD
(3 different dates
for isotopes)
21
Na+, K+, Ca2+, Mg2+, Cl), SO2)
4
32
TP, SRP
Lake Ohrid (0, 5, 50, 250 m)
4/2001–10/2002
4
d18O, dD
St. Naum springs
4/2001–9/2003
7
d18O, dD
Other springs
4/2001–9/2003
11
Na+, K+, Ca2+, Mg2+, Cl), SO2)
4
10
TP
12
SRP
11
10
d18O, dD
Na+, K+, Ca2+, Mg2+, Cl), SO2)
4
10
TP
21
Rain samples
5/2001–10/2002
8
7/2002–9/2003
79
5/2002 and 4/2003
76
13
SRP
d18O, dD
Effect on Lake Ohrid a
Lake Ohrid (0, 10, 20, 30, 40, 50,
TP
75, 100, 150, 200, 250, 275 m)
Lake Ohrid sediment cores
a
TP, Water content
Cs, 210Pb
137
Titles in italics correspond with the subtitles in ‘Materials and methods’ and ‘Results and discussion’ sections.
Lake Prespa underground outflow
Before we can assess the potential effect of Lake
Prespa, the underground outflow must be quantified. Whereas a rough water balance for Lake
Prespa can be drawn from earlier publications, a
simple model was developed to account for the
effect of the observed level decline. The continuous
decrease (Fig. 3) indicates that a large share of the
underground channels is located below the present
lake surface, making Lake Prespa particularly
vulnerable to changes in water input. As a result,
any additional water output Qout,add, for example
for irrigation, will lead to a decrease in water level
z until a new equilibrium is reached at the level
z = h (z is the vertical coordinate, positive
upward; highest level: z = 0 m, deepest spot:
z = )54 m). In our model, equilibrium level h is
attained when
Qout;add ¼ Qout;undgr ðz ¼ 0Þ Qout;undgr ðz ¼ hÞ
þ Enet ðAðz ¼ 0Þ Aðz ¼ hÞÞ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl {zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl }
ð1Þ
i
3
where Qout,add (m yr ) is the additional water
consumption, which leads to a change in the water
balance, Qout,undgr (m3 yr)1) is the underground
outflow for water level z, A (m2) is the lake surface
area, and Enet (m yr)1) is the net evaporation
from the lake surface (=evaporation ) precipitation = 0.37 m yr)1). If we assume that the
outflow channels are distributed evenly over the
)1
94
western half of the lake bottom and that the
groundwater velocity scales with the hydrostatic
pressure, the underground outflow Qout,undgr in
Equation (1) can be expressed for an arbitrary
water level z = h as
Qout;undgr ðz ¼ hÞ
¼ Qout;undgr ðz ¼ 0Þ
0
1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
BAðz ¼ hÞ
Vðz ¼ hÞ Aðz ¼ 0Þ C
B
C
B
C
@Aðz ¼ 0Þ
Vðz ¼ 0Þ Aðz ¼ hÞ A
|fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl {zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl }
ii
ð2Þ
iii
3
where V (m ) is the lake volume. If we combine
Equations (1) and (2), three stabilizing processes
are considered that are effective during the water
level decline [letters correspond with the terms in
Equations (1) and (2)]: (i) reduced evaporation
due to smaller lake surface, (ii) karst outflow
channels that are set dry, (iii) decrease in hydrostatic pressure (assumed relative to average lake
depth, V/A).
If the recent water level decline is the result of
an additional output (e.g., irrigation) rather than
an increase in evaporation, a decrease in the lake
level will change the total water outflow Qout,tot
from Lake Prespa. Using Equations (1) and (2) we
find:
Vðz ¼ hÞ
Qout;tot ðz ¼ hÞ
Zh
XðzÞAðzÞdz f
ð3Þ
ð4Þ
will decrease with declining water level.
Phosphorus load from Lake Prespa
The concentration of phosphorus (P) in Lake
Prespa was monitored in order to know the export
loads from the upstream lake by underground
outflow. In addition nitrogen (N) species and dissolved oxygen (DO) were measured to assess the
Zh
AðzÞdzg1 ð5Þ
z¼zmax
z¼zmax
According to Equation (3), Qout,tot will increase
during water level decline by the reduction in net
evaporation from the lake surface, although
Qout,undgr decreases. Under that assumption, the
level-dependent bulk water residence time
sðz ¼ hÞ ¼
Water samples
At one of the deep sites in Lake Prespa
(depth 30 m; Fig. 1) water samples were collected bimonthly at 5 m depth intervals using a
Niskin bottle (Table 2). The campaign started in
December 2002 and was continued until September 2003. Samples were stored in new or acidrinsed plastic bottles and cooled for transport.
Total phosphorus (TP), total nitrogen (TN),
soluble reactive phosphorus (SRP), nitrite (NO)2 )
and ammonium (NH+
4 ) were analyzed colorimetrically using standard analytical methods (DEW,
1996). Mean measurement errors were 1.9 mg
P m)3 for TP, 20 mg N m)3 for TN, 0.5 mg P m)3
for SRP, 0.2 mg N m)3 for NO)2 and 2 mg N m)3
for NH+
4 . DO was measured with the Winkler
method (Table 2).
For average lake concentrations, annual or
perennial averages of the volume-integrated profiles were used. Volume integration was performed
numerically using
hXih ¼
Qout;tot ðz ¼ hÞ ¼ Qout;add þ Qout;undgr ðz ¼ hÞ
¼ Qout;undgr ðz ¼ 0Þ þ Enet
ðAðz ¼ 0Þ Aðz ¼ hÞÞ
extent of eutrophication in Lake Prespa. Furthermore, sediment samples were analyzed and sedimentation rates determined to reconstruct the
history of eutrophication and the related potential
increase in the outflow of P.
where X (mg m ) is the concentration of the
considered parameter, e.g., TP, ÆXæh is the volumeaverage of X within the lake water between the
surface level z = h and the maximal depth
z = zmax. The cross-sectional areas A(z) are taken
from the function plotted in Figure 2.
)3
Sediment sampling
One sediment core was retrieved from a depth of
20 m in Lake Prespa, close to the water sampling
site (Fig. 1), using a gravity corer (Kelts et al.,
1986) and subsequently sectioned to 1–2 cm long
vertical segments. For each segment, the water
content was measured by weight loss after freezedrying. TP was measured photometrically after
digestion with K2S2O8 in an autoclave for 2 h at
120 °C (DEW, 1996). Total carbon (TC) and total
nitrogen (TN) were analyzed with a combustion
95
CNS-Analyzer (EuroVector Elemental Analyzer).
Total inorganic carbon (TIC) was measured by
infrared absorption of CO2 after acidifying the
sample with 3 M HCl (Skoog et al., 1996). Total
organic carbon (TOC) was calculated as
TOC = TC ) TIC.
For the dating of the core 137Cs and 210Pb
activities were established from c-counting in
Ge–Li borehole detectors (Hakanson & Jansson,
1983). As no clear peaks could be identified in the
137
Cs profile, only 210Pb was used for the dating. In
the top few cm of the core, the 210Pb activity was
practically constant, probably because of bioturbation by benthic organisms. Below the
homogeneous layer, the 210Pb signal decreased
exponentially to background activity. The
exponential fit led to a sedimentation rate SR of
0.075 ± 0.008 cm yr)1 [correlation R2 = 0.96; the
method is detailed in Wieland et al. (1993) and
Doskey & Talbot (2000)].
Sediment accumulation rates were calculated
using the following equation:
SM ¼ ð1 PORÞ SR qsed
ð6Þ
where SM (kg m yr ) is the mass accumulation
rate, POR ()) is porosity calculated from the
water content, SR (m yr)1) is the sedimentation
rate from 210Pb dating and qsed = 2500 kg m)3
is the sediment density established by pycnometer. The average SM in the dated section of the
core (top 13 cm) was used to calculate the current accumulation rates for TP, TN and TOC,
by multiplication with their respective measured
proportions. For total lake accumulation the
rates above were multiplied with the surface area
A(z = h).
)2
)1
Phosphorus balance
A balance of TP was used to assess the effect of
anthropogenic changes. The phosphorus balance
of a lake is generally expressed as:
V
@hTPi
¼ Pinp Psed;net Pout
@t
ð7Þ
where ÆTPæ (mg m)3) is the volume-averaged
concentration of TP (Equation 5), ¶ÆTPæ/¶t
(mg m)3 yr)1) is the rate of change of ÆTPæ, Pinp
(mg yr)1) is the annual phosphorus input, Psed,net
(mg yr)1) is the area-integrated net sedimentation,
and Pout (mg yr)1) is the outflow of TP. Given the
simplicity of Equation (7), all parameters represent
annual or perennial averages. For our purposes,
we used the linear model by Vollenweider (1969):
@hTPi 1
b
¼ Pinp r hTPi hTPi
@t
V
s
ð8Þ
where Psed,net from Equation (7) is assumed proportional to the total phosphorus content with the
sedimentation rate constant r (yr)1) and the outflow Pout from Equation (7) is expressed by the
water outflow (V/s) times the average outflow
concentration b ÆTPæ (where b = TPsurface/ÆTPæ).
From Equation (8), the residence time of P in the
water column results as s* = ((b/s) + r))1, where
s* quantifies the e-folding time with which the
equilibrium concentration s* Æ Pinp/V is approached. For Lake Prespa, V and s depend on the
water level h. For changes in h, V was adapted
according to the function in Figure 2 and s was
replaced by Equation (4).
Transfer to Lake Ohrid
In order to understand the transfer from Lake
Prespa to Lake Ohrid, different tracers, as well as
nutrients, were measured in samples from Lake
Prespa and in springs which flow into Lake Ohrid.
Water from eight different karst spring areas
(Fig. 1), Lake Prespa and Lake Ohrid was collected from 2001 to 2003. In addition, rain water
was collected at different altitudes (Table 2). The
springs at St. Naum were sampled at a 3 month
interval. The other springs were sampled a total of
between two and eight times, whenever opportunity was given, as some of them are hard to access.
Samples were stored in new or acid-washed plastic
bottles. TP and SRP were measured as described
for Lake Prespa above. Na+, K+, Ca2+, Mg2+,
Cl), SO2)
4 were determined with ion chromatography (IC 690 equipped with a Super-Sep column
for cations; IC 733 with 753 suppression module
for anions, all Metrohm, Switzerland; methods in
Weiss, 2004). Because of an unfortunate loss of
samples, only four springs were analyzed for TP
and ions (Table 2).
Water samples for stable isotope analysis were
stored in glass vials and cooled immediately after
sampling. 18O/16O and D/H ratios were analyzed
by isotope ratio mass spectrometry (GV Instruments IRMS (Manchester, UK) in continuous
96
flow mode) at the Limnological Research Center
Kastanienbaum. d18O and dD isotope compositions of the water samples were expressed using the
delta-notation (d) as a per mille deviation from the
internationally accepted Vienna Standard Mean
Ocean Water (VSMOW). The analytical errors are
0.3 and 0.8& for d18O and dD, respectively.
Samples were equilibrated with a CO2 and He
mixture for d18O and with a H2 and He mixture for
dD analysis at 40 °C for at least 12 h prior to the
measurement. The method used is adapted from
Werner & Brand (2001).
The share of Lake Prespa water in the
St. Naum Springs flow was calculated using different tracers with the following equation:
rPR ¼
CSN CPS
100%
CPR CPS
ð9Þ
where rPR (%) is the share of Lake Prespa water
and C (mg m)3) is the tracer concentration; subscript SN stands for St. Naum Springs, PS for
precipitation-fed springs and PR for Lake Prespa.
Deviations from the mean were individually calculated for each place and parameter. For rPR, the
individual errors were combined through error
propagation. Based on these errors DrPR, an errorweighted average was calculated from all conservative tracers.
Effect on Lake Ohrid
In order to evaluate potential effects on Lake
Ohrid, its P balance of the form of Equation (8)
was also established. For that reason, TP was also
measured in the water column and in the sediment
of Lake Ohrid.
Water profiles were drawn from May 2002 to
September 2003 (Table 2). Water samples were
treated and analyzed for TP using the same
methods as for Lake Prespa. Three sediment cores
were taken along the North-South axis of Lake
Ohrid (Fig. 1) in 2002 and 2003. An exponential fit
to the measured 210Pb activities (R2 = 0.97) of one
core resulted in a sedimentation rate of
0.089 ± 0.006 cm yr)1. For the analysis of TP in
the sediment, the same methods described for Lake
Prespa were applied. P accumulation rates Psed,net
were calculated by multiplying measured TP
contents with SM (Equation 6) for all three cores.
For Equation (8), the average, estimated between
1.5 and 3.5 cm sediment depth, was used as the
current value of Psed,net.
Results and discussion
Lake Prespa underground outflow
To assess the current underground outflow from
Lake Prespa through the karst aquifers and its
potential variability, we need to understand the
overall lake water balance. The three major contributors to water input are river runoff from
numerous small streams, direct precipitation onto
the lake surface and inflow from Small Lake
Prespa. The loss terms in the water balance are
evaporation, diversion for irrigation as well as
outflow through karst aquifers and fissures on its
western shore (Eftimi et al., 2001; Fig. 1).
Anovski et al. (2001) compiled measurements
of hydraulic loads q [l km)2 s)1] (runoff generated
per land surface area) from various authors and
combined them with their own measurements to
calculate a water balance (Table 3). For a consistency check, we compared this water balance with
the seasonal level fluctuations of Lake Prespa
measured by Hollis & Stevenson (1997) (Fig. 3).
The de-trended water level shows a very regular
seasonal oscillation with a period of one year and
average amplitude of 0.3 m. The lake level
oscillations are the result of the seasonal changes
in the relative contribution of the two major terms
in the water balance, evaporation and precipitation (Fig. 4). If we assume that river runoff follows
a similar pattern as precipitation at lake level, we
would expect a net water input from October to
March and a net water loss from April to September. Each term of the water balance was split
up between the dry and wet seasons, based on the
relative fractions of evaporation and precipitation
in Figure 4 (third and fourth columns in Table 3).
Groundwater in- and outflows were assumed to be
constant, whereas irrigation was considered only
during dry seasons. The separate balance for two
time periods resulted in a seasonal volume change
of ±145 Mio m3. This corresponds to a surface
height variation of ±0.29 m, which fits with the
observed annual oscillation in Figure 3. This
97
Table 3. Water balance of Lake Prespa from Anovski et al. (2001)
Process
Runoff from subarea in E
and SE (q = 13 l km)2 s)1)
Runoff from karst subarea in W and
Inflow
Outflow
Fraction during
Fraction during
[Mio m3 yr)1]
[Mio m3 yr)1]
dry seasona [%]
wet seasona [%]
199
37
63
31
37
63
69
37
63
186
37
63
49
50
50
279
78
22
10
100
0
245
50
50
SW (10% of total precipitation)
Runoff from subarea in plain in N
(q = 9 l km)2 s)1)
Precipitation on lake surface
(average from 6 stations along
shore 735 mm yr)1)
Inflow from Small Lake Prespa
and from aquifers in Greece
Evaporation (average from three
different approaches 1100 mm yr)1)
Irrigation with lake water
Underground outflow (from balance)
Total
534
534
a
Dry season is from April to September, wet season from October to March. Relative contributions of river runoff, precipitation and
evaporation are based on Figure 4.
220
200
Average evaporation
Average precipitation
at lake level
-1
Water flux [mm month ]
180
160
140
120
100
80
60
40
20
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 4. Seasonal evaporation (from Anovski et al., 2001) and
precipitation (from Hollis & Stevenson, 1997; Ristevski et al.,
1997) at the surface of Lake Prespa. 100 mm month)1 corre3
sponds to 0.8 m s)1.
agreement indicates that the water balance by
Anovski et al. (2001) is within the correct range.
However, this consistent water balance cannot
explain the recent decrease of the Lake Prespa
water level. This water loss, apparent in Figure 3,
continued resulting in a total decrease of 6 m
between 1986 and 1996 (Chavkalovski, 1997;
Löffler et al., 1998; Anovski et al., 2001). From
1996 to 1999, the level stabilized (Anovski et al.,
2001) but decreased further from 2000 to 2002
(Chavkalovski, pers. comm., 2002). According to
Figure 2, a decrease of 5 m corresponds to a loss
of 1.2 km3 of water, which is roughly 25% of the
total lake volume. Different explanations have
been proposed for the observed decrease (Chavkalovski, 1997; Hollis & Stevenson, 1997; Anovski
et al., 2001): (a) a tectonic subduction of the lake
bottom, (b) an opening of underground channels,
due to seismic activity, (c) climatic variability or
(d) increased irrigation. Comparing a topographic
map (probably from the late 1940s) with our own
measurements, the lake depth has decreased by
5–6 m, which contradicts point (a). Archaeological findings imply that the water level has been
several meters below the present elevation for extended time periods during the past 1000 years
(Milevski et al., 1997). While options (b) and (c)
cannot be ruled out completely regarding the
current level decline, Chavkalovski (1997) calculated that the combined irrigation water consumed
by the three riparian countries since the mid 1960s
could indeed lead to several meters decrease in lake
98
level. He assumed consumption from Lake Prespa
of 10 Mio m3 yr)1 from Macedonian territory,
and a water uptake from Small Lake Prespa of 35
and 10 Mio m3 yr)1 from Albanian and Greek
territory, respectively.
Using Equations (1) and (2), it is possible to
extrapolate the underground outflow to the level
of 1965 and to estimate the effect of an additional
water outlet Qout,add on the water level. In such a
scenario a former underground outflow
of 313 Mio m3 yr)1 would be 68 Mio m3 yr)1
higher than today. The consumption of 55 Mio m3
for irrigation water each year would decrease the
surface water level in Figure 2 by 2 m in 10 years
and by 4.5 m, until a new equilibrium is reached
after several decades. This rudimentary analysis
shows that
Phosphorus load from Lake Prespa
To assess the role of Lake Prespa in the eutrophication of Lake Ohrid, we need to know how the
phosphorus loads from Lake Prespa have changed
in the recent past and how they are likely to
change in the near future. A key to these questions
lies in the change of the trophic state of Lake
Prespa. Several indicators of eutrophication in
Lake Prespa are subsequently discussed.
Dissolved oxygen (DO)
Figure 5a shows the development of the vertical
distribution of DO in the course of one year for a
deep site in the Macedonian part of Lake Prespa
(Fig. 1). Anoxic conditions prevail below 15 m
from July to September 2003. This pattern has also
been documented by others (Naumoski et al.,
1997; Löffler et al., 1998; Jordanoski et al., 2002).
Along with the depletion of DO, the accumulation
of mineralization products such as SRP, NO)2 and
NH+
4 (Fig. 5b, c and d) indicates a significant input
of organic material to the lake sediment. Figure 6
compares historic data from the work of Jakovljevic (1935) with recent measurements of DO from
the same location. The most striking difference
(i) any additional water loss, artificial or natural, can lead to a drastic decline in the water
level of Lake Prespa;
(ii) the long-term underground outflow of Lake
Prespa will be reduced by such an additional
water loss;
(iii) although the exact reasons for the observed
decline are not known, irrigation practices
are a plausible hypothesis.
25-Mar-03
7-May-03
Depth [m]
(a)
12-Jun-03
17-Jul-03
(b)
26-Aug-03
24-Sep-03
(c)
(d)
0
0
0
0
5
5
5
5
10
10
10
10
15
15
15
15
20
20
20
20
25
25
25
25
30
30
30
30
0 2 4 6 8 10 12
-1
DO [mg L ]
0
20
40
60
0
-3
SRP [mg P m ]
80
100
-3
N-NO2 [mg N m ]
0 2 4 6 8 10 12
-3
N-NH4 [mg N m ]
Figure 5. Development of (a) DO, (b) SRP, (c) NO)2 and (d) NH+
4 in the water column of Lake Prespa in the course of the productive
season 2003.
99
appears in the summer profile (Fig. 6c). In 1931,
DO close to the lake bottom never dropped below
6 mg l)1. Such a strong change over the past
70 years indicates that the observed summer
anoxia is partly of recent anthropogenic origin.
50
-3
TP concentration [mg P m ]
45
Phosphorus in the water column
As in Lake Ohrid, P is the growth-limiting element
in Lake Prespa, as the molar N:P ratio in the top
10 m is about 25:1 on average, well above the
Redfield ratio of 16:1 (Redfield, 1958). The average concentration of TP in Lake Prespa was 31 mg
P m)3 in 2003, which is consistent with findings by
Jordanoski et al. (2002) for the years 2000–2002
(Fig. 7). Naumoski et al. (1997) measured an
average TP concentration of 20 mg P m)3 between
January 1992 and January 1993, which signifies an
increase of 11 mg P m)3 within 10 years. The
concurrent 0.6 km3 decrease in lake volume would
explain a TP increase of 3.3 mg P m)3, if we assume that the P-content of the lake and the P-input to the lake have both remained constant for
this 10-year period. Thus an increase of
11 ) 3.3 = 7.7 mg P m)3 or 0.8 mg P m)3 yr)1
must be caused by a raised P-load to Lake Prespa.
30
25
20
15
2000
2001
11-Aug-1931
26-Aug-2003
0
(b)
(c)
5
5
5
10
10
10
15
15
15
20
20
20
25
25
25
30
30
30
35
60
80
100
2003
Phosphorus in the sediments
Another approach investigating the eutrophication
history is provided by sediment cores. In a
sediment core, taken at a depth of 20 m, an increase in the contents of TOC, TN and TP was
observed over the top 12 cm (150 years; Fig. 8).
0
35
2002
Figure 7. Seasonal development of average TP concentration in
the top 15 m of Lake Prespa. The full circles are measurements
by Jordanoski et al. (2002), the empty squares are our measurements. The horizontal line is the average concentration over
the whole period.
27-Apr-1932
7-May-2003
(a)
Depth [m]
35
10
4-Dec-1931
26-Dec-2002
0
40
35
60
80
100 120 140
0
40
80
120
Oxygen saturation [%]
Figure 6. Comparison of oxygen saturation in profiles of Lake Prespa between historic results from Jakovljevic (1935) and recent
measurements, for winter (a), spring (b) and summer (c). The surface point from December 1931 in Figure 6a is most probably a
measurement error.
100
area for the top of the core. The results show an
increase over the past 150 years from 46 to 72 t
P yr)1 for TP, from 253 to 346 t N yr)1 for TN
and from 1710 to 2440 t C yr)1 for TOC.
Surprisingly the contents of TOC, TN and TP
increase with growing depth of the core below
20 cm (Fig. 8). A possible reason is again water
level fluctuation, which could change the mixing
regime of the lake. At much lower lake level,
benthic–pelagic coupling would be expected to be
enhanced throughout the summer season,
increasing the available nutrients and thus lake
productivity. The increase in organic components
in Figure 8 towards the bottom of the core might
thus indicate much lower water levels several
centuries ago. Archaeological findings of settlements from the 11th century in the lake, located
below the present water level (Sibinovik, 1987;
Milevski et al., 1997), support such a scenario
(Fig. 2).
Such an increase in organic components towards
the sediment surface is expected as a result of early
diagenesis (e.g., Hupfer et al., 1995; Meyers &
Ishiwatary, 1995). However, while mineralization
stays at high rates for a few years only, nutrient
release should be terminated after two decades at
the most (Lotter et al., 1997; Urban et al., 1997).
Thus, apart from the top few centimeters a real
increase in TOC, TP and TN content has occurred.
Nevertheless the reason for the increase is not
straightforward. The most obvious explanations
would be a change in allochthonous input or
eutrophication. In the case of the latter a concurrent increase in calcite-precipitation, and thus a
dilution effect, would be expected for Lake Prespa
hard water (Dittrich & Koschel, 2002). Indeed,
such an effect can be seen in the measured TIC
concentrations (Fig. 8). On the other hand, the
observed increase in TIC supports the scenario of
eutrophication. Furthermore the level fluctuations
of Lake Prespa could have a strong influence on
sedimentation. Net sedimentation rates were estimated with linear lake surface correction based on
Equation (6), using the lake surface area as in
Figure 2 for 12 cm depth and the current surface
0.6
0.8 TP (
2
4 TN (
Lake Prespa phosphorus balance
In order to establish a phosphorus balance for
Lake Prespa, the influence of the recent water level
fluctuations on the terms V and s in Equation (8)
-1
) [mg g ] 1.0
1.2
-1
) [mg g ] 6
8
0
Sediment depth [cm]
10
20
30
40
50
60
70
30
5
TIC (
TOC (
-1
) [mg g ]
) [mg g-1] 10
45
15
Figure 8. Content of TP, TN, TOC and TIC as measured in the sediment core from Lake Prespa, taken at 20 m depth in 2002 (Fig. 1).
101
Table 4. P balance of Lake Prespa and Lake Ohrid
P Balancea
Lake Prespa
b
Equilibrium P concentration
Lake Ohrid
Historic
2003
2003
20
31
46
0.5
60
0.5
38
0.15
1
1
0.77
52
70
6
9
4.5
[mg P m)3]c
Net sedimentation [t P yr)1]
r(sedimentation rate constant = net
P sedimentation per total lake content) [yr)1]
b(concentration in outflow
divided by volume-average concentration) [)]
P-input (Pinp [t P yr)1])
P-outflow (Pout [t P yr)1])
40
3.5
a
P balance of Lake Prespa based on Equations (2, 4 and 8) and of Lake Ohrid based on Equation (8). Input data are in italics, results
are in bold.
b
For the historic situation of Lake Prespa the minimum P sedimentation rate and the 1992 TP concentration by Naumoski et al. (1997)
were used.
c
Equilibrium P concentration is the annual average of mean lake concentrations (Equation 5).
have to be included. If we know the water level, TP
concentration in the lake water and Psed,net, Pinp
can be estimated for equilibrium conditions based
on Equations (4) and (8).
Results of these calculations are shown in
Table 4. In this table the historic situation, before
the recent increase of TP in the sediment core
(Fig. 8), is compared to 2003. Because there is little
historic data, the TP concentration in 1992
(Naumoski et al., 1997) was used for the historic
conditions. Table 4 implies an increase in TP input
from 52 to 70 t P yr)1 from 1992 to 2003 and a
current net sedimentation rate of 60 t P yr)1. This
net sedimentation rate indicates that about
60 ) 46 = 14 t P yr)1, or about 50% of the
observed 26 t P yr)1 increase of TP in the sediment
core, is likely the result of the higher P input. It
further implies that Lake Prespa is reducing the
potential P loads to Lake Ohrid by Psed,net/
Pinp 86% (Table 4).
Results from Equations (4 and 8) can be
checked for plausibility with the observed internal
balance. Figure 7 shows the seasonal P dynamics
of Lake Prespa. During the productive season, the
average TP concentration decreases by 28 mg
P m)3 in the top 15 m, which indicates loss by
sedimentation. This loss corresponds to a gross
sedimentation (Psed,gross) of 90 t P yr)1. If the
calculated net sedimentation Psed,net in Table 4 is
correct, the fraction of the released P from sedi-
menting particles and from the sediment,
Prel = Psed,gross ) Psed,net, equals 30 t P yr)1. Such
a release of phosphorus can be observed in an
increase of SRP below the thermocline over the
summer season (Fig. 5b). By depth-integrating the
SRP profiles below 10 m (Equation 5), an increase
of 31 t P was found from March to August/September 2003, which compares very well with
the result of our internal balance. The ratio
Prel/Psed,gross would then be 30%, which is within
the expected range (Hupfer et al., 1995; Moosmann et al., 2005). The sum of the calculated input
fluxes Pinp and Prel is 101 t P yr)1, which is very
close to the sum of the output fluxes Psed,gross and
Pout of 99 t P yr)1.
Summary
Both the development of anoxic conditions in the
bottom water and the increases in sedimentary P,
as well as the linear lake P balance, point towards
a eutrophication of Lake Prespa. This development is most probably the result of a combination of intensified agriculture (Chavkalovski,
1997; Hollis & Stevenson, 1997), lack of sewage
treatment and increased use of P-containing detergents. However, the increase of P input cannot
be linked directly to the population, which has
decreased in the past 50 years (Berxholi, 1997;
Catsadorakis & Malakou, 1997; Macedonian
State Statistical Institute, pers. comm., 2003).
102
Apart from the increased P loads, lake level
changes can have a large influence on the water
quality. Changes in the lake volume have a direct
effect on the concentration of dissolved nutrients.
If indeed the lake surface was much lower historically, P concentrations may have been even
higher than today.
This leaves us with two main conclusions: (i)
Lake Prespa is in a process of eutrophication and (ii)
the water level of Lake Prespa has a large effect on its
water quality. A further level drop could surpass the
effect from increased nutrient input.
Transfer to Lake Ohrid
Having established a model for calculating the
phosphorus load from Lake Prespa, we focus now
on the fraction of P that is transferred to Lake
Ohrid.
Water transfer
Within a current IAEA project (Anovski, 2001)
color tracer experiments have revealed that two
large springs at the south–eastern end of Lake
Ohrid have significant connections to Lake Prespa
(Zoto, pers. comm., 2003). These two sources
provide a total inflow of 10 m3 s)1 (Watzin
et al., 2002). Part of this flow may not originate
from Lake Prespa but instead from local precipitation seeping through the karst rocks and mixing
with the Prespa water that flows downwards into
Lake Ohrid. Several studies have been undertaken
on the hydraulic connection between the two lakes
using stable isotopes (Anovski et al., 1980;
Anovski et al., 1992; Eftimi & Zoto, 1997; Eftimi
et al., 2001). The results for dD and d18O from
Anovski et al. (1980) and Eftimi & Zoto (1997) are
plotted in Figure 9, together with our own measurements from 2001 to 2003. Despite the
dynamics of the underground system, the values
have been surprisingly constant over more than
two decades.
Lake Prespa and Lake Ohrid are enriched in
heavy isotopes due to evaporative processes.
Springs fed from precipitation are evenly distributed between the prediction bands of the empirical
meteoric water line (MWL), which is based on
local rain samples by Anovski et al. (1980) and by
ourselves (Table 2, Fig. 9). St. Naum Springs, the
largest spring area, which is partly charged by
Lake Prespa, is also enriched in heavy isotopes
compared to the MWL, which further confirms
that parts of its waters have been subject to
evaporation.
The measurement of stable isotopes has been
used to estimate the ratio of water that originates
from Lake Prespa emerging at St. Naum (Fig. 1)
and of rainwater percolating through the karst
rocks, based on Equation (9). A similar balance
can be done for all substances which are transported by underground water. In Table 5, the results for the stable isotopes dD and d18O, several
ions and TP are shown. The calculations are based
on 5 to 30 measurements per site from the years
2001–2003 (Table 2). Measurements from the top
15 m of Lake Prespa were included without volume-averaging, because no distinct profiles were
observed and the position of the outlets is
unknown. Most of the water constituents, listed in
Table 5 (e.g., SO2)
4 or SRP), are not conservative
since they can be influenced by biological processes
or by ion exchange, as most cations. Thus, only
stable isotopes and Cl) were used to establish an
error-weighted average of the mixing ratio. The
result implies that 43 ± 5% of the spring water at
St. Naum is fed from Lake Prespa and 57 ± 5%
from local precipitation.
Eftimi & Zoto (1997) found similar numbers
for St. Naum Springs and a slightly increased
share of Lake Prespa water in the Tushemishti
Springs in Albania (Fig. 1). In Table 6, the percentages of Lake Prespa water in St. Naum
Springs from our own findings and in Tushemishti
Springs from Eftimi & Zoto (1997) are multiplied
with their respective inflows, presented by Watzin
et al. (2002). In total, the two spring areas were
found to contribute about 4.5 m3 s)1 of Lake
Prespa water to Lake Ohrid. This corresponds to
58% of the total underground outflow from
Lake Prespa (245 Mio m3 yr)1 = 7.8 m3 s)1) and
leaves about 7.8 – 4.5 = 3.3 m3 s)1 (100 Mio m3 yr)1)
unidentified (Tables 3 and 6). Although there are
several small springs to the south of Lake Ohrid,
outside of its catchment area, Eftimi & Zoto
(1997) found that these springs are mainly fed
from local precipitation. However, we have
located subaquatic springs below the surface of
Lake Ohrid along its eastern shoreline with CTD
transects. Moreover, the water balance of Lake
Ohrid indicates that as much as 10 m3 s)1 are
103
from Anovski et al. (1980)
from Eftimi et al. (1997)
own measurements 2001 / 2002
empirical meteoric water line
95% prediction limits
20
0
MWL
Lake Prespa
δD [‰]
-20
-40
Lake Ohrid
-60
St. Naum Springs
Springs without significant
connection from Lake Prespa
-80
-100
-12
-10
-8
-6
-4
-2
0
18
δ O [‰]
18
Figure 9. Ratios of stable isotopes dD and d O in the two lakes, St. Naum Springs with a proven connection from Lake Prespa and
various springs fed by precipitation only. The meteoric water line (MWL) is based on linear regression of local precipitation measurements by us and Anovski et al. (1980).
expected from subaquatic springs (Matzinger
et al., 2004). As a result, it will be assumed in the
following that the total underground outflow of
7.8 m3 s)1 from Lake Prespa flows to Lake Ohrid.
Table 5. Water composition at St. Naum springs
Substance
Share
Share from
Standard
from Lake
precipitationa
deviation
respaa [%]
[%]
[%]
18
d O
47
53
6
Table 6. Water flow from Lake Prespa in two spring areas
dD
45
56
4
Average
Share from
Discharge from
Cl)
35
65
8
dischargea
Lake
Lake Prespa
SO2)
4
12
88
4
[m3 s)1]
Prespab [%] [m3 s)1]
Na+
K+
27
30
73
70
6
4
7.5
43
3.2
Ca
56
44
49
Mg2+
68
32
3
Tushemishti
Springs (AL)
2.5
52
1.3
Total
10
45
4.5
2+
a
Phosphorus transfer
For assessing the contribution of this underground
connection to the eutrophication of Lake Ohrid,
the transfer of phosphorus from Lake Prespa to
Lake Ohrid needs to be quantified. As for the
determination of the water origin, we distinguish
between springs in the catchment of Lake Ohrid
with and without a proven connection to Lake
Prespa. For the latter an average TP concentration
SRP
32
68
14
TP
11
89
5
Ratios calculated, based on measured concentrations in Lake
Prespa, in precipitation-fed springs and in St. Naum Springs
using Equation (9).
Spring area
St. Naum
Springs (MK)
a
From Watzin et al. (2002).
From own measurements for St. Naum Springs and from
Eftimi & Zoto (1997) for the Tushemishti springs.
b
104
of 4 ± 1 mg P m)3 was found based on our
sampling (Table 2, Fig. 10). According to this
value and the water loads estimated from stable
isotope and chloride measurements (Tables 5 and
6), a TP concentration of 11 mg P m)3 is calculated for the Lake Prespa water of the St. Naum
Springs, which is significantly lower than in the
lake itself. Based on the oxygen saturation in the
spring water arising on the Lake Ohrid side, we
presume that organic matter is exposed to oxic
mineralization during underground transport.
Accordingly, virtually all P that reaches Lake
Ohrid through underground transport is
bio-available, which is also indicated by the
insignificant difference between measured SRP and
TP spring concentrations, based on a two sample
paired t-test at level 0.05 (Fig. 10). Lake Prespa
outflow obviously passes a coarse filter, such as a
sediment layer, which retains most particles.
Within such a layer P mineralization and incorporation or sorption of SRP seem probable
(Gächter et al., 2004). Moreover, SRP can be
14
adsorbed to mineral surfaces such as Fe-oxyhydroxides, silicates or Ca2+ on its way through the
underground (Diaz et al., 1994; Ioannou &
Dimirkou, 1997; Dittrich & Koschel, 2002).
We conclude that 65% of the TP leaving Lake
Prespa (average of 31 mg P m)3) entering into the
karst underground is retained. We assume this
ratio to be constant for higher P concentrations
as well. This seems plausible for filtering as well
as for adsorption processes, since any solubility
product between PO43) and Ca2+ species will be
dominated by Ca2+ (Ca:P > 1000).
Effect on Lake Ohrid
Given the calculations above, about 20% of the
Lake Ohrid water inflow and 7% of the TP load
originate from Lake Prespa. The recent rapid
water level decline of Lake Prespa most likely is
the result of increased irrigation or net evaporation, rather than of an opening of additional
underground channels. Hence, the water level
Springs without connection to Lake Prespa
St. Naum Springs
12
-3
TP concentration [mg P m ]
10
8
6
4
1:1
2
0
-2
-4
0
2
4
6
8
10
12
-3
SRP concentration [mg P m ]
Figure 10. SRP and TP concentrations measured in St. Naum Springs and in springs without connection from Lake Prespa. Error bars
show the respective measurement errors.
105
1.0
0.5
0.0
Lake Prespa water level as a result of three
competing causes (compare Equation 8). (i) A
smaller subterranean water flow Qout,undgr
(Equation 2) consequently reduces Pinp, (ii) the
decreasing Lake Prespa volume leads to an increase in P concentration in Qout,undgr and finally
(iii) a shorter hydraulic residence time of Lake
Prespa decreases this concentration. The last
point is valid as long as r in Equation (8) remains
constant. This estimate of r was shown to be
reasonable within the lake level range of the past
decade and can be expected to provide sensible
results under a further level decline. However, it
is not known how a dramatic decrease to an
empty lake would influence the phosphorus
transport to Lake Ohrid. In Figure 11, the three
effects (i) to (iii) on the phosphorus transport to
Lake Ohrid are summarized using Equations (2, 4
and 8). For a 20 m decrease in the water level,
the P concentration in Lake Prespa increases almost five-fold, which leads to a 30% increase of
the phosphorus load from Lake Prespa to Lake
Ohrid, although the groundwater outflow
decreases. With further level decline, the pre-
current water level
historic water level from settlements
10
300
8
250
200
6
150
4
100
2
50
Underground outflow
6
3
-1
from Lake Prespa [10 m year ]
1.5
Relative change in Lake Prespa P concentration [-]
Relative change in P load to Lake Ohrid [-]
drop most likely is reducing the underground
flow. Such a decrease has no influence on the
water level of Lake Ohrid. However, its outflow
is reduced, and consequently the hydraulic residence time s of Lake Ohrid increases. Based on
Equation (2), the underground flow can be calculated for different water levels (Fig. 11). The
decline of the Lake Prespa water level since
1960s has reduced the annual Lake Ohrid inflow
by DQ = 68 Mio m3 yr)1. This decrease in
inflow results in a 9% increase in the hydraulic
residence time of Lake Ohrid (1/snew = 1/
sold ) DQ/V). Based on our measured TP concentration in the water column of 4.5 mg P m)3
and a net P sedimentation of 38 t P yr)1 in Lake
Ohrid (Table 2), we found the parameters
r = 0.15 yr)1 and b = 0.77 for Equation (8)
(Table 4). Using the linear P model of Equation
(8) with those parameters, a smaller water exchange leads to a P residence time scale s* of
6.06 yr, which corresponds to an increase in the TP
concentration of Lake Ohrid by merely 0.6%.
However, the P load Pinp to Lake Ohrid does
not remain constant, but changes with decreasing
0
0
0
10
20
30
40
50
Water level decrease of Lake Prespa [m]
Figure 11. Effect of water level decline of Lake Prespa on its underground outflow, its P concentration and the P load to Lake Ohrid
based on Equations (1), (2) and (8). The P load to Lake Ohrid is proportional to underground outflow times Lake Prespa TP
concentration. Note that the figure is only valid if water is extracted from Lake Prespa, but not if more water flows to Lake Ohrid. The
dotted lines indicate that the use of Equations (2) and (3) may no longer be appropriate.
106
dicted concentration in Lake Prespa goes up to
160 mg P m)3 but, as the groundwater flow
declines rapidly, the P load to Lake Ohrid is also
reduced (Fig. 11). This decline in groundwater
flow occurs only under the assumption of an
additional outlet (or a reduced inflow) on the
Lake Prespa side, such as water abstraction for
irrigation or increased evaporation. In the case of
an increased underground outflow to Lake Ohrid,
the load would simply increase with the concentration in Lake Prespa.
Apart from the effects of the level decrease,
the observed increase in P loads to Lake Prespa
has to be taken into account. Adding both
effects we find that the P loads to Lake Ohrid
have increased by about 50% since the 1960s.
Based on Equation (8), this combined increase in
the P load from Lake Prespa will augment the
TP concentration of Lake Ohrid by merely
0.1 mg P m)3. Even if we assume that the P
load from Lake Prespa increases four times, the
total increase of the TP concentration in Lake
Ohrid would still be below 1 mg P m)3. The
transferred P is predominantly in readily
bio-available form, which might amplify its effect
on lake productivity. Still, the eutrophication
potential of the underground connection between
Lake Prespa and Lake Ohrid seems limited,
thanks to the purifying effect of the underground. However, in historic times, when 90%
of the inflow to Lake Ohrid was from direct
precipitation and the karst aquifers, the SRP
input from the latter could have been the major
P source.
Conclusions
Lake Ohrid
Presently, Lake Prespa does not pose an immediate threat on Lake Ohrid, despite seven times
higher TP concentrations and despite the large
share of the Lake Ohrid catchment. The karst
aquifers connecting the two lakes not only mineralize the entering phosphorus to SRP but also
retain 65% of the P load from Lake Prespa. As a
result, even a four-fold increase of the P load from
Lake Prespa would lead to only 20% increase
(+0.9 mg P m)3) of the actual 4.5 mg P m)3 TP
concentration of Lake Ohrid. However, in ultraoligotrophic Lake Ohrid, an increase by 0.9 mg
P m)3 might have significant effects on its fragile
endemic community (Watzin et al., 2002). Moreover, the eutrophication potential of the spring
inflow is increased, since almost all the P arrives as
bio-available SRP.
Historically, the P load from Lake Prespa
could have been important for the formation of
the unique aquatic ecosystem of Lake Ohrid, due
to the constant supply in directly bio-available
SRP. This importance is indicated by specialized
endemic phytoplankton species (e.g., Cyclotella
fottii), which are dominant between 20 and 50 m
depth (Ocevski & Allen, 1977; Mitic, 1985;
Patceva, 2001), exactly where the karst spring
water intrudes during the productive summer
season. This endemic community could change
drastically with a substantial increase in the
spring water P load.
Lake Prespa
The P balances above reveal that Lake Prespa
would need to become eutrophic or even
hypertrophic before it affects Lake Ohrid
significantly. In fact, Lake Prespa is a very vulnerable system, because any additional consumption of water has a direct effect on its water
level, which in turn affects not only the lake
hydraulics but the entire lake ecosystem. A level
decrease alone can cause an increase in the trophic state of Lake Prespa. If the external P loads
increase simultaneously, the two combined processes can amplify. Such amplification is a realistic scenario in the case of further intensification
of agriculture, where water consumption and
fertilization increase in parallel. The recently
observed anoxia in the deeper layers of the lake
will most probably have a significant effect on its
biodiversity.
Future stakes
In the future, P and DO in the water column as
well as the water level of Lake Prespa must be
monitored on a regular basis, as they are essential
indicators for the lake ecology. Moreover, the
implications of these abiotic changes in Lake
107
Prespa on its flora and fauna should be examined.
Regarding Lake Ohrid, occasional measurements
of P concentrations in the springs flowing to Lake
Ohrid would help quantify changes in the P
transfer from Lake Prespa. The main anthropogenic factor influencing Lake Prespa is agricultural
development and in particular the abstraction of
water, application of fertilizers, and enhanced soil
erosion. Thus, it is important that the agricultural
practices in all three riparian states are assessed in
detail and management options to control the
water use and the application of fertilizers are
proposed.
Acknowledgements
This work was only possible thanks to the cooperation between the Hydrobiological Institute in
Ohrid (HBI) and the Swiss Federal Institute of
Environmental Science and Technology (EAWAG). Particular thanks to B. Cakalovski,
M. Krstanovska, Z. Brdaroski, K. Paloski,
M. Schurter, R. Stierli, B. Sinnet, A. Zwyssig,
I. Holderegger, M. Fette and C. Schubert for their
great help in the field and the laboratory. C. Hoyle
and M. Stubbs kindly improved the English of this
manuscript. An earlier version of the manuscript
profited significantly from suggestions by B.
Wehrli and two anonymous reviewers. The work
was financially supported by the Swiss State Secretariat for Economic Affairs (seco) and the Swiss
National Science Foundation Grant 2000067091.01.
References
Anovski, T. (ed.), 2001. Progress in the Study of Prespa Lake
using Nuclear and Related Techniques. Project Report,
IAEA Regional Project RER/8/008, ISBN 9989–650-21-7,
Skopje, Macedonia.
Anovski, T., B. Andonovski & B. Minceva, 1992. Study of the
hydrological relationship between Lake Ohrid and Prespa.
Proceedings of Symposium on Isotope Techniques in Water
Resources Development, IAEA, Vienna, Austria, March
1991.
Anovski, T., M. Kolaneci, J. Milevski, P. Ristevski &
A. Stamos, 2001. Hydrological aspects and water balance of
Prespa Lakes. In Anovski, T. (ed.), Progress in the Study of
Prespa Lake using Nuclear and Related Techniques. Project
Report, IAEA Regional Project RER/8/008, ISBN 9989650-21-7, Skopje, Macedonia, 53–66.
Anovski, T., J. Naumovski, D. Kacurkov & P. Kirkov, 1980. A
study of the origin of water of St. Naum Springs, Lake Ohrid
(in Macedonian). Fizika 12: 76–86.
Berxholi, A., 1997. Human pressure in Prespa basin: problems
and alternatives. Proceedings of the International Symposium Towards Integrated Conservation and Sustainable
Development of Transboundary Macro and Micro Prespa
Lakes, PPNEA, ILAR Typography. Tirana, Albania,
203–206.
Catsadorakis, G. & M. Malakou, 1997. Conservation and
management issues of Prespa National Park. Hydrobiologia
351: 175–196.
Chavkalovski, I., 1997. Hydrology of Prespa Lake. Proceedings
of International Symposium Towards Integrated Conservation and Sustainable Development of Transboundary Macro
and Micro Prespa Lakes, PPNEA, ILAR Typography,
Tirana, Albania, 9–14.
Crivelli, A. J. (ed.), 1996. Action plan for the Dalmatian Pelican. European Commission, http://europa.eu.int/comm/
environment/nature/directive/birdactionplan/pelecanuscrispus.htm.
Crivelli, A. J., G. Catsadorakis, M. Malakou & E. Rosecchi,
1997. Fish and fisheries of the Prespa lakes. Hydrobiologia
351: 107–125.
DEW, 1996. German Standard Methods for Water Analysis (in
German) II, VCH, Weinheim, Germany.
Diaz, O. A., K. R. Reddy & P. A. Moore Jr., 1994. Solubility of
inorganic phosphorus in stream water as influenced by pH
and Calcium concentration. Water Research 28: 1755–1763.
Dittrich, M. & R. Koschel, 2002. Interactions between calcite
precipitation (natural and artificial) and phosphorus cycle in
the hardwater lake. Hydrobiologia 469: 49–57.
Doskey, P. V. & R. W. Talbot, 2000. Sediment chronologies of
atmospheric deposition in a precipitation-dominated seepage
lake. Limnology and Oceanography 45: 895–904.
Eftimi, R., E. Micevski & A. Stamos, 2001. Geological and
hydrogeological conditions of the Prespa Region. In Anovski, T. (ed.), Progress in the Study of Prespa Lake Using
Nuclear and Related Techniques. Project Report, IAEA
Regional Project RER/8/008, ISBN 9989-650-21-7, Skopje,
Macedonia, 11–22.
Eftimi, R. & J. Zoto, 1997. Isotope study of the connection of
Ohrid and Prespa lakes. Proceedings of International Symposium Towards Integrated Conservation and Sustainable
Development of Transboundary Macro and Micro Prespa
Lakes. PPNEA, ILAR Typography, Tirana, Albania: 32–37.
Ernst Basler & Partners, 1995. Feasibility Study on the Lake
Ohrid Conservation Project. The World Bank, Zollikon,
Switzerland.
Gächter, R., S. M. Steingruber, M. Reinhardt & B. Wehrli,
2004. Nutrient transfer from soil to surface waters: differences between nitrate and phosphate. Aquatic Sciences 66:
117–122, doi: 10.1007/s00027-003-0661-x.
Gilbert, J. J. & S. Hadzisce, 1984. Taxonomic notes on the
shallow-water endemic sponges of Lake Ohrid, Yugoslavia
with a description of two new species and a redescription of
Spongilla stankovici. Archiv für Hydrobiologie 99: 331–339.
108
Golterman, H. I., 2001. Editorial – a call for help: the Prespa
Lakes – a call for help in a distressing situation. Hydrobiologia 450: 3–4.
Hadzisce, S. D., 1966. The mixi-phenomenon of Lake Ohrid
over the period 1941/42–1964/65 (in German). Verhandlungen der Internationalen Vereinigung für theoretische und
angewandte Limnologie 16: 134–138.
Hakanson, L. & M. Jansson, 1983. Principles of Lake Sedimentology. Springer-Verlag, Germany.
Hollis, G. E. & A. C. Stevenson, 1997. The physical basis of the
Lake Mikri Prespa systems: geology, climate, hydrology and
water quality. Hydrobiologia 351: 1–19.
Hupfer, M., R. Gächter & R. Giovanoli, 1995. Transformation
of phosphorus species in settling seston and during early
sediment diagenesis. Aquatic Sciences 57: 305–324.
Ioannou, A. & A. Dimirkou, 1997. Phosphate adsorption on
hematite, kaolinite and kaolinite-hematite (k-h) systems as
described by a constant capacitance model. Journal of Colloid and Interface Science 192: 119–128.
Jakovljevic, S., 1935. Limnological research on Lake Prespa (in
French). Verhandlungen der Internationalen Vereinigung für
theoretische und angewandte Limnologie 7: 187–226.
Jordanoski, M., V. Mitic, Z. Krstanoski, G. Kostoski, T.
Stafilov, N. Joseska, T. Naumoski, L. Lokoska, T. Anovski,
V. Novevska, D. Gusevska, M. Talevska, S. Trajanovski, S.
Patceva, E. Veljanoska-Sarafiloska, 2002. The Influence of
the Anthropogenic Factor on the Trophic State of Lake
Prespa (in Macedonian). National Project Report, Hydrobiological Institute, Ohrid, Macedonia.
Karaman, M., 1971. Zoogeographical relations between Lake
Prespa and Lake Ohrid (in Serbo-Croatian). Zavod za
Ribarstvo na CRM, Skopje IV: 1–16.
Kelts, K., U. Briegel, K. Ghilardi & K. Hsü, 1986. The limnogeology-ETH coring system. Schweizerische Zeitschrift
für Hydrologie 48: 104–115.
Löffler, H., E. Schiller, E. Kusel & H. Kraill, 1998. Lake
Prespa, a European natural monument, endangered by irrigation and eutrophication? Hydrobiologia 384: 69–74.
Lotter, A. F., M. Sturm, J. L. Teranes & B. Wehrli, 1997. Varve
formation since 1885 and high-resolution varve analyses in
hypertrophic Baldeggersee (Switzerland). Aquatic Sciences
59: 304–325.
Matzinger, A., E. Veljanoska-Sarafiloska, M. Jordanoski &
T. Naumoski, 2004. Lake Ohrid – a unique Ecosystem
endangered by Eutrophication? International Conference
on Water Observation and Information System for Decision Support – BALWOIS, May 25–29, Ohrid, Macedonia.
Meybeck, M., 1995. Global distribution of lakes. In Lerman,
A. D. Imboden & J. R. Gat (eds), Physics and Chemistry of
Lakes. Springer, Berlin Heidelberg, Germany, 1–32.
Meyers, P. A. & R. Ishiwatari, 1995. Organic matter accumulation records in lake sediments. In Lerman, A. D. Imboden
& J. R. Gat (eds), Physics and Chemistry of Lakes.
Springer, Berlin Heidelberg, Germany, 279–323.
Milevski, J., P. Ristevski, D. Tanusevska, 1997. Influence of
rainfalls on the oscillations at Prespa Lake. Proceedings of
International Symposium Towards Integrated Conservation
and Sustainable Development of Transboundary Macro and
Micro Prespa Lakes, PPNEA, ILAR Typography, Tirana,
Albania, 52–57.
Mitic, V., 1985. Vertical distribution of the biomass of Cyclotella Fottii Hust, in Lake Ohrid during April–November
1981 (in Macedonian). Edition jubilaire, Station hydrobiologique – Ohrid 1: 227–235.
Moosmann, L., R. Gächter, B. Müller & A. Wüest, 2005. Is
phosphorus retention in autochthonous lake sediments
controlled by oxygen or phosphorus? Limnology and
Oceanography 50, in press.
Nastov, A. D., 1997. Implementation of international criteria
for the integral protection of Prespa ecosystem. Proceedings
of International Symposium Towards Integrated Conservation and Sustainable Development of Transboundary Macro
and Micro Prespa Lakes, PPNEA, ILAR Typography, Tirana, Albania, 193–197.
Naumoski, T., V. Novevska L. Lokoska & V. Mitic, 1997.
Trophic state of Prespa Lake. Proceedings of International
Symposium Towards Integrated Conservation and Sustainable Development of Transboundary Macro and Micro
Prespa Lakes, PPNEA, ILAR Typography, Tirana, Albania,
132–137.
Ocevski, B. T. & H. L. Allen, 1977. Limnological studies in a
large, deep, oligotrophic lake (Lake Ohrid, Yugoslavia):
seasonal and annual primary production dynamics of the
pelagial phytoplankton. Archiv für Hydrobiologie 79: 429–
440.
Patceva, S., 2001. The phytoplankton and the chlorophyll a as
indicators of the trophic state of the water in the littoral and
pelagic region of Lake Ohrid (in Macedonian). M.S. Thesis,
University ‘‘St. Kiril i Metodij’’, Skopje, Macedonia.
Ramsar 2000. Lake Prespa. www.ramsar.org.
Redfield, A. C., 1958. The biological control of chemical factors
in the environment. American Scientist 46: 205–221.
Ristevski, P., S. Monevska & B. Popovski, 1997. Characteristics
of temperature, pluviometric and evaporation regime of
Prespa lake basin. Proceedings of International Symposium
Towards Integrated Conservation and Sustainable Development of Transboundary Macro and Micro Prespa Lakes,
PPNEA, ILAR Typography, Tirana, Albania, 58–64.
Sell, J. & Z. Spirkovski, 2004. Mitochondrial DNA differentiation between two forms of trout Salmo letnica, endemic to
the Balkan Lake Ohrid, reflects their reproductive isolation.
Molecular Ecology 13: 3633–3644, doi: 10.1111/j.1365294X.2004.02362.x.
Serafimova-Hadzisce, J., 1985. 50 years since the establishment
of the Hydrobiological Institute in Ohrid, Edition jubilaire,
Station hydrobiologique – Ohrid 1: 49–75.
Shapkarev, J., 1997. Qualitative and quantitative composition
of the macrozoobenthos in Prespa Lake. Proceedings of
International Symposium Towards Integrated Conservation
and Sustainable Development of Transboundary Macro and
Micro Prespa Lakes, PPNEA, ILAR Typography, Tirana,
Albania, 118–125.
Sibinovik, M., 1987. Lakes Ohrid and Prespa (in Macedonian).
University Press ‘‘Kiril i Metodij’’, Skopje, Macedonia.
Skoog, D. A., D. M. West & F. J. Holler, 1996. Fundamentals
of Analytical Chemistry 7th, Saunders College Publishing,
New York, USA.
109
Stankovic, S., 1960. The Balkan Lake Ohrid and its Living
World. Monographiae Biologicae IX, Uitgeverij Dr. W.
Junk, Den Haag, Netherlands.
UNESCO, 1979. Ohrid Region with its Cultural and Historical
Aspect and its Natural Environment. http://whc.unesco.org.
Urban, N. R., C. Dinkel & B. Wehrli, 1997. Solute transfer
across the sediment surface of a eutrophic lake: I. Porewater
profiles from dialysis samplers. Aquatic Sciences 59: 1–25.
Vollenweider, R. A., 1969. Possibilities and limits of elementary
models for lake material balances (in German). Archiv für
Hydrobiologie 66: 1–36.
Watzin, M. C., V. Puka & T. B. Naumoski (eds), 2002. Lake
Ohrid and its Watershed, State of the Environment Report.
Lake Ohrid Conservation Project, Tirana, Albania and
Ohrid, Macedonia.
Weiss, J., 2004. Handbook of Ion Chromatography (3rd edn).
Wiley-VCH, Weinheim Germany.
Werner, R. A. & W. A. Brand, 2001. Referencing strategies and
techniques in stable isotope ratio analysis. Rapid Communications in Mass Spectrometry 15: 501–519.
Wieland, E., P. H. Santschi, P. Höhener & M. Sturm, 1993.
Scavenging of Chernobyl 137Cs and natural 210Pb in Lake
Sempach, Switzerland. Geochimica et Cosmochimica Acta
57: 2959–2979.