Site description

Lake Kinneret watershed is characterized by a typical Mediterranean climate of hot dry summers (June–October) and cool rainy winters (Nov.–March). The watershed (area of ~2600 Km²) extends mostly to the north of the lake, up to the Hermon Mountains, with the Hula Valley at its center (Fig. 1). A major source of water for LK is the northern branch of the Jordan River, which supplies ca. 450 x 10⁶ m³ year^-1. Three tributaries feed this river; Dan (~235 x 10⁶ m³ y^-1), Banias (~120 x 10⁶ m³ y^-1) which emerge from springs located at the southern margins of the Hermon mountainous area and the Hatsbani (~70 x 10⁶ m³ y^-1) which emerges from springs located west of this mount. The three tributaries converge at the northern part of the Hula Valley to form the Jordan River. Additional water sources to this lake include direct rain (~70 x 10⁶ m³ y^-1), smaller streams not within the Jordan River catchment and saline water springs. The Meshoshim stream (Fig. 1) is an example of a small perennial stream (~25 x 10⁶ m³ y^-1) that drains 160 Km² of basaltic terrain in the Golan Heights⁷ and flows into the lake from the north-eastern side.

Figure 1

Lake Kinneret watershed location map (left) and schematic view (right) of the watershed.

Anthrophogenic Preturbations in the watershed

Until the early 1950s, the southern and central parts of the Hula Valley were covered by the shallow Lake Hula and adjacent marshes (Fig. 2). These water bodies served as a wetland for Lake Kinneret, which is located further downstream, retaining some of the nutrients that originated in the northern part of the watershed. In the 1950s, Lake Hula was drained requiring the removal of 1.5 x 10⁶ ton peat from the bed sediments of the marshes,⁸ and later scattering this over parts of the valley.

Figure 2

The Hula Valley with historic Lake Hula (light blue) & Marshes (dark grey).

The drainage of the Hula was performed in order to increase arable land, to eradicate malaria and to reduce evaporation losses.^9,10 The soils exposed were mostly peat (~26 Km²) underlying the marshes^11–1213 and mineral soils (mostly marl) underlying the historical lake (~14 Km²:.¹⁴

Until the early 1970s these farm land were irrigated by sub-surface flooding, thus maintaining soil-water levels at more or less the depth of the vegetation root zone. Fertlizers introduction to the soils, particularly phosphorus, was increased during he early 1970s.

The drainage of the Hula valley, and the subsequent agro-engineering activities resulted in exposure of the peat to atmospheric oxygen thereby accelerating its biodegradation.¹⁵ Cotton was grown over most of the valley (an area of up to 60 Km²), as a mono-culture leaving the ploughed soils denuded of vegetation for ~5-6 months of the year (October through March). The combination of both degradation of peat and exposure of surface soils resulted in extensive erosion and heavy dust storms. As long as the groundwater table was maintained at the depth of the vegetation root zone (some 30-50 cm) this degradation was restricted to the top soil layer. Surveys made after drainage showed that the soils contained gypsum (CaSO₄*2H₂0; 10%-20% of the dry weight) and an abundance of iron oxides which are products of processes associated with the aerobic decomposition of the peat.^16–1718

Toward the end of the 1970s, the combined effect of the decreasing cotton prices with the poor quality of the soils, made cultivation of part of the peat soils no longer profitable and local farmers gradually abandoned them. In the 1980s these soils were covered by a succession of dense natural vegetation, including phragmites reeds and as a result, the frequency and strength of dust storms declined.

Monitoring Stations, Sampling and Analytical procedures

Lake Kinneret (LK) data was collected at St.A, the central deepest part of the lake. The four monitoring stations in the Jordan River catchment are located on the main river (at the Huri Bridge: St. H.) and on the three tributaries of the river, the Dan, the Banias and the Hatsbani. Another station was located away from this catchment, and on the Meshoshim stream, which flows directly from the Golan Heights to LK (Fig. 1). St. H was sampled daily, whereas the other stations were sampled, weekly.

The analytical techniques common for both databases used here are summarized in Table 1. Some differences between the sampling procedures in each monitoring program were noted. LK samples were collected with a Rhode sampler, stored in 1 L plastic bottles and brought back to the lab within 2-3 hours. Oxygen was fixed on board (Winkler reagents I+II) in a BOD bottle and samples for pH measurements were taken in chilled opaque BOD bottles. Both parameters were measured within 2-3 hours of sampling. Upon arrival in the lab, sub-samples were filtered (0.45 µm and GFF for soluble species and for suspended matter, respectively) and stored at ~4 ^°C until further analysis. The analysis of Soluble Reactive Phosphorus (SRP) and NH₄ was performed within 5 hours of sampling while NO₃, organic nitrogen (total and dissolved), total N and total phosphorus, were determined on the following day.

Table 1

Lake Kinneret Database (LKDB) and Mekorot Water Company parameters used in this study.

Sampling of the watershed was performed manually, with a bucket lowered from a nearby bridge into the middle of the stream (at about 1 m depth). Water samples were stored in closed and cooled 1 L plastic bottles. pH measurements were performed in the lab in sub-samples drawn from the original 1 L bottles several hours after sampling. The remaining samples were stored at 4 ^°C for further analysis, within 48 hours of sampling.

Sediment samples taken from a pelagic core were analyzed for total sulfur (S_T), total iron (Fe_T) as well as for few additional trace metals using ICP-AES, (Geological survey, Jerusalem) after solubilization of the sediment in a sodium peroxide sinter.

Monthly average data presented for chemical parameters measured in the epilimnion of LK represent an average of 16 water samples taken weekly at 1,3,5 and 10 m depth.

Monitoring of the watershed

Station H is a major monitoring station for the LK watershed as it is located on the Jordan River at the southern end of the Hula Valley (Fig. 2) and thus represents the combined contribution from most of the streams in the watershed. In the 1970s, the monthly average concentration of Norg at St.H (~1.3 mgN/L, Fig. 3A) was higher than that measured during the 1980s and 1990s (~0.4 mgN/L). The long-term pattern of pH is a mirror image (Fig. 3A) to that of N_ORG, characterized by lower levels (ave. 7.95) in the 1970s with higher levels (ave. 8.25) in the following decades.

At St.H, long-term trends for the concentrations of other parameters such as suspended matter (SM), particulate phosphorus (PP), total iron and total organic carbon (C_ORG) (Table 2) were similar to those of N_ORG. The multi annual average concentration of SM, PP and Fe dropped from 75, 0.17, 0.08 mg/L during the 1970s to 40, 0.1, 0.02 mg/L, respectively, during the 1980s. However, SRP showed a different temporal pattern, characterized by a gradual increase during the 1970s and a leveling off during the 1980s.¹⁹

The major tributary of the Jordan River is the Dan stream, which flows north of the Hula Valley (Fig. 1). However, the long-term patterns of N_ORG, pH and C_ORG in this stream (Fig. 3, bottom and Table 2) were similar to those at St.H (Fig. 3, top). ie, higher N_ORG and C_ORG and lower pH levels in the 1970s, compared to the ensuing decades. Again, both organic constituents and pH showed mirror image profiles, suggesting a general tendency that when N_ORG (or C_ORG) increases, the pH decreases and vice versa. Similar long-term patterns have been observed in the Banias and the Hatsbani streams (Table 2), during the transition from the 1970s to the 1980s. The pH rise observed in the Hatsbani was somewhat smaller than in the other two streams. In each decade, however, the concentration of N_ORG (and of C_ORG) at St.H has been much larger than that in the northern tributaries (Table 2). This suggests that most of the organic constituents detected in St.H, were acquired by processes occurring within the Hula Valley (Fig. 1).

The Meshoshim stream (Fig. 1) drains 160 Km² of basaltic terrain in the Golan Heights⁷ and is not included within the Jordan River catchment. Nevertheless, the long-term pattern of N_ORG in this stream (Fig. 4) was similar to the Jordan River catchment streams (Table 2 and Fig. 3). In addition, the Meshoshim stream exhibited a phenomenon that had not been apparent in the catchment streams, sulfate concentrations were three fold higher during the 1970s than in the following decades;(~9 and ~3 mg SO₄ L^-1), respectively (Fig. 4).

Monitoring of LK

The running average (±36 weeks) of weekly epilimnetic data (depths; 1, 3, 5 and 10 m) for N_ORG species in Lake Kinneret (LK) is shown in Fig. 5. Total N_ORG revealed a long-term pattern similar to that of the watershed streams, with about 0.8 mgN/l in the 1970s compared to 0.4 mgN/l in the ensuing decades. In the lake both total and soluble N_ORG were measured, while articulate N_ORG concentration was calculated by subtracting the soluble form from the total. It thus became possible to distinguish between long-term trends of both species. The long-term pattern of soluble Norg resembled that of total Norg while particulate N_ORG did not show such a long-term pattern.

During the 1970s, soluble N_ORG was two fold larger than particulate N_ORG, ~0.52 and ~0.26 mgN/l, respectively). In the mid-1980s, with the drop in the level of total N_ORG, this difference became less apparent. However, there were distinct short-term seasonal peaks in the concentration of the particulate form (week or two a year) that appeared mostly in April, during the spring bloom.

Routine measurements of C_ORG in LK begun in 1978. The long-term profile of C_ORG showed somewhat higher levels in 1978 and 1979 (~5 mg/L) compared to the following decades (Table 2). Similar to the streams, also LK epilimnion reveals lower pH levels in the 1970s, as compared to the 1980s.²⁰ In the case of LK where dissolved oxygen was also monitored, lower epilimnetic DO levels were also recorded in the 1970s (Fig. 6) as compared to the 1980s and 1990s.

The Spring-Summer 1974 event

The seasonal dynamics of N_ORG in LK epilimnion during the 1980s can be best represented by the 1987 profile (Fig. 7, top). It indicates relatively high winter-spring levels, approaching ~1 mgN L^-1, while the crash of the spring bloom had been followed by a decline to relatively low levels of ~0.25 mgN L^-1, in both particulate and soluble forms. The respective profile for 1974 (Fig. 7, bottom) is however different for the summer in which an abrupt rise from ~0.7 mgN-N_ORG^-1 to ~1.3 mg L^-1 is shown in June, particularly in the soluble form. This steep summer rise is equivalent to an increase in the overall lake inventory of approximately 2400 ton N, which cannot be accounted for by the concomitant one order of magnitude smaller riverine inflows. The summer of 1974 was also characterized by a noticeable rise in the turbidity (not shown) and in Corg concentrations in LK. Special profiles taken on the 16th and the 23rd of June 1974 revealed extremely high concentrations of Corg all along the water column, with an average of 23.4 ± 2.2 mg C_ORG L^-1. This was a seven-fold higher concentration than the typical mid 1980's levels of 3 mg L^-1. Consequently, in June 1974, the amount of C_ORG in the lake was estimated at about 92.000 tons as compared to typical inventories in the 1980s and 1990s of only 12,000 tons, suggesting an unexplained C_ORG excess of 80,000 tons. Again, judging by the levels of Corg in the Jordan river (St.H) this huge reservoir of Corg in LK cannot be attributed to riverine inflows. The size of this reservoir was comparable to the annual primary production estimated for LK in the 1970s, which was in the order of magnitude of 100,000 tons carbon.²¹

Table 2

General information on monitored parameters and typical decadal values in mg/L. Watershed parameters taken from MEK data base (Y. Geifman), LK parameters from LKDB.

Figure 4

Monthly average concentration of Norg and SO₄ (Runing average ±12 months) in the Meshoshim stream during 1971-1999.

As riverine inflows are ruled out dust is the only other possible source of organic constituents to the streams and to LK. In the Hula Valley 25 kilometers to the north of LK, the last week of March 1974 was characterized by an extreme dust storm, induced by a strong easterly wind, which lifted up a 5 cm thick layer of the local peat soils. A significant rise in Norg was noticed right after the storm in St.H, located on the Jordan River at the southern part of this valley (Figs. 1 and 2). About three months later, in June–July, there was a significant rise in Norg in the northern tributaries (Fig. 3B and Table 2) and in LK (Fig. 5).

Figure 5

Lake Kinneret surface water (1 m depth) running averages, ±36 weeks, of total Norg (solid line: 1970-1999), DON (dashed line, 1972-1999) and PON (thin line, 1972-1999).

The prevailing directions and timing of wind events in the LK drainage basin may well be suited for the distribution of dust formed in the Hula Valley. In spring and fall, easterly and southerly winds predominate in the region. The easterly wind prevailing in March 1974 probably carried peat dust toward the Naftali Mountains range, bordering the west of the Hula Valley (Fig. 1). The southerly winds that followed may have carried the dust northward. In summer and fall, the predominant regional wind is from the north-west may have carried the dust directly to LK and to the Meshoshim watershed. Thus, the prevailing wind directions may also support the dust supposition.

Figure 6

Monthly average concentration of DO measured in the 1 m depth layer of Lake Kinneret during 1970-1992. Solid line represent running average of ±12 months.

The long-term record of SO₄ in the Meshoshim stream provides an additional support to the dust supposition. During the 1970s (Fig. 4) SO₄ levels were three fold higher (~9 vs. ~3 mg SO₄/L, respectively) than in the subsequent decades. A most probable source for the additional SO₄ in this freshwater stream was the dissolution of Aeolian gypsum particles, which are a major constituent of Hula peat.¹⁸ The reason why a similar affect had not been documented in the larger northern tributaries of Banias and Dan was perhaps their much higher and seasonally variable background SO₄ levels, ranging between 10 and 50 mg/l in the Banias tributary, and therefore masking the smaller contribution from the dissolution of gypsum.

Figure 7

Weekly Norg species concentrations in LK surface water (1 m depth) during 1987 (top) and during 1974 (bottom).

According to Brenner et al¹³ dry peat in Hula Valley (soil surface layer) contains between 15% to 29% C_ORG and between 0.8% to 1.2% N_ORG, with average ratios (w/w) of 4.5 and 21 for dry peat/C_ORG and C/N, respectively. Thus, the excess Corg inventory estimated in LK is equivalent to 330,000 tons of dry peat.

Most of the excessive N_ORG detected in LK water during the 1970s appeared in the “soluble” form. This suggests that within several after the deposition of the peat dust on the lake surface, the particles would disintegrate to sizes small enough to pass the 0.45 µm filter used to distinguish between particulate and soluble phases. It should be noted that loads of DIN (mostly NO₃ and less NH₄) in the watershed did not show any significant long-term trends between 1970 and 1990.

Relationship between Organic Material, ph and do levels

The inverse temporal relationship detected between the concentrations of organic constituents and pH, and in the case of LK, also of DO, may be given a straightforward qualitative explanation. The general form of the photosynthesis (Pr) and Respiration (Re) equation²² is given by:

Sources of organic matter may be both autochthonous and allochtanous. Upon the introduction of allochtonous biodegradable organic matter to an aquatic system Re is expected to increase and consequently DO is expected to decrease. In a carbonate buffered system, such as that of LK and the streams,¹ the pH is expected to decrease due to the CO₂ released during respiration. LK shows considerable daily, and seasonal variations in both DO, and pH levels in response shifts in balance between production and respiration. As an example of daily variations, the pH in LK surface water during the peak spring bloom may vary between ~8.8 during the night to ~9.6 at noon, while the respective DO levels vary between 120 to 200% saturation (LKDB-Russ profiler data).

Despite the apparent rise in winter and summer algal biomass in the early 1980s,^5,23 primary productivity in LK (¹⁴C uptake) did not show a parallel rise.² In fact, Pr, which has been measured continuously since 1970, does not show distinct long-term trends at all.²⁴ It is therefore concluded that the rise in DO and pH which was recorded during the 1980s resulted from smaller respiratory fluxes as compared to the previous decade. In other words the decline in inflow loads of allochthanous organics was sufficiently large to lead to the higher levels of both pH and DO in LK epilimnion.

Evidence for Peat Deposited in lk Bed sediments

Evidence for the occurrence of massive peat erosion in the Hula Valley during the 1970s is not only suggested from summer peaks in levels of N_ORG in LK in 1974, 1976, 1977 and 1981 (Fig. 5) as well as through elevated levels during the 1970s in the streams (Figs. 3 and 4). The indication we have for peat transport to LK, prior to the initiation of the routine monitoring in the 1970s, is obtained from a pelagic sediment core, that was sampled in November 1993. Sub samples from this core were processed by Dubowski⁸ who dated (²¹⁰Pb) and measured Corg concentrations (Corg-m; shown here in Fig. 8), and Δ¹³C-Corg values in this core. Another set of sub-samples was used here to analyze the respective profiles of sulfur and iron (Figs. 8 and 9). The dating of this core was therefore adapted from Dubowski et al²⁵ but with one additional date attributed here to a layer deposited in 1982. A “green” lamina containing an abundance of chlorophyll preserved within diatom cells (Melosira aulccelera) was identified between 10.2 and 10.4 cm depth. This lamina is assumed to represent sedimentation in 1982 because the only large bloom of this alga in LK between 1969 and 1988 was documented in winter 1982.²

Apparently, there is a large difference between the Δ¹³C (C_ORG) values in the upper layers of the Hula peat of -18%o¹⁶ and of autochthonous C_ORG from LK, of -28 and -29%o.²⁶ Duboswki et al²⁵ had shown that the 25 cm depth layer in this core represented sedimentation in the early 1950s and that Δ¹³C (C_ORG) levels below this depth were ~-28%₀ while above this depth the respective levels were in-between these two end members. The upper layer is hence assumed to represent mixtures of Corg from both sources. As an example, in the 13 cm depth layer, which was deposited in the mid 1970s, C_ORG concentration was 4.3% (Fig. 8) and Δ¹³C (C_ORG) equal to -24%o. Transferring the Δ¹³C notation into relative concentrations of ¹³C and ¹²C, and considering the Δ¹³C values of the above two end members, it is estimated that 32% of the C_ORG in this layer, ie, 1.38% of the sediment dry weight, originated from the deposition of un-altered Hula peat (C_ORG-p), while the remainder 2.94% of the Corg represents autochthonous sources. Applying the same procedure all along this core (Fig. 8) suggests that in LK C_ORG-p first appeared during the drainage of the Hula and that later its concentration gradually increased, with peak levels in the mid 1970s followed by a decline to lower levels during the 1980s. The temporal pattern of C_ORG-p after 1970 thus resembles the respective pattern of N_ORG in LK water and in the watershed, presented above for the period between 1970 and 1993 (Figs. 3, 4 and 7). Notice that the concentration of C_ORG-p in the surface sediment layer (Fig. 8), which represents sedimentation in 1993, is estimated at 0.93%, hence significantly higher than that of the pre-1950s level of 0.65%. This suggests that even in 1993 peat derived C_ORG was still accumulating in LK pelagic bed sediments

Figure 8

Corg and Fe concentrations (dw) in core St.A, Lake Kinneret pelagic zone. Corg_p: Corg of unaltered peat calculated from measured Corg and Δ¹³C data p(Dubowski et al 2003). Fe_T - measured by AES after Na-peroxide sinter dissolution, Corg-m measured Corg. Sediment deposition time adapted from Dubowski et al (2003).

Following the seasonal oxygen depletion, the degradation of organic mater in LK hypolimnion is mostly through bacterial sulfate reduction (SRB:.^27,28 As indicated from the activity of sulfate reducing bacteria,^27,29 from sedimentary profiles of sulfide in pore-water³⁰ and from the sulfur content of the pelagic bed sediments.²⁸ SO₄ reduction occurs not only in the hypolimnion but also in the upper ~5 cm layer of the bed sediments. The source of the sulfur is SO₄ diffusing from the overlying lake water. Sulfide formed is entrapped in the bed sediments through pyritization, initially as mackinawite, FeS_0.9, which later converts to pyrite, FeS₂ in the LK pelagic zone.³⁰ The sedimentary profile of sulfur may therefore supply additional information on the past deposition flux of Corg.

Total sulfur (S_T) concentration in subsamples taken here from the core varied considerably. Below 25 cm depth (Fig. 9), ie, before the Hula drainage, it was relatively low and stable, at 0.65 ± 0.03%. The layer above showed a steep raise in S_T to a maximum of 2.25% at 13 cm depth, which represents the mid-1970s and marks a peak in Corg-p (Fig. 8). The uppermost layers were characterized by a decrease in S_T, approaching c. 0.96% in the sediment surface layer.

Sulfur in the bed sediments of LK is assumed to have two potential sources; sulfur accumulated due to activity of sulfate reducing bacteria (SRB), associated with the decomposition of either autochthonous or allochthonous organic matter followed by pyritization, S-_SRB, and sulfur bound to un-altered Hula peat, S_p. Hula peat is enriched with biogenic sulfur, with S concentrations several fold that of the precursor vegetation of phragmites, having an average C/S ratio of 6.15 (w/w).^13,17 Using this ratio for the peat deposited on the LK floor the S_p profile in this sediment core was estimated from the respective C_ORG-p data. S_-SRB, was then calculated by subtracting S_p from S_T (Fig. 9).

Figure 9

Core St.A, LK, Sedimentary profiles of total sulfur (in percentage dry weight), S_T (squares), formed in-situ through SRB, S_-SRB (triangles) and S associated with un-altered peat, S_p (filled circles). The analytical error for the determination of S_T is estimated at ±5%. The dating of layers deposition adapted from Dubowski et al (2003).

The expected Corg/S molar ratio in SRB is 2:1.²² Hence organic carbon that had been removed from the sediment column through SRB activity followed by pyritization, C_ORG-S, can be estimated from S-_SRB data. However, C_ORG-S may not represent all of the sulfur formed due to SRB activity because it takes into account only S entrapped as FeS_x while some of the S may have migrated through H₂S diffusion to the overlying lake water. Another mechanism that may cause the removal of Corg buried in LK bed sediments and is not accounted for in C_ORG-S is through methanogenesis.²⁸ Accordingly, the sum of Corg-m and Corg-s should be smaller than the original flux of Corg buried in the bed sediments.

Pyritization in the sediments may proceed as long as there is a sufficient amount of sedimentary iron oxi-hydroxides to interact with the sulfide formed through SRB and to trap it. In the case of LK pelagic sediments, the concentration profile of total iron, Fe_T (Fig. 8), in the aforementioned core increases from 2.9 at 22 cm to a maximum of 3.4% at 13 cm and it then decreases to1.6% at 3 cm depth. ie, in layers deposited after the Hula drainage and until mid 1970s it increased but decreased later during the 1980s. In the upper 13 cm layer a significant positive correlation was found between Corg-p and Fe_T (R² = 0.92, P < 0.001): Fe_T = 2 x Corg-P + 0.28. Considering the abundance of Fe-oxides in the Hula peat soils^18,31 and the long-term reduction in Fe_T levels in St.H (Table 2) it is suggested that the decline in LK sedimentary Fe_T also resulted from the reduced loads of particles originating in the Hula valley.

In the upper 13 cm sedimentary layer S_SRB correlates with Fe_T (R² = 0.78, P < 0.001: Fe_T = 1.24 S_SRB + 078, both in % dw). In the molar scale the slope obtained here is equal to 2.15, close to the expected stochiometry of FeS₂. Assuming that FeS₂ is the solid phase which represents the highest possible ratio of S_SRB/Fe, this slope suggests that iron oxide resources controls S_SRB accumulation in LK bed sediments. The long-term decline in Fe sources to LK therefore allowed for a larger portion of the H₂S formed in the bed sediments through SRB activity, to release to the overlying lake water during the 1980s. The long-term increase in the hypolimnetic inventory of H₂S between 1970 and 1992²⁰ that can not be attributed to increased sedimentation of organic matter to the hypolimnion because there was no concomitant rise in primary productivity,²⁴ may support this hypothesis.

Synthesis and implications

The drainage of the Hula wetlands in the 1950s led to two direct consequences. Firstly, it removed the shallow water bodies (volume approx. 50 x 10⁶ m³) that previously restricted downstream transport to LK of nutrients that originated in the northern part of the watershed. This is shown from the abundance of vegetation and peat that characterized those marshes prior to drainage. Overall, these shallow Hula water bodies may have been more efficient as traps for nutrients during the summer, which is characterized by much smaller river discharge. Secondly, the drainage of these wetlands resulted in the formation of new arable, heavily cultivated and fertilized land, which contributed to bioavailable phosphorus loads to LK.³² These observations probably form the basis for Dubowski's general claim that the drainage of the Hula basin greatly increased primary production in LK:⁸ in abstract).

As indicated from the C_ORG-p concentration profile, the LK pelagic sediments, peat erosion began concomitantly with the drainage of the Hula. Two decades later, erosion increased considerably due to unsuitable agricultural practices including the growth of cotton as a monoculture and the poor management of the groundwater table. from the conclusions drawn for 1974, the predominant transport mechanism of the eroded peat was by dust, which was efficiently scattered all over the drainage basin, including LK itself. Peat decomposition in the oxidized water bodies resulted in the decline of DO and pH levels of the ambient water. When peat supply became restricted, as in the 1980s, both pH and DO levels raised back close to what might have been the steady state levels prevailing in these water bodies prior to the Hula drainage.

The 1970s to 1990s witnessed a twofold decline in zooplankton biomass in LK^3,4 resembling the respective temporal pattern of Norg, and suggesting the possibility of a link between both. During these two decades phytoplankton composition reminded relatively, stable with Peridinium predominance in spring and green algae predominance in summer-fall. Hence, the long-term decline in zooplankton can not be attributed to a change from more eatable to less eatable algae. The decline in zooplankton can neither be attributed to to a parallel decline in the primary production in LK because no such decline was recorded.²⁴ Zooplankton respiration in LK between 1970 and 1990 was calculated²⁰ by considering the biomass (LKDB), average epilimnetic water temperature (LKDB) and taxon-specific rates, based on published information.^33,34 It showed that the mean decadal oxygen respiratory consumption by zooplankton during the 1970s was ~40% of oxygen production through photosynthesis, while the respective respiratory mean for the 1980s was only 20%. The later is similar to the average of 17.5 ± 5% between 2001 and 2007 estimated by Berman et al.³⁵ A most probable explanation for the decadal drop of 20% in zooplankton respiration is therefore that during the 1970s about half of the zooplankton activity was supported by allochthonous peat resources.

The decline in particulate phosphorus (PP) associated with the reduction in peat resources toward the 1980s (Table 2) would have had a minor effect on bio-available P inflows to LK because most of the PP in the Hula peat appears as some form of hydroxy or fluoro-apatite.³¹ In the Jordan River, soluble P (SRP) accounts for almost half of total P (MEK database) and is considered the major source of bioavailable P to LK.³⁶ The 1970s gradual rise in SRP loads in the Jordan River¹⁹ was suggested to trigger the increase, of algal biomass in LK during the early 1980s.^21,23 It is hypothesized here that not only the long-term rise in SRP but also the decline in zooplankton, that led to reduced grazing pressure, affected the 1980s algal growth.

Another possible factor in controlling long-term algal activity in LK is related to the available nitrogen sources. During the 1970s and 1980s the mean annual inflow loads of DIN to LK were relatively stable while the corresponding loads of N_ORG declined from ~570 tons during the 1970s to ca. 200 tons, during the 1980s.²⁰ The respective drop in decadal mean inventories of Norg in LK was from 3000 ± 500 to 1400 ± 400 tons (LKDB). Owing to the reduced inflows monitored after the 1970s, part of the drop in LK may have been due to a shift in balance between Norg removal, through water pumping, and the inflows. An additional cause for the removal of Norg from LK could be that which is related to the biodegradation of the allochthonous peat, suggested from the inverse relationship between DO and pH vs. allochthanous Norg. However such mechanism of mineralization of organic matter is expected also to lead to a parallel long-term decline in NH₄ (and for NO₃ through nitrification), but no such trends were recorded. A possible explanation may be that the biodegradation of the allochthanous Norg occurred through some sort of a bypass over the NH₄ pathway, where by a portion of this Norg was directly consumed by bcateria³⁷ and zooplankton.

The long-term decline in aeolian transport of peat resulted in a general decrease in the N/P ratio in the inflows to LK. A concomitant decline in the ratio between PON to PP in the LK epilmnion was suggested³⁸ to have triggered blooms of the N₂ fixing blue-green cyanobacteria Aphanizomenon Ovalisporum since 1994.⁶