Towards a quanTiTaTive reconsTrucTion of lake TroPhic sTaTe in TemPeraTe lakes usinG subfossil cladocera and diaToms: comPosiTion of a TraininG seT from ne Poland

We present a training set, the database involving physical-chemical water parameters together with the subfossil Cladocera and diatoms community composition in the surface sediments of 64 postglacial lakes in NE Poland sampled along a wide trophic gradient (from oligo- to highly eutrophic). The most important water parameters measured in water were chlorophyll- a , electrical conductivity (EC) and oxygen concentration. In ad-dition, total phosphorus (TP) and Secchi depth (SD) were determined for the surface water layer. The data col-lected will be used to calculate a transfer-function for quantitative reconstruction of trophic state in freshwater temperate lakes.


Introduction
Lakes are important components of postglacial landscapes, sensitively reacting to environmental changes from local to global scale. Lake catchment features such as land cover and land use have a profound impact on the amount of nutrients reaching lake water and therefore they are very influential in terms of lake trophic status. The global climate changes also play a great role in shaping the habitats for water organisms as well as physical and chemical processes in lakes. The information on the current and past states of the lake ecosystem is preserved in lake sediments. By using a plethora of analytical methods including paleoecology, geochemistry, stable isotopes etc. paleolimnologists acquire this information and translate it into reconstructions of past environments on different time scales. In Central and Eastern European Lowlands lake bottom sediments have been accumulating since the termination of the last glaciation and consequently sediment cores from the deepest parts of lake basins often contain a high-resolution record of environmental changes since the fall of the last glaciation (Ralska-Jasiewiczowa et al., 1998;Lauterbach et al., 2011;Apolinarska et al., 2012;Zawiska et al., 2018).
The remains of planktonic organisms such as Cladocera and diatoms are known for being reliable paleoecological bioindicators (Battarbee et al., 2001;Korhola & Rautio, 2001). They respond sensitively to different environmental parameters such as pH, salinity, trophic state, water depth etc. and are very common in the sediments. The classical way of interpreting results of paleolimnological analysis is based on indicative species as well as a full assessment of the species composition of the entire subfossil communities (Korhola & Rautio, 2001) to perform qualitative reconstructions of the past environments. However, in order to get quantitative information on past environmental changes from the species composition preserved in the sediment cores, it is necessary to find relationships between environmental factors and the population composition. This approach has been successfully applied in paleolimnology during the last decades (Chalie & Gasse, 2002;Lamentowicz et al., 2009;Larocque-Tobler et al., 2015;Zawiska et al., 2015;Rzodkiewicz, 2018). Calculating a transfer function requires a good quality training set, consisting of lakes covering a wide range of values of parameters that will be reconstructed from the sediment cores. Such a training set act as a database involving in-situ and/or laboratory measurements of present-day environmental parameters and the composition of corresponding local biocenoses (Juggins & Telford, 2012).
There are few existing training-sets for Central and Eastern Europe for the diatoms (Sienkiewicz & Gąsiorowski, 2017;Rzodkiewicz, 2018;Sienkiewicz et al., 2021) and Chironomidae (Kotrys et al., 2020). However, so far there is no Cladocera-based training set from temperate latitudes. The existing ones were created in the regions with different climatic and geological conditions as well as vegetation and landuse changes, which made it unsuitable for paleoenvironmental reconstructions in temperate European latitudes (Brodersen et al., 1998;Lotter et al., 1998;Chen et al., 2010;Davidson et al., 2010;Nevalainen et al., 2013). In Poland the first attempt to bind subfossil Cladocera composition with environmental parameters was made for the group of dystrophic lakes in the Wigierski National Park (Zawisza et al., 2016). The ongoing research is centered on creating a Cladocera-based transfer function for reconstructing trophic state in freshwater temperate lakes. The first step was to built a training set for the above application. The current paper aim is to provide the information on the composition and structure of this training set. The training set encompasses a hydrochemical and ecological (primarily diatom and Cladocera composition) data from a number of lakes throughout the NE part of Poland. Lakes involved in the database represent a wide range of trophic status (from oligo-to highly eutrophic) which makes the dataset representative for the temperate lakes as a whole. The eutrophication processes in lakes in northeastern Poland are climate-and humaninduced (Marszelewski, 2005) and it is important to determine the reference conditions to provide scientific advice for restoration programs applied to these ecosystems (Bennion et al., 2011;Luoto et al., 2013).

Study area
This study involves 64 lakes distributed over the area of around 30,000 km 2 , located in NE Poland and at the transition between Mid and East European Plains. The lakes are located between Polish/Lithuanian/Belarussian/Russian border and the Vistula river valley (Fig. 1). The landscape in this part of Europe is dominated by glacial landforms from the Pomeranian phase of the Weichselian glaciation (Marks, 2012) and consequently, the lakes are of glacial origin (Pochocka-Szwarc, 2010;Kondracki, 2013). The Late Weichselian maximum ice sheet limit in Poland was timetransgressive and occurred at 24-19 kyrs BP becoming younger to the east (Pochocka-Szwarc, 2010;Marks, 2012). The geological substrate for lakes in this region consists of glacial till and varigrained glaciofluvial clastic sediments (Pochocka-Szwarc, 2013). The climate in the area is temperate, the average July temperature is 19.0 °C, while in January is between -2.7 °C in the southern Warmia to -3.4 °C in the Suwałki Lake District. The average precipitation is from 699 to 734 mm/yr, most of which occurs during the summer (pl.climate-data.org). The lakes of study are small to medium size and moderately deep with only a few having maximum depth below 10 m (Tab. 1). They also differ in terms of catchment lithology, land use and trophic state (Tab. 1). The latter displays gradient from oligotrophic via eutrophic to dystrophic lakes (Jańczak, 1999;Jekatierynczuk-Rudczyk et al., 2014).

Methods
Each lake was sampled once during the summer field campaigns in July 2018, 2019 and 2020. Epilimnion water samples were collected 1 m below the water surface with UWITEC water sampler of 2 liter capacity, from the deepest part of the lake. Samples were stored unpreserved in darkness in a cool room until chemical analysis. The several water parameters were measured in laboratory: total phosphorus (P tot ), sulfates (SO 4 2-), bicarbonates (HCO 3 -), chlorides (Cl -), calcium (Ca 2+ ), magnesium (Mg 2+ ), potassium (K + ), sodium (Na + ).
Along with water sampling the Secchi depth (SD; m) was assessed using a standard disk. On-site measurements at 1m depth of physical-chemical parameters (lake water temperature (t), electric conductivity (EC), total dissolved solids (TDS), dissolved oxygen (DO), chlorophyll-a (Chl-a), phycocyanin (PC)) Table 1. Physical features (range) of the lakes studied. Morphometric parameters were taken from The Atlas of Polish Lakes (Jańczak, 1999). Catchment characteristics were obtained from Corine Land Cover (CLC) 2018 (Corine, 2018) and Geological Map of Poland (GMP) 1:500,000 (Marks et al., 2006) with the procedure described by Jasiewicz et. al. (2022)   were performed using a multiparameter probe EXO 1 by YSI calibrated and checked with certified reference material (Harbour water, NWHAMIL-20.2). The reference temperature used for calculating temperature compensated electric conductivity is 25°C. In order to get information on the temperature and oxygen stratification in each lake measurements were conducted in the entire water column every 1 m. The stratification profiles of oxygen, temperature, chlorophyll-a, electric conductivity were prepared using C2 freeware (Juggins, 2007).
The sediment samples (upper 4 cm) which were used to assess sedimentary species composition of Cladocera and diatoms were also taken from the deepest part of the lake, alike the water sample, using Limnos sediment corer. The deepest point is regarded as representative for the whole lake for limnological and paleolimnological studies (Apolinarska et al., 2020;Davidson et al., 2010;Frey, 1988;Heggen et al., 2012;Tylmann et al., 2012). In case of a few lakes of complex morphology (Lake Ełckie, Żabińskie, Łazduny, Kamenduł, Kortowskie, Szurpiły, Dmitrowo, Długie), the bottom sediments were sampled in second deepest site.
In order to conduct subfossil Cladocera analysis, the 2 cm 3 of homogenized fresh sample from each lake was prepared in the laboratory according to the standard procedure described by Frey (1986). Samples were first treated with hot 10% KOH and HCl in order to remove carbonates. Chemical treatment was followed by sieving with a 38 μm mesh size. Microscope slides were examined with a light microscope under magnifications of ×100, ×200, and ×400. All skeletal elements (head shields, shells, and postabdomens) were counted until 70-100 individuals were found, which is regarded as an adequate number to characterize the assemblages (Kurek et al., 2010). Identification of Cladocera remains was based on the key by Szeroczyńska and Sarmaja-Korjonen (2007). Distinction of the Eubosmina species was based on publication of Faustová et al. (2011) and Błędzki & Rybak (2016).

Results
The studied lakes differed considerably from each other with regard to morphometric features (Tab. 1), however it should be underlined that small lakes were preferably chosen, because of their high sensitivity to environmental changes. Despite that all the lakes were located in post-glacial landscape their catchments displayed different lithologies from primarily fine-grained (up to 100% coverage by till and/or clay) to predominantly coarse-grained (up to 100% coverage by sands) with appreciably high shares of paludified areas. The catchments were usually weakly urbanized. The highest share of urbanized areas of 30.5% was in Lake Kortowskie catchment.
Even more important for the current study was that the lakes studied covered appreciably wide gradient of physical-chemical water parameters during summer maximum of biological activity in temperate lakes (Tab. 2; Fig. 4-9). The key variables measured were (i) bulk water mineralisation-related indicators (EC/TDS, Na + , Ca 2+ ), (ii) red-ox and alkalinity conditions (pH, O 2 concentration, HCO 3 -) and (iii) trophy-related indicators such as total phosphorus (TP), chlorophyll-a (Chl-a) and Secchi depth (SD) (Fig. 4), On the basis of the Gibbs' diagram (Gibbs', 1970) (Fig. 10) it appears that water chemistry in the lakes were primarily shaped by chemical weathering reactions, which can      Figure 5. The vertical distribution of temperature (t) in studied lakes during summer stagnation period. The x axis represents the rage of t values (°C), y axis represents depth (m)      Figure 8. The vertical distribution of electrical conductivity (EC) in studied lakes during summer stagnation period. The x axis represents the range of EC values (μS/cm), y axis represents depth (m) be found typical of temperate lakes. The contribution from evaporation and especially from atmospheric precipitation was very weak. The former showed some, albeit minor, importance in Lake Sarąg, L. Skanda, L. Linowskie, L. Kiersztanowskie. It seems however, that the enhanced TDS and Na/Na+Ca values in these lakes were owing to cultural eutrophication rather than due to enhanced evaporation. Epilimnetic waters in the lakes were well oxygenated, as expected for surface layers of lakes, however O 2 concentrations changed in a broad range (Tab. 2). O 2 variability seems to be related to bioproductivity. The highest DO values (~17.5 mg L -1 ) were noted in two lakes with high Chl-a (Olecko Małe and Morliny). Lake water pH was alkaline albeit alkalinity (in terms of HCO 3 -), and thus pH buffering capacity, varied greatly from low values in Lake Tyrsko to high values in L. Leszczewek.
Total phosphorus (P tot ) is considered as a main limiting factor of biological productivity in lakes (Wetzel, 2001). The P tot gradient in studied lakes was wide (Fig. 4) ranging from very low values (nearly 0) typical of low trophic state and removal of P by phytoplankton in the lake to 70 µg L -1 typical of eutrophic conditions (OECD, 1982). The Chl-a acts as a phytoplankton community biomass index and is positively related to the P tot availability (Kalff, 2002). However, high P tot values not always results in high Chl-a values (eg.: lakes Wymój, Zajdy). The SD values is broadly used to characterize water transparency however it is also closely related to water trophic state because light penetration depth in lake water columns are very often determined by the occurrence of phytoplankton (Kalff, 2002;OECD, 1982). On the other hand, the relationship between SD and trophic state can be modified by the occurrence of dissolved organic matter as well as inorganic suspension (Kalff, 2002;Borowiak, 2015). The measured SD range from considerably high values of ca. 6.00 m in the lake Białe Wigierskie to as low as 0.5 m in Lake Wymój. The very low transparency in the latter can be attributed to the abundance of cyanobacteria, as evidenced by highest phycocyanin values (PC) of 5.88 μg L -1 (Brient et al., 2008).
The vertical hydrochemical profiles provide additional information on the lake functioning. The summer hydrochemical profiles show vertical thermal stratification in the lake water columns (Fig. 5). In addition, in lakes > 10 m deep the water column was stratified into epi-, meta-and hypolimnion. In lakes < 10 m deep the stratification was rather weak albeit temperature gradients were present from the top to the bottom. Based on temperature distribution it appeared that the thickness of epilimnion varied between c.a. 2 m in Lake Kukowino and Małe to c.a. 7 m in L. Białe Wigierskie. Metalimnion, where thermocline (defined as ≥ 1°C difference per 1 m depth) was located, was 3-9 m-thick. The minimum thickness occurred in L. Białe Wigierskie, the deepest lake in our data set. On the other hand, the highest thickness of the metalimnion was noted in Lake Linowskie (Fig. 5), where also the highest value of epilimnetic TP and Chl-a were measured. Among many factors controlling the thickness of epi-and metalimnion the wind fetch seems to be of crucial importance, however, some influence from lake area and water transparency was also reported (Kalff, 2002). It was shown that the abundance of solid particles in suspension hampers penetration of the water column by solar energy thus limiting the depth of heat transfer. Midsummer oxygen (DO) profiles showed overall O 2 downward decreasing trends in most lakes studied (Fig. 6). Clinograde O 2 curves predominated with only few exceptions. Near bottom waters were usually anoxic/suboxic. Oxic conditions in the bottom part of the water column were found in deep, stratified lakes such as L. Białe Wigierskie and Berżniki, as well as in several shallow lakes e.g L. Płaskie, L. Zelwa, L. Tobołowo (Fig. 6). In the two deep lakes the presence of O 2 is related to nutrient-poor conditions therein (Jasiewicz et al., 2022) and thus low O 2 consumption for organic matter degradation. The oxic conditions throughout the whole water column in few shallow lakes was presumably owing to low water depth and polymictic character, as well as low trophic status. In 6 out of 64 studied lakes (Białe Wigierskie, Tyrsko, Szurpiły, Brożane, Łazduny, Dmitrowo; Fig. 9) we found positive heterograde O 2 distribution. This type of distribution is characterized by a maximum oxygen concentration occurring in metalimnion. The DO maxima are formed both by biological processes resulting from photosynthethic O 2 production, as well as physical processes such as warming of gasses trapped below the thermocline (Wilkinson et al., 2015). The co-occurrence of DO and Chl-a maxima in L. Brożane suggest that O 2 peak in metalimnion formed as a result of photosynthethic production. In other 5 lakes, however, DO and Chl-a show different patterns, thus, suggesting physical forcing on DO maximum formation (Wilkinson et al. 2015). This process presumably involves the offshore O 2 transport by water currents into deeper layers Kalff (2002).
Vertical distribution of chlorophyll-a (Chl-a) in the water columns vary greatly between lakes and show very week relationships to measured hydrochemical parameters.
However, in most of the lakes studied maximum values occurred in the epilimnion (Fig. 7). Provided that chlorophyll-a was found to be very labile compound, which could be microbially degraded within a time span of days to hours (Rydberg et al., 2020), we presume that its distribution in the water column indicates the zones of its in-situ production. Interestingly, in 10 stratified lakes the maximum Chl-a values occurred in meta-or hypolimnion indicating forming of deep chlorophyll maxima (DCM) or deep chlorophyll layer (DCL). The primary production within DCM may contribute to large amount (~60%) of total algal production in the lake (Camacho, 2006;Scofield, Watkins, Osantowski, & Rudstam, 2020). The mechanisms responsible for DCL formation are various (Camacho, 2006;Scofield, 2020), but its depth increase with water transparency (Scofield, 2020). In low productivity lake, Białe Wigierskie (SD 5.96 m), DCL occurs in oxygenated hypolimnion at 11,5-18,5 m depth. Whereas in Lake Łazduny (SD 3.45 m) it was located within a layer with a very low oxygen concentration at 7.5-11.5 m. We hypothesize that occurrence of DCL in anoxic waters may be caused by the presence of chlorophyllous phototrophs conducting anoxygenic photosynthesis (Camacho, 2006;Scofield, 2020).
Together with temperature electrical conductivity (EC) is known to reflect density stratification of lake waters (Fig. 8). As a rule, when vertical stratification is established, EC shows a downward increase and the EC gradient zone occurs in metalimnion. The difference between low-EC epilimnion and higher EC bottom/hypolimnetic waters is owing to the predominance of photosynthesis and CaCO 3 precipitation over respiration in the surface waters while in bottom waters organic matter degradation and carbonate dissolution became more important (Wetzel, 2001). With only few exceptions, the lake studied follow the above well-established tendency. However, in L. Kiersztanowskie, L. Sunowo, L. Skazdubek, L. Siekierowo, L. Dmitrowo and L. Sarąg EC culminations occurred in metalimnetic waters (Fig. 8). We speculate that Geographia Polonica 2022, 95, 3, pp. 227-253  Figure 9. The combined results of vertical distribution of temperature (t; °C), dissolved oxygen (DO; mg/L), chlorophyll-a (Chl-a; μg/L), electrical conductivity (EC; μS/cm) for selected lakes during summer stagnation period this unusual EC distribution can be related to enhanced oxidation of anoxic chemical species (e.g. CH 4 ) which is often found in the oxycline (Schubert et al., 2010).

Subfossil diatoms
In total, 129 diatom species, including varies and forms were identified. The structure of diatom assemblages in terms of habitat was very diverse. Among the lakes studied there were those in which planktonic diatoms predominated and the lakes with high shares of tychoplanktonic (i.e. accidental plankton/ pseudoplankton) species. The latter feature in lakes is a result of vertical water mixing and sediment resuspension, which causes benthic diatoms to detach from the lake floor and float in the water (Kawecka & Eloranta, 1994;Witkowski et al., 2009). The most abundant planktonic species were Aulacoseira italica, Aulacoseira granulata, Aulacoseira ambigua, Pantocsekiella comensis, Stephanodiscus parvus. The most abundant tychoplanktonic species were Pseudostaurosira brevistriata, Staurosira construens, Staurosirella lapponica. Sixteen species were rare (occurring only once), while 37 species (28.7% all diatoms) occurred at low (< 1%) relative abundance. Thirty-one species were abundant (i.e. ≥ 5% maximum relative The diagram is used to show the major processes affecting lake water chemistry. From the diagram it appears that lake water composition in Polish lowland lakes is primarily shaped by weathering reactions in the substrate. The influence from evaporation (upper edge) and atmospheric precipitation (lower edge) is negligible