• Open Access

Feeding strategies and elemental composition in Ponto-Caspian peracaridans from contrasting environments: can stoichiometric plasticity promote invasion success?


Kęstutis Arbačiauskas, Nature Research Centre, Akademijos St. 2, LT-08412 Vilnius, Lithuania. E-mail: arbas@ekoi.lt


1. Ponto-Caspian peracaridans, and mysids and amphipods in particular, are among the most successful aquatic invaders. However, species differ in the trophic-status range of ecosystems they can invade while establishment rates and impacts can vary substantially between habitats. There is limited knowledge of the environmental factors and species characteristics that drive such variation in invasion success.

2. Here we test how trophic level and body stoichiometry vary among peracaridan species and in relation to body size. The amphipod Pontogammarus robustoides and the mysids Limnomysis benedeni and Paramysis lacustris were investigated in ecosystems differing considerably in productivity and nutrient supply, namely an N-limited eutrophic lagoon and P-limited mesotrophic lakes.

3. As revealed by stable isotope (15N/14N) analysis, herbivory was inferred to be the main feeding mode of L. benedeni. In contrast, the mysid P. lacustris and the amphipod P. robustoides displayed a higher propensity for predatory feeding at larger body sizes, a pattern that was more pronounced in the eutrophic lagoon than in the mesotrophic lakes.

4. Their mean stoichiometric composition (P. robustoides C:N:P 108:20:1, L. benedeni 92:21:1 and P. lacustris 93:22:1) demonstrates that these peracaridans are rich in nutrients, especially nitrogen. They all exhibited the same ontogenetic pattern of reduced stoichiometric regulation during juvenile stages and stricter homoeostasis at older stages.

5. The higher P content in juveniles of all peracaridan species from the lagoon indicates higher potential somatic and population growth rates than those in the mesotrophic lakes. Such a difference may explain the substantially faster rates of invader establishment observed in the lagoon in comparison with lakes of low trophy.

6. Due to differences in ontogenetic and habitat-induced variation, the study species differed significantly in stoichiometric variability, which was lowest in L. benedeni and highest in P. robustoides. The ranges of species-specific variation in stoichiometric ratios corresponded to the trophic (by chlorophyll a) and nutrient stoichiometry (N:P) ranges of lentic waters successfully invaded by these species in Lithuania.

7. Stoichiometric plasticity, which should be associated with flexibility of feeding strategy, may enhance the potential of peracaridan species to successfully invade habitats with differing trophy and nutrient supply. The optimal feeding strategy should be omnivory with a propensity for predatory feeding, which can be adjusted with respect to ontogenetic nutrient demands and resource availability. Invading species may have a stronger effect on the local biota in ecosystems with high P levels, which promote growth, and N limitation that should favour predation.


The success of invading species in new communities is highly dependent on their somatic growth and reproduction rates and consequent ability for rapid population growth (Shea & Chesson, 2002), although biotic interactions such as predation or competition may also be important (de Rivera et al., 2005; Blight et al., 2010). These biological traits are tightly associated with enhanced demands for phosphorus (P) and nitrogen (N) that may be limiting in new environments (González et al., 2010). Phosphorus as an important constituent of RNA and nitrogen as an essential element of protein may control the rate of biosynthesis. Thus, attempts have recently been made to explain the success of biological invasions in the context of ecological stoichiometry (Acharya, Jack & Smith, 2006; Naddafi, Eklöv & Pettersson, 2009; González et al., 2010). The Growth Rate Hypothesis predicts that organisms with the highest P concentrations in their bodies would grow faster in P-rich environments and would be sensitive to P limitation (Sterner & Elser, 2002). Predators have been reported to generally be the most N-rich trophic group (Gonzáles et al., 2011), and N limitation may drive them to search for N-rich food, frequently even resulting in intraguild predation (Denno & Fagan, 2003). Theory predicts that primary producers have a wide range of variation in elemental composition, whereas consumers keep their elemental composition relatively constant and have relatively strict stoichiometric homoeostasis (Sterner & Elser, 2002). Thus, for herbivores it might be challenging to meet high nutrient demands in nutrient-poor habitats, whereas a shift to omnivory might relax such limitations by reducing the stoichiometric imbalance between the body and the resource (Gaedke, Hochstädter & Straile, 2002; Elser & Hessen, 2005; Laspoumaderes, Modenutti & Balseiro, 2010). Nutritional quality in terms of ratios of carbon (C):nutrient shows substantial variation in pelagic and benthic primary producers between ecosystems with different nutrient supplies (Sterner et al., 1997; Qin et al., 2007). This may set limits for the trophic-status range or stoichiometric range of aquatic ecosystems where particular species could invade and could also influence the effect of the invader on the native biota.

The peracaridans of the Ponto-Caspian region, in particular the mysids and amphipods, are among the most successful aquatic invaders, presumably because they typically possess a combination of biological traits including broad environmental tolerance, omnivory and r-selected life-history traits (Bij de Vaate et al., 2002; Hänfling, Edwards & Gherardi, 2011). With increasing body size, these crustaceans usually undergo ontogenetic diet shifts from herbivory at early life stages to omnivory and even carnivory in adulthood (Grossnickle, 1982). However, they frequently remain opportunistic omnivores that probably adjust their diet in response to quality and quantity of available resources (Viherluoto et al., 2000; Kelly, Dick & Montgomery, 2002; Lesutiene et al., 2008; Fink et al., 2012). Peracaridans can select foods with high nutrient content (Kelly et al., 2002; Aßmann, von Elert & Gergs, 2009). Consequently, the nutrient content of available food sources could be an important factor governing the shifts from herbivory to carnivory or even intraguild predation and cannibalism that are seen in this group. Nevertheless, the reliance of peracaridan feeding strategies on resource nutrient stoichiometry still warrants investigation. Such information is of key importance in the prediction of possible effects on invaded communities and ecosystems (e.g. Ketelaars et al., 1999; van der Velde et al., 2009; Ricciardi, Avlijas & Marty, 2011).

For several decades, mysids, amphipods and other aquatic fauna from the Ponto-Caspian region have been well known for their high invasion success in European and North American inland waters (Mordukhay-Boltovskoy, 1964; Ricciardi & MacIsaac, 2002). There is strong evidence for greater success of these crustaceans in ecosystems with higher nutrient supply (see Wikström & Hillebrand, 2012). In Lithuanian waters, these peracaridans were deliberately introduced during the early 1960s to improve fish-food resources, and their expansion across lentic waters proceeded mainly through secondary introductions (Gasiūnas, 1972; Arbačiauskas et al., 2011). Attempts to introduce Ponto-Caspian mysids and amphipods into more than 100 lakes of different trophy were undertaken, but they established only in 14, mainly mesotrophic, lakes with variable species-specific success (Arbačiauskas, Rakauskas & Virbickas, 2010). It has been hypothesised that establishment of non-indigenous peracaridans, especially that of amphipods in eutrophic lentic waters, may be restricted by low oxygen under ice, as all these crustaceans have a high oxygen demand (Arbačiauskas, 2005; Arbačiauskas & Gumuliauskaitė, 2007). However, the reasons for the inappropriateness of some low trophy lakes for certain species, and the substantially different rate of population establishment of the same species in different waterbodies, are unclear.

Here we test whether the nutrient status of environments affects the feeding strategy and body stoichiometry of invader species. Thus, we address the hypothesis that plasticity of a species’ response to varying environments is related to its ability to invade ecosystems whose primary producers have different stoichiometric properties. For this purpose, the ontogenetic (size-related) patterns of feeding modes and body C:N:P stoichiometry in two mysid and one amphipod species, and the relationship of those patterns to resource nutrient content, were investigated. Each species was studied in two ecosystems that exhibit the most extreme differences in trophic status across Lithuanian lentic waters, namely a eutrophic lagoon and mesotrophic lakes. To differentiate between herbivorous, omnivorous and carnivorous crustacean size classes, stable isotope analysis (SIA) was applied. The ratio of 15N/14N stable isotopes in the consumer indicates the ratio of assimilated food items taking into account a known enrichment factor (3.4‰). Therefore, the δ15N analysis is widely used in limnological research to reconstruct the diet and estimate the trophic position of consumers in food chains (Post, 2002).


Species and study sites

Ponto-Caspian amphipods and mysids were deliberately introduced into the Kaunas Water Reservoir, located on the Nemunas River, during 1960–1961. By downstream dispersal, these crustaceans reached the Curonian Lagoon and quickly established high population densities there (Gasiūnas, 1972; Arbačiauskas, 2002). Dispersal across other lentic waterbodies mainly proceeded via secondary introductions. The most successful species in inland waters appear to be the amphipod Pontogammarus robustoides Sars and mysid Paramysis lacustris Czerniavsky. In addition to Kaunas Reservoir, they each currently inhabit two other reservoirs, and the amphipod also occurs in 11 lakes while the mysid occurs in 14. In contrast, another introduced mysid, Limnomysis benedeni Czerniavsky, managed to establish only in one reservoir and two lakes (Arbačiauskas et al., 2011). Generally, P. lacustris is a bentho-pelagic feeder capable of utilising the fine fraction of seston, as well as plant or animal food (Lesutiene et al., 2007, 2008), while L. benedeni usually is considered a ‘truly’ herbivorous feeder (Komarova, 1991; Berezina et al., 2011; but see Fink et al., 2012). P. robustoides has been reported to be largely detritivorous and herbivorous (frequently with a preference for filamentous algae), but is capable of carnivory at large body sizes (Berezina, Golubkov & Gubelit, 2005; Arbačiauskas & Gumuliauskaitė, 2007; Lesutienė, 2009).

We selected three study sites, representing environments with contrasting nutrient supply, and at the end-points of species occurrence in the trophic gradient of Lithuanian waters. Important characteristics of these ecosystems are summarised in Table 1. Curonian Lagoon (55° 20′32.5′N, 21°11′19.8′E) is a large, shallow eutrophic waterbody connected to the South-Eastern Baltic Sea via a narrow strait. It is largely fresh water but the salinity in the northern part varies from 0 to 7 ppt (Gasiūnaitėet al., 2008). Summer phytoplankton is characterised by nitrogen limitation inducing a phytoplankton community shift towards dominance of diazotrophic cyanobacteria (Pilkaitytė & Razinkovas, 2007). Lakes Daugai (54°19′55.3′N, 24°22′11.6′E) and Plateliai (56°1′41.5′N, 21°51′15.7′E) have mean depths greater than 10 m and are classified as mesotrophic (Table 1). These are the lakes of the lowest trophy where both mysid species and the amphipod have established populations.

Table 1. Main morphometric and trophic characteristics (summer values) of the waterbodies examined in this study
Study speciesCuronian LagoonLake DaugaiLake Plateliai
  1. Means ± SD of replicate measurements are given where possible. TSM – total suspended matter, POC – particulate organic carbon. Species abbreviations: LB –Limnomysis benedeni, PL –Paramysis lacustris and PR –Pontogammarus robustoides.

  2. aGasiūnaitėet al. (2008).

  3. bKalytytė (2010).

  4. cThis study.

Area, km215849.1111.82
Mean (max) depth, m3.8 (5.8)13.2 (44)11.4 (46)
Secchi depth, m0.3–2.2a6.0c7.0c
Trophic statusEutrophicaMesotrophicbMesotrophicb
TSM, mg L−153.4 ± 3.5c2.97 ± 0.20c1.28 ± 0.12c
POC, mg L−124.5 ± 3.5c0.86 ± 0.05c0.40 ± 0.01c
Chlorophyll a, μg L−183.7 ± 13.5c3.00 ± 0.23c2.20 ± 0.12c

Mysids and amphipods are key species in vegetated littoral and open sandy bottoms of the Curonian Lagoon (Daunys & Zettler, 2006; Lesutienė, 2009). Both mysid species were transferred into Lake Daugai in 1965 and repeatedly in 1970 and 1972. The establishment of P. lacustris was reported in the early 1970s, but the establishment of L. benedeni was noted only in late 1990s (Arbačiauskas, 2002). The amphipod P. robustoides was released into Lake Plateliai in 1964, and another translocation of the species together with the mysid P. lacustris was undertaken in 1968. The establishment and expansion of a population of this amphipod in the lake were rather slow, while the mysid did not survive (Arbačiauskas, 2005; Gumuliauskaitė & Arbačiauskas, 2008).

Substantial differences between the lagoon and the lakes in the rate of invader population establishment cannot easily be attributed to variation in biotic interactions, such as competition or predation. The amphipod P. robustoides is a superior competitor to native amphipod species (Arbačiauskas et al., 2010), while Ponto-Caspian mysids do not have ecologically equivalent native species in the shallow waters of Lithuania. The invasive peracaridans are consumed by various fish species, especially juvenile percids (Arbačiauskas et al., 2010). However, higher predation pressure is to be expected in the lagoon, which is a eutrophic waterbody characterised by high fish production and abundant populations of perch (Perca fluviatilis) and pikeperch (Sander lucioperca) (Ložys, 2003).

Sampling and sample treatment

All crustacean samples were taken in July 2011. P. lacustris was collected at 1.5–2.5 m depths from a boat using an epibenthic sledge, whereas L. bendeni was collected in vegetated littoral areas at depths of <1 m. A hand net was used to sample the amphipod P. robustoides. Live specimens were brought to the laboratory where they were kept overnight in filtered water for gut content evacuation (as recommended in Hill & McQuaid, 2011) and stored frozen at −20 °C until later treatment. Thawed specimens were measured to the nearest 0.1 mm using a stereomicroscope equipped with an ocular micrometer. Total body length was estimated from the tip of the rostrum to the end of the telson. Each specimen was dried individually at 60 °C for 48 h. The whole body of all crustaceans and the abdominal part of large mysids (to avoid the ovary) were used for stoichiometric and stable isotope analysis. Three to 10 individuals in each 0.5-mm size interval were ground to a fine powder in an agate mortar. A 0.8–1.0 mg portion of each sample was weighed into tin capsules for C and N elemental analysis and SIA, and a 0.05–0.20 mg amount was weighed for particulate phosphorus analysis, using a Sartorius SE 2-OCE ultra-micro analytical balance (±0.0002 mg). When available, four replicates for each 0.5-mm size interval were analysed.

Filamentous algae were collected by hand from stones and submerged plants in the lagoon and Lake Plateliai as food for the amphipods. They were washed with filtered water and frozen. Thawed algal matter was dried and prepared for elemental analysis as described for the peracaridans.

Seston was sampled by a 2-L water sampler. In the lagoon, three replicate samples were taken from the middle of the water column at about 1 m depth, whereas in the lakes three integrated samples were taken by pooling subsamples from the surface and at each of 0.5, 1 and 2 times the Secchi depth. Water was filtered through 80-μm Nitex to remove zooplankton. For C and N elemental analysis and SIA, the seston was concentrated on pre-combusted and pre-weighed glass fibre filters (Whatmann GF/F, 47 mm diameter), which were dried for 48 h at 60 °C and then weighed. From each filter, two subsamples were prepared by cutting a 7-mm-diameter circle using a hole punch, or cutting a quarter of the filter, depending on organic matter content. These were then packed into the tin capsules, weighed and used as analytical replicates. Filter subsamples of equivalent size were used for particulate P analysis.

For the estimation of chlorophyll a concentration, water samples of 0.3–2.0 L (depending on a visual estimate of phytoplankton density) were concentrated on GF/F filters and kept frozen at −20 °C until analysis. Chlorophyll a was measured after extraction in 90% acetone. Total phosphorus was estimated using a molybdate–ascorbic acid method, after potassium persulfate digestion at 120 °C for 30 min. Samples were centrifuged (at 4000 g for 10 min) before they were analysed spectrophotometrically as in Færøvig & Hessen (2003).

Ratios of 13C/12C and 15N/14N as well as %C and %N in the samples were determined using continuous-flow isotope mass spectrometry provided in automated NC analysis (ANCA) SL 20-20, PDZ Europa at the Stable Isotope Facility, UC Davis, U.S.A. The standard reference materials were Vienna PDB and atmospheric N2. Isotope ratios are expressed in units of per mille (‰). Repeat analyses of internal standards indicated a typical measurement precision of ±0.1‰ for carbon and ±0.3‰ for nitrogen.

Data analysis

The trophic level (TL) of peracaridans was calculated with respect to δ15N values using the equation inline image, where δ15Nc is the nitrogen isotope ratio in crustaceans and δ15Nb is the nitrogen isotopic baseline of primary consumers (i.e. trophic level 2; Post, 2002). The ideal baseline should be biologically related to the consumer and have similar temporal and spatial integration of food-source isotopic signatures (Matthews & Mazumder, 2003). As appropriate baseline organisms belonging to the same food chain and having tissue turnover rates comparable to those of peracaridans were not available in all studied ecosystems, it was assumed that juvenile individuals of smallest body size (body length ≤5.5 mm) were ‘true’ herbivores and consequently, their δ15N values were considered as baselines for each species in each waterbody separately. In a previous study, δ15N values of P. lacustris juveniles did not differ significantly from herbivorous zooplankton over the entire season (Lesutiene et al., 2007, 2008). Elemental C, N and P contents were calculated as proportions of dry weight, and C:N, C:P and N:P ratios were converted to molar units.

For further analysis of trophic modes and C, N and P content, specimens were placed into three size classes (with equal sample sizes where possible), hereafter referred to as Small, Medium and Large. For the mysid P. lacustris, which exhibited the largest range of body lengths, the largest specimens were analysed as a separate size class ‘Largest’ (Table 2). The size classes Large and Largest included adult individuals with only a few small mysid adults attributed to the class Medium.

Table 2. Feeding modes of peracaridan size classes in the Curonian Lagoon and lakes Daugai (mysids) and Plateliai (amphipod), classified by their inferred trophic levels (TL)Thumbnail image of

Means and standard deviations were calculated from replicate samples for each size class, whereas grand means and their standard deviations were calculated from data pooled over habitats and size classes, unless otherwise specified. Within species, the effect of size class and waterbody on measured characteristics was analysed by two-way anovas, and difference between size classes was tested applying the post hoc Tukey’s unequal N HSD test (referred further as HSD test). When appropriate, data were Box-Cox transformed prior to calculations to achieve normality and homoscedasticity. Separation between herbivorous, omnivorous and carnivorous size classes was based on HSD tests; size classes that showed significantly higher δ15N isotopic signatures in comparison with Small were considered to be omnivores, and those that had trophic level estimates significantly higher than 3 were classified as predators. Elemental composition and ratios of particulate organic matter (POM) and filamentous algae between habitats were compared by one-way anovas.

The relationship in stoichiometry between consumer and putative resources can be analysed using the model of homoeostasis proposed by Sterner & Elser (2002). This model is applicable for consumers of similar age or body size. When analysis includes measurements from individuals with different body sizes, the model can be extended to account for possible ontogenetic patterns in variation. Assuming constant proportionality, we determined that consumer stoichiometry (y) depends upon resource stoichiometry (x) and an ontogenetic variable (z). This relationship can be given by the differential equation


Integration of this equation gives


and further linearisation using logarithms results in an equation suitable for analysis of stoichiometric data:


where c is a constant, k1 is a coefficient indicating the effect of resource on the consumer’s stoichiometry (i.e. the level of homoeostasis) and k2 is a coefficient showing the effect of ontogenetic stage (i.e. the ontogenetic pattern). With respect to the original model (Sterner & Elser, 2002), k= 1/H, where H is a regulation coefficient with values from 1 (no homoeostasis, k= 1) to infinity (strict homoeostasis, k= 0). Analysing field data, k1 may assume values above 1 and below 0. When this bias is small, it may be attributed to measurement errors. Consequently, a k1 estimate slightly above 1 can be interpreted as an indication of no homoeostasis, whereas a non-significantly negative estimate would indicate strict homoeostasis. If the k1 estimate is substantially >1 or significantly negative, reasons for such an outcome can be associated with the utilisation of other food resources or reliance on multiple food items. For convenience, in this work we expressed the regulation coefficient as H′ = 1 − k1 with variation between 0 (no homoeostasis) and 1 (strict homoeostasis). The other coefficient of equation (1), namely k2, can be used as a test for the presence or absence of an ontogenetic pattern in a consumer’s stoichiometry.

An appropriate characteristic of ontogeny in the extended model would be body weight or length, which are linearly related on a logarithmic scale. We used body length estimates, as these were available. In this study, we collated stoichiometry of juveniles and their putative resources using simple regression models. Stoichiometry of larger-sized individuals was analysed in relation to resource stoichiometry and body length, and all individuals were collated against resource and body length to look for an ontogenetic pattern using multiple regression analysis, with all calculations performed using Statistica 6.1 software (StatSoft, Inc., Tulsa, OK, USA).


Feeding modes

The inferred trophic position of the peracaridans ranged from herbivory (trophic level (TL) = 2) in juveniles to ‘true’ carnivory (TL ≥ 3) in the largest adults (Table 2). TL in all species generally increased with body size (Fig. 1). In L. benedeni, significant differences between sizes classes were absent in the lagoon and the lake (Table 2). However, the effect of size class on TL emerged in the anova (Table 2). This result was due to the significant difference between Small and Large classes when pooling the measurements from both habitats (HSD test, P = 0.033) and suggests that large individuals were generally consuming more animal food. The amphipod P. robustoides and the mysid P. lacustris were more predatory, and the ontogenetic increase in carnivorous feeding in these species was larger in the Curonian Lagoon than in the lakes. While the mysid P. lacustris also showed a gradual increase in carnivory with increasing body size in Lake Daugai, this was not the case for P. robustoides in Lake Plateliai (Table 3, Fig. 1).

Figure 1.

 Trophic level values (mean, SE, SD of replicate measurements) in peracaridans from the Curonian Lagoon (closed squares) and lakes (open squares) Daugai (mysids) and Plateliai (amphipod). Significant size class differences between lagoon and lake indicated by asterisks (HSD tests, P < 0.05). For size class intervals see Table 2.

Table 3. Results of two-way anovas testing for the effect of waterbody (W, lagoon vs. lake) and size class (S) on trophic level of the peracaridans Limnomysis benedeni, Paramysis lacustris and Pontogammarus robustoides in the Curonian Lagoon and lakes Daugai (mysids) and Plateliai (amphipod)
SpeciesSourcedf F P
  1. Significant probabilities are in bold.

L. benedeni W10.0020.97
S25.40 0.012
W × S20.880.43
P. lacustris W124.5 <0.001
S362.4 <0.001
W × S312.1 <0.001
P. robustoides W141.6 <0.001
S29.53 <0.001
W × S26.56 0.003

The feeding modes of size classes for each species can be summarised as follows: (i) herbivory –P. lacustris, Small and Medium [body length (BoL) ≤ 8 mm] in both habitats; P. robustoides, Small (BoL ≤ 5.5 mm) in the lagoon, all size classes in Lake Plateliai; and L. benedeni, all size classes in both habitats; (ii) omnivory –P. lacustris, Large (BoL = 8.1–10.0 mm) in the lagoon, Large and Largest (BoL > 8 mm) in Lake Daugai; and P. robustoides, Medium and Large (BoL > 5.5 mm) in the lagoon; (iii) carnivory –P. lacustris, Largest (BoL > 10 mm) in the lagoon (Table 2). Thus far, variation in TL was present between habitats and among ontogenetic stages, and these effects differed between the peracaridan species.

Mean resource and peracaridan stoichiometry

Understanding variation in peracaridan feeding patterns also requires the characterisation of species-specific consumer and resource stoichiometry as imbalances between them may affect observed differences. The highest particulate organic carbon (POC) and chlorophyll a concentrations were measured for the Curonian Lagoon, while the lowest were in Lake Plateliai (Table 1). Carbon:nutrient ratios of POM decreased with decreasing productivity of the waterbody, whereas N:P ratio showed the opposite trend (Fig. 2) with substantial deviation from the Redfield ratio (16:1) indicating P limitation, particularly in the lakes. The C:N ratio in lagoon POM exceeded N deficiency levels (C:N > 8.3) for phytoplankton (see Hecky, Campbell & Hendzel, 1993), indicating N limitation. Filamentous algae had lower N and P concentrations and substantially higher C:nutrient ratios than POM (two-way anovas, food type effect, F ≥ 27.8, P ≤ 0.001). The extremely poor nutrient content of filamentous algae contrasted with the nutrient-enriched POM in Lake Plateliai (Table 4), whereas the difference in elemental contents and ratios between the two potential food sources was less pronounced in the lagoon (two-way anovas, food type × habitat interaction effect, F ≥ 102.1, P ≤ 0.001, Fig. 2).

Figure 2.

 Mean stoichiometric ratios of N:P, C:P and C:N in mysids Paramysis lacustris, Limnomysis benedeni and amphipod Pontogammarus robustoides (±SD of all measurements), and filamentous algae and particulate organic matter (POM) (±SD of replicate measurements) from the Curonian Lagoon and lakes Daugai and Plateliai.

Table 4. Elemental composition (C%, N% and P%) and N:P, C:P and C:N ratios (mean ± SD of replicate measurements) in particulate organic matter (POM) and filamentous algae in the Curonian Lagoon and lakes Daugai and Plateliai
VariablesLagoonLake DaugaiLake Plateliai F
  1. F values and probability levels (<0.05*, <0.01** and <0.001***) indicate results of one-way anovas for habitat effect; differences between habitats are denoted by non-matching letters (HSD tests, P < 0.05).

 C%45.6 ± 5.0a29.2 ± 1.1b31.8 ± 3.8b17.7**
 N%5.57 ± 0.60a6.56 ± 0.35a12.5 ± 0.6b156***
 P%0.31 ± 0.03a0.27 ± 0.02a0.44 ± 0.00b28.7**
 C:P389.8 ± 98.9a274.3 ± 30.6ab188.7 ± 21.8b8.20*
 N:P39.3 ± 4.6a52.9 ± 6.8b63.6 ± 3.0b17.5**
 C:N9.89 ± 1.89a5.19 ± 0.29b2.96 ± 0.23b30.1***
Filamentous algae
 C%30.1 ± 0.741.7 ± 0.5560***
 N%2.21 ± 0.282.05 ± 0.021.05
 P%0.19 ± 0.010.05 ± 0.01359***
 C:P407.6 ± 9.72302 ± 2614281***
 N:P25.7 ± 3.296.9 ± 1.01336***
 C:N16.0 ± 1.623.8 ± 0.267.2**

Mean elemental stoichiometry differed significantly between the amphipod and the two mysids (Table 5). The highest C:nutrient ratios and lowest elemental C, N and P contents were observed for P. robustoides (Table 5). Most of the mean elemental characteristics were similar between the two mysid species, with the exception of N proportion and C:N ratio, which were slightly higher in P. lacustris. The C:P, C:N and N:P ratios were generally lower and less variable in the crustaceans than in the POM and filamentous algae, showing substantial stoichiometric imbalances between the body and this food (Fig. 2). The largest stoichiometric imbalance was observed between the amphipod and filamentous algae in Lake Plateliai, and between the peracaridans and POM (C:nutrient ratios) in the lagoon. The POM of Lake Plateliai was over-enriched with N, resulting in a lower C:N ratio than that in the amphipods (Fig. 2). These results clearly indicate that peracaridans inhabited environments differing with respect to trophy and stoichiometry, and these habitat properties must have affected the performance of the species.

Table 5. Grand means (±SD of all measurements) of the study species’ elemental compositions and elemental ratios
  Limnomysis benedeni Paramysis lacustris Pontogammarus robustoides
  1. Significantly different means (HSD test, P < 0.05) are denoted by non-matching letters.

C%44.5 ± 0.8a45.1 ± 1.4a36.1 ± 1.5b
N%11.6 ± 0.1a12.4 ± 0.4b7.76 ± 0.59c
P%1.26 ± 0.07a1.27 ± 0.12a0.88 ± 0.14b
C:P91.8 ± 5.0a92.9 ± 10.4a108.1 ± 15.5b
N:P20.7 ± 1.3ab21.9 ± 2.5b20.0 ± 3.5a
C:N4.46 ± 0.07a4.25 ± 0.12b5.46 ± 0.36c

Size and habitat-specific variation in peracaridan stoichiometry

Elemental content was rather constant in L. benedeni over different body sizes and both habitats except that juveniles of this mysid species were significantly richer in P in the lagoon than in the lake (Fig. 3). In contrast, there was clear size-related variation and a pronounced response to contrasting environments in nutrient content of the mysid P. lacustris and the amphipod P. robustoides (Table 6, Fig. 3). In addition to P enrichment in the lagoon, juveniles of these species also showed variation with respect to N content. For P. lacustris, N content was higher in the lagoon than in the lake, while for P. robustoides it was lower. The overall decrease in P content with increasing body size in both P. lacustris and P. robustoides was less pronounced in the lakes. On average, the content of C, N and P in the mysid P. lacustris and that of C and P in the amphipod P. robustoides was lower in the lakes (Table 6, Fig. 3).

Figure 3.

 Proportions (mean, SE, SD of replicate measurements) of carbon, nitrogen and phosphorus in different size classes of peracaridans from the Curonian Lagoon (closed squares) and lakes (open squares) Daugai (mysids) and Plateliai (amphipod). Significant size class differences between lagoon and lake indicated by asterisks (HSD tests, P < 0.05). For size class intervals see Table 2. Note different scales for mysids and amphipods.

Table 6. Results of two-way anovas testing for the effect of waterbody (W, lagoon vs. lake) and size class (S) on proportions of elements C, N and P in the peracaridans Limnomysis benedeni, Paramysis lacustris and Pontogammarus robustoides from the Curonian Lagoon and lakes Daugai (mysids) and Plateliai (amphipod)
SpeciesElementSourcedf F P
  1. Only analyses with significant effects are shown. Significant probabilities are in bold.

L. benedeni PW15.81 0.027
W × S25.97 0.008
P. lacustris CW116.2 <0.001
S334.0 <0.001
W × S33.81 0.015
NW148.4 <0.001
S323.6 <0.001
W × S316.4 <0.001
PW114.4 <0.001
S315.8 <0.001
W × S34.00 0.013
P. robustoides CW15.36 0.024
W × S20.330.72
S23.81 0.028
W × S29.97 <0.001
PW110.0 0.002
S224.4 <0.001
W × S24.71 0.013

The C:P and N:P ratios for juveniles of all species were significantly lower in the lagoon due to enrichment with P (Fig. 4). Juveniles of the amphipod P. robustoides also exhibited significantly lower C:N ratio in the lake, suggesting that in the lagoon they were exceptionally depleted in N. The significantly higher estimate of this ratio in the lake was measured for the Large size class of P. lacustris (Fig. 4). A body size effect on C:P and N:P ratios was absent in L. benedeni, but was significant in the other mysid species and the amphipod; both ratios increased with increasing body size while exhibiting significant variation between habitats (Table 7, Fig. 4). The effect of body size on the C:N ratio was significant in all species. A slight increase in this ratio without variation between habitats was observed for L. benedeni, while the response of the other species differed between habitats (Fig. 4). On average, the C:P ratio in P. lacustris and the N:P ratio in all study species were higher in the lakes, whereas the C:N ratio in P. robustoides was higher in the lagoon (Table 7, Fig. 4). Consequently, the study species varied in their stoichiometric response to the environment, and the ontogenetic pattern of this response differed between the habitats.

Figure 4.

 Elemental stoichiometric ratios (mean, SE, SD of replicate measurements) C:P, N:P and C:N in different size classes of peracaridans from the Curonian Lagoon (closed squares) and lakes (open squares) Daugai (mysids) and Plateliai (amphipod). Significant and marginally significant size class differences between lagoon and lake revealed by HSD tests indicated by asterisks (P < 0.05) and asterisks in brackets (P < 0.10), respectively. All these differences are significant at P < 0.05 when applying pairwise Mann–Whitney U-tests with sequential Bonferroni adjustment. For size class intervals see Table 2. Note different scales for mysids and amphipods.

Table 7. Results of two-way anovas testing for the effect of waterbody (W, lagoon vs. lake) and size class (S) on elemental ratios C:P, N:P and C:N in the peracaridans Limnomysis benedeni, Paramysis lacustris and Pontogammarus robustoides from the Curonian Lagoon and lakes Daugai (mysids) and Plateliai (amphipod)
SpeciesRatioSourcedf F P
  1. Significant probabilities are in bold.

L. benedeni C:PW14.080.055
W × S26.34 0.006
N:PW15.52 0.027
W × S24.73 0.019
S25.15 0.014
W × S20.480.63
P. lacustris C:PW18.24 0.006
S326.1 <0.001
W × S33.19 0.032
N:PW14.76 0.033
S321.3 <0.001
W × S35.76 0.002
S34.27 0.009
W × S38.53 <0.001
P. robustoides C:PW11.480.23
S226.8 <0.001
W × S25.25 0.008
N:PW17.17 0.010
S226.3 <0.001
W × S211.5 <0.001
C:NW115.5 <0.001
S24.24 0.019
W × S214.6 <0.001

Consumer stoichiometry vs. resource stoichiometry and body size

Inferences about stoichiometric ‘regulation’ require regression analysis of the consumer’s stoichiometry against the elemental composition of a putative resource. It is of particular interest that the P content in the body of juveniles of both mysid species varied across habitats in proportion to its content in the POM, suggesting that stoichiometric regulation may be weak (k1 close to 1 and consequently, stoichiometric regulation H’ ≤ 0.1; Table 8). On the other hand, the relationship in P content between juvenile amphipods and putative food sources was weaker, suggesting stricter homoeostasis of this element; the relationship with POM was stronger than that for filamentous algae. With respect to the N:P ratio in the POM, juvenile mysids exhibited stoichiometric regulation at 0.6, while that for P. robustoides was at 0.4. The C:N ratio in juvenile mysids did not vary between habitats (Fig. 4) and consequently was unrelated to that in the POM. Although juvenile amphipods in the lagoon were depleted in N, the relationship in C:N between their body and POM was weakly positive or negative with respect to the filamentous algae (Table 8).

Table 8. Results of regression analysis of dependence of stoichiometric characteristics of peracaridans Limnomysis benedeni, Paramysis lacustris and Pontogammarus robustoides upon putative resource stoichiometric characteristics (i.e. POM for all species and additionally filamentous algae for the amphipod (denoted by FA)) and body size
  1. Provided regression coefficients: k1 for juveniles (Re-Juv, simple regression) and larger-sized individuals (Re-Other, multiple regression) against resource; k2 for species against body length (BoL-Sp, multiple regression over all-sized individuals). Probability levels are denoted as follows: <0.10(*), <0.05*, <0.01** and <0.001***. Note that stoichiometric regulation can be assessed as H′ = 1 − k1 (see Methods).

L. benedeni P%0.89(*)−0.05−0.06
P. lacustris P%1.10***0.12−0.13***
P. robustoides P%0.51**−0.02−0.42***
P. robustoidesFA P%0.13**0.01 

As the P content in peracaridan juveniles, especially mysids, was related to that in the POM, the analogous relationship also was expected for the C:P ratio. However, surprisingly the C:P ratio in juveniles of all peracaridan species was negatively related to that in the POM (Table 8). The higher absolute P concentration associated with poor nutritional quality of the seston (high C:P ratio) in the lagoon may have resulted in such a pattern (see Discussion).

In contrast to juveniles, the larger-sized individuals of all peracaridan species exhibited independence of their stoichiometry from putative resources suggesting strict homoeostasis (H’ ≥ 0.9). Regression analyses also confirmed that the mysid L. benedeni had no detectable pattern in its elemental composition in relation to body size, but the ontogenetic pattern was highly significant in the mysid P. lacustris and especially evident in the amphipod P. robustoides (Table 8).

Species-specific variation in peracaridan stoichiometry

The overall variation of all elemental ratios measured over all body size ranges and both habitats differed significantly between species and increased in the following order: L. benedeni < P. lacustris < P. robustoides (F tests for variances, P ≤ 0.003). Variation within habitats was similar for the mysids (excluding that for C:N ratio in P. lacustris which was larger in the lake; F test, P = 0.008), whereas the range of variability in elemental ratios for the amphipod was significantly wider in the lagoon (F tests, P < 0.001). The robust ranges of variation in L. benedeni, P. lacustris and P. robustoides (with 10 and 90 percentiles applied to exclude possible analytical errors) were C:P, 85.0–97.1, 79.9–107.5 and 83.6–129.5; C:N, 4.4–4.6, 4.1–4.4 and 5.1–6.0; and N:P, 19.2–21.6, 18.8–25.6 and 13.7–24.4, respectively.

Furthermore, the estimated ranges of variation of elemental ratios corresponded well with the ranges of trophic status (seasonal mean of chlorophyll a) and N:P ratio (seasonal mean of ratio of total N and total P) of lentic waterbodies successfully invaded by the study species in Lithuania (Fig. 5). For this analysis, we used data from the Lithuanian Environmental Protection Agency collected over a decade, excluding any other sources of information to ensure the same method of measurements. It included all lentic habitats for L. benedeni and P. robustoides, and 15 of 18 known habitats of P. lacustris to enable a robust estimate throughout the species’ distributions with respect to trophy and stoichiometry. Thus, the observed pattern suggests that the ability to alter body elemental composition indeed can affect invasion success of the peracaridan species across variable environments.

Figure 5.

 Variation (median, quartiles, range of all measurements) of elemental C:P (a) and N:P (c) ratios in peracaridans Limnomysis benedeni (LB), Paramysis lacustris (PL) and Pontogammarus robustoides (PR), and ranges of trophic statuses (b, seasonal mean of chlorophyll a) and N:P ratios (d, calculated on seasonal means of total N and total P) of lentic water bodies in which these species have established in Lithuania. Included are those waterbodies for which 2001–2011 data were available (Lithuanian Environmental Protection Agency, http://vanduo.gamta.lt/cms/); multiple assessments for each waterbody were averaged. Large circles indicate waterbodies studied here. Note the logarithmic scale for waterbody characteristics.


Invasive Ponto-Caspian peracaridan crustacean species were investigated in two types of ecosystem differing considerably in productivity (30- to 40-fold difference in chlorophyll a and 30- to 60-fold difference in POC concentrations) and nutrient supply. Phosphorus was the limiting nutrient in lakes, whereas nitrogen limitation was characteristic of the Curonian Lagoon. Invasion histories also suggest that establishment dynamics of peracaridan populations differed between these habitats. Within a few years of invasion, all species established high densities in the lagoon but increased rather slowly in the lakes (Gasiūnas, 1972; Arbačiauskas, 2005). The results of the present study clearly suggest that some properties of each habitat had an impact on species stoichiometry and feeding strategies, and that these responses were species specific. Furthermore, the ranges of species variation in elemental ratios over both habitats corresponded to the ranges in productivity and N:P ratio of the lentic waters that these species have successfully invaded. Such a pattern may suggest that stoichiometric plasticity facilitates invasion of a wider range of environments with respect to trophic status or stoichiometric properties.

Average stoichiometric composition (C:N:P) was rather similar between the three study species (P. robustoides 108:20:1, L. benedeni 92:21:1 and P. lacustris 93:22:1), as was N:P ratio. By contrast, the C:nutrient ratios were slightly higher in the amphipod. The C:N:P ratio in most phosphorus-rich daphnids is around 85:14:1, and in calanoid copepods, 234:25:1 (Andersen & Hessen, 1991; Sterner & Elser, 2002; Villar-Argaiz, Medina-Sánchez & Carrillo, 2002). In comparison with these zooplankton taxa, Ponto-Caspian peracaridans have intermediate values of N:P ratio. However, they also can be characterised as nutrient-rich species (particularly nitrogen, providing a low C:N ratio) with enhanced nitrogen demands. Nitrogen content in peracaridans, particularly in mysids (L. benedeni 11.6% and P. lacustris 12.4%), is even higher than that reported for herbivorous and carnivorous terrestrial arthropods (i.e. 9.8 and 11.4%; Gonzáles et al., 2011). However, there is significant variation due to ontogenetic and habitat effects on their elemental contents and ratios. This variation probably depends on availability and quality of food resources, ontogenetic stage-specific nutrient demands and the species’ ability to utilise various food items that is possible only with a flexible feeding strategy.

The weakest response to environmental properties was exhibited by the mysid L. benedeni. As revealed by 15N content, this species was largely herbivorous across a broad range of productivity and contrasting nutrient supplies in the habitats studied here. Although L. benedeni is capable of predation on small zooplankton under laboratory conditions (Fink et al., 2012), this was not evident from our field δ15N data. Only a slight increase in animal food was observed in large specimens compared to juveniles. This small-bodied mysid inhabits shallow littoral waters with submerged vegetation, which usually contain low zooplankton abundance due to predation by juvenile fish (Gliwicz & Rykowska, 1992), and the species’ ability to consume other animal prey is probably very limited.

In contrast to L. benedeni, the other mysid P. lacustris exhibited habitat-specific feeding strategies. The increase in carnivory with increasing body size in P. lacustris in the lagoon was substantially larger than that seen in the lake. The largest lagoon mysids may be characterised as true carnivores as they were more than one trophic level above juveniles. Such a difference suggests that they largely relied on zooplankton, as has been shown previously (Lesutiene et al., 2007, 2008), and may have included cannibalism, as is well known in mysids (Quirt & Lasenby, 2002). Meanwhile, the large-sized mysids showed omnivorous feeding in the lake. It seems that P. lacustris feeding on nitrogen-depleted phytoplankton in the lagoon faced a larger C:N imbalance than did those in the lake, and a flexible feeding strategy that enables an increase in carnivory might help to fulfil nutrient demands and maintain optimal body stoichiometry. It has been shown that a shift to protein-rich animal food relieves copepods of the shortage of nitrogen in algal food and supports further growth and reproduction (Laspoumaderes et al., 2010).

Our results clearly showed that the feeding performance of the amphipod P. robustoides also differed between the habitats. The species was more predatory in the lagoon than in the lake. Previous gut content analysis has shown that this species shifts from feeding on POM to filamentous algae at a size of about 7 mm (Berezina et al., 2005). This feeding change may increase the C:N imbalance between the animal tissues and the food, as filamentous algae typically produce organic matter that is poor in nitrogen (high C:N). Interestingly, the shift of P. robustoides from herbivory to omnivory was observed at approximately the same body size in the lagoon. The nitrogen demand in this species most likely was supplemented by N derived from animal proteins. It is noteworthy that the high nitrogen demand in amphipods, especially under P-rich conditions that do not limit high growth rates, probably explains the very common phenomenon of intraguild predation among these crustaceans, which is usually responsible for the extermination of native amphipod species (Dick & Platvoet, 2000; MacNeil & Platvoet, 2005; Arbačiauskas & Gumuliauskaitė, 2007; van der Velde et al., 2009).

By contrast, evidence of carnivory (with respect to 15N) by P. robustoides was absent in Lake Plateliai despite extreme depletion in N of filamentous algae there. However, the POM in this lake was highly enriched in N, and juvenile P. robustoides had significantly higher N content than those in the lagoon. Thus, the POM of Lake Plateliai may be a sufficient, or even stoichiometrically over-enriched, nitrogen source for this invader that hampers or mitigates its shift to filamentous algal diet and omnivory. It has been shown that surplus nutrient concentration in food can negatively affect consumer growth rates (Boersma & Elser, 2006).

Of particular interest is that P. robustoides (which usually induces the complete extirpation of local Gammarus species in invaded ecosystems; Arbačiauskas & Gumuliauskaitė, 2007) until now has not exterminated the native Gammarus lacustris in Lake Plateliai (Gumuliauskaitė & Arbačiauskas, 2008). The stoichiometric imbalance of resources, which presumably suppresses the growth of the invader, may be favouring the coexistence of these amphipod species in the lake.

High phytoplankton biomass in the Curonian Lagoon also had a high C:P ratio, which may be a summer characteristic of highly productive shallow waterbodies with dominance of cyanobacteria (Hessen, Van Donk & Gulati, 2005). This ratio (∼390) however was only slightly higher than the C:P threshold ratio of 200–300 for highly phosphorus demanding Daphnia (Sterner & Hessen, 1994; Vrede, Persson & Aronson, 2002). The threshold elemental ratio that limits growth rate is a function of food quantity (amount of carbon) and quality (nutrient content) (Sterner & Elser, 2002), while the less severe growth limitation by high C:P ratios has been observed at high food levels (Ferrão-Filho, Tessier & DeMott, 2007). Therefore, growth limitation due to P is not likely in the lagoon, as confirmed by significantly higher P content in juveniles of all the study species. Furthermore, the enhanced P content of peracaridan juveniles suggests that in the lagoon they were growing faster than in the lakes of low productivity. As juvenile growth rate may be considered a proxy for species fitness (Lampert & Trubetskova, 1996), it becomes evident why the establishment of populations of peracaridan species in the Curonian Lagoon occurred much more rapidly than in the mesotrophic lakes.

In the study species that showed an ontogenetic pattern in their stoichiometry, P content generally decreased, and consequently N:P and C:P ratios increased, with body size. Such a trend is common among consumers due to decreasing growth rates at later ontogenetic stages (Carillo, Villar-Argaiz & Medina-Sánchez, 2001; DeMott, 2003). The approximate 1.3-fold difference in P content between juvenile and adult stages of P. lacustris and 1.9-fold difference in that for P. robustoides is comparable to a 1.4-fold ontogenetic decrease in P in mayflies (Frost & Elser, 2002) or 1.1-fold and 1.9-fold P declines in Daphnia fed on P-sufficient and P-deficient diets, respectively (DeMott, 2003). The extent of the ontogenetic increase in C:P ratio (from 80 to ∼130) in peracaridans is lower than that observed in copepods (i.e. from 80 in nauplii to ∼290 in adults; Laspoumaderes et al., 2010), while the N:P ratio in our study ranged from about 12–24 as opposed to 3.3–24.6 in copepods (Carillo et al., 2001).

The deviation from strict homoeostasis in peracaridans also may occur due to an ontogenetic change of nutrient allocation strategies (i.e. allocation to growth during juvenile stages) as is well demonstrated by this study, or allocation to survival and reproduction later on (Villar-Argaiz et al., 2002). C content may increase in adulthood due to more C storage in fat (Walve & Larsson, 1999), while the enhanced N demand may be related to the increased proportion of chitinous (carbohydrate containing N but no P) exoskeleton in large specimens (Sterner & Elser, 2002). The decrease in phosphorus observed in the lakes in large P. lacustris females of the overwintered generation was probably due to reproduction. These females probably allocated relatively larger P amounts to embryo production to support their early growth in a food-limited environment. Reproductive tissues generally have lower C:nutrient ratios than do somatic ones (Færøvig & Hessen, 2003), while the difference between mother and offspring nutrient content is more pronounced under nutrient-deficient feeding conditions (Boersma & Kreutzer, 2002).

The study species exhibited substantial differences in variability of their body stoichiometry. Despite this, they all showed the same ontogenetic pattern of stoichiometric regulation, particularly the reduced regulation for juveniles and tighter homoeostasis at older stages of life. An interesting response to the properties of the habitat was observed in juveniles. Our results suggest that juvenile mysids of both species were acquiring P that is essential for rapid growth in proportion to its concentration in the POM. Stricter regulation was observed for the N:P ratio. However, unexpectedly, the relationship between consumer and resource appeared to be negative for the C:P ratio, and this was also detected in the amphipod. The most likely reason for such an outcome may be the high-resource C:P ratio (poor quality) in association with generally higher production fuelled by P and N enrichment in the Curonian Lagoon. The higher C content observed in the juveniles of P. lacustris (also in the larger-sized mysids and the amphipod) here indicates the accumulation of extra carbon in the body tissues. Consequently, extraction of P from the POM while eliminating the excess of C in juveniles, at least in the mysid juveniles, can be anticipated. This probable response of juvenile stages resulting in proportionality between body P content and P availability may be interpreted as an adjustment to the environment, rather than reduced regulation of homoeostasis. It has been proposed that such regulation of P assimilation may be common in species with rapid growth and high P demand at high C:P ratios in primary producers, although it may require enhanced energy costs for elimination of the excess of C (Hessen & Anderson, 2008).

What is worthy of notice also is that the comparison of stoichiometry of the consumer and the putative resource using field data as done here also may help identify the most likely sources of nutrients or even individual elements. If variation in consumers between habitats is observed, then a relationship to the resource is to be expected. Consequently, our results also suggest that the POM was the more likely source of P for juvenile P. robustoides than was filamentous algae, which is to be expected in juvenile amphipods, while the acquisition of N also included other food sources.

This study supports the conclusion from mesocosm studies that higher invasion success is to be expected in environments with higher nutrient levels (Wikström & Hillebrand, 2012). It seems that invading species with a propensity for predatory feeding may have a stronger effect on the local biota in environments with high P levels, which promote growth and N limitation and which may favour a carnivorous diet. Moreover, our study suggests that stoichiometric plasticity may enhance the potential of peracaridan species to successfully invade aquatic ecosystems with differing trophy and nutrient supply. Advantages of stoichiometric plasticity should be associated with a species’ ability to regulate body elemental composition, which, in turn, depends on flexibility of feeding strategy allowing for assimilation of various food items. It seems that the best feeding strategy may be omnivory, with propensity to predatory feeding that can be adjusted with respect to ontogenetic nutrient demands and resource availability. The narrowest variation of stoichiometric composition and similar trophic level of different ontogenetic stages indicates a rather uniform algal diet in the mysid L. benedeni, which corresponds to its low invasion success over the lentic waters of Lithuania, and its absence from low productivity lakes. By contrast, other peracaridan species that have substantially larger stoichiometric plasticity and habitat-specific flexibility of feeding strategy were able to invade a wider trophic as well as stoichiometric range of aquatic ecosystems. This hypothesis, which may be described as the stoichiometric plasticity and invasivity hypothesis, definitely warrants further examination over a range of environments and species.


Research was supported by Research Council of Lithuania, Project No. LEK-06/2010. We acknowledge the assistance of R. Žilienė, T. Ruginis, I. Lubienė and J. Petkuvienė in the field and laboratory work. We are especially grateful to W.R. DeMott for critical comments and advices, and thank the peer reviewer for remarks that all substantially improved the manuscript.