Introduction

Natural convection in lentic water bodies can influence water quality (Naghib et al. 2018) and may have substantial ecological impacts (Mao et al. 2019). In a lake, mixing and transport of particles can be promoted by density currents driven by variations in water temperature (Mortimer 1974; Okely and Imberger 2007; Tsydenov et al. 2016; Mao et al. 2019). Horizontal differences (limnetic vs littoral areas) in temperature create density gradients, which promote the establishment of horizontal surface exchange flows (Farrow 2004; Okely and Imberger 2007). The faster heating of the water in shallower areas than in the deeper areas generates convective currents from the littoral to the open lake at the surface. Temperature differences as small as 0.5 °C lead to velocity magnitudes of the order of a few centimetres per second (Pálmarsson and Schladow 2008). Also, aquatic plants are able to induce convective motion by promoting differential shading and by reducing wind in shallower regions (Lovstedt and Bengtsson 2008; Lightbody et al. 2008; Zhang and Nepf 2009). Lovstedt and Bengtsson (2008) observed average temperature differences of 0.8 °C and mean velocities of 0.8 ± 0.5 cm/s in surface currents towards the vegetated littoral in a shallow lake in southern Sweden. These thermal flows can transport nutrients, chemicals and pollutants through the surface of water across lentic water bodies (Mao et al. 2019), and can influence zooplankton distribution patterns (Podsetchine and Schernewski 1999).

The distribution of zooplankters has been argued to be driven by abiotic and/or biotic factors (Viljanen and Karjalainen 1993; Pinel-Alloul 1995; Thackeray et al. 2004; Gabaldón et al. 2019; Rollwagen-Bollens et al. 2020), relating to two major zooplankton movement patterns: diel horizontal migration (DHM) and diel vertical migration (DVM) (Pinel-Alloul 1995; Hembre and Megard 2003; Pinel-Alloul et al. 2004; Emily et al. 2017; Ermolaeva et al. 2019). DVM describes the movement into deeper and darker sites in the water column during the light time (day) to avoid visual predators, but at night the opposite movement typically occurs towards the water surface for improved acquisition of food resources (O’Brien 1979; Viljanen and Karjalainen 1993; Cuker and Watson 2002). In shallow lakes, DVM is less meaningful, since light may easily reach the bottom of the water column (Lauridsen et al. 1998; Burks et al. 2002). Rather, DHM is a major mechanism, described as the movement of zooplankton from limnetic zones into the littoral zones where macrophyte stands occur or vice versa (Wojtal et al. 2003; Jensen et al. 2010). DHM has been argued to depend on factors such as oxygen availability, pH, water turbidity, food resource availability and predatory pressure (Lauridsen et al. 1996; Burks et al. 2002, 2006; Ermolaeva et al. 2019). Fish predators can have a strong role in inducing zooplankton to move horizontally (Wojtal et al. 2003), and this pressure is higher in shallow than in deep lakes, since the relative density of fish is higher (Jeppesen 1998; Burks et al. 2002). Zooplankters indeed perceive chemical cues released by predators, as demonstrated by the induction of morphological and behavioural defences, as well as physiological adjustment (Engelmayer 1995; Lauridsen and Lodge 1996; Sakwińska 1998; Laforsch et al. 2006; Pijanowska et al. 2006; Castro et al. 2007a; Bell et al. 2019).

Although there is evidence of the establishment of typical DHM (see above), consistent preference for the limnetic zone by zooplankters regardless of the light cycle has also been found in shallow lakes and during spring, when predatory cues are greatest (Antunes et al. 2007; Castro et al. 2007b). The inconsistent evidence in DHM patterns may relate to the limited examination of conditioning factors, while DVM has been studied in more detail. For example, the role of thermal-driven surface exchange flows in modulating DHM of zooplankters in shallow lakes has not been comprehensively studied, despite occasional reference related to DVM (Pinel-Alloul 1995). This is an important gap considering the swimming activity constraints of zooplankters; they can float, drift or indeed swim, but their active swimming capacity has been assumed to be limited, leaving them at the mercy of currents to a significant extent (Thorp and Covich 2001). In fact, zooplankton exhibit movement patterns that can be generally understood as their swimming mode, which can be active (e.g. for cruise feeders and feeding-current zooplankters) or passive (e.g. for ambush feeders; Tiselius and Jonsson 1997; Visser and Thygesen 2003; Kiørboe 2011; Kjellerup and Kiørboe 2012; van Someren Gréve et al. 2017). There are species-specific, gender-specific and size-specific variations regarding swimming capacity and patterns (Folt and Burns 1999; Bianco et al. 2014; Almeda et al. 2017; Heuschele et al. 2017; McCloud et al. 2018; Ekvall et al. 2020), but the true capacity of actively swimming against flows to follow a given stimulus (e.g. food availability or predation avoidance) can be questioned.

In this context, the use of thermal surface exchange flows as a locomotory mechanism to zooplankters seems reasonable. This is an underlying hypothesis of the present study, which we explore by conducting an experimental campaign in Lake Vela, a shallow lake that exhibits temperature differences between littoral and pelagic areas, mostly when the former is covered with Myriophyllum.

Materials and methods

Study area

The study area is located in the southern end of Lake Vela (40.259984, −8.792322; Fig. 1), where dense, non-coincident littoral stands of the emergent macrophyte Schoenoplectus lacustris (hereinafter coded as S) or the submerged macrophyte Myriophyllum aquaticum (hereinafter coded as M) are established. A 48-h-long sampling campaign covering these two stands was run during July 2019. The time frame defined for the study was defined considering the typical plankton dynamics in Lake Vela and the climatic evolution of the spring season observed in 2019. In this way, a period of high abundance of zooplankton of a wide range of sizes, in its typical growing season, was targeted and successfully verified.

Fig. 1
figure 1

Location of the study area at Lake Vela (Quiaios, Portugal). The positioning of the temperature sensors in the limnetic (out) or littoral areas (in), relating to the Myriophyllum aquaticum stand (M) or to the Schoenoplectus lacustris stand (S) is shown, and a schematic detail of the deployment at each site is provided

Full size image

Lake Vela is a freshwater shallow lake with 70 ha surface area included in the Natura 2000 network (PTCON0055) (CM—Ministries Council 2000), which is integrated in the Quiaios system of interconnected reservoirs (Figueira da Foz, Portugal; Fig. 1).

The west bank of the lake is covered by a Pinus pinaster and Acacia spp. stand, and the east bank comprises agricultural fields (Abrantes et al. 2006b, 2010). Littoral vegetation across the lake includes Phragmites australis, Cladium mariscus, Nymphaea alba, Myriophyllum sp. and species of the Poaceae family (Antunes et al. 2003; Abrantes et al. 2006a; Castro et al. 2007b). The lake harbours several fish species including several dominant non-indigenous fish species: the pumpkinseed sunfish Lepomis gibbosus, the mosquitofish Gambusia holbrooki, the carp Cyprinus carpio and the largemouth bass Micropterus salmoides (Abrantes et al. 2006a; Castro et al. 2007a, b). The pumpkinseed sunfish, and even more the mosquitofish, are voracious visual omnivorous/planktivorous fish that typically exert strong predatory pressure on zooplankton (García-Berthou 1999; García-Berthou and Moreno-Amich 2000). The pumpkinseed sunfish was reported to comprise more than half of the fish assemblage in Lake Vela about 15 years ago (Castro et al. 2007b). No systematic records are available since then, but occasional surveys show that this species is declining, a trend possibly related to the increase in catfish populations following introduction, although large populations of mosquitofish are frequently observed (J Pereira and F Gonçalves, personal observation). The introduction of these fish species has displaced native species such as Barbus spp., Chondrostoma spp. and Squalius spp. (Castro et al. 2007a).

Water temperature measurements

The water temperature was monitored every minute during the 48 h of the field campaign through dedicated data loggers (HOBO® Pendant, MX2201) placed 5 cm below the water surface. Lake Vela is confirmedly polymictic (Castro et al. 2007b); hence major vertical temperature gradients are not expected. Four temperature loggers were deployed to record water temperature, inside the stand of each plant species (‘S-in’ and ‘M-in’, respectively) and in the corresponding limnetic zone (‘S-out’ and ‘M-out’) following littoral-limnetic axes; the sensors were placed in the lower surface of floating boards, which were moving structures guided by poles fixed in the lake sediment (Fig. 1). For the data analysis, the water temperature was sampled in time intervals of 7 min, which is the response time of the sensors in water.

Physiochemistry: sampling and analysis

Water physiochemistry was monitored along the 48 h of each field campaign at each sampling site (i.e. near each temperature logger), every 4 h (i.e. two measurements through the campaign at 0:00, 4:00, 8:00, 12:00, 16:00, 20:00). Dissolved oxygen, pH, total dissolved solids (TDS) and electrical conductivity were recorded in situ using a multiparameter field probe (AquaRead AP2000). Water samples were collected at a depth of 20 cm below the free surface. Water was collected into plastic bottles and vacuum-filtered (Nalgene 6132-0020, 36 cc) in the field through glass microfibre filters (1.2 µm pore size; 47 mm ø). Filtered and unfiltered sub-samples, as well as the filters, were stored and transported into the laboratory at 4 °C in the dark for further analysis. The residue was used to quantify total suspended solids (TSS; APHA 2017) and chlorophyll a (Chl a) content (Lorenzen 1967). The filtrate was used to quantify coloured dissolved organic carbon (cDOC; Williamson et al. 1999). Unfiltered water samples were used to estimate turbidity through the absorption coefficient at 450 nm (Brower et al. 1998). Additional aliquots of unfiltered samples were mineralized with potassium persulfate (Ebina et al. 1983) for quantification of total phosphorous content through the tin(II) chloride method (APHA 2017) and total nitrogen content through the cadmium reduction method (Lind 1979).

Zooplankton: sampling and analysis

Zooplankton was sampled following the same design and schedules as described above for water samples. For this purpose, 20 L of lake water was pumped (underwater electric pump, Nuova Rade 500 GPH, 1900 LPH, 12 V) and filtered in situ through a 200-µm mesh sieve. The sieve was then washed, and the retained organisms were preserved with alcohol (70% v/v). The sample volume was based on previous experience of the research team in the same lake, considering the feasibility of seasonal and long-term zooplankton sampling with diverse research focuses using volumes ranging from 5 to 25 L (Antunes et al. 2003; Abrantes et al 2006a; Castro et al. 2007b). Preserved samples were sorted under a stereoscope (Olympus SZX9), and the organisms belonging to the four major zooplankter groups in the lake assemblage (Copepoda, Bosminidae, Daphniidae and Chydoridae) were counted. Whole samples were analysed, and reference zooplankton keys were used as a support to assign organisms to these groups (Sandercock and Scudder 1996; Bledzki and Rybak 2016).

Data analysis

Data were grouped by diel period (night samples assumed as replicates for the night group and day samples assumed as replicates for the day group) and by macrophyte stand focused, both for value averaging or ranging and for the statistical analysis. A two-way analysis of variance (ANOVA) approach was applied to assess the effect of these two factors (diel period, with two levels; site, with four levels) and their interaction. Each dataset was transformed as necessary to meet ANOVA assumptions of normality of the distribution and homoscedasticity. Provided a significant effect of a given factor, the main effects for this factor within each level of the other were further inspected using the post hoc Tukey test. All analyses considered an alpha level of 0.05. A water temperature value was assigned to each sample, which was obtained as the average temperature of the readings obtained in the 15-min period preceding the sampling.

Results

Physiochemical context

The field campaign for the present study was carried out at the end of the spring season, when the daily mean atmospheric temperatures were around 20 °C, as measured at 5-min intervals by a nearby weather station. Over the period of 10 days before the study, the station recorded a mean air temperature of 18.3 ± 3.5 °C (mean ± standard deviation; n = 3328; data not provided). The water temperatures in the same period ranged within 21.0–28.2 °C (22.6 ± 1.7 °C). Although these ranges are typical in Lake Vela, they are slightly higher than the records commonly observed earlier in the season (Castro et al. 2007b). Electrical conductivity and pH varied within 600–643 µS/cm and 6.61–8.22, respectively (Table 1), the latter fulfilling the national criteria for good ecological potential (INAG 2009). However, high values for total dissolved solids were recorded (391–417 mg/L), as well as high TSS and cDOC levels, consistent with the observed low transparency of the water and nutrient levels (Table 1). Total N and total P were found at concentrations above national criteria for good ecological potential in northern lakes (INAG 2009) and typical of eutrophic to hypereutrophic water bodies, with Chl a levels as a phytoplankton biomass surrogate confirming this picture (Thomas et al. 1996; USEPA 2000).

Table 1 Mean ± standard deviation and ranges (within parenthesis) of records for physiochemical parameters obtained for the four study sites in Lake Vela, during the day (n = 8) or the night (n = 4)
Full size table

In general, there is no remarkable spatial (among sites) or diel (day vs night) variation regarding the physiochemical variables recorded in the present study (see Table 1 for measured ranges). The single exception to this finding was barely found for pH, which differed significantly among sites (Table S1) only during the night, but these differences were between unrelated sites (Fig. 2).

Fig. 2
figure 2

Mean pH, dissolved oxygen and chlorophyll a (Chl a) levels recorded at the study sites during the day (n = 8; white bars) or during the night (n = 4; grey bars). The error bars represent the standard deviation and the letters denote significant differences between sites during the night (Tukey test; p < 0.05)

Full size image

Dissolved oxygen was found to change significantly with the diel period by the omnibus two-way ANOVA (Table S1), with the mean levels always lower during the night (Fig. 2). The lowest dissolved oxygen levels (2.10–3.38 mg/L) at all sites occurred at 8:00 following the second night of the field campaign. Generally, the levels rose and peaked between 12:00 and 20:00 (maximum = 9.74 mg/L).

Since the variables that characterize sites and diel periods did not exhibit significant variations (Table S1, supplementary material), there is no evidence that spatial and temporal variation in the studied physiochemical parameters (except in some cases, discussed in Sect. 3.2, Chl a as a surrogate for food availability) can constrain horizontal zooplankton distribution patterns, serving as DHM drivers. There was room then to consider that predatory pressure (Castro et al. 2007a, b) could be a main biotic driver of any potential DHM patterns in Lake Vela, and that thermally driven currents could also be an important factor to explain the DHM patterns.

Horizontal distribution of zooplankton in Lake Vela

In the present study, no significant differences depending on the diel period or the sites were found in the zooplankton abundance (Table S2, supplementary material). The graphical analysis of sample abundance through time shows that theoretical patterns of whole zooplankton DHM driven by visual predators are followed very occasionally (Fig. 3). For example, within the Myriophyllum axis, zooplankton indeed seem to prefer (higher density) the littoral over the limnetic zone during the day in the first and last diurnal period, but there is one inversion of this preference in the second day time period (24 h). The expected (based on potential predation cues) reversion of the trend during the night was not observed; within the Schoenoplectus axis, preference patterns of whole zooplankton are even less clear, since it was common to observe similar abundance records in the littoral and the limnetic zones through time. It is noteworthy that sample size can be claimed as a bias source in the capturing of migration patterns in general. In the present study, we established sample volumes based on previous experience in Lake Vela (see Sect. 2.4). Moreover, our abundance records (see Fig. 3) are within the same order of magnitude as those obtained in previous studies in Lake Vela for equivalent seasonal periods and the same taxa (Antunes et al. 2003; Abrantes et al. 2006a); in some cases, we recorded lower abundance, but recent seasonal sampling campaigns confirm that the zooplankton assemblage in Lake Vela is declining coincidently with an increasing frequency of cyanobacteria blooms.

Fig. 3
figure 3

Abundance of zooplankters, and then group-specific zooplankters in the littoral (in) or the limnetic (out) area of Lake Vela along the Myriophyllum (M) and the Schoenoplectus (S) axis through the sampling period. Grey lines were added between the data points for clarity purposes only, hence not reflecting record continuity. Grey shadowing was used to mark the nocturnal period

Full size image

In summary, our results suggest that no DHM can be clearly recognized as a mechanism driven by interspecific relationships adopted by the zooplankton in Lake Vela.

Group-specific abundance patterns are not consistent with the trends found for the overall zooplankton assemblage, and the site had a significant effect in the abundance of Copepoda, Daphniidae and Chydoridae (Table S2). Copepods always preferred the littoral over the limnetic zone regardless of the diel period (Fig. 3); their abundance was significantly different between the two zones within the Myriophyllum axis in both the diurnal (Tukey test; p < 0.001) and the nocturnal (Tukey test; p = 0.005) period, and within the Schoenoplectus axis during the day (Tukey test; p < 0.001).

Daphnids showed an inconsistent abundance pattern, preventing any conclusions on their preference towards the littoral or the limnetic zone. In the Myriophyllum axis, abundance was higher or similar in the littoral throughout the whole sampling period, except at the beginning of the second night, when the organisms apparently preferred the limnetic zone (Fig. 3); this particular time point apparently configured the expected DHM pattern, but the statistics could not distinguish M-in and M-out in any of the diel periods (Tukey test: p = 0.372 for diurnal data; p = 0.853 for nocturnal data). In the Schoenoplectus axis, a significant preference for the limnetic zone was observed during the night (Fig. 3 and Tukey test with p = 0.036), consistent with the expected DHM, but this preference did not invert during the day (or abundance was similar in both zones).

Chydorids were consistent throughout the whole sampling period by preferring the littoral zone regardless of the macrophyte stand involved, with very little numbers found in the limnetic zone. This translated into significantly higher abundance records in the littoral for the Myriophyllum axis (Tukey test: p < 0.001 for diurnal data; p = 0.019 for nocturnal data), but the abundance did not differ statistically for the Schoenoplectus axis. Apart from particular time points at the beginning of the sampling period concerning the Schoenoplectus axis, bosminids tended to preferentially concentrate in the limnetic zone (Fig. 3), but the statistics did not confirm that site had a significant effect constraining abundance (Table S2).

Surface thermal flows in Lake Vela and their relationship with zooplankton distribution

Figure 4 presents the water temperature measured during the field campaign in the sampling sites (see Fig. 1), while Fig. 5 shows the discrete difference in the 15-min average water temperature in littoral and limnetic areas, for both axes, for each sampling time, allowing a more direct comparison with zooplankton discrete data.

Fig. 4
figure 4

Water temperature measured by the temperature sensors in the Myriophyllum axis (top) and in the Schoenoplectus axis (bottom) sampled with an interval of 7 min. The vertical dotted lines indicate the campaign sampling time points

Full size image
Fig. 5
figure 5

Difference between water temperature in littoral (In) and limnetic (Out) areas for Myriophyllum (circles) and Schoenoplectus (triangles) axes. Average temperatures for the 15-min period preceding the sampling hours were considered. The grey-shaded areas indicate the nocturnal sampling period

Full size image

The temperature records in the Schoenoplectus axis (S-in and S-out) indicate that the surface water is always warmer in the limnetic area than in the littoral where there is a vegetation cover, even during the night. The only exception was registered at the time of the first sampling (12:00 on 01/07/2019) when the records showed a short period with water temperature higher in the littoral than in the limnetic area. This might be a consequence of perturbations induced by the start of the campaign. The temperature differences observed vary between 0.2 and 0.5 °C during the day and are around 0.1 °C during the night. Although the temperature difference reduces during the night, the expected behaviour of the vegetation in relevantly slowing the cooling during the night was not observed, which may be related to the specific heat features of Schoenoplectus. Regarding the Myriophyllum axis (M-in and M-out), the temperature difference pattern is not consistent with the expected: during the day the water was warmer in the littoral than in the limnetic area, and during the night, temperatures in the littoral and in the limnetic area did not differ. The temperature differences observed during the day are significantly larger than those registered in the Schoenoplectus axis, with nearly 2 °C difference registered on the first day, corresponding to a mean density variation of 996.8 – 997.2 = −0.4 kg/m3 or −0.04%. During the night, the differences between surface water temperature in the littoral and limnetic areas were also around −0.1 °C.

Considering the temperature differences as the main driving force to induce surface exchange flows, no obvious density current would be expected in the Schoenoplectus axis during the entire observation period. This obviously excludes thermally driven currents as a contributor to explain whole and group-specific horizontal distribution of zooplankton in the Schoenoplectus axis. In the Myriophyllum axis, a surface current from the littoral to the limnetic area should be expected during the day, basically ceasing from late afternoon to morning. Correlating the observed distribution of the temperature differences with the horizontal variation in whole zooplankton abundance, no strong relation was found, as no consistent increase in the zooplankton abundance in the limnetic area and corresponding decrease in the littoral was observed in the Myriophyllum axis during the day. Furthermore, significant variations in zooplankton abundance were identified during the night period when weak or no surface currents were expected. Higher abundance of copepods, daphnids and chydorids was found in the Myriophyllum littoral area during the day, which is contrary to what would be expected if DHM patterns of these species were driven by thermal currents. Bosminids are always more abundant in limnetic areas, which meets the thermal flow expected in the Myriophyllum axis during the day, but is inconsistent with the expected absence of thermal flows therein during the night and in the Schoenoplectus axis. Thus, the acquired dataset does not allow definitive conclusions on the role of thermally driven exchange flows on the zooplankton horizontal migration patterns in shallow lakes.

Discussion

Concerning the physiochemical context of the study, the observed trend of higher pH values in vegetated littoral areas than in limnetic zones has been documented previously (Kairesalo 1980; Frodge et al. 1990; Arcifa et al. 2013). This generally relates to whether macrophytes are submerged, and thus rooted in the substrate (normally prevalent in shallower areas), or floating-leaf forms, for example, and thus relatively unaffected by the size of the water column. Regarding the dissolved oxygen, the lower values observed during the night correspond to the typical pattern in lakes, reflecting continuous oxygen consumption by respiration through the night, with limited input compensation versus maximum phytoplankton photosynthesis releasing oxygen to the water during the day (Andersen et al. 2017). While dissolved oxygen levels below 3.5 mg/L may challenge the survival of aquatic animals in the long term, zooplankters generally withstand these hypoxia levels for short periods (Burks et al. 2002). The diel variation observed in oxygen levels was consistent with the corresponding variation in Chl a (Fig. 2), but no consistent spatial pattern between littoral zones and limnetic zones was observed, which is in line with previous records in Lake Vela (Castro et al. 2007b) and in other lakes (Šorf and Devetter 2011; Špoljar et al. 2011).

High zooplanktivorous fish density is known to indirectly impact zooplankton communities due to the corresponding predation pressures, stimulating zooplankton movement to escape predation (Castro et al. 2007b; Jensen et al. 2010). In polymictic lakes, there is no effective depth gradient of abiotic conditions (e.g. temperature, light penetration), which is why it is frequently assumed that zooplankton move horizontally to find refuge, following a DHM featuring movement towards limnetic zones at night and towards the vegetated littoral during the day (Lauridsen et al. 1996; Burks et al. 2002). Reverse DHM patterns have also been reported and argued to be related to non-visual benthic invertebrate predators such as Chaoborus (Ringelberg 2009; Antón-Pardo et al. 2021) that typically swim at lower depth closer to lake banks, but the depth of the sampling sites in Lake Vela is homogeneous (50–70 cm, depending on the macrophyte stand covered) within the littoral–limnetic axes in our sampling area, rendering this alternative pattern unlikely. Moreover, in water bodies with high fish load, invertebrate predators are also preyed upon, which decreases the likelihood of a reverse DHM driven by non-visual invertebrate predators (Nurminen et al. 2007; Montiel-Martínez et al. 2015), and in Lake Vela in particular. No evidence of DHM was observed in Lake Vela at first; however, this analysis was based on the entire zooplankton assemblage. Recognizing that different zooplankters have different ‘swimming’ abilities and hence distinct capacity to actively move along horizontal axes (Tiselius and Jonsson 1997; Visser and Thygesen 2003; Kiørboe 2011; Kjellerup and Kiørboe 2012; van Someren Gréve et al. 2017), it is reasonable to examine potential spatial patterns among the different zooplankton groups.

The abundance of copepods was higher in the littoral than in the limnetic zone regardless of the diel period. Our expectation was that this group would consistently reflect predator-based DHM patterns regardless of any other potential drivers due to their long antennae allowing propulsive force of suitable magnitude and direction to enable an active protective behaviour. Indeed, copepods are able to exploit the mechanical energy to capture as many food items (algae) as possible, while expending as little energy as possible (Jiang et al. 2002b, 2002a; Jiang and Strickler 2007); they can detect and locate food particles (chemoreception and mechanoreception), swimming towards it (Bundy and Vanderploeg 2002). This suggests that food availability, which was slightly higher in the littoral as indicated by higher Chl a concentration in both macrophyte axes and regardless of the diel period, may drive copepod preference for the littoral over the limnetic zone.

Daphnids, as the largest cladocerans, are particularly susceptible to visual predation; hence, this group is generally expected to exhibit diel migration responding to refuge needs. It has been shown that patches of vascular macrophytes growing in littoral zones constitute a refuge to large cladocerans, and that the pursuit for protection is directly proportional to the density and structural complexity of these habitats, i.e. to the predatory capacity by fish within the patches (Manatunge et al. 2000; Padial et al. 2009). Submerged macrophytes are a more effective refuge for zooplankton than emergent plants (Hanson and Butler 1994; Jacobsen et al. 1997; Stansfield et al. 1997; Jeppesen et al. 1998; Ardohain et al. 2021), and our results are consistent in this context. First, the submerged density and complexity of the Myriophyllum patch was much higher and likely provided a more efficient refuge than the Schoenoplectus patch, largely limiting the capacity of even small fish exploring the littoral to actively prey upon zooplankters (e.g. Gambusia sp.). This may explain the preference of daphnids for the littoral regardless of the diel period within the Myriophyllum axis. Second, the shadowing within the Schoenoplectus patch is lower and would likely provide similar visual fields for predators between the limnetic and the littoral zone relative to the Myriophyllum stand. This may explain the preference, or similar distribution, of daphnids through these two zones within the Schoenoplectus axis during diurnal or nocturnal periods, respectively.

Chydorids and bosminids have similar mean body size ranges, being amongst the smallest cladoceran representatives, and definitively smaller than Daphniidae (Rizo et al. 2019). In this context, clear patterns of DHM driven by predatory cues would not be expected for these two groups, as they would be inherently protected from visual predators by their small size (Rizo et al. 2019), although this perspective has been challenged (Jeppesen et al. 1998). In the present study, chydorids preferred the littoral zone, with higher abundance generally recorded within Myriophyllum paths, and bosminids preferred the limnetic zone in general; differences should likely be due to their ecology. This difference can be primarily explained by the ecological niches involved. As compiled e.g. by Nevalainen (2010) or Klemetsen et al. (2020), chydorids evolved coupled with the diversification of aquatic plants, and the group is composed mostly of littoral species exploiting different substrates through their various locomotor and feeding adaptations. Our results agree with the preference for the littoral by chydorids (note that limnetic abundance of these organisms was very low throughout the sampling campaign) and also with the role of vegetation in shaping distribution (Tremel et al. 2000; Adamczuk 2014), with higher abundance generally recorded within the Myriophyllum stand than the Schoenoplectus stand. Bosmina have a different feeding flexibility and locomotory behaviour than other cladocerans because they feed more like a raptorial predator than a passive collector, and they can select food items upon availability, which is an energy-efficient mechanism allowing them to share habitat with competitors without the need for costly spatial migration (DeMott and Kerfoot 1982).

Challenges to and inconsistencies with zooplanktonic DHM theory are well known and relate to the contrast between predation and prey refuge (Burks et al. 2002; Nurminen and Horppila 2002; Meerhoff et al. 2006; Castro et al. 2007b; Jensen et al. 2010; Arcifa et al. 2013; Antón-Pardo et al. 2021), as well as to the role of water transparency in moderating the relationship. Turbidity, which is high in Lake Vela, has a consistent negative effect on prey capture by visually oriented predators, and there is also evidence that high turbidity leads to reduced prey capture in non-visual predators (Ortega et al. 2020). The behaviour of dominant planktivorous fish in Lake Vela may also contribute to the inconsistencies, because young pumpkinseed sunfish tend to prey in the littoral (García-Berthou and Moreno-Amich 2000), as do mosquitofish, mostly upon littoral cladocerans (García-Berthou 1999). Unfortunately, there are no systematic records on the fish assemblage of Lake Vela at the time of the sampling, and mosquitofish were consistently observed near both vegetated areas during the sampling period, while pumpkinseed sunfish were more rarely observed. The role of wind and thermal currents in modulating the spatial heterogeneity of zooplankton distribution in lakes has been postulated (Okely and Imberger 2007), but these ideas have also been questioned as factors affecting zooplankton spatial distribution (Lévesque et al. 2010).

In a shallow lake with littoral regions populated by emergent vegetation, differential solar heating can produce near-surface temperature differences between vegetated and non-vegetated regions. During the day, especially on sunny days, the shadowing effect on the littoral areas should reduce surface water heating relative to the limnetic areas, leading to the generation of horizontal exchange flows towards the vegetated areas (Zhang and Nepf 2009). During the night, the cooling effect is expected to be more efficient in the limnetic areas, leading to a surface flow from littoral to lake open areas. Nevertheless, the water temperature measurements of the present work did not follow that expected pattern. In the Schoenoplectus axis, the surface water in the limnetic area was always warmer than in the littoral. In the Myriophyllum axis, the water was warmer in the littoral than in the limnetic area during the day, whereas the temperatures were similar in the littoral and limnetic areas during the night. This behaviour is likely related to the type of vegetation and its specific heat features. Myriophyllum aquaticum is characterized by very dense plant distributions with short canopies. A potential large heat absorption by the plants might contribute to a faster water heating process compared with the limnetic area.

The correspondence between the expected surface exchange flow based on the water temperature differences and the abundance of the whole and group-specific zooplankton was not definitive with respect to the role of thermally driven exchange flows on the zooplankton horizontal migration patterns. Lake circulation is prone to the influence of several factors leading to complex flow (Zhang and Nepf 2009; Mao et al. 2019; Naghib et al. 2018). In shallow lakes in Mediterranean regions with high temperatures and exposure to annually prevalent winds (North Atlantic Anticyclone combined with North Atlantic Oscillation), the wind pattern and intensity may have a significant influence on the surface exchange flow and, therefore, on the spatial distribution of zooplankton. The potential role of the wind in this process will require further analysis and additional study.