Chironomidae (Insecta: Diptera) are the most widespread of all freshwater macroinvertebrate families (Ferrington, 2008) occurring in almost all kinds of freshwater bodies of all zoogeographic regions over the world (Sæther, 2000). They are also often the most abundant and diverse group in freshwaters, representing up to 50% of the macroinvertebrate community (Armitage et al., 1995). Chironomids are widely regarded as effective indicators of water quality and changes in habitat conditions since they exhibit a great variety of ecological traits and can be present over a wide range of environmental conditions (Battarbee, 2000; Walker, 2001; Cañedo-Argüelles et al., 2012).
Temperature has been traditionally considered one of the most important factors controlling chironomid distribution (Eggermont and Heiri, 2012; Marziali and Rossaro, 2013), but other factors such as water depth (Korhola et al., 2000; Engels et al., 2012), dissolved oxygen (Little and Smol, 2001) or trophic status (Brodersen and Quinlan, 2006) can also be important. Moreover, Chironomidae communities experience marked temporal and spatial variations in their composition. The temporal variability of the Chironomidae community is mainly related to the life cycle characteristics (Heinis and Davids, 1993), which lead to different emergence patterns that are known to be principally controlled by water temperature and light intensity (Kureck, 1980). Chironomidae spatial changes result from different factors (e.g., aquatic vegetation, dissolved oxygen) affecting the community at different depths and habitats (Prat and Rieradevall, 1995; Eggermont et al., 2008; Frossard et al., 2013).
In each lake the key factors controlling Chironomidae communities may change according to local and global conditions. Understanding the response of the current communities can be helpful to interpret the past and predict future changes, since Chironomidae capsules are well preserved through time in the sediment, allowing us to conduct paleolimnological studies (e.g., Walker, 2001). In this regard, local and regional studies of the actual fauna are necessary to calibrate how the community has changed and will change in the future if Chironomidae want to be used for predictive studies of global change (e.g., using transfer functions) (Luoto, 2010). At the same time, by exploring the spatial and temporal variations in community composition, the influence of local factors (e.g., land use, organic enrichment) can be disentangled. Thus, Chironomidae can provide useful information for water managers, helping them to adopt appropriate measures to achieve and maintain the good ecological status of lakes, e.g., as required in Europe by the Water Framework Directive.
Lakes have been widely considered as sentinels of climate and environmental changes (Adrian et al., 2009; Schindler, 2009). Among them, high mountain lakes are especially sensitive to external forcing due to their characteristics (high altitude, high UV radiation, low nutrients, etc.) (Catalan et al., 2006). Thus, special scientific attention has been paid to this type of lakes (Battarbee, 2005; Catalan et al., 2009a). In the Iberian Peninsula, there are several studies focused on chironomid communities of high mountain lakes (Real and Prat, 1992; Rieradevall and Prat, 1999; Real et al., 2000), although the majority took place in the Pyrenees (Catalan et al., 2009b; De Mendoza and Catalan, 2010) or in the Central Range (Toro et al., 2006). Furthermore, these studies usually focus on the littoral part of the lake, with only a few covering the whole depth transect (Rieradevall et al., 1999; Rieradevall and Prat, 1999) and usually only sampling once or twice, which implies the loss of information related to seasonal variability. Here we address seasonal and spatial variability of the Chironomidae communities, providing with certainty the most intensive work conducted on a single high mountain lake of the Iberian Peninsula to date.
Our main hypothesis was that depth would play an important role in explaining chironomid community changes in Enol Lake through changes in environmental factors. Together with depth, we hypothesized that the presence of macrophytes and algae in the lake should be important, since the macrophyte and algae-related chironomid species tend to be very different than the sediment-related ones (Langdon et al., 2010; Vermaire et al., 2013). The knowledge of the factors that control the community in the lake today are a preliminary step in understanding the past changes in the community (using the subfossil midges present in the sediment) and the possible future changes related to local, regional and global drivers, which will be the objective of the forthcoming studies in the lake.
Enol Lake (43° 16’ N, 4° 59’ W, 1070 m asl) (Fig. 1) is a karstic lake of glacial origin located in the northern part of Spain (Asturias), in the western massif of Picos de Europa National Park. It has a water surface of 12.2 ha, a maximum depth of 22 m and a small watershed (1.5 km2). The lake is fed by groundwater and surface runoff and it has no permanent inlets. Water looses occur through evaporation, groundwater discharges and an outlet located at the northeast border of the lake, which is regulated by a small dam. Previous surveys (Velasco et al., 1999; Moreno et al., 2011) and the data collected in our study (see sampling subsection) characterize the lake as warm monomictic (with a thermocline located between 8 and 12 m from early July until early November). The lake is oligotrophic (total phosphorous 8 µg L–1, Chl-a 0.5-1 µg L–1), moderately hard (alkalinity 2.4 meq L–1 and 24-37 mg Ca L–1) and with a conductivity between 150 µS cm– 1 at the surface and 227 µS cm–1 at the bottom. It is almost fully covered with a great carpet of Chara sp. between 2 and 8 m of depth, while Potamogeton natans occurs between 1 and 3 m of depth. Despite this oligotrophy the bottom of the lake is devoid of oxygen for 4 months every year during the stratification period (García-Criado and Martínez-Sanz, 2010).
Samplings were performed in two consecutive years (2013 and 2014), in May, July, September and November of each year. In every sampling campaign, both littoral and bottom samples were collected. In the upper littoral zone, we took 3 samples per campaign belonging to the three different dominant habitats found in the lake: i) sediment, ii) stones and iii) blocks. Samples were collected using the kick-sampling method (sampling surface: 1 m2) with a 250 µm mesh net and the filtered sample was preserved in formaldehyde at 4%. In the case of the sublittoral and profundal zones (2-22 m), we used an Ekman grab (sampling surface: 225 cm2), with three replicates per sample following a depth transect, collecting samples every 2 m. These samples were also sieved in the field using a 250 µm mesh net and preserved in formaldehyde at 4%.
In the laboratory, we sorted all the Chironomidae larvae present in the sample up to a maximum of 300 individuals. Initially, we sorted the chironomid specimens by morphotypes. A certain number of larvae of each morphotype were treated with 10% potassium hydroxide (KOH) at 70°C and, after dehydration, were mounted on microscope slides in Euparal®. The Chironomidae specimens were identified using an optical microscope (Olympus CX41) at 400x magnification and several taxonomic keys (Wiederholm, 1983; Rieradevall and Brooks, 2001; Brooks et al., 2007). Later, when morphotypes were identified, several taxa were counted directly from samples without further processing. The identification of larvae to species level was validated through the examination of a large collection of pupal exuviae from the lake using the key of Langton and Visser (2003).
Environmental variables were collected by researchers of the Department of Geological Resources Research of the Geological and Mining Institute of Spain (IGME) in the same 8 campaigns. Temperature, pH, dissolved oxygen and conductivity were measured every 0.5 m of depth using a multiparametric sensor (Hydrolab MS5 and DS5). Water samples were collected every 5 m of depth for total organic carbon (TOC), total inorganic carbon (TIC), nutrients, major ions and trace metals analyses. These samples were analyzed in the laboratory following standard methods (Rice et al., 2012).
The general distribution of the chironomid taxa along depth was represented by the program Psimpoll (Bennett, 2009), using the mean of the relative abundances of each taxa per depth. Statistical analyses were performed using R software (R Core Team, 2016). Rare taxa, defined by those that did not reach a relative abundance of ≥2% in at least two samples, were removed from the analyses. For the statistical analyses, we used total abundances, which were previously transformed to log (x+1).
Assemblage ordination was explored through Nonmetric Multidimensional Scaling (NMDS) of the species matrix using the function ‘metaMDS’ of the R package ‘vegan’ version 2.4-0 (Oksanen et al., 2016) and the Bray Curtis dissimilarity index to calculate the resemblance matrix. We classified the samples in 3 groups through a k-means clustering, then the indicator taxa of each group were identified through an IndVal analysis (Dufrêne and Legendre, 1997) using the ‘multipatt’ function in the R package ‘indicspecies’ (De Cáceres and Legendre, 2009). This analysis assigned each taxon to a most probable group based on its relative abundance and relative frequency and provided an indicator value (IV, varying between 0 and 1) and an alpha value obtained by Monte Carlo permutations (9999 runs). The differences in the environmental characteristics of each group were tested by means of one-way ANOVA of normalized variables. In this case, we used temperature, pH, dissolved oxygen, conductivity and Chara sp. abundance as environmental descriptors. When ANOVA yielded significant results, Tukey tests were performed to evaluate pairwise comparisons. Finally, temporal variations in the chironomid assemblages were explored through a Redundancy Analysis (RDA) at 5 depths (0.5 m, 5m, 10m, 15m and 20m) with forward selection of explanatory variables using the ‘rda’ and ‘ordistep’ functions of the ‘vegan’ 2.4-0 package (Oksanen et al., 2016) respectively. For these analyses, all environmental variables measured were included. Prior to this, Spearman correlation matrices were calculated in order to discard those variables that were significantly correlated (i.e., ρ≥0.85).
We identified a total of 14,248 chironomid larvae belonging to 27 species from 4 sub-families. The most diverse and abundant was the Chironominae sub-family, with 12 species of the Tribe Chironomini and 4 species of the Tribe Tanytarsini, followed by Orthocladiinae (7 species), Tanypodinae (3 species) and Prodiamesinae (1 species). Among them, we identified Einfeldia pagana (Meigen, 1838), which is considered a rare species in the Iberian Peninsula with only one previous citation (de Mendoza, 2013). All the identified taxa are shown in Tab. 1, with their names following the Fauna Europaea database nomenclature (Sæther and Spies, 2013) and the equivalent code proposed by Schnell et al. (1999) (Tab. 1).
According to the k-means clustering, samples were classified as follows: i) profundal samples (10-22 m), ii) Chara-dominant samples (2-8 m) and iii) littoral samples (0.5 m). All environmental variables included in the ANOVA significantly differed between these groups (Fig. 2). Temperature, conductivity, pH and dissolved Oxygen showed non-significant differences between the littoral and the Chara-zone group and significant differences between these two groups and the profundal samples. Chara sp. abundance showed non-significant differences between the littoral and the profundal groups and significant differences between these two groups and the Chara-dominant one.
Only a few taxa were present in the profundal zone, mainly represented by Chironomus (Chironomus) plumosus (Linnaeus, 1758) (Fig. 3). However, the IndVal analysis did not get any taxa exclusively associated with this group, as C. plumosus is also present in other zones (although with lower abundances). E. pagana was considered to be exclusively indicator of the Chara-dominant zone while Corynoneura lobata (Edwards, 1924), Polypedilum (Polypedilum) nubeculosum (Meigen, 1804), Stictochironomus sticticus (Fabricius, 1781), Procladius (Holotanypus) choreus (Meigen, 1804), Ablabesmyia (Ablabesmyia) monilis (Linnaeus, 1758), Parakiefferiella bathophila (Kieffer, 1912), Paratanytarus bituberculatus (Edwards, 1929) and Endochironomus albipennis (Meigen, 1830) were significantly associated with both Chara and littoral zones, although the last two species were significantly more abundant where Chara sp. was present (45 and 22 % respectively for P. bituberculatus and 12 and 5% for E. albipennis). Finally, Cladotanytarsus (Cladotanytarsus) atridorsum (Kieffer, 1924), Microtendipes pedellus (De Geer, 1776), Polypedilum (Pentapedilum) nubens (Edwards, 1929), Endochironomus tendens (Fabricius, 1775) and Eukiefferiella coerulescens (Kieffer, 1926), were significantly and exclusively associated with the littoral zone, with C. atridorsum being the most abundant (27%).
According to the NMDS 3 groups of assemblages could be identified (Fig. 4), similar to those obtained with the k-means clustering (littoral, Chara-dominant and profundal assemblages). The 8 samples belonging to the littoral group appear well separated in the NMDS plot. The Chara-dominant group included samples located between 4 and 8 m of depth, but also few samples at 10 m depth, on the edge of Chara sp. zone. Finally, the profundal group mainly included samples located below 10 m of depth, although three samples included in this group were collected at lower depth. These lower-depth samples were collected in soft sediment substrate, devoid of vegetation.
According to forward selection, temperature and dissolved oxygen significantly explained the temporal variation in the Chironomidae assemblages in the littoral zone (Fig. 5). As a consequence, the sites were ordered by sampling date, with higher temperatures and dissolved oxygen summer samples plotted in the upper right part of Fig. 5. Although magnesium (Mg) was selected at 5 m of depth and sulphates (SO42–) and bicarbonate (HC03–) were significantly selected at 10 m, they showed very narrow concentration ranges (i.e., 0-3 mg L–1 of Mg at 5 m, 2-15 mg L–1 of SO42– and 84-101 mg L–1 of HC03– at 10 m). Finally, no environmental variable was significantly related with the temporal changes in Chironomidae assemblages in the profundal zone (i.e., 15 and 20 m).
Our results show that, according to Chironomidae assemblages, Enol Lake is clearly differentiated in 3 zones: a littoral zone, a sublittoral zone dominated by Chara sp., and a profundal zone. In the littoral zone, both abundance and composition of Chironomidae changed in time. In the Chara-zone, Chironomidae assemblages were constantly dominated by a single species: P. bituberculatus. The profundal zone was dominated by C. plumosus, indicator of low oxygen concentration. Overall the diversity of midges decreased with depth, as previously reported in many other lakes (Lindegaard, 1992; Korhola et al., 2000; Chen et al., 2014). Furthermore, the spatial variation in the assemblages’ composition was consistent with the known autoecology of the identified taxa (Pinder and Reiss, 1983; Brodin, 1986; Pinder, 1986; Prat and Rieradevall, 1995; Brodersen and Lindegaard, 1997; Brodersen et al., 2001; Brooks et al., 2007). C. plumosus is known to live in water with low oxygen content (Brooks et al., 2007), whereas P. bituberculatus has been often associated with the presence of macrophytes (Brodersen et al., 2001).
C. plumosus is also commonly regarded as an indicator of eutrophication (Brodersen and Quinlan, 2006) due to its tolerance to low oxygen concentrations and even anoxia (Real et al., 2000). The dominance of this species in the deepest part of Enol Lake should suggest that the lake is eutrophic, but according to primary production rates the lake is oligotrophic. Since the nutrient content of the water is low (Velasco et al., 1999), the deoxygenation of the bottom should be related to a source of organic matter different from phytoplankton production. The lake basin is used for intensive pasture of cow (Rodríguez Castañón, 1996), which leads to important soil erosion and a direct input of cow manure into the lake. These factors increase the sediment and organic matter direct inputs into the lake bottom, as they are directly transported to the sediment. This process is very likely responsible for the high oxygen consumption of hypolimnetic waters. Organic matter measures in sediment cores from Enol Lake emphasized an increase of TOC levels in the sediment during the recent decades (Ortiz et al., 2016), supporting the idea of an increase in organic matter input to the lake due to human land uses in the basin area. The oxygen consumption is enhanced by the relative low volume of water in the hypolimnion and the extended period of the thermocline presence, lasting from early July to early November (García-Criado and Martínez-Sanz, 2010). The absence of oxygen in the deep zone has been also noted in many mountain lakes but usually without such dominance of C. plumosus, as it is not commonly found in these type of lakes (Walker, 1993). In the Pyrenees (Capblancq and Laville, 1983) and in Laguna Cimera (Central Range) (Granados and Toro, 2000), its presence was associated with oxygen depletion during the ice-cover period. However, this can’t explain the anoxic conditions found in Enol Lake, since the stratification period caused by ice-cover is very short if present. Thus, Enol Lake might constitute a good example of what could occur in mountain lakes if human activities occurring in the lake basin are not properly managed. This is a hypothesis that should be further tested since other factors (e.g., water temperature increase) could play an important role in oxygen depletion.
The presence of Chara sp. beds was mainly responsible for explaining changes in the chironomid community of the littoral and sublittoral zones. P. bituberculatus, E. albipennis, E. pagana and C. lobata were more abundant in Chara sp. beds, whereas C. atridorsum was present in littoral areas devoid of Characeae, similar to the results found by Van Den Berg et al. (1997). Since the littoral and Chara-dominant zones were not different in terms of key environmental variables (e.g., oxygen contents), our results suggest that the presence of Chara sp. was the key driver of Chironomidae assemblages in the littoral and sublittoral zones (i.e., the species commonly associated with macrophytes increased their dominance when Chara sp. was present). The importance of aquatic vegetation as a prime driver of Chironomidae biodiversity in lakes has been pointed out by several studies (Brodersen et al., 2001; Langdon et al., 2010; Cañedo-Argüelles and Rieradevall, 2011). Characeae density has also been reported to structure lake macroinvertebrate communities (Van Den Berg et al., 1997; Van Den Berg et al., 1998).
On the other hand, temperature has been widely acknowledged as one of the main drivers affecting chironomid community composition in high mountain lakes (Heiri and Lotter, 2003). In Enol Lake, the temporal variability of the littoral zone assemblage was mainly driven by seasonality. This is not surprising, since seasonal changes in temperature, light incidence and food availability, are known to have a great influence on Chironomidae assemblages through changes in adult emergence (Armitage, 1995; Tokeshi, 1995) and habitat conditions, e.g., higher temperature levels can enhance macrophyte production (highest Oxygen values) during the summer (Hering et al., 2010). In our case, spring and autumn samples were mainly represented by C. atridorsum and P. bathophila, whereas P.bituberculatus, M. pedellus and E. albipennis dominated the summer samples, suggesting different emergence patterns of the littoral community.
The results obtained in this study are crucial to understand the temporal and spatial dynamics of Chironomidae assemblages in Enol Lake, constituting an essential step forward to improve the interpretation of the past changes of the lake by means of the subfossil chironomid community. Future studies should focus on the relationship between the living and the recent subfossil communities in order to understand head capsules’ transportation and deposition processes. It would constitute the key point to understand how the living community is recorded in the sediments, which would allow a correct interpretation of downcore changes. Disentangling the importance of local, regional and global factors affecting the community is necessary to calibrate past changes and making future predictions. In this regard, we have shown that the littoral community is mainly driven by regional and global processes (e.g., temperature), whereas the profundal assemblages are mainly driven by local factors, i.e., sediment and organic matter discharge into the lake caused by pasture, leading to oxygen depletion in the hypolimnion. Our results also provide valuable insights for the managers to understand the current ecological status of Enol Lake and to evaluate which measures should be implemented to preserve or improve it. For example, the dominance of pollution tolerant species in the bottom of the lake suggests that its trophic status may be changing from oligotrophy to mesothrophy in the near future if no measures are taken to counteract the direct entrance of organic matter to the lake bottom caused by pasture in the lake basin.