Journal of Plankton Research academic.oup.com/plankt J. Plankton Res. (2023) 45(2): 266–277. First published online January 13, 2023 https://doi.org/10.1093/plankt/fbac074 ORIGINAL ARTICLE Self-filling enclosures to experimentally assess plankton response to pulse nutrient enrichments PAU GIMÉNEZ-GRAU 1,2 , LLUÍS CAMARERO 3, CARLOS PALACÍN-LIZARBE 1,4 , MARC SALA-FAIG 1,3 , AITZIBER ZUFIAURRE 1,5 , SERGI PLA-RABÉS 1, MARISOL FELIP 1,6 AND JORDI CATALAN 1,7, * 1 creaf, cerdanyola del vallès, catalonia e-08193, spain, 2 department of bioscience, aarhus university, aarhus c dk-8000, denmark, 3 ceab, csic, blanes, catalonia e-17300, spain, 4 department of environmental and biological sciences, university of eastern finland, kuopio fi-70211, finland, 5 área de biodiversidad, gestión ambiental de navarra-nafarroako ingurumenkudeaketa (gan-nik), pamplona-iruñea, navarra e-31001, spain, 6 departament de biologia evolutiva, ecologia i ciències ambientals, universitat de barcelona, barcelona, catalonia e-08028, spain and 7 csic, bellaterra e-08193, spain *corresponding author: j.catalan@creaf.uab.cat Received May 13, 2022; editorial decision December 7, 2022; accepted December 8, 2022 Corresponding editor: Beatrix E. Beisner Experimental nutrient additions are a fundamental approach to investigating plankton ecology. Possibilities range from whole-lake fertilization to flask assays encompassing a trade-off between closeness to the “real world” and feasibility and replication. Here we describe an enclosure type that minimizes the manipulation of planktonic communities during the enclosure filling. The enclosure (typically ∼100 L volume) consists of a narrow translucent cylinder that can comprise the entire photic zone (or a large part of it in clear deep lakes, e.g. 20-m long) and holds a sediment trap at the bottom for recovering the sinking material. The enclosures are inexpensive and straightforward to build. Thus, many can be used in an experiment, favoring the diversity of treatments and the number of replicates. They also are lightweight with easy transport and use in lakes that cannot be reached by road. The enclosures are fundamentally aimed at investigating the short-term response of the planktonic community, integrated across the photic zone, to pulse perturbations using before and after comparisons and multiple replication and treatments. The pros and cons of the enclosure design are evaluated based on experience gained in Lake Redon, a high mountain ultraoligotrophic deep lake in the Pyrenees. KEYWORDS: microcosms; mesocosms; nutrient additions; experimental plankton perturbations; phytoplankton; protists; lake field experiments available online at academic.oup.com/plankt © The Author(s) 2023. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE INTRODUCTION case, the ability to deal with high trophic levels and approximate hydrodynamics found in nature is always Planktonic ecosystems commonly show episodes of limited (Riebesell et al., 2010). Trophic interactions are higher productivity embedded within prolonged periods affected by the size of the experimental units (Marrase of reduced activity. This pattern is particularly discernible et al., 1992), and phytoplankton development appears in oligotrophic ecosystems where phytoplankton growth more consistent across enclosures of different sizes than increases during periods of water column mixing when bacterioplankton and zooplankton (Kuiper et al., 1983). nutrients are loaded into the photic zone from the deep Episodic nutrient enrichments in lakes occur at tempo- layers (Sommer, 1986; Goldman, 1988; Domis et al., ral scales of days to weeks. Therefore, hypothesis testing 2013). The study of these episodic enrichments is critical concerning the response of primary production, phys- for the assessment of the long-term biogeochemical iological traits, and community processes, up to protist dynamics and the understanding of the ecology and level, to nutrient additions can be adequately performed evolution of planktonic organisms and their assemblage using multiple enclosures (>10) of relatively small size into communities (Harris, 1980; Falkowski et al., 1998; (50–1000 L) during days to weeks (Riebesell et al., 2010). Mcandrew et al., 2007; Spivak et al., 2011). One of Here, we describe an enclosure system for in situ lake the challenges of these studies is untangling the role experiments of a relatively short duration that minimizes of co-occurring driving factors (Richardson et al., 2017; the manipulation of the planktonic community during Morabito et al., 2018). The nutrient enrichment coincides the enclosure filling and allows for many replicates at with a shift in various physical conditions, e.g. mixing a low cost. The enclosures are lightweight and folded depth, mixing intensity, light regime and temperature for transportation, so they can be used in remote loca- (Reynolds, 2006). Experimental additions are a way to tions without road access and deployed using small boats. disentangle the effects of targeted nutrients and other Although the method to enclose a water volume by lifting substances of biological influence (e.g. dissolved organic and unfolding a bag has been previously used—e.g. with matter, vitamins, toxicants) from other environmental the aid of divers (Menzel and Case, 1977)—the system we factors (Downing et al., 1999; Benton et al., 2007; describe is straightforward and valuable for perturbation Faithfull et al., 2011; Spivak et al., 2011). The available experiments of a few days to weeks. methods for nutrient enrichments cover a wide range The experience gained during the “Episodic Nutrient of laboratory and field settings, from flask assays to Enrichment Experiment (ENEX)” in Lake Redon, an whole-lake fertilization, each with its pros and cons, ultraoligotrophic deep high-mountain lake in the Pyre- encompassing a trade-off between closeness to the “real nees (Ventura et al., 2000), is used to illustrate the suit- world” and feasibility and replication (Riebesell et al., ability and limitations of the enclosure system described, 2010). The common feature of the different methods is and discuss the possibilities of alternative designs that that they aim to manipulate an environmental gradient could be applied in other systems following the same and explore its role in structuring communities and general design concept. The planktonic community in functional traits (Fraser and Keddy, 1997). Lake Redon is P-limited due to the long-range atmo- Designs of enclosures used in aquatic ecological spheric transport of reactive nitrogen (Camarero and research have been repeatedly reviewed (Parsons, 1982; Catalan, 2012). The primary objective of the ENEX Sanders, 1985; Scor-Working-Group-85, 1991; Riebesell experiment was to evaluate the effects of the fluctuations et al., 2010; Stewart et al., 2013; Legendre et al., 2018). in P availability and N:P imbalance on the structure of Enclosures of less than 1 L are limited to prokaryote the planktonic protist community (Zufiaurre et al., 2021) and protist experiments, whereas 10–100 L are required and its stoichiometry (Giménez-Grau et al., 2020). We first to consider zooplankton dynamics and larger enclosures describe the enclosure design and use as performed in the are necessary for higher trophic levels (Parsons, 1982). ENEX experiment, and, in the discussion, we comment The Scor-Working-Group-85 (Scor-Working-Group-85, on alternative options and designs according to other 1991) divided pelagic experimental enclosures into objectives and lake characteristics. three size classes: microcosms (<1 m3 ), mesocosms (1– Ecological field experiments are inevitably subject to 1000 m3 ) and macrocosms (>1000 m3 ). Microcosms trade-offs between what is desirable and what can be provide more experimental control (i.e. replication, afforded. Whatever the size of the experiment, replication controllability and homogeneity) and lower cost, and is a critical issue. The lower the cost, the higher the macrocosms are closer to reality, capture better some potential number of replicates and treatments and, thus, biological complexity (e.g. trophic interaction, dispersal), more robust results (Riebesell et al., 2010). The enclo- and allow recurrent sampling (i.e. time series). In any sure described here is low cost (in our case, ∼60e per 267 JOURNAL OF PLANKTON RESEARCH VOLUME 45 NUMBER 2 PAGES 266–277 2023 Fig. 1. (A) Enclosure sketch, deployment and filling method. Underwater view of the upper (B), middle (C) and bottom (D) parts of the enclosures. enclosure), simple in its components and deployment, and the bag and help the enclosure sink during deployment. highly reliable in results. Each enclosure is self-filled (see below) with ∼100 L of water from 0 to 20-m lake depths. The complete list of items and some extra details necessary for the enclosure METHOD assembly are provided in Table I. Enclosures cannot be reused without changing the polythene bags and cleaning Enclosure design the PVC tubes. The enclosures are constructed using tubular-shaped clear polythene bags (diameter: 8.5 cm; length: 20 m) and two polyvinyl chloride (PVC) tubes, attached one at each Enclosure deployment extreme of the bag (Fig. 1). The PVC tube at the lower The deployment of the enclosure consists of the following end (length: 0.5 m) is closed with a sealed stopper and steps. (i) The enclosure is assembled: the polythene bag serves as a sediment trap, while the upper tube (length: is folded around the upper tube, and this tube is held 1.5 m) enables the gaseous exchange with the atmosphere, together with the sediment trap by a rope lacing using an prevents the bag from collapsing, and facilitates the escape knot (Fig. 1A; Supplementary Information Video, attachments. An extruded polystyrene float is attached 0:19 min). (ii) The folded enclosure is placed horizon- to this upper tube to hold the enclosure at the water tally on the lake surface and filled with water (Supple- surface. A weight is tied to the sediment trap to stretch mentary Information Video, 1:05 min). (iii) The folded 268 P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE Table I: Items, materials and guiding comments to build, deploy and use the enclosure system described Item Materials Function Comments Upper holding tube PVC, 7.5 cm Ø, 1.5 m long Tube for attachment of the flexible The tube length depends on the bag, floats, rope and mooring bag’s length since it acts as a pole connection. It acts as the mast for to fold the bag around during the bag during deployment. deployment. The diameter must be adapted to the bag diameter. Transparent materials can be considered (for example, methacrylate) Tube bag roll Polythene, 8.5 cm Ø, 20 m long Defines the enclosure growth The polythene bags are sold in rolls environment of many meters (e.g. 200 m) and are available in many diameters, commonly from 4 to 30 cm, allowing for alternative designs without specific customer manufacturing demands. Lower sediment trap PVC, 7.5 cm Ø, 1.5 m long. Sealed Collect the sinking material. A funnel connecting the bag and tap bottom Attachment for the guiding rope tube should be introduced if the and weight net bag. bag diameter is enlarged. Floats Extruded polystyrene. Size tailored Maintain upraising buoyancy They have to be adapted to the to the weight of the enclosure and upper tube diameter. Two sets are lake wind conditions. used. One is located below the rope attachment, used for uprising during deployment. Another is above the clamp attachment, added after filling the enclosure to increase buoyancy. Clamps Stain steel. Adapted to PVC tube They hold together the bag and diameter tubes and are the attachment points of the rope and weight net bag. Net bag Any suitable material Bag for extra weight. Depending on the transport facilities, weight can be local material (e.g. littoral rocks) or specific items of known weight. Rope Non-elastic material It keeps the system folded before deployment. Controls the deployment from the boat, and the upper part is released once the desired depth is reached Addition and sampling Plastic tubes. 20-m long. About 1 L Initial nutrient addition (1 tube) and Taps are necessary to keep the tubes volume. With taps at the extremes. final sampling (5 tubes). nutrient solution within the tube during the introduction in the enclosure. No need for sampling. Fishing sinkers Lead and stainless steel. Used for sinking the nutrient addition and sampling tubes. Vacuum flask Polypropylene or glass. For emptying the sampling tubes by siphoning vacuum. Vacuum pump Hand- or battery-operated with Producing vacuum in the sampling Keep pressure as low as possible to gauge flask. maintain a sufficient flow, and only pump the tube volume for a balanced water column sampling. ENEX enclosure dimensions and features are indicated with comments on some alternative options. enclosure is allowed to sink until the depth equivalent the upper part to increase the buoyancy (Supplementary to the maximum length of the enclosure (20 m in the Information Video, 2:38 min). case described here), where the rope stops it (Supple- mentary Information Video, 1:45 min). (iv) When the rope tensions, the lacing, which holds the tubes together, Nutrient addition unties (Supplementary Information Video, 0:45 min). Once the enclosure is filled with lake water and attached The upper tube starts to float toward the lake surface, thus to the mooring, one may proceed to the specific addition gently filling the bag with water from the upper 20 m of of the planned treatment. A 20-m long thin plastic tube is the water column (Fig. 1A; Supplementary Information filled with ∼0.9 L of nutrient-enriched water. The tube is Video, 2:14 min). (v) Finally, extra floats are added to introduced inside the enclosure and is dropped to the 269 JOURNAL OF PLANKTON RESEARCH VOLUME 45 NUMBER 2 PAGES 266–277 2023 bottom with the help of weight at the front end trap is detached from the bag, and its water content is (Supplementary Information Video, 2:58 min). The tube collected and kept for later filtration in the laboratory. is open in the front end and closed at the back end using a tap so that it does not leak its content during the deployment within the enclosure. Once the enclosure’s DISCUSSION bottom is reached, the tube’s rear end at the surface Experimentation in difficult-to-reach sites is opened. The solution is released homogeneously along the enclosure’s water column as the tube is Large experimental enclosures in lakes require transport constantly withdrawn from the enclosure (Supplementary and operational facilities that are usually only available in Information Video, 3:14 min). Finally, the enclosure well-connected areas. Large mesocosm experiments have is attached to a rope that holds it to an all-enclosure been typically carried out not far away from limnological mooring (Supplementary Information Video, 3:37 m). laboratories [e.g. Belham Tarn, England (Lack and Lund, The enclosure deployment and nutrient additions can be 1974); Lake Biwa, Japan (Nakano et al., 2001)]. When the performed from a small boat without problems. experiments are performed across many sites (Scharfen- berger et al., 2019) or in locations difficult to reach (Halac Mooring et al., 1997), the enclosure volume quickly declines to smaller volumes, and replication is limited. If field exper- The enclosures are individually hung from a line of a imentation is too difficult, water samples are brought to mooring system, maintaining a distance between them labs, and flask assays are performed (Jacquemin et al., sufficient to minimize direct shadowing (Fig. 1B). An east– 2018). In the case of the ENEX experiment in Lake west alignment of the mooring system guarantees equiv- Redon, located in an alpine mountain landscape with- alent exposure of each enclosure to solar irradiance. out access by road or easy tracks, all the material was Placement needs to consider also potential surrounding transported in backpacks. Even with that restriction, 22 relief shadows. Depending on the lake dimensions, the enclosures were deployed. Therefore, one of the strengths mooring system can consist of buoys and anchor weights of the described system is the affordability of an elevated or be attached to existing platforms or the shore. The number of treatments and replicates even in remote weight at the base of each enclosure should be sufficient to locations. avoid vertical deflection of the enclosure during windy sit- uations. Using a net bag and on-site stones allow adjusting the necessary weight without transporting extra weights Experiment duration to the site (Fig. 1D). If strong currents are expected, more The enclosure system described has been thought of as a buoys and anchor weights interspersed between every few way to perform short perturbation experiments on the enclosures can be added. In Lake Redon, winds can be planktonic community. Colonization is excluded, and relatively strong, but the fetch is short, we were using the response depends on the existing species assemblage only mooring at the extremes, and the line sustained 22 enclosed during deployment. Therefore, the duration enclosures (Giménez-Grau et al., 2020). of the experiment has to be tailored to the response capacity of the community. Ideally, the investigation will Enclosure final sampling focus primarily on growth responses, thus avoiding decay These enclosures should be considered microcosms phases and secondary successions within the enclosure. (<1 m3 ) according to the Scor-Working-Group-85 (Scor- The amount of material collected in the sediment trap Working-Group-85, 1991) classification. Intermediate compared with the material remaining in the water sampling during the experiment will cause too much column of the transparent bag will inform about the disturbance in the relatively small volume (∼100 L). growth phase of the community. However, this will At the end of the experiment, an integrated (0–20 m be a piece of information gathered at the end of the deep) water sample can be obtained from each enclosure experiment. Alternatively, a fluorescence sensor can by sinking 20-m long plastic tubes with weight. In the be installed in one enclosure where higher growth is ENEX experiment, we used five clean tubes, each of 1-L expected. Another option is to dedicate an extra enclosure capacity, deployed simultaneously. The integrated sample for determining the end of the experiment, in which was recovered by pumping the exact volume retained in fluorescence profiles are performed regularly to follow each deployed tube into a vacuum flask using a hand the temporal sequence of the response. According to pump to reduce potential cell damage (Table I). Once the the objectives, this enclosure should be treated with the water column part is sampled, the enclosure is uplifted expected more informative additions. However, it should gently until reaching the sediment trap. The sediment not be used in the intercomparison of the treatments, as 270 P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE profiling will introduce a highly artificial mixing in the stress (e.g. light, temperature, pressure), and the verti- enclosure. cal heterogeneity in the photic zone is fully captured In extremely oligotrophic cold systems such as Lake during the regular and slow-pace filling. There is an Redon, the duration can last several weeks without signif- initial community representative of the photic zone in icant decay of the responding populations. In the ENEX each enclosure. Enclosure and incubation systems that experiment, the enclosures were deployed on 5–6 August require pumping or weighted integration of samples in 2013 and recovered 25 days later. Given the starting low containers may introduce initial biases (e.g. high mortality populations in the epilimnetic waters and the relatively of abundant but sensitive organisms) compared with the low-temperature conditions (14–6◦ C), a few weeks were self-filling system used here (Riebesell et al., 2010). a reasonable approach (Parsons, 1982). Shorter exper- The main power of experimental approaches is reduc- imental durations are expected in warmer and more ing the causal effects to those manipulated by the exper- productive locations. In highly productive systems, the imenter. Unfortunately, field experiments can hardly fol- experiment should last from a few hours to a few days. low the optimal experimental designs equivalent to those The most complicated systems could be those that expe- obtained in a lab. The ENEX experiment took about 12 h rience large periodical fluctuations in productivity. They to deploy and perform the additions in the 22 enclosures. may show extremely clear waters but hold a community The relatively long-lasting and self-filling system involves capable of significant and rapid blooms, like some marine some risk of increasing the variation in initial conditions. systems (Estrada et al., 1988). Because of the low cost of The enclosures are not filled from a homogenized com- the enclosures, a preliminary test for adjusting the time of mon pool and could cause noisy responses to the treat- the experiment is recommended. ments. It is assumed that the treatment effects would be stronger than the random variation in initial conditions. Sampling and measurements However, the only way to assess this risk is by checking the final coherence of the results and the similarity between The enclosure design is conceived to integrate changes replicates. In the ENEX experiment, which dealt with a across the epilimnetic or photic zone gradients. The tubes’ P-limited planktonic community, we expected an increase system described above delivers 5 L of an integrated in primary productivity across the P-enrichment gradient sample, which allows for many measurements, but other and no significant productivity changes with N imbal- volumes can be considered, including pumping the entire ance. Protist community composition changes were more enclosure. For instance, in the ENEX experiment, we unpredictable, but we expected similar responses between measured in each enclosure: total dissolved phosphorus treatment replicates. Indeed, there was a coherent general (TDP), NH4 + , NO2 − , NO3 − ; dissolved organic carbon response across the P- and N-addition gradients and (DOC), particulate C (PC), particulate N (PN), particulate similarity in the replicates (Giménez-Grau et al., 2020). P (PP), chlorophyll a (Chla) and other pigments by liquid It may be argued that these general patterns may have chromatography, protists abundance, and biovolume, and been obtained with other methods, for instance, incuba- the prokaryote abundance estimated as 4 ,6-diamidino-2- tion flasks. Although these may be true, the enclosure fenilindol stain (DAPI) counts. The same biogeochemical system provides a more robust result because it minimizes variables were measured in the sediment traps, and the potential cell damage during manipulation than field diatom valves and chrysophyte cysts were assessed. The or lab flask assays. Despite being an artificial system, it is critical point is to maintain top-to-bottom integrated still much closer to the real world than the lab’s small flask water samples, as the heterogeneity within the enclosure assays. In flask experiments, water must be obtained either may be vertically substantial. The method’s strength from single depths or integrated across the water column is comparing treatments rather than a time series of and mixed. In the first case, many samples are required changes within the enclosure. The initial conditions to view the photic zone comprehensively; in the second, must be assessed with a similar integrated sample of the plankton populations are subjected to substantial stress. lake water column at the beginning of the experiment. In the enclosure system, the haphazard initial differences Comparing the initial and final conditions provides during filling could be expected to have more influence on information about rates as the system is basically closed the species composition than on the bulk phytoplankton to dissolved and particulate matter exchanges (Zufiaurre (Chl) growth. However, it was not the case in the ENEX et al., 2021). experiment, and the changes in composition were also coherent between replicates (Zufiaurre et al., 2021). This Coherence and replication coherence is shown by the close position of the replicate The enclosure’s filling method is quite friendly for the pairs in an ordination of the protists’ composition in that organisms as they are not submitted to any particular experiment (Fig. 2). 271 JOURNAL OF PLANKTON RESEARCH VOLUME 45 NUMBER 2 PAGES 266–277 2023 Table II: Nutrient initial treatment values that defined gradients of P-enrichment and N-imbalance in the ENEX experiment Gradient Treatment TDP NO3 − NH4 + DIN DIN:TDP μmol L−1 μmol L−1 μmol L−1 μmol L−1 (a/a) P-enrichment NO_P++ 1.9 16.8 0.2 17 9 P-enrichment NH_P++ 1.9 4.2 12.8 17 9 P-enrichment NO_P+ 0.21 16.8 0.2 17 81 P-enrichment NH_P+ 0.21 4.2 12.8 17 81 P-enrich/N-imbal NO_P 0.06 16.8 0.2 17 283 P-enrich/N-imbal NH_P 0.06 4.2 12.8 17 283 N-imbalance NO+_P 0.06 34.8 0.2 35 583 N-imbalance NH+_P 0.06 4.2 30.8 35 583 N-imbalance NO++_P 0.06 72.8 0.2 73 1217 N-imbalance NH++_P 0.06 4.2 68.8 73 1217 Control/Initial NA 0.022 4.2 0.2 4.4 200 The two gradients combined resulted in two similar N:P stoichiometric gradients, differing in the N primary source (Giménez-Grau et al., 2020). The control values indicate the average characteristics of the initial filling water Chrysophytes are the dominant group in the phy- toplankton of Lake Redon (Felip et al., 1999). These organisms produce resistant silica cysts for which spe- cific catalogs have been developed (Pla, 2001) to study their sediment records and ecological indicator value (Pla-Rabes and Catalan, 2011). The sediment traps at the end of the enclosures allow evaluation of the response to the nutrient enrichments in terms of the cyst fluxes and composition. The chrysophytes’ cyst fluxes in all the additions were higher or similar to the control enclo- sures (Fig. 3). However, the response patterns along the gradients differed from those shown by Chla. The most significant fluxes were found in the low-addition, implying that some other phytoplankton groups were responsible for the Chla increase observed. In fact, there was an increase in cryptophytes, diatoms, and chlorophytes at Fig. 2. Ordination of the protist species composition in the treatment enclosures of the ENEX experiment using a principal component higher nutrient levels (Zufiaurre et al., 2021). analysis with the Hellinger distance (Legendre and Gallagher, 2001). Although the enclosure characteristics are not appro- Treatment codes and characteristics are indicated in Table II. The first priate for maintaining diatoms in suspension, there was a axis mostly discriminates between phosphorus treatment levels, and the second axis nitrogen additions. See Zufiaurre et al. (2021) for details on significant growth of Fragilaria nanana, a small and thin life species composition. Note the proximity between replicas. form, in higher P treatments of the ENEX experiment. In the initial community, another planktonic diatom, Cyclotella pseudostilligera, was relatively abundant; however, The added value of the sediment trap this species apparently did not proliferate within the The enclosures are intended for before/after experi- enclosures in any treatment according to the specimens ments. We may wonder if what is found at the end accounted for in the water column of the enclosures. corresponds to the primary response to the treatment or, However, we found that the patterns of the two species in perhaps, the experiment has been too long, and we are the sediment traps were quite similar (Fig. 4). Therefore, sampling a decaying phase of the community response. Cyclotella was actually responding to the treatments but The comparison between the material accumulated in the sank much faster than Fragilaria, leaving an apparent sediment traps and the material remaining in the water negative growth in the bags. The ratio between the two column is highly informative. In the ENEX experiment, species in the sediment traps indicated that F. nanana grew the amount of particulate organic matter in the water relatively better in P-enriched treatments than Cyclotella column of the enclosures was two orders of magnitude and, on the contrary, showed more inhibition by N higher than in the sediment traps, thus indicating that excess. Previous field observations had shown that these most of the initial response had not been sedimented. two species grew during the period of the thermocline 272 P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE Fig. 3. Effects of P enrichment and N imbalance on chlorophyll-a levels (A) and chrysophyte cyst fluxes (B) during the ENEX experiment. The sediment trap provides a complementary view to the bag. Details of the treatments are indicated in Table II. Replicates are plotted individually except in the case of the initial conditions (In), for which the average value is shown as a linear reference. Note that controls (NA) and the samples for the lowest P and N additions are the same in the P enrichment and N imbalance gradients. Red and dark blue refer to treatments with ammonium as the nitrogen source, and yellow and light blue to treatments with nitrate. deepening in Lake Redon, when mixing starts to contact content (Gust, 1977). However, it will remain under- the first fine sediment layers, increasing nutrients in the mixed relative to the external water or, at least, differently water column: Fragilaria peaked in September and was mixed. If the community (e.g. diatom dominance) or the followed by Cyclotella in October (Catalan et al., 2002). topic to be studied [e.g. grazing rates (Marrase et al., The enclosure experiment confirmed a quick positive 1992)] are highly dependent on mixing conditions, the response of the two species to P increase in oligotrophic enclosure described does not appear as a good option. waters. Without the sediment trap, we would conclude The light environment is also greatly modified by the opposed responses for the two species. enclosure walls. The relative irradiance gradient across the photic zone may be maintained, but the light quality is markedly modified because of the absorption proper- The enclosure intrinsic effects ties of the bag material. In the ENEX experiment, the The enclosure tubes have a large surface-to-volume ratio polythene walls drastically reduced the high UV radia- that markedly modifies the hydrodynamic conditions, and tion levels existing in high mountain lakes (Sommaruga, the plastic bags affect light conditions. Hydrodynamic 2001). Indeed, preserving UV radiation in field experi- conditions are always far from natural conditions in enclo- ments is not easy and requires low UV-absorbing mate- sures and mesocosms. Artificial mixing in enclosures usu- rials (Halac et al., 1997). Furthermore, polythene reduces ally results in hydrodynamic conditions that may be as photosynthetic active radiation (PAR) between 20% and unrealistic as no mixing (Riebesell et al., 2010). In fact, 30% depending on wavelength (Cutroneo and Torrisi, there is an inherent problem in generating turbulent 2014). Therefore, light conditions are drastically mod- spectra comparable to those occurring in lakes and oceans ified. One positive aspect is that the large surface-to- in any enclosure or laboratory experiments. The charac- volume ratio provides homogenous light conditions at teristic length within which energy dissipation is averaged each depth within the enclosure (Berg et al., 1999). is much shorter in the enclosures. The controls, enclosures without additions, account for Consequently, either the rate of energy dissipation is the enclosure effect. In the ENEX experiment, the first too high or, if realistic, the Reynolds number (inertial forc- axis of variation shows that control enclosures were the ing against viscosity) is too low (Osborn, 1996; Catalan, most similar to the initial communities at the end of the 1999). In the enclosures described, if the wind induces experiment (Fig. 2), indicating that enclosure effects were some lake currents, the flexible walls and the long and not playing a determinant conditioning effect upon the thin shape of the enclosures can gently stir the inner water primary response. However, initial and control samples 273 JOURNAL OF PLANKTON RESEARCH VOLUME 45 NUMBER 2 PAGES 266–277 2023 Fig. 4. Sedimentation fluxes of the two main diatom species growing during the ENEX experiment evaluated with the sediment traps at the enclosure’s bottom. (A) Fragilaria nanana, (B) Cyclotella pseudostelligera). The sediment trap helps to assess the response of rapidly sinking organisms. differed markedly concerning the second axis, probably In four enclosures, we did not find any adult crustaceans summarizing the significant environmental change within (∼20%), <50 individuals in 62% of the enclosures, <100 the enclosure. Only a few species were strongly negatively in 90% and <300 individuals in all of them. The dif- affected in all treatments (<5%). They included species ferences were mainly due to Cyclops, in which juvenile with contrasting life forms, such as the chrysophycean, and adults are mainly predators of Daphnia in this lake Pseudokephyrion inflatum, and the ciliate, Askenasia acrostomia. (Ventura and Catalan, 2008). In addition to different Most species grew in many treatments (Zufiaurre et al., avoidance reactions, the patchy nature of the populations 2021). In any case, the enclosure effect can be accurately in the water column may add some noise. Despite the evaluated because of the possibilities of high replication. variation, no significant influence of large zooplankton composition on the results obtained was found. The zooplankton exclusion can be enhanced by insert- Large zooplankton ing a mesh in the frontal opening of the enclosure. How- ever, the finer the mesh, the larger the risk of excluding As mentioned, the volume and duration of the experi- other organisms. This is an aspect that any experimental ments with these enclosures are appropriate for protist method has to face and handle accordingly with the par- community dynamics (i.e. phytoplankton and small zoo- ticular characteristics of the community studied and the plankton) (Parsons, 1982). Large zooplankton could be experiment goals. Large zooplankton has to be considered a source of undesired variation between experimental a random factor in the enclosures described. Studying treatments. The small diameter of the enclosures and the multitrophic interactions, including tertiary trophic levels, frontal turbulence generated during the uprising filling ideally requires large volumes (>100 m3 ) and probably should partially exclude large zooplankton that shows longer experiments (Riebesell et al., 2010). avoidance reaction to turbulence (Franks, 2001). The degree of avoidance may be different for each species. In the ENEX experiment, three crustacean species were present in the lake plankton community: Cyclops abyssorum, Fouling and elemental mass balance Diaptomus cyaneus and Daphnia pulicaria. Comparing the One of the particular features of the enclosures described outside concentration with the average inside the enclo- is the high aspect ratio (wall area vs. water volume). This sures indicated a marked avoidance difference between feature introduces the risk of becoming an experiment on the three species. On average, the inside Diaptomus density benthic colonization if the duration is too long or some was <10% than outside, Daphnia about 60%, and Cyclops treatments stimulate rapid wall colonization. Adjusting showed no significant decrease. In fact, because of the the experiment to short periods minimizes the undesired low numbers, the variation between enclosures was high. effects of periphyton growth on the enclosure surfaces 274 P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE (Chen et al., 1997). Wall growth depends not only on the enclosure design and the duration of the experiment but also on the trophic state and size of the water body. Open waters hold fewer benthic organisms as potential inoculum for the enclosure walls than waters in the littoral zone. In the ENEX experiment, we did not observe any significant development of microeukaryotic biofilms on the surface of the bags after a microscope inspection at the end of the experiment, except for spots of zygnemato- phyceae filaments (i.e. Spirogyra) in the inner part of the treatments with the highest P-addition. These algae are common in the littoral epilithon of the lake. The growth and influence of bacterial biofilms are more difficult to be evaluated by direct inspection unless they are huge. Elemental mass budgets are a way to evaluate some biases deriving from the material characteristics and experimental method. In the ENEX experiment, we compared the total initial and final elemental amounts of the nutrients manipulated, P and N (Fig. 5). Although the average difference between initial and final elemental inventories was low, the variation among replicates was relatively large (mean ± standard deviation; P, 8% ± 24; N 7% ± 14). A significant bias in one of the controls (Fig. 5A) was in the positive direction; more P was estimated at the end than at the beginning, which is not possible. This paradox was due to the assumption that the initial TP in the bags was identical to the integrated measure in the lake water column at the filling time, which could be correct on average, but deviations in each Fig. 5. Balance between the initial and final total phosphorus (A) and enclosure existed, being extreme in the specific case of total nitrogen (B) in the enclosures. Nitrogen shows a random variation, while the phosphorus pattern indicates some bias at high P additions, that control. Because of the added P in other treatments, probably related to some adsorption to the plastic walls. Initial values the deviations from the assumption were only noticeable were estimated based on two 20-m integrated samples in the lake water in the controls. column at the beginning of the period. Therefore, the exact initial values in each enclosure could differ. Negative values indicate lower amounts In the P case, the budget deviations indicated a at the end of the period. The size of the symbols shows the initial level systematic bias at high initial TP (Fig. 5A), whereas N of the complementary element, TN in A and TP in B. deviations were random (Fig. 5B). The random error in these budgets was related to the assumption of a common compare total nutrient budgets between initial and final initial composition, which is not exact, and the many conditions. fractions analyzed, each with their analytical error. The N case indicates that deviations lower than ±20–30% could be considered potentially random, which was also Alternative designs the case for the low P additions. The high negative P The enclosure design described allows modifications budgets at elevated P additions (Fig. 5B) could be due according to the system to be studied and the purpose to the P adsorption to the polythene walls (Hassenteufel of the experiment. For instance, the length of the bags et al., 1963) and the previously mentioned fouling growth. can be shortened for experiments in shallow lakes or Whatever the reason, the experiment would have to be thinner photic layers. In these cases, transparent material shorter for the high P-addition treatments. However, (e.g. methacrylate) can be used for the upper tube to despite the anomalous P budget in the highest P additions, minimize light reduction. The diameter can also be no benthic organisms were found in the microscopy wider: 20–30 cm appears manageable for transport and counting of the plankton samples (Zufiaurre et al., 2021). deployment from small boats. Polythene bag rolls are still The concentrations of TDP at the end of the experiment commercially available with these diameters. The upper were still high (∼390 nM) in those high P addition tube should accommodate the new dimensions, but the enclosures, not limiting the growth. Therefore, the results sediment trap can be kept as currently designed, just were reliable. Any nutrient addition experiment should placing an appropriate funnel connecting the bag and 275 JOURNAL OF PLANKTON RESEARCH VOLUME 45 NUMBER 2 PAGES 266–277 2023 trap. Changing the bag material can also be helpful for Berg, G. M., Glibert, P. M. and Chen, C.-C. (1999) Dimension effects of experiments on light/dark expositions or investigating enclosures on ecological processes in pelagic systems. Limnol. Oceanogr., irradiance effects. It can also be considered to sample the 44, 1331–1340. internal gradient that develops within the long bags rather Camarero, L. and Catalan, J. (2012) Atmospheric phosphorus deposi- tion may cause lakes to revert from phosphorus limitation back to than integrate it. Another option could be modifying nitrogen limitation. Nat. Commun., 3, 1118. the bottom tub for benthic-pelagic coupling experiments Catalan, J. (1999) Small-scale hydrodynamics as a framework for plank- (Dimitriou et al., 2017). Application of the system in ton evolution. Jpn. J. Limnol., 60, 469–494. marine conditions is also feasible, probably, enhancing Catalan, J., Felip, M., Giménez-Grau, P., Zufiaurre, A., Camarero, the robustness of the construction and planning for short- L. and Pla-Rabés, S. (2020) Effects of episodic nutrients duration deployments during forecasted good weather. enrichments on P-limited planktonic communities—Lake Redon ENEX 2013 experiment. Dataset. Dryad Digital Repository. https://doi.org/10.5061/dryad.tx95x69vh. CONCLUSIONS Catalan, J., Pla, S., Rieradevall, M., Felip, M., Ventura, M., Buchaca, T., Camarero, L., Brancelj, A. et al. (2002) Lake redo ecosystem response The enclosure type described is an addition to the exper- to an increasing warming in the Pyrenees during the twentieth imental toolbox for plankton investigation. The enclo- century. J. Paleolimnol., 28, 129–145. sures are particularly suitable for studying the immediate Chen, C. C., Petersen, J. E. and Kemp, W. M. (1997) Spatial and tem- response of the planktonic community to pulse pertur- poral scaling of periphyton growth on walls of estuarine mesocosms. bations of nutrients, or other dissolved substances, using Mar. Ecol. Prog. Ser., 155, 1–15. before and after comparisons. The design offers the pos- Cutroneo, M. and Torrisi, A. (2014) Characterization of thin films for TNSA laser irradiation. J. Phys. Conf. Ser., 508, 012012. sibility of obtaining an integrated assessment across the entire or most of the photic zone, facilitating multiple Dimitriou, P. D., Papageorgiou, N., Geropoulos, A., Kalogeropoulou, V., Moraitis, M., Santi, I., Tsikopoulou, I., Pitta, P. et al. (2017) A novel replicates and treatments. Although the enclosures were mesocosm setup for benthic-pelagic coupling experiments. Limnol. initially thought for experiments in remote oligotrophic Oceanogr. Methods, 15, 349–362. systems, the design can be easily tailored and expanded Domis, L. N. D., Elser, J. J., Gsell, A. S., Huszar, V. L. M., Ibelings, B. W., for applications to a large variety of water masses, main- Jeppesen, E., Kosten, S., Mooij, W. M. et al. (2013) Plankton dynamics taining its essential differential features: low perturbation under different climatic conditions in space and time. Freshw. Biol., 58, of the planktonic organisms during filling and possibilities 463–482. for high replication. Downing, J. A., Osenberg, C. W. and Sarnelle, O. (1999) Meta-analysis of marine nutrient-enrichment experiments. Ecology, 80, 1157–1167. Estrada, M., Marrase, C. and Alcaraz, M. (1988) Phytoplankton SUPPLEMENTARY DATA response to intermittent stirring and nutrient addition in marine Supplementary data can be found at Journal of Plankton Research online. microcosms. Mar. Ecol. Prog. Ser., 48, 225–234. Faithfull, C. L., Bergström, A.-K. and Vredeb, T. (2011) Effects of nutri- ents and physical lake characteristics on bacterial and phytoplankton DATA AVAILABILITY production: a meta-analysis. Limnol. Oceanogr., 56, 1703–1713. The ENEX experiment data is available in the Dryad Digital Repository, Falkowski, P. G., Barber, R. T. and Smetacek, V. (1998) Biogeochemical doi: 10.5061/dryad.tx95x69vh (Catalan et al., 2020). controls and feedbacks on ocean primary production. Science, 281, 200–206. ACKNOWLEDGEMENTS Felip, M., Bartumeus, F., Halac, S. and Catalan, J. (1999) Microbial P.G.-G. (FPU AP2010-3596), C.P.-L. (FPU12-00644) and A.Z. plankton assemblages, composition and biomass, during two ice-free periods in a deep high mountain lake (Estany Redó, Pyrenees). J. (FPI BES-2014-070196) acknowledge their respective doctoral schol- Limnol., 58, 10. arships. Franks, P. J. S. (2001) Turbulence avoidance: an alternate explanation of turbulence-enhanced ingestion rates in the field. Limnol. Oceanogr., 46, 959–963. FUNDING Fraser, L. H. and Keddy, P. (1997) The role of experimental microcosms The Spanish Government (Ministerio de Ciencia e Innovación): in ecological research. Trends Ecol. Evol., 12, 478–481. NitroPir (CGL2010-19737) and Transfer (CGL2016-80124-C2-1-P) Giménez-Grau, P., Felip, M., Zufiaurre, A., Pla-Rabès, S., Camarero, projects. L. and Catalan, J. (2020) Homeostasis and non-linear shift in the stoichiometry of P-limited planktonic communities. Ecosphere, 11, e03249. https://doi.org/10.1002/ecs2.3249. REFERENCES Goldman, J. C. (1988) Spatial and temporal discontinuities of biological Benton, T. G., Solan, M., Travis, J. M. J. and Sait, S. M. (2007) Micro- processes in pelagic surface waters. In Rothschild, B. J. (ed.), Towards cosm experiments can inform global ecological problems. Trends Ecol. a Theory on Biological-Physical Interactions in the World Ocean, Kluwer Evol., 22, 516–521. Academic Publishers, Dordrecht, pp. 273–296. 276 P. GIMÉNEZ-GRAU ET AL. SELF-FILLING ENCLOSURES TO EXPERIMENTALLY ASSESS PLANKTON RESPONSE Gust, G. (1977) Turbulence and waves inside flexible-wall systems Parsons, T. R. (1982) The future of controlled ecosystem enclosure designed for biological studies. Mar. Biol., 42, 47–53. experiments. In Grice, G. D. and Reeve, M. R. (eds.), Marine Meso- Halac, S., Felip, M., Camarero, L., Sommaruga-Wograth, S., Psen- cosms: Biological and Chemical Research in Experimental Ecosystems, Springer ner, R., Catalan, J. and Sommaruga, R. (1997) An in situ US, New York, NY, pp. 411–418. enclosure experiment to test the solar UVB impact on plank- Pla, S. (2001) Chrysophycean Cysts from the Pyrenees, Cramer, Berlin. ton in a high-altitude mountain lake. I. Lack of effect on phy- Pla-Rabes, S. and Catalan, J. (2011) Deciphering chrysophyte responses toplankton species composition and growth. J. Plankton Res., 19, to climate seasonality. J. Paleolimnol., 46, 139–150. 1671–1686. Reynolds, C. S. (2006) The Ecology of Phytoplankton, Cambridge University Harris, G. P. (1980) Temporal and spatial scales in phytoplankton Press, Cambridge. ecology. Mechanisms, methods, models, and management. Can. J. Fish. Aquat. Sci., 37, 877–900. Richardson, D. C., Carey, C. C., Bruesewitz, D. A. and Weathers, K. C. (2017) Intra- and inter-annual variability in metabolism in an Hassenteufel, W., Jagitsch, R. and Koczy, F. F. (1963) Impregnation of oligotrophic lake. Aquat. Sci., 79: 319–333. glass surface against sorption of phosphate traces. Limnol. Oceanogr., 8, 152–156. Riebesell, U., Lee, K. and Nejstgaard, J. C. (2010) Pelagic mesocosms. In Riebesell, U., Fabry, V. J., Hansson, L. and Gattuso, J.-P. (eds.), Jacquemin, C., Bertrand, C., Oursel, B., Thorel, M., Franquet, E. and Guide to Best Practices for Ocean Acidification Research and Data Report- Cavalli, L. (2018) Growth rate of alpine phytoplankton assemblages ing, Publications Office of the European Union, Luxembourg, pp. from contrasting watersheds and N-deposition regimes exposed to 95–112. nitrogen and phosphorus enrichments. Freshw. Biol., 63, 1326–1339. Sanders, F. S. (1985) Use of large enclosures for perturbation experi- Kuiper, J., Brockmann, U. H., Groenewoud, H. V., Hoornsman, G. ments in lentic ecosystems: a review. Environ. Monit. Assess., 5, 55–99. and Hammer, K. D. (1983) Influences of bag dimensions on the development of enclosed plankton communities during POSER. Mar. Scharfenberger, U., Jeppesen, E., Beklioglu, M., Sondergaard, M., Ecol. Prog. Ser., 14, 9–17. Angeler, D. G., Cakiroglu, A. I., Drakare, S., Hejzlar, J. et al. (2019) Effects of trophic status, water level, and temperature on shallow lake Lack, T. J. and Lund, J. W. G. (1974) Observations and experi- metabolism and metabolic balance: a standardized pan-European ments on phytoplankton of Belham Tarn, English-Lake-District— mesocosm experiment. Limnol. Oceanogr., 64, 616–631. experimental tubes. Freshw. Biol., 4, 399–415. Scor-Working-Group-85 (1991) Manual on Marine Experimental Ecosystems Legendre, L., Rivkin, R. B. and Jiao, N. (2018) Advanced experimental Unesco Technical Papers in Marine Science, UNESCO, Paris, p. 178. approaches to marine water-column biogeochemical processes. ICES J. Mar. Sci., 75, 30–42. Sommaruga, R. (2001) The role of solar UV radiation in the ecology of alpine lakes. J. Photochem. Photobiol. B-Biol., 62, 35–42. Legendre, P. and Gallagher, E. D. (2001) Ecologically meaningful trans- formations for ordination of species data. Oecologia, 129, 271–280. Sommer, U. (1986) The periodicity of phytoplankton in Lake Constance (Bodensee) in comparison to other deep lakes of Central Europe. Marrase, C., Lim, E. L. and Caron, D. A. (1992) Seasonal and daily Hydrobiologia, 138, 1–7. changes in bacterivory in a coastal plankton community. Mar. Ecol. Prog. Ser., 82, 281–289. Spivak, A. C., Vanni, M. J. and Mette, E. M. (2011) Moving on up: can results from simple aquatic mesocosm experiments be applied across Mcandrew, P. M., Björkman, K. M., Church, M. J., Morris, P. J., broad spatial scales? Freshw. Biol., 56, 279–291. Jachowski, N., Williams, P. J. L. B. and Karl, D. M. (2007) Metabolic response of oligotrophic plankton communities to deep water nutri- Stewart, R. I. A., Dossena, M., Bohan, D. A., Jeppesen, E., Kordas, ent enrichment. Mar. Ecol. Prog. Ser., 332, 63–75. R. L., Ledger, M. E., Meerhoff, M., Moss, B. et al. (2013) Mesocosm experiments as a tool for ecological climate-change research. In Woodward, Menzel, D. W. and Case, J. (1977) Concept and design: controlled G. and O’gorman, E. J. (eds.), Advances in Ecological Research, Vol. 48, ecosystem pollution experiment. Bull. Mar. Sci., 27, 1–7. Academic Press, Cambridge, pp. 71–181. Morabito, G., Mazzocchi, M. G., Salmaso, N., Zingone, A., Bergami, Ventura, M., Camarero, L., Buchaca, T., Bartumeus, F., Livingstone, D. C., Flaim, G., Accoroni, S., Basset, A. et al. (2018) Plankton dynamics M. and Catalan, J. (2000) The main features of seasonal variability across the freshwater, transitional and marine research sites of the in the external forcing and dynamics of a deep mountain lake (redo, LTER-Italy network. Patterns, fluctuations, drivers. Sci. Total Environ., Pyrenees). J. Limnol., 59, 97–108. 627, 373–387. Ventura, M. and Catalan, J. (2008) Incorporating life histories and diet Nakano, S., Hayakawa, K., Frenette, J. J., Nakajima, T., Jiao, C. M., quality in stable isotope interpretations of crustacean zooplankton. Tsujimura, S. and Kumagai, M. (2001) Cyanobacterial blooms in a Freshw. Biol., 53, 1453–1469. shallow lake: a large-scale enclosure assay to test the importance of diurnal stratification. Arch. Hydrobiol., 150, 491–509. Zufiaurre, A., Felip, M., Giménez-Grau, P., Pla-Rabès, S., Camarero, L. and Catalan, J. (2021) Episodic nutrient enrichments stabilise Osborn, T. (1996) The role of turbulent diffusion for copepods with protist coexistence in planktonic oligotrophic conditions. J. Ecol., 109, feeding currents. J. Plankton Res., 18, 185–195. 1717–1729. 277