Mission possible: diatoms can be used to infer past duckweed (lemnoid Araceae) dominance in ponds

Compared to larger lakes, ponds have rarely been the focus of palaeoecological studies. A common feature of ponds, especially those subject to eutrophication, is mass surface coverings of lemnoid Araceae (duckweed) which have severe implications for ecological processes in small waterbodies, in particular lowered oxygen content. To help understand the implications of duckweed dominance for the long-term ecology of ponds, and to determine the potential for palaeoecological studies in ponds more generally, we develop a new diatom-based Lemna-indicator metric. Recent studies of diatom host-plant relationships have shown significant associations between duckweed and the epiphytes Lemnicola hungarica and Sellaphora saugerresii (formally known as Sellaphora seminulum). To determine the potential of these species as palaeo-indicators of long-term duckweed dynamics in ponds, we investigated the diatom composition of surface sediment assemblages in sets of duckweed and non-duckweed-dominated ponds in Norfolk, eastern England. In addition, we undertook diatom analysis of two cores from a small farmland pond (Bodham Rail Pit) subject to a known duckweed dominance event (1999–2005). Both L. hungarica and S. saugerresii were significant predictors of past Lemna dominance in the surface sediments. Further, in the core study, both diatom species accurately and closely tracked the documented “on–off” duckweed cycle. Our study suggests huge potential for using ponds in palaeoecological studies and for diatom-based investigations of floating plant histories.


Introduction
To date, both ecological and palaeoecological research has focused almost entirely on large, deep and shallow lakes, with minimal attention devoted to the 'poor cousins' of shallow lakes, namely ponds. The definition of a pond is currently fluid (Biggs et al. 2005), but in this study, we use it to refer to waterbodies with a maximum diameter of less than 100 m, where insufficient wind-mixing permits chemical stratification of the water-column (Sayer et al. 2013). In terms of palaeoecology, the lack of studies in ponds is particularly acute (Edwards et al. 2011;Emson 2015). While a number of palaeo-studies have been undertaken in arctic ponds (Douglas and Smol 1994;Keatley et al. 2011;Bouchard et al. 2012;Stewart et al. 2013), few if any studies have focused on temperate European lowland ponds-as in this study. Nonetheless, in recent years, a number of studies have shown the predominantly agricultural pond resource of lowland Europe to be of major importance in terms of aquatic biodiversity conservation (Ruggiero et al. 2008;Sayer et al. 2012), aquatic-terrestrial energy subsidies (Davies et al. 2016) and carbon sequestration (Downing et al. 2008), among other roles. Given increasing interest in pond conservation, palaeoecology could have an important role in inferring longterm changes in pond ecology and function in response to environmental impacts.
A key impact, particularly in ponds subject to eutrophication and/or enhanced shading (terrestrialisation), is the development of dense mats of lemnoid Araceae (formally Lemnaceae, and hereafter referred to as duckweed) which typically results in severe water-column anoxia (Lewis and Bender 1961;Pokornŷ and Rejmánková 1983;Janes 1998) in turn causing fish kills (Lewis and Bender 1961), diurnal temperature stratification (Dale and Gillespie 1976), reductions in pH (McLay 1976) and loss of submerged macrophytes and macroinvertebrates (Janse and Van Puijenbroek 1998). The invasion of dense mats of freefloating plants is acknowledged as an important threat to the functioning and biodiversity of freshwater systems (Scheffer et al. 2003), with this especially true of ponds (Sayer et al. 2011). In this study, we test the potential for using sedimentary diatoms to indirectly infer the past presence of duckweed dominance in ponds.
For many diatom species, preferences for particular plant structures are generally poorly understood (Bennion et al. 2010). Indeed, despite several studies, potential species-habitat associations between epiphytic diatoms and different macrophyte species have remained relatively elusive, and it is probable that species-specific plant-diatom relationships are rare (Eminson and Moss 1980). There are some studies which do show strong host-plant links, however. For example Reavie and Smol (1997) revealed a close association between Cocconeis pediculus Ehrenb., and the filamentous algae Cladophora sp. in the St. Lawrence River, Canada. Additionally, Comte and Cazaubon (2002) showed a significant degree of homogeneity of the diatom communities associated with Chara vulgaris L., Groenlandia densa L. and Cladophora glomerata Kütz., in a large Mediterranean river, likely linked to substantial variation in plant surface micro-habitat.
Perhaps the best known example of diatom-habitat specificity is between Lemnicola hungarica (Grunow) Round and P.W. Basson (Synonym: Achnanthes hungarica (Grunow) Grunow in Cleve and Grunow) and free-floating lemnoid (duckweed) plants (Goldsborough 1993;Round and Basson 1997). Grunow first described A. hungarica in 1863 (hereafter referred to as L. hungarica), and it was noted that this species was found in lakes where duckweed was present. Hustedt (1930) commented that L. hungarica had a predisposition towards Lemna spp., and Round (1973) confirmed the observed association between Lemna spp. and L. hungarica. This habitat association was also used by Round and Basson (1997) to justify the establishment of Lemnicola as a new genus distinct from Achnanthes sensu lato. Associations between L. hungarica and free-floating plants were also observed by Bowker and Denny (1980), Goldsborough and Robinson (1985), and Goldsborough (1993Goldsborough ( , 1994 from field studies of diatom epiphytes sampled from Lemna minor L. Further, Zuberer (1984) identified L. hungarica in Scanning Electron Microscopy (SEM) studies of L. minor, as did Buczkó (2007) from herbarium specimens of various free-floating plants including L. minor, and Desianti (2012) from field collections of diatom epiphytes on free-floating plants.
In a host-diatom study of 131 macrophyte samples from 63 sites, covering North and South America, northern Europe, southern Africa, Australia and Asia (Emson 2015), L. hungarica showed a significant association with free-floating plants, as did Sellaphora seminulum (recently named Sellaphora saugerresii (Desm.) C.E. Wetzel et D.G. Mann comb. nov., and hereafter referred to as S. saugerresii- Wetzel et al. 2015). Thus, several studies suggest a strong and robust association between L. hungarica and duckweed, and potentially S. saugerresii also.
Given the poor preservation of Lemna spp. fronds in sediments, and because species in the lemnoid Araceae rarely produce flowers, pollen or seeds (Hillman 1961;Landolt 1986), it is likely infeasible to reconstruct past duckweed dominance directly. Hence, this study will investigate the potential for using diatoms, specifically L. hungarica and S. saugerresii, as indirect indicators of duckweed in ponds. Firstly, we conducted a study of surface sediment diatom assemblages from Lemna-covered ponds and from ponds with no Lemna surface coverage to establish contemporary diatom-duckweed relationships and thereby assess the potential for inferring changes in free-floating macrophyte cover, in small waterbodies. Secondly, we tested the ability of diatoms to track a documented duckweed ''on-off'' cycle in a small pond (Bodham Rail Pit) through diatom analysis of a sediment core.

Study sites
The surface sediment study focuses on a set of small ponds in Norfolk, eastern England. For the most part, the ponds are situated in arable agricultural land and have their origins as 'marl pits' being formed by marl extraction largely in the eighteenth-nineteenth centuries (Prince 1964). Other ponds have their origins as pingos formed at the end of the Last Glaciation 12.7-11.5 cal kyr BP (Flemel 1976). The ponds were all small (\ 40 m max. diameter), shallow (generally \ 1 m) and, in the case of the farmland ponds, largely alkaline and eutrophic (Table 1). Further details on the history, geography and ecology of ponds in the study area are given in Sayer et al. (2012).
Some 12 Lemna-dominated ponds were selected for study based on local knowledge of duckweed occurrence. In these ponds L. minor and/or Lemna minuta Kunth., was known to have been prevalent over at least 2-3 years at between 25 and 100% of water-surface cover. At the time of sampling (2012) 1, 3 and 8 sites were dominated by L. minor, L. minuta and codominated by L. minor and L. minuta respectively. For comparison 14 non-Lemna ponds were selected at random from a comparable study area. In all cases, it was known that these sites had not experienced duckweed domination in the recent (about the last 5 years) past.
The focus of the palaeoecological investigation was the Bodham Rail Pit (52°54 0 20. 62 00 N; 1°09 0 21. 23 00 E), hereafter referred to as Rail Pit, a typical small (350 m 2 ), shallow (mean depth 103 cm) pond located in Bodham, North Norfolk (Fig. 1). The Rail Pit is located in arable farmland and surrounded by a 10 m rough grassland buffer zone. The pond is primarily ground-water fed but also receives surface runoff via a small road. Water levels in the pond fluctuate seasonally by about 1 m but it never completely dries out. The age of the pond is unknown, but it is shown on a tithe map dating to 1841, suggesting that it is at least 150-200 years old. Importantly, the pond underwent a major duckweed dominance event over 1999-2005, before and after which it is known to have been largely duckweed free. Over the period 1994-1998, prior to high duckweed coverage, the Rail Pit had abundant stands of Potamogeton natans L., but more recently (2008-present) it has been dominated by Ceratophyllum submersum L (Fig. 2).

Historical plant records: Rail Pit
A recent history of aquatic macrophyte abundance for the Rail Pit is available from 1994 to 2016 (Sayer and Emson, unpublished data). In each year, plants were recorded on the DAFOR scale (Dominant-5, Abundant-4, Frequent-3, Occasional-2, Rare-1) based on visual observations assisted by plant collections made with a double-headed rake.
A 118-cm sediment core (RAIL1) was collected from the Rail Pit using a 'Big Ben' wide-bore piston corer (Patmore et al. 2014) from a southern central location at a depth of 108 cm (Fig. 1). The core was sliced on site at 1-cm intervals. On extrusion the core length was reduced to 75 cm, likely due to compaction of the very soft upper sediments which were very flocculent. Given the fluid upper section of RAIL1, and as the key aim was to track recent duckweed coverage, a short (22 cm) Glew core was also collected from an adjacent location. This core (RAIL2) was sliced at 0.5-cm intervals throughout.
All sediment samples were placed in sealed plastic bags and subsequently stored at 4°C prior to analysis.

Radiometric analysis
Dried sediment samples from cores RAIL1 and RAIL2 were analysed for 210 Pb, 226 Ra, 137 Cs and 241 Am by J Paleolimnol (2018) 60:209-221 211 direct gamma assay at the Bloomsbury Environmental Isotope Facility (BEIF), University College London, using an ORTEC HPGe GWL series well-type coaxial low background intrinsic germanium detector. 210 Pb was determined via its gamma emissions at 46.5 keV, and 226 Ra using the 295 keV and 352 keV gamma rays emitted by its daughter isotope 214 Pb following three weeks storage in sealed containers to allow radioactive equilibrium. 137 Cs and 241 Am were measured by their emissions at 662 keV and 59.5 keV respectively (Appleby et al. 1986). The absolute efficiencies of the detector were determined using calibrated sources and sediment samples of known activity. Corrections were made for the effect of self-absorption of low energy gamma rays within each sample (Appleby et al. 1992).
Chronologies and sediment accumulation rates for both of cores RAIL1 and RAIL2 were calculated using the constant rate of 210 Pb supply (CRS) model (Appleby and Oldfield 1978). Use of the constant initial 210 Pb concentration (CIC) model was precluded because of irregular declines in unsupported 210 Pb activities resulting in non-monotonic features in the 210 Pb profiles of both cores (Fig. 3).

Diatom analysis
A total of 48 sediment samples were analysed for diatoms from core RAIL1. The uppermost 10-20 cm of the core was analysed at contiguous 1-cm intervals and below this samples were analysed at 2-cm intervals. A total of 44 contiguous samples were analysed for diatoms in core RAIL2 at 0.5-cm intervals covering the full length of the 22-cm core. Samples were prepared for diatom analysis using standard methods (Battarbee 1986;Battarbee et al. 2001). Given that eutrophication can potentially swamp rare species by abundance increases of mesoeutraphentic and hyper-eutraphentic species (Van Dam and Mertens 1993), and as the aim was to determine how well the diatom stratigraphy tracked past duckweed dominance using two indicator species, large diatom counts (range 500-4700 valves per sample) were undertaken to reduce the dominance of common diatom species, whilst allowing responses of rarer taxa to be more accurately captured. All samples were mounted on microscope slides using Naphrax TM and absolute numbers of diatoms present in 0.1 g of sediment were counted at 10009 magnification with oil immersion under phase-contrast illumination using a Leitz 'Laborlux S' light microscope. Diatoms were identified to species level and diatom data for both cores were presented as percentage relative abundances. Images of the two indicator species used in the study are given in Fig. 4.

Data manipulation and analysis
Binomial logistic regression was performed using SPSS version 22 (2013) to determine whether L. hungarica and S. saugerresii could faithfully predict Lemna dominance in the ponds. Pearson's correlation analysis was applied to L. hungarica and S. saugerresii relative abundances to determine the strength of their bivariate relationship in both of cores RAIL1 and RAIL2. Diatom core stratigraphic data were plotted using C2 software (Juggins 2007).

Radiometric dating of cores RAIL1 and RAIL2
The simple CRS model places the 1963 maximum nuclear weapons fallout layer ( 137 Cs peak) at 23.5 cm, slightly above the 137 Cs peak (26.5 cm) in core RAIL1. There is a sudden and dramatic increase in sedimentation rate at 34.5 cm (1940s) to 0.47 g cm -2 year -1 which may be due to a sediment slumping event (Fig. 3a). This is followed by a relatively uniform period covering the 1950s to the 1970s of 0.04-0.07 g cm -2 year -1 and an increased relatively uniform sedimentation rate of 0.08-0.1 g cm -2 year -1 for the last 30 years. For core RAIL2 the CRS dating model places 1963 at 12.5 cm, just above that suggested by the 137 Cs and 241 Am records at 15.5 cm. This may be due to noncontinuous sedimentation before 1963. The CRS calculated chronology indicates that sediment accumulation rates gradually increased from 0.02 g cm -2 year -1 in the 1940s to 0.06 g cm -2year -1 since the 1980s (Fig. 3b).
A further Lemna-dominance phase was evident in core RAIL1 with L. hungarica and S. saugerresii peaks over 48-40 cm (Phase 2 in Fig. 6a). Finally, a potential earlier Lemna phase was evidenced by increases in S. saugerresii (2-4%) and to a lesser extent L. hungarica (\ 1%) at around 68 cm (Phase 1 in Fig. 6a). Despite some scatter and a lesser degree of agreement, especially in the lower sections of RAIL1, a positive linear relationship was found between percentages of L. hungarica and S. saugerresii in both of cores RAIL1 (r = 0.57, p \ 0.01, Fig. 7a) and RAIL2 (r = 0.72, p \ 0.01, Fig. 7b).

Discussion
The first formal documented record of a diatom was in 1703 (thought to be Tabellaria flocculosa) sampled from the roots of Lemna (Round et al. 1990), but on Christmas day in 1702 Van Leeuwenhoek looked at the roots of Lemna and also probably saw diatoms, although he described his findings as 'animalcula' in the Philosophical Transactions of the Royal Society (Van Leeuwenhoek 1703). Since these early days, many workers have investigated potential hostmacrophyte relationships for diatoms (Prowse  Siver 1977;Eminson and Moss 1980;Millie and Lowe 1983). Round (1973), Bowker and Denny (1980), Goldsborough and Robinson (1985) and Goldsborough (1993Goldsborough ( , 1994 were amongst the first researchers to acknowledge the association between L. minor and L. hungarica. Following these pioneering studies other workers have investigated the consistency and robustness of diatom-duckweed relationships. Buczkó (2007) found L. hungarica dominated diatom assemblages attached to L. minor, Lemna gibba L., Spirodela polyrhiza L. Schleid., and Wolffia arhiza L. Horkel ex Wimm., in marked contrast to assemblages found on L. trisulca which were dominated by Cocconeis placentula Ehrenb. Buczkó (2007) concluded that L. hungarica was tightly attached to well definable taxa of the lemnoid Araceae. Desianti (2012) also reported L. hungarica for L. minor, S. polyrhiza and Wolffia spp. Clearly, several studies have concluded that L. hungarica has a strong habitat preference for duckweed, while Emson (2015) showed a similar association for S. saugerresii (Synonym: Sellaphora seminulum (Grunow) D.G. Mann). This good understanding of diatom ecology begs the question: is it possible to employ these diatoms as palaeo-indicators of past Lemna dominance? To this end, our study clearly shows both L. hungarica and S. saugerresii to be successfully transferred from duckweed mats to the sediment strongly suggesting a positive answer to this question. Further, this study shows that, even at low percentages (\ 5%), these diatoms can indicate high duckweed coverage (Fig. 6). Importantly, this indicator status for L. hungarica and S. saugerresii applies not only to the lemnoid Araceae (with the exception of L. trisulca) but can also be applied to other freefloating plants such as Azolla spp. and Riccia spp. (Emson 2015).
Confirmation of a strong Lemna-diatom association comes from comparing the RAIL1 and RAIL2 diatom stratigraphies with historical observations in the Rail Pit. In both cores the documented arrival and termination of high Lemna coverage over 1999-2005 is clearly tracked by peaks of both L. hungarica and S. saugerresii (Phase 4 in Fig. 6). Thus, even despite highly flocculated upper sediments for RAIL1 after collection, the Lemna signal remained intact. Core RAIL1, which extended to the beginning of sedimentation in the Rail Pit, showed some degree of bloom versus low-abundance (boom-bust) cyclicity for both L. hungarica and S. saugerresii, with three further potential Lemna-covered phases (Fig. 6a). For the penultimate peak (Phase 3 in Fig. 6) reasonable temporal accordance was evident between cores RAIL1 (32-27 cm = 1950(32-27 cm = -1960(32-27 cm = ) and RAIL2 (19-17 cm = 1948(32-27 cm = -1958 giving faith in the reconstruction. The two earlier inferred duckweed peaks in RAIL1 (Phase 2-48-40 cm; Phase 1-68-60 cm) were less well defined for S. saugerresii compared to L. hungarica, however, potentially suggesting that L. hungarica is the most reliable Lemna indicator. On this basis low percentages of L. hungarica for Phase 1 make an inference of duckweed-dominance more questionable. Indeed, despite a significant positive correlation between L. hungarica and S. saugerresii in both cores, differences in their responses in RAIL1 suggest that these species may have slightly different ecological preferences and further studies in this respect could be fruitful. Given this, the merit of Fig. 7 Relationships between Lemnicola hungarica and Sellaphora saugerresii percentage relative abundances for cores RAIL1 (a) and RAIL2 (b) combining L. hungarica and S. saugerresii to provide a Lemna-indicator metric (Fig. 6) is open to debate.
The ability to infer periods of free-floating plant dominance through time is of particular relevance to small waterbodies where duckweed proliferation, often symptomatic of high-nutrient loading, results in extensive surface coverings of duckweed (Portielje and Roijackers 1995). These dense mats physically prevent oxygen from being dissolved from the atmosphere and increase carbon dioxide levels (Janes 1998) causing a reduction in pH (McLay 1976;Janes 1998) and increased conductivity (Greaves et al. unpublished data). In addition, lemnid dominance can attenuate photosynthetic active radiation (PAR) by up to 99% with associated temperature fluctuations which can lead to diurnal temperature stratification (Dale and Gillespie 1976;Goldsborough 1993). Dense duckweed mats can cause significant species losses in submerged macrophyte and invertebrate communities (Janse and Van Puijenbroek 1998), and associated cold, anoxic conditions can cause fish kills (Lewis and Bender 1961). Thus, a diatom-Lemna inference tool has much utility in terms of reconstructing the past ecology of ponds, with the same potentially true of shallow lakes and indeed larger lakes under certain circumstances across the globe (Duarte and Kalff 1990) where high Lemna-coverage occurs. For example, dominance of free-floating mats of lemnids have also occurred in Lake Maracaibo, Venezuela and Lake Titicaca, Peru due to increased nutrient inputs (Cruz et al. 2006), and seasonal blooms of duckweeds have also been observed in cut-off meanders in large rivers (Beauger et al. 2015). Significant relationships between L. hungarica and duckweed have been demonstrated for a variety of climatic conditions and waterbody types in many parts of the world including Canada (Goldsborough 1993(Goldsborough , 1994, USA (Zuberer 1984;Desianti 2012), Brazil (Garcia and Fonseca de Souza 2006) and substantial tracts of Europe (Buczkó 2007;Emson 2015). Consequently, with the caveat that more work on habitat preferences is needed for S. saugerresii, compared to L. hungarica, we strongly suggest that both diatoms might be reliably used to infer past free-floating plant dominance much more widely. Will it be ''mission possible'' in other waterbodies in the world? Certainly, we urge other workers to go out and test this idea.

Conclusions
This study suggests that ponds in lowland European agricultural landscapes have considerable palaeolimnological potential. Indeed, the close agreement of diatom-Lemna signals in two cores from the Rail Pit strongly suggest that palaeo-techniques can be confidently applied to these often over-looked waterbodies, at least where ponds are permanent.
The study also shows the huge advantage of incorporating 'rare' species in diatom analysis and of undertaking large ([ 500 valves per slide) counts to better characterize diatom signals in cores. In this case the two Lemna-indicator diatoms occurred at relatively low percentages (\ 5%) in both stratigraphic sequences and in the surface sediments, even where duckweed was entirely dominant. This is particularly pertinent given that conventional diatom studies, based on counts of 300 or at best 500 diatom valves, tend to delete species at \ 5% relative abundance from summary plots. Thus, as a 'rule of thumb', the duckweed-covered signal afforded by L. hungarica and S. saugerresii would typically be missed.
Finally, in developing two diatom-indicators of free-floating plants, our research highlights the huge value of understanding diatom species ecologies for interpreting palaeo-data. In this respect, similar to previous studies (Bennion et al. 2010;Sayer et al. 2010), we suggest that a greater understanding of diatom responses to habitat and chemical variation across and between different littoral zone structures in lakes and ponds would greatly improve the interpretation of sediment records dominated by non-planktonic diatoms.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.