Rarity of microalgae in macro, meso, and microhabitats

ABSTRACT

Climate change and human-induced habitat degradations result in loss of species diversity in natural ecosystems. While the extinction of macroscopic organisms has been well documented in the scientific literature and public media, we have only limited knowledge on the loss of microscopic elements of the ecosystems. Because rarity coincides with the increased risk of extinction, we investigated the commonness and rarity of microalgae in the Pannonian ecoregion. We reviewed the published literature of microalgal research in Hungary over the last 140 years and created a species-by-site matrix containing 2489 algae species and 1145 localities. Analysing this dataset, we found that although the core-satellite hypothesis suggests a bimodal site occupancy distribution, microalgae displayed a unimodal pattern with a high number of rarely occurring species. We also demonstrated that the well-known negative relationship between body size of organisms and number of occupied habitats holds for microalgae. Rarity values of taxa have a phylogenetic signal indicating that in terms of rarity, closely related species (desmids, dinoflagellates, euglenophytes) show considerable similarities. The various habitat types showed differences in the number of rare taxa. Small- and medium-sized habitats (bog lakes, streams, oxbows) hosted the majority of rare species. These results highlight the conservation importance of small habitats in preserving microbial diversity.

KEYWORDS:

• habitat specificity
• habitat types
• microalgae
• phytoplankton
• rarity
• rarity metrics

Introduction

Loss of biodiversity is one of the greatest ecological problems of the 21st century. The current biodiversity crisis is primarily driven by changes in land use, direct exploitation of organisms, climate change, pollution, and appearance of invasive alien species (IPBES 2019), which accelerate species extinctions. Whereas biodiversity investigations have focused primarily on terrestrial systems, closer attention should be given to other extremely diverse and endangered natural systems, such as freshwaters (Lydeard and Mayden 1995, Vári et al. 2022), which have far greater biodiversity loss than terrestrial systems (Reid et al. 2019, Sala et al. 2000). Freshwaters cover only 1% of the Earth’s surface, but the nearly 125 000 species of freshwater animals described to date represent 9.5% of all known animal species on the planet (Balian et al. 2008, Strayer and Dudgeon 2010); thus freshwaters are considered one of the most endangered ecosystems in the world. Freshwater conservation research and actions have focused on more charismatic species groups visible to the naked eye, with particular emphasis on vertebrates, such as amphibians and fish (Grenyer et al. 2006, Wake and Vredenburg 2008, Abell et al. 2011, Jenkins et al. 2015), but relatively few studies have investigated freshwater invertebrate or microbial diversity (Stomp et al. 2011, Collen et al. 2014).

As primary producers, algae make up the foundation of the aquatic food web. They play a significant role in oxygen production and in natural purification processes of aquatic ecosystems and should therefore be more important research targets in conservation planning than top predators (Abell 2002). Currently, however, few data are available for the distribution of freshwater algae and even fewer for rare species of this group (Németh 2005, Padisák et al. 2016, T-Krasznai and B-Béres 2021), partly because of the applied sampling methods. Phytoplankton investigation uses spatial snapshot samplings in which the sample coverage of conventional examinations (1–10 cm³) is several orders of magnitude smaller than the volume of the studied waterbody (Görgényi et al. 2016). Because of this low sample coverage, rare species are difficult, and often impossible, to catch, potentially resulting in large omission errors (Rondinini et al. 2006). Another explanation is that most basic functions of phytoplankton (production of organic matter and oxygen) or annual successional changes can be safely described at the level of dominant algal groups. Therefore, many studies consider only those species that contribute a minimum of 5% (sometimes 10%) to the total biomass (Padisák et al. 2010). However, biodiversity depends not only on richness, but also on dominance and rarity (Hillebrand et al. 2018). Distinguishing between rare and dominant phytoplankton species is difficult because species proportions are not exactly quantified (Padisák et al. 2010). Rarity must be taken into account to understand aquatic ecological processes.

Commonness and rarity of species in communities are recurrent topics in ecology (Rochelle and Martins 2009). The core-satellite framework is one way to examine this distribution pattern. Based on this theory, the community is made up of core (common) species found on nearly every suitable site, and satellite (rare) species are found on a few sites within the same habitat types. Moreover, some intermediate taxa are also found in the community. Hanski’s dynamic model predicts a bimodal distributional pattern of species, which he termed the core-satellite hypothesis (Hanski 1982), where core (common) and satellite (rare) species are numerous in a community.

Another approach used to investigate the commonness and rarity pattern of species is the “seven forms of rarity” framework, in which Rabinowitz (1981) established rarity categories based on different combinations of 3 attributes: (1) geographic range (small vs. large), (2) local population size (small vs. large), and (3) habitat specificity (generalist vs. specialist). These variables allow species to be classified into 8 categories that reflect different types of rarity and commonness. The rarity of species depends on several factors, such as their tolerances and preferences to environmental constraints, the type and quality of available habitats, and dispersal capabilities, for example. Body size is a key trait of species, which is often positively associated with habitat size (Brown 1984) and correlates with rarity and susceptibility to extinction (Terborgh 1974, Willis 1979, Kattan 1992). In general, the minimum habitat size requirement of species increases with body size (Brown 1984). However, this relationship is nonlinear because although the habitat size of small-bodied species ranges from smallest to largest, large-bodied species prefer exclusively large habitats, explaining why habitat fragmentation especially threatens the persistence of large species (Gaston 1994). These findings have been demonstrated for larger (10⁻²–10¹ m) organisms (Nee et al. 1991) but might not apply to smaller-bodied (10⁻⁶–10⁻² m) organisms.

In our study, we aimed to explore the rare and common elements of the microalgal flora in Hungary (Pannonian ecoregion), focusing on (1) how various water types share rare species; (2) whether any taxonomic groups are threatened with extinction, and (3) whether size of microalgae impacts their distribution.

We hypothesised that:

(H1) the various water types show differences in terms of the number of rare taxa, and therefore the conservation value of the habitats is different;

(H2) in accordance with the core-satellite hypothesis, microalgae concerning the number of occupied habitats display bimodal habitat preference in the Pannonian ecoregion;

(H3) rarity of microalgae depends on their phylogenetic relatedness;

(H4) as has been observed for macroscopic taxa, a negative relationship exists between the body size of algae and the number of occupied habitats.

Compilation of a species inventory of local aquatic habitats and their microflora is the first step in understanding regional processes that shape community assembly and also in planning and implementing conservation practices. In this study, we present a species inventory of microalgae collected in the Pannonian ecoregion and provide an overview of the species and habitat characteristics that determine how species perform under different habitat conditions.

Materials and methods

During the compilation of a regional species inventory, we focused on the Pannonian ecoregion (Illies 1978, EEA Coppenhagen 2004), which includes the lowland part of the Carpathian Basin. Climatically, the region belongs to the Eurasian Steppe zone with 8–13 °C annual mean temperature and 500–800 mm annual mean precipitation (Mezősi 2017).

Data sources

We collected data from various algal handbooks that included species descriptions in addition to species occurrences for the region (Felföldy 1972, 1981, 1985, Németh 1997, Schmidt and Fehér 1998, 1999, 2001, Grigorszky et al. 1999). We supplemented this database with data obtained from recently published literature on floristic studies (Hajdu and Zupkó 1982, Padisák and Hegewald 1992, Uherkovich et al. 1995, Komáromi and Padisák 1999, Padisák 1999) and data derived from grey literature such as unpublished Ph.D. dissertations or governmental reports. We focused primarily on phytoplankton species and benthic soft algae, but benthic diatoms were not included in this database. Finally, we expanded the database by including phytoplankton monitoring data of the General Directorate of Water Management and biovolume abundance data.

Habitat types

To classify occurring habitat types, we considered the following criteria: water categories (surface and subsurface waters, water currents, and standing waters), chemical composition (organic, high alkalinity, moderate alkalinity), origin (artificial or natural), water balance (stationary or temporary), size (>10 km²), and depth (deep: mean depth >4 m) (Supplemental Material S1).

Statistical analyses

We used quantile regression to examine the relationships between the biovolume of algae species and the number of occupied localities (and water types). We applied this approach because we predicted meaningful differences in slope values along the upper and lower boundaries of the conditional distribution values (here the number of occupied localities and waterbody types). The analyses were conducted for the 50%, 75%, 90%, and 95% quantiles. Algae with very small (<20 µm³) and very large (>500 000 µm³) biovolumes were excluded from the analysis because data in these size ranges are insufficient to estimate the parameters. Quantile regressions were done in R using the quantreg package (Koenker 2017).

To characterise the rarity of species, we aimed to create a frequency term and a rarity metric, which represent both habitat specificity and population abundance. Having a presence/absence species-by-site matrix, habitat specificity can be numerically characterised by the number of occupied water types, but because no information on local abundance was available, only the number of occupied localities within the types were calculated as a proxy of the abundance. This approachcan be feasible if a positive relationship exists between these variables. According to Brown (1984), species that occasionally attain a large abundance in a given habitat have a higher probability of dispersion and occur in more localities within this habitat. While Rabinowitz et al. (1986) found these variables to be independent, results from other studies (Kattan 1992) suggest a positive relationship between population size and distribution. To test this relationship, we used phytoplankton monitoring data with available, accurate abundance values and numbers of occupied localities within a habitat type. We analysed 8835 phytoplankton samples (containing 880 species) to demonstrate this positive relationship (Fig. 1).

Figure 1. Relationship between the maximum relative abundance of 880 species and number of samples in which they occurred. The analysis is based on Hungarian lake and river phytoplankton monitoring data for 2000–2020.

To give a frequency term and rarity metric for each species, we first calculated the proportion of occupied localities within each water type (maximum of this ratio may differ among water types). Because complete colonization of isolated localities cannot be achieved, the ratio is <1, even for the most common species. These proportions were divided by the maximal ratio observed in the given habitat type, and the occurrence values were then summed and divided by the number of types (24). The given values fall between 0 and 1 and can be considered a standardised frequency term (equation 1). To have a rarity metric, these values were subtracted from 1 (equation 2): (1) $F_i=\sum _{j=1}^N{\displaystyle \frac{N_{OLij}/N_{Lj}}{max{N}_{OLj}}/N,}$(1) where F_i is standardised frequency terms of ith species, N_OLij is the number of occupied localities by ith species in the jth water type, N_Lj is the number of localities in the jth water type, maxN_OLj is the number of occupied localities by species that occupied the most localities in the in the jth water type, and N is the number of water types (here 24): (2) $R_i=1-\sum _{j=1}^N{\displaystyle \frac{N_{OLij}/N_{Lj}}{max{N}_{OLj}}/N,}$(2) where R_i is rarity metric of ith species.

Comparisons of rarity metrics (equation 2) among the 24 habitat types were analysed by using Kruskal-Wallis test followed by the Wilcoxon signed-rank test for multiple comparisons (at p < 2.2 × 10⁻¹⁶ significance level). The analyses were completed using R statistical programming.

To identify the satellite and core microalgal species (Hanski 1982) at habitat and at locality levels, we investigated the shape of species occupancy frequency distribution (OFD) and ranked species occupancy curves (RSOC). Both approaches are suited to assess the assemblage patterns of large numbers of sites at a regional scale (Hui 2012).

The Rabinowitz model (1981) at the global scale presents 8 possible outcomes with 7 types of rarity and 1 form of total commonness. Because we worked in 1 geographic region, we had only 4 outcomes, but instead of using 4 discrete quadrants (3 rarities and 1 commonness), we evaluated habitat specificity and population size on a continuous scale. Because no information is available on the abundance of occurring species in most floristic studies, we calculated the standardised frequency values (equation 1) for each species (as a proxy of abundance) and plotted the values against the occupied number of habitat types by the species (as a proxy of habitat specificity). To indicate the potential range of frequency values, we calculated their theoretical minima and maxima for each number site. Minimum values were calculated considering that only one site is occupied in the types (N_OLij = 1 in equation 1); maximum values were calculated considering that each site has been occupied in the given type (N_OLij = maxN_OLij in equation 1).

To evaluate the relationship between phylogenetic relatedness and the habitat specificity of species, we constructed a phylogenetic tree following the taxonomic classification given in Algaebase (Guiry and Guiry 2008). During the construction for each species, we retrieved the taxonomic hierarchy levels up to the “empire” level as the first step. Next, we applied the as.phylo function on the: empire/kingdom/phylum/subphylum/class/order/family/genus/species nested formula, available in the ape package in R providing a rooted phylogenetic tree without branch lengths. To evaluate the phylogenetic signal in the species rarity, we first computed branch lengths considering Grafen’s calculation of branch lengths, defined as the difference between the height of lower and upper nodes. In the next step, we applied the phylosig function available in the phytools R package (Revell 2012). The calculated Pagel’s λ values, which measured the degree of phylogenetic dependence in the data (Pagel 1999), ranged from 0 to 1, revealing no or strong phylogenetic signal.

Results

Description of the database

The compilation of available literature data resulted in a presence/absence species-by-site matrix of 2489 rows (taxa) and 1145 columns (localities). The size of Hungarian waterbodies ranged from thelmas to the large rivers (e.g., Danube) and lakes (e.g., Fertő and Balaton). The database included representatives of all major groups characteristic for freshwater habitats found in Hungary: Euglenozoa, Cyanobacteria, Miozoa, Bigyra, Charophyta, Chlorophyta, Cryptophyta, Haptophyta, Ochrophyta, and planktonic diatom species. Algae were distributed among 12 phyla (Fig. 2), 31 classes, 81 orders, 205 families, and 528 genera.

Figure 2. Taxonomic distribution of algae in the species-by-side matrix.

Species richness of the habitats

The number of localities across water types and the species richness of the 24 habitat types also displayed considerable differences. Oxbows (1124), bog lakes (921) and streams (865) contained the most species, while pit lakes (106), snow (89), cave waters (83), and thelmas (25) were the most species-poor habitats (Fig. 3).

Figure 3. Number of species and localities in the different habitat types.

Assemblage pattern at regional scale

The pattern of RSOC followed similar (log normal) distribution both when localities were considered and when the analyses were conducted at the level of habitat types (Fig. 4a–b). This “satellite-mode” distribution model indicated that a high proportion of species occurred at few locations and types. Of the 927 species that occurred exclusively in one habitat type (Supplemental Material S2), 787 species were found exclusively in one locality, all of which can be considered satellite species. Satellite species found exclusively in one habitat type fell into 10 phyla, mostly Chlorophyta, Charophyta, and Ochrophyta (Table 1).

Figure 4. Ranked species occupancy curves (RSOC) based on (a) habitat types and (b) localities.

Table 1. Taxonomic distribution of rare (satellite) species (occupied exclusively in 1 habitat type).

Taxonomic classification (phylum)	Number of rare species
Bacillariophyta	15
Bigyra	3
Charophyta	213
Chlorophyta	268
Cryptista	1
Cryptophyta	3
Cyanobacteria	108
Euglenozoa	145
Haptophyta	0
Miozoa	24
Ochrophyta	147
Rhodophyta	0

Based on the number of occupied localities, only a few taxa could be considered core species, mostly belonging to Chlorophyta and Bacillariophyta (Table 2). Investigating the species occurrences at the level of habitat types (Fig. 5a), we found that no species occurred in all 24 water types, and only one species, Actinastrum hantzschii Lagerheim (Chlorophyta, Trebouxiophyceae), occupied 23 water types. At the level of localities (Fig. 4b) the species occupancy values ranged from 1 to 409. Among core organisms, Monoraphidium contortum Komárková-Legnerová (Chlorophyta, Chlorophyceae) was present in the most habitats, occupying 409 localities.

Figure 5. Species occupancy frequency distribution (OFD) curves calculated for (a) habitat types and (b) localities.

Table 2. List of the common (occupied >300 localities) species with the number of their occupied localities.

Species	Number of occurred localities
Monoraphidium contortum Komárková-Legnerová	409
Nitzschia acicularis (Kützing) W. Smith	397
Monoraphidium griffithii (Berkeley) Komárková-Legnerová	373
Stephanocyclus meneghinianus (Kützing) Kulikovskiy, Genkal & Kociolek	343
Lemmermannia tetrapedia (Kirchner) Lemmermann	346
Melosira varians C. Agardh	334
Lepocinclis acus (O. F. Müller) B. Marin & Melkonian	318

The results of species OFD (Fig. 5a–b) were also unimodal, satellite-mode dominant. Bimodality occurred in neither localities nor habitat types.

Microalgae and cyanobacteria in the modified Rabinowitz rarity framework

Applying the Rabinowitz framework to our data-matrix, we distinguished 4 rarity categories (Fig. 6): species that occur in several habitat types in high abundance (SH); species that occur in several habitat types but in low abundance (SL); species occurring only in few habitat types but with high abundance (FH); and finally, species that occur only in few habitats with low abundance (FL). These categories occupy different regions of the plot (Fig. 6).

Figure 6. Distribution of species in the modified Rabinowitz (1981) model. The frequency values of species are indicated with black squares, the theoretical minima of frequency values are marked with stars, and the theoretical maxima with triangles. The rarity categories occupy different regions of the plot. These categories indicated by different labels on the graph: SH = species occupying several habitat types with high abundance; SL = species occupying several habitat types with low abundance; FH = species occupying few habitat types with high abundance; FL = species occupying few habitat types with low abundance.

The minimum frequency terms of several species (mostly of those that occurred in 1–4 habitat types) belonging to the FL category are close to the theoretical minimum, indicating that these species can be considered especially rare (Table 3).

Table 3. List of the taxa having the lowest frequency values occurred in the different number of habitat types.

Number of occupied habitat types	Species	Frequency values	Theoretical minima of frequency values
1	Acanthochloris hungarica Palik	0.0003	0.0003
	Anabaena constricta f. minima Palik	0.0003	0.0003
	Ankylonoton pyreniger Pascher	0.0003	0.0003
	Aphanizomenon klebahnii (Elenkin) Pechar & Kalina	0.0003	0.0003
	Botrydiopsis intercedens Pascher	0.0003	0.0003
	Bumilleria sicula Borzì	0.0003	0.0003
	Bumilleriopsis simplex Pascher	0.0003	0.0003
	Chlamydomonas falsipertusa Nakada	0.0003	0.0003
	Chlorochytrium lemnae Cohn	0.0003	0.0003
	Chrysococcus heverlensis Conrad	0.0003	0.0003
	Closterium pritchardianum var. oligopunctatum Y.V.Roll	0.0003	0.0003
	Coronastrum ellipsoideum Fott	0.0003	0.0003
	Cyanodictyon reticulatum (Lemmermann) Geitler	0.0003	0.0003
	Dactylococcopsis hungarica (Kol) Hindák	0.0003	0.0003
	Dictyochlorella reniformis (Korshikov) P.C.Silva	0.0003	0.0003
	Dinobryon attenuatum (Hilliard) Hilliard	0.0003	0.0003
	Dinobryon eurystoma (A.Stokes) Lemmermann	0.0003	0.0003
	Dinobryon saccatum Prescott	0.0003	0.0003
	Dinobryon suecicum Lemmermann	0.0003	0.0003
	Dolichospermum smithii (Komárek) Wacklin, L.Hoffmann & Komárek	0.0003	0.0003
	Epipyxis utriculus (Ehrenberg) Ehrenberg	0.0003	0.0003
	Granulocystopsis elegans (Fott) Hindák	0.0003	0.0003
	Gymnodinium paradoxum A.J.Schilling	0.0003	0.0003
	Kephyrion globosum (Czosnowski) Bourrelly	0.0003	0.0003
	Kephyrion petasatum Conrad	0.0003	0.0003
	Klebsormidium rivulare (Kützing) M.O.Morison & Sheath	0.0003	0.0003
	Microcystis natans Lemmermann ex Skuja	0.0003	0.0003
	Mischococcus confervicola Nägeli	0.0003	0.0003
	Monoraphidium nanum (Ettl) Hindák	0.0003	0.0003
	Monoraphidium obtusum (Korshikov) Komárková-Legnerová	0.0003	0.0003
	Neocystis brevis (Vischer) Kostikov & Hoffmann	0.0003	0.0003
	Phormidium retzii var. nigroviolaceum (Wille) V.Poljanskij	0.0003	0.0003
	Protoperidinium bulla (Meunier) Balech	0.0003	0.0003
	Pseudokephyrion conicum Schiller	0.0003	0.0003
	Pseudokephyrion poculum W.Conrad	0.0003	0.0003
	Spirogyra inflate (Vaucher) Dumortier	0.0003	0.0003
	Strombomonas gibberosa (Playfair) Deflandre	0.0003	0.0003
	Strombomonas helix (Deflandre) R.A.Townsend & Huisman	0.0003	0.0003
	Strombomonas praeliaris (T.C.Palmer) Deflandre	0.0003	0.0003
	Synura globose (Schiller) Starmach	0.0003	0.0003
	Trachelomonas asiatica (Skvortzov) Deflandre	0.0003	0.0003
	Trachelomonas volzii Lemmermann	0.0003	0.0003
	Volvox carteri f. weismannia (J.H.Powers) M.O.P.Iyengar	0.0003	0.0003
2	Carteria turfosa Fott	0.0006	0.0006
	Desmatractum delicatissimum Korshikov	0.0006	0.0006
	Jaaginema profundum (Schröter & Kirchner) Anagnostidis & Komárek	0.0006	0.0006
	Oscillatoria angusta Koppe	0.0006	0.0006
	Pediastrum biradiatum var. emarginatum (Ehrenberg) Lagerheim	0.0006	0.0006
	Pediastrum duplex var. cohaerens (Bohlin) Ergashev	0.0006	0.0006
	Phacus lismorensis f. minor Hortobágyi	0.0006	0.0006
	Strombomonas ovalis (Playfair) Deflandre	0.0006	0.0006
	Trachelomonas obtusa var. papillata Drezepolski	0.0006	0.0006
3	Oscillatoria rupicola (Hansgirg) Hansgirg ex Forti	0.0015	0.0015
	Pormidium lutescens (Hansgirg) Anagnostidis & Komárek	0.0015	0.0015
	Trachelomonas lemmermannii Wołoszyńska	0.0015	0.0015
4	Pediastrum duplex var. lividum Raciborski	0.0026	0.0026
5	Eremosphaera gigas (W.Archer) Fott & Kalina	0.0063	0.0043
6	Cartusia fontana (Hansgirg) Mai, J.R.Johansen & Pietrasiak	0.0075	0.0062
7	Elakatothrix gelatinosa Wille	0.0129	0.0081
8	Conticribra guillardii (Hasle) Stachura-Suchoples & D.M.Williams	0.0210	0.0122
9	Volvox globator Linnaeus	0.0221	0.0132
10	Lagerheimia balatonica (Scherffel) Hindák	0.0338	0.0164
11	Chlamydomonas pertusa Chodat	0.0413	0.0196
12	Desmodesmus denticulatus var. Linearis (Hansgirg) Hegewald	0.0491	0.0231
13	Phacus longicauda f. tortuosus (Skvortzov) Rai & Uherkovich	0.0542	0.0283
14	Closterium aciculare T.West	0.0763	0.0352
15	Didymogenes palatina Schmidle	0.0695	0.0422
16	Trachelomonas hispida (Perty) F.Stein	0.0971	0.0491
17	Lyngbya martensiana Meneghini ex Gomont	0.1238	0.0575
18	Trachelomonas scabra Playfair	0.1302	0.0658
19	Oocystis borgei J.W.Snow	0.1542	0.0741
20	Gonium pectorale O.F.Müller	0.1365	0.0825
21	Euglenaformis proxima (P.A.Dangeard) M.S.Bennett & Triemer	0.3900	0.0929
22	Pandorina morum (O.F.Müller) Bory	0.3339	0.1033
23	Actinastrum hantzschii Lagerheim	0.5169	0.1241

The minimum frequency values of species belonging to the SL group move farther away from the theoretical minimum values as the number of occurrences increases. The maximum frequency values of SH and FH group species seemed considerably lower than the theoretical maximum (Fig. 6).

Rarity–body size relationship

The occupied number of water types and localities showed an apparently nonlinear relationship with the biovolume of microalgae. Quantile regressions, however, indicated significant differences at all quantiles of the distributions. The results from quantile regression showed significant decreasing tendencies in the case of each quantile, from the lower (25%) to the uppermost (95%) quantiles (Fig. 7a–b). These outcomes imply that phytoplankton species with smaller biovolumes are able to populate more habitats than larger ones.

Figure 7. Rarity–body size relationship in (a) number of water types and (b) number of localities for the quantiles 50% (red), 75% (magenta), 90% (blue), and 95% (cyan).

Rarity of taxa in the different habitat types

We observed significant differences in rarity metric values among the water types (Fig. 8). High rarity metric values (>0.9975) characterised the following habitat types: streams, brooks, rivers, canals, oxbows, reservoirs, soda pans, fishponds, pit lakes, and lakes. Low rarity metric values (<0.9950) were found exclusively in snow. In other habitat types, rarity values varied between 0.9950 and 0.9975.

Figure 8. Distribution of rarity metrics in the 24 habitat types. The boxplots summarize the average rarity metric values in the different habitat types (the centre line within the box represents the median; the upper and lower edges of the box represent lower and upper quartiles; circles are outliers). Values on the type axis are codes of habitat types as specified in Supplemental Material S1. Comparing the different rarity metrics in the different habitat types resulted in significant differences between the habitat types (Supplemental Material S3).

The Wilcoxon signed-rank test denoted statistically significant differences (p < 0.00001) between several habitat types (values referring to significant differences are in red in Supplemental Material S3). The number of rare taxa differed considerably among the types. Bog lakes (11), oxbows (7), and soda pans (9) contained the most species that did not occur in any other types (Table 4).

Table 4. Numerical distribution of rare (occupied exclusively 1 habitat type) species by habitat types.

Code of water types	Habitat types	Number of rare species
1	Cave water	5
2	Springs	4
3	Streams	95
4	Brooks	7
5	Rivers	55
6	Canals	47
7	Oxbows	138
8	Reservoirs	16
9	Soda pans	111
10	Soil surfaces	3
11	Bog lakes	244
12	Ponds (<1 ha)	37
13	Small temporary wetlands	15
14	Thelmas	2
15	Pool	5
16	Fishponds	36
17	Water-filled pits	13
18	Lake Balaton	33
19	Ricelands	19
20	Snow	2
21	Marshlands	11
22	Pit lakes	2
23	Large saline lakes	11
24	Lakes	16

Effects of the phylogenetic relatedness

The phylosig test resulted in a strong phylogenetic signal, which significantly differed from 0 (λ = 0.208; logL(λ) = 4266.6; LR(λ = 0) = 19.78; p < 0.00001; Supplemental Material S4), clearly indicating that evolutionary relations should be considered when evaluating species rarity.

Discussion

The species inventory compiled for this study contained 2489 algae species and 1145 localities, a similar number of taxa to that published for other countries or regions of similar size (Iraq 2013: Maulood et al. 1993; Ukraine 3708: Tsarenko et al. 1999; Thailand 1001: Ariyadej et al. 2004). The REBECCA database (Moe et al. 2008) that compiles phytoplankton monitoring data from 13 countries (1450 lakes) contains ∼2300 species. However, these values are only rough estimates of the real species richness of a region because of differences in sampling effort (both in terms of the temporal and spatial scales), types of localities covered, taxonomic expertise of the data providers, and taxonomic coverage of the inventories; therefore, the usefulness of direct comparisons among species inventories is highly limited. Learning more about large-scale ecological processes would require more information on species distributions, but the number of available species inventories for larger areas is low because, unfortunately, a compilation of inventories is not particularly rewarding for scientists.

The considerable differences in species richness among the water types supported our first hypothesis (H1) and can be traced back to various methodological and ecological reasons. Species richness should correlate with the number of localities within the types, sampling frequency, size, habitat diversity, or connectedness of localities. Original data sources do not provide information on sampling methods or frequency, but several are worth discussing. The large species richness of streams and canals is only partly accounted for by the large number of localities within these types. Small water currents serve as sink habitats for macroscopic (Roberts and Rahel 2008) as well as microscopic organisms (Bolgovics et al. 2017, Borics et al. 2021). They are connected with various other aquatic habitats of the watershed, and, because of their small size, microflora of the incoming waters considerably increases their diversity (Borics et al. 2015). Large river diversity is much less affected by the incoming waters, but these rivers are monitored by water management authorities, and thus their intensive sampling explains their high species richness (Kiss and Schmidt 1998). In accordance with the old ecological paradigm suggesting that species richness increases with area (Arrhenius 1921, Gleason 1922, MacArthur and Wilson 1967), small waterbodies had significantly lower numbers of species than larger ones (Smith et al. 2005). Recent studies, however, demonstrated that in phytoplankton, species richness increases with surface area, but because of their extended littoral zone and large habitat diversity, lakes in the 10⁵–10⁶ m² size range occasionally have more microalgal diversity than the larger lakes (Várbíró et al. 2017). Although Lake Balaton or large saline lakes like Fertő (Neusidlersee) and Velence have been being monitored for decades, their phytoplankton diversity is considerably smaller than the cumulative species richness of oxbows and bog lakes. This finding is in line with the results reported by Bolgovics et al. (2019), who found that the species richness of small isolated waterbodies exceeds that of a large one of similar size.

The shape of RSOC and OFD curves suggests that at the regional scale, phytoplankton consists of a few frequent and extraordinarily large number of rare species. In line with the common–rare species categories proposed by Raunkiaer (1918) and the core–satellite model suggested by Hanski (1982) that predict bimodal OFDs, we expected similar patterns (H2), but this pattern was not characteristic for our data. Most studies report right-skewed unimodal OFDs (Malmqvist et al. 1999, Soininen and Heino 2005, Heino and Virtanen 2006). The occurrence of bimodality can be traced to a sampling artefact (Papp and Izsák 1997), good connectedness, and colonization dynamics within metacommunities (Hanski 1982), or to the effect of the spatial scale on species occupancy (Collins and Glenn 1997, Hui and McGeoch 2007). Note that Hanski’s core–satellite model explains species distributions among interconnecting habitats that belong to similar habitat types. Unpredictable spatiotemporal fluctuations in environmental conditions may occur in these habitats, but their basic environmental characteristics are identical. Our results suggest that most of the proposed water types have considerable differences in their environmental properties, and several do not provide suitable habitats for the incoming algae. The other explanation for the lack of bimodal OFDs is that the increasing spatial scale of investigations coincides with a reduction in connectedness, which reduces the number of common species even in the case of microorganisms with high dispersal capabilities. Investigating the phytoplankton composition in the cross-section (small scale) and longitudinal-section (large scale) of an oxbow, Görgényi et al. (2019) found truncated log-normal RSOC curves in both cases, but bimodal OFDs could be found only for the cross-section samples because longitudinal differences were so large within the oxbow that core species could not be identified.

While RSOC and OFD curves can be considered 1-dimensional representations of rarity, Rabinowitz (1981) applied 3 rarity dimensions to analyse the flora of British Isles (Rabinowitz 1981): geographic distribution, habitat specificity, and local population size of species. During the analysis of the microalgal species inventory of Hungary (covering most of the Pannonian ecoregion), rarity of species could be characterised by 2 dimensions: habitat specificity and population size. When positioning the species in the modified Rabinowitz space we applied continuous scales, and we did not apply boundaries to separate taxa. Distribution of species in the occupied habitats/frequency metric plot, however, clearly indicated that most microalgal species belong to the high habitat specificity but low abundance group. The literature demonstrates that most macroscopic species belong to one of the rarity categories, as has been proved for birds (Kattan 1992), rainforest (Caiafa and Martins 2010), and savannah trees (Maciel and Martins 2021), or for deep sea bivalves (McClain 2021). However, the large number of stenoecious microalgal species that occurred only in one locality was surprising, especially because of the large number of phytoplankton samples investigated in the last century in Hungary. From the late 19th century, famous algologists (Jenő Cholnoky, Géza Entz, Erzsébet Kol, Gábor Uherkovich, Tibor Hortobágyi) revealed the microflora of the region, and since the late 1960s, phytoplankton in hundreds of waterbodies have been monitored regularly; thus, the estimated number of studied samples highly exceed hundreds of thousands.

The strong phylogenetic signal in the rarity metrics of the studied microalgae indicates that the frequency of occurrence is substantially influenced by the evolutionary history of the species (H3). The vast majority of rare taxa belong to the desmids. Representatives of this group evolved sometime in the Ordivician, and they were the first algae to conquer freshwater habitats (Becker and Marin 2009). These taxa have high habitat specificity, frequently occurring in the metaphyton of littoral areas of lakes and in small bog lakes, which can be considered small islands in the terrestrial landscape. Colonization of these habitats is difficult because of their small size and low number in the Pannonian ecoregion. Desmids, however, could effectively invade these habitats because their spores have strong, pigmented cell walls composed of the resistant polymer sporopollenin (Delwiche et al. 1989). The spores of this group became resistant to chemical and mechanical damage and to mutagenic UV light, which facilitates their successful dispersal.

Several rare taxa occurred in the Miozoa phylum. Although some representatives of this group are widespread in waters (Grigorszky et al. 2003), most species have high habitat specificity and only occur in a few waterbodies in the region (Grigorszky et al. 1999). Similar to desmids, dinoflagellates also have strong thick-walled resting forms, hypnozygotes, that help them colonize new localities (Matsuoka and Fukuyo 2003).

Rarity and uniqueness of centric diatoms Cyclotella gamma Sovereign, Cyclotella polymorpha Meyer et Håkansson, Stephanodiscus agassizensis Håkansson et Kling, and Stephanodiscus binderanus (Kützing) Krieger in the Pannonian ecoregion, however, cannot be necessarily traced back to colonization difficulties or high habitat specificity. The use of high-resolution scanning electron microscopy revealed new species and large cryptic diversity in the region (Ács et al. 2016, 2017a, 2017b). These investigations, however, were restricted to a few waterbodies, and thus uniqueness of this taxa is supposedly an artefact of differences in resolution of taxa identification across the waterbodies.

Body size is a key trait of species, associated negatively to population density and positively to habitat size (Yu and Dobson 2000). A variety of ecological theories predicts that small-bodied species are generally more abundant and less threatened by extinction than large-bodied species (McKinney 1997, Oindo et al. 2001). This theory is clearly and widely correlated in natural animal communities and has been proved for birds (Blackburn and Gaston 1994, 1996, Owens and Bennett 2000) and mammals (Blackburn and Gaston 1998, Cardillo et al. 2005), but the theory remains untested for microalgae. In this study, we calculated the mean body volume for 2489 algae species to test the hypothesis. Interestingly, in line with our fourth hypothesis (H4), the pattern seemed similar to that described previously for macroscopic organisms. The wedge-like distribution of the biovolume–occupancy relationship implies that among the small-sized microalgae, in addition to the large number of habitat specialist species, several subcosmopolitan taxa can also be found in large numbers. For large-sized taxa, however, the ratio of tolerant subcosmopolitan species is considerably smaller. This pattern theoretically can be explained by differences in dispersal capabilities and tolerance spectra of species. While dispersal limitation is a more likely constraining factor for macroscopic organisms, this factor is irrelevant in the size range of microalgae. Large-celled taxa like dinoflagellates or several desmid species have thick-walled hypnozygotes, or resting cysts, which efficiently aid their dispersal because microalgae as dormant walled cysts can survive long periods of drought, starvation, and other threats (Schaap and Schilde 2018). Likely more important is that most of these taxa are K strategists (Sommer 1981, Reynolds 2006) and thus are good competitors and dominate the phytoplankton in late successional phases. The small aquatic habitats are frequently disturbed (Borics et al. 2013), providing a competitive advantage for small-celled, r-selected species; however, small-celled K strategists may also occur in small waterbodies. Most silica-scaled chrysophyte species have been reported from small, sheltered waterbodies in Hungary (Péterfi et al. 1998a, 1998b, Padisák et al. 2000). An additional explanation is that, similarly to macroscopic organisms, microalgae also have habitat size preferences (Borics et al. 2016). Limnological characteristics of small habitats (<10² m²) are not favourable for large-celled taxa lacking good buoyancy regulation mechanisms (e.g., large-celled centric diatoms).

Conservation outlook

The Birds and Habitats Directives together with the Water Framework Directive provide a legislative basis to protect and restore diversity and good status of aquatic- and water-related ecosystems in the EU member states. Many projects were implemented in recent years to improve the quality of these systems, but the vast majority of financial sources has been spent to protect birds and mammals, and the share for invertebrates was much smaller (Mammola et al. 2020). Microscopic organisms (to our knowledge) were never targets of restoration actions. Although benthic microflora and phytoplankton are among biological elements used to monitor the ecological status of surface waters, these organisms serve as indicators of nutrient enrichment and not as elements of an ecosystem that should be protected (B-Béres et al. 2021). Also problematic is that the good ecological status suggested by microalgae does not coincide with their high diversity. Diversity of microalgae reflects the fluctuation of resources and not their absolute quantity (Sommer 1984, Padisák et al. 1993), which is why diversity metrics are not included in phytoplankton indices used by EU member states (Carvalho et al. 2013). Species diversity can be influenced by a great variety of system properties, but the nature conservation value of aquatic ecosystems is an intrinsic biological quality that should be linked more strongly to uniqueness and rarity of species than to diversity. This study revealed the importance of small waterbodies in maintaining the populations of rare microscopic algae, yet small habitats are ignored by WFD. Although several of these waterbodies are part of the Natura 2000 sites, this status is attributable to macroscopic taxa, and monitoring of their microflora is not an obligation for the official bodies. Small aquatic ecosystems (bogs and marshlands) are increasingly threatened by undesirable consequences of anthropogenic loads and climate change. Oxbows, marshlands bogs or bog lakes, and soda pans are groundwater-related ecosystems. Groundwater depletion has become a global problem that threatens groundwater-dependent ecosystem stability and biodiversity worldwide (Famiglietti 2014, Devitt et al. 2019). Because of the considerable decrease in groundwater level in the region (Mezősi 2017), these systems might disappear, especially bogs and bog lakes, which need stable eustatic conditions to existence. Drying of these systems cannot be stopped without artificial water supply, and as a last resort, although it raises ethical concerns (Minteer and Collins 2010), translocation of species must be considered.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by Hungarian Scientific Research Fund (NKFIH OTKA) project no. K-132150, as well as the KDP, Ministry of Technology and Innovation, and the National Research, Development and Innovation Office project no. KDP-2020.