Effects of sampling site, season, and substrate on foraminiferal assemblages grown from propagule banks from lagoon sediments of Corfu Island (Greece, Ionian Sea)

Foraminiferal propagule banks occur in fine sediment fractions that contain small individuals of benthic foraminifera. These sediments include locally sourced juveniles and propagules, as well as allochthonous propagules that have dispersed from surrounding areas. Such propagules can remain viable even under unfavorable local conditions. When exposed to more favorable conditions, they may grow to adult stages. Accordingly, during environmental changes, propagule banks have the potential to function as species pools and allow quick assemblage reactions. The propagule method was designed to study responses of foraminiferal assemblages by exposing propagule banks to controlled conditions in the laboratory, an approach that is applicable to a variety of ecological questions. Therefore it is important to understand the nature and dynamics of propagule banks, including local and seasonal influences. To obtain insights into the composition of local propagule banks, we studied experimentally grown assemblages from two shallow-water lagoons on Corfu Island in western Greece, and compared the results with in situ assemblages. We sampled in spring and autumn of 2017 and experimental treatments included the use of different substrates in our experiments to account for potential effects on assemblage compositions. Results revealed that sediments from each lagoon contained a distinct propagule bank. We found abundant allochthonous taxa among specimens grown in all experimental treatments, indicating dispersal of propagules, and possibly also juveniles, from adjacent regions into both lagoons. The time of sampling had a significant effect on experimental assemblages, indicating that the composition of propagule banks can vary throughout the year. However, no significant differences were found in assemblages grown in different substrata, suggesting a stronger influence of water variables (e.g., temperature or salinity) on assemblage compositions. Moreover, the experimental set-ups favored small, fast-growing, sediment-dwelling species tolerant of relatively high organic content. Our findings highlight the potential of propagule banks as species pools and will help to refine and improve future applications of the method.


Introduction
Assemblages of benthic foraminifera are widespread and nearly ubiquitous in the modern oceans. In shallow-water, coastal environments they are found in almost all habitats and form habitat-specific assemblages, depending on environmental conditions and available microhabitats (e.g., [1,2]). As such, they are widely applicable in ecological and paleoecological research and are increasingly used as indicators for bio-monitoring [2][3][4][5][6][7][8][9]. Most studies focus on adult foraminiferal assemblages.
The fine size fraction of the sediment, however, contains numerous living small juvenile forms and cryptobiotic propagules. Such specimens, which can either derive from reproduction of local populations or be transported to the respective sites from adjacent areas, form the so-called propagule bank. The term "propagule" refers to tiny juveniles (maybe just the proloculus [10]) that can easily be dispersed due to their small size and be transported well beyond where they were produced [10][11][12]. A variety of dispersal mechanisms are known for benthic foraminifera including transport in suspension by currents [13,14], rafting on objects (e.g., [15]) or carried in the intestinal tracts of metazoans, including fish [16] or on feet and feathers of birds [17,18]. Thus, dispersal in form of propagules appears to be quite widespread among many species of benthic foraminifera [11,19,20]. If propagules are transported into habitats outside their normal environmental conditions, they can become dormant [21], which has also been observed in adult foraminifera [22], and can remain viable within the local propagule bank for at least two years [19]. If local environmental conditions should become suitable (e.g., due to climate change), subsequent growth of the propagules to adult populations is possible [10,19]. Recent observations in the Bottsand lagoon (Baltic Sea) have provided field evidence of the presence of allochthonous propagules, as the previously absent Elphidium incertum suddenly appeared among local assemblages after a short period of increased salinity and even prevailed after conditions returned to their original state [23].
The propagule method is an experimental procedure in which propagule banks are concentrated in the fine fraction by removing the coarser sediments (e.g., using sediment sieves). Subsamples of the fine sediments are then placed under different environmental conditions (i.e., experimental treatments) in the laboratory [11,12]. The exposure to "new" conditions can lead to growth and even reproduction of faunal assemblages that differ in diversity and faunal composition from the in situ assemblage at the respective collection site [10,12,19,20,24]. Allochthonous species have been documented in several growth experiments using the propagule method (e.g., [21,25,26]). Some studies have recorded shallow-water species that grew from sediments collected from deep-water environments [10,19]. Others documented the occurrence of marsh or open-shelf species in experimental assemblages from shallow-water or intertidal study sites [12,20,24].
The presence of allochthonous species within the propagule banks, which are not normally observed within the in situ assemblages (>63 μm), significantly increases the overall diversity of foraminiferal communities in various environments (e.g., [19,24]). As such, they can contribute to community structures in the sense of a "species pool", a hypothesis introduced by Buzas and Culver [27]. It postulates that any community at a certain time can be considered as a subset of a larger species pool that includes all potential immigrants and emigrants from adjacent habitats over time, as long as these habitats are connected [27]. This explains rapid faunal responses in benthic foraminiferal assemblages that were previously observed (e.g., [28]). The hidden diversity of the propagule banks can contribute to such responses, since dormant propagules might be present at a site at any time. In the wake of ongoing global ecological alterations to aquatic environments globally, it becomes more important to analyze and quantify community structures and dynamics for potential response mechanisms, including local propagule banks.
The Mediterranean Sea is a highly suitable area for such studies, because it is one of the most severely impacted marine regions with regard to ongoing environmental changes (e.g., [29]). Climate change leads to increasing temperatures and sea-level rise. The latter will strongly affect shallow-water environments such as coastal lagoons, which face the possibilities of a continuous "marinization" [30]. Coastal lagoons, including those in Greek waters, harbor specific foraminiferal communities (e.g., [31][32][33][34][35][36][37][38][39]), which are modified as a result of ecological alterations.
Under conditions of global environmental change, the role of propagule banks as potential species pool contributors may become even more important. As such, an important research target is the analysis of structures and dynamics of those propagule banks. A better knowledge and understanding of propagule banks may improve and refine future growth experiments using the propagule method, which provides a wide range of potential applications [11].
To enhance our understanding of the composition and dynamics of local propagule banks, we conducted growth experiments on material from two shallow-water lagoons on the eastern coast of Corfu Island (Western Greece): Chalikiopoulou and Antinioti. Our experimental setups were designed to target the following research questions: 1) How does the sampling site influence the faunal composition of local propagule banks? Previous studies have already demonstrated the presence of individually composed propagule assemblages at sites that differ environmentally (e.g., with regard to temperature and salinity [20,24]). Yet differences between sites that are ecologically similar have not been documented. Furthermore, adult foraminiferal assemblages in shallow-marine environments often exhibit patchy distributions [40][41][42] and it has not yet been determined if propagule banks are also patchy.
2) How does the time of sampling influence the composition of local propagule banks? A comparison between two previous growth experiments using propagule banks from sediments that were sampled at the same site but at different times of the year, showed noticeable differences in their respective experimental assemblages, even under comparable experimental conditions [12,20]. As shallow-water benthic foraminifera exhibit high variability in reproduction cycles (e.g., [41,[43][44][45]), this should lead to seasonal variations in propagule production and subsequent dispersal and settling.
3) How do different types of substrate influence the composition of assemblages grown from the same propagule banks? Since shallow-water foraminiferal species exhibit a large variety of life-modes (e.g., epiphytic, infaunal, etc.) [2], the availability of offered additional microhabitats during the experiments may affect the resulting assemblage compositions.

Sampling area
Sediment samples were taken from two shallow-water lagoons at Corfu Island: Chalikiopoulou lagoon on the eastern coast and Antinioti lagoon on the northeastern coast (Fig 1). Sampling was permitted through collaborative research with the National and Kapodistrian University Athens. No additional specific permissions were required for this work, which also does not involve endangered or protected species. Both lagoons exhibited calm and sheltered conditions for the accumulation of fine sediments, which is important for the collection of propagule banks. At the same time, both lagoons were linked to the open ocean, featuring a high connectivity with surrounding habitats.
Chalikiopoulou lagoon is situated south of Corfu Town. To the east, it is separated from the sea by the Kanoni Peninsula, leaving only a small inlet (approximately 300 m) to the south. The lagoon has a north-south extension of 2.1 km and a maximum east-west extension of 1.4 km. On its eastern coast, it is crossed by the runway of Corfu Airport. Nevertheless, the ecological status of the lagoon is good and it is a Natura 2000 site (http://ec.europa.eu/ environment/nature/natura2000/). Local fishing activities are common. The average water  Table 1). (D) Antinioti lagoon with location of two subsamples taken in May and October 2017 (see also depth of the lagoon is <1 m [46] and tidal influence is low with a range of~20 cm (personal observation and www.tide-forecast.com). The inlet is bridged by a narrow wall with several larger openings to allow tidal currents to flow in and out of the lagoon. Near the inlet, a continuous current is visible. The sediment at the sampling site near the inlet is greyish in color and is composed of mostly silt, mud, and fine sand. Bioturbation is common. Organic material (mostly plant debris from algae and surrounding vegetation) is visible, and the sediments turn black below 1 cm of sediment depth.
Antinioti lagoon is located on the northernmost point of the island, approximately 25 km north of Corfu Town. The main lagoon has an east-west extension of 1 km and a north-south extension of 270 m. It is connected to the open ocean by an elongated inlet to the northwest and a shorter inlet to the east. Water depth is comparable to Chalikiopoulou lagoon and the tidal range is only a few centimeters (personal observation and www.tide-forecast.com). The lagoon is used for fish and shrimp farming and is also a Natura 2000 site. Unlike Chalikiopoulou, no current is visible~100 m behind the inlet mouth and the water is very calm. The sediment at the sampling site near the eastern inlet is brown-grey in color and is composed of silt, mud, and very fine sand. Especially in the still-water areas behind small shrubs of vegetation, organic material (mostly plant debris) is clearly visible and the sediment turns dark after a few millimeters. Bioturbation is less distinct compared to Chalikiopoulou lagoon.

Sample collection and treatment
Samples were taken at both sampling sites in spring (May 26) and autumn of 2017 (October 13) from mudbanks that were partially exposed during low tide. Water temperature and salinity values measured onsite with a digital thermometer (Extech Instruments, 0.1˚C resolution, ±1˚C accuracy) and a hand-held seawater refractometer (Red Sea) are provided in Table 1. Samples were taken by scraping the upper-most sediment layers (max. 5 mm) within two areas of approximately 5 m 2 per lagoon, resulting in two subsamples of~1-2 L per site. The material was placed into 5 L plastic containers and carefully homogenized.
After collection, the sediments were sieved on site using ambient seawater and a 53-μm stainless-steel sieve (as in [12,20,24]). The fractions >53 μm were preserved in 90% ethanol (buffered with sodium carbonate to avoid acidic conditions) with 2 grams of rose Bengal per liter of ethanol to distinguish the living and dead assemblages (modified from [47]). Despite some challenges, the staining with rose Bengal is deemed a sufficiently reliable technique in warm and oxygenated environments [48] and can serve as an estimate of the living assemblage. After two weeks, the stained samples were washed over a 63-μm sieve and dried for at least 48h Effects of site, season, and substrate on foraminiferal assemblages grown from propagule banks at 50˚C. Foraminifera from each sample (see Table 1) were picked (until~300 benthic specimens), identified, and counted (S1 Table). Species identification was mainly based on [49][50][51][52][53][54][55]. The fractions <53 μm were collected in 5 L plastic containers. After the suspended sediment settled, the subsamples contained approximately 100-200 ml of fine sediment. On May 27 and October 14 2017, the containers were sealed with parafilm and placed in insulated boxes for transport. A temperature logger (LOG200, Dostmann electronic) was placed within one of the boxes for the monitoring of ambient temperature during the 18h transport. In May, temperatures during transport varied between [16][17][18][19][20][21][22].8˚C and in October they ranged from 17.8-22.6˚C. After arrival at the micropaleontological laboratory of the University of Bonn (Germany), the containers were opened and the suspension was left to settle for approximately 28h before being processed for the growth experiments. Temperatures in the laboratory ranged from 21.

Growth experiments
Growth experiments were performed using the Propagule Method [11,12] with some modifications to adjust for the local conditions of the sampling sites and the research questions. From the fine fraction (<53 μm), 10-ml subsamples of sediment were taken and placed into translucent containers (100 ml; polypropylene), along with 60 ml of artificial seawater (Coral Pro salt, Red Sea). The concentration of the Coral Pro was adjusted to a salinity of 40 ppt, which was close to ambient conditions, except for Antinioti in October (see Table 1). Salinity values within the containers were checked twice per week with a hand-held seawater refractometer (Red Sea). To simulate different additional substrata, 3 leaves of an artificial water plant and 15 pieces of rubble-sized hydrocarbonate (Aqua Medic, size fraction 5 mm to 1 cm) were placed into one third of the containers respectively while the remaining third was left as plain mud sediments. The containers were sealed with tight-fitting lids and placed into an incubator (ST2Basic, Pol-Eko Aparatura) set at a daily temperature cycle varying between 26 and 30˚C to simulate climate conditions of the northern Red Sea in summer (see [56]). The experimental containers were illuminated on a daily cycle (14h of light and 10h of darkness) to simulate photic conditions in the summer and to promote algal growth. We did not use additional aeration of the containers, since previous studies showed that the use of polypropylene containers ensured sufficient gas exchange and oxic conditions throughout the experiments [10-12, 20, 24]. The oxygen content of each individual container was checked twice during the course of the experiment (after 3 and 6 weeks, respectively; O 2 -Test for aquaria, color scale, JBL) and values did not drop below 183 μmol/kg. At the same interval, pH was checked and values remained around 8 throughout the experiment (pH-indicator paper, neolab). Every 10-11 days, one third of the artificial seawater was replaced with the same amount of fresh Coral Pro, which was aerated until the day before the water change. No additional food was added but algal growth was observed throughout the course of the experiment. The algae and assorted bacteria provided nutrition. Goldstein and Alve [12] and Weinmann and Goldstein [20] also reported signs of feeding in the algal mats of their propagule experiments.
The experimental design resulted in three treatments (3 substrate types) for each of the two subsamples from both sampling locations, which were sampled twice (May and October). Each treatment had a replicate, resulting in a total of 48 treatments. The treatments were harvested after 6 weeks (July 11 and November 27 2017, respectively) by sieving over a 63-μm stainless steel sieve. The use of a 63-μm sieve ensured that foraminifera retained on the sieve must have grown by at least 10 μm over the course of the experiment, which also reduces the possible bias due to different mesh-passing capabilities in differently shaped forms. The resulting material was preserved in 90% ethanol with 2 grams of rose Bengal per liter ethanol. After 2 weeks, the material was washed with tap water over 63 μm and all foraminifera were picked wet. Polythalamous foraminifera were identified, counted and stored in 70% ethanol (S2 Table and S3 Table). The staining of the material allowed distinguishing those specimens that were likely alive at the end of the experiment. Unstained or empty tests reflected reproduction and/or death of specimens during the experiment [11,48].
All samples were coded for their respective sampling locations, subsamples, season, and substrate types, as listed in Table 2.

Data analysis and statistics
All foraminiferal faunal analyses such as diversity calculations of assemblages (species richness (S), Shannon (H), and Berger-Parker dominance indices (max p i ); see [57]), ternary plots, cluster analysis, and multi-dimensional scaling (nMDS) were performed with PAST 3.13 software [58]. Relationships between assemblage characteristics and sampling or experimental variables (sampling site, sampling season, subsamples, substrate type) were tested using analysis of variance (ANOVA) or analysis of similarity (ANOSIM), both performed with PAST 3.13 software. Unless otherwise stated, total abundances of pooled replicates were used in all analyses (stained & unstained assemblages of replicates a & b) as in [12,20,24]. For cluster analyses and MDS plots, abundances were square-root transformed and resemblances were calculated using the Bray-Curtis similarity index. Bray-Curtis similarity was also used for ANOSIM analyses. PASTplots were transferred to Inkscape (V. 0.92.3, www.inkscape.org) and prepared as figures. For a better understanding of the nature and influence of foraminiferal dispersal, we categorized the benthic taxa found in our study into "autochthonous", "sporadic", and "allochthonous" taxa (S4 Table). Based on a classification used by Weinmann and Goldstein ([24] see Table 3 therein), we categorized species as autochthonous if they were found among the stained (living) in situ assemblages or if they were commonly present (>1%) among the dead in situ assemblages. Species that were only sporadically present (<1%) within the in situ dead assemblages were deemed sporadic, indicating that they might be part of the in situ communities at some time of the year and that their status was uncertain. Species that were absent from the in situ assemblages but grew during the experiments were categorized as allochthonous.
For additional analyses, both in situ and experimental benthic taxa were sorted into ecological and functional groups (S5 Table). We applied the ecological groups defined for the calculation of the Foram-AMBI index. Foram-AMBI was based on Borja et al. [59] and developed by the Fobimo group. It was first described by Alve et al. [8] from Atlantic and Arctic areas and further developed for the Mediterranean Sea by Jorissen et al. [9]. To calculate the index, species are distinguished based on their sensitivity to organic enrichment: Group 1 is deemed "sensitive", Group 2 is deemed "indifferent" and Groups 3-5 are categorized as third-, second-, and first-order opportunists [9]. The latter generally increase in abundances with increasing organic enrichment, whereas the sensitive species will disappear [9]. We assigned our species data to the five ecological groups and calculated their relative abundances within each sample. We only included species that could be clearly assigned according to the species list presented by Jorissen et al. [9], terming the remaining taxa as "unassigned".
We grouped the benthic foraminiferal taxa into "functional groups" based on their predominant mode of life (S5 Table). We applied this to evaluate the assemblages for possible effects of the three simulated substrate types muddy, phytal, and rubble. For this, we adjusted the categorization of Langer [60], which includes four types of epiphytic species: Epiphytic a (predominantly sessile), epiphytic b (temporary mobile), epiphytic c (predominantly mobile) and epiphytic d (permanently mobile). We further included the groups epifaunal (predominantly on sediment) and infaunal (predominantly within sediment). We included all species in this analysis and chose one predominant life mode for those species that are known to exhibit more than one.

Experimentally grown assemblages
During the course of the experiments (6 weeks), the water quality was tested regularly (S6 Table). Small increases in salinity within individual containers (~0.5-2 ppt) were recorded and were balanced by adding small amounts of distilled water. The experimental containers remained well oxygenized throughout the experiments (concentrations >183 μmol/kg) and pH remained constant (S6 Table). Green and brown algae grew during experiments and built mats on the sediment surface and on rubble (if present). At the termination of the experiments, foraminiferal shells were visible on the sediment surface.
Between 517 and 4487 individuals grew per replicate with 1157-7482 individuals per combined set of replicates (Table 4). Season (time of sampling) was significant for the number of individuals found (ANOVA F (1,22) = 8.3, p = 0.01), whereas numbers did not differ significantly between sampling sites or substratum treatments.
Of the 88 species of benthic foraminifera that grew within the treatments, the majority were rare. Fewer than 10 individuals were found for each of these 45 rare species and for 12 of those, only one individual was found. The species that grew most abundantly in the treatments were Ammonia tepida, Quinqueloculina seminula, Pseudotriloculina rotunda, Pseudotriloculina cf. P. oblonga, Textularia bocki, Haynesina depressula, Rosalina bulloides, and Miliammina fusca.
Species richness (S) varied between 20-46 species within combined replicates. Higher species richness was found in treatments with sediments from Chalikiopoulou lagoon (samples 16, 17, 59 and 60; Table 4). Sampling site and season (time of sampling) were significant for species richness (Table 5) Season had a significant effect on Shannon Index values (Table 5), which were higher in spring. If treatments from both sampling sites were analyzed separately, season was significant for species richness in Chalikiopoulou, while it was significant for Shannon diversity in Antinioti (Table 5). No tested variable was significant for dominance (Berger-Parker index; Table 5). The different substrate types revealed no significant effects on faunal diversity (Table 5). Only one significant correlation was found for Shannon diversity between muddy and phytal substrate in treatments from the Chalikiopoulou lagoon (Table 5).
Q-mode cluster and nMDS analysis (Bray-Curtis similarity on square-root transformed combined replicates using the 23 most common species) of treatments from both sampling Table 4 2a and 2b, Fig 2), containing results of treatments from the autumn samples from Chalikiopoulou (2a) and the spring samples from Antinioti (2b) respectively. Sampling site and sampling season both had significant effects on assemblage compositions ( Table 6). The use of different substrata did not significantly affect the assemblage composition. When analyzed separately, only sampling season was significant for differences in both Chalikiopoulou and Antinioti (Table 6). No significant effects could be observed for subsamples or substrata (Table 6).   Table 2.

Comparison of the in situ and experimental assemblages
Experimentally grown assemblages differed from in situ assemblages (Fig 4). While both stained and dead in situ assemblages were dominated by hyaline taxa, many experimental assemblages contained more porcellaneous and agglutinated taxa. ANOSIM analysis of the wall structure distributions showed significant differences between in situ dead and experimental assemblages (R = 0.74, p<0.001) as well as in situ stained and experimental assemblages (R = 0.48, p<0.001). In situ stained and dead assemblages were also different from each other (R = 0.24, p = 0.024), but with a lower significance.
Of the 111 benthic species that were found in the in situ samples from Chalikiopoulou lagoon, 63 species also grew in experimental treatments. Twenty of the 22 stained species encountered from the in situ samples also grew during experiments; only Brizalina ? sp. 1 and Conorbella patelliformis were not found in the experimental assemblages. On the other hand, 19 species grew within experimental treatments from Chalikiopoulou lagoon that were completely absent from the in situ assemblages. Of those, 6 taxa contributed with more than 1% to experimental assemblages within at least one treatment: Adelosina carinatastriata,   In samples from Antinioti lagoon, 61 species were found within the in situ assemblage and 43 of those species also grew in the experimental treatments from that site. Of the 20 stained species from the in situ samples, 17 species also grew during the experiments and only Brizalina ? sp. 1, Bulimina costata and Pseudotriloculina jugosa did not. Within experimental assemblages, 25 species were documented that did not occur within any of the in situ assemblages from Antinioti. Seven of those species contributed more than 1% in a least one treatment: Ammobaculites sp. 1, Bolivina pseudoplicata, Cymbaloporetta plana, Pseudotriloculina laevigata, Quinqueloculina cf. Q. laevigata, Textularia bocki, and Triloculina schreiberiana.
Of the 8 species that grew most abundantly in the experimental treatments, Ammonia tepida, Haynesina depressula, Pseudotriloculina cf. P. oblonga, Pseudotriloculina rotunda, and Quinqueloculina seminula were also present within stained in situ assemblages from both sites. Miliammina fusca was present in both stained and dead in situ assemblages from Antinioti, but completely absent from all in situ assemblages of Chalikiopoulou lagoon. Textularia bocki Effects of site, season, and substrate on foraminiferal assemblages grown from propagule banks and Rosalina bulloides were absent from the stained assemblages of both sites and T. bocki was also absent from the dead assemblage of Antinioti.
Within in situ and grown assemblages from Chalikiopoulou, 46 species were defined as autochthonous, 65 as sporadic, and 19 as allochthonous (S4 Table). In Antinioti, we found 37 autochthonous, 24 sporadic, and 25 allochthonous species (S4 Table). Autochthonous species were further divided into those that were present in situ and either did or did not grow during experiments (group 1a and 1b). Sporadic species were also divided into 2 subcategories depending if they did or did not grow during experiments (group 2a and 2b). Allochthonous species represented taxa that were only present within experimental assemblages but did not occur within in situ assemblages from the respective sampling sites (group 3).
Within in situ assemblages from Chalikiopoulou, 89-91% of all individuals were autochthonous (Fig 5A), although the relative abundances of autochthonous species varied between 55-63% ( Fig 5C). In Antinioti, 93-96% of individuals and 70-76% of species were autochthonous (Fig 5B and 5D). The number of autochthonous species that did not grow during experiments (group 1b) was higher at Antinioti than Chalikiopoulou (Fig 5C and 5D). Differences between both sites were significant (ANOSIM R = 0.95, p = 0.03 for individuals and R = 1, p = 0.03 for species). Among the sporadic taxa, the relative abundances of group 2b that did not grow during experiments were higher than sporadic species that did grow (group 2a, Fig 5).
Within experimental assemblages, the abundances of autochthonous individuals were still relatively high, although more variable with 53-91% in Chalikiopoulou ( Fig 5A) and 30-97% in Antinioti (Fig 5B). The percentages of species belonging to group 1a (autochthonous species that grew from experiments) were comparable to in situ assemblages (Fig 5C and 5D). Relative abundances of both individuals and species belonging to sporadic species that grew (group 2a) were significantly higher in experimental assemblages (Fig 5). Furthermore, individuals belonging to those species that grew during experiments but were absent from in situ assemblages (allochthonous, group 3) were detected within all treatments from both sites (Fig 5A  and 5B). In Antinioti, those individuals were especially abundant in October with 14-66% ( Fig  5B). Season was statistically significant for assemblages from Antinioti (R = 0.7, p = 0.002).
For the analysis of ecological groups based on the Foram-AMBI approach, we only grouped species that were mentioned by name in the list of Jorissen et al. [9], leading to relatively high percentages and to a high variability of unassigned taxa (Fig 6, S5 Table). We found 31 species belonging to Group 1, 17 species belonging to Group 2, 9 species belonging to Group 3 and only 1 species (Ammonia tepida) belonging to Group 4. We did not find species from Group 5. We plotted the percentages of Foram-AMBI groups among individuals (Fig 6A and 6B) and species (Fig 6C and 6D).
The in situ assemblages of Chalikiopoulou and Antinioti lagoon contained relatively high percentages of individuals belonging to Foram-AMBI Group 1 with 36-45% in Chalikiopoulou and 15-21% in Antinioti (Fig 6A and 6B). Percentages of Groups 2 and 3 were comparable between sites, while Antinioti assemblages contained more Group 4 specimens (Ammonia tepida) with 24-34% compared to 8-16% in Chalikiopoulou (Fig 6A and 6B). Differences between sampling sites were statistically significant with regard to specimens (ANOSIM, R = 1, p = 0.03). However, when the distribution of in situ species was analyzed, the sites were not statistically different (p = 0.06, Fig 6C and 6D). Between 22 and 31% of all species were assigned to Group 1, while only 1-2% of all species belonged to Group 4 (only Ammonia tepida, Fig 6C and 6D).
The distributions of the four Foram-AMBI Groups within the experimentally grown individuals did not differ significantly between Chalikiopoulou and Antinioti (p = 0.14; Fig 6A  and 6B). Overall, high percentages of Groups 3 and 4 grew within the experimental treatments. However, differences between in situ and experimental assemblages were only significant for Distribution of relative abundances of autochthonous, sporadic, and allochthonous foraminifera within in situ and experimental assemblages. Group 1a refers to autochthonous taxa that were alive and/or common within in situ assemblages and grew during experiments. Group 1b refers to autochthonous taxa that did not grow during experiments. Group 2a refers to taxa that were sporadically present (<1%) within in situ assemblages and grew during experiments. Group 2b refers to sporadic taxa that did not grow during experiments. Group 3 refers to allochthonous taxa that were absent from in situ assemblages but did grow during experiments. Effects of site, season, and substrate on foraminiferal assemblages grown from propagule banks Chalikiopoulou (R = 0.84, p<0.001; Fig 6A). When examined separately, individuals of both Chalikiopoulou and Antinioti displayed significant differences between sampling seasons (R = 0.38, p = 0.01 for Chalikiopoulou, R = 0.3, p = 0.03 for Antinioti). In Chalikiopoulou samples, relatively few individuals from Groups 1 and 2 were found in assemblages grown in May (0.3-4%). However, Group 2 percentages increased considerably in the experimental assemblages grown in October (Fig 6A). In Antinioti samples, Group 1 was only rare in May experimental treatments (0.1-0.5%) but increased strongly in October (Fig 6B). Subsamples and substrates did not differ significantly. With regard to species distribution (Fig 6C and 6D), only Chalikiopoulou showed a significant difference with regard to sampling season (R = 0.39, p = 0.02). Percentages of Group 1 species were higher in May, while percentages of Groups 3 and 4 species increased in October (Fig 6C).
We further separated the assemblages into six "functional groups" according to their known mode of life (S5 Table). We found species within each of the functional groups: 24 species were epifaunal (predominantly on sediment), 34 species were infaunal (predominantly in sediment), one species belonged to epiphytic a (predominantly sessile), 16 species to epiphytic b (temporary mobile), 12 to epiphytic c (predominantly mobile) and 52 species to epiphytic d (permanently mobile).
The in situ assemblages included specimens from all groups except for epiphytic a. Most were epifaunal with 30-40% in sediments from Chalikiopoulou and 66-78% in those from Antinioti (Fig 7A and 7B). Percentages of infaunal specimens were comparable between the two sites, whereas all epiphytic groups were more common at Chalikiopoulou (Fig 7A and  7B). In situ assemblages differed significantly between sampling sites (ANOSIM, R = 1, p = 0.03) and between relative abundances of species belonging to the functional groups (R = 0.97, p = 0.03), with Antinioti showing overall higher abundances in epi-and infaunal species, but lower abundances in epiphytic c and d (Fig 7C and 7D).
The experimental assemblages were characterized by significantly higher percentages of epiphytic d specimens in addition to overall lower percentages of epiphytic b and c as well as infaunal specimens compared to in situ assemblages (Fig 7A and 7B). Differences between in situ and experimental assemblages were statistically significant for both sites (R = 0.95, p<0.001 for Chalikiopoulou, R = 0.43, p = 0.009 for Antinioti). However, there were no significant differences between sampling sites; epifaunal specimens were common to dominant in both sites. Experimental assemblages from Chalikiopoulou did not show significant differences between season and substrate type. However, subsamples 17 and 59 showed higher percentages of epiphytic c species than subsamples 16 and 60, which highlighted a difference between local subsamples (R = 0.28, p = 0.04). In Antinioti, experimental assemblages grown in October had lower percentages of epifaunal specimens and more epiphytic d specimens (ANOSIM for season: R = 0.43, p = 0.003). As in Chalikiopoulou, there was no statistically significant difference between substrate types. With the relative abundances of species belonging to the functional groups, there was a significant difference between assemblages grown from Chalikiopoulou and Antinioti (R = 0.63, p<0.001), with more epifaunal species and fewer epiphytic c species in Antinioti (Fig 7C and 7D). For both sites, differences in experimental and in situ compositions were significant (R = 0.39, p = 0.02 for Chalikiopoulou, R = 0.89, p<0.01 in Antinioti), but season was only significant for samples from Antinioti (R = 0.51, p = 0.002)

In situ assemblages
Foraminiferal assemblages in the Chalikiopoulou and Antinioti lagoons of Corfu Island are generally dominated by hyaline taxa such as Ammonia spp., Haynesina depressula, Asterigerinata mamilla and small porcellaneous miliolids such as Quinqueloculina seminula and Pseudotriloculina spp. As such they are similar to those found in other shallow-water lagoon or sheltered habitats comparable to the studied sites (e.g., [34,[61][62][63][64]). Diversity values of the studied sites are also comparable to shallow-marine areas of the Greek mainland (e.g., [39,65]), and the assemblages reveal a composition that has previously been described from the inlet areas of other lagoonal sites such as the Gulf of Kalloni (Aegean Sea, [34]) or Amvrakikos Gulf (Ionian Sea, [65]). From the latter, it was suggested that the prevalence of miliolids and epifaunal taxa could be attributed to adjacent seagrass and algal vegetation and local hydrodynamic regimes [65], which might also be a factor in the Corfu lagoons, whose shores are extensively vegetated.
The high abundance of planktonic taxa, especially in the in situ assemblages (21-71%) is surprising. Some shells appeared worn, abraded, and "old looking", suggesting a fossil source. Deposits of planktonic foraminifera are known from Pliocene formations of the island (e.g., [66]), and eroded individuals could have been transported into the lagoons. However, other planktonic specimens appeared more pristine and included a few stained individuals, suggesting a more recent origin. This is particularly surprising, because neither lagoon entrance opens directly into the Ionian Sea, but rather into the Corfu Channel between the island and the mainlands of Albania and Greece. The width of the channel varies between 2 and 22 km and the water depth does not exceed 100 m (portal.emodnet-bathymetry.eu). However, the north-flowing surface current along the Greek coast is probably sufficient to transport small and/or empty planktonic shells from the open sea into the Corfu Channel and subsequently into the lagoons.
Some transport of empty benthic foraminiferal tests into both study sites also likely occurs. The local assemblages consisted of <10% living (stained) individuals. The numbers of species found alive were comparable at both sites (22 and 20 respectively), while 89 species found in Chalikiopoulou and 41 in Antinioti were only present as empty shells. Some of them were abundant and might represent time-averaged accumulations resulting from seasonal blooms. However, the abundance of sporadic species, with rare occurrences in the dead assemblages, were higher in Chalikiopoulou than Antinioti, highlighting the more restricted nature of the latter. While large storm events are uncommon on the eastern coast of Corfu [46], it is possible that specimens are frequently transported into the lagoons from adjacent bays or the deeper channel. This can occur either in suspension or on dislodged seagrass leaves, especially during winter [13,67].

Experimentally grown assemblages
The propagule banks of both lagoons contain a high diversity of autochthonous and allochthonous propagules, as evidenced by the 88 species that grew in experimental treatments. Diversity was higher in assemblages that grew from Chalikiopoulou sediments, suggesting that the more open nature of the lagoon facilitates propagule dispersal. Further, the taxonomic composition of the assemblages grown from Chalikiopoulou and Antinioti sediments also differed significantly. Previous studies have already demonstrated the influence of different study sites on the resulting experimental assemblages (e.g. from coastal areas off Georgia and Florida [20]). However, those study sites differed significantly in terms of temperature and salinity. These variables were much more comparable between Chalikiopoulou and Antinioti lagoons, except for the lower salinity values at Antinioti in October 2017 (25 instead of [39][40]. The main difference between the two sites appears to be the greater restriction of Antinioti. An effect of site exposure on species richness has previously been described [12]. Although the composition of experimentally grown assemblages differed significantly between the two sites, within the same sampling site, no significant difference was detected between the two subsamples analyzed. This suggests that specimens in propagule banks might be less patchy and more evenly distributed, which would correspond to their "dynamic" nature and the overall strong indications of propagule dispersal. Both cluster analysis and nMDS plot revealed a significant effect of seasonality, specifically the time of sampling, indicating that propagule banks may vary throughout the year. They are probably influenced by species-specific reproductive cycles, which can be quite variable among taxa [68,69]. Those are not necessarily bound to annual cyclicity [40,41], but reproduction can be triggered by other factors, such as a substantial increase in food availability (e.g., [70]). In another example, Adelosina carinatastriata exhibited maximum abundances in samples collected during autumn from the Atlantic French coast, suggesting that a large reproduction event occurred during the warmest times of that particular year [71]. In addition, it is likely that seasonal changes in dispersal vectors, such as currents strengths or weather, have an effect on the local propagule assemblages. Differences in the seasonal colonization of natural and artificial seagrasses have previously been described by Ribes et al. [68].
The simulated substrates did not significantly influence the foraminiferal assemblages that developed in the treatments, based on either abundance, diversity or assemblage composition. In case of the artificial substrates (both phytal and rubble), the foraminifera may have rejected both additional substrates and therefore assemblages did not differ from those grown only in the mud or they did not migrate from the muddy substrate to the artificial leaves or rubble. Selectivity in the colonization of artificial substrates was previously demonstrated in an experiment using deep-sea assemblages, where the studied species revealed a preference for one or more types of artificial substrate over others [72]. Ribes et al. [68] showed that epiphytic foraminifera settled preferentially on certain types of artificial substrates. However, they found no significant differences between artificial and natural plant material [68]. Successful settlement on artificial hardgrounds has also been demonstrated (e.g., [73]), although both studies were conducted over longer time periods [68,73]. On the other hand, variations in temperature and salinity have been shown to strongly influence assemblages grown from propagule experiments [12,20,24]. The absence of a significant effect of different simulated substrates on the assemblages grown in our study could suggest that for early life stages, water variables might be more important for the growth and development of foraminiferal specimens.
Within experimental assemblages from both sites and seasons, specimens of Rosalina bulloides and Cymbaloporetta spp. were found displaying inflated last chambers (i.e., float chambers). Meroplanktonic stages have previously been described for both genera (e.g., [74,75]) and they are associated with gametogenesis [75,76]. When asexually produced megalospheres of Rosalina (Tretomphalus) bulloides reach maturity, they produce float chambers and assume a temporary pelagic life-style before gamete release [76]. This adaptation of the life-cycle has also later been described from other taxa such as Cymbaloporetta spp. [75]. Many of the specimens with a float chamber found in the present study were empty, suggesting successful gametogenesis during the course of the experiment. The presence of specimens with float chambers at the termination of the experiment suggests the presence of megalospheric juveniles within the original fine fraction of sediment. It is also possible that the megalospheres originated from microspheric propagules, although it cannot be determined if the duration of the experiment provided enough time for the completion of the life cycle [74]. Alve and Goldstein [10] suggested that propagules can be either micro-or megalospheric, although microspheric individuals were more common.

Comparison of the in situ and experimental assemblages
Experimental assemblages differed significantly from in situ assemblages at both sites. Differences between assemblage compositions could be assessed by analysis of allochthonous taxa, which were absent from the in situ assemblages but grew during experiments. Between 16 and 46% of the experimentally grown species were deemed allochthonous. As such, they represent taxa that were previously "hidden" from the in situ assemblages and could only have been present in the fine fraction of the sediment as small juveniles or propagules. These species included agglutinated taxa such as Textularia bocki (allochthonous only in Antinioti) and Miliammina fusca (allochthonous only in Chalikiopoulou), which were among the most commonly grown species during the experiments. Miliammina fusca is well known from higher estuarine or marsh environments [61,77] and has previously been found to grow during propagule experiments from Georgia and Florida [12,20,24]. It has also been described from marginal areas of the Kalloni Gulf in eastern Greece [32]. As such, small juveniles or propagules could possibly have been transported from the vegetated marginal areas deeper in the lagoon towards the inlet in Chalikiopoulou. Its presence within in situ samples from the Antinioti lagoon suggests that it can occur within the vegetated areas of the Corfu lagoons and is able to tolerate high salinity conditions (at least up to 40 ppt). Textularia bocki has been described as an infralittoral or circalittoral taxon in the central Tyrrhenian Sea [78], indicating that its propagules have been transported from outside the lagoon, comparable to findings reported from the US coast [20,24]. Other allochthonous taxa include several species of small, epiphytic miliolids, which possibly originate from the extensive Posidonia meadows along the coast of Corfu Island, where those species have been found (own observations and [79]).
Many species, which were classified as "sporadically present" in the in situ assemblages at both sites were absent from the experimentally grown assemblages. On the other hand, the latter revealed significant percentages of allochthonous taxa not found in the >63 μm fractions at the study sites. This suggests that many sporadic species of the in situ assemblages could have been transported into the lagoons as adults, for example, in suspension or attached to plant debris [13,14,67]. The presence of "exotic" taxa belonging to the allochthonous group 3 (S4 Table) suggests that propagules of different species were transported into the lagoons and grew only under experimental conditions. The more restricted nature of Antinioti lagoon is probably responsible for the lower abundance of sporadic and very rare taxa within the in situ assemblages as well as the overall lower diversity. However, the calmer conditions in the more sheltered environment of Antinioti might be advantageous for the settlement of small exotic propagules, which could be an explanation for the high percentages of allochthonous individuals grown from the October samples. At the same time, the number of individuals >63 μm were almost twice as high as in May, which could suggest that the "fine"assemblages were relatively impoverished in autochthonous taxa, which might in turn have resulted in the relatively higher percentages of allochthonous specimens. This would be another argument for the strong influence of seasonality on the composition of local propagule banks.
The ternary plot further highlighted differences in assemblage compositions between in situ and experimental assemblages, which has also been observed in previous studies from the US coast [20,24]. In those studies, differences in salinities during experiments were interpreted to have been one of the main reasons for assemblage differences between treatments [20,24]. In our study, the known environmental differences between in situ and experimental conditions were slightly higher temperatures during experiments and differences in substrate. While the three variations in substrate type (muddy, phytal, rubble) did not result in significant differences among assemblages, the restriction towards the size fraction <53 μm could have influenced assemblages grown during experiments. The organic content within sediment is often higher within the fine fractions such as mud or silt, since organic matter is often adsorbed on or within clay minerals (e.g., [5,9]). As such, using this fine fraction in our experiments could have increased organic enrichment within the experimental treatments.
Our analysis of the distribution of ecological groups based on Foram AMBI within all assemblages showed that most of the experimental assemblages contained higher percentages of groups 3 and 4, which generally react positively to higher organic content [9]. This was mostly observed on the individual level, as the differences were less distinct on the species level (although significant for Antinioti). While we did not encounter first order opportunists (group 5), the significant increase in enrichment-tolerant specimens in Chalikiopoulou treatments suggests an increase in organic content compared to the original sediment composition. Differences in Antinioti lagoon were less distinct, which might be because the original sediment from that site was already more enriched in organic material than Chalikiopoulou due to its lower energy and more restricted circulation conditions (see sample description). Nevertheless, the percentages of group 3 increased during experimental growth. Interestingly, at both sites the percentages of sensitive and indifferent individuals (groups 1 and 2, S5 Table) were higher in experimental treatments grown in October (e.g., Adelosina carinatastriata, which has also been described to occur in fine sediments; [39,71]). This might reflect seasonal variations in organic content, with a potentially higher accumulation during winter months, perhaps due to more land-derived input, and higher rates of consumption over summer. We also found more living individuals >63 μm within in situ assemblages sampled in October, which mainly include Ammonia tepida. The higher abundances could have led to more organic material being consumed and thus provided better conditions for the sensitive or indifferent species.
The increased organic content of the experimental conditions did not lead to a significant decrease in oxygen levels during the experiments, which never dropped below 183 μmol/kg. However, together with the increased temperatures, more organic matter would favor group 3 and 4 taxa, with many of them also exhibiting fast growth and reproduction rates [5]. As such, it appears reasonable to assume that they were able to accumulate much faster during the course of the experiment (6 weeks) compared to taxa which might be sensitive to organic content or have slower growth rates. As an example, Quinqueloculina seminula, which was among the most common species that grew in our experiments, has previously been described as an early colonizer due to its temporarily opportunistic behavior in terms of growth and reproduction (e.g., [5,80]). Together with Ammonia tepida, it has been deemed as the primary pioneer in paralic environments [80].
Analysis of the functional groups also revealed higher percentages of permanently mobile specimens belonging to the epifaunal and epiphytic d groups. Taxa belonging to the latter group are often associated with plant rhizomes and the surrounding muddy sediments [81]. They can also reproduce within the sediment and, unlike most other epiphytic groups, they have short life-spans [68], which could explain their success within the experimental conditions that featured mostly fine sedimentary substrates, enriched with organic material. Of the eight most abundant species grown from the experiments, all were grouped within either epifaunal or epiphytic d, except Rosalina bulloides (epiphytic b). Most epiphytic species are sensitive to eutrophication, although they are less effected by organic accumulation related to the presence of plant debris in seagrass meadows [5]. Since algae, other plant material, and bacteria were probably the main food sources within the experimental conditions, these might favor the growth and development of epifaunal and epiphytic d groups that exhibited the highest abundances within our experimental assemblages.

Conclusions
In our study on growth experiments on foraminiferal assemblages originating from fine sediments from two shallow-water lagoons of Corfu Island, we gained new insights into the assemblage composition and dynamics of local propagule banks, specifically: 1. Chalikiopoulou and Antinioti lagoons contain diverse but distinct propagule banks, which are influenced by the availability and viability of allochthonous taxa. Within each site, the distribution of small juveniles and propagules appeared to be more uniform (less patchy) than that of adult assemblages.
2. The time of sampling had a significant effect on grown assemblages, suggesting a seasonal influence on the composition and dynamics of local propagule banks, which is probably related to species-specific reproductive cycles and dispersal mechanisms.
3. Although different simulated substrates did not reveal a significant impact, the fine, organic-enriched sediment fraction used in the experiments appeared to influence the resulting foraminiferal assemblages. Together with differences in temperature and salinity, sediment quality (including organic content) may be a key factor in the differences in assemblage compositions between in situ and experimental assemblages. Future experiments that include different sediment compositions could further elucidate this relationship.
Our results shed new light on the structure of foraminiferal propagule banks, their role in local species pools, and the potential responses of foraminiferal communities to ongoing local and global environmental change.
Supporting information S1