The Holocene History of the Diatom Community in a Small Water Body on Shemya Island (Aleutian Arc, USA): The Inﬂuence of Global and Local Environmental Changes

: A diatom analysis of a peat deposit from Shemya Island (Aleutian Arc, USA) is performed, and the dynamics of the diatom community are described. According to the radiocarbon dating, the formation of the deposit began 9300 cal. years BP. Principal component analysis made it possible to relate the dynamics of the diatom community to certain environmental conditions and the factors that inﬂuenced the coastal ecosystems during its formation. The following factors (predictors) were considered: the inﬂuence of age, zoo- and anthropogenic effects, and changes in climatic conditions. Sea level change was the main driver of the diatom community in the studied water body having a continuous direct and indirect inﬂuence on the studied small water body, i.e., by bird colony formation and more humid and coastal conditions. Since 3000–2000 cal. years BP, the anthropogenic factor (hunting depression of the bird colony) also became signiﬁcant. During the whole water body lifetime and following peat formation, the diatom community was inﬂuenced by groups of factors: global factors (e.g., sea level rise) caused gradual change of local factors, which resulted in smooth shifts in community. In contrast, local factor inﬂuence (bird colony rise and fall due to human activity) caused abrupt and transient shifts. We can hypothesize that the relatively stable global environmental conditions in the Late Holocene were an auspicious background to see abrupt changes due to inﬂuence of the zoogenic and anthropogenic factors. We believe that further works on the material from other islands will make it possible to form a general picture of changes in the diatom communities in the Holocene and interpret it in connection with climatic changes in the region. of Chamaepinnularia spp., Luticola spp., Staurosira spp., and Staurosurella spp.—the taxa inhabiting moist soil and mosses. It did not show a linear time trend ( R 2 = 0.0001, F 1,14 = 0.002, p = 0.97), peaking at 6000–8000 cal. years BP and at the present time, and ambient signif-icantly decreased with an increase in δ 15 N ( R 2 = 0.29, F 1,11 = 4.5, slope = − 0.54 ± 0.25, p = 0.057) and was strongly and negatively related to the decomposition rate ( R 2 = 0.41, F 1,9 = 6.2, slope = 0.64 ± 0.26, p = 0.03).


Introduction
A great number of palaeoecological publications of the two last decades have been focused on the whole Holocene period with special attention to human-induced transformations of ecosystems in the second half of the Holocene [1][2][3][4][5]. The Aleutian Arc is an area where such studies have intensified over the past two decades on several large and small islands [6][7][8][9][10]. The terrestrial communities of the Aleutian Islands represent a good model for palaeoecological reconstructions for several important reasons.
First of all, the Islands were all formed by volcanic and tectonic activity [11,12]. Terrestrial and freshwater communities of each island have originated by colonization from the continent or from the adjacent islands and are simplified as compared to the mainland communities [13,14] as commonly occurs in relatively small islands [15]. The sea level in the region has been rising during the Holocene, and we expect its strong influence on the island communities. Volcanic eruptions during different periods of the Holocene also had a significant impact on the flora and fauna of these islands, and their influence was recorded even in localities distant from eruption sites [16][17][18]. Volcanic activity was more intense in the eastern part of the Aleutian Arc, while more western islands were less vulnerable to its impact [11]. Finally, climatic changes in the region strongly influenced the biota through changes in the annual temperature and humidity, storm activity, snow cover in winter, etc. [19,20].
Along with global factors (volcanism, sea level, and climate changes), local biotic factors have strongly influenced the flora and fauna of the Aleutian Islands. Except for a few islands closest to Alaska and Kamchatka, the Aleutians are lacking top terrestrial mammal predators [21][22][23], which favored colonization of islands by seabirds forming great coastal colonies [22,24]. In turn, the activity of birds (active guano deposition) influenced the chemical composition of soil and water, and the structure of islands' flora [8,[25][26][27][28]. Since the beginning of the Early Holocene, the Aleutian Islands began to be colonized by ancient Aleuts, whose activities influenced the local flora and fauna, but not so crucially as people colonization in Europe [29]. Humans colonized different islands at different times from the East, and the seabirds were among their main prey [10,22,30].
Thus, the terrestrial and aquatic biota of each island was influenced by a specific combination of local and global environmental factors, and a comparison of different islands can be helpful for understanding the influence of environmental changes in the Holocene on island biota.
A relatively small island of the Aleutian Arc, Shemya Island, where several archeological sites were detected [31], has attracted the attention of the investigators of the terrestrial communities and their dynamics during the Holocene. Savinetsky et al. [8] analyzed a peat deposit located at the sea shore of this island. They revealed a significant δ 15 N enrichment of layers formed at 4700-2800 years BP due to a high seabird guano deposition (due to global sea level rise and seashore approaching to the water body that produced peat). Then, a sharp decline in δ 15 N coincided with the first appearance of human settlement on the island and reflected a drastic reduction in seabird abundance. Therefore, the terrestrial fauna of Shemya Island was greatly influenced by a combination of factors specific for this island due to its small size and position in the Aleutian Arc.
In contrast to terrestrial communities, the Holocene history of the freshwater communities in the Aleutians is poorly studied [32]. The remains of several invertebrate (Cladocera, Copepoda, Ostracoda, Rotifera) and algal (cyanobacteria, diatoms) groups are well preserved in the lake sediments and can be used efficiently to reconstruct the Holocene history. In particular, diatom analysis is widely recognized as a useful tool to indicate past environmental changes and to study the effects of environmental factors on the dynamics of the water-body communities. Diatoms, modern and fossil, are ideal proxies allowing assembly of training and calibration sets for different geographical regions and a wide range of purposes [5,[33][34][35][36][37]. The diatoms of the Aleutian Islands are poorly studied, although some studies have been conducted on the diatoms of Adak Island [32,38] or the Carlisle Islands [39]. Species composition of Bering Island (Commander Islands) water bodies was also investigated [40]. Fossil diatoms of the Shemya peat deposit were studied previously in a taxonomic aspect [41,42], but no studies on diatom (or any other freshwater) community dynamics in the Holocene on the small Shemya Island have been conducted up to date.
This work is aimed to elucidate the dynamics of the diatom community of a small water body located in Shemya Island during the Holocene and to identify the main global and local factors driving these changes.

Study Area and Sampling
Shemya Island is a relatively small (its total area does not exceed 15.3 km 2 ) island in the Near Islands group of the western part of the Aleutian-Commander Islands chain ( Figure 1). Unlike most other islands of the Aleutian Arc, it has no volcanoes. In 1999, a peat deposit with a total depth of 385 cm (lab name "McDonald Point") was sampled from the ledge protruding into the sea near Fox Beach in the southeast end of the island (52 • 42 19 N, 174 • 09 11 E) by a joint expedition of the IEE RAS (A.B. Savinetsky, N.K. Kiseleva, B.F. Khasanov) and the US Western Aleutians Archaeological and Paleobiological Project (WAAPP) [6]. The samples were collected from 76 layers (each with thickness of 3-5 cm). [8,43] and Smyshlyaeva et al. [27] (see Table S1). Additional information about la and "Bchron" package calculations can be found in Table S2 [44,45]. According diocarbon dating, the formation of the deposit began at 9300 cal. years BP. The a layers was estimated using previous radiocarbon dating data [8], and, as e showed a strong and reliable correlation with the layer depth (r = 0.97, p < 0.000 16). Some layers were represented by sand masses with little organic matter and on the nitrogen content were available for them.
Based on previous studies [8,25,28], we assumed that δ 15 N (a difference in 15 N/ 14 N rate from that in atmospheric N2) reflects the strength of the zoogenic ef ciated with the abundance of avian guano inputted into the soil. The degree of composition was calculated by Kiseleva et al. [46]. The degree of decomposition o material is supposed to reflect climatic patterns-its high values indicate warm tions [46][47][48], but this correlation could be complicated by different factors [49], e for volcanic islands. We tested the possibility of its application as an additional the diatom-based reconstruction.

Diatom analysis
The samples were treated in the laboratory following the hot peroxide me scribed by Kelly et al. [50]. Permanent slides were made using aniline-forma The lithological description of the deposit, results of the calibrated radiocarbon dating, and data on δ 15 N (sometimes with gaps) have been published by Savinetsky et al. [8,43] and Smyshlyaeva et al. [27] (see Table S1). Additional information about layers age and "Bchron" package calculations can be found in Table S2 [44,45]. According to the radiocarbon dating, the formation of the deposit began at 9300 cal. years BP. The age of the layers was estimated using previous radiocarbon dating data [8], and, as expected, showed a strong and reliable correlation with the layer depth (r = 0.97, p < 0.000001, N = 16). Some layers were represented by sand masses with little organic matter and no data on the nitrogen content were available for them. Based on previous studies [8,25,28], we assumed that δ 15 N (a difference in ‰ in the 15 N/ 14 N rate from that in atmospheric N 2 ) reflects the strength of the zoogenic effect associated with the abundance of avian guano inputted into the soil. The degree of peat decomposition was calculated by Kiseleva et al. [46]. The degree of decomposition of organic material is supposed to reflect climatic patterns-its high values indicate warmer conditions [46][47][48], but this correlation could be complicated by different factors [49], especially for volcanic islands. We tested the possibility of its application as an additional proxy in the diatom-based reconstruction.

Diatom Analysis
The samples were treated in the laboratory following the hot peroxide method described by Kelly et al. [50]. Permanent slides were made using aniline-formaldehyde mount medium (n = 1.67-1.68). The slides were examined under high-power light microscopes Leica 2500 and Leica DM500 at 1000× magnification and under scanning electron microscopy using a Jeol JSM-6380 microscope. For the scanning electron microscopy investigation, drops of cleaned material were air-dried on pieces of aluminium foil, mounted on brass stubs with double sided carbon tape, and coated with Au-Pd in a Giko IB-3 ion coater.
In total, 80 taxa and variations of diatoms were recorded (Table S3). The diatoms were classified into ecological groups according to Van Dam et al. [51], Lange-Bertalot et al. [52], Spaulding et al. [53], and Guiry and Guiry [54]. Proper diatom taxon names were verified according to Guiry and Guiry [54]. Light microscope (LM) and scanning electron microscope (SEM) images of dominant species are represented in Figure 2.
The composition of the diatom community and the abundance of different species for each layer were estimated from the number of valves counted in five slides. For some layers, we counted less than 300 valves in the five slides [34]. In these cases, if neighboring poor layers were similar in their species composition and belonged to similar stratigraphic layers according to Savinetsky et al. [8], we merged them with each other to have 300 or more valves for each of the resulting combined layers. Especially poor (="empty") layers (usually associated with mostly sand intercalations) were excluded from the analysis. After merging neighboring valve-poor layers and excluding "empty" layers with no diatoms found, 16 combined layers were included in the analysis (Table S1). For each combined layer, δ 15 N and degree of decomposition of plant material were calculated as the average of estimates for all initial layers. The mean age of each combined layer was determined according to Table S2.
Most of the diatom species found here are widely distributed taxa, which allowed us to use the data on their ecological preferences known from previous studies [51,52,[54][55][56]. The taxa included in the analysis were selected and grouped according to the following scheme: (1) abundant indicator species with specific ecology were included as separate taxa; (2) ecologically similar species from the same genera were combined at the genera level and in one case (close genera Staurosira and Staurosirella) in an artificial supra-genera group; and (3) rare taxonomic groups for which <30 valves were detected over the entire deposit were excluded from the analysis. As a result, 17 groups of taxa (6 individual species, 10 genera-level groups, and 1 supra-generic group were used for the analysis (Table 1, Figure 2)). Unfortunately, different taxa in these groups have mainly different trophic preferences (see [51]), and such data could not be used in the analysis of trophic conditions in the water body. Tilia 2.6.1 and Tilia-Graph software [57] were used to plot a diagram of the layer-bylayer distribution of the taxa of the diatoms and to perform the stratigraphically constrained incremental sum of squares cluster analysis (CONISS) to identify the stages of the diatom community dynamics. The stratigraphic zones were visually identified using the results from CONISS [58]. Table 1. Species and taxonomic groups of diatoms and their ecological preferences (according to [51,54,55,[59][60][61]) included in the analysis.   To describe the dynamics of the diatom community, we assessed the taxonomic diversity and relative abundances of the 17 taxonomic groups included in the analysis for each of the 16 layers. As a measure of taxonomic diversity, we used Shannon's diversity index, (H') [62]. We assessed the relative abundances of taxonomic groups as the number of valves of each group divided by the total number of valves of all taxa found in the layer. The latter varied slightly (from 300 to 306), which made it possible to obtain adequate comparable estimates of the relative abundances of taxa in different layers. To reduce the dimensionality of correlated measures of the relative abundances of 17 taxa, we used a principal component analysis. The first three components, which explained 65.8% of the total variation in taxa abundances, and Shannon's H' were all normally distributed (Shapiro-Wilk test: p > 0.05) and used in GLMs (generalized linear models) as dependent variables to assess the effects of the (1) age of the layer, (2) nitrogen content, and (3) decomposition rate (factors) on their variation. Accordingly, we constructed three models with a full set of four dependent variables and one factor in each.

Environmental Factor Dynamics
The decomposition rate was fluctuated strongly through the time (from 4 to 60%). Decomposition rate values, fluctuated in the range of 4-25.7%, before Zones changing, reached a maximum at approximately 265 cm (~4800 cal. years BP) (Figure 3a). Then, the decomposition rate declined, not reaching its minimum values, but stopping at 20-31%. No data were available on decomposition rate in the two uppermost layers (0-6 cm).

Environmental Factor Dynamics
The decomposition rate was fluctuated strongly through the time (from 4 to 60%). Decomposition rate values, fluctuated in the range of 4-25.7%, before Zones changing, reached a maximum at approximately 265 cm (~4800 cal. years BP) (Figure 3a). Then, the decomposition rate declined, not reaching its minimum values, but stopping at 20-31%. No data were available on decomposition rate in the two uppermost layers (0-6 cm). , and diatom community parameters: first three principal components (c-e) and Shannon Index (f). PC1 is associated with cold-water and terrestrial diatoms, PC2 with benthic and epiphytic diatoms, and PC3 with taxa inhabiting moist soil and mosses (see Tables 1 and 2). Dashed line marks the border between Zone 1 and Zone 2 (at 265 cm), while areas of gray color mark layers of sandy content with no diatom valves found.
The δ 15 N (Figure 3b) began to accumulate in the peatland soil at ~8800 cal. years BP (~350 cm) and, with accelerating pace, reached high values at ~4800 cal. years BP (~270 cm), and a peak of concentration at ~3900-3800 cal. years BP (235-245 cm, Figure 3b). Then, , and diatom community parameters: first three principal components (c-e) and Shannon Index (f). PC1 is associated with cold-water and terrestrial diatoms, PC2 with benthic and epiphytic diatoms, and PC3 with taxa inhabiting moist soil and mosses (see Tables 1 and 2). Dashed line marks the border between Zone 1 and Zone 2 (at 265 cm), while areas of gray color mark layers of sandy content with no diatom valves found.
The δ 15 N (Figure 3b) began to accumulate in the peatland soil at~8800 cal. years BP (~350 cm) and, with accelerating pace, reached high values at~4800 cal. years BP (~270 cm), and a peak of concentration at~3900-3800 cal. years BP (235-245 cm, Figure 3b). Then, it rapidly declined (by more than three times by 2000 cal. years BP) and continued to decrease gradually to nearly zero by the present time. No data on the nitrogen content were available for some layers-they were represented by continuous sandy layers, with extremely low concentrations of organic matter (layers 285-296, 300-305, 305-310 cm).
An abundance of Pinnularia borealis and Eunotia curtagrunowii increased in Subzone 1b (350-318 cm,~8800-7400 cal. years BP). Epiphytic forms, represented by Cavinula, Staurosira, and Staurosirella species, with Stauroforma exuguiformis as dominant, appeared. An abundance of epiphytic taxa decreased towards the top of the subzone, while an abundance of Diadesmis + Humidophila, preferring terrestrial habitats, increased, reaching its maximum. The valves of Eunotia cf. bidens, presented in the underlying layers, disappeared in this zone, while Aulacoseira spp. was observed sporadically. valves increased noticeably. At the same time, the core of the diatom community was formed by Stauroforma exiguiformis, Cavinula (mainly C. cocconeiformis f. elliptica and C. davisiae), and Encyonema (E. silesiacum and E. minutum) species. Diadesmis + Humidophila, Eunotia curtagrunowii, and Pinnularia borealis completely disappeared. A complex of species from the genus Pinnularia (including mainly P. intermedia, P. incostans, and P. arkadii) appeared at the top of this zone. In Subzone 1c (318-265 cm,~7400-4800 cal. years BP), the number of Aulacoseira valves increased noticeably. At the same time, the core of the diatom community was formed by Stauroforma exiguiformis, Cavinula (mainly C. cocconeiformis f. elliptica and C. davisiae), and Encyonema (E. silesiacum and E. minutum) species. Diadesmis + Humidophila, Eunotia curtagrunowii, and Pinnularia borealis completely disappeared. A complex of species from the genus Pinnularia (including mainly P. intermedia, P. incostans, and P. arkadii) appeared at the top of this zone.
The interval between Zones 1 and 2 (265-245 cm,~4800-4000 cal. years BP) was represented by a layer of sand where the diatom valves were absent or sporadically recorded.
Zone 2 covered 245 cm (~4000 cal. years BP to present). The dominant taxa of Zone 2 were represented by two ecologically different species complexes of the genus Pinnularia.
In Subzone 2a (245-235 cm,~4000-3900 cal. years BP), a dramatic change in the species composition and structure of the diatom complex was observed. All usual species of the underlying layers disappeared here. Pinnularia cf. islandica, Pinnularia incostans, and Pinnularia intermedia became the obvious dominants.
Subzone 2b (164-65 cm,~2000-680 cal. years BP) contained an interval without any valves (75-144 cm,~1700-690 cal. years BP). The similarity of the species composition in the layers above (65-75 cm,~700-650 cal. years BP) and below (144-164 cm) allowed us to combine them into a single zone. The zone was characterized by the dominance of Pinnularia cf. islandica, P. viridiformis, and P. inconstans. Their abundance began to decrease from the base to the top of this subzone. At the same time, Rexlowea navicularis, whose valves were found only in this subzone, and representatives of the genus Diploneis (D. krammeri, D. ovalis) were among the dominants.
The interval of 65-6 cm (~650-41 cal. years BP) was represented by a thick layer of sand without diatom valves.
Subzone 2c (6-0 cm,~41 cal. years BP-present time) was characterized by an increased species number and the presence of Pinnularia lagerstedtii among the dominants. In addition, species of the genus Chamaepinnularia (C. krookii and C. krookiiformis), inhabiting moist and aerated habitats, and species of the genus Luticola (L. arctica and L. cf. mutica), were present only in this subzone, where they reached a high abundance. The presence of Staurosira and Staurosirella spp. was also noticeable in this subzone.

Principal Component Analysis
The first three principal components (PC) explained 65.8% of the total variation in initial variables reflecting the relative abundances of 17 diatom groups and individual taxa. Analysis of the loadings of each taxon to each PC ( Table 2) allowed us to interpret their ecological meaning. PC1 (27% of the total variance) (Figures 3c and 5a,b) reflected an increase in the abundances of the cold-water taxa living in aerated water with low pH and conductivity (Eunotia, see Table 2 and Table S3) and terrestrial (=soil) species (Pinnularia borealis and P. lata). Among the acidophiles, these are Eunotia curtagrunowii, Pinnularia borealis, and P. lata, cold-water terrestrial species that also need silica available in the environment to form their frustules. In addition, PC1 was associated with an increase in abundance of Diadesmis + Humidophila, also preferring moist soils, but negatively associated with epiphytic Cocconeis, Chamaepinnularia, and Pinnularia lagerstedtii. PC1 was strongly and positively correlated with the layer age (i.e., showed a significant negative linear time trend, R 2 = 0.94, F 1,14 = 220.8, slope = 0.97 ± 0.07, p < 0.000001), weakly negatively and non-significantly correlated with decomposition rate (R 2 = 0.27, F 1,9 = 3.26, p = 0.1), and varied independently of δ 15 N (R 2 = 0.04, F 1,11 = 0.4, p = 0.5).

Diatom Community Dynamics: Environmental Effects
Short paleoreconstruction based on the diatom analysis is represented in Appendix A.
PC1 associated with the diatoms of aerated waters with a low pH and conductivity, and with terrestrial taxa, is the only PC that shows a significant negative linear time trend (Figure 3c). These ecological groups were most abundant in the lowest time Zone 1 and then disappeared. This may reflect a general trend in the abiotic changes, in particular the growing influence of the coastal conditions (i.e., the microclimate), as a result of a gradual advance of the coast towards the water body due to the global sea level rise in the Holocene [12]. Landward shift of the coastline should lead to a more humid and milder microclimate as compared to severe and drier conditions at the earlier phases of the water body development when the sea level was lower. But note that PC1 is non-significantly related to decomposition rate (Figure 5a), a possible indicator of warmer conditions [46] ( Figure 4). The local trend of PC1 was apparently caused by the global effect of the gradual

Diatom Community Dynamics: Environmental Effects
Short paleoreconstruction based on the diatom analysis is represented in Appendix A. PC1 associated with the diatoms of aerated waters with a low pH and conductivity, and with terrestrial taxa, is the only PC that shows a significant negative linear time trend (Figure 3c). These ecological groups were most abundant in the lowest time Zone 1 and then disappeared. This may reflect a general trend in the abiotic changes, in particular the growing influence of the coastal conditions (i.e., the microclimate), as a result of a gradual advance of the coast towards the water body due to the global sea level rise in the Holocene [12]. Landward shift of the coastline should lead to a more humid and milder microclimate as compared to severe and drier conditions at the earlier phases of the water body development when the sea level was lower. But note that PC1 is non-significantly related to decomposition rate (Figure 5a), a possible indicator of warmer conditions [46] ( Figure 4). The local trend of PC1 was apparently caused by the global effect of the gradual sea level rise in the Holocene on all coastal ecosystems, not just individual water bodies. In particular, the observed shift in diatom zones is consistent with vegetation transformations and the humidification of local conditions reported in other studies [27].
PC2 indicates an abrupt rise of benthic and epiphytic diatoms~7000 cal. years BP (Figure 3d) which coincides in time with an abrupt shift in community structure at the upper Zone 1 (the transition from Subzone 1b to Subzone 1c, Figure 4). Then, its abundance sharply declined at~5000 cal. years BP, at the next abrupt transition of diatom community from Zone 1 to Zone 2, which suggests that the period of its peak values (~7000-5000 cal. years BP) was the time of maximum development of the pool when it was relatively deep and had a well-developed macrophyte zone.
Macrophytes may also indicate an input of nutrients (including nitrogen) required for their growth. Moreover, "benthic and epiphytic" PC2 demonstrates positive (close to significant) correlation with δ 15 N (Figure 5a), and its rise roughly coincides with the beginning of active nitrogen accumulation, an indicator of seabird impact (Figure 3b). Thus, the development of epiphytic assemblages may indicate the introduction of additional organic material into the ecosystem by seabirds and, consequently, the development of higher plants due to a strong increase in the nutrient inflow. In parallel, the excessive nitrogen inflow depressed the community and reduced the diatom diversity as indicated by the negative correlation between δ 15 N and Shannon's index. Diversity was lowest during the peak of nitrogen inflow, the "benthic and epiphytic period" of community history (7000-4000 cal. years BP), and high before and after this period ( Figure 3). The growing influence of the seabird colony on the diatom community after 7000 cal. years BP may be related to the coastline advance towards the pool as the sea level rose [8]. In turn, the decrease of the seabird impact (after 4000 cal. years BP) may be attributed to the decline of the seabird colony due to the colonization of the island by Aleuts [8]. Thus, the "benthic and epiphytic period" of the diatom community history may be the result of a combination of abiotic, zoogenic, and anthropogenic effects. PC3 (Figure 3e), associated with an abundance of diatoms inhabiting moist soil and mosses, may inversely reflect the pool depth. High values at the bottom of the column may indicate a shallow water body if not just moist soil, while the following decrease after 7000 cal. years BP, in parallel with an abrupt increase of "benthic-epiphytic" PC2, suggests the development of the pool.
"Moss" PC3 is the only component showing a negative relationship with δ 15 N (Figure 5a)-its minimum coincides with the peak of the organic inflow as indicated by an increase in δ 15 N enrichment (Figure 5a). On the other hand, PC3 is negatively related to the decomposition rate, which could depend on the temperature (Figure 5b). The increase of the decomposition rate can be explained by a shift to more coastal (mild) conditions in the vicinity of the water body. In addition, PC3 is associated with an increase in the abundance of epiphytic taxa. Decline of epiphytic and soil diatoms after 6000 cal. years BP and their low abundances in Zone 2 can also indicate wetter conditions due to a shift of the coastline towards land, unfavorable for soil-and moss-living diatoms.

Environmental Factors as Drivers of Gradual and Abrupt Changes in Diatom Community Y
The presence of thick layers of sand could be, theoretically, a trace of tsunamis, which are quite frequent in the Aleutian region [67,68]. However, we found only two valves of truly marine species (Arachnodiscus sp. and Cocconeis cf. scutellum in layer 210-215 cm) in the entire deposit. Such a small number of marine valves allows us to exclude the possible influence of tsunamis on diatom community dynamics. Volcanic ash or other traces of volcanic activity were not found in the studied peat deposit [8,27,43] and, thus, volcanism also did not affect local diatom communities.
The abundance of cold-water diatoms, dominant taxa in the lower Zone 1, had been decreasing during the Holocene (Figures 3c and 4). This is in agreement with the gradual decline of dry dwarf shrub tundra communities at the same point in Shemya Island in the Holocene, indicating the transition from severe to more mild and humid coastal conditions [27]. This local climatic trend which could be proposed based on our diatom analysis contradicts the global trend of cooling after 4000 BP as the Neoglacial period started [32,69] and can be explained by the growing ocean-coast effect on the local diatom community due to the sea level rise. Humidification of local conditions is also supported by the gradual decline in terrestrial soil-living taxa. Thus, the continuous rise of the ocean level produced a stronger effect on the local microclimate than the global climate changes.
Together with the direct influence on the microclimate and hence the diatom community, the sea level rise had a significant indirect impact on the pool environment. The sea level rise began around 6000 cal. years BP, leading to the flooding of the old shoreline, and forced seabirds to move inland for nesting. This influence was not linear-at first, the δ 15 N level had been increasing gradually, but then, at about 4500 cal. years BP it rose sharply, which corresponds to a global sea level rise that reached the present values around 5000-4000 cal. years BP [12,70,71]. The zoogenic factor explains the development of higher plants with additional nitrogen inflow as fertilizer, and, consequently, the abundance of epiphytic diatom species in 6000-4000 cal. years BP.
Soils on the islands are known to be exposed to colonial seabirds and have high levels of NO 3 , NH 4 + , and total nitrogen (N) abundance [25,72]. The massive introduction of organic agents (by the influx of water, animals, and mammals) leads to the eutrophication of the water bodies, which is confirmed for other regions [73][74][75]. In our case, the eutrophication was not so obvious, because the observed diatom species mostly prefer oligotrophic to dystrophic water (Pinnularia borealis, Pinnularia intermedia, Pinnularia lagerstedtii, Pinnularia lata, Cavinula spp.). This means that the water body was located out of the bird colony itself, although it received a high input of organic matter. On the other hand, too high concentrations of the nutrients (hyper-eutrophication) can inhibit the growth and development of the diatoms and all vegetation in general [76,77].
The effect of an increase of the organic matter inflow on the diatom community of the pool is clearly detectable. In particular, we found a positive relationship between δ 15 N and the "benthic-epiphytic" component (PC2), and a negative one with the Shannon's index. A rather sharp decline in the diversity index during the period between~4000 and~2000 cal. years BP (154-242 cm) is associated with the peak of nutrients introduced, presumably, with the guano of birds nesting close to the peat deposits (see Figure 3b). After that, diatom diversity again increased as the bird colony on Shemya Island was depressed by human hunting [8]. Therefore, after colonization of the island at about 3000 cal. years BP, humans became one of the main drivers of changes in the coastal environment and fauna, even indirectly influencing the diatoms. The vegetation communities of the Shemya coast also radically changed with the high guano input and after the reduction of avian influence [27]. Thus, combined effects of zoogenic and anthropogenic factors can cause abrupt changes in the coastal ecosystems.
Zone 1 and Zone 2 in diatom community history represent two basically different types of communities. They are separated by 20 cm of layers with high sand content (about 400 years) with no diatoms, suggesting that at the transition point between two regimes (c.a. 4500 cal. years BP), the pool had dried out, and the newly formed pool community was not related by origin to the former, primary community. This abrupt transition in community structure coincided with the peaks of decomposition rate and δ 15 N 5 (Figure 3a,b), and sharp decline in diatom diversity (Figure 3f). These synchronized abrupt shifts in community and environmental parameters suggest that, at the transition point between the Zone 1 and Zone 2, the properties of the water body have changed drastically due to the regime shift.
Moreover, patterns of community dynamics in the lower Zone 1 and upper Zone 2 differ strongly in terms of the amplitude of fluctuations and graduality. In the continuous Zone 1, not disrupted by sand-filled layers with no diatoms, we observed abrupt rapid threshold-type shifts in community composition (e.g., between Subzones 1b and 1c), strong fluctuations of the Shannon index, and all three principal components. In contrast, in Zone 2, the community changed gradually without sharp transitions from one state to another. The interrelatedness of successive communities and the continuity of community dynamics was maintained despite long (6000-1000 years, Figure 4) intervals with pure sand with no diatoms, suggesting that the pool was dried out several times during this period. This difference in dynamic patterns between lower and upper zones may be explained by the more stable climatic conditions and sea level in the second half of Holocene compared to the first half, when they had changed significantly and sometimes fluctuated [69,78,79]. These changes might cause rapid shifts in ecosystem states, indicated by several "tipping points" (as per Gladwell [80]; see also [81,82]) in diatom community dynamics observed in the Early Holocene (between Subzones 1a and 1b, 1b and 1c, and Figures 3 and 4) and at the transition point between Zone 1 and Zone 2. We can also hypothesize that the relatively stable global environmental conditions were an auspicious background to see abrupt changes due to influence of the zoogenic and, then, anthropogenic factors.

Conclusions
Sea level change was the main driver of the diatom community producing direct (i.e., climatic) and indirect (i.e., zoogenic) influences on the studied small water body. After 3000-2000 cal. years BP, the anthropogenic factor (hunting depression of the bird colony) also became significant. During the entire water body history, the diatom community was influenced by groups of factors: global factors (e.g., sea level rise) caused gradual change of local factors, which resulted in smooth shifts in community. In contrast, local factor influence (bird colony rise and fall due to human activity) caused abrupt and transient shifts. We can hypothesize that relatively stable global environmental conditions in the Late Holocene were an auspicious background to see abrupt changes caused by zoogenic and anthropogenic factors.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/w13213134/s1, Table S1. Information for the combined layers of McDonald Point peat deposition included in the analysis, Table S2: Information for the layers age carried out using the Bchron package for R environment, Table S3: Species found with their ecological preferences.

Paleoreconstruction Based on the Diatom Analysis
Based on our diatom analysis and taking into consideration earlier obtained data on abiotic factors, we can reconstruct the history of the studied water body during the Holocene.
Zone 1 (~9500 cal. years BP to -4700 cal. years BP). At this time, earlier stages of a pool formation and then its subsequent filling by water with a gradual depth increasing took place. During all this time, the pool was fresh, with a low conductivity and low pH. The earlier Holocene was a relatively warm period, but the winters were cooler than now [69]. This fact alone explains the abundance of several cold-tolerant taxa in Zone 1.
The diatom community of Subzone 1a indicates a weak acidification of the water body, and, apparently, very earlier stages of its formation since dominant species in the community are usually characteristic of moist soils and very shallow waters (Table 1). Pinnularia borealis is a cold-and desiccation-resistant species that can withstand temperatures down to −20 • C [83]. Pinnularia lata is an indicator species of siliceous waters; its abundance can mean a high availability of biogenic silica [59]. The presence of few valves of the genus Aulacoseira (mainly A. canadensis and A. crassipunctata), which are planktonic forms, may indicate an increase in pool depth. In general, beginning of the freshwater sedimentogenesis took place at this time.
Aulacoseira spp. valves in Subzone 1b are rare, which is most likely associated with the cessation of the depth increasing. An increase of the epiphytic form abundances and a great abundance of Eunotia, as well as a decrease of terrestrial diatom abundances, are evidence of waterlogging beginning as well as of a pH lowering in the pool. The diatom assemblage of Subzone 1b corresponds to an initial stage of the bog development.
A noticeable increase of Aulacoseira abundance in Subzone 1c indicates a rise of the water level. Since this genus occurs throughout the entire subzone with a strong abundance which increases towards its top, the water body level can be considered as constantly increasing. In general, only centric diatoms are well-represented in the diatom communities of sphagnum bogs [84]; however, in very shallow bog basins (with depth of 5-10 cm 2 ), centric diatoms are not present [85]. Interestingly, Heart Lake from Adak Island (located 635 km west of Shemya Island) was also characterized by the dominance of planktonic forms (Aulacoseira subarctica, Cyclotella spp.) in the period from 9300 to 4400 BP [32]. The authors proposed that it was warmer than now at 9600 BP. They also hypothesized possible strong winds which promoted the mixing of water masses and explain the presence of a large amount of biogenic silicon in the environment.
Staurosira exiguiformis was among the dominants at that time. This is a widespread freshwater benthic cosmopolitan species which also prefers near-neutral waters with a low nutrient content [86]. The water level in the pool was increased at that time, the pH value also increased, and the level of humic acids fell. The observed species of the genus Cavinula prefer oligotrophic or dystrophic near-neutral or slightly acidic waters with a low conductivity [87]. An increase of Encyonema species abundances is an indicator of depth increasing, as well as of the presence of possible weak water flows [88]. P. intermedia is widespread both in swampy water bodies and in sediments of shallow lakes and rivers; it prefers slightly acidified waters (optimal pH = 6.6-7) with a low conductivity [89].
The transition to Zone 2 is marked lithologically by a layer of sand and can be characterized by the absence of any diatom valves. This sand mass was probably transferred by wind from the ocean shore, as, due to the sea level rise, the peat bog became closer to the edge of the island and could be strongly affected by the ocean. Moreover, the beginning of the Neoglacial was characteristic of the winter storm intensification [69]. Most likely, due to the abundant sand drift, this pool was completely filled by sand and dried up, and a newly appeared water body with its diatom community has no continuity with the pool which existed there during Phase 1. A new water body was formed in changed climatic conditions as a cooler Neoglacial climatic phase began. For example, according to the study of the pollen spectra of the peat bog from another island of the Aleutian Arc, Carlisle Island, some changes in the pollen spectra after 4500 cal. years BP reflected an increase in humidity and the climate cooling [26].
Zone 2 (-4700 cal. years BP-present). Zone 2 corresponds to the formation of a peatland with a change of its water chemistry towards near-neutral conditions.
The absolute dominants of the Subzone 2a (Pinnularia cf. islandica, Pinnularia incostans, Pinnularia intermedia) are freshwater benthic species, so they could not be introduced with sea sand. They form the core of the diatom community in the new reservoir.
The species composition of Subzone 2b is similar to that of Subzone 2a, but freshwater species preferring waters with moderate conductivity also appear (Diploneis krammeri, D. ovalis), which could indicate an increase in the ocean influence on the water body, i.e., due to salty water splashes. Subzone 2c is characterized by the presence of the northern alpine aerophile Pinnularia lagerstedtii, which often inhabits wet mosses and the soil surface. P. lagerstedtii is a terrestrial species preferring near-neutral or acid waters [90]. Luticola cf. mutica is a terrestrial species, which indicates the development of soil processes in the surrounding area. An increase in the number of valves of the species Staurosira and Staurosirella may indicate the development of water macrophyte vegetation.
Thus, we can assume that after disappearance of the pool existing at the time of Zone 1, active waterlogging began, accompanied by some peat-forming processes. At the final stage of its development, the water body was represented by a medium-sized boggy swamp with well-developed vegetation and well-moistened adjacent soil. The water varied from acid (at the very beginning) to slightly acidic (at the end of existence), and the conductivity remained low throughout the entire time.