Responses of attached bacterial communities to blooms of the swimming shelled pteropod Creseis acicula in Daya Bay, southern China

Abstract The shelled pteropod Creseis acicula is a marine pelagic shellfish widely distributed from temperate to tropical seas around the world. From June to July 2020, a C. acicula bloom first happened in the Daya Bay, southern China, and its density reached the highest value (5600 ind. m−3) ever recorded around the world. However, few studies have investigated the responses of bacterial communities to the C. acicula bloom. In the present study, we examined the community profiles of three communities of bacteria including the free-living and particle-attached bacteria in the blooming and reference waters, and bacteria attached to the whole body and shell of C. acicula using a high-throughput sequencing method. The results indicated that the C. acicula bloom had a greater impact on particle-attached bacteria than free-living bacteria. Among the bloom-sensitive particle-attached bacteria, the predominant bacterial phyla were Pseudomonadota, Bacteroidota and Verrucomicrobiota in the blooming areas, whereas they were Actinomycetota and Planctomycetota in the reference areas. Specifically, fecal bacteria Haloferula and Halioglobus spp. were significantly enriched in the blooming waters and accumulated on C. acicula shells. Conversely, the significantly lower relative abundance of Nocardioides sp. in the blooming area and accumulated on the whole body of C. acicula indicated their attachment to particles consumed by C. acicula. Overall, our results suggested that the C. acicula bloom influenced marine bacteria, particularly particle-attached bacteria, by increasing (e.g. providing shells and feces) or decreasing (e.g. filter-feeding the suspended particles) the abundance of available substances.


Introduction
Rapid and vigorous proliferation of any marine organisms, culminating in predominance within the local population, exerts profound impacts upon both the environment and the other species, thus a compr ehensiv e understanding of these impacts is essential (Blauw et al. 2010 , Zohdi andAbbaspour 2019 ).Extensive attention and r esearc h hav e been focused on blooms of algae and jellyfish to clarify direct and indirect impacts on bacterial communities (Sun et al. 2020, Xia et al. 2020, Peng et al. 2021, Shi et al. 2023 ).Ho w e v er, our current understanding of these impacts remains insufficient.
The shelled pteropod Creseis acicula is a marine pelagic shellfish widely distributed around the world.Creseis acicula is primarily found at depths of less than 500 m in both tropical and subtropical waters of the Pacific, Atlantic and Indian Oceans (Albergoni 1975, López-Arellanes et al. 2018, Zeng et al. 2021 ).Creseis acicula blooms hav e occurr ed globall y, with the highest fr equenc y in the Karw ar, Goa and Bengal bays of the Indian Ocean (Krishna-Murthy 1967, Peter and Paulinose 1978, Naomi 1988 ), the north Pacific waters around Sado Island (Nishimura 1965 ), the Gulf of Mexico (Hutton 1960 ) and the Mediterranean Sea (Kokelj et al. 1994, Tunçer et al. 2021 ) .Creseis acicula is commonly the dominant pteropod species in the Yellow Sea, East China Sea and South China Sea (Zhang 1966, Qi et al. 2021 ).Although widely distributed in the coastal waters of China, its abundance has remained relatively lo w, usually lo w er than 0.50 individuals (ind.)m −3 , and has ne v er bloomed (Qi et al. 2021 ).Ho w e v er, in the summer of 2020, from mid-June to mid-July, a substantial outbreak of C. acicula was observed in southwestern Da ya Ba y, Guangdong Pro vince , northern South China Sea.This was the first recorded outbreak of C. acicula in Chinese waters and also is the highest density (5600 ind.m −3 ) recor ded w orldwide to date (Qi et al. 2021 ).
Bacterial communities, either floating in seawater or attached to the surfaces of particulate matter, play crucial roles in the biogeochemical cycles of almost all elements in the ocean ecosystem (Ruiz-Gonzalez et al. 2013, Shi et al. 2023 ).Additionally, activ e bacterial exc hange occurs with c hanges to the environmental conditions of two or more habitats, where each habitat favors the pr olifer ation of specific taxa that are typically unre presentati ve in the other (De Corte et al. 2014, 2018, Liu et al. 2023 ).Therefore, we hypothesized that the outbreak of C. acicula may exert various impacts on the bacterial communities.For example, the sudden bloom of C. acicula provided a substantial number of particles (e.g. the pteropod and fecal pellets) to which some bacterial species could attach, thus favoring their numbers increas- ing, whereas the abundance of some other bacterial species may decrease due to the filter-feeding of C. acicula , which markedly decreases the amount of suspended particulate organic matter (such as diatoms, dinoflagellates and microcrustaceans) to which these bacterial species are attached.Ho w ever, these hypotheses and assumptions have not yet been conclusiv el y inv estigated, potentially due to the lo w frequenc y of C. acicula blooms .T he associations and relationships between bacteria and C. acicula are far fr om full y understood.
In the present study, we examined the profiles of three communities of bacteria, that is, the pelagic species floating in the waters (r eferr ed to as "free-living bacteria"), the species attached to particles such as the fecal pellets of C. acicula (referred as "particleattached bacteria") and the bacteria attached to the whole body (soft tissue + shell) and shell of C. acicula using a high-throughput sequencing method to gain a better understanding of the potential impacts on other organisms in coastal waters.

Study sites and sampling protocols
This study was carried out in Daya Bay, located in Guangdong Pro vince , southern China, which is a 650 km 2 semi-enclosed embayment in the northeast of the South China Sea.Its av er a ge water depth is 10 (range: 5 −20) m and the annual mean air temperature is 22 • C. The minimum sea surface temperature occurs in winter (15 • C), and the maximum in summer and fall (30 • C) (Wang et al. 2006 ).A bloom of C. acicula occurred in southw est Day a Bay from mid-June to mid-July 2020.On 9 and 10 July, triplicate water samples were collected from each station in both blooming areas (n = 15) and adjacent non-blooming (termed as r efer ence) areas (n = 15) (as shown in Fig. 1 ); 5 L of surface seawater was collected in each station and immediately transported to our seaside laboratory using an ice box for determining the bacterial community structure, picoplankton abundance and the concentrations of NO 3 − , NO 2 − , NH 4 + , PO 4 3 − , SiO 3 2 − and c hlor ophyll a .Additionall y, the temper atur e, salinity, dissolv ed oxygen (DO) and pH of the surface seaw ater w er e measur ed in situ using a m ultipr obe sonde (Yellow Springs Instrument Company, Inc., Yellow Springs, OH, USA).
In the laboratory, 2.5 L of surface seawater was pr e-filter ed with a 20-μm bolting cloth to r emov e debris and lar ger or ganisms .T hen the number of C. acicula individuals retained in the bolting cloth was quantified.For bacterial DNA extraction, 2.5 L of seawater was passed through 3-μm polycarbonate membranes (EMD Millipor e Cor por ation, Billerica, MA, USA), and 1.5 L of the filtrate was passed through 0.2-μm polycarbonate membranes (EMD Millipore Cor por ation).The particle-attac hed bacteria (attac hed to particles with a diameter of 3-20 μm) and free-living bacteria (with a size of 0.2-3 μm) that were captured with the 3-and 0.2-μm membr anes, r espectiv el y, wer e fr ozen in liquid nitr ogen and stor ed at -80 • C until DNA extraction (Ortega-Retuerta et al. 2013, Shi et al. 2023 ).
For the bacteria attached to the whole body (soft tissue + shell) of C. acicula , intact C .acicula samples (n = 8-10) collected from sites B1, B3 and B4 were transferred to a centrifuge tube containing 5 ml of sterile seawater after washing twice with sterile seawater to r emov e an y fr ee-living bacteria.Samples (sterile seawater containing C .acicula ) were filtered through 0.2-μm membranes after crushing and oscillating at 1000 r/m for 20 sec with a vortex oscillator (Scientific Industries , Bohemia, NY, USA).T he membranes wer e fr ozen in liquid nitr ogen and stor ed at -80 • C until DNA extr action.Similarl y, for the bacteria attached on the shell of C. acicula , shells were gently separated from the soft tissues of C. acicula (n = 8-10) with a tweezer, then collected and stored as described abo ve .

Abundance of picoplankton
Triplicate 2-mL seawater samples wer e tr eated with 1% glutar aldehyde and stor ed in liquid nitr ogen to determine the abundance of picoplankton (picophytoplankton and free-living bacteria).Picophytoplankton populations of Proc hlorococcus , Synec hococcus and picoeukaryotes were identified by flow cytometry (FAC-Scalibur flow cytometer, Becton Dickinson, San J ose , C A, USA) based on analysis of the parameters SSC, FL2 and FL3 (Olson et al. 1990, Li et al. 2017 ).Total free-living bacterial abundance (TBA) was determined by staining with SYBR-Green I (Molecular Probes, Eugene, OR, USA) and quantifing cell numbers by flow cytometry.Based on gr een fluor escence intensity (FL1), the free-living bacteria were segregated into two categories: free-living bacteria with low nucleic acid content (LNA) and free-living bacteria with high nucleic acid content (HNA).

DN A extr action and high-throughput sequencing
Bacterial DN A w as extracted using the FastDNA ® Spin Kit for Soil (MP Biomedicals, Inc., Santa Ana, CA, USA), and assessments of its concentration and quality used a NanoDrop™ 2000 UV-vis spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and 1% a gar ose gel electr ophor esis, r espectiv el y.The V4-V5 hyperv ariable regions of the bacterial 16S rRNA gene were amplified using the universal bacterial primers 515F (5 -GTG YCA GCM GCC GCG GTA A-3 ) and 926R (5 -CCG YCA ATT YMT TTR AGT TT-3 ) with a GeneAmp™ PCR System 9700 (Applied Biosystems , Carlsbad, C A, USA) (P ar ada et al. 2015 ).The reaction system and reaction conditions were according to Shi et al. ( 2023 ).PCR products were extr acted fr om 2% a gar ose gel and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) according to the manufacturer's instructions, and were quantified using a Quantus™ Fluorometer (Promega Co., Madison, WI, USA).Purified amplicons were pooled in equimolar amounts and pairedend sequenced on an Illumina MiSeq platform (Illumina Inc., San Diego, C A, USA).T he sequences are publicly available at the NCBI Sequence Read Arc hiv e ( http:// www.ncbi.nlm.nih.gov/Traces/ sra ) under BioProject number PRJNA988167.

Environmental factors analysis
Samples for c hlor ophyll a wer e collected by filtering 0.3-L seawater samples to a Whatman GF/F filter (47 mm in diameter), and c hlor ophyll a concentr ations wer e measur ed with a fluorometer, as described by Parsons et al. ( 1984 )

Da ta anal ysis
T he ra w 16S rRNA gene sequencing reads were demultiplexed, quality filtered and merged using fastp version 0.20.0 (Chen et al. 2018 ) and FLASH version 1.2.7 (Mago č and Salzberg 2011 ), and the applied criteria were according to Shi et al. ( 2023 ).Operational taxonomic units (OTUs) were clustered at a 97% similarity cutoff using UPARSE version 7.1 (Edgar 2013 ) and rarefied based on the smallest number of sequences (16 405) to standardize une v en sequence depth, and the taxonomy of each OTU re presentati ve sequence was analyzed with RDP Classifier version 2.2 against the 16S rRNA gene database (Silva v138.1) with a confidence threshold of 0.7 (Wang et al. 2007 ).Statistical analysis was performed using R version 3.6.3(R Core Team 2018 ).The bacterial comm unity structur e was anal yzed using non-metric multidimensional scaling (NMDS) based on Bray-Curtis distances with the R pac ka ge "v egan".ANOSIM was used to compar e differ ences both within and between groups with 999 permutations (Oksanen et al. 2019 ).Normal distribution and homogeneity of v ariance wer e anal yzed with the R pac ka ges "stats" and "car".A co-occurrence network was established using the R pac ka ges "psyc h" and "igr a ph" based on the genus le v el (Csárdi andNepusz 2006 , Re v elle 2021 ).Spearman corr elation scor es wer e calculated, and onl y r obust (Spearman's r > 0.7 or r < -0.7) and statistically significant (Benjamini and Hochberg-adjusted P < 0.01) corr elations wer e k e pt in the network (Benjamini et al. 2006, Ma et al. 2016 ).In the co-occurr ence network, bloom-sensitiv e particle-attached bacteria whose distributions were significant differences ( P < 0.05) between the blooming and reference areas were identified using indicator species and edgeR analysis and marked in the plot with R softwar e. Differ ences in the distribution of the bloom-sensitiv e particle-attac hed bacteria (with a relative abundance of > 0.1%) between the whole body (soft tissue + shell) and shell of C .acicula were calculated using STAMP (Parks et al. 2014 ).Additionall y, corr elation coefficients between environmental factors and bloom-sensitive particle-attached bacteria, with significant differences in distribution patterns between the whole body (soft tissue + shell) and shell of C .acicula , were visualized in heatmaps generated with the R package "pheatmap" (Kolde 2019 ).

Bacterial community structures in the waters between C. acicula blooming and reference areas
There was a clear separation of particle-attached bacteria between the blooming and r efer ence ar eas (Fig. 2 A).The differ ences between blooming and r efer ence ar eas wer e gr eater than that within eac h ar ea (Fig. 2 B).The fr ee-living bacterial comm unities from either blooming or reference areas were mixed (Fig. 2 C), thus no obvious differences were observed (Fig. 2 D).T herefore , C. acicula bloom had a greater impact on particle-attached bacteria than that on free-living bacteria (i.e. the particle-attached bacteria were more sensitive to C. acicula bloom).

Effects of C. acicula bloom on co-occurrence patterns of particle-attached bacteria
As shown in Fig. 3 , there was a clear separation of the bloomsensitiv e particle-attac hed bacteria between the blooming and r efer ence ar eas.Furthermor e, Module 1 mainl y consisted of bacteria significantly enriched in the reference areas, with Actinomycetota and Planctomycetota dominating with mean cum ulativ e r elative abundances of 62.13% and 26.78%, r espectiv el y (Fig. 3 A and  B).Bacteria in Modules 2 and 3 were significantly enriched in the blooming areas, and these bacteria mainly belonged to Pseudomonadota, Bacteroidota and Verrucomicrobiota [mean cumulativ e r elativ e abundances of 54%, 23.70% and 17.84%, respectiv el y (Fig. 3 B and C)].

Bacteria attached to the whole body (soft tissue + shell) or shell of C. acicula
The distribution differences of the bloom-sensitive particleattached bacteria on the whole body (soft tissue + shell) and shell of C. acicula were determined to clarify potential relationships between these bacteria and C. acicula (Fig. 4 ).Of the bacteria enriched in the r efer ence ar eas, onl y Nocardioides sp.(Actinomycetota) had a significantly ( P = 0.028) higher abundance on the whole body than on the shell of C. acicula (Fig. 4 ).Of the bacteria enriched in the blooming areas, the abundance of unclassified Flavobacteriaceae bacteria (Bacteroidota) on the whole body was significantly ( P = 0.015) higher than that on the shell of C. acicula .Conv ersel y, Bacter oidota, Pseudomonadota (Alpha pr oteobacteria and Gamma pr oteobacteria), Verrucomicr obiota and Planctomycetota wer e significantl y ( P < 0.05) mor e abundant on the shell than on the whole body of C. acicula .

Rela tionships betw een bloom-sensiti v e particle-attached bacteria and environmental factors
For those bloom-sensitive particle-attached bacteria that had significantl y differ ent distribution patterns between the whole body and shell of C. acicula , their relationships with environmental factors are shown in Nautella sp.

HIMB11
Whole body (soft tissue + shell) or ganic nitr ogen (NO 3 − , NH 4 + and NO 2 − ), but significantly ( P < 0.05) and negativ el y corr elated with LN A, TB A, HN A and temperature.Nautella , Silicimonas and Planktotalea spp.were significantly ( P < 0.01) and positiv el y corr elated with nutrient contents, salinity and CA, but significantly ( P < 0.01) and negatively correlated with temper atur e and DO.

Discussion
Compared with algal blooms (Sun et al. 2020, Xia et al. 2020, Shi et al. 2023 ), the impacts of C. acicula bloom on bacterial communities remain unclear.The results of the present study clearly demonstrated that a C .acicula bloom also had direct or indirect influences on bacterial communities.In addition, compared with algae, C. acicula had some species-specific impacts .T hey could migrate in both horizontal and vertical directions (Zeng et al. 2021 ), instead of passiv el y tr ansporting with curr ent as algae do, and this ability restricted long-term exposure of bacteria to their metabolites (Kornicker 1959, Han et al. 2022 ).Additionally, unlike the dead algae that floated in the surface waters and decomposed there, the heavy straight-needle shells made the dead C. acicula sink rapidly to the seabed where they decomposed, thereby minimizing their impacts on bacteria in the surface waters (Blauw et al. 2010, Zohdi and Abbaspour 2019, Zeng et al. 2021 ).
Algal blooms a ppear ed to mainl y affect the fr ee-living bacteria in the waters instead of those species attached to particles (Shi et al. 2023 ), whereas C. acicula bloom could have various impacts on particle-attached bacteria communities in se v er al ways; for example, on the one hand, C. acicula were filter-feeders, they fed on suspended particles including diatoms, dinofla gellates and micr ocrustaceans, whic h wer e substr atum for particle-attached bacteria.This process consequently decreased the abundance of these bacteria (Sakthivel andHaridas 1974 , Zeng et al. 2021 ).On the other hand, the fecal pellets released by C. acicula provided the substratum for particleattac hed bacteria, whic h had the potential to incr ease the abundance of these bacterial species .Furthermore , C. acicula bloom excr eted inor ganic nutrients (e.g.NO 3 − , NH 4 + , PO 4 3 − and CO 2 into water; Table S1 ) and stimulated the growth of phytoplankton.This also increased the av ailable substr atum for particleattac hed bacteria.Additionall y, bioav ailable phosphorus (mainl y PO 4 3 − ) could be dir ectl y absorbed and utilized by some particleattached bacterial species such as copepod-attached bacteria, which harbor large numbers of phosphate transport-related genes (De Corte et al. 2018 ).Ther efor e, the PO 4 3 − excr eted by C. acicula may favor the growth of copepod-attached bacterial communities.P article-attac hed bacteria with significant differences in distribution patterns between the r efer ence and blooming areas were classified as bloom-sensitive particle-attached bacteria; the distribution of these bacteria into distinct modules of the cooccurrence netw ork w as corr elated with the pr esence or absence of C. acicula bloom, which indicated that the distributions of these bacteria may relate to C. acicula , whose exoskeleton and gut are nutrient-and carbon-enric hed micr ohabitats for colonization by bacteria (Carman and Dobbs 1997, Tang et al. 2010, De Corte et al. 2018 ).Additionally, these bacteria may originate from the feces or intestines of C. acicula , or attach to particles consumed by C. acicula .Ther efor e, it is helpful to analyze the differences in the distribution patterns of bloom-sensitive particle-attached bacteria between the whole body (soft tissue + shell) and shell of C. acicula .
For the bloom-sensitive particle-attached bacteria belonging to the phylum Actinomycetota, only the endophytic Nocardioides sp. had a significantly higher abundance on the whole body (soft tissue + shell) rather than on the shell of C. acicula .Its abundance was positiv el y corr elated with the picoeukaryotes abundance, but negativ el y with the density of C. acicula , suggesting that its abundance increases within the C. acicula microbiome by "hitchhiking" into the gut on ingested food particles (Benson andSilvester 1993 , Anandan et al. 2016 ).Ultimately, Nocardioides sp.accumulated on the whole body of C. acicula instead of its shell.
The reasons for the higher abundance of particle-attached bacteria in blooming waters and on C. acicula shells could potentially be: (1) as discussed abo ve , the high population of C. acicula in blooming areas produced a large number of fecal pellets, which could be the substratum for these bacteria to attach to.Reasonabl y, some bacteria attac h to C. acicula shells.For example, in the present study, the abundance of Halioglobus and Haloferula spp., whic h wer e pr ov en to be the dominant bacteria in m ussel feces (Griffin et al. 2021 ), was positive, although the correlation efficiency did not r eac h a significant le v el, r elated to the density of C. acicula in the waters; (2) the C. acicula shells also served as the substratum for some epizoic diatom Licmophora sp., which have specialized m ucila ge pads and stalks favoring them to adhere on shells (Russell and Norris 1971, Hiromi et al. 1985, Gómez et al. 2018, 2020 ).Ther efor e, the abundance of Licmophora sp. and bacteria attached to it, such as diatom-associated Winogradskyella (Bacter oidota), substantiall y incr eased in C. acicula bloom areas (Zhang et al. 2023 ); and (3) some C. acicula -associated bacteria, such as members of the famil y Rhodobacter aceae, significantl y pr oliferated with the C. acicula bloom and subsequently adhered to other particles in the envir onment.Rhodobacter aceae was the dominant family of zooplankton-associated bacteria and could adhere to coral, sponges and microalgae (Roder et al. 2014, De Corte et al. 2018 ).In the pr esent study, Rhodobacter aceae str ains, includ-ing Nautella , Silicimonas , Planktotalea and Roseovarius spp., as well as HIMB11, had accumulated on the shells of C. acicula and proliferated in the blooming areas as particle-attached bacteria.Additionally, the abundances of these taxa were significantly and positiv el y corr elated with the density of C. acicula , whic h may be because they are contributing to biofilm formation, and thus play a crucial role in the colonization of other micr oor ganisms to gain a competiti ve ad vantage for colonization on the shells of C. acicula (Amaral-Zettler et al. 2020, Zhao et al. 2023 ).
Members of the family Fla vobacteriaceae , whic h ar e able to degrade high molecular organic matter, such as proteins and pol ysacc harides (Bennke et al. 2016, Lapebie et al. 2019 ), exhibit commensal or par asitic inter actions with zooplankton (Cottrell andKirchman 2000 , Beier andBertilsson 2013 ).In this study, the abundance of unclassified Flavobacteriaceae on the whole body of C. acicula was significantly greater than that on the shells.Flavobacteriaceae are known to be the dominant species in shellfish guts and feces (Griffin et al. 2021 ); similarly, they might also be the dominant species in C. acicula intestine.Some members of the Flavobacteriaceae family are aerobic or facultative anaerobic species (Lv et al. 2014 ), and they could pr olifer ate in the gut of C. acicula , which is hypoxic but rich in nutrients (De Corte et al. 2018 ).These bacteria could be released into the seawater along with the excretions of feces.

Conclusion
The present study sho w ed that C. acicula bloom can directly and indir ectl y influence bacterial communities through physiological pr ocesses suc h as filter feeding, r espir ation, metabolite r elease, excretion and decomposition.The influences on particle-attached bacteria are more pronounced than on free-living bacteria.Our findings indicate that the C. acicula bloom can lead to various ecological effects; further studies are needed to improve our understanding of its potentiall y pr ofound effects on the biogeochemical circulation of the marine ecosystem.

Figure 1 .
Figure 1.Sampling stations ( •) in Daya Bay.Red dots (R1-R5) r epr esent the r efer ence ar eas and blue dots (B1-B5) r epr esent the C. acicula blooming areas .T he dot with half-r ed and half-blue r epr esents a blooming area on July 9 (B4) and a reference area on July 10 (R4), r espectiv el y.

Fig. 5 .Figure 2 .Figure 3 .
Figure 2. NMDS and distance boxplots of particle-attached bacteria (A, B) and free-living bacteria (C, D) at the genus le v el.

Figure 4 .
Figure 4. Extended error bar plot of C. acicula bloom-sensitive particle-attached bacteria (relative abundance of > 0.1%) at the genus level.Genera with significant differences between the whole body (soft tissue + shell) and shell of C. acicula ( P > 0.05) are shown.The error bars represent standard errors of the bacterial proportions.

Figure 5 .
Figure 5. Correlation coefficients between environmental factors and C. acicula bloom-sensitive particle-attached bacteria (at the genus level) with significant difference in distribution patterns on the whole body and shell of C. acicula .* P < 0.05, * * P < 0.01.CA, Creseis acicula density; DO, dissolved o xygen; HNA, free-li ving bacteria with high nucleic acid content; LNA, free-living bacteria with low nucleic acid content; TBA, total free-living bacteria.