Lakes-scale pattern of eukaryotic phytoplankton diversity and assembly process shaped by electrical conductivity in central Qinghai-Tibet Plateau

Abstract Phytoplankton are the main primary producers in aquatic ecosystems and play an important role in food web and geochemical cycles. Its diversity, community structure, and assembly process are influenced by several factors. Alpine lake ecosystems are relatively weak and extremely sensitive to global climate change. However, the impact of climate change on phytoplankton in Qinghai-Tibet Plateau lakes and their responses are still unclear. In this study, we analyzed the diversity, environmental drivers, and assembly process of phytoplankton community in the central QTP lakes. The phytoplankton of these lakes can be primarily distinguished into freshwater and brackish types, with significant differences in species diversity and community dissimilarity. Both shared nearly same key environmental factors that significantly affecting phytoplankton such as EC, and brackish lakes were also positively correlative with TN. Stochastic process was predominant in phytoplankton assembly. Additionally, freshwater and brackish lakes were dominated by dispersal limitation and heterogeneous selection respectively. Alpine lakes had significant EC thresholds, and their diversity and assembly processes changed significantly around the thresholds. The present findings have important implications for understanding and predicting the response of lake phytoplankton communities to climate change and for making decisions to protect the ecological resources of alpine lakes.


Introduction
Eukaryotic phytoplankton are a complex group of diverse microorganisms widely scattered in various aquatic habitats, and are the fundamental feature of aquatic systems, which can also be used to c har acterize water quality (Soballe and Kimmel 1987 ).Although eukaryotic phytoplankton play a variety of k e y ecosystem roles as important primary producers, they can also occasionally cause harmful algal blooms under certain conditions (Tranvik et al. 2009, Borics et al. 2021 ).A large number of studies have focused on the dynamics of phytoplankton communities and their envir onmental driv ers, and the abundance, structure, and assembly of these phytoplankton communities are known to be sensitive to climate-mediated physical forces and changes in nutrient enrichment (Winder and Sommer 2012, Klais et al. 2017, Borics et al. 2021, David et al. 2021 ).Ho w e v er, onl y a few studies have been performed on phytoplankton communities in alpine lakes, whic h ar e consider ed to be sensitiv e indicators of climate c hange (Elser et al. 2020 ).Recently, studies focused on alpine freshwater lakes hav e r eported that changes in phytoplankton communities may be driven by nitrogen and phosphorus deposition (Wolfe et al. 2003, Tolotti et al. 2006, Brahney et al. 2015 ).In peri-alpine lakes, phytoplankton community is mainly driven by factors that are str ongl y influenced by climate change and eutrophication (Anneville et al. 2005, Gallina et al. 2013 ).In br ac kish alpine lakes, salinity is the critical driving factor regulating both bacterial and micr oeukaryotic comm unities (Liu et al. 2020 ).The shifts in salin-ity and other nutrients caused by expansion or shrinkage of these lakes and human activities can str ongl y contribute to c hanges in the plankton communities and gross primary productivity (Jia et al. 2021, Li et al. 2021, Liang et al. 2021 ).Ov er all, climate c hange, associated with changes in hydrological condition and nutrition shifts, is considered to be one of the possible reasons for phytoplankton comm unity c hanges in alpine lakes.Man y studies hav e demonstrated that the mechanistic links between climate change and phytoplankton dynamics are very important to assess the impacts of climate change on aquatic ecosystems (Winder and Sommer 2012 ); ho w e v er, r esearc h on phytoplankton communities in Qinghai-Tibet (Xizang) Plateau (QTP) lakes is limited.
It has been reported that the phytoplankton communities in QTP lakes have a low alpha diversity and high beta diversity owing to the effects of harsh environmental conditions (Yang et al. 2018 ).The changes in both lake salinity and water temper atur e hav e been r eported to significantl y affect the tr ophic structur e, tr ophic inter actions, and biodiv ersity in aquatic ecosystems of lakes around Siling Co (Zhu et al. 2019 ).Another key issue in microbial ecology is to quantify the ratio of deterministic niche processes and stochastic neutral processes in microbial community assembly (Stegen et al. 2013, Dini-Andreote et al. 2015 ).The debate on microbial community assembly process has been long and mainly focused on the coexistence of nic he pr ocesses and neutr al pr ocesses, and pr e vious studies hav e shown that ther e is no general consensus on the process of phytoplankton assembly (Zhou et al. 2014 ).In gener al, the r elativ e contributions of both the assembl y pr ocesses ar e r elated to envir onmental v ariables (Dini-Andreote et al. 2015, Feng et al. 2018 ).Phytoplankton have distinct abundance, and studies on the effects of ecological processes on the phytoplankton community structure and environmental variables gradient in alpine lakes, especially QTP lakes, are scarce.
In the past three decades, QTP experienced evident climate changes and sho w ed ov er all surface warming and moistening (Kang et al. 2010, Yao et al. 2019 ).Such significant warming and humidification are very critical to the ecological systems across the plateau, including the aquatic ecosystems.Many lake ecosystems have experienced long-term changes as a result of other ecological pr essur es intr oduced by climate c hanges that could affect phenology.There is strong evidence indicating that climate change has a significant impact on the r epr oductiv e phenology of QTP fish as well as other species in terrestrial ecosystems (Zhuang et al. 2010, Tao et al. 2018 ); ho w e v er, their impact on phytoplankton in QTP lakes as well as the responses of these phytoplankton are still unclear.The hydrological changes in adjacent but unconnected lakes can be used to infer phytoplankton community dissimilarity, envir onmental driv ers, as well as assembly processes, and are crucial to determine the outcome of environmental changes .T he Changtang endorheic region is the central part of the QTP, which is the largest uninhabited area far from human activities and with numerous lakes in China.We considered these lakes with a certain gradient in terms of v arious envir onmental factors as an ideal model to investigate phytoplankton diversity and their envir onmental driv ers, and thus further study the effects of climate change on aquatic ecosystems, especially phytoplankton.
In recent years, the Uthermoll sedimentation method and micr oscopic observ ation for the identification of algae has pr esented man y c hallenges owing to the phenotypic plasticity and cryptic diversity of algae (Leliaert et al. 2014, Verbruggen 2014 ).Metabarcoding sequencing can better avoid these deficiencies caused by tr aditional micr oscopic identification (Blaxter 2004 ), although this method also has some limitations, such as variation in gene copy number (GCN) between different species (Angly et al. 2014 ).With regard to QTP lak es, lak es-scale patterns of molecular di versity and assembl y pr ocess of eukaryotic phytoplankton still r emain poorl y pr ofiled, despite their importance in understanding alpine lak e producti vity.Furthermore, studies on the diversity and ecology of QTP are largely focused on terrestrial en vironments , and those on aquatic ecosystem are mainly aimed at fishes or bacteria.Owing to the limited r esearc h on the molecular diversity and ecology of phytoplankton, the applicability of ecological theories obtained from metazoans , embryophytes , or bacteria still remain unclear for eukaryotic micr oor ganisms.Accordingl y, the primary aims of present study were: (1) to comprehensively profile the diversity and assembly process of eukaryotic phytoplankton community in those ultra-alpine lakes, and reveal the k e y dri ving factors; (2) to assess impacts of climate change induced decrease in electrical conductivity on phytoplankton diversity in various Qinghai-Tibet Plateau lakes, and thus to predict how phytoplankton diversity patterns will change with fluctuations in electrical conductivity across different alpine lakes.

Sampling and e v alua tion of environmental variables
A total of 16 lakes, including 5 freshwater lakes and 11 br ac kish lakes, in the central Changtang endorheic region were selected (Fig. 1 ).At least two samples fr om differ ent pela gic zones in the same lake were collected from May to July 2015.The altitude and coor dinates w ere obtained using a GPS tr ac ker (Garmin or egon 750), and detailed sampling information is provided in Table S1 .Water temper atur e, salinity, electrical conductivity (EC), pH, total dissolved solids (TDS), and dissolved oxygen concentration (DO) wer e measur ed in field using a Hydrolab Hash (Austin, TX, USA).The concentrations of total nitrogen (TN), ammonium aitrogen (NH 4 -N), nitr ate nitr ogen (NO 3 -N), nitrite nitrogen (NO 2 -N), total phosphorus (TP), soluble r eactiv e phosphate (SRP), and other biogenic elements and heavy metal ions were determined as described pr e viousl y (Xiong et al. 2020 ).For envir onmental DNA extraction, 500 ml of the collected samples were immediately filter ed thr ough 0.22-μm (in diameter) Dur a por e membr anes (Millipore) using a peristaltic pump, and all the membr anes wer e instantaneously placed into liquid nitrogen and stored.

Sequencing and annotation of metabarcoding data
The DN A samples w er e pr ocessed using OMEGA Water DNA Kit according to the manufacturer's instructions .T he universal primer (F: 5 -CC AGC ASCYGCGGT AA TTCC-3 ; R: 5 -ACTTTCGTTCTTGATYRA-3 ) was used to amplify the SSU rRNA V4 r egion fr om the genomic DNA extracted from each sample (Stoeck et al. 2010 ).The total volume of the reaction mixture for PCR was 10 μL, and the PCR conditions were as follows: initial denaturation at 95 • C for 5 min; 25 cycles of denaturation at 95 • C for 30 s, annealing at 50 • C for 30 s, and extension at 72 • C for 40 s; and a final step at 72 • C for 7 min.The PCR amplicons were purified using Agencourt AMPure XP Beads (Beckman Coulter, Indianapolis, IN, USA) and quantified using Qubit dsDNA HS Assay Kit and Qubit 4.0 Fluorometer (In vitrogen, T hermo Fisher Scientific, Oregon, USA).Subsequently, the amplicons were pooled into equal amounts.For sequencing and construction of library, Illumina HiSeq 2500 was used, and the raw data of all the samples have been deposited in the SRA of the NCBI database under the accession number, PRJNA865087.The raw data were primaril y filter ed and mer ged, and r emov al of primer base pairs was performed using USEARCH (Edgar and Fl yvbjer g 2015 ).The high-quality r eads gener ated fr om the abov e-mentioned steps were used in the following analysis.Sequences with similarity ≥ 97% were clustered into the same operational taxonomic unit (OTU) by USEARCH (v10.0), and OTUs with abundance < 0.005% wer e filter ed, followed by c himer a r emov al using UCHIME (version 8.1) (Edgar et al. 2011 ).The r epr esentativ e sequences of each OTU were annotated by USEARCH (v10.0), and only eukaryotic algal OTUs were subsequently processed (Edgar 2010 ).

Sta tistical anal ysis
The sequence matrix was aligned using Mafft online service (Katoh et al. 2019 ), and then trimmed using TrimAl (Ca pella-Gutierr ez et al. 2009 ).Phylogenetic tr ee was constructed using IQtr ee (Lam-Tung et al. 2015 ), and the obtained consensus tree was used for calculating phylogenetic diversity and assembly process.To explore the differences in phytoplankton community patterns, principal coordinate analysis (PCoA) and non-metric multidimensional scaling (NMDS) clustering based on Bray-Curtis distance were performed.The richness, Shannon-Wiener diversity, and phylogenetic diversity indices were used to measure alpha diversity, while the community dissimilarity (based on Bra y-Curtis , Jaccard, and Morisita distances) was calculated to determine beta diversity.All these indices were computed and compared using VEGAN and PICANTE pac ka ges (K embel et al. 2010(K embel et al. , Oksanen et al. 2017 ) ).To identify the k e y en vironmental factors , redundancy anal ysis (RDA), v ariance partitioning anal ysis (VPA), and Mantel test were performed using VEGAN packages.To explore the mechanism of community assembly in different lake groups, the ecological processes affecting phytoplankton assembly were quantified using Picante , iC AMP, and NST pac ka ges (Stegen et al. 2013 ).The positive (z + ) and negative (z −) responding OTUs (threshold indicator taxa analysis, TITAN) for both br ac kish and fr eshwater lakes were identified, and the change points of phytoplankton comm unity in r esponse to k e y envir onmental v ariables (c hangepoint anal ysis, nCPA) wer e detected using TITAN2 (Baker and King 2010 ).All numerical ecological analyses were performed in R (R Core Team 2021 ).

Comparison of phytoplankton di v ersity in brackish and freshwater lakes
The obtained clean reads from 16 lakes were clustered into 449 O TUs , of which 185 OTUs were annotated as eukaryotic phytoplankton belonging to 8 phyla ( Tables S2 -S3 ; Fig. S1 ).The detailed information, including closest species, GenBank accession number, and % identity of each phytoplankton OTU is provided in Table S4 .The plateaued r ar efaction gr a ph indicated that the data are adequate for further analysis ( Fig. S2 ).Both PCoA and NMDS or dinations clearly sho w ed that the phytoplankton comm unities separ atel y gr ouped into br ac kish lakes and fr eshwater lakes (Fig. 2 A).Dinophytes and Chrysophytes were dominant with respect to relative abundance and number of observed O TUs , respectiv el y (Fig. 2 B).Ov er all, the abundance of Bacillariophytes in freshw ater lakes w as higher than that in br ac kish lakes, wher eas the abundances of Dictyochophytes and Chlorophytes in freshwater lakes were lower than those in br ac kish lakes .T he phylogenetic diversity and alpha diversity of phytoplankton communities in freshwater lakes wer e significantl y higher than those of phytoplankton communities in brackish lakes (Fig. 2 C).Ho w ever, community dissimilarities based on three distances of freshwater lakes were significantly lo w er than those in br ac kish lakes (Fig. 2 C).Based on these results, the phytoplankton communities in QTP lakes could be primarily classified into two types according to salinity or EC, namel y, br ac kish and fr eshwater phytoplank-ton comm unities.Subsequentl y, these two types of phytoplankton comm unities wer e r espectiv el y anal yzed.

Correla tion betw een phytoplankton di v ersity and en vironmental v ariables
Following calculation of variance inflation factor (only factors with variance inflation factor < 10, and Pearson coefficient < 0.75 wer e r etained), 10 and 7 envir onmental factors wer e detected in br ac kish and fr eshwater lakes, r espectiv el y ( Figs S3 -S4 ).In br ac kish lakes, RDA based on the abundance of phytoplankton and filter ed envir onmental v ariables indicated an ordination of lakes related to physicochemical indices (PCI, such as temperature, EC, and pH) and biogenic elements (N, P, Ca, and Si).The first two axes accounted for 25.58% and 16.22% of the v ariance, r espectiv el y (Fig. 3 A), and VPA r esults sho w ed that these se v en v ariables accounted for 46% variance, of which PCI accounted for 19% variance and nitrogenic nutrition accounted for 11% variance (Fig. 3 B).In freshwater lakes, the first two axes in RDA accounted for 29.54% and 21.22% of the v ariance, r espectiv el y (Fig. 3 A), and the phytoplankton communities in freshwater lakes clearly sho w ed an ordination related to EC and temperature .T he VPA results suggested that temper atur e and EC accounted for 6% v ariance, and nitr ogenic nutrition and phosphorus nutrition accounted for 7% and 2% v ariance, r espectiv el y (Fig. 3 B).The Mantel test r esults r e v ealed no significant relationship between environmental variables and phylogenetic diversity, and only EC and TN were correlated with Shannon-Wiener diversity index in brackish lakes .T he dissimilarity based on Bray-Curtis distance of phytoplankton community in br ac kish lakes pr esented significant corr elation with temper atur e, EC, silicon, TN, and calcium (Fig. 3 C).In freshwater lakes, both phylogenetic diversity and Shannon-Wiener diversity indices exhibited no significant correlation with any environmental factors.The dissimilarity of phytoplankton community sho w ed significant correlation with temperature and EC (Fig. 3 C).In summary, the results of Mantel test were closely consistent with those of RDA and VPA, thus indicating that the phylogenetic diversity of phytoplankton in central QTP lakes is independent of most of the en vironmental factors .Ho w ever, the phytoplankton community dissimilarity was significantly correlated with PCI, especially temper atur e and EC, in both br ac kish and freshwater lakes.In particular, temper atur e significantl y affected the dominant algal group in freshwater lakes (namely, Chrysophytes) (Fig. 3 D), whereas EC and TN were the major factors influencing the dominant algal group

T hreshold indica tors and change points along ke y en vironmental v ariables gr adient
Although six envir onmental v ariables (temper atur e, EC, TN, silicon, and calcium) were identified as significant k e y factors influencing phytoplankton communities in brackish lakes, only two exhibited significant change points (Table 1 ).A total of 5 and 6 OTUs were annotated as negative indicators across the full gradients along temper atur e and EC, r espectiv el y, wher eas 5 and 5 OTUs were identified as positive indicators across the gradients of the abov e-mentioned fiv e v ariables, r espectiv el y (Table 1 ; Figs S5 -S8 ).There was no significant threshold indicators and change points across the five k e y environmental variables in freshwater lakes.
Ov er all, temper atur e and EC were the two most significant variables affecting phytoplankton diversity and assembly in freshwater and br ac kish lakes in QTP.As the daily and annual variation in water temper atur e in the same lake is very significant, this study mainly focused on the effects of EC gradient on phytoplankton.Only EC exhibited a gradient variation among the 16 lakes, and played a k e y role in shaping the phytoplankton community structure in both brackish and freshwater lakes; hence, EC was analyzed in the subsequent nCP analysis.All brackish lakes were classified into three subtypes according to the change points of each k e y v ariables, namel y, pr e-c hange point lakes (BC), inter-c hange point lakes (MC), and post-change point lakes (AC).All freshwater lakes were classified as one group (FW) together with the other thr ee br ac kish lake types.Subsequentl y, the alpha div ersity, beta di versity, and quantitati ve process of assembly of phytoplankton in all the lake types were respectively compared.The results revealed that the alpha diversity of phytoplankton in freshwater lakes was significantly higher than that in the other three brackish lakes subgroups, with no significant difference noted among the three brackish lakes subgroups ( Fig. S9 ).The similarity of the phytoplankton communities gradually decreased with increasing EC and was the highest in MC lakes ( Fig. S10 ), suggesting that the phytoplankton community dissimilarity exhibited a normal distribution trend along the EC, with maximum diversity noted in the EC range of about 14-17 ms/cm.

Quantitati v e estimation of phytoplankton assembly process
The proportions of the four assembl y pr ocesses significantl y v aried among the four types of lakes.Dispersal limitation was predominant in freshwater lakes, and only accounted for about 30% in br ac kish lakes, with no significant difference in the three types of br ac kish lakes (Fig. 4 ).The pr oportion of ecological drift increased with the EC, with no significant difference between BC and MC lakes; ho w e v er, the v alue was significantl y higher than that in freshwater lakes and lower than that in AC lakes (Fig. 4 ).The proportion of selection (including homogeneous and heterogeneous selection) was the highest in MC lakes, but was significantly lo w er in the other three types of lakes.Ov er all, it can be concluded that the ecologicall y neutr al pr ocesses wer e dominant in both br ac kish and freshwater lakes.

Discussion
In the present study, Dinoflagellates and Chrysophytes were the pr edominant gr oups in both alpine fr eshwater and br ac kish lakes.Ho w e v er, the r elativ e abundance of Bacillariophytes was higher in freshwater lakes, while that of Cryptophytes and Chlorophytes was higher in br ac kish lakes.A pr e vious study on phytoplankton assembly in coastal lakes had reported that diatoms and Cryptophytes are indicators that prefer freshwater lakes, whereas Cyanobacteria and green algae are predominant in brackish lakes (Obolewski et al. 2018 ).On the contr ary, micr oscopic inv estigations hav e r e v ealed that Bacillariophytes ar e the absolutel y dominant phytoplankton in lakes with different salinity across the QTP  (Li et al. 2021 ).These inconsistent results may be attributed to the huge difference in the identification methods emplo y ed.It must be noted that Uthermoll sedimentation method and microscopic counting can easily detect Bacillariophytes taxa, and may miss some fr a gile taxa suc h as Chrysophytes as well as some pico algae (with diameter less than 2 μm).Besides, these variations in results could also be attributed to the differences in the study area and ele v ation of the lake.Phytoplankton diversity is known to be correlated with altitudes and coordinates, mainly owing to the changes in lake PCI and nutrient le v els along with the variations in altitudes and coor dinates (Kamyko wski et al. 2002 , Bergström andKarlsson 2019 ).
In the present study, the investigated lakes are located in the same geogr a phical r egion, under same climatic conditions and almost fr ee fr om human activities .Hence , the effects of both geogr a phic differences and human activities on phytoplankton community in these lakes were not discussed.Although EC (or salinity) can significantly distinguish lakes into freshwater and brackish, only few studies had focused on comparisons of phytoplankton in freshwater and br ac kish lakes, especiall y in alpine lakes .T he species F igure 4. P er centage of each assembly process in each lake group.FW, freshwater lakes; BC, brackish lakes with EC < 14 ms/cm; MC, brackish lakes with EC = 14-17 ms/cm; AC, br ac kish lakes with EC > 17 ms/cm.DL, Dispersal limitation; ED, Ecological drift; HD, Homogenous dispersal; HES, Heterogeneous selection; HOS, Homogeneous selection.div ersity, comm unity structur e, and dissimilarity of phytoplankton in these lakes significantly differ, with freshwater lakes having higher species diversity and brackish lakes having higher community dissimilarity.Pr e vious studies hav e shown that phytoplankton diversity is higher in semi-brackish water than in brackish and freshwater, and this phenomenon is well explained by the intermediate disturbance hypothesis (Irena et al. 2013 ).In coastal r eservoirs, the slight incr ease in salinity could decr ease the specific diversity of eukaryotic phytoplankton (Mo et al. 2021 ).Howe v er, it m ust be noted that the sampling sites with different salinity emplo y ed in these studies had been in the same reservoir or lake in gener al. Ther efor e, these conclusions may not applicable to the m utuall y inde pendent and unconnected lak es in central QTP.
PCI can dir ectl y or indir ectl y affect the metabolic activity of algae, and can thus r esha pe the phytoplankton community structur e in gener al (Zohary et al. 2021 ).Hence, we presumed that phytoplankton community is strongly influenced by physicochemical factors (temper atur e, EC, etc.) in ultr a-alpine fr eshwater lakes.It has been reported that phytoplankton communities in mountain lakes are mainly related to phosphorus, nitrogen, and silicon contents, with Bacillariophytes pr edominantl y closel y r elated to silicon content and green algae mainly related to salinity (Krupa and Barinova 2016 ), consistent with the findings of the present study in br ac kish lakes.In addition, salinity and EC have been noted to str ongl y affect phytoplankton structur e and abundance in the Great Salt Lak es, QTP lak es, and Baltic Sea coastal lakes (Barrett and Belovsky 2020 ;Obolewski et al. 2018;Li et al. 2021 ), which is in agreement with the results of the present study in br ac kish lakes.Although both br ac kish and fr eshwater lakes ar e affected by PCI such as water temper atur e and EC, phytoplankton communities as well as their main groups in brackish and freshwater lakes have been observed to show different environmental drivers.In the present study, a significant positive correlation was found between the four algal groups (Bacillariophytes, Chloroph ytes, Chrysoph ytes, and Dictyochoph ytes) and silicon concentr ation, whic h may be owing to the fact that Bacillariophytes, Dictyochoph ytes, Chrysoph ytes, and some Haptoph ytes form silica theca or silica scale during their life cycle (Eikrem et al. 2017, Kristiansen and Škaloud 2017, Mann et al. 2017 ).Furthermore, Chrysophytes, the predominant group, were affected by temperature, as most of them occur in the cold water of br ac kish and fr eshwater lakes.In the alpine freshwater lakes , pH, phosphorus , and EC have been found to be the three major variables (Grossmann et al. 2016 ).Besides nitrogen, phosphorus has been noted to primarily play a k e y role in determining the phytoplankton community of subpolar freshwater lakes (Arvola et al. 2011 ).It has been reported that total phosphorus, alkalinity, and water color are the major factors that influence the large-scale distribution patterns of dominant phytoplankton groups across European lakes (Maileht et al. 2013 ).Ho w e v er, the r esults of the pr esent study sho w ed that phytoplankton in QTP freshwater lakes were not affected by phosphorus or nitrogen contents, unlike those in QTP brackish lakes or freshwater lakes in other regions.
Stoc hastic pr ocesses wer e observ ed to gov ern the micr obial comm unity and r egulate its assembl y in alpine freshwater and br ac kish lakes.Ho w e v er, the dominance of selection or speciation in lakes with different EC sho w ed variation.A previous study indicated that phytoplankton in QTP lakes have low alpha diversity and high beta diversity owing to the harsh environmental conditions, suggesting that assembly process is mainly dominated by pr obabilistic dispersal, especiall y dispersal limitation (Yang et al. 2018 ).Likewise, phytoplankton in the lakes of Inner Mongolia Plateau have also been reported to be influenced by stochastic processes (Liu et al. 2022 ).Similar results have also been obtained in some alpine freshwater lakes in Europe and North America (Gendron et al. 2019, Monchamp et al. 2019 ).The contribution of deterministic processes in eukaryotic phytoplankton community assembl y incr eased with incr easing salinity, similar to that reported by Liu et al. ( 2022 ).In addition, the proportion of deterministic processes increased with increasing EC in brackish lakes until EC > 17 ms/cm, and the ecological drift and dispersal limitation were the two main mechanisms across the central QTP lakes.
The phytoplankton diversity and structure are known to show regular shifts along with changes in different environmental factors (Vallina et al. 2017 ).Ho w e v er, it is difficult to explore the underl ying mec hanism of these envir onmental v ariables and determine their change points (Edw ar ds et al. 2013 ).The findings of the present study sho w ed that phytoplankton in both freshwater and br ac kish lakes in the centr al QTP ar e significantl y affected by EC, especially in lakes with EC around 6, 14, and 17 ms/cm.The TITAN results revealed no significant change thresholds and repr esentativ e indicator in freshwater lakes .T hus , we hypothesized that there may not be any significant differences in phytoplankton comm unity c hanges in fr eshwater lakes, and that the impact of climate change on freshwater lakes may be r elativ el y small.Ho w e v er, our r esults sho w ed that the decreased EC did not lead to significant changes in the phytoplankton species diversity in most of the br ac kish lak es, exce pt in lak es with EC ≤ 6 ms/cm, whereas an obvious increase in species diversity was noted in the semibr ac kish lakes.In lakes with EC of around 14 ms/cm, a decrease in EC and a subsequent increase to around 17 ms/cm caused a decrease in phytoplankton community dissimilarity.Such obvious shifts in phytoplankton diversity and structure are schematically summarized in Fig. 5 A. Mor eov er, a gr adual decr ease in EC also affected the phytoplankton assembly process, and species dispersal limitation pr ogr essiv el y incr eased in semi-br ac kish lakes.In lakes with EC of ar ound 14 ms/cm, the pr oportion of selection decr eased.Ov er all, the effect of EC on the phytoplankton assembly process in alpine lakes mainly regulated the proportion of specific pr ocess, while both fr eshwater and br ac kish lakes wer e gener all y dominated by stochastic processes (Fig. 5 B).

Conclusion
The present study revealed that EC (or salinity) is po w erful structuring force on phytoplankton structuring and assembly process in central QTP lakes .T he phytoplankton in these lakes could be extensiv el y classified into freshwater and brackish types, with sig- nificant difference in species and community dissimilarity between them.Although phytoplankton communities in both freshwater and br ac kish lakes wer e significantl y affected by EC and temper atur e, br ac kish phytoplankton were also significantly influenced by TN, silicon, and calcium contents.Stochastic processes were dominant in central QTP lakes, with the proportion of species dispersal limitation gr aduall y decr easing and ecological drift gr aduall y incr easing with the incr easing EC.Significant differ ences in EC gr adients wer e noted in br ac kish lakes, and these lakes could be distinguished into three types based on EC, with an EC threshold of about 14-17 ms/cm.The phytoplankton species div ersity, comm unity dissimilarity, and assembl y pr ocess sho w ed ob vious differ ences acr oss this EC thr eshold.T hus , these results suggest that the decrease in lake EC due to warming and moistening from climate change may have almost no effect on phytoplankton in freshwater lakes, but could afftect phytoplankton in br ac kish lakes, particularl y those with an EC of ar ound 14-17 ms/cm.Ov er all, the findings r e v eal that the effects of lake desalination on eukaryotic phytoplankton depend on the EC regime, and provide a mechanistic basis for understanding global climate change in Qinghai-Tibet Plateau aquatic ecosystems.

Figure 1 .
Figure 1.Map of the studied areas in central QTP and the aerial view of two typical lake groups.(A) Locations of 16 lakes in Changtang endorheic r egion.(B) Repr esentativ e fr eshw ater lake, Gy aring Co. (C) Re presentati ve brackish lake, Siling Co.

Figure 2 .
Figure 2. Comm unity structur e of eukaryotic phytoplankton acr oss Changtang endorheic r egion lakes and comparison of their assembl y, div ersity, and community dissimilarity.(A) Grouping of phytoplankton communities according to compositional similarity (Bray-Curtis distance) using PCoA (adonis analysis r 2 = 0.29, P = 0.001) and NMDS (stress = 0.10), respectively.Red and blue dots represent brackish and freshwater lak es, respecti vely.(B) Heatmaps indicating the r elativ e abundance (z-value) and number of observed OTUs of phytoplankton composition across all the studied lakes.(C) Comparison of species diversity (including phylogenetic div ersity, ric hness, and Shannon-Wiener diversity indices) and community dissimilarity (including Bra y-Curtis , J accar d, and Morisita distances) between br ac kish and freshwater phytoplankton communities.

Figure 3 .
Figure 3. Corr elation anal ysis between comm unity distance and explanatory v ariables .(A) Biplot (samples and en vir onmental v ariables) of RDA based on br ac kish and fr eshwater phytoplankton abundance matrix, r espectiv el y. (B) VPA-based Venn dia gr ams showing variation according to Bray-Curtis community distance explained by distinct and combined effects of PCI, phosphorus nutrition (P-nutrition), and nitrogen nutrition (N-nutrition).(C) Mantel test results of environmental variables with alpha diversity (Shannon-Wiener index), phylogenetic diversity, and beta diversity (Bray-Curtis distance), r espectiv el y. * , * * , and * * * r epr esent P < 0.05, P < 0.01, and P < 0.001, r espectiv el y. (D) P airwise comparisons of envir onmental factors with a color gradient denoting Pearson's correlation coefficient, and Mantel test results of major phytoplanktonic groups with en vironmental factors , the size of circle r epr esent the correlation between environmental variables.

Figure 5 .
Figure 5. Sc hematic illustr ation of v ariations in the phytoplankton diversity and assembly process along EC. (A) Trends of alpha diversity (red line) and beta diversity (blue line) along EC. (B) Trends of four basal assembl y pr ocesses (in %) in phytoplankton comm unities along EC.ED, Ecological drift; DL, Dispersal limitation; HES, Heterogeneous selection; HOS, Homogeneous selection.