After the acceptance of our manuscript, we became aware of an error in our approach used to analyze the bacterial community associated with Microcystis. We only removed DNA sequence reads from Microcystis after the calculation of relative abundance rather than before. In turn, Microcystis abundance may have introduced errors into results on community composition in the ordination and hierarchical clustering.

To assess this, we re-computed the hierarchical clustering and ordination analysis using relative abundance values for the microbiome calculated after Microcystis read removal. The results of the analysis with single Microcystis colonies alone do not change substantially. The correlations with community dissimilarity and sampling date and Microcystis oligotype are still significant and ANOSIM R values increased from 0.41 to 0.43 (for sampling date) and 0.49 to 0.53 (for Microcystis oligotype), respectively (p = 1x10^-4 in both cases; figure 1, correction of previous figure 5). Hierarchical clustering assignments changed for 2 of the 44 colonies sequenced (colonies MC2019-108 and MC2019-126). The clusters that now contain these colonies correspond better with the sampling date and Microcystis oligotype of the other member colonies. Temporal trends in the relative abundance of key phycosphere taxa as well as their specificity to certain Microcystis oligotypes were not impacted by the analysis error (figure 2, correction of previous figure 4). The relative abundances reported in original Figure 2 (depicted as percent of total community on single colonies, including Microcystis) remain unaltered to show the contribution of non-Microcystis bacterial reads to total reads in single colonies.

The results of the ordination and hierarchical clustering analysis including bulk samples changed more substantially (Results and Discussion subsection “Microcystis phycosphere communities are dissimilar from bulk community samples” in the original publication). Microbial communities were separated into 6 major hierarchical clustering groups rather than 4 (figure 3, correction of previous figure 3). The membership of the clustering is still significantly correlated with sample type (ANOSIM R = 0.6758, P = 0.001), but the correlation is lower than previously calculated (ANOSIM R = 0.7534, P = 1x10^-5). The lower correlation can be explained by the 100 μm retentate communities, which were divided among multiple clusters (clusters A, C, and F) that included single colony and other bulk community sample types in some cases (figure 4, correction of previous figure S11). The clustering of samples within these three clusters is correlated with the relative abundance of Microcystis (Mantel ρ = 0.5037, P = 1x10^-4, figure 5), which suggests that non-Microcystis communities in bulk phytoplankton seston heavily dominated by Microcystis can more closely resemble non-Microcystis communities associated with some single Microcystis colonies than described in the original manuscript. However, the clustering of single colony-associated communities together with 100 μm retentate communities was limited to 17 colonies in hierarchical cluster C (figures 3 and 4), and non-Microcystis communities ordinated along NMDS1 are separated along a gradient transitioning from single colonies to bulk community samples (figure 3). When viewed with NMDS2, only 2 single colony communities appear to overlap with 100 μm retentate communities along NMDS1, while 5 single colony communities appear to overlap with 100 μm retentate samples along NMDS1 when viewed with NMDS3 (figure 3). Furthermore, the relative abundance of Microcystis in 100 μm retentate samples that clustered with whole water and < 105 μm samples was not significantly higher than that of 100 μm retentate samples that clustered with single Microcystis colonies (Pairwise Wilcox Test, P > 0.05, figure 5). Therefore, even when the relative abundance of Microcystis in bulk samples is high, differences in bacterial community composition between bulk phytoplankton seston and single Microcystis colonies cannot be explained solely by Microcystis relative abundance in bulk seston, which supports the original conclusion of the manuscript that bacterial communities not associated with the Microcystis phycosphere are a major constituent of bacterial communities in bulk seston.

Shifts in non-Microcystis communities in whole Lake Erie water are still significantly correlated with month (ANOSIM R remained at ~0.36, P = 1x10^-4; figure 6, correction of previous figure S9) and chlorophyll a concentration (Mantel ρ decreased from ~0.40 to ~0.37 while P remained at 1x10^-4; figure 7, correction of previous figure S10).

The updated results do not substantially change the implications or interpretation of the study, and the updated analysis is included in the GitHub site (https://github.com/Geo-omics/Characterizing-individual-Microcystis-colony-phycosphere-communities) in addition to the originally submitted analysis. We sincerely apologize for any inconvenience that this oversight has caused readers.

在我们的手稿被接受后，我们意识到我们用于分析与微囊藻相关的细菌群落的方法存在错误。我们只是在计算相对丰度之后而不是之前从微囊藻中删除了 DNA 序列读数。反过来，微囊藻丰度可能会在排序和层次聚类中对群落组成的结果引入错误。

为了评估这一点，我们使用微囊藻读取删除后计算的微生物组的相对丰度值重新计算层次聚类和排序分析。单独使用单个微囊藻菌落的分析结果没有显着变化。与群落差异和采样日期和微囊藻寡型的相关性仍然显着，ANOSIM R 值分别从 0.41 增加到 0.43（采样日期）和 0.49 增加到 0.53（微囊藻寡型）（p = 1x10 ^-4在这两种情况下；图 1，对之前图 5 的修正）。已测序的 44 个菌落中的 2 个（菌落 MC2019-108 和 MC2019-126）的分级聚类分配发生了变化。现在包含这些菌落的集群与其他成员菌落的采样日期和微囊藻寡型更好地对应。关键藻类群相对丰度的时间趋势以及它们对某些微囊藻寡型的特异性不受分析误差的影响（图 2，对之前图 4 的更正）。原始图 2 中报告的相对丰度（描绘为单个菌落上总群落的百分比，包括微囊藻）保持不变，以显示非微囊藻的贡献 单个菌落中的细菌读数到总读数。

包含大量样本的排序和层次聚类分析的结果发生了更大的变化（原始出版物中的结果和讨论小节“微囊藻藻类群落与大量群落样本不同”）。微生物群落被分成 6 个主要的层次聚类组，而不是 4 个（图 3，对之前图 3 的更正）。聚类的成员仍然与样本类型显着相关（ANOSIM R = 0.6758，P = 0.001），但相关性低于先前计算的（ANOSIM R = 0.7534，P = 1x10 ^-5）。较低的相关性可以通过 100 μm 滞留物群落来解释，这些群落被分成多个集群（集群 A、C 和 F），在某些情况下包括单菌落和其他大量群落样本类型（图 4，更正之前的图 S11 ）。样品的这三个簇内的群集与相对丰度相关的微囊（曼特尔ρ= 0.5037，P = 1×10 ^-4，图5），这表明非微囊社区散装浮游植物浮游物很大程度上受支配微囊能更紧密地类似于与某些单一微囊藻相关的非微囊藻群落殖民地比原始手稿中描述的要多。然而，单个菌落相关群落与 100 μm 滞留群落的聚类仅限于分层聚类 C 中的 17 个菌落（图 3 和 4），并且沿 NMDS1 排列的非微囊藻群落沿从单个菌落到大宗社区样本（图 3）。当用 NMDS2 观察时，只有 2 个单菌落群落似乎与 NMDS1 沿线的 100 μm 滞留物群落重叠，而当使用 NMDS3 观察时，5 个单菌落群落似乎与 NMDS1 沿线的 100 μm 滞留物样本重叠（图 3）。此外，微囊藻的相对丰度在与全水和 < 105 μm 样品聚集的 100 μm 滞留物样品中，与与单个微囊藻菌落聚集的 100 μm 滞留物样品相比没有显着升高（Pairwise Wilcox Test，P > 0.05，图 5）。因此，即使散装样品中微囊藻的相对丰度很高，散装浮游植物群落和单个微囊藻群落之间细菌群落组成的差异也不能仅仅用散装群落中微囊藻的相对丰度来解释，这支持了手稿的原始结论，即细菌与微囊藻无关的社区 藻圈是大块沉积物中细菌群落的主要组成部分。

在非转移微囊在整个伊利湖水社区仍然显著与月相关（ANOSIMř保持在〜0.36，P = 1×10 ^-4 ;图6中，上图S9的校正）和叶绿素一个浓度（曼特尔ρ从降低〜 0.40 到 ~0.37，而P保持在 1x10 ^-4；图 7，对之前图 S10 的修正）。

更新的结果不会实质性地改变研究的含义或解释，更新的分析包含在 GitHub 站点 (https://github.com/Geo-omics/Characterizing-individual-Microcystis-colony-phycosphere-communities)除了最初提交的分析。对于此疏忽给读者带来的任何不便，我们深表歉意。