Latitudinal changes in RMR and body mass
We found that the mass-adjusted RMR exhibited significant geographic differences during the winter, with a significant positive correlation between mass-adjusted RMR and latitude (Fig. 2A; p = 0.0011, r = 0.33). Although this relationship was statistically nonsignificant in the breeding season, the trend was also positive (Fig. 2A; p = 0.065, r = 0.25). Consistently, body mass was also positively related to the latitude in the both breeding and wintering seasons, consistent with the prediction of Bergmann's rule (Fig. 2B; breeding season: p = 0.045, r = 0.27, wintering season: p = 0.0012, r = 0.25).
In addition, body mass and mass-adjusted RMR varied between seasons. The mass-adjusted RMR of the two populations (NC and BJ) were significantly elevated in the wintering season compared to the breeding season (Fig. 2C; t test, p < 0.05), while no difference between the two seasons was found of in the GL and DH populations (Fig. 2C; t test, p > 0.05). The body mass of the four geographic populations was also significantly higher in the wintering season than that in the breeding season (Fig. 2D, t test, p < 0.05). Together, our results showed both latitudinal and seasonal variation in both mass-adjusted RMR and body mass, with the variation in body mass slightly more dramatic than that in mass-adjusted RMR.
Latitudinal Variation In Microbial Diversity In Different Geographic Populations
To identify whether the gut microbiota exhibited dynamic variations in response to latitudinal gradient, we analyzed 16S rRNA sequences of 135 fecal samples from four groups (GL, NC, DH and BJ) in the both breeding and wintering seasons (Table S1). After quality control, a total of 6,170,643 high-quality sequences were retained for all samples and an average of 45,708 sequences were obtained per sample (Table S2).
Linear mixed-effects model analysis indicated that latitude was the major factor shaping the microbial community diversity of light-vented bulbul (Table 1; FASV = 52.6421, p < 0.001; FShannon = 83.709, p < 0.001). In the breeding season, although the ASV index was not significantly different among populations (Fig. 3A; Kruskal-Wallis test, p = 0.21), the value was higher in the southern populations (GL and NC) than that in the northern populations (DH and BJ). The Shannon index was significantly higher in the southern populations (GL and NC) than that in the northern populations (DH and BJ) (Fig. 3A; Wilcoxon test, p < 0.05). In addition, based on PERMANOVA of the Bray-Curtis distance matrix, we found that microbial communities were significantly different among the four groups (p = 0.001). The PCoA plot showed that the northern populations (DH and BJ) were clustered together, and were dramatically different from the others (Fig. 3B). We further found that the relative abundances of the four most abundant phyla (Proteobacteria, Firmicutes, Bacteroidetes and Desulfobacterota) were significantly different among the four groups (Fig. 3C and Fig. S1; Kruskal-Wallis test, p < 0.05). Core microbial analysis showed that only 3 species (Acinetobacter radioresistens, Bacillus nealsonii and Chryseobacterium stagni) had significantly increased relative abundances in the northern populations (DH and BJ) compared with the southern population (GL and NC), while 7 species were significantly decreased in the northern populations (Fig. 3D; Wilcoxon test, p < 0.05). Overall, during the breeding season, the diversity of the northern populations was lower than that in the southern populations, and the microbial composition was also different among populations.
Table 1
Linear mixed-effects model by restricted maximum likelihood (REML) for alpha diversity of ASV and Shannon index.
|
d.f.
|
F
|
p-value
|
Main effects – ASV
|
AIC = 1492.819
|
Latitude
|
132
|
52.6421
|
< 0.001
|
Season
|
132
|
9.1736
|
0.003
|
Main effects – Shannon
|
AIC = 323.6687
|
Latitude
|
132
|
83.709
|
< 0.001
|
Season
|
132
|
99.083
|
< 0.001
|
d.f., degrees of freedom; AIC, Akaike information criterion |
Likewise, a latitudinal pattern of gut microbiota was observed during the wintering season. Both ASV and Shannon indices were significantly varied among the four groups (Fig. 4A; Kruskal-Wallis test, p = 1e-09, p = 2.3e-09), and the two indices were significantly higher in the southern populations (GL and NC) than that in the northern populations (DH and BJ) (Wilcoxon test, p < 0.05). Microbial communities (i.e., beta-diversity) were significantly different across the four populations, based on PERMANOVA of the Bray-Curtis distance matrix (p = 0.001). The principal-coordinate analysis (PCoA) graphs clearly illustrated that the samples of northern birds (DH and BJ) were clustered together, and GL samples were clustered together, whereas NC samples were discrete in the two clusters (Fig. 4B). In addition, the relative abundances of the top five phyla (Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteriota and Desulfobacterota) were significantly different in all four groups (Fig. 4C and Fig. S1; Kruskal-Wallis test, p < 0.05). The core microbiome (occurring in > 50% of individuals) showed geographic differences (Fig. 4D). Specifically, the relative abundance of 11 species were significantly decreased in northern populations (DH and BJ), while those of 5 species (Alistipes shahii, Bacteroides stercoris, Bacteroides vulgatus, Parabacteroides merdae and Bacteroides spp.) were significantly elevated in southern populations (GL and NC) (Fig. 3D; Wilcoxon test, p < 0.05). Therefore, during the wintering season, the diversity of the northern populations decreased compared with that in the southern populations, and the microbial composition was also different between the northern and southern populations.
Furthermore, bacterial functions were predicted using the MetaCyc database. In both seasons, more pathways associated with metabolism and biosynthesis increased as latitude increased (Table S3). In the breeding season, the abundance of 143 pathways increased along latitudinal gradients (Mann-Kendall test, FDR < 0.01, z > 0), while 54 pathways decreased (Mann-Kendall test, FDR < 0.01, z < 0). In the wintering season, the abundance of 171 pathways increased with increasing latitude (Mann-Kendall test, FDR < 0.01, z > 0), whereas only 61 pathways decreased (Mann-Kendall test, FDR < 0.01, z < 0).
Identification Of Gut Microbiota Associated With The Rmr
We further analyzed the gut microbiota in relation to the RMR of the light-vented bulbul. In the breeding season, 23 genera were significantly correlated with the mass-adjusted RMR (22 in positive correlation and 1 in negative correlation), of which only 4 genera were also related to latitude (Fig. 5A). However, in the wintering season, we found more genera (71 genera: 20 in positive correlation, 51 in negative correlation) that were significantly correlated with the mass-adjusted RMR, of which 69 genera were related to latitude (Fig. 5B). Importantly, we found the overlapping 19 genera, such as Bacteroides, Lachnospira, Alistipes and Faecalibacterium, that were positively related to both latitude and mass-adjusted RMR in the wintering season to be significantly enriched (Fig. 5B). Our results showed that more genera correlated with the mass-adjusted RMR and latitude in the wintering season than that in the breeding season, which may play a critical role in metabolic thermogenesis, especially in winter.
Variation In Microbial Diversity Between The Breeding And Wintering Seasons
To evaluate the dynamics of the gut microbiota in the breeding and wintering seasons, a Venn diagram of amplicon sequence variants (ASVs) was constructed, as shown in Fig. 6A. We found a total of 359 shared ASVs, and 989 and 654 ASVs were specific to breeding samples and wintering samples, respectively. Linear mixed-effects model analysis indicated that seasonal change was also important to the microbial community diversity (Table 1; FASV = 9.1736, p = 0.003; FShannon = 99.083, p < 0.0001). Specifically, in northern populations (DH and BJ), the ASV index was dramatically higher in the breeding season than that in the wintering season (Fig. 3A and 4B; Wilcoxon test, p < 0.01), while there was no significant difference in southern populations (GL and NC) (Fig. 3A and 4B; Wilcoxon test, p > 0.05). The Shannon index was strikingly lower in the breeding season than that in the wintering season in the three groups (NC, DH and BJ) (Fig. 3A and 4B; Wilcoxon test, p < 0.01), while there was no significant difference in the GL group (Fig. 3A and 4B; Wilcoxon test, p > 0.05). Additionally, PERMANOVA analysis revealed that microbial communities (beta-diversity) were significantly different between the two seasons (p = 0.001). The PCoA plot showed that the breeding season samples were clustered together, and were obviously different from the wintering samples (Fig. 6B). The constrained PCoA1 and PCoA2 analysis explained 34.83% and 24.86% of the variation in birds, respectively. Therefore, the diversity of the microbiota was different in both seasons.
Significant differences were also detected at the phylum level during the breeding and wintering seasons of light-vented bulbul. Firmicutes abundance was significantly higher in the wintering season than that in the breeding season in the four groups (Fig. 3C and 4C; Wilcoxon test, p < 0.01). Similarly, Bacteroidetes abundance was significantly elevated in the wintering season in the three groups (Fig. 3C and 4C; Wilcoxon test, p < 0.01), except for the GL group. In contrast, Proteobacteria abundance was significantly decreased in the wintering season (Fig. 3C and 4C; Wilcoxon test, p < 0.01).