Table 1 Contributions of the 22 biotic and abiotic predictors to boosted regression tree models relating latitudinal trends to the (relative) abundance of the ten secondary superclasses + CO2-fixation pathways and PP at Depth 1 (n = 75) along the GO-SHIP P15S transect.

From: Metabolic pathways inferred from a bacterial marker gene illuminate ecological changes across South Pacific frontal boundaries

Pathway/process

Parameter 1

Parameter 2

Parameter 3

Parameter 4

Parameter 5

cv

PP

[Nanoplankton] (26%)

[Hex.Fuco] (26%)

[Pras] (16%)

[POC] (6%)

[NH4+] (4%)

0.91 ± 0.02

CO2-fixation

Temperature (25%)

[Total pigment] (20%)

[Nanoplankton] (15%)

[chl b] 13%)

[O2] (7%)

0.842 ± 0.04

Energy metabolism

Temperature (12%)

Salinity (12%)

[NH4+] (10%)

[chl b] (9%)

[Microplankton] (8%)

0.757 ± 0.08

Nucleotide biosynthesis

Salinity (20%)

δ13C (16%)

Temperature (15%)

[Nanoplankton] (7%)

[Silicate] (6%)

0.725 ± 0.13

Cell structure and cell wall biosynthesis.

Temperature (56%)

[NH4+] (18%)

[Zeaxanthin] (4%)

δ13C (3%)

[Picoplankton] (3%)

0.955 ± 0.01

Amino acid biosynthesis

[Picoplankton] (34%)

Temperature (11%)

[Total pigment] (11%)

chl b (8%)

δ15N (5%)

0.796 ± 0.05

Secondary metabolite biosynthesis

chl b (29%)

[Total pigment] (20%)

[Nanoplankton] (8%)

δ15N (7%)

C:N ratios (5%)

0.822 ± 0.05

Cofactor biosynthesis

[Total pigment] (43%)

Salinity (8%)

[PON] (7%)

[Silicate] (7%)

δ15N (4%)

0.837 ± 0.03

Vitamin biosynthesis

Temperature (30%)

[NO3] (17%)

[Total pigment] (13%)

[NO2] (9%)

[Fuco] (6%)

0.946 ± 0.02

Lipid biosynthesis

Temperature (45%)

[NO3] (9%)

[NO2] (6%)

Salinity (5%)

[DV.Chl-a] (5%)

0.956 ± 0.01

Carbohydrate biosynthesis

Temperature (56%)

[NH4+] (18%)

[Zeaxanthin] (4%)

δ13C (3%)

Salinity (3%)

0.945 ± 0.01

Fermentation

Temperature (28%)

[Picoplankton] (23%)

[NO2] (11%)

[Fuco] (10%)

δ13C (4%)

0.928 ± 0.02

  1. The relative contribution of each parameter is shown in brackets (%). The cumulative contribution of all listed parameters within a pathway/PP amounts to ~60–80% of the model predictability, except for the energy metabolism pathway, which shows a cumulative predictability of 51%. Square brackets are used to represent concentrations. A full breakdown of all predictor variables can be found in Supplementary Data 2.
  2. cv coefficient of variation.