Linking phytoplankton community metabolism to the individual size distribution

Abstract Quantifying variation in ecosystem metabolism is critical to predicting the impacts of environmental change on the carbon cycle. We used a metabolic scaling framework to investigate how body size and temperature influence phytoplankton community metabolism. We tested this framework using phytoplankton sampled from an outdoor mesocosm experiment, where communities had been either experimentally warmed (+ 4 °C) for 10 years or left at ambient temperature. Warmed and ambient phytoplankton communities differed substantially in their taxonomic composition and size structure. Despite this, the response of primary production and community respiration to long‐ and short‐term warming could be estimated using a model that accounted for the size‐ and temperature dependence of individual metabolism, and the community abundance‐body size distribution. This work demonstrates that the key metabolic fluxes that determine the carbon balance of planktonic ecosystems can be approximated using metabolic scaling theory, with knowledge of the individual size distribution and environmental temperature.


Figure S1. Photosynthesis irradiance curves of warm (red) and ambient (black) mesocosm communities in the (a) ambient and (b) warmed incubators. Values for
the photosynthetic maximum (Eq S1) were used as the metabolic rate data to which Eq 2 was fitted. Faded points and lines represent the raw measurements and individual fits to each photosynthesis irradiance curve. The bold, thicker lines represent the fit of the average parameter values from all of the individual parameter values. tops and bottoms of box-whisker plots represent the 75 th and 25 th percentiles and the white horizontal line represents the median.

Additional methods on mesocosm setup, measurements of metabolic flux and the individual size distribution
Overview of long-term mesocosm experiments

Analysing sequencing data
Sequence data was analysed in R (v 3.3.2) (Team 2014) using the packages 'dada2' and 'phyloseq' (Callahan et al. 2015(Callahan et al. , 2016. Reads were truncated at 250 bp. We then followed the full stack workflow to estimate error rates, infer and the merge sequences, construct a sequence table, remove chimeric sequences and assign taxonomy (Callahan et al. 2016). Sequence inference was done by pooling all the samples to improve the detection of rare variants that are seen just once or twice in an individual sample, but many times across all samples. We combined multiple rRNA databases to create a new database from which taxonomy was assigned. PhytoREF (Decelle et al. 2015), provides a reference database for the plastidial 16S rRNA gene for photosynthetic eukaryotes.
Consequently, in a single amplicon sequencing run of the 16S v4 region we quantified both bacterial and eukaryotic autotroph diversity. We combined the PhytoREF database with the Ribosomal Database Project (Cole et al. 2014) that contains ribosomal RNA sequences of prokaryotes and 2700 16S rDNA cyanobacterial references (Decelle et al. 2015). Using CD-HIT (Li & Godzik 2006) we created a clustered database that aligned sequences with >97% similarity. We then preferentially assigned clustered sequences as originating from 1) PhytoREF, 2) cyanobacteria or 3) the Ribosomal Database Project as previous work has shown erroneous assignments of chloroplast plastidial sequences as being of bacterial origin (Decelle et al. 2015). We used the R package 'taxise' (Chamberlain & Szöcs 2013) to reconcile each species in the reference database with its higher taxonomy. Samples were removed if represented by fewer than 1000 reads and the remaining samples were standardised to the total number of amplicon sequence variants (ASVs; Callahan et al. 2017) through rarefaction to account for biases associated with differences in sequencing effort. We then selected ASVs (amplicon sequence variants) corresponding to autotrophic taxa, which resulted in samples from 37 of the 40 communities that could be used for downstream analysis.

Measuring community metabolism
After ~30 days of culture (enough time for acclimation responses to short-term warming to occur [1-10 generations in phytoplankton] (Staehr and Birkeland, 2006)), we measured metabolism at incubator temperature (16 ºC or 20 ºC) on communities below carrying capacity. Aliquots (30 mL) of each community were concentrated through centrifugation (~1500 rpm for 30 minutes at 4 ºC) and resuspended in 5 mL and acclimatised to the measurement temperature for 15 minutes in the dark prior to measuring metabolic flux. Primary production and community respiration were measured through oxygen evolution in the light and oxygen consumption in the dark respectively on a Clark-type electrode (Hansatech Ltd, King's Lynn UK Chlorolab2).
Rates of community respiration were measured for two minutes in the dark at the end of each PI curve to ensure respiration was not limited by available photosynthate during the measurement period.
Each individual PI curve was fit to a modification of the Eiler's curve for photoinhibition that incorporates community respiration. This model allows for negative rates of net primary production at low light levels even when community respiration is greater than gross primary production (Eilers and Peeters, 1988): where ( ), is the rate of net primary production at irradiance, , RST is the maximal rate of net primary production at optimal light, UVW , controls the gradient of the initial slope and is community respiration, the rate of oxygen consumption in the dark. Gross primary production (GPP) at light saturation was then found by adding community respiration onto maximal net primary production ( = RST + ).
Gross primary production at light saturation and measured community respiration were subsequently used in the metabolic scaling framework.