skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


  • NSF Public Access
  • Search Results
  • Can machine learning extract the mechanisms controlling phytoplankton growth from large-scale observations? – A proof-of-concept study
Title: Can machine learning extract the mechanisms controlling phytoplankton growth from large-scale observations? – A proof-of-concept study
Abstract. A key challenge for biological oceanography is relating the physiologicalmechanisms controlling phytoplankton growth to the spatial distribution ofthose phytoplankton. Physiological mechanisms are often isolated by varyingone driver of growth, such as nutrient or light, in a controlled laboratorysetting producing what we call “intrinsic relationships”. We contrastthese with the “apparent relationships” which emerge in the environment inclimatological data. Although previous studies have found machine learning(ML) can find apparent relationships, there has yet to be a systematic studyexamining when and why these apparent relationships diverge from theunderlying intrinsic relationships found in the lab and how and why this may depend on the method applied. Here we conduct a proof-of-concept studywith three scenarios in which biomass is by construction a function oftime-averaged phytoplankton growth rate. In the first scenario, the inputsand outputs of the intrinsic and apparent relationships vary over thesame monthly timescales. In the second, the intrinsic relationships relateaverages of drivers that vary on hourly timescales to biomass, but theapparent relationships are sought between monthly averages of these inputsand monthly-averaged output. In the third scenario we apply ML to the outputof an actual Earth system model (ESM). Our results demonstrated that whenintrinsic and apparent relationships operate on the same spatial andtemporal timescale, neural network ensembles (NNEs) were able to extract theintrinsic relationships when only provided information about the apparentrelationships, while colimitation and its inability to extrapolate resulted in random forests (RFs) diverging from the true response. Whenintrinsic and apparent relationships operated on different timescales (aslittle separation as hourly versus daily), NNEs fed with apparentrelationships in time-averaged data produced responses with the right shapebut underestimated the biomass. This was because when the intrinsicrelationship was nonlinear, the response to a time-averaged input differedsystematically from the time-averaged response. Although the limitationsfound by NNEs were overestimated, they were able to produce more realisticshapes of the actual relationships compared to multiple linear regression.Additionally, NNEs were able to model the interactions between predictorsand their effects on biomass, allowing for a qualitative assessment of thecolimitation patterns and the nutrient causing the most limitation. Futureresearch may be able to use this type of analysis for observational datasetsand other ESMs to identify apparent relationships between biogeochemicalvariables (rather than spatiotemporal distributions only) and identifyinteractions and colimitations without having to perform (or at leastperforming fewer) growth experiments in a lab. From our study, it appearsthat ML can extract useful information from ESM output and could likely doso for observational datasets as well.  more » « less
Award ID(s):
1756568
PAR ID:
10338099
Author(s) / Creator(s):
;
Date Published:
Journal Name:
Biogeosciences
Volume:
18
Issue:
6
ISSN:
1726-4189
Page Range / eLocation ID:
1941 to 1970
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; (, Geoscientific Model Development)
    Abstract. Earth system models (ESMs) are useful tools forpredicting and understanding past and future aspects of the climate system.However, the biological and physical parameters used in ESMs can have widevariations in their estimates. Even small changes in these parameters canyield unexpected results without a clear explanation of how a particularoutcome was reached. The standard method for estimating ESM sensitivity isto compare spatiotemporal distributions of variables from different runs ofa single ESM. However, a potential pitfall of this method is that ESM outputcould match observational patterns because of compensating errors. Forexample, if a model predicts overly weak upwelling and low nutrientconcentrations, it might compensate for this by allowing phytoplankton tohave a high sensitivity to nutrients. Recently, we demonstrated that neuralnetwork ensembles (NNEs) are capable of extracting relationships betweenpredictor and target variables within ocean biogeochemical models. Beingable to view the relationships between variables, along with spatiotemporaldistributions, allows for a more mechanistically based examination of ESMoutputs. Here, we investigated whether we could apply NNEs to help usdetermine why different ESMs produce different spatiotemporal distributionsof phytoplankton biomass. We tested this using three cases. The first andsecond case used different runs of the same ESM, except that the physicalcirculations differed between them in the first case, while the biologicalequations differed between them in the second. Our results indicated thatthe NNEs were capable of extracting the relationships between variables fordifferent runs of a single ESM, allowing us to distinguish betweendifferences due to changes in circulation (which do not changerelationships) from changes in biogeochemical formulation (which do changerelationships). In the third case, we applied NNEs to two different ESMs.The results of the third case highlighted the capability of NNEs to contrastthe apparent relationships of different ESMs and some of the challenges itpresents. Although applied specifically to the ocean components of an ESM,our study demonstrates that Earth system modelers can use NNEs to separatethe contributions of different components of ESMs. Specifically, this allowsmodelers to compare the apparent relationships across different ESMs andobservational datasets. 
    more » « less
  2. ; ; ; (, Limnology and Oceanography Letters)
    Abstract Surface‐ocean mixing creates dynamic light environments with predictable effects on phytoplankton growth but unknown consequences for predation. We investigated how variations in average mixed‐layer (ML) irradiance shaped plankton trophic dynamics by incubating a Northwest‐Atlantic plankton community for 4 days at high (H) and low (L) light, followed by exposure to either sustained or reversed light intensities. In deep‐ML (sustained L), phytoplankton biomass declined (μ= −0.2 ± 0.08 d−1) and grazing was absent. In shallow‐ML (sustained H), growth exceeded grazing (μ= 0.46 ± 0.07 d−1;g= 0.32 ± 0.04 d−1). In rapidly changing ML‐conditions simulated by switching light‐availability, growth and grazing responded on different timescales. During rapid ML‐shoaling (L to H),μimmediately increased (0.23 ± 0.01 d−1) with no change in grazing. During rapid ML‐deepening (H to L),μimmediately decreased (0.02 ± 0.09 d−1), whereas grazing remained high (g= 0.38 ± 0.05 d−1). Predictable rate responses of phytoplankton growth (rapid) vs. grazing (delayed) to measurable light variability can provide insights into predator‐prey processes and their effects on spatio‐temporal dynamics of phytoplankton biomass. 
    more » « less
  3. ; ; ; (, Global Biogeochemical Cycles)
    Abstract Phytoplankton stoichiometry modulates the interaction between carbon, nitrogen and phosphorus cycles. Environmentally driven variations in phytoplankton C:N:P can alter biogeochemical cycling compared to expectations under fixed ratios. In fact, the assumption of fixed C:N:P has been linked to Earth System Model (ESM) biases and potential misrepresentation of responses to future change. Here we integrate key elements of the Adaptive Trait Optimization Model (ATOM) for phytoplankton stoichiometry with the Carbon, Ocean Biogeochemistry and Lower Trophics (COBALT) ocean biogeochemical model. Within a series of global ocean‐ice‐ecosystem retrospective simulations, ATOM‐COBALT reproduced observations of phytoplankton N:P, and compared to static ratios, exhibited reduced phytoplankton P‐limitation, enhanced N‐fixation, and increased low‐latitude export, improving consistency with observations and highlighting the biogeochemical implications of dynamic N:P. We applied ATOM‐COBALT to explore the impacts of different physiological mechanisms hypothesized to underlie N:P variation, finding that two mechanisms together drove the observed patterns: proportionality of P‐rich ribosomes in phytoplankton cells to growth rates and reductions in P‐storage during scarcity. A third mechanism which linked temperature with phytoplankton biomass allocations to non‐ribosomal proteins, led only to relatively modest impacts because this mechanism decreased the temperature dependence of phytoplankton growth rates, compensating for changes in N:P. We find that there are quantitative response differences that associate distinctive biogeochemical footprints with each mechanism, which are most apparent in highly productive low‐latitude regions. These results suggest that variable phytoplankton N:P makes phytoplankton productivity and export resilient to environmental changes, and support further research on the physiological and environmental drivers of phytoplankton stoichiometry and biogeochemical role. 
    more » « less
  4. ; ; ; ; ; ; ; (, Journal of Plankton Research)
    Campbell, Lisa (Ed.)
    Abstract Biomass and composition of the phytoplankton community were investigated in the deep-water Gulf of Mexico (GoM) at the edges of Loop Current anticyclonic eddies during May 2017 and May 2018. Using flow cytometry, high-performance liquid chromatography pigments and microscopy, we found euphotic zone integrated chlorophyll a of ~10 mg m−2 and autotrophic carbon ranging from 463 to 1268 mg m−2, dominated by picoplankton (<2 μm cells). Phytoplankton assemblages were similar to the mean composition at the Bermuda Atlantic Time-series Study site, but differed from the Hawaii Ocean Times-series site. GoM phytoplankton biomass was ~2-fold higher at the deep chlorophyll maximum (DCM) relative to the mixed layer (ML). Prochlorococcus and prymnesiophytes were the dominant taxa throughout the euphotic zone; however, other eukaryotic taxa had significant biomass in the DCM. Shallower DCMs were correlated with more prymnesiophytes and prasinophytes (Type 3) and reduced Prochlorococcus. These trends in ML and DCM taxonomic composition likely reflect relative nutrient supply—with ML populations relying on remineralized ammonium as a nitrogen source, and the taxonomically diverse DCM populations using more nitrate. These spatially separated phytoplankton communities represent different pathways for primary production, with a dominance of picoplankton in the ML and more nano- and microplankton at the DCM. 
    more » « less
  5. ; ; (, Journal of Geophysical Research: Oceans)
    In nutrient-limited conditions, phytoplankton growth at fronts is enhanced by winds, which drive upward nutrient fluxes via enhanced turbulent mixing and upwelling. Hence, depth-integrated phytoplankton biomass can be 10 times greater at isolated fronts. Using theory and two-dimensional simulations with a coupled physical-biogeochemical ocean model, this paper builds conceptual understanding of the physical processes driving upward nutrient fluxes at fronts forced by unsteady winds with timescales of 4–16 days. The largest vertical nutrient fluxes occur when the surface mixing layer penetrates the nutricline, which fuels phytoplankton in the mixed layer. At a front, mixed layer deepening depends on the magnitude and direction of the wind stress, cross-front variations in buoyancy and velocity at the surface, and potential vorticity at the base of the mixed layer, which itself depends on past wind events. Consequently, mixing layers are deeper and more intermittent in time at fronts than outside fronts. Moreover, mixing can decouple in time from the wind stress, even without other sources of physical variability. Wind-driven upwelling also enhances depth-integrated phytoplankton biomass at fronts; when the mixed layer remains shallower than the nutricline, this results in enhanced subsurface phytoplankton. Oscillatory along-front winds induce both oscillatory and mean upwelling. The mean effect of oscillatory vertical motion is to transiently increase subsurface phytoplankton over days to weeks, whereas slower mean upwelling sustains this increase over weeks to months. Taken together, these results emphasize that wind-driven phytoplankton growth is both spatially and temporally intermittent and depends on a diverse combination of physical processes. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email