Research ArticleAPPLIED ECOLOGY

Phosphorus, not nitrogen, limits plants and microbial primary producers following glacial retreat

See allHide authors and affiliations

Science Advances  23 May 2018:
Vol. 4, no. 5, eaaq0942
DOI: 10.1126/sciadv.aaq0942
  • Fig. 1 Three different measures of plant growth in response to a one-time addition of nutrients to soils (in August 2010) that were uncovered by the Puca Glacier 3 years preceding the start of the experiment.

    This site was initially barren, but plots with added P exhibited exponential increase in the number of plants within (A). By the fifth year (2015), plant biomass was high enough that it was feasible to measure the percent cover by plants within the plots. This was done manually using the p-i method and digitally using mNDVI in 2015 and 2016 (B). The p-i method yielded almost identical results for 2015 (left) and 2016 (right), and analysis of variance (ANOVA) showed an overwhelmingly large effect size of P addition (ω2, inset, pie charts). mNDVI analysis showed very similar results (bottom of Fig. 3B), although the effect size of P was smaller (***P < 0.001, **P < 0.01). The effect of N addition was not statistically significant in any of these four analyses.

  • Fig. 2 Near-infrared–enhanced photographs of field plots and mNDVI comparison.

    In these false-color photographs, the visible light channels of the sensor have been averaged and rendered as a grayscale background, and the near-infrared (NIR) channel has been overlayed in red. Control plots (the dashed line is the plot boundary) (A) did not contain any visible plants, but some pixels in the plot area still had mNDVI values above the threshold of 0.1 (B), possibly indicating detection of microbial phototrophs. However, plots with P added (C) contained many plants, which were detected with the NIR sensor and statistically differentiated from barren areas using mNDVI. The mNDVI distribution plot (B) shows that the +P plot has many more pixels above the mNDVI threshold (dashed red line) than the control plot.

  • Fig. 3 Shifts in cyanobacterial communities from nutrient addition and nutrient limitation of microbial phototrophs.

    Both N and P addition changed cyanobacterial community structure, but P addition caused a marked increase in the relative abundance of N-fixing Nostocales (pink) (A). Nostoc were visually apparent across several orders of magnitude in samples from the field plots as large spherical colonies (B) or strands of cells (C) and were identified using shotgun-metagenomic and amplicon sequencing as Nostoc commune (16S ribosomal RNA gene was 100% identical to GenBank accession no. KY380004). Microscopic counts of spherical Nostoc colonies showed P limitation (D), as did microscopic percent cover based on chlorophyll autofluorescence (E and F), similar to the results for macroscopic plants (Figs. 1 and 2) (***P < 0.001, **P < 0.01).

  • Fig. 4 Results of laboratory bioassays.

    Laboratory bioassay experiments (28, 29) showed P limitation of early successional soils from the High Andes (left) and from the receding Toklat Glacier in central Alaska (right). The percent cover of microbial phototrophs over time is significantly higher in treatments with P added and is lower when P is not added (A). Repeated-measures ANOVA shows a much larger effect size of P than of N as well (B) (***P < 0.001, **P < 0.01). Furthermore, when N is added in the absence of added P, that N was not assimilated, indicating that the microcosms were P-limited even in their ability to immobilize N (C). FOV, fields of view [see Darcy and Schmidt (29)].

Supplementary Materials

  • Supplementary Materials

    This PDF file includes:

    • table S1. Element analysis results from x-ray fluorescence analysis.
    • fig. S1. Aerial view of the Puca Glacier field site, showing plot layout and orientation.

    Download PDF

    Files in this Data Supplement: