ULEENTARY INORATION - Nature Research · 2017-04-28 · join from ea-utils (Aronesty et al. 2011)....

In the format provided by the authors and unedited.

1

Divergent plant-soil feedbacks could alter future elevation ranges and ecosystem dynamics

SUPPLEMENTARY INFORMATION

Michael E. Van Nuland1*, Joseph K. Bailey1, and Jennifer A. Schweitzer1

1Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville,

Tennessee, 37996, United States of America.

*Corresponding author

Author emails:

[email protected]

[email protected]

[email protected]

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

SUPPLEMENTARY INFORMATIONVOLUME: 1 | ARTICLE NUMBER: 0150

NATURE ECOLOGY & EVOLUTION | DOI: 10.1038/s41559-017-0150 | www.nature.com/natecolevol 1

http://dx.doi.org/10.1038/s41559-017-0150

2

Soil conditioning effects Measures of total soil C, total soil N, and pH had significant soil location effects (conditioned versus unconditioned), but little or no effects of range location (interior versus edge) or soil location × range location interactions (Supplementary Table 2). Conditioned soils that were collected beneath trees had higher amounts of C (F1,50 = 9.0; p < 0.01), N (F1,50 = 4.8; p = 0.03), and more basic pH (F1,50 = 6.4; p = 0.02) compared to unconditioned soils (Supplementary Fig. 2a-c). However, post-hoc analyses reveal that interior trees are responsible for creating these effects. Within range interiors, conditioned soil C was 52% higher (mean = 5.7 ± 0.7 standard error [SE]; z = 3.5, p < 0.01), conditioned soil N was 40% higher (mean = 0.33 ± 0.04 SE; z = 2.8, p = 0.03), and conditioned soil pH was 5% higher (mean = 5.6 ± 0.04 SE; z = 4.5, p < 0.001) relative to paired unconditioned soils. By comparison, edge trees had no conditioning effect on soil C, N, or pH, suggesting there may be less plant-soil conditioning at range limits. In surveying the diversity of dominant rhizosphere bacteria (using next-generation 16S amplicon sequencing) associated with a representative subset of P. angustifolia trees in the field, only the relative abundance of Betaproteobacteria (7.5% overall) showed a difference in conditioned soils between interior and edge range locations (Supplementary Fig. 2d). Betaproteobacteria abundance was 63% higher in edge (mean = 0.09 ± 0.01 SE) versus interior tree-conditioned soils (F1,11 = 6.4, p = 0.05). This observation could be due to differences in plant-soil conditioning or to site differences between range interiors and edges. No other dominant bacterial class (including Actinobacteria [9%], Alphaproteobacteria [19%], Gammaproteobacteria [18%], or Synergistia [4%]) differed by range location (see Supplementary Table 3). Moreover, differences in conditioning between range locations could be related to tree age since there is a marginal effect of interior trees being 20.7% larger (and presumably older) than edge trees (F1,210 = 3.7, p = 0.06), and this pattern is consistent across populations (population × range location: F6,210 = 1.4, p = 0.2). Allometric equation for aboveground biomass of Populus angustifolia cuttings Allometric equations use plant traits to predict biomass when destructive sampling is not possible. Recent work has constructed allometric equations for Populus spp. that are close relatives of P. angustifolia (Chojnacky et al. 2014), as well as for P. angustifolia saplings (Lojewski et al. 2009). However, both approaches use mature trees or saplings that relate DBH to plant biomass; therefore, these equations are not intended to accurately scale down to the size of trees in our greenhouse experiment. As a result, we established an allometric equation using six different P. angustifolia genotypes that were collected at three time periods (June 2012, 2013, and 2014) and grown under equal greenhouse conditions at the University of Tennessee. We measured height and basal stem diameter from these six plants in September 2014 before the aboveground portion was dried at 72° C for 48 h and dry biomass was measured. We calculated plant cross sectional areas from basal stem diameter measurements (π(0.5*diameter)2), and multiplied area by plant height to quantify total stem volume (mm3). We then used a linear regression to test the relationship between stem volume and aboveground biomass. Stem volume predicted more than 98% of the variation in aboveground biomass (Supplementary Fig. 1); as a result, we created the following allometric equation: Aboveground biomass (g) = (stem volume (mm3) * 0.41899) - 2.40137. Measurements of height and basal stem diameter from trees in the feedback experiment were recorded in May 2015 to calculate final aboveground biomass using this allometric equation; measures of total growth responses were calculated as final - initial biomass. Calculations of biomass growth that were negative (due to calculation errors) were




http://dx.doi.org/10.1038/s41559-017-0150

3

removed from analyses, resulting in a total of 609 trees that were incorporated in experimental analyses. Analysis of maternal effects on tree growth traits Maternal effects transferred from trees in the field to tree cuttings used in the greenhouse experiments may influence growth traits measured in this study. We tested whether “maternal” tree size (DBH) corresponded to measures of initial aboveground biomass, final aboveground biomass, or aboveground biomass growth in our experiment. First we calculated genotype-means for each growth traits across all treatment types to compare to DBH of the corresponding genotype in the field (N = 209). Next, we created a model with the three growth traits as the response variable, DBH as the fixed effect, and population as a random effect. Any relationship between DBH and growth traits in the greenhouse would suggest that maternal tree size could be influencing growth responses in the feedback experiments and might be confounding effects from the various treatments. We find no patterns of DBH relating to any of the growth traits measured in the greenhouse. Specifically, there was no relationship between DBH and initial aboveground biomass (r2 = 0.0, p = 0.4), final aboveground biomass (r2 = 0.01, p = 0.9), or aboveground biomass growth (r2 = 0.03, p = 0.3). These results suggest that maternal effects likely play a minimal role affecting variation in P. angustifolia growth in our study. Amplicon sequencing and bioinformatic processing Soil samples were sent to the Environmental Genetics and Genomics Laboratory at Northern Arizona University for sequencing of marker gene amplicon pools. Samples were amplified in triplicate PCR reactions for the 16S v4 region using the universal prokaryotic primers 515F and 806R. First round reactions were performed in triplicate in 384 well plates. The 8 µL volumes contained the following: 1 µM each primer (Eurofins MWG Operon, LLC, Louisville, KY), 200 µM each dNTP (Phenix Research, Candler, NC), 0.01 U/µL Phusion Hot Start II DNA Polymerase (Life Technologies, Carlsbad, CA), 1X HF Phusion Buffer (Life Technologies), 3 mM MgCl2, 6% glycerol, and 1 µL normalized template DNA. Cycling conditions were: 2 minutes at 95°C followed by 20 cycles of 30 seconds at 95°C, 30 seconds at 55°C, 4 minutes at 60°C. Triplicate reactions for each sample were pooled by combining 4 µL from each, and 2 µL was used to check for results on a 1% agarose gel. The remainder was diluted 10-fold and used as template in a second PCR reaction in which 12 base Golay indexed sequencing tails were added. Second round reaction conditions were identical to the first round except only one reaction was conducted per sample and only 15 total cycles were performed. Indexed PCR products were purified using a 1:1 ratio of 18% PEG and carboxylated magnetic beads, quantified by PicoGreen fluorescence, and an equal mass of each sample was combined into a final sample pool. The pool was purified and concentrated, and subsequently quantified by quantitative PCR against Illumina DNA Standards (Kapa Biosystems, Wilmington, MA). Sequencing was carried out on a MiSeq Desktop Sequencer (Illumina Inc, San Diego, CA) running in paired end 2x250 mode. We processed 16S amplicon data using akutils (https://github.com/alk224/akutils-v1.2), which includes modifications to a QIIME 1.9.1 workflow (Caporaso et al. 2010) since it has been shown that the default settings can significantly overestimate microbial diversity (Krohn et al. 2016a). Prior to sample processing, primers were trimmed and PhiX Control was identified and removed using the “akutils phix_filtering” command (Krohn et al. 2016b). Paired end reads were




http://dx.doi.org/10.1038/s41559-017-0150

4

joined with the “akutils join_paired_reads” command (mean length = 253 bp), which uses fastq-join from ea-utils (Aronesty et al. 2011). The raw joined data was demultiplexed with the “split_libraries_fastq.py” script in QIIME and underwent stricter minimum quality thresholds of q20 (q = 19), 0-3 low quality base calls allowed (r = 1-3), and each read required to be at least 95% high quality (p = 0.95) (Krohn et al. 2016b). We used vsearch 1.1.1 (Rognes et al. 2015) to remove chimeras against the Gold database (http://drive5.com/uchime/gold.fa). Demultiplexing and quality filtering resulted in a total of 203,259 high-quality reads (average per sample = 16,938 ± 468 SE). We used the “akutils pick_otus 16S” command for OTU picking and taxonomy assignment (Krohn et al. 2016b). Specifically, the prefix_suffix OTU picker in QIIME was used to dereplicate sequences on the first 100 bases. OTUs were picked at 97% similarity using Swarm (d1) and taxonomy assignment was performed with BLAST (Altschul et al. 1990). To assess microbial taxonomic diversity across the samples, we used the “akutils core_diversity” command that incorporated an OTU table having been filtered at the Kircher threshold (0.3% per sample; Kirchner et al. 2012). References in Supplementary Information Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic Local alignment

search tool. J. Mol. Biol., 215, 403-410 (1990). Aronesty, E. ea-utils: command-line tools for processing biological sequencing data. Expression

Analysis, Durham, NC (2011). Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data.

Nature Methods, 7, 335-336 (2010). Chojnacky, D. C., Heath, L. S., & Jenkins, J. C. (2014). Updated generalized biomass equations

for North American tree species. Forestry, 87, 129-151. Kirchner, M., Sawyer, S. & Meyer, M. Double indexing overcomes inaccuracies in multiplex

sequeuncing on the Illumina platform. Nucleic Acids Res., 40, e3, (2012). Krohn, A. akutils-v1.2: facilitating analyses of microbial communities through QIIME. Zenodo,

10.5281/zenodo.56764 (2016b). Krohn, A., Stevens, B., Robbins-Pianka, A., Belus, M., Allan, G. J. & Gehring, C. Optimization

of 16S amplicon analysis using mock communities: implications for estimating community diversity. Peer J, 10.7287/peerj.preprints.2196v2 (2016a).

Lojewski, N. R., Fischer, D. G., Bailey, J. K., Schweitzer, J. A., Whitham, T. G., & Hart, S. C. Genetic basis of aboveground productivity in two native Populus species and their hybrids. Tree Physiology, 29, 1133-1142 (2009).

Rognes, T., Mahé, F., Flouri, T., Quince, C. & Nichols, B. vsearch: VSEARCH version 1.0.16. Zenodo. 10.5281/zenodo.15524 (2015).




http://dx.doi.org/10.1038/s41559-017-0150

5

Supplementary Table 1. Summary of soil nutrients, pH, and relative abundance of dominant bacteria measured in this study Conditioned/Unconditioned Proteobacteria Site a Carbon (% total) Nitrogen (% total) pH Actinobacteria a b g Synergistia BL

Interior 2.2/1.0 0.16/0.10 5.6/5.3 0.09 0.19 0.08 0.09 0.04 Edge 1.4/2.4 0.10/0.17 5.4/5.3 0.18 0.12 0.15 0.03 0.0 Beyond 1.2 0.11 5.0

OC Interior 2.7/1.9 0.17/0.14 5.2/5.3 Edge 2.5/2.1 0.16/0.16 5.2/5.3 Beyond 4.0 0.27 5.3

SJ Interior 5.5/4.0 0.40/0.40 5.4/5.1 0.07 0.11 0.04 0.10 0.04 Edge 3.8/1.5 0.26/0.20 5.4/5.5 0.20 0.10 0.06 0.05 0.03 Beyond 1.2 0.14 5.5

SMIG Interior 8.1/4.7 0.44/0.30 5.6/5.4 Edge 4.1/3.2 0.26/0.18 5.6/5.1 Beyond 24.6 1.4 5.0

OGC Interior 6.3/4.5 0.34/0.25 5.7/5.0 Edge 4.1/1.6 0.25/0.13 5.4/5.5 Beyond 4.9 0.40 5.7

WR Interior 6.3/4.1 0.40/0.26 5.7/5.2 Edge 8.4/3.0 0.40/0.22 5.5/5.4 Beyond 5.5 0.32 5.11

SNR Interior 8.1/3.4 0.41/0.16 5.6/5.5 0.17 0.23 0.02 0.08 0.02 Edge 4.8/1.3 0.26/0.14 5.6/5.8 0.24 0.11 0.05 0.23 0.06 Beyond 4.04 0.23 5.3

a Sites are from the following Populus angustifolia populations: BL = Blue River, AZ; OC = Oak Creek, AZ; SJ = San Juan River, CO; SMIG = San Miguel River, CO; OGC = Ogden Canyon, UT; WR = Weber River, UT; and Snake River, WY.




http://dx.doi.org/10.1038/s41559-017-0150

6

Supplementary Table 2. Summary of soil nutrient and pH conditioning results by P. angustifolia trees in the field Carbon Nitrogen pH Model df F df F df F Soil location 50 9.0 ** 50 4.8 * 50 6.4 * Range location 50 3.7 † 50 3.0 † 50 1.0 Soil x range 50 0.1 50 0.2 50 3.3 † † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 Supplementary Table 3. Summary of rhizopshere soil microbial variation across P. angustifolia elevation ranges in the field

Actinobacteria a-proteobacteria b-proteobacteria g-proteobacteria Synergistia Model df F df F df F df F df F Range Location 11 1.8 11 1.7 11 6.4 * 11 0.1 11 0.0 Population 10 0.5 10 0.8 10 8.9 ** 10 1.3 10 0.4 Range x population 10 0.1 10 0.4 10 0.7 10 1.8 10 1.2 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 Supplementary Table 4. Results of plant-soil feedbacks generated by P. angustifolia conditioning of soil communities

Interior & edge Model df F Soil location 359 0.6 Soil treatment 391 0.4 Location x treatment 368 4.0 * Interior Edge Model df F df F Soil location 270 0.6 88 0.1 Soil treatment 288 0.0 95 0.7 Location x treatment 284 2.3 84 1.6 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001




http://dx.doi.org/10.1038/s41559-017-0150

7

Supplementary Table 5. Summary of the relationships between soil microbial relative abundance and P. angustifolia growth Actinobacteria a-proteobacteria b-proteobacteria g-proteobacteria Synergistia

Model df F df F df F df F df F Rel. abundance 11 0.7 11 1.2 11 11.1 ** 11 0.1 11 0.8 Range Location 11 0.7 11 1.2 11 1.7 11 0 11 0 Abundance x Range 11 0.7 11 1.3 11 0 11 0.7 11 1.1 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 Supplementary Table 6. Results of how range shift categories and soil treatments influence range shift PSF Home-away Local-foreign Model df F df F Range shift 213 13.1 *** 141 2.5 Soil treatment 211 2.5 144 0.0 Shift x treatment 211 2.3 † 144 0.1 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 Supplementary Table 7. Results of how soil conditioning (difference in soil traits between soil locations) predicts PSF

Carbon Nitrogen pH Model df F df F df F Soil conditioning 104 0.53 76 0.2 73 0.05 Range location 111 0.17 114 1.4 114 0.04 Conditioning x location 76 5.2 * 73 4.7 * 95 0.7 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001




http://dx.doi.org/10.1038/s41559-017-0150

8

Supplementary Table 8. Results of how residual conditioned soil differencesa between elevation sites predict range shift PSF Model Carbon Nitrogen pH Interior-Interior (I->I) df F df F df F Residual conditioned soil differences 113 0.2 113 1.7 113 15.7*** Range shift 112 10.2*** 112 12.6*** 112 3.0 Resid. soil difference x range shift 112 0.8 112 1.8 112 3.6* † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 a This metric refers to conditioned soil differences across elevation sites after background soil variation was removed. Supplementary Table 9. Results of field soil chemical variation related to aboveground biomass growth in the experiment. Modela df F p Growth ~ Soil Carbon 608 0.3 0.6 Growth ~ Soil Nitrogen 608 1.6 0.2 Growth ~ Soil pH 608 2.2 0.2 a All models include population, genotype, soil treatment (live or sterile inoculum), and location (conditioned/unconditioned or range shift category) as random effects. Supplementary Table 10. Results of soil chemical variation across populations, elevation, and multiple sampling years. Carbon Nitrogen pH Model df F df F df F Population 315 5.3 ** 315 5.2 ** 218 3.3 * Year 316 4.3 * 316 8.1 *** 220 493.9 *** Elevation 317 1.4 317 0.3 220 7.7 ** Population X elevation 315 3.8 * 315 0.6 218 0.5 Year X elevation 316 0.7 316 1.3 220 0.9 Population X Year 312 0.9 312 1.1 218 3.0 * Population X Elevation X Year 312 0.5 312 0.7 218 1.0 † p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 Results from 2012, 2013, and 2014 sampling of the following sites: Blue River, AZ, San Juan, CO, San Miguel, CO, and Snake River, WY. Note, pH data were compared between 2012 and 2014 collection dates since a different protocol was used for 2013 samples.




http://dx.doi.org/10.1038/s41559-017-0150

9

Supplementary Figure 1. Plant-soil conditioning in the field varies by range location. Soils conditioned beneath interior P. angustifolia trees have higher soil carbon (a), soil nitrogen (b), and soil pH (c) compared to unconditioned interspace locations away from trees. In contrast, no conditioning effects were found at the range edge. The relative abundance of Betaproteobacteria (d) in soils collected beneath P. angustifolia trees was found to be lower within the range interior compared to the range edge (unconditioned soils outside of the influence of trees were not sequenced). It should be noted that although these soils were collected directly from the rhizosphere and range location differences may be caused by variation in conditioning, it is also possible that Betaproteobacteria are more abundant at higher elevations regardless of P. angustifolia conditioning effects. Bars depict least square means ± 1 standard error.




http://dx.doi.org/10.1038/s41559-017-0150

10

Supplementary Figure 2. Stem volume of Populus angustifolia cuttings is an excellent predictor of aboveground biomass. Using a combination of basal diameter and plant height traits, total stem volume predicts more than 98% variation in aboveground biomass of trees that were collected over three years (2012-2014) and grown under equivalent greenhouse conditions. We used this relationship to establish the following allometric equation: Aboveground biomass (g) = (stem volume (mm3) * 0.41899) - 2.40137. Grey area represents 95% confidence interval.




http://dx.doi.org/10.1038/s41559-017-0150

11

Supplementary Figure 3. Tree size in the field (DBH) is unrelated to the initial biomass of tree cuttings, their final biomass, or their aboveground biomass growth. (a) Although trees varied in size in the field, there was no pattern between field and greenhouse size (i.e., larger trees did not produce cuttings with more biomass and vice versa). (b) Similarly, DBH does not correlate with final aboveground biomass at the end of the experiment. (c) Tree DBH in the field does not relate to the amount of aboveground biomass growth measured in this experiment. Together, these results suggest that maternal effects in this study do not heavily control these growth traits.




http://dx.doi.org/10.1038/s41559-017-0150

12

Supplementary Figure 4. Aboveground biomass growth responses to conditioned versus unconditioned soil locations across populations and soil treatments. (a) Populations consistently had larger growth in live conditioned soils than live unconditioned soils, indicating a positive feedback in soils that were modified by trees in the field. (b) There was no clear pattern in how populations responded to soil location when soil inoculum was sterilized, which suggests that variation in growth responses are due to conditioning of soil biota. Colors represent different populations, closed circles represent live soil treatments, open circles represent sterile soil treatments, and mean growth values ± 1 standard error are shown.




http://dx.doi.org/10.1038/s41559-017-0150

13

Supplementary Figure 5. Aboveground growth responses to simulated range shifts across populations and soil treatments. Populations generally have less growth with lower interior versus higher interior soil communities (I->I) in live treaments (a), but no clear response to this simulated range shift in sterile treatments (b). Similarly, there were no clear patterns in population responses to higher elevation interior trees grown with their ‘home’ versus higher elevation edge soil communities (I->E) in either live (c) or sterile treatments (d). Overall, populations generally had larger growth when their edge trees were grown with live soil communities from beyond range limits than live edge soil communities (E->B, e), but this pattern was not found with the equivalent sterile treatment (f). Colors represent different populations, closed circles represent live soil treatments, open circles represent sterile soil treatments, and mean growth values ± 1 standard error are shown.




http://dx.doi.org/10.1038/s41559-017-0150

14

Supplementary Figure 6. Plant-soil feedbacks (PSF) are related to the strength of soil conditioning across elevation gradients. The effect of (a) soil carbon (C) and (b) soil nitrogen (N) conditioning (i.e., the standardized difference between conditioned and unconditioned soil locations) positively relates to feedback effects for interior trees, but not edge trees. Soil pH conditioning (c) did not predict interior or edge PSF. Solid lines depict significant regressions with grey areas representing 95% confidence interval, dashed lines represent insignificant regressions.




http://dx.doi.org/10.1038/s41559-017-0150

15

Supplementary Figure 7. The relative abundance of Betaproteobacteria is related to P. angustifolia conditioned soil nutrients. (a) Conditioned soil carbon (C) and (b) nitrogen (N) are negatively related to the relative abundance of Betaproteobacteria. (c) There is no relation between conditioned soil pH and Betaproteobacteria abundance. Points are from soils collected and analyzed in 2014 from a subset of genotypes that were sampled in 2012 and 2013 from BL, SJ, and SNR populations. Solid lines depict significant linear regression trends with grey areas representing 95% confidence interval.




http://dx.doi.org/10.1038/s41559-017-0150

ULEENTARY INORATION - Nature Research · 2017-04-28 · join from ea-utils (Aronesty et al. 2011)....

Documents

Transcript of ULEENTARY INORATION - Nature Research · 2017-04-28 · join from ea-utils (Aronesty et al. 2011)....