Philippa Kaur

Literature search and study selectionA systematic literature search was conducted in the ISI Web of Science database for observational and experimental studies published from 1975 to 13 January 2020 using the following search terms: TOPIC: (grassland* OR prairie* OR steppe* OR shrubland* OR scrubland* OR bushland*) AND TOPIC: (drought* OR ‘dry period*’ OR ‘dry condition*’ OR ‘dry year*’ OR ‘dry spell*’) AND TOPIC: (product* OR biomass OR cover OR abundance* OR phytomass). The search was refined to include the subject categories Ecology, Environmental Sciences, Plant Sciences, Biodiversity Conservation, Multidisciplinary Sciences and Biology, and the document types Article, Review and Letter. This yielded a total of 2,187 peer-reviewed papers (Supplementary Fig. 1). At first, these papers were screened by title and abstract, which resulted in 197 potentially relevant full-text articles. We then examined the full text of these papers for eligibility and selected 87 studies (43 experimental, 43 observational and 1 that included both types) on the basis of the following criteria:

(1)

The research was conducted in the field, in natural or semi-natural grasslands or shrublands (for example, artificially constructed (seeded or planted) plant communities or studies using monolith transplants were excluded). We used this restriction because most reports on observational droughts are from intact ecosystems, and experiments in disturbed sites or using artificial communities would thus not be comparable to observational drought studies.

(2)

In the case of observational studies, the drought year or a multi-year drought was clearly specified by the authors (that is, we did not arbitrarily extract dry years from a long-term dataset). Please note that some observational data points are from control plots of experiments (of any kind), where the authors reported that a drought had occurred during the study period. We did not involve gradient studies that compare sites of different climates, which are sometimes referred to as ‘observational studies’.

(3)

The paper reported the amount or proportion of change in annual or growing-season precipitation (GSP) compared with control conditions. We consistently use the term ‘control’ for normal precipitation (non-drought) year or years in observational studies and for ambient precipitation (no treatment) in experimental studies hereafter. Similarly, we use the term ‘drought’ for both drought year or years in observational studies and drought treatment in experimental studies. In the case of multi-factor experiments, where precipitation reduction was combined with any other treatment (for example, warming), data from the plots receiving drought only and data from the control plots were used.

(4)

The paper contained raw data on plant production under both control and drought conditions, expressed in any of the following variables: ANPP, aboveground plant biomass (in grassland studies only) or percentage plant cover. In 79% of the studies that used ANPP as a production variable, ANPP was estimated by harvesting peak or end-of-season AGB. We therefore did not distinguish between ANPP and AGB, which are referred to as ‘biomass’ hereafter. We included the papers that reported the production of the whole plant community, or at least that of the dominant species or functional groups approximating the abundance of the whole community.

(5)

When multiple papers were published on the same experiment or natural drought event at the same study site, the most long-term study including the largest number of drought years was chosen.

In addition to the systematic literature search, we included 27 studies (9 experimental, 17 observational and 1 that included both types) meeting the above criteria from the cited references of the Web of Science records selected for our meta-analyses, and from previous meta-analyses and reviews on the topic. In total, this resulted in 114 studies (52 experimental, 60 observational and 2 that included both types; Supplementary Note 9, Supplementary Fig. 2 and ref. 25).Data compilationData were extracted from the text or tables, or were read from the figures using Web Plot Digitizer26. For each study, we collected the study site, latitude, longitude, mean annual temperature (MAT) and precipitation (MAP), study type (experimental or observational), and drought length (the number of consecutive drought years). When MAT or MAP was not documented in the paper, it was extracted from another published study conducted at the same study site (identified by site names and geographic coordinates) or from an online climate database cited in the respective paper. We also collected vegetation type—that is, grassland when it was dominated by grasses, or shrubland when the dominant species included one or more shrub species (involving communities co-dominated by grasses and shrubs). Data from the same study (that is, paper) but from different geographic locations or environmental conditions (for example, soil types, land uses or multiple levels of experimental drought) were collected as distinct data points (but see ‘Statistical analysis’ for how these points were handled). As a result, the 114 published papers provided 239 data points (112 experimental and 127 observational)25.For the observational studies, normal precipitation year or years specified by the authors was used as the control. If it was not specified in the paper, the year immediately preceding the drought year(s) was chosen as the control. When no data from the pre-drought year were available, the year immediately following the drought year(s) (14 data points) or a multi-year period given in the paper (22 data points) was used as the control. For the experimental studies, we also collected treatment size (that is, rainout shelter area or, if it was not reported in the paper, the experimental plot size).For the calculation of drought severity, we used yearly precipitation (YP), which was reported in a much higher number of studies than GSP. We extracted YP for both control (YPcontrol) and drought (YPdrought). For the observational studies, when a multi-year period was used as the control or the natural drought lasted for more than one year, precipitation values were averaged across the control or drought years, respectively. Consistently, in the case of multi-year drought experiments, YPcontrol and YPdrought were averaged across the treatment years. When only GSP was published in the paper (63 of 239 data points), we used this to obtain YP data as follows: we regarded MAP as YPcontrol, and YPdrought was calculated as YPdrought = MAP − (GSPcontrol − GSPdrought). From YPcontrol and YPdrought data, we calculated drought severity as follows: (YPdrought − YPcontrol)/YPcontrol × 100.For production, we compiled the mean, replication (N) and, if the study reported it, a variance estimate (s.d., s.e.m. or 95% CI) for both control and drought. In the case of multi-year droughts, data only from the last drought year were extracted, except in five studies (17 data points) where production data were given as an average for the drought years. When both biomass and cover data were presented in the paper, we chose biomass. For each study, we consistently considered replication as the number of the smallest independent study unit. When only the range of replications was reported in a study, we chose the smallest number.To quantify climatic aridity for each study site, we used an aridity index (AI), calculated as the ratio of MAP and mean annual PET (AI = MAP/PET). This is a frequently used index in recent climate change research27,28. AI values were extracted from the Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v.2 for the period of 1970–2000 (aggregated on annual basis)29.Because we wanted to prevent our analysis from being distorted by a strongly unequal distribution of studies between the two study types regarding some potentially important explanatory variables, we left out studies from our focal meta-analysis in three steps. First, we left out studies that were conducted at wet sites—that is, where site AI exceeded 1. The value of 1 was chosen for two reasons: above this value, the distribution of studies between the two study types was extremely uneven (22 experimental versus 2 observational data points with AI  > 1)25, and the AI value of 1 is a bioclimatically meaningful threshold, where MAP equals PET. Second, we left out shrublands, because we had only 14 shrubland studies (out of 105 studies with AI  More

IPCC. IPCC Fifth Assessment Report (AR5). 10–12 (IPCC, 2014).Cailleret, M. et al. A synthesis of radial growth patterns preceding tree mortality. Glob. Chang. Biol. 23, 1675–1690 (2017).PubMed 

Google Scholar 
Forzieri, G. et al. Emergent vulnerability to climate-driven disturbances in European forests. Nat. Commun. 12, 1–12 (2021).
Google Scholar 
Bonan, G. B. Forests and climate change: forcings, feedbacks, and the climate benefits of forests. Science https://doi.org/10.1126/science.1155121 (2008).Article 
PubMed 

Google Scholar 
Buras, A. & Menzel, A. Projecting tree species composition changes of European forests for 2061–2090 under RCP 4.5 and RCP 8.5 scenarios. Front. Plant Sci. 9, 1–13 (2019).
Google Scholar 
van der Maaten, E. et al. Species distribution models predict temporal but not spatial variation in forest growth. Ecol. Evol. 7, 2585–2594 (2017).PubMed 
PubMed Central 

Google Scholar 
Lebaube, S., Le Goff, N. L., Ottorini, J. M. & Granier, A. Carbon balance and tree growth in a Fagus sylvatica stand. Ann. Sci. 57, 49–61 (2000).
Google Scholar 
Dobbertin, M. Tree growth as indicator of tree vitality and of tree reaction to environmental stress: a review. Eur. J. For. Res. 124, 319–333 (2005).
Google Scholar 
Büntgen, U. Re-thinking the boundaries of dendrochronology. Dendrochronologia 53, 1–4 (2019).
Google Scholar 
Klesse, S. et al. Continental-scale tree-ring-based projection of Douglas-fir growth: Testing the limits of space-for-time substitution. Glob. Chang. Biol. 26, 5146–5163 (2020).PubMed 

Google Scholar 
Zhao, S. et al. The International Tree-Ring Data Bank (ITRDB) revisited: data availability and global ecological representativity. J. Biogeogr. 46, 355–368 (2019).
Google Scholar 
Babst, F. et al. When tree rings go global: challenges and opportunities for retro- and prospective insight. Quat. Sci. Rev. 197, 1–20 (2018).
Google Scholar 
Klesse, S. et al. Sampling bias overestimates climate change impacts on forest growth in the southwestern United States. Nat. Commun. 9, 1–9 (2018).
Google Scholar 
Yousefpour, R. et al. Realizing mitigation efficiency of European commercial forests by climate smart forestry. Sci. Rep. 8, 1–11 (2018).CAS 

Google Scholar 
Giesecke, T., Hickler, T., Kunkel, T., Sykes, M. T. & Bradshaw, R. H. W. Towards an understanding of the Holocene distribution of Fagus sylvatica L. J. Biogeogr. 34, 118–131 (2007).
Google Scholar 
Fang, J. & Lechowicz, M. J. Climatic limits for the present distribution of beech (Fagus L.) species in the world. J. Biogeogr. 33, 1804–1819 (2006).
Google Scholar 
Luterbacher, J., Dietrich, D., Xoplaki, E., Grosjean, M. & Wanner, H. European seasonal and annual temperature variability, trends, and extremes since 1500. Science 303, 1499–1503 (2004).CAS 
PubMed 

Google Scholar 
Luterbacher, J. et al. European summer temperatures since Roman times. Environ. Res. Lett. 11, 24001 (2016).Nabuurs, G. J. et al. By 2050 the mitigation effects of EU forests could nearly double through climate smart forestry. Forests 8, 1–14 (2017).
Google Scholar 
Walentowski, H. et al. Assessing future suitability of tree species under climate change by multiple methods: a case study in southern Germany. Ann. Res. 60, 101–126 (2017).
Google Scholar 
Mäkelä, A. et al. Process-based models for forest ecosystem management: current state of the art and challenges for practical implementation. Tree Physiol. 20, 289–298 (2000).PubMed 

Google Scholar 
Leech, S. M., Almuedo, P. L. & Neill, G. O. Assisted migration: adapting forest management to a changing climate. BC J. Ecosyst. Manag. 12, 18–34 (2011).
Google Scholar 
Sass-Klaassen, U. G. W. et al. A tree-centered approach to assess impacts of extreme climatic events on forests. Front. Plant Sci. 7, 1069 (2016).PubMed 
PubMed Central 

Google Scholar 
Bowman, D. M. J. S., Brienen, R. J. W., Gloor, E., Phillips, O. L. & Prior, L. D. Detecting trends in tree growth: not so simple. Trends Plant Sci. 18, 11–17 (2013).CAS 
PubMed 

Google Scholar 
Hacket-Pain, A. J. et al. Climatically controlled reproduction drives interannual growth variability in a temperate tree species. Ecol. Lett. 21, 1833–1844 (2018).PubMed 
PubMed Central 

Google Scholar 
Dorji, Y., Annighöfer, P., Ammer, C. & Seidel, D. Response of beech (Fagus sylvatica L.) trees to competition-new insights from using fractal analysis. Remote Sens. 11, 2656 (2019).Petit-Cailleux, C. et al. Combining statistical and mechanistic models to unravel the drivers of mortality within a rear-edge beech population. bioRxiv https://doi.org/10.1101/645747 (2019).Weigel, R., Gilles, J., Klisz, M., Manthey, M. & Kreyling, J. Forest understory vegetation is more related to soil than to climate towards the cold distribution margin of European beech. J. Veg. Sci. 30, 746–755 (2019).
Google Scholar 
Etzold, S. et al. Nitrogen deposition is the most important environmental driver of growth of pure, even-aged and managed European forests. Forest Ecol. Manag. 458, 117762 (2020).
Google Scholar 
Martínez-Sancho, E. et al. The GenTree dendroecological collection, tree-ring and wood density data from seven tree species across Europe. Sci. Data 7, 1–7 (2020).PubMed 
PubMed Central 

Google Scholar 
Hartl-Meier, C., Dittmar, C., Zang, C. & Rothe, A. Mountain forest growth response to climate change in the Northern Limestone Alps. Trees 28, 819–829 (2014).
Google Scholar 
Way, D. A. & Montgomery, R. A. Photoperiod constraints on tree phenology, performance and migration in a warming world. Plant Cell Environ. 38, 1725–1736 (2015).PubMed 

Google Scholar 
Martínez del Castillo, E. et al. Spatial patterns of climate – growth relationships across species distribution as a forest management tool in Moncayo Natural Park (Spain). Eur. J. Res. 138, 299 (2019).
Google Scholar 
Hacket-Pain, A. J., Cavin, L., Friend, A. D. & Jump, A. S. Consistent limitation of growth by high temperature and low precipitation from range core to southern edge of European beech indicates widespread vulnerability to changing climate. Eur. J. Res. 135, 897–909 (2016).
Google Scholar 
van der Maaten, E. Climate sensitivity of radial growth in European beech (Fagus sylvatica L.) at different aspects in southwestern Germany. Trees 26, 777–788 (2012).
Google Scholar 
Decuyper, M. et al. Spatio-temporal assessment of beech growth in relation to climate extremes in Slovenia – an integrated approach using remote sensing and tree-ring data. Agric. Meteorol. 287, 107925 (2020).
Google Scholar 
Kraus, C., Zang, C. & Menzel, A. Elevational response in leaf and xylem phenology reveals different prolongation of growing period of common beech and Norway spruce under warming conditions in the Bavarian Alps. Eur. J. Res. 135, 1011–1023 (2016).
Google Scholar 
Martínez del Castillo, E. et al. Living on the edge: contrasted wood-formation dynamics in Fagus sylvatica and Pinus sylvestris under mediterranean conditions. Front. Plant Sci. 7, 370 (2016).PubMed 
PubMed Central 

Google Scholar 
Čufar, K. et al. Temporal shifts in leaf phenology of beech (Fagus sylvatica) depend on elevation. Trees 26, 1091–1100 (2012).
Google Scholar 
Bontemps, J. D., Hervé, J. C. & Dhôte, J. F. Dominant radial and height growth reveal comparable historical variations for common beech in north-eastern France. Forest Ecol. Manag. 259, 1455–1463 (2010).
Google Scholar 
Latte, N., Lebourgeois, F. & Claessens, H. Increased tree-growth synchronization of beech (Fagus sylvatica L.) in response to climate change in northwestern Europe. Dendrochronologia 33, 69–77 (2015).
Google Scholar 
Zimmermann, J., Hauck, M., Dulamsuren, C. & Leuschner, C. Climate warming-related growth decline affects Fagus sylvatica, but not other broad-leaved tree species in central european mixed forests. Ecosystems 18, 560–572 (2015).CAS 

Google Scholar 
Tegel, W. et al. A recent growth increase of European beech (Fagus sylvatica L.) at its Mediterranean distribution limit contradicts drought stress. Eur. J. Res. 133, 61–71 (2014).
Google Scholar 
Hacket-Pain, A. J. & Friend, A. D. Increased growth and reduced summer drought limitation at the southern limit of Fagus sylvatica L., despite regionally warmer and drier conditions. Dendrochronologia 44, 22–30 (2017).
Google Scholar 
Dulamsuren, C., Hauck, M., Kopp, G., Ruff, M. & Leuschner, C. European beech responds to climate change with growth decline at lower, and growth increase at higher elevations in the center of its distribution range (SW Germany). Trees 31, 673–686 (2017).
Google Scholar 
Spiecker, H., Mielikäinen, K., Köhl, M. & Skovsgaard, J. P. Growth trends in European forests: studies from 12 countries. European Forest Institute Research Report (1996).Cavin, L. & Jump, A. S. Highest drought sensitivity and lowest resistance to growth suppression are found in the range core of the tree Fagus sylvatica L. not the equatorial range edge. Glob. Chang. Biol. 23, 1–18 (2016).
Google Scholar 
Mette, T. et al. Climatic turning point for beech and oak under climate change in Central Europe. Ecosphere 4, 1–19 (2013).
Google Scholar 
Michelot, A., Simard, S., Rathgeber, C. B. K., Dufrêne, E. & Damesin, C. Comparing the intra-annual wood formation of three European species (Fagus sylvatica, Quercus petraea and Pinus sylvestris) as related to leaf phenology and non-structural carbohydrate dynamics. Tree Physiol. 32, 1033–1045 (2012).PubMed 

Google Scholar 
Meier, I. C. & Leuschner, C. Belowground drought response of European beech: Fine root biomass and carbon partitioning in 14 mature stands across a precipitation gradient. Glob. Chang. Biol. 14, 2081–2095 (2008).
Google Scholar 
Leuschner, C. & Ellenberg, H. Ecology of Central European Forests. Vegetation Ecology of Central Europe. Vol. I (Springer, 2017).Allen, C. D., Breshears, D. D. & McDowell, N. G. On underestimation of global vulnerability to tree mortality and forest die-off from hotter drought in the Anthropocene. Ecosphere. 6, 1–55 (2015).
Google Scholar 
Pechanec, V., Purkyt, J., Benc, A., Nwaogu, C. & Lenka, Š. Ecological Informatics Modelling of the carbon sequestration and its prediction under climate change. https://doi.org/10.1016/j.ecoinf.2017.08.006 (2017).Speer, J. H. Fundamentals of Tree-Ring Research (University of Arizona Press, 2010).Biondi, F. & Qeadan, F. A theory-driven approach to tree-ring standardization: defining the biological trend from expected basal area increment. Tree-Ring Res. 64, 81–96 (2008).
Google Scholar 
Biondi, F. & Qeadan, F. Removing the tree-ring width biological trend using expected basal area increment. in USDA Forest Service RMRS-P-55 124–131 (2008).Karger, D. N. et al. Climatologies at high resolution for the earth’s land surface areas. Sci. Data 4, 1–20 (2017).
Google Scholar 
De Martonne, E. Une nouvelle fonction climatologique: L’indice d’aridité. La Meteorol. 2, 449–458 (1926).Martínez del Castillo, E., Longares, L. A., Serrano-Notivoli, R. & de Luis, M. Modeling tree-growth: assessing climate suitability of temperate forests growing in Moncayo Natural Park (Spain). Ecol. Manag. 435, 128–137 (2019).
Google Scholar 
Bolker, B. M. et al. Generalized linear mixed models: a practical guide for ecology and evolution. Trends Ecol. Evol. 24, 127–135 (2009).PubMed 

Google Scholar 
Calcagno, V. & Mazancourt, C. De. glmulti: an R package for easy automated model selection with (generalized) linear models. J. Stat. Softw. 34, 1–29 (2010).
Google Scholar 
Detry, M. A. & Ma, Y. Analyzing repeated measurements using mixed models. JAMA J. Am. Med. Assoc. 315, 407 (2016).CAS 

Google Scholar 
Harrison, X. A. et al. A brief introduction to mixed effects modelling and multi-model inference in ecology. PeerJ 2018, 1–32 (2018).
Google Scholar 
Johnson, J. B. & Omland, K. S. Model selection in ecology and evolution. Trends Ecol. Evol. 19, 101–108 (2004).PubMed 

Google Scholar 
Caudullo, G., Welk, E. & San-Miguel-Ayanz, J. Chorological maps for the main European woody species. Data Brief 12, 662–666 (2017).PubMed 
PubMed Central 

Google Scholar 
Meinshausen, M. et al. The shared socio-economic pathway (SSP) greenhouse gas concentrations and their extensions to 2500. Geosci. Model Dev. 13, 3571–3605 (2020).CAS 

Google Scholar 
Karger, D. N. & Zimmermann, N. E. CHELSAcruts – High Resolution Temperature And Precipitation Timeseries For The 20th Century And Beyond. https://doi.org/10.16904/envidat.159 (2018).Norinder, U., Rybacka, A. & Andersson, P. L. Conformal prediction to define applicability domain – a case study on predicting ER and AR binding. SAR QSAR Environ. Res. 27, 303–316 (2016).CAS 
PubMed 

Google Scholar 
Metzger, M. J., Bunce, R. G. H., Jongman, R. H. G., Mücher, C. A. & Watkins, J. W. A climatic stratification of the environment of Europe. Glob. Ecol. Biogeogr. 14, 549–563 (2005).
Google Scholar  More

Our general strategy was to compare the performance of four approaches for inferring microbial associations from abundance data with overlying time-series signals. The approaches were (1) pairwise spearman correlation analysis (SCC) [1, 29], (2) Graphical lasso analysis (Glasso) [30, 31], (3) pairwise SCC analysis with a pre-processing step where seasonal and long-term splines were fit to and subtracted from each variable using a GAM (GAM-SCC), and (4) Glasso with the same GAM subtraction approach (GAM-Glasso). Our validation strategy for the GAM transformation consisted of generating mock datasets with underlying associations, masking those associations by adding seasonal and long-term signals to the abundance data, and comparing the predicted associations obtained from each network inference method to the true species-species associations.Data simulation: generating mock abundance data with time-series propertiesWe generated mock abundance datasets that had a predetermined, underlying network structure and contained long-term and seasonal species abundance patterns. First, a covariance matrix was generated to describe the relationships between species in a mock dataset (Fig. S1, Panel 1). The covariance matrices were constructed with underlying network structures that followed either a scale-free Barabási-Albert model, a random Erdős-Rényi model, or a model of network topology based on a real microbial dataset (American Gut dataset; Fig. S1) [32, 33]. The Erdős-Rényi and Barabási-Albert model datasets were generated so that each dataset contained 400 species and 200 samples, and the American Gut datasets were created so that each dataset contained 127 species and 200 samples. A random Bernoulli distribution was used to simulate the covariance matrix for the Erdős-Rényi networks. We set the probability of interactions occurring between species in a given Erdős-Rényi network to 1%. The Barabási-Albert networks were generated using the “sample_pa” function in the igraph package [34]. The “graph2prec” function in the SpiecEasi package was used to predict the covariance matrix of the American Gut dataset [33]. The covariance between species in a dataset was considered “high” or “low” when the true associations in the covariance matrix were set to 100 or 10 respectively (Fig. S1, Panel 1). These covariance matrices describe the “real”, underlying species interactions in our mock datasets.After generating a covariance matrix, the mean abundance for each species was generated from a normal distribution with a mean of 10 and a variance of 1. These mean abundance values and the covariance matrix were used to parameterize a multivariate normal distribution from which species abundance values for all 200 samples in a dataset were drawn (Fig. S1, Panel 2). The values generated from this multivariate normal distribution were the species abundance values without time-series features confounding the relationship between two associated species (Fig. S1, Panel 2).“Gradual” or “abrupt” seasonal trends were added to 0%, 25%, 50% or 100% of the species in each mock dataset. The gradual seasonal trend increased over 5 months, peaked at a specific month, and decreased over 5 months. Conversely, the abrupt seasonal signal increased over 2 months, peaked at a specific month, and decreased over 2 months (Fig. S1, Panel 3). These seasonal signals were generated by plugging a vector of consecutive integers of length 200 (Nt) into the gradual (Eq. (1)) or abrupt (Eq. (2)) seasonal equations (Fig. S1, Panel 3)…$$Gradual:S_t = left( {frac{{cos left( {N_t ast 2 ast frac{pi }{{12}}} right)}}{2}} right) + 0.5$$
(1)
$$Abrupt:,S_t = left( {left( {frac{{cos left( {N_t ast 2 ast frac{pi }{{12}}} right)}}{2}} right) + 0.5} right)^{10}$$
(2)
where N is the random vector of consecutive integers, S is the output seasonal vector, and t is the index of vectors N and S. The starting value of vector Nt was drawn at random for each species to allow the seasonal peaks to be centered at different months. Each element in the seasonal vector (St) was then multiplied by the corresponding element in the abundance vector (Xt) of a specific species to obtain mock species abundance values with a gradual or abrupt seasonal trend (Fig. S1, Panel 3).A long-term time-series trend was added to the abundance values of 0% or 50% of the species in the mock datasets (Fig. S1, Panel 4). When a long-term signal was applied to 50% of the species in a dataset, half of the species were randomly selected to have this long-term trend. Then, a vector of linear values was generated following Eq. (3) such that…$$Long – term,trend:,L_t = pm mleft( {L_{t – 1}} right) + 0.01$$
(3)
where Lt is the point in the line at the next time point and m is the slope of the line. The slope parameter (m) was generated from a random normal distribution with a mean of 0.01 and a variance of 0.01. The slope parameter (m) was also multiplied by −1 half of the time to ensure that half of the long-term trends increased over time and half decreased over time (Fig. S1, Panel 4). After generating the vector of linear values (Lt), each element of this vector was added to each element of the abundance vector (Xt) of a specific species to simulate long-term time-series trends (Fig. S1, Panel 4).Time-series predictor columns were added to each dataset after applying monthly and long-term abundance trends to a portion of the species in the mock datasets. The predictors that were used in the downstream GAM-based data transformation were the month of the year (i.e., 1–12) and the day of the time-series (i.e., 1–200). In total, we generated 100 mock datasets for every combination of conditions (84 combinations total; Table S1), resulting in 8400 mock time-series datasets that were used in the downstream count data transformation, GAM subtraction, and network analysis procedures.Data simulation: Simulating count data from abundance valuesThe 8400 time-series datasets that were generated using the methods described above were transformed to make the abundance values resemble high-throughput sequencing data because microbial time-series sampling efforts are often processed using such molecular methods (e.g., tag-sequencing, meta-omics). Analysis of high-throughput sequencing data is complicated by the compositional (i.e., relative) nature of the data and by the high number of zeros that may be prevalent in a dataset (i.e., zero-inflation; see Supplementary Information) [35, 36]. Relative abundances of different species in natural communities are also highly skewed, so that relatively few species constitute most of the organisms in a sample although many rare species are also present [37, 38]. Therefore, species abundances were first exponentiated to increase the prevalence of abundant species and to decrease the prevalence of rare species (Fig. S1, Panel 5). The exponentiated species abundance values were then converted to relative abundance values by dividing each species count by the sum of all species counts in a sample (Fig. S1, Panel 6). The resulting relative abundance values and time-series predictor variables were used in data normalization and GAM-transformation steps prior to carrying out the network analyses.Network inference: Count data normalization and GAM transformationSeveral steps were taken to back out the species-species relationships in the mock datasets. We advocate these steps to infer network structure from a real time-series dataset. A centered log-ratio (CLR) transformation was first applied to the species relative abundance values to normalize the mock species abundance data across samples using the “clr” function in the compositions package in R (Fig. 1) [39]. This transformation step is important to avoid spurious inferences induced by the inherent compositionality of relative abundance data [31, 33, 36]. In addition to the CLR transformation used in our main network iterations, we carried out additional network iterations using the modified CLR [40], cumulative sum scaling [41], and total sum scaling [42] transformations (see Supplementary Information). In all cases, the normalized dataset was copied, with one copy subjected to a subsequent GAM transformation, and the other one not GAM-transformed.Fig. 1: Steps used to carry out the GAM-based transformation of time-series species abundance data prior to carrying out pairwise spearman correlation (SCC) and graphical lasso (Glasso) ecological network analyses.The raw, species abundance data were first CLR-transformed (1). Generalized additive models (GAMs) were then fit to each species in the dataset (2) and the residuals of each GAM were checked for significant autocorrelation (3). The residuals of each GAM were extracted (4) and were used as input in the SCC and Glasso network analysis methods (5). Finally, the GAM-transformed network outputs were obtained (6; see text for additional details).Full size imageThe GAM transformation was carried out by fitting GAMs to each individual species in the dataset to remove monthly signals, long-term trends, and autocorrelation from the species abundance data. These GAMs were fit using the “gamm” function in the mgcv package in R [43, 44]. The GAMs that were used included the “month of year” parameter as a cyclical spline predictor and the “day of time-series” parameter as a penalized thin-plate spline predictor (“ts” in the mgcv package; Fig. 1), which given our one-dimensional data is analogous to a natural cubic spline. In addition, the first GAM included a continuous AR1 (“corCAR1” in the mgcv package) correlation structure term in the model. This corCAR1 model was revised for specific species when the GAM could not be resolved or when significant autocorrelation was detected in the GAM residuals (Fig. 1). The GAM revision step fit 4 new GAMs with different correlation structure terms (i.e., “AR1”, “CompSymm”, “Exp”, and “Gaus”) to the species that could not be fit using the corCAR1 model or that contained significant autocorrelation in the corCAR1 GAM residuals. Then, the correlation structure term that addressed these issues for the largest number of individuals was used as the GAM model for this group of species. After fitting a GAM to all of the species in the input dataset, the residuals of each GAM were extracted and were used as the new, GAM-transformed abundance values (Fig. 1). These GAM residuals represent species abundance values with a reduced influence of time (Fig. 2) and were used as input in the downstream GAM-SCC and GAM-Glasso network analyses.Fig. 2: A conceptual figure that demonstrates how the GAM transformation can remove seasonal signals while preserving ecologically relevant species co-occurrence patterns.In this example, the co-occurrence pattern between Species A and Species B persists even after the seasonal signals are removed by the GAM transformation.Full size imageNetwork inference: Network runs and statistical analysesThe pre-processed species abundance data with and without the GAM-removal of time-series signals were used in SCC and Glasso networks in order to compare the outputs of the SCC, GAM-SCC, Glasso, and GAM-Glasso network inference approaches (Fig. 1). Additional network iterations were also carried out using the CCLasso [45] and SPRING [40] network inference approaches (see Supplementary Information). For the SCC and Glasso network iterations, a nonparanormal transformation was applied to the species abundance datasets with and without the GAM transformation using the “huge.npn” function in the huge package in R [46]. Spearman correlation networks were then constructed by calculating the correlation between every pair of species in the mock abundance datasets. A Bonferroni-corrected p value of 0.01 was used as a cutoff to identify edges in these SCC networks. The Glasso networks were constructed by testing 30 regularization parameter values (i.e., lambdas) in each network using the “batch.pulsar” (criterion = “stars”; rep.num = 50) function in the pulsar package in R [47]. The lambda that resulted in the most stable network output was selected using the StARS method [48]. Finally, the graph that resulted from the StARS output was used to obtain a species adjacency matrix for the Glasso networks.The species-species associations predicted by the SCC, GAM-SCC, Glasso, and GAM-Glasso networks were compared to the true species-species associations and the F1 scores of the network predictions were calculated. The F1 score is a measure of classification performance (presence or absence of an edge) that accounts for uneven classes, which is essential when dealing with sparse networks. The F1 scores of the GAM-transformed networks were compared to the networks that did not undergo GAM transformation using paired Wilcoxon tests with Bonferroni correction. An adjusted p value of 0.01 was used as a cutoff to identify under what circumstances the GAM significantly improved the F1 score of a Glasso or SCC network.Network inference: Comparison of predicted network structuresAdditional networks were generated using the methods described above to compare the predicted network structures obtained from the GAM-Glasso, Glasso, GAM-SCC, and SCC approaches to the real network structures. These additional networks were constructed using smaller mock datasets to allow for better visualization of the network outputs and contained species with a gradual seasonal signal and high species-species covariance (see Supplementary Information). The average clustering coefficient and the degree distribution of these additional network outputs were calculated and used for the network structure comparisons. The average clustering coefficient of a network describes the likelihood that two species that are both associated with a third species are also associated with each other [49], and in a sense describes the “clumpiness” of a network. The network degree distributions describe the probability distribution of the number of interactions per node in a network [50]. More

Marshall, C. R. Explaining the Cambrian “explosion” of animals. Annu. Rev. Earth Planet. Sci. 34, 355–384 (2016).ADS 

Google Scholar 
Bush, A. M. & Bambach, R. K. Paleoecologic megatrends in marine metazoa. Annu. Rev. Earth Planet. Sci. 39, 241–269 (2011).ADS 
CAS 

Google Scholar 
Smith, M. P. & Harper, D. A. Causes of the Cambrian explosion. Science 341(6152), 1355–1356 (2013).ADS 
PubMed 

Google Scholar 
Daley, A. C., Antcliffe, J. B., Drage, H. B. & Pates, S. Early fossil record of Euarthropoda and the Cambrian explosion. Proc. Natl. Acad. Sci. 115(21), 5323–5331 (2018).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Fu, D. et al. The Qingjiang biota—A Burgess Shale–type fossil Lagerstätte from the early Cambrian of South China. Science 363(6433), 1338–1342 (2019).ADS 
CAS 
PubMed 

Google Scholar 
Nanglu, K., Caron, J. B. & Gaines, R. R. The Burgess Shale paleocommunity with new insights from Marble Canyon, British Columbia. Paleobiology 46(1), 58–81 (2020).
Google Scholar 
Sepkoski, J. J. Jr. The Ordovician radiations: Diversification and extinction shown by global genus-level taxonomic data. In Ordovician Odyssey: Short Papers, 7th International Symposium on the Ordovician System (eds Cooper, J. D. et al.) 393–396 (Pacific Section Society for Sedimentary Geology (SEPM), 1995).
Google Scholar 
Servais, T., Cascales-Miñana, B. & Harper, D. A. The Great Ordovician Biodiversification event (GOBE) is not a single event. Paleontol. Res. 25(4), 315–328 (2021).
Google Scholar 
Harper, D. A., Cascales-Miñana, B., Kroeck, D. M. & Servais, T. The palaeogeographical impact on the biodiversity of marine faunas during the Ordovician radiations. Glob. Planet. Change 207, 103665 (2021).
Google Scholar 
Harper, D. A. et al. The Furongian (late Cambrian) biodiversity gap: Real or apparent?. Palaeoworld 28(1–2), 4–12 (2019).
Google Scholar 
Saleh, F. et al. Taphonomic bias in exceptionally preserved biotas. Earth Planet. Sci. Lett. 529, 115873 (2020).CAS 

Google Scholar 
Saleh, F. et al. A novel tool to untangle the ecology and fossil preservation knot in exceptionally preserved biotas. Earth Planet. Sci. Lett. 569, 117061 (2021).CAS 

Google Scholar 
Vizcaïno, D. & Lefebvre, B. Les échinodermes du Paléozoïqueinférieur de Montagne Noire: Biostratigraphie et paléodiversité. Geobios 32(2), 353–364 (1995).
Google Scholar 
Vizcaïno, D. & Álvaro, J. J. Adequacy of the Early Ordovician trilobite record in the southern Montagne Noire (France): Biases for biodiversity documentation. Earth Environ. Sci. Trans. R. Soc. Edinb. 93(4), 393–401 (2002).
Google Scholar 
Lefebvre, B. et al. Palaeoecological aspects of the diversification of echinoderms in the Lower Ordovician of central Anti-Atlas, Morocco. Palaeogeogr. Palaeoclimatol. Palaeoecol. 460, 97–121 (2016).
Google Scholar 
Lefebvre, B. et al. Exceptionally preserved soft parts in fossils from the Lower Ordovician of Morocco clarify stylophoran affinities within basal deuterostomes. Geobios 52, 27–36 (2019).
Google Scholar 
Martin, E. L. O. et al. Biostratigraphic and palaeoenvironmental controls on the trilobite associations from the Lower Ordovician Fezouata Shale of the central Anti-Atlas, Morocco. Palaeogeogr. Palaeoclimatol. Palaeoecol. 460, 142–154 (2016).
Google Scholar 
Waisfeld, B. G. & Balseiro, D. Decoupling of local and regional dominance in trilobite assemblages from northwestern Argentina: New insights into Cambro-Ordovician ecological changes. Lethaia 49(3), 379–392 (2016).
Google Scholar 
Serra, F., Balseiro, D. & Waisfeld, B. G. Diversity patterns in upper Cambrian to Lower Ordovician trilobite communities of north-western Argentina. Palaeontology 62(4), 677–695 (2019).
Google Scholar 
Serra, F., Balseiro, D., Vaucher, R. & Waisfeld, B. G. Structure of trilobite communities along a delta-marine gradient (lower Ordovician; Northwestern Argentina). Palaios 36(2), 39–52 (2021).ADS 

Google Scholar 
Saleh, F., Lefebvre, B., Hunter, A. W. & Nohejlová, M. Fossil weathering and preparation mimic soft tissues in eocrinoid and somasteroid echinoderms from the Lower Ordovician of Morocco. Microsc. Today 28(1), 24–28 (2020).
Google Scholar 
Saleh, F. et al. Insights into soft-part preservation from the Early Ordovician Fezouata Biota. Earth Sci. Rev. 213, 103464 (2021).
Google Scholar 
Saleh, F. et al. Large trilobites in a stress-free Early Ordovician environment. Geol. Mag. 158(2), 261–270 (2021).ADS 

Google Scholar 
Vilmi, A. et al. Dispersal–niche continuum index: A new quantitative metric for assessing the relative importance of dispersal versus niche processes in community assembly. Ecography 44(3), 370–379 (2021).
Google Scholar 
Hubbell, S. P. A Unified Theory of Biodiversity and Biogeography (Princeton University Press, 2001).
Google Scholar 
Gravel, D., Canham, C. D., Beaudet, M. & Messier, C. Reconciling niche and neutrality: The continuum hypothesis. Ecol. Lett. 9(4), 399–409 (2006).PubMed 

Google Scholar 
Bergström, S. M., Chen, X., Gutiérrez-Marco, J. C. & Dronov, A. The new chronostratigraphic classification of the Ordovician system and its relations to major regional series and stages and to δ13C chemostratigraphy. Lethaia 42, 97–107 (2009).
Google Scholar 
Lefebvre, B. et al. Age calibration of the Lower Ordovician Fezouata Lagerstätte, Morocco. Lethaia 51(2), 296–311 (2018).
Google Scholar 
Servais, T. et al. The onset of the ‘Ordovician Plankton Revolution’in the late Cambrian. Palaeogeogr. Palaeoclimatol. Palaeoecol. 458, 12–28 (2016).
Google Scholar 
Lee, J. H. & Riding, R. Marine oxygenation, lithistid sponges, and the early history of Paleozoic skeletal reefs. Earth Sci. Rev. 181, 98–121 (2018).ADS 
CAS 

Google Scholar 
Servais, T., Danelian, T., Harper, D. A. T. & Munnecke, A. Possible oceanic circulation patterns, surface water currents and upwelling zones in the Early Palaeozoic. GFF 136(1), 229–233 (2014).
Google Scholar 
Rasmussen, C. M. et al. Onset of main Phanerozoic marine radiation sparked by emerging Mid Ordovician icehouse. Sci. Rep. 6(1), 1–9 (2016).
Google Scholar 
Edwards, C. T. Links between early Paleozoic oxygenation and the Great Ordovician Biodiversification Event (GOBE): A review. Palaeoworld 28(1–2), 37–50 (2019).
Google Scholar 
Buatois, L. A. et al. Quantifying ecospace utilization and ecosystem engineering during the early Phanerozoic—The role of bioturbation and bioerosion. Sci. Adv. 6(33), eabb0618 (2020).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Mángano, M. G. et al. Were all trilobites fully marine? Trilobite expansion into brackish water during the early Palaeozoic. Proc. R. Soc. B 288(1944), 20202263 (2021).PubMed 
PubMed Central 

Google Scholar 
Park, T. Y. S. et al. Ontogeny of the Furongian (late Cambrian) trilobite Proceratopyge cf. P. Lata Whitehouse from northern Victoria Land, Antarctica, and the evolution of metamorphosis in trilobites. Palaeontology 59(5), 657–670 (2016).
Google Scholar 
Laibl, L. & Fatka, O. Review of early developmental stages of trilobites and agnostids from the Barrandian area (Czech Republic). J. Natl. Mus. (Prague) Nat. Hist. Ser. 186(1), 103–112 (2017).
Google Scholar 
Laibl, L., Cederström, P. & Ahlberg, P. Early post-embryonic development in Ellipsostrenua (Trilobita, Cambrian, Sweden) and the developmental patterns in Ellipsocephaloidea. J. Paleontol. 92(6), 1018–1027 (2018).
Google Scholar 
Laibl, L., Maletz, J. & Olschewski, P. Post-embryonic development of Fritzolenellus suggests the ancestral morphology of the early developmental stages in Trilobita. Pap. Palaeontol. 7(2), 839–859 (2021).
Google Scholar 
Chatterton, B. D. E. & Speyer, S. E. Ontogeny in Treatise on Invertebrate Paleontology. Part O, Arthropoda 1, Trilobita 1, revised. 7–11 (Geological Society of America and University of Kansas Press, Lawrence, 1997).Bignon, A., Waisfeld, B. G., Vaccari, N. E. & Chatterton, B. D. Reassessment of the order Trinucleida (Trilobita). J. Syst. Palaeontol. 18(13), 1061–1077 (2020).
Google Scholar 
Torsvik, T. H. & Cocks, L. R. M. The Palaeozoic palaeogeography of central Gondwana. Geol. Soc. Lond. Spec. Publ. 357(1), 137–166 (2011).ADS 

Google Scholar 
Torsvik, T. H. & Cocks, L. R. M. New global palaeogeographical reconstructions for the early Palaeozoic and their generation. Geol. Soc. Lond. Mem. 38(1), 5–24 (2013).
Google Scholar 
Bahlburg, H., Moya, M. C. & Zeil, W. Geodynamic evolution of the early Palaeozoic continental margin of Gondwana in the Southern Central Andes of Northwestern Argentina and Northern Chile. In Tectonics of the Southern Central Andes. 293–302 (Springer, 1994).McEdward, L. R. & Miner, B. G. Larval and life-cycle patterns in echinoderms. Can. J. Zool. 79(7), 1125–1170 (2001).
Google Scholar 
Lefebvre, B. et al. Palaeobiogeography of Ordovician echinoderms. Geol. Soc. Lond. Mem. 38(1), 173–198 (2013).
Google Scholar 
Signor, P. W. & Vermeij, G. J. The plankton and the benthos: Origins and early history of an evolving relationship. Paleobiology 20, 297–319 (1994).
Google Scholar 
Davis, M. A., Grime, J. P. & Thompson, K. Fluctuating resources in plant communities: A general theory of invasibility. J. Ecol. 88(3), 528–534 (2000).
Google Scholar 
Franeck, F. Perspectives on the Great Ordovician Biodiversification Event-local to global patterns (2020).Pulliam, H. R. Sources, sinks, and population regulation. Am. Nat. 132(5), 652–661 (1988).
Google Scholar 
Kröger, B., Franeck, F. & Rasmussen, C. M. The evolutionary dynamics of the early Palaeozoic marine biodiversity accumulation. Proc. R. Soc. B 286(1909), 20191634 (2019).PubMed 
PubMed Central 

Google Scholar 
Penny, A. & Kröger, B. Impacts of spatial and environmental differentiation on early Palaeozoic marine biodiversity. Nat. Ecol. Evol. 3(12), 1655–1660 (2019).PubMed 

Google Scholar 
Rasmussen, C. M., Kröger, B., Nielsen, M. L. & Colmenar, J. Cascading trend of Early Paleozoic marine radiations paused by Late Ordovician extinctions. Proc. Natl. Acad. Sci. 116(15), 7207–7213 (2019).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Stigall, A. L. The invasion hierarchy: Ecological and evolutionary consequences of invasions in the fossil record. Annu. Rev. Ecol. Evol. Syst. 50, 355–380 (2019).
Google Scholar 
Stigall, A. L., Edwards, C. T., Freeman, R. L. & Rasmussen, C. M. Coordinated biotic and abiotic change during the Great Ordovician Biodiversification Event: Darriwilian assembly of early Paleozoic building blocks. Palaeogeogr. Palaeoclimatol. Palaeoecol. 530, 249–270 (2019).
Google Scholar 
Stigall, A. L. How is biodiversity produced? Examining speciation processes during the GOBE. Lethaia 51(2), 165–172 (2018).
Google Scholar 
Servais, T. & Harper, D. A. T. The great Ordovician biodiversification event (GOBE): Definition, concept and duration. Lethaia 51(2), 151–164 (2018).
Google Scholar 
Trotter, J. A. et al. Did cooling oceans trigger Ordovician biodiversification? Evidence from conodont thermometry. Science 321(5888), 550–554 (2008).ADS 
CAS 
PubMed 

Google Scholar 
Vizcaïno, D., Álvaro, J. J. & Lefebvre, B. The lower Ordovician of the southern Montagne Noire. Ann. Soc. Géol. Nord 8(4), 213–220 (2001).
Google Scholar 
Hsieh, T. C., Ma, K. H. & Chao, A. iNEXT: An R package for rarefaction and extrapolation of species diversity (H ill numbers). Methods Ecol. Evol. 7(12), 1451–1456 (2016).
Google Scholar 
Suchéras-Marx, B., Escarguel, G., Ferreira, J. & Hammer, Ø. Statistical confidence intervals for relative abundances and abundance-based ratios: Simple practical solutions for an old overlooked question. Mar. Micropaleontol. 151, 101751 (2019).ADS 

Google Scholar 
Hammer, Ø., Harper, D. A. T. & Ryan, P. D. PAST-palaeontological statistics, ver. 1.89. Palaeontol. Electron 4(1), 1–9 (2001).
Google Scholar 
Gibert, C. & Escarguel, G. PER-SIMPER—A new tool for inferring community assembly processes from taxon occurrences. Glob. Ecol. Biogeogr. 28(3), 374–385 (2019).
Google Scholar 
Gibert, C. DNCImper: Assembly Process Identification Based on SIMPER Analysis. R package ver. 0.0.1.0000. https://github.com/Corentin-Gibert-Paleontology/DNCImper (2019). More

Microcosm preparation and incubationLeaf litters were collected from Lilly-Dickey Woods, a mature eastern US temperate broadleaf forest located in South-Central Indiana (39°14′N, 86°13′W) using litter baskets and surveys for freshly senesced litter as described in Craig et al.52. Of the 19 species collected in Craig et al. (2018), we selected litter from 16 tree species with the goal of maximizing variation in litter chemical traits (Table S1). Litters were air-dried and then homogenized and fragmented such that all litter fragments passed a 4000 µm, but not a 250 µm mesh. Whereas leaf litters had a distinctly C3 δ13C signature of −30.1 ± 1.5 (mean, standard deviation), we used a 13C-rich (δ13C = −12.6 ± 0.4) soil obtained from the A horizon of a 35-yr continuous corn field at the Purdue University Agronomy Center for Research and Education near West Lafayette, Indiana (40°4′N, 86°56′W). The soil is classified as Chalmers silty clay loam (a fine-silty, mixed, superactive, mesic Typic Endoaquoll). Prior to use in the incubation, soils were sieved (2 mm) and remaining recognizable plant residues were thoroughly picked out. Soils were mixed with acid-washed sand—30% by mass—to facilitate litter mixing (see below) and to increase the soil volume for post-incubation processing. The resulting soil had a pH of 6.7 and a C:N ratio of 12.0.We constructed the experimental microcosms by mixing the 16 litter species with the 13C-enriched soil. Each litter treatment was replicated four times in four batches (i.e., 16 microcosms per species, 272 total microcosms including 16 soil-only controls). Two batches (C budget microcosms) were used to monitor CO2 efflux and to track litter-derived C into SOM pools, and two batches were used to quantify microbial biomass dynamics.Incubations were carried out in 50 mL centrifuge tubes modified with an O-ring to prevent leakage and a rubber septum to facilitate headspace sampling. To each microcosm, we added 5 g dry soil, adjusted moisture to 65% water-holding capacity, and pre-incubated for 24 h in the dark at 24 °C. Using a dissecting needle, 300 mg of leaf litter were carefully mixed into treatment microcosms and controls were similarly agitated. This corresponds to an average C addition rate of 27.1 ± 1.1 g C kg−1 dry soil among the 16 species. During incubation, microcosms were loosely capped to retain moisture while allowing gas exchange, and were maintained at 65% water-holding capacity by adding deionized water every week.Carbon budget in microcosmsRespiration was quantified with an infrared gas analyzer (LiCOR 6262, Lincoln, NE, USA) coupled to a sample injection system. Our first measurement was taken about 12 h after litter addition (day 1) and subsequent measurements were taken on days 2, 4, 11, 19, and 30 for both batches and on days 46, 64, 79, 92, 109, 128, 149, and 185 for the second batch. Prior to each measurement, microcosms were capped, flushed with CO2-free air, and incubated for 1–8 h depending on the expected efflux rate. Headspace was sampled with a gas-tight syringe and the CO2-C concentration was converted to a respiration rate (µg CO2-C day−1). Total cumulative CO2-C loss was derived from point measurements by numerical integration (i.e., the trapezoid method). To evaluate soil-derived CO2-C efflux, we measured δ13C in two gas samples per litter type or control on a ThermoFinnigan DELTA Plus XP isotope ratio mass spectrometer (IRMS) with a GasBench interface (Thermo Fisher Scientific, San Jose, CA). Isotopes were measured on days 1, 4, 11, 30, 64, 109, and 185. On each of these days, a two-source mixing model70 was applied to determine the fraction of total CO2-C derived from soil organic matter vs. litter:$$frac{{F}^{l}(t)}{F(t)}=frac{delta Fleft(tright)-,delta {F}^{c}(t)}{delta {C}_{l}-delta {F}^{c}(t)}$$
(1)
where (frac{{F}^{l}(t)}{F(t)}) is the fraction total CO2-C efflux [(F(t))] derived from litter [({F}^{l}(t))] at time (left(tright)), (delta Fleft(tright)) is the δ13C of the CO2 respired by each litter-soil combination, (delta {F}^{c}(t)) is the average δ13C of the CO2 respired by the control soil, and (delta {C}_{l}) is the δ13C of each litter type. These data were used to calculate cumulative soil-derived C efflux via numerical integration and, for each litter type, average soil-derived C efflux was subtracted from total cumulative CO2-C loss to determine cumulative litter-derived CO2-C loss.Carbon budget microcosms were harvested on days 30 and 185 to track litter-derived C into mineral-associated SOC at an early and intermediate stage of decomposition. To do this, we used a size fractionation procedure71,72 modified to minimize the recovery of soluble leaf litter compounds or dissolved organic matter in the mineral-associated SOC fraction. For each sample, we first added 30 mL deionized water, gently shook by hand to suspend all particles, and then centrifuged (2500 rpm) for 10 min. Floating leaf litter was carefully removed, dried for 48 h at 60 °C, and weighed; and the clear supernatant was discarded to remove the dissolved organic matter. The remaining sample was dispersed in 5% (w/v) sodium hexametaphosphate for 20 h on a reciprocal shaker and then washed through a 53 µm sieve. The fraction retained on the sieve was added to the floating leaf litter sample and collectively referred to as particulate SOC, while the fraction that passed through the sieve was considered the mineral-associated SOC. Both fractions were dried, ground, and weighed; and analyzed for C concentrations and δ13C values on an elemental combustion system (Costech ECS 4010, Costech Analytical Technologies, Valencia, CA, USA) as an inlet to an IRMS. As above, litter-derived C in the particulate and mineral-associated SOC was determined as follows:$$frac{{C}_{s}^{l}(t)}{{C}_{s}(t)}=frac{delta {C}_{s}left(tright)-,delta {C}_{c}(t)}{delta {C}_{l}-delta {C}_{c}(t)}$$
(2)
where ({C}_{s}(t)) is the total particulate or mineral-associated SOC content in the sample at time ((t)), ({C}_{s}^{l}(t)) is the litter-derived C in the soil, (delta {C}_{s}left(tright)) is the measured δ13C value for each soil fraction, (delta {C}_{c}left(tright)) is the average δ13C for each fraction in control samples, and (delta {C}_{l}) is the δ13C of each litter type. In a few cases, mineral-associated δ13C was slightly less negative in the treatment than in the control soil. In these cases, litter-derived mineral-associated SOC was considered zero.Total litter-derived SOC at each harvest date was calculated by subtracting the cumulative litter CO2-C from initial added litter C. The difference between this value and the sum of litter-derived particulate and mineral-associated SOC was considered the residual pool which we assume mostly represents water-extractable dissolved organic matter.Microbial biomass dynamics during incubationSample batches were harvested at days 15 and 100 to capture early- and intermediate-term microbial biomass responses to litter treatments. These times were selected to correspond with the middle of early and intermediate C budget microcosm incubations. We quantified microbial biomass as well as MGR, CUE, and MTR using 18O-labeled water73,74 as in Geyer et al.75.Microbial biomass C (MBC) was determined on two ~2 g subsamples using a standard chloroform fumigation extraction76. One subsample was immediately extracted in 0.5 M K2SO4 and one was fumigated for 72 h before extraction. After shaking for 1 h, extracts were gravity filtered through a Whatman No. 40 filter paper, and filtrates were analyzed for total organic C using the method of Bartlett and Ross77 as adapted by Giasson et al.78. The difference between total organic C in the fumigated and unfumigated subsamples was used to calculate MBC (extraction efficiency KEC = 0.45).To determine MGR, CUE, and MTR, we first pre-incubated two 0.5 g soil subsamples (one treatment and one control) for 2 d at 24 °C. Prior to this pre-incubation, samples were allowed to evaporate down to 53 ± 6% (mean, sd) water-holding capacity. After the pre-incubation, water was injected with a 25 µL syringe to bring each sample to 65% water-holding capacity. For one subsample, we used unlabeled deionized water. For the second subsample, enriched 18O-water (98.1 at%; ICON Isotopes) was mixed with unlabeled deionized H2O to achieve approximately 20 at% of 18O in the final soil water. Each sample was placed in a centrifuge tube (modified for gas sampling), flushed with CO2-free air, and incubated for 24 h. Headspace CO2 concentration was then measured, and samples were flash frozen in liquid N2 and stored at −80 °C until DNA extraction.DNA was extracted from each sample using a DNA extraction kit (Qiagen DNeasy PowerSoil Kit, Venlo, Netherlands) following the protocol described in Geyer et al. (2019) which sought to maximize the recovery of DNA. The DNA concentration was determined fluorometrically using a Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). DNA extracts (80 µL) were dried at 60 °C in silver capsules spiked with 100 µL of salmon sperm DNA (42.5 ng µL−1), to reach the oxygen detection limit, and sent to the UC Davis Stable Isotope Facility for quantification of δ18O and total O content.Microbial growth rate (MGR) was calculated following Geyer et al. (2019). Specifically, atom % of soil DNA O (at% ODNA) was determined using the two-pool mixing model:$${at} % ,{O}_{{DNA}}=,frac{left[left({at} % ,{O}_{{DN}A+{ss}}times {O}_{{DNA}+{ss}}right)-left({at} % ,{O}_{{ss}}times {O}_{{ss}}right)right]}{{O}_{{DNA}}},$$
(3)
where at% is the atom % 18O and ODNA+ss, ODNA, and Oss are the concentration of O in the whole sample, soil DNA, and salmon sperm, respectively. Atom percent excess of soil DNA oxygen (APE Osoil) was calculated as the difference between at% ODNA in the treatment and control samples. Total microbial growth in terms of O (Total O; µg) was estimated as:$${Total},O=frac{{O}_{{soil}}times ,{{APE},O}_{{soil}}}{{at} % ,{soil},{water}}$$
(4)
where at% soil water is the atom % 18O in the soil water. MGR in terms of C (µg C g−1 soil d−1) was calculated by applying conversion mass ratios of oxygen:DNA (0.31) and MBC:DNA for each sample, dividing by the soil mass, and scaling by the incubation time. Assuming uptake rate (Uptake) is equal to the sum of MGR and respiration, CUE and MTR were calculated by the following equations.$${CUE}=,frac{{MGR}}{{Uptake}},$$
(5)
$${MTR}=,frac{{MGR}}{{MBC}}$$
(6)
Data analysis for microcosm experimentLitter decay constants were calculated for each species using litter-derived CO2-C values to estimate litter mass loss over time. After it was determined that a single exponential decay model provided a poor fit, we fit litter decomposition data using the double exponential decay model:$$y=s{e}^{{-k}_{1}t}+(1-s){e}^{{-k}_{2}t}$$
(7)
where s represents the labile or early stage decomposition fraction that decomposes at rate k1, and k2 is the decay constant for the remaining late stage decomposition fraction.To reduce the dimensionality of litter quality and microbial indicators, indices were derived by principal component analysis (PCA; Fig. S1A, B) using the ‘prcomp’ function in R. The first axis of a PCA of decomposition parameters (s, k1, and k2) and litter chemical properties (soluble and AUR contents; AUR-to-N and C-to-N ratios; and the lignocellulose index [LCI]) was taken as a litter quality index. Whereas this index highly correlated with indicators of C quality (AUR, soluble content, and LCI), the second axis of this PCA correlated with C:N and AUR:N and was therefore taken as a second litter quality index representing variation in N concentration. The first axis of a PCA of MGR, CUE, and MTR was taken as a microbial physiological trait index.Bivariate relationships were examined using simple linear regressions on average species values at each harvest (n = 16). To examine relationships between microbial physiological traits and mineral-associated SOC, data from the early-term (day 15) and intermediate-term (day 100) microbial harvest were matched with early-term (day 30) and intermediate-term (day 185) C budget microcosms, respectively. In addition to examining total mineral-associated SOC formation, we also estimated the efficiency of litter C transfer into the mineral-associated SOC pool as the fraction of lost litter C (i.e., litter C lost as CO2, recovered in the mineral-associated SOC fraction, or in the residual pool) retained in the mineral-associated SOC. Path analyses were used to evaluate the hypothesis that microbial physiological traits mediate the effect of litter quality on mineral-associated SOC formation and mineral-associated and particulate SOC decay. We hypothesized that the litter quality index would be positively associated with the microbial physiological trait index (representing faster and more efficient microbial growth) and microbial physiological traits would, in turn, be positively associated with the rate and efficiency of mineral-associated SOC formation. We expected that this mediating pathway would reduce the direct relationship between litter quality and SOC. This analysis was conducted using the LAVAAN package79 to run path analyses for total litter-derived mineral-associated SOC, mineral-associated SOC formation efficiency, and soil-derived mineral-associated and particulate SOC for both early and intermediate stage harvests. All analyses were performed using R version 3.5.2.Field study design and soil samplingWe worked in the Smithsonian’s Forest Global Earth Observatory (ForestGEO) network80 in six mature U.S. temperate forests varying in climate, soil properties, and tree community composition (Fig. 1a): Harvard forest (HF; 42°32′N, 72°11′W) in North-Central Massachusetts, Lilly-Dickey Woods (LDW; 39°14’N, 86°13’W) in South-Central Indiana, the Smithsonian Conservation Biology Institute (SCBI; 38°54′N, 78°9′W) in Northern Virginia, the Smithsonian Environmental Research Center (SERC; 38°53′N, 76°34′W) on the Chesapeake Bay in Maryland, Tyson Research Center (TRC; 38°31′N, 90°33′W) in Eastern Missouri, and Wabikon Lake Forest (WLF; 45°33′N, 88°48′W) in Northern Wisconsin, USA. Land use history across the six sites consisted mostly of timber harvesting which ceased in the early 1900s. Soils are mostly Oxyaquic Dystrudepts at HF, Typic Dystrudepts and Typic Hapludults at LDW, Typic Hapludalfs at SCBI, Typic or Aquic Hapludults at SERC, Typic Hapludalfs and Typic Paleudalfs at TRC, and Typic and Alfic Haplorthods at WLF. Further site details are reported in Table S5.Each site contains a rich assemblage of co-occurring arbuscular mycorrhizal (AM)- and ectomycorrhizal (ECM)-associated trees (Table S6), which we leveraged to generate environmental gradients in factors hypothesized to predict microbial physiological traits within each site. Specifically, the dominance of AM vs. ECM trees within a temperate forest plot has been shown to be a strong predictor of soil pH, C:N, inorganic N availability, and litter quality52,53,54. We established nine 20 × 20 m plots in each of our six sites in Fall 2016 (n = 54) distributed along a gradient of AM- to ECM-associated tree dominance. Plots were selected to avoid obvious confounding environmental factors. Where possible, we established our nine-plot gradient in three blocks (5 cm) at HF, which was removed before coring. Samples were also collected at 5–15 cm depth for soil texture analysis. We sampled from an inner 10 × 10 m square in each plot to avoid edge effects. All samples from the same plot were composited, sieved (2 mm), picked free of roots, subsampled for gravimetric moisture (105 °C), and air-dried, or refrigerated (4 °C) until analysis for microbial physiological variables and N availability.Soil propertiesWe determined several physicochemical properties known to predict mineral-associated SOC. We measured soil pH (8:1 ml 0.01 M CaCl2:g soil) and soil texture using a benchtop pH meter and a standard hydrometer procedure82, respectively. Organic matter content was high in some upper surface soils, so plot-level soil texture was determined from 5 to 15 cm depth samples. We quantified oxalate-extractable Al and Fe pools (Alox and Feox) in all soil samples as an index of poorly crystalline Al- and Fe-oxides83, which is one of the strongest predictors of SOM content in temperate forests84. Briefly, we extracted 0.40 g dried, ground soil in 40 mL 0.2 M NH4-oxalate at pH 3.0 in the dark for 4 h before gravity filtering and refrigerating until analysis (within 2 w) on an atomic-adsorption spectrometer (Aanalyst 800, Perkin Elmer, Waltham, MA, USA), using an acetylene flame and a graphite furnace for the atomization of Fe and Al, respectively.We quantified potential net N mineralization rates as an index of soil N availability. One 5 g subsample per plot was extracted immediately after processing by adding 10 mL 2 M KCl, shaking for 1 h, and filtering through a Whatman No. 1 filter paper after centrifugation at 3000 rpm. A second subsample from each plot was incubated under aerobic conditions at field moisture and 23 °C for 14 d before extraction. Extracts were frozen (−20 °C) until analysis for NH4+-N using the salicylate method and for NO3−-N plus NO2−-N after a cadmium column reduction on a Lachat QuikChem 8000 flow Injection Analyzer (Lachat Instruments, Loveland, CO, USA). Potential net N mineralization rates (mg N g dry soil−1 d−1) were calculated as the difference between pre- and post-incubation inorganic N concentrations.Microbial biomass dynamics in field plotsMicrobial biomass carbon and microbial physiological traits were quantified within 10 days of collection as described above, with four minor differences. First, 30 g soil subsamples were covered with parafilm and pre-incubated for 2 d near the field soil temperature measured at the time of sampling (16.5 °C for WLF and HF, and 21.5 °C for LDW, TRC, SCBI, and SERC). Second, for CO2 analysis, samples were placed in a 61 mL serum vial crimped with a rubber septum. Third, DNA concentrations were determined using a Qubit dsDNA BR Assay Kit (Life Technologies) and a Qubit 3.0 fluorometer (Life Technologies). Fourth, 14.5 g subsamples were used for microbial biomass analysis.Soil organic matter characterization in field plotsMineral-associated SOC was quantified as in the microcosm experiment, but without a pre-fractionation leachate removal step. We additionally measured soil amino sugar concentrations to estimate microbial necromass contributions to SOM. Amino sugars are useful microbial biomarkers because they are found in abundance in microbial cell walls, but are not produced by higher plants and soil animals19. Moreover, amino sugars can provide information on the microbial source of necromass. For example, glucosamine (Glu) is produced mostly by fungi whereas muramic acid (MurA) is produced almost exclusively by bacteria61,85. Amino sugars were extracted, purified, converted to aldononitrile acetates, and quantified with myo-inositol as in Liang et al.86. We used the concentrations of Glu and MurA to estimate total, fungal, and bacterial necromass soil C using the empirical relationships reported in Liang et al.8.$${Bacterial},{necromass},C,=,{MurA},times ,45$$
(8)
$${Fungal},{necromass},C,=,({mmol},{GluN},{-},2,times ,{mmol},{MurA})times ,179.17,times ,9$$
(9)
Leaf litter and fine roots in field plotsIn Fall 2017, we collected leaf litter on two sample dates from four baskets deployed in the inner 10 × 10 m of each plot. Litter was composited by plot, dried (60 °C), sorted by species, weighed, and ground. We performed leaf litter analyses on at least three samples of each species at each site —unless a species was only present in one or two plots— to get a site-specific mean for each species. Some non-dominant species were not included in these analyses because an insufficient amount of material was collected. Fine roots ( 0.5). Feox and Alox were correlated above this threshold and final models were selected to contain only Feox based on AIC. Residuals were screened for normality (Shapiro-Wilk), heteroscedasticity (visual assessment of residual plots), and influential observations (Cook’s D). Based on this, MGR, MTR, and mineral-associated SOC were natural log transformed. For all mixed models, we centered and standardized predictors (i.e., z-transformation) so that the slopes and significance levels of different predictors could be compared to one another on the same axis88. More

Hannam, K. D. et al. Wood ash as a soil amendment in Canadian forests: what are the barriers to utilization?. Can. J. For. Res. 48, 442–450 (2018).
Google Scholar 
Hope, E. S., McKenney, D. W., Allen, D. J. & Pedlar, J. H. A cost analysis of bioenergy-generated ash disposal options in Canada. Can. J. For. Res. https://doi.org/10.1139/cjfr-2016-0524 (2017).Article 

Google Scholar 
Bowd, E. J., Banks, S. C., Strong, C. L. & Lindenmayer, D. B. Long-term impacts of wildfire and logging on forest soils. Nat. Geosci. 12, 113–118 (2019).ADS 
CAS 

Google Scholar 
Adotey, N., Harrell, D. L. & Weatherford, W. P. Characterization and liming effect of wood Ash generated from a biomass-fueled commercial power plant. Commun. Soil Sci. Plan. 49, 38–49 (2018).CAS 

Google Scholar 
Royer-Tardif, S., Whalen, J. & Rivest, D. Can alkaline residuals from the pulp and paper industry neutralize acidity in forest soils without increasing greenhouse gas emissions?. Sci. Total Environ. 663, 537–547 (2019).ADS 
CAS 
PubMed 

Google Scholar 
Reid, C. & Watmough, S. A. Evaluating the effects of liming and wood-ash treatment on forest ecosystems through systematic meta-analysis. Can. J. For. Res. 44, 867–885 (2014).CAS 

Google Scholar 
López, R., Díaz, M. J. & González-Pérez, J. A. Extra CO2 sequestration following reutilization of biomass ash. Sci. Total Environ. 625, 1013–1020 (2018).ADS 
PubMed 

Google Scholar 
Emilson, C. E. et al. Short-term growth response of jack pine and spruce spp. to wood ash amendment across Canada. GCB Bioenergy 12, 158–167 (2020).
Google Scholar 
Azan, S. S. E. et al. Could a residential wood ash recycling programme be part of the solution to calcium decline in lakes and forests in Muskoka (Ontario, Canada)?. FACETS 4, 69–90 (2019).
Google Scholar 
Gorgolewski, A. et al. Responses of eastern red-backed salamander (Plethodon cinereus) abundance 1 year after application of wood ash in a northern hardwood forest. Can. J. For. Res. 46, 402–409 (2016).
Google Scholar 
McTavish, M. J., Gorgolewski, A., Murphy, S. D. & Basiliko, N. Field and laboratory responses of earthworms to use of wood ash as a forest soil amendment. For. Ecol. Manag. 474, 118376 (2020).
Google Scholar 
Mortensen, L. H., Rønn, R. & Vestergård, M. Bioaccumulation of cadmium in soil organisms: with focus on wood ash application. Ecotox. Environ. Safe. 156, 452–462 (2018).CAS 

Google Scholar 
Bélanger, N., Palma Ponce, G. & Brais, S. Contrasted growth response of hybrid larch (Larix × marschlinsii), jack pine (Pinus banksiana) and white spruce (Picea glauca) to wood ash application in northwestern Quebec, Canada. iForest. 14, 155 (2021).
Google Scholar 
Santás-Miguel, V. et al. Use of biomass ash to reduce toxicity affecting soil bacterial community growth due to tetracycline antibiotics. J. Environ. Manage. 269, 110838 (2020).PubMed 

Google Scholar 
Fritze, H. et al. A microcosmos study on the effects of cd-containing wood ash on the coniferous humus fungal community and the cd bioavailability. J Soils Sediments 1, 146–150 (2001).CAS 

Google Scholar 
Coleman, D., Callaham, Jr., M. A. & Crossley, Jr., D. A. Fundamentals of Soil Ecology. (Elsevier, 2018). https://doi.org/10.1016/C2015-0-04083-7.Smenderovac, E. E. et al. Does intensified boreal forest harvesting impact soil microbial community structure and function?. Can. J. For. Res. 47, 916–925 (2017).CAS 

Google Scholar 
Joseph, R. et al. Limited effect of wood ash application on soil quality as indicated by a multisite assessment of soil organic matter attributes. GCB Bioenergy. 00, 1–22. https://doi.org/10.1111/gcbb.12928 (2022).CAS 
Article 

Google Scholar 
Noyce, G. L. et al. Soil microbial responses to wood ash addition and forest fire in managed Ontario forests. Appl. Soil Ecol. 107, 368–380 (2016).
Google Scholar 
Liiri, M., Ilmarinen, K. & Setälä, H. Variable impacts of enchytraeid worms and ectomycorrhizal fungi on plant growth in raw humus soil treated with wood ash. Appl. Soil Ecol. 35, 174–183 (2007).
Google Scholar 
Brais, S., Bélanger, N. & Guillemette, T. Wood ash and N fertilization in the Canadian boreal forest: Soil properties and response of jack pine and black spruce. For. Ecol. Manag. 348, 1–14 (2015).
Google Scholar 
Gömöryová, E., Pichler, V., Tóthová, S. & Gömöry, D. Changes of chemical and biological properties of distinct forest floor layers after wood ash application in a Norway spruce stand. Forests 7, 108 (2016).
Google Scholar 
Hannam, K., Great Lakes Forestry Centre, Canada, Ressources naturelles Canada & Canadian Forest Service. Regulations and guidelines for the use of wood ash as a soil amendment in Canadian forests. (2016).Hannam, K. D. et al. AshNet: Facilitating the use of wood ash as a forest soil amendment in Canada. Forest. Chron. 93, 17–20 (2017).
Google Scholar 
Klavina, D. et al. The ectomycorrhizal community of conifer stands on peat soils 12 years after fertilization with wood ash. Mycorrhiza 26, 153–160 (2016).PubMed 

Google Scholar 
Bang-Andreasen, T. et al. Wood ash induced pH changes strongly affect soil bacterial numbers and community composition. Front. Microbiol. 8, 1400 (2017).PubMed 
PubMed Central 

Google Scholar 
Vestergård, M. et al. The relative importance of the bacterial pathway and soil inorganic nitrogen increase across an extreme wood-ash application gradient. GCB Bioenergy 10, 320–334 (2018).
Google Scholar 
Ekenler, M. & Tabatabai, M. A. β-glucosaminidase activity as an index of nitrogen mineralization in soils. Commun. Soil Sci. Plan. 35, 1081–1094 (2004).CAS 

Google Scholar 
Margalef, O. et al. Global patterns of phosphatase activity in natural soils. Sci Rep 7, 1337 (2017).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Vestergaard, G., Schulz, S., Schöler, A. & Schloter, M. Making big data smart—how to use metagenomics to understand soil quality. Biol. Fertil. Soils 53, 479–484 (2017).
Google Scholar 
Emilson, C. et al. Synthesis of current AshNet study designs and methods with recommendations towards a standardized protocol. Information Report GLC-X-22. (2018).Baldwin, K. et al. Vegetation zones of Canada: A biogeoclimatic perspective – Open Government Portal. (2019).Findlay, S. CHAPTER 11: Dissolved organic matter. In: Methods in Stream Ecology (Second Edition) (eds. Hauer, F. R. & Lamberti, G. A.) 239–248 (Academic Press, 2007). https://doi.org/10.1016/B978-012332908-0.50013-9.Saiya-Cork, K. R., Sinsabaugh, R. L. & Zak, D. R. The effects of long term nitrogen deposition on extracellular enzyme activity in an Acer saccharum forest soil. Soil Biol. Biochem. 34, 1309–1315 (2002).CAS 

Google Scholar 
Porter, T. M. & Hajibabaei, M. METAWORKS: A flexible, scalable bioinformatic pipeline for multi-marker biodiversity assessments. bioRxiv 2020.07.14.202960 (2020) https://doi.org/10.1101/2020.07.14.202960.Wang, Q., Garrity, G. M., Tiedje, J. M. & Cole, J. R. Naïve Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 73, 5261–5267 (2007).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Porter, T. M. & Hajibabaei, M. Automated high throughput animal CO1 metabarcode classification. Sci. Rep-UK 8, 4226 (2018).ADS 

Google Scholar 
Kõljalg, U., Abarenkov, K., Nilsson, R. H., Larsson, K. & Taylor, A. F. S. The UNITE Database for molecular identification and for communicating fungal species (2019). https://doi.org/10.3897/BISS.3.37402.Porter, T. M. UNITE ITS Classifier. (2020). https://github.com/terrimporter/UNITE_ITSClassifierLouca, S., Parfrey, L. W. & Doebeli, M. Decoupling function and taxonomy in the global ocean microbiome. Science 353, 1272–1277 (2016).ADS 
CAS 

Google Scholar 
Nguyen, N. H. et al. FUNGuild: An open annotation tool for parsing fungal community datasets by ecological guild. Fungal Ecol. 20, 241–248 (2016).
Google Scholar 
Hedde, M. et al. BETSI, a complete framework for studying soil invertebrate functional traits. (2012). https://doi.org/10.13140/2.1.1286.6888.McKenney, D. W. et al. Customized spatial climate models for North America. Bull. Am. Meteor. Soc. 92, 1611–1622 (2011).ADS 

Google Scholar 
R Core Team. R: A Language and Environment for Statistical Computing. (R Foundation for Statistical Computing, 2021).Fernandes, A. D., Macklaim, J. M., Linn, T. G., Reid, G. & Gloor, G. B. ANOVA-like differential expression (ALDEx) analysis for mixed population RNA-Seq. PLOS ONE 8, 15 (2013).
Google Scholar 
Wickham, H. et al. Welcome to the {tidyverse}. J. Open Source Softw. 4, 1686 (2019).ADS 

Google Scholar 
Oksanen, J. et al. Vegan: Community ecology package. https://CRAN.R-project.org/package=vegan (2020).Domes, K. A. et al. Short-term changes in spruce foliar nutrients and soil properties in response to wood ash application in the sub-boreal climate zone of British Columbia. Can. J. Soil. Sci. 98, 246–263 (2018).CAS 

Google Scholar 
Pugliese, S. et al. Wood ash as a forest soil amendment: The role of boiler and soil type on soil property response. Can. J. Soil. Sci. 94, 621–634 (2014).CAS 

Google Scholar 
Bang-Andreasen, T. et al. Total RNA sequencing reveals multilevel microbial community changes and functional responses to wood ash application in agricultural and forest soil. FEMS Microbiol. Ecol. 96, fiaa016 (2020).PubMed 
PubMed Central 

Google Scholar 
Haimi, J., Fritze, H. & Moilanen, P. Responses of soil decomposer animals to wood-ash fertilisation and burning in a coniferous forest stand. For. Ecol. Manag. 129, 53–61 (2000).
Google Scholar 
Aronsson, K. A. & Ekelund, N. G. A. Biological effects of wood ash application to forest and aquatic ecosystems. J. Environ. Qual. 33, 1595–1605 (2004).CAS 
PubMed 

Google Scholar 
Omil, B., Piñeiro, V. & Merino, A. Trace elements in soils and plants in temperate forest plantations subjected to single and multiple applications of mixed wood ash. Sci. Total Environ. 381, 157–168 (2007).ADS 
CAS 
PubMed 

Google Scholar 
Taylor, A. F. S. & Finlay, R. D. Effects of liming and ash application on below ground ectomycorrhizal community structure in two Norway spruce forests. WAFO 3, 63–76 (2003).CAS 

Google Scholar 
Wallander, H., Fossum, A., Rosengren, U. & Jones, H. Ectomycorrhizal fungal biomass in roots and uptake of P from apatite by Pinus sylvestris seedlings growing in forest soil with and without wood ash amendment. Mycorrhiza 15, 143–148 (2005).PubMed 

Google Scholar 
Kjøller, R., Cruz-Paredes, C. & Clemmensen, K. E. Ectomycorrhizal fungal responses to forest liming and wood ash addition: Review and meta-analysis. In Soil Biological Communities and Ecosystem Resilience (eds Lukac, M. et al.) 223–252 (Springer International Publishing, Berlin, 2017).
Google Scholar 
Peltoniemi, K., Pyrhönen, M., Laiho, R., Moilanen, M. & Fritze, H. Microbial communities after wood ash fertilization in a boreal drained peatland forest. Eur. J. Soil Biol. 76, 95–102 (2016).CAS 

Google Scholar 
Boisvert-Marsh, L., Great Lakes Forestry Centre, Canada & Resources naturelles Canada. The Island Lake biomass harvest experiment: early results. (2016).Couch, R. L., Luckai, N., Morris, D. & Diochon, A. Short-term effects of wood ash application on soil properties, growth, and foliar nutrition of Picea mariana and Picea glauca seedlings in a plantation trial. Can. J. Soil. Sci. 101, 203–215 (2021).CAS 

Google Scholar 
Perkiömäki, J. & Fritze, H. Cadmium in upland forests after vitality fertilization with wood ash—a summary of soil microbiological studies into the potential risk of cadmium release. Biol Fertil Soils 41, 75–84 (2005).
Google Scholar 
Paredes, C. et al. Bacteria respond stronger than fungi across a steep wood ash-driven pH gradient. Front. For. Glob. Change 4, 781884 (2021).
Google Scholar 
Kļaviņa, D. et al. Fungal communities in roots of scots pine and Norway spruce saplings grown for 10 years on peat soils fertilized with wood ash. Balt. For. 22, 10 (2016).
Google Scholar 
Hansen, M., Bang-Andreasen, T., Sørensen, H. & Ingerslev, M. Micro vertical changes in soil pH and base cations over time after application of wood ash on forest soil. For. Ecol. Manag. 406, 274–280 (2017).
Google Scholar 
Fu, X. et al. Understory vegetation leads to changes in soil acidity and in microbial communities 27years after reforestation. Sci. Total Environ. 502, 280–286 (2015).ADS 
CAS 
PubMed 

Google Scholar 
Pitman, R. M. Wood ash use in forestry – a review of the environmental impacts. Forestry 79, 563–588 (2006).
Google Scholar 
Cruz-Paredes, C., Tájmel, D. & Rousk, J. Can moisture affect temperature dependences of microbial growth and respiration?. Soil Biol. Biochem. 156, 108223 (2021).CAS 

Google Scholar  More

NATURE PODCAST
09 March 2022

The AI that deciphers ancient Greek graffiti

An artificial intelligence that restores illegible inscriptions, and the project that’s reintroducing lost species in Argentina.

Nick Petrić Howe

&

Benjamin Thompson

Nick Petrić Howe

View author publications

You can also search for this author in PubMed
 Google Scholar

Benjamin Thompson

View author publications

You can also search for this author in PubMed
 Google Scholar

Twitter

Facebook

Email

Subscribe
Subscribe

iTunes
Google Podcast
acast
RSS

Listen to the latest from the world of science, with Nick Petrić Howe and Benjamin Thompson.

Your browser does not support the audio element.

Download MP3

In this episode:00:46 The AI helping historians read ancient textsResearchers have developed an artificial intelligence that can restore and date ancient Greek inscriptions. They hope that it will help historians by speeding up the process of reconstructing damaged texts. Research article: Assael et al.News and Views: AI minds the gap and fills in missing Greek inscriptionsVideo: The AI historian: A new tool to decipher ancient textsIthaca platform08:53 Research HighlightsPollinators prefer nectar with a pinch of salt, and measurements of a megacomet’s mighty size.Research Highlight: Even six-legged diners can’t resist sweet-and-salty snacksResearch Highlight: Huge comet is biggest of its kind11:10 Rewilding ArgentinaThis week Nature publishes a Comment article from a group who aim to reverse biodiversity loss by reintroducing species to areas where they are extinct. We speak to one of the Comment’s authors about the project and their hopes that it might kick start ecosystem restoration.Comment: Rewilding Argentina: lessons for the 2030 biodiversity targets21:02 Briefing ChatWe discuss some highlights from the Nature Briefing. This time, giant bacteria that can be seen with the naked eye, and how record-breaking rainfall has caused major floods in Australia.Science: Largest bacterium ever discovered has an unexpectedly complex cellNew Scientist: Record flooding in Australia driven by La Niña and climate changeThe Conversation: The east coast rain seems endless. Where on Earth is all the water coming from?Subscribe to Nature Briefing, an unmissable daily round-up of science news, opinion and analysis free in your inbox every weekday.Never miss an episode: Subscribe to the Nature Podcast on Apple Podcasts, Google Podcasts, Spotify or your favourite podcast app. Head here for the Nature Podcast RSS feed.

doi: https://doi.org/10.1038/d41586-022-00701-7

Related Articles

Read the paper: Restoring and attributing ancient texts using deep neural networks

Rewilding Argentina: lessons for the 2030 biodiversity targets

Subjects

Machine learning

History

Ecology

Latest on:

Machine learning

AI minds the gap and fills in missing Greek inscriptions
News & Views 09 MAR 22

The evolution, evolvability and engineering of gene regulatory DNA
Article 09 MAR 22

Gran Turismo champion, reimagined urine — the week in infographics
News 15 FEB 22

History

AI minds the gap and fills in missing Greek inscriptions
News & Views 09 MAR 22

Police rely on radiocarbon dating to identify forged paintings
News 09 MAR 22

Restoring and attributing ancient texts using deep neural networks
Article 09 MAR 22

Ecology

Discovery of a Ni2+-dependent guanidine hydrolase in bacteria
Article 09 MAR 22

Rewilding Argentina: lessons for the 2030 biodiversity targets
Comment 07 MAR 22

How itchy vicuñas remade a vast wilderness
Research Highlight 04 MAR 22

Jobs

Principal Investigator / Faculty

San Diego Biomedical Research Institute (SDBRI)
Multiple locations

Research Associate / Light Microscopy Imaging Specialist (m/f/x)

Technische Universität Dresden (TU Dresden)
01069 Dresden, Germany

Bioinformatician (m/f/div)

Max Planck Institute for Plant Breeding Research (MPIPZ)
Cologne, Germany

Research Coordinator / Institute Manager (div/f/m)

Max Planck Institute for Astrophysics (MPA)
Garching near Munich, Germany More

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript. More