Ecology

Bradford, M. A. et al. Soil carbon science for policy and practice. Nat. Sustain. 2, 1070–1072 (2019).Article 

Google Scholar 
Lehmann, J. & Kleber, M. The contentious nature of soil organic matter. Nature 528, 60–68 (2015).Article 

Google Scholar 
Liang, C., Schimel, J. P. & Jastrow, J. D. The importance of anabolism in microbial control over soil carbon storage. Nat. Microbiol. 2, 17105 (2017).Article 

Google Scholar 
Liang, C., Amelung, W., Lehmann, J. & Kästner, M. Quantitative assessment of microbial necromass contribution to soil organic matter. Glob. Change Biol. 25, 3578–3590 (2019).Article 

Google Scholar 
Wang, B. R., An, S. S., Liang, C., Liu, Y. & Kuzyakov, Y. Microbial necromass as the source of soil organic carbon in global ecosystems. Soil Biol. Biochem. 162, 108422 (2021).Article 

Google Scholar 
Kästner, M. & Miltner, A. in The Future of Soil Carbon (eds Garcia, C. et al.) Ch. 5 (Academic Press, 2018).Buckeridge, K. M. et al. Sticky dead microbes: rapid abiotic retention of microbial necromass in soil. Soil Biol. Biochem. 149, 107929 (2020).Article 

Google Scholar 
Kallenbach, C. M., Grandy, A. S., Frey, S. D. & Diefendorf, A. F. Microbial physiology and necromass regulate agricultural soil carbon accumulation. Soil Biol. Biochem. 91, 279–290 (2015).Article 

Google Scholar 
Kallenbach, C. M., Frey, S. D. & Grandy, A. S. Direct evidence for microbial-derived soil organic matter formation and its ecophysiological controls. Nat. Commun. 7, 13630 (2016).Article 

Google Scholar 
Emerson, J. B. et al. Schrödinger’s microbes: tools for distinguishing the living from the dead in microbial ecosystems. Microbiome 5, 86 (2017).Article 

Google Scholar 
Zhang, Y. et al. Simulating measurable ecosystem carbon and nitrogen dynamics with the mechanistically defined MEMS 2.0 model. Biogeosciences 18, 3147–3171 (2021).Article 

Google Scholar 
Ackermann, M., Stearns Stephen, C. & Jenal, U. Senescence in a bacterium with asymmetric division. Science 300, 1920–1920 (2003).Article 

Google Scholar 
Aguilaniu, H., Gustafsson, L., Rigoulet, M. & Nyström, T. Asymmetric inheritance of oxidatively damaged proteins during cytokinesis. Science 299, 1751–1753 (2003).Article 

Google Scholar 
Maheshwari, R. & Navaraj, A. Senescence in fungi: the view from Neurospora. FEMS Microbiol. Lett. 280, 135–143 (2008).Article 

Google Scholar 
See, C. R. et al. Hyphae move matter and microbes to mineral microsites: integrating the hyphosphere into conceptual models of soil organic matter stabilization. Glob. Change Biol. 28, 2527–2540 (2022).Article 

Google Scholar 
Pusztahelyi, T. et al. Comparative studies of differential expression of chitinolytic enzymes encoded by chiA, chiB, chiC and nagA genes in Aspergillus nidulans. Folia Microbiologica 51, 547–554 (2006).Article 

Google Scholar 
Bartoszewska, M. & Kiel, J. A. The role of macroautophagy in development of filamentous fungi. Antioxid. Redox Signal. 14, 2271–2287 (2011).Article 

Google Scholar 
Josefsen, L. et al. Autophagy provides nutrients for nonassimilating fungal structures and is necessary for plant colonization but not for infection in the necrotrophic plant pathogen Fusarium graminearum. Autophagy 8, 326–337 (2012).Article 

Google Scholar 
Heaton, L. L., Jones, N. S. & Fricker, M. D. Energetic constraints on fungal growth. Am. Nat. 187, E27–E40 (2016).Article 

Google Scholar 
Taiz, L. & Zeiger, E. Plant Physiology 4th edn (Spektrum Akademischer Verlag, 2008).Bowman, E. J. & Bowman, B. J. in Cellular and Molecular Biology of Filamentous Fungi (eds Borkovich, K. & Ebbole, D.) 179–190 (ASM Press, 2010).Voigt, O. & Pöggeler, S. Self-eating to grow and kill: autophagy in filamentous ascomycetes. Appl. Microbiol. Biotechnol. 97, 9277–9290 (2013).Article 

Google Scholar 
Grimmett, I. J., Shipp, K. N., Macneil, A. & Barlocher, F. Does the growth rate hypothesis apply to aquatic hyphomycetes? Fungal Ecol. 6, 493–500 (2013).Article 

Google Scholar 
Camenzind, T., Philipp Grenz, K., Lehmann, J. & Rillig, M. C. Soil fungal mycelia have unexpectedly flexible stoichiometric C:N and C:P ratios. Ecol. Lett. 24, 208–218 (2021).Article 

Google Scholar 
Mason-Jones, K., Robinson, S. L., Veen, G. F., Manzoni, S. & van der Putten, W. H. Microbial storage and its implications for soil ecology. ISME J. 16, 617–629 (2022).Article 

Google Scholar 
Gow, N. A. R., Latge, J. P. & Munro, C. A. The fungal cell wall: structure, biosynthesis, and function. Microbiol. Spectr. 5, FUNK-0035–2016 (2017).Article 

Google Scholar 
Steiner, U. K. Senescence in bacteria and its underlying mechanisms. Front. Cell Dev. Biol. 9, 668915 (2021).Article 

Google Scholar 
Allocati, N., Masulli, M., Di Ilio, C. & De Laurenzi, V. Die for the community: an overview of programmed cell death in bacteria. Cell Death Dis. 6, e1609 (2015).Article 

Google Scholar 
Peeters, S. H. & de Jonge, M. I. For the greater good: programmed cell death in bacterial communities. Microbiol. Res. 207, 161–169 (2018).Article 

Google Scholar 
Wang, J. & Bayles, K. W. Programmed cell death in plants: lessons from bacteria? Trends Plant Sci. 18, 133–139 (2013).Article 

Google Scholar 
Nagamalleswari, E., Rao, S., Vasu, K. & Nagaraja, V. Restriction endonuclease triggered bacterial apoptosis as a mechanism for long time survival. Nucleic Acids Res. 45, 8423–8434 (2017).Article 

Google Scholar 
Kysela, D. T., Brown, P. J. B., Huang, K. C. & Brun, Y. V. Biological consequences and advantages of asymmetric bacterial growth. Annu. Rev. Microbiol. 67, 417–435 (2013).Article 

Google Scholar 
Bayles, K. W. Bacterial programmed cell death: making sense of a paradox. Nat. Rev. Microbiol. 12, 63–69 (2014).Article 

Google Scholar 
Flemming, H.-C. & Wuertz, S. Bacteria and archaea on Earth and their abundance in biofilms. Nat. Rev. Microbiol. 17, 247–260 (2019).Article 

Google Scholar 
Coleman, D. C. & Wall, D. H. in Soil Microbiology, Ecology and Biochemistry 4th edn (ed. Paul, E. A.) Ch. 5 (Academic Press, 2015).Hungate, B. A. et al. The functional significance of bacterial predators. mBio 12, e00466-21 (2021).Article 

Google Scholar 
Kuzyakov, Y. & Mason-Jones, K. Viruses in soil: nano-scale undead drivers of microbial life, biogeochemical turnover and ecosystem functions. Soil Biol. Biochem. 127, 305–317 (2018).Article 

Google Scholar 
Williamson, K. E., Fuhrmann, J. J., Wommack, K. E. & Radosevich, M. Viruses in soil ecosystems: an unknown quantity within an unexplored territory. Annu. Rev. Virol. 4, 201–219 (2017).Article 

Google Scholar 
Sokol, N. W. et al. Life and death in the soil microbiome: how ecological processes influence biogeochemistry. Nat. Rev. Microbiol. 20, 415–430 (2022).Article 

Google Scholar 
Bonkowski, M. & Clarholm, M. J. A. P. Stimulation of plant growth through interactions of bacteria and protozoa: testing the auxiliary microbial loop hypothesis. Acta Protozool. 51, 237–247 (2012).
Google Scholar 
Potapov, A. M., Pollierer, M. M., Salmon, S., Šustr, V. & Chen, T.-W. Multidimensional trophic niche revealed by complementary approaches: gut content, digestive enzymes, fatty acids and stable isotopes in Collembola. J. Anim. Ecol. 90, 1919–1933 (2021).Article 

Google Scholar 
Esteban, G. F. & Fenchel, T. M. in Ecology of Protozoa: The Biology of Free-living Phagotrophic Protists (eds Esteban, G. F. & Fenchel, T. M.) 33–54 (Springer, 2020).Koksharova, O. A. Bacteria and phenoptosis. Biochemistry 78, 963–970 (2013).
Google Scholar 
Tilman, D. Resource Competition and Community Structure (Princeton Univ. Press, 1982).Boddy, L. Interspecific combative interactions between wood-decaying basidiomycetes. FEMS Microbiol. Ecol. 31, 185–194 (2000).Article 

Google Scholar 
Hibbing, M. E., Fuqua, C., Parsek, M. R. & Peterson, S. B. Bacterial competition: surviving and thriving in the microbial jungle. Nat. Rev. Microbiol. 8, 15–25 (2010).Article 

Google Scholar 
Müller, S. et al. Predation by Myxococcus xanthus induces Bacillus subtilis to form spore-filled megastructures. Appl. Environ. Microbiol. 81, 203–210 (2015).Article 

Google Scholar 
Laskowska, E. & Kuczynska-Wisnik, D. New insight into the mechanisms protecting bacteria during desiccation. Curr. Genet. 66, 313–318 (2020).Article 

Google Scholar 
Rillig, M. C., Ryo, M. & Lehmann, A. Classifying human influences on terrestrial ecosystems. Glob. Change Biol. 27, 2273–2278 (2021).Article 

Google Scholar 
Dörr, T., Moynihan, P. J. & Mayer, C. Bacterial cell wall structure and dynamics. Front. Microbiol. 10, 02051 (2019).Article 

Google Scholar 
Corredor, B., Lang, B. & Russell, D. Effects of nitrogen fertilization on soil fauna in a global meta-analysis. Preprint at Res. Sq. https://doi.org/10.21203/rs.3.rs-1438491/v1 (2022).Blankinship, J. C., Niklaus, P. A. & Hungate, B. A. A meta-analysis of responses of soil biota to global change. Oecologia 165, 553–565 (2011).Article 

Google Scholar 
Manzoni, S., Chakrawal, A., Spohn, M. & Lindahl, B. D. Modeling microbial adaptations to nutrient limitation during litter decomposition. Front. For. Glob. Change 4, 686945 (2021).Article 

Google Scholar 
Frank, D. et al. Effects of climate extremes on the terrestrial carbon cycle: concepts, processes and potential future impacts. Glob. Change Biol. 21, 2861–2880 (2015).Article 

Google Scholar 
Gunina, A. & Kuzyakov, Y. From energy to (soil organic) matter. Glob. Change Biol. 28, 2169–2182 (2022).Article 

Google Scholar 
Fernandez, C. W. & Koide, R. T. Initial melanin and nitrogen concentrations control the decomposition of ectomycorrhizal fungal litter. Soil Biol. Biochem. 77, 150–157 (2014).Article 

Google Scholar 
Kästner, M., Miltner, A., Thiele-Bruhn, S. & Liang, C. Microbial necromass in soils—linking microbes to soil processes and carbon turnover. Front. Environ. Sci. 9, 756378 (2021).Article 

Google Scholar 
Buckeridge, K. M., Creamer, C. & Whitaker, J. Deconstructing the microbial necromass continuum to inform soil carbon sequestration. Funct. Ecol. 36, 1396–1410 (2022).Article 

Google Scholar 
Lehmann, J. et al. Persistence of soil organic carbon caused by functional complexity. Nat. Geosci. 13, 529–534 (2020).Article 

Google Scholar 
Blazewicz, S. J. et al. Taxon-specific microbial growth and mortality patterns reveal distinct temporal population responses to rewetting in a California grassland soil. ISME J. 14, 1520–1532 (2020).Article 

Google Scholar 
Kallenbach, C. M., Wallenstein, M. D., Schipanksi, M. E. & Grandy, A. S. Managing agroecosystems for soil microbial carbon use efficiency: ecological unknowns, potential outcomes, and a path forward. Front. Microbiol. 10, 1146 (2019).Article 

Google Scholar 
Liang, C. Soil microbial carbon pump: mechanism and appraisal. Soil Ecol. Lett. 2, 241–254 (2020).Article 

Google Scholar 
Sinsabaugh, R. L., Manzoni, S., Moorhead, D. L. & Richter, A. Carbon use efficiency of microbial communities: stoichiometry, methodology and modelling. Ecol. Lett. 16, 930–939 (2013).Article 

Google Scholar 
van Groenigen, J. W. et al. Sequestering soil organic carbon: a nitrogen dilemma. Environ. Sci. Technol. 51, 4738–4739 (2017).Article 

Google Scholar 
Greenlon, A. et al. Quantitative stable-isotope probing (qSIP) with metagenomics links microbial physiology and activity to soil moisture in Mediterranean-climate grassland ecosystems (in the press).Mafla-Endara, P. M. et al. Microfluidic chips provide visual access to in situ soil ecology. Commun. Biol. 4, 889 (2021).Article 

Google Scholar 
Schaible, G. A., Kohtz, A. J., Cliff, J. & Hatzenpichler, R. Correlative SIP-FISH-Raman-SEM-NanoSIMS links identity, morphology, biochemistry, and physiology of environmental microbes. ISME Commun. 2, 52 (2022).Article 

Google Scholar 
See, C. R. et al. Distinct carbon fractions drive a generalisable two-pool model of fungal necromass decomposition. Funct. Ecol. 35, 796–806 (2021).Article 

Google Scholar 
Wang, C. et al. Stabilization of microbial residues in soil organic matter after two years of decomposition. Soil Biol. Biochem. 141, 107687 (2020).Article 

Google Scholar 
Veresoglou, S. D., Halley, J. M. & Rillig, M. C. Extinction risk of soil biota. Nat. Commun. 6, 8862 (2015).Article 

Google Scholar 
Potapov, A. M. et al. Feeding habits and multifunctional classification of soil-associated consumers from protists to vertebrates. Biol. Rev. 97, 1057–1117 (2022).Article 

Google Scholar 
Trap, J., Bonkowski, M., Plassard, C., Villenave, C. & Blanchart, E. Ecological importance of soil bacterivores for ecosystem functions. Plant Soil 398, 1–24 (2016).Article 

Google Scholar 
Dooley, S. R. & Treseder, K. K. The effect of fire on microbial biomass: a meta-analysis of field studies. Biogeochemistry 109, 49–61 (2012).Article 

Google Scholar 
Muñoz-Leoz, B., Ruiz-Romera, E., Antigüedad, I. & Garbisu, C. Tebuconazole application decreases soil microbial biomass and activity. Soil Biol. Biochem. 43, 2176–2183 (2011).Article 

Google Scholar 
Meyer, M., Diehl, D., Schaumann, G. E. & Muñoz, K. Agricultural mulching and fungicides—impacts on fungal biomass, mycotoxin occurrence, and soil organic matter decomposition. Environ. Sci. Pollut. Res. 28, 36535–36550 (2021).Article 

Google Scholar 
Thiery, S. & Kaimer, C. The predation strategy of Myxococcus xanthus. Front. Microbiol. 11, 2 (2020).Article 

Google Scholar 
Laloux, G. Shedding light on the cell biology of the predatory bacterium Bdellovibrio bacteriovorus. Front. Microbiol. 10, 3136 (2020).Article 

Google Scholar  More

Figure 1 presents the average monthly energy expenditure at the household level based on USD across the 37 surveyed nations. The households in Singapore expend the most amount of energy, that is, 748 USD each month on average. The energy consumption appears positively associated with the economic development level; for example, households from high-income countries, including France, Italy, Japan and the US, tend to consume more energy than those from low-income countries (e.g., Kazakhstan, Myanmar, and Mongolia). In India, Indonesia, and Vietnam, households with higher income expend more on energy than rural/slum households. For the energy expenditure to household income ratio, strong trends were not found between developing and developed countries. Notably, middle-income countries (e.g., Greece, Chile, Brazil, Egypt) spend a relatively higher share of total income on energy.Figure 1Average monthly energy expenditure at the household level across the 37 surveyed nations. Data source: Original survey.Full size imageThe relationship between subjective well-being and energy consumption expenditure based on the ordered logit, ordered probit, and OLS models is shown in Table 2, panel A. The LR Chi-Square test and Pseudo R-squared for the ordered logistic regression model and the ordered probit model were applied to measure the goodness of the fit, whereas F-statistics and adjusted R-squared were used for the OLS model. For the validation of the measurement of subjective well-being, life satisfaction and happiness measures were used. Importantly, the results from variated regression models are consistent, indicating a positive relationship between household energy consumption expenditure and the improvement of individuals’ subjective well-being. Regarding the model’s goodness of fit, the LR Chi-Square test with ordered logit and probit models, and the F-statistic in the OLS model are all statistically significant at 0.1%, which validates the regression model. As the consistency of the robustness results is derived from different models, the ordered logit model is applied in Table 2 (Panel B).Table 2 Association between energy consumption expenditure and subjective well-being in high- and non-high-income countries.Full size tableWith the control variables being constant, energy consumption expenditure improves subjective well-being, including life satisfaction and happiness. The coefficients for the relationship of energy consumption with life satisfaction and with happiness are 0.018 and 0.008, respectively, and they are statistically significant at the 1% level; in other words, there is increased energy consumption for people who are satisfied with their lives and are happier. This is because electricity, water, gas, or gasoline are indispensable consumption goods in daily life. The results suggest that when policies lead to a reduction in the consumption of these goods at the household level, the life satisfaction of citizens is likely to decrease. When reducing energy consumption at the household level to reduce the emission of greenhouse gases, the conflicts of interest of individuals in these households (given that they derive life satisfaction from energy consumption) pose a challenge to policymakers; therefore, policymakers should devise strategies to improve both citizens’ living standards and environmental preservation.Referring to the criteria developed by the World Bank, the standard classification of high-income nations and non-high-income nations is as follows. Based on the 2017 gross national income (GNI) per capita, the World Bank List of Economies (June 2018) presented the following criteria for nations to be classified as high-income and non-high-income nations, respectively: a GNI per capita of $12,056 or higher, and less than $12,056. According to this standard of classification, in this study, high-income nations comprise Japan, Singapore, Chile, Australia, the United States, Germany, the United Kingdom, France, Spain, Italy, Sweden, Canada, Netherlands, Greece, Hungary, Poland, and the Czech Republic, whereas non-high-income nations comprise Thailand, Malaysia, Indonesia, Vietnam, Philippines, Mexico, Venezuela, Brazil, Colombia, South Africa, India, Myanmar, Kazakhstan, Mongolia, Egypt, Russia, China, Turkey, Romania, and Sri Lanka.Regarding the comparison of high- and non-high-income countries, energy consumption at the household level is more likely to lead to life satisfaction in non-high-income than in high-income countries. In high-income countries, the coefficients for the relationship of energy consumption with life satisfaction and with happiness are 0.010 and 0.003, respectively; these coefficients are 0.035 and 0.015, respectively, among non-high-income countries. Hence, in both high-income and non-high-income countries, an increase in energy consumption leads to an increase in life satisfaction; nonetheless, energy consumption is more crucial for households in non-high-income countries. Compared to the effect of energy consumption on satisfaction in high-income countries and non-high-income countries, individuals living in less urbanized countries appear more satisfied with energy consumption.Table 3 presents the association between life satisfaction and energy consumption expenditure at the household level in each country by estimating Eq. (2) based on the ordered logit model for each country. There is a positive relationship between energy consumption expenditure and life satisfaction in 27 out of the 37 nations. For example, the coefficient of this relationship is 0.062 in Brazil, and is statistically significant at the 1% level. An increase in energy consumption expenditure positively impacts the life satisfaction of households in Brazil, meaning that individuals with greater energy expenditure tend to be satisfied with their lives. Similar results are found in other countries: Canada, Chile, China, Egypt, France, Germany, Greece, India, Indonesia, Italy, and Japan. As life satisfaction is a proxy of well-being, energy consumption is expected to increase when households can afford more energy to obtain higher life satisfaction. These results indicate that most of the developed and developing countries analyzed face a conflict of interest in addressing individuals’ life satisfaction and environment conservation goals; these countries include China and India that are home to large populations that have a positive desire for energy consumption.Table 3 Relationship between energy expenditure and life satisfaction for each country.Full size tableHowever, the association between life satisfaction and energy consumption expenditure at the household level was non-significant across some countries. In Australia, the coefficient of this association is positive but not statistically significant; hence, an increase in energy expenditure is not completely associated with life satisfaction at the household level here. Similar results are found in the Netherlands, Hungary, Sweden, Singapore, Poland, the Czech Republic, and Colombia. In these countries, energy consumption is at an adequate level, and additional energy consumption does not lead to higher life satisfaction. It may be that households consume an adequate amount of energy with their income and energy price.Tables 4, 5, 6, and 7 display the determinant factors of household energy consumption in 37 nations by estimating the energy demand equation for each country using Eq. (3). The key energy consumption metric is the quantity of energy consumed (e.g., kWh) across the targeted households. Since price information is limited, transforming consumption expenditure into a quantity (e.g., kWh) is problematic. As explained earlier, this study adopted the energy demand equation.Table 4 Household socioeconomic and demographic determinants of household energy consumption expenditure I.Full size tableTable 5 Household socioeconomic and demographic determinants of household energy consumption expenditure II.Full size tableTable 6 Household socioeconomic and demographic determinants of household energy consumption expenditure III.Full size tableTable 7 Household socioeconomic and demographic determinants of household energy consumption expenditure IV.Full size tableThere are positive relationships between energy consumption expenditure at the household level and household income across countries. If the coefficients for household income are positive and statistically significant, this means that energy consumption expenditure at the household level would increase with an increase in household income ensuing from economic development in the country, ceteris paribus. The positive coefficients for the association between energy consumption expenditure and household income range from 0.756 (Japan) to 3.613 (the Philippines) in our sample, indicating that an additional 10,000 USD would lead to an additional energy consumption expenditure at the household level of approximately 17.3% (Japan) – 445% (Mongolia). The number is calculated using the magnitude of the coefficient/energy consumption expenditure. The results also show that homeowners tend to consume more energy than renters in Australia, Brazil, Canada, Chile, China, Colombia, Germany, India, Italy, Japan, Malaysia, Mexico, Russia, the United States, and Vietnam. This indicates that if individuals live in their own houses, the household energy consumption expenditure tends to be higher owing to the wealth effect, as energy is a normal consumption good. Overall, the wealth effect on energy consumption expenditure at the household level is increasing in our sample, and with economic development, energy consumption may increase.The following factors are confirmed to reduce energy consumption at the household level: (1) energy-curtailment behavior regarding electricity, (2) higher education, and (3) age. The energy-saving effect is confirmed in households. In Canada, the coefficient of energy-saving behaviors is -0.642, indicating that households consume 12.5% less energy when they adopt both energy curtailment behavior and non-saving groups (64.2/513). The Canadian household average energy consumption is 513 USD. Similar results are seen in Colombia, Germany, India, Indonesia, Italy, Japan, the Netherlands, Poland, Russia, Turkey, the United Kingdom, and the United States. The magnitude of the effect of energy curtailment behavior ranged from 6.4% (Russia) to 32% (India) less energy consumption expenditure. Hence, energy-saving behaviors have a favorable effect on environmentally preferable outcomes. By contrast, households in Indonesia save electricity as they tend to spend more on purchasing energy.Individuals with higher education tend to save energy in 23 out of the 37 nations. For instance, the coefficient for individuals with university-level education is -2.292 and statistically significant at the 1% level. This suggests that households with individuals who have university-level education have less energy consumption expenditure than households with individuals with junior high school or lower levels of education. Similar results are seen in Brazil, Canada, Chile, Colombia, the Czech Republic, France, Germany, Hungary, India, Indonesia, Japan, Malaysia, the Netherlands, the Philippines, Poland, Russia, Singapore, South Africa, Spain, Sweden, Turkey, the United Kingdom, and the United States. Encouraging households to engage in energy curtailment behaviors and higher educational attainment may lead to environment-friendly outcomes.Surprisingly, purchasing energy-saving household products has a limited effect on reducing energy consumption expenditure at the household level. The coefficients for purchasing energy-saving household products are negative, ranging between -0.044 and -0.763, and are statistically significant in Australia, Canada, the Czech Republic, Italy, and Kazakhstan. Hence, the purchase of these products in these five countries decreases energy expenditure from 2.9% (China) to 14% (Australia). However, the relationship between energy consumption expenditure at the household level and purchasing energy-saving household products is non-significant in the other countries. Moreover, in Poland and Turkey, households that purchase these products consume more energy than those that do not. Therefore, purchasing energy-saving household products has a limited contribution to energy saving at the household level.The findings also show that older individuals tend to have lower energy consumption. The coefficients for the age variable are negative and statistically significant in 30 countries (out of 37). The effect of age on energy consumption expenditure ranges between -0.003 and -0.148, indicating that as the average age of individuals increases by one year, their monthly energy consumption expenditure reduces from 0.3–14.8 USD. This may be because older individuals are more likely to live frugally. More

DatasetsImage acquisitionFor this study, we assembled three datasets: one for training of the RootDetector Convolutional Neural Network (Training-Set), one for a performance comparison between humans and RootDetector in segmenting roots in minirhizotron images (Comparison-Set), and one for the validation of the algorithm (Validation-Set). The Training-Set contained 129 images comprised of 17 randomly selected minirhizotron images sampled in a mesocosm experiment (see “Mesocosm sampling” Section), 47 randomly selected minirhizotron images sampled in a field study (see “Field sampling” Section) as well as the 65 minirhizotron images of soy roots published by Wang et al.15. The Comparison-Set contained 25 randomly selected minirhizotron images from the field-study which all were not part of the images included in the Training- and Validation-Sets. The Validation-Set contained 10 randomly selected minirhizotron images from the same field study, which had not been used in the Training-Set. All images were recorded with 2550 ✕ 2273 pixels at 300 dpi with a CI-600 In-Situ Root Imager (CID Bio-Science Inc., Camas, WA, USA) and stored as .tiff files to reduce compression loss. For all training and evaluation purposes we used raw, unprocessed output images from the CI-600.Mesocosm samplingThe mesocosm experiment was established in 2018 on the premises of the Institute for Botany and Landscape Ecology of the University of Greifswald (Fig. S1). It features 108 heavy duty plastic buckets of 100 l each, filled to two thirds of their height with moderately decomposed sedge fen peat. Each mesocosm contained one minirhizotron (inner diameter: 64 mm, outer diameter: 70 mm, length: 650 mm) installed at a 45°angle and capped in order to avoid penetration by light. The mesocosms were planted with varying compositions of plant species that typically occur in north-east German sedge fens (Carex rostrata, Carex acutiformis, Glyceria maxima, Equisetum fluviatile, Juncus inflexus, Mentha aquatica, Acorus calamus and Lycopus europaeus). The mesocosms were subjected to three different water table regimes: stable at soil surface level, stable at 20 cm below soil surface and fluctuating between the two levels every two weeks. The minirhizotrons were scanned weekly at two levels of soil depth (0–20 cm and 15–35 cm) between April 2019 and December 2021, resulting in roughly 9500 minirhizotron images of 216 × 196 mm. Manual quantification of root length would, based on own experience, take approximately three hours per image, resulting in approximately 28,500 h of manual processing for the complete dataset. Specimens planted were identified by author Dr. Blume-Werry, however no voucher specimen were deposited. All methods were carried out in accordance with relevant institutional, national, and international guidelines and legislation.Field samplingThe field study was established as part of the Wetscapes project in 201716. The study sites were located in Mecklenburg-Vorpommern, Germany, in three of the most common wetland types of the region: alder forest, percolation fen and coastal fen (Fig. S2). For each wetland type, a pair of drained versus rewetted study sites was established. A detailed description of the study sites and the experimental setup can be found in Jurasinski et al.16. At each site, 15 minirhizotrons (same diameter as above, length: 1500 mm) were installed at 45° angle along a central boardwalk. The minirhizotrons have been scanned biweekly since April 2018, then monthly since January 2019 at two to four levels of soil depth (0–20 cm, 20–40 cm, 40–60 cm and 60–80 cm), resulting in roughly 12,000 minirhizotron images of 216 × 196 cm, i.e. an estimated 36,000 h of manual processing for the complete dataset. Permission for the study was obtained from the all field owners. Figure 1Overview of the RootDetector system. The main component is a semantic segmentation network based on the U-Net architecture. The root length is estimated by skeletonizing the segmentation output and applying the formula introduced by Kimura et al.17. During training only, a weight map puts more emphasis on fine roots.Full size imageThe CNN RootDetectorImage annotationFor the generation of training data for the CNN, human analysts manually masked all root pixels in the 74 images of the Training-Set using GIMP 2.10.12. The resulting ground truth data are binary, black-and-white images in Portable Network Graphics (.png) format, where white pixels represent root structures and black pixels represent non-root objects and soil (Fig. 2b). All training data were checked and, if required, corrected by an expert (see “Selection of participants” for definition). The Validation-Set was created in the same way but exclusively by experts.Figure 2Example of segmentation and result of skeletonization. A 1000 by 1000 pixel input image (a), the manually annotated ground truth image (b), the RootDetector estimation image (c), the combined representation image (error map, d with green indicating true positives, red indicating false positive, blue indicating false negatives), the skeletonized RootDetector estimation image (e), and the skeletonized ground truth image (f).Full size imageArchitectureRootDetector’s core consists of a Deep Neural Network (DNN) based on the U-Net image segmentation architecture[27]nd is implemented in TensorFlow and Keras frameworks18. Although U-Net was originally developed for biomedical applications, it has since been successfully applied to other domains due to its generic design.RootDetector is built up of four down-sampling blocks, four up-sampling blocks and a final output block (Fig. 1). Every block contains two 3 × 3 convolutional layers, each followed by rectified linear units (ReLU). The last output layer instead utilizes Sigmoid activation. Starting from initial 64 feature channels, this number is doubled in every down-block and the resolution is halved via 2 × 2 max-pooling. Every up-block again doubles the resolution via bilinear interpolation and a 1 × 1 convolution which halves the number of channels. Importantly, after each up-sampling step, the feature map is concatenated with the corresponding feature map from the down-sampling path. This is crucial to preserve fine spatial details.Our modifications from the original architecture include BatchNormalization19 after each convolutional layer which significantly helps to speed up the training process and zero-padding instead of cropping as suggested by Ronneberger, Fischer, & Brox20 to preserve the original image size.In addition to the root segmentation network, we trained a second network to detect foreign objects, specifically the adhesive tape that is used as a light barrier on the aboveground part of the minirhizotrons. We used the same network architecture as above and trained in a supervised fashion with the binary cross-entropy loss. During inference, the result is thresholded (predefined threshold value: 0.5) and used without post-processing.TrainingWe pre-trained RootDetector on the COCO dataset21 to generate a starting point. Although the COCO dataset contains a wide variety of image types and classes not specifically related to minirhizotron images, Majurski et al.22 showed, that for small annotation counts, transfer-learning even from unrelated datasets may improve a CNNs performance by up to 20%. We fine-tuned for our dataset with the Adam optimizer23 for 15 epochs and trained on a total of 129 images from the Training-Set (17 mesocosm images, 47 field-experiment images, 65 soy root images). To enhance the dataset size and reduce over-fitting effects, we performed a series of augmentation operations as described by Shorten & Khoshgoftaar24. In many images, relatively coarse roots ( > 3 mm) occupied a major part of the positive (white) pixel space, which might have caused RootDetector to underestimate fine root details overall. Similarly, negative space (black pixels) between tightly packed, parallel roots was often very small and might have impacted the training process to a lesser extent when compared to large areas with few or no roots (Fig. 2). To mitigate both effects, we multiplied the result of the cross-entropy loss map with a weight map which emphasizes positive–negative transitions. This weight map is generated by applying the following formula to the annotated ground truth images:$$omega left( x right) = 1 – left( {tanh left( {2tilde{x} – 1} right)} right)^{2}$$
(1)
where ω(x) is the average pixel value of the annotated weight map in a 5 × 5 neighborhood around pixel x. Ronneberger, Fischer, & Brox20 implemented a similar weight map, however with stronger emphasis on space between objects. As this requires computation of distances between two comparatively large sets of points, we adapted and simplified their formula to be computable in a single 5 × 5 convolution.For the loss function we applied a combination of cross-entropy and Dice loss 25:$${mathcal{L}} = {mathcal{L}}_{CE} + lambda {mathcal{L}}_{Dice} = – frac{1}{N}sumnolimits_{i} {wleft( {x_{i} } right)y_{i} log left( {x_{i} } right) + lambda frac{{2sumnolimits_{i} {x_{i} y_{i} } }}{{sumnolimits_{i} {x_{i}^{2} sumnolimits_{i} {y_{i}^{2} } } }}}$$
(2)

where x are the predicted pixels, y the corresponding ground truth labels, N the number of pixels in an image and λ a balancing factor which we set to 0.01. This value was derived empirically. The Dice loss is applied per-image to counteract the usually high positive-to-negative pixel imbalance. Since this may produce overly confident outputs and restrict the application of weight maps, we used a relatively low value for λ.Output and post-processingRootDetector generates two types of output. The first type of output are greyscale .png files in which white pixels represent pixels associated with root structures and black pixels represent non-root structures and soil (Fig. 2c). The advantage of .png images is their lossless ad artifact-free compression at relatively small file sizes. RootDetector further skeletonizes the output images and reduces root-structures to single-pixel representations using the skeletonize function of scikit-image v. 0.17.1 (26; Fig. 2e,f). This helps to reduce the impact of large diameter roots or root-like structures such as rhizomes in subsequent analyses and is directly comparable to estimations of root length. The second type of output is a Comma-separated values (.csv) file, with numerical values indicating the number of identified root pixels, the number of root pixels after skeletonization, the number of orthogonal and diagonal connections between pixels after skeletonization and an estimation of the physical combined length of all roots for each processed image. The latter is a metric commonly used in root research as in many species, fine roots provide most vital functions such as nutrient and water transport3. Therefore, the combined length of all roots in a given space puts an emphasis on fine roots as they typically occupy a relatively smaller fraction of the area in a 2D image compared to often much thicker coarse roots. To derive physical length estimates from skeletonized images, RootDetector counts orthogonal- and diagonal connections between pixels of skeletonized images and employs the formula proposed by Kimura et al.17 (Eq. 3).$$L = left[ {N_{d}^{2} + left( {N_{d} + N_{o} /2} right)^{2} } right]^{{1/2}} + N_{o} /2$$
(3)
where Nd is the number of diagonally connected and No the number of orthogonally connected skeleton pixels. To compute Nd we convolve the skeletonized image with two 2 × 2 binary kernels, one for top-left-to-bottom-right connections and another for bottom-left-to-top-right connections and count the number of pixels with maximum response in the convolution result. Similarly, No is computed with a 1 × 2 and a 2 × 1 convolutional kernels.Performance comparisonSelection of participantsFor the performance comparison, we selected 10 human analysts and divided them into three groups of different expertise levels in plant physiology and with the usage of digital root measuring tools. The novice group consisted of 3 ecology students (2 bachelor’s, 1 master’s) who had taken or were taking courses in plant physiology but had no prior experience with minirhizotron images or digital root measuring tools. This group represents undergraduate students producing data for a Bachelor thesis or student assistants employed to process data. The advanced group consisted of 3 ecology students (1 bachelor’s, 2 master’s) who had already taken courses in plant physiology and had at least 100 h of experience with minirhizotron images and digital root measuring tools. The expert group consisted of 4 scientists (2 PhD, 2 PhD candidates) who had extensive experience in root science and at least 250 h of experience with digital root measuring tools. All methods were carried out in accordance with relevant institutional, national, and international guidelines and legislation and informed consent was obtained from all participants.Instruction and root tracingAll three groups were instructed by showing them a 60 min live demo of an expert tracing roots in minirhizotron images, during which commonly encountered challenges and pitfalls were thoroughly discussed. Additionally, all participants were provided with a previously generated, in-depth manual containing guidelines on the identification of root structures, the correct operation of the root tracing program and examples of often encountered challenges and suggested solutions. Before working on the Comparison-Set, all participants traced roots in one smaller-size sample image and received feedback from one expert.Image preparation and root tracingBecause the minirhizotron images acquired in the field covered a variety of different substrates, roots of different plant species, variance in image quality, and because tracing roots is very time consuming, we decided to maximize the number of images by tracing roots only in small sections, in order to cover the largest number of cases possible. To do this, we placed a box of 1000 × 1000 pixels (8.47 × 8.47 cm) at a random location in each of the images in the Comparison-Set and instructed participants to trace only roots within that box. Similarly, we provided RootDetector images where the parts of the image outside the rectangle were occluded. All groups used RootSnap! 1.3.2.25 (CID Bio-Science Inc., Camas, WA, USA;27), a vector based tool to manually trace roots in each of the 25 images in the comparison set. We decided on RootSnap! due to our previous good experience with the software and its’ relative ease of use. The combined length of all roots was then exported as a csv file for each person and image and compared to RootDetector’s output of the Kimura root length.ValidationWe tested the accuracy of RootDetector on a set of 10 image segments of 1000 by 1000 pixels cropped from random locations of the 10 images of the Validation-Set. These images were annotated by a human expert without knowledge of the estimations by the algorithm and were exempted from the training process. As commonly applied in binary classification, we use the F1 score as a metric to evaluate the performance RootDetector. F1 is calculated from precision (Eq. 4) and recall (Eq. 5) and represents their harmonic mean (Eq. 6). Ranging from 0 to 1, higher values indicate high classification (segmentation) performance. As one of the 10 image sections contained no roots and thus no F1 Score was calculable, it was excluded from the validation. We calculated the F1 score for each of the nine remaining image sections and averaged the values as a metric for overall segmentation performance.$$Precision;(P) = frac{{tp}}{{tp + fp}}$$
(4)
$$Recall;(R) = frac{{tp}}{{tp + fn}}$$
(5)
$$F1 = 2*frac{{P*R}}{{P + R}}$$
(6)
where P = precision, R = recall, tp = true positives; fp = false positives, fn = false negatives.Statistical analysisWe used R Version 4.1.2 (R Core Team, 2021) for all statistical analyses and R package ggplot2 Version 3.2.128 for visualizations. Pixel identification-performance comparisons were based on least-squares fit and the Pearson method. Root length estimation-performance comparisons between groups of human analysts (novice, advanced, expert) and RootDetector were based on the respective estimates of total root length plotted over the minirhizotron images in increasing order of total root length. Linear models were calculated using the lm function for each group of analysts. To determine significant differences between the groups and the algorithm, 95% CIs as well as 83% CIs were displayed and RootDetector root length outside the 95% CI were considered significantly different from the group estimate at α = 0.0529. The groups of human analysts were considered significantly different if their 83% CIs did not overlap, as the comparison of two 83% CIs approximates an alpha level of 5%30,31.This study is approved by Ethikkommission der Universitätsmedizin Greifswald, University of Greifswald, Germany. More

Foley, J. A. et al. Global consequences of land use. Science 309, 570–574 (2005).Article 
CAS 

Google Scholar 
Rockström, J. et al. A safe operating space for humanity. Nature 461, 472–475 (2009).Article 

Google Scholar 
Geiger, F. et al. Persistent negative effects of pesticides on biodiversity and biological control potential on European farmland. Basic Appl. Ecol. 11, 97–105 (2010).Article 
CAS 

Google Scholar 
Zhang, W., Ricketts, T. H., Kremen, C., Carney, K. & Swinton, S. M. Ecosystem services and dis-services to agriculture. Ecol. Econ. 64, 253–260 (2007).Article 

Google Scholar 
Tilman, D., Cassman, K. G., Matson, P. A., Naylor, R. & Polasky, S. Agricultural sustainability and intensive production practices. Nature 418, 671–677 (2002).Article 
CAS 

Google Scholar 
Helming, K. et al. Managing soil functions for a sustainable bioeconomy-assessment framework and state of the art. Land Degrad. Dev. 29, 3112–3126 (2018).Article 

Google Scholar 
Rockström, J. et al. Sustainable intensification of agriculture for human prosperity and global sustainability. Ambio 46, 4–17 (2017).Article 

Google Scholar 
Lehmann, J., Bossio, D. A., Kögel-Knabner, I. & Rillig, M. C. The concept and future prospects of soil health. Nat. Rev. Earth Environ. 1, 544–553 (2020).Article 

Google Scholar 
Smith, J., Pearce, B. D. & Wolfe, M. S. Reconciling productivity with protection of the environment: Is temperate agroforestry the answer? Renew. Agric. Food Syst. 28, 80–92 (2013).Article 

Google Scholar 
European Commission. A Greener and Fairer CAP (EC, 2021).Grass, I. et al. Trade-offs between multifunctionality and profit in tropical smallholder landscapes. Nat. Commun. 11, 1186 (2020).Article 
CAS 

Google Scholar 
Mayer, S. et al. Soil organic carbon sequestration in temperate agroforestry systems – a meta-analysis. Agric. Ecosyst. Environ. 323, 107689 (2022).Article 
CAS 

Google Scholar 
Pardon, P. et al. Juglans regia (walnut) in temperate arable agroforestry systems: effects on soil characteristics, arthropod diversity and crop yield. Renew. Agric. Food Syst. 35, 533–549 (2020).Article 

Google Scholar 
Schmidt, M. et al. Nutrient saturation of crop monocultures and agroforestry indicated by nutrient response efficiency. Nutr. Cycl. Agroecosyst. 119, 69–82 (2021).Article 
CAS 

Google Scholar 
Beule, L. & Karlovsky, P. Tree rows in temperate agroforestry croplands alter the composition of soil bacterial communities. PLoS ONE 16, e0246919 (2021).Article 
CAS 

Google Scholar 
Palma, J. H. N. et al. Modeling environmental benefits of silvoarable agroforestry in Europe. Agric. Ecosyst. Environ. 119, 320–334 (2007).Article 

Google Scholar 
Kay, S. et al. Spatial similarities between European agroforestry systems and ecosystem services at the landscape scale. Agroforest Syst. 92, 1075–1089 (2018).Article 

Google Scholar 
Swieter, A., Langhof, M., Lamerre, J. & Greef, J. M. Long-term yields of oilseed rape and winter wheat in a short rotation alley cropping agroforestry system. Agroforest Syst. 93, 1853–1864 (2019).Article 

Google Scholar 
Ivezić, V., Yu, Y. & van der Werf, W. Crop yields in European agroforestry systems: a meta-analysis. Front. Sustain. Food Syst. 5, 606631 (2021).Article 

Google Scholar 
Cardinael, R. et al. High organic inputs explain shallow and deep SOC storage in a long-term agroforestry system – combining experimental and modeling approaches. Biogeosciences 15, 297–317 (2018).Article 
CAS 

Google Scholar 
Smith, P. Carbon sequestration in croplands: the potential in Europe and the global context. Eur. J. Agron. 20, 229–236 (2004).Article 
CAS 

Google Scholar 
Kay, S. et al. Agroforestry creates carbon sinks whilst enhancing the environment in agricultural landscapes in Europe. Land Use Policy 83, 581–593 (2019).Article 

Google Scholar 
Cardinael, R. et al. Impact of alley cropping agroforestry on stocks, forms and spatial distribution of soil organic carbon — a case study in a Mediterranean context. Geoderma 259–260, 288–299 (2015).Article 

Google Scholar 
Cardinael, R. et al. Spatial variation of earthworm communities and soil organic carbon in temperate agroforestry. Biol. Fertil. Soils 55, 171–183 (2019).Article 
CAS 

Google Scholar 
Boinot, S. et al. Alley cropping agroforestry systems: reservoirs for weeds or refugia for plant diversity? Agric. Ecosyst. Environ. 284, 106584 (2019).Article 

Google Scholar 
Barnes, A. D. et al. Direct and cascading impacts of tropical land-use change on multi-trophic biodiversity. Nat. Ecol. Evol. 1, 1511–1519 (2017).Article 

Google Scholar 
Kehoe, L. et al. Biodiversity at risk under future cropland expansion and intensification. Nat. Ecol. Evol. 1, 1129–1135 (2017).Article 

Google Scholar 
DuPont, S. T., Culman, S. W., Ferris, H., Buckley, D. H. & Glover, J. D. No-tillage conversion of harvested perennial grassland to annual cropland reduces root biomass, decreases active carbon stocks, and impacts soil biota. Agric. Ecosyst. Environ. 137, 25–32 (2010).Article 
CAS 

Google Scholar 
Bengtsson, J. et al. Grasslands-more important for ecosystem services than you might think. Ecosphere 10, e02582 (2019).Article 

Google Scholar 
Beule, L. et al. Conversion of monoculture cropland and open grassland to agroforestry alters the abundance of soil bacteria, fungi and soil-N-cycling genes. PLoS ONE 14, e0218779 (2019).Article 
CAS 

Google Scholar 
Borrelli, P., Ballabio, C., Panagos, P. & Montanarella, L. Wind erosion susceptibility of European soils. Geoderma 232–234, 471–478 (2014).Article 

Google Scholar 
Amundson, R. et al. Soil and human security in the 21st century. Science 348, 12610711–12610716 (2015).Article 

Google Scholar 
Olson, K. R., Al-Kaisi, M., Lal, R. & Cihacek, L. Impact of soil erosion on soil organic carbon stocks. J. Soil Water Conserv. 71, 61A–67A (2016).Article 

Google Scholar 
Larney, F. J., Bullock, M. S., Janzen, H. H., Ellert, B. H. & Olson, E. C. S. Wind erosion effects on nutrient redistribution and soil productivity. J. Soil Water Conserv. 53, 133–140 (1998).
Google Scholar 
de Jong, E. & Kowalchuk, T. E. The effect of shelterbelts on erosion and soil properties. Soil Sci. 159, 337–345 (1995).Article 

Google Scholar 
Deutsch, M. & Otter, V. Nachhaltigkeit und förderung? Akzeptanzfaktoren im Entscheidungsprozess deutscher Landwirte zur Anlage von Agroforstsystemen. Berichte über Landwirtschaft – Zeitschrift für Agrarpolitik und Landwirtschaft Aktuelle Beiträge (2021).Tsonkova, P., Böhm, C., Quinkenstein, A. & Freese, D. Ecological benefits provided by alley cropping systems for production of woody biomass in the temperate region: a review. Agroforest Syst. 85, 133–152 (2012).Article 

Google Scholar 
Lehmann, J., Weigl, D., Droppelmann, K., Huwe, B. & Zech, W. Nutrient cycling in an agroforestry system with runoff irrigation in Northern Kenya. Agroforestry Syst. 43, 49–70 (1998).Article 

Google Scholar 
Shao, G. et al. Impacts of monoculture cropland to alley cropping agroforestry conversion on soil N2O emissions. GCB Bioenergy https://doi.org/10.1111/gcbb.13007 (2022).Isaac, M. E. & Borden, K. A. Nutrient acquisition strategies in agroforestry systems. Plant Soil 444, 1–19 (2019).Article 
CAS 

Google Scholar 
Cannell, M. G. R., van Noordwijk, M. & Ong, C. K. The central agroforestry hypothesis: the trees must acquire resources that the crop would not otherwise acquire. Agroforestry Syst. 34, 27–31 (1996).Article 

Google Scholar 
Beule, L., Vaupel, A. & Moran-Rodas, V. E. Abundance, diversity, and function of soil microorganisms in temperate alley-cropping agroforestry systems: a review. Microorganisms 10, 616 (2022).Article 
CAS 

Google Scholar 
Thevathasan, N. V. & Gordon, A. M. in New Vistas in Agroforestry, Vol. 1 (eds Nair, P. K. R., Rao, M. R. & Buck, L. E.) 257–268 (Springer Netherlands, 2004).Veldkamp, E. & Keller, M. Fertilizer-induced nitric oxide emissions from agricultural soils. Nutr. Cycling Agroecosyst. 48, 69–77 (1997).Article 
CAS 

Google Scholar 
Luo, J., Beule, L., Shao, G., Veldkamp, E. & Corre, M. D. Reduced soil gross N2O emission driven by substrates rather than denitrification gene abundance in cropland agroforestry and monoculture. JGR Biogeosciences 127, e2021JG006629 (2022).Article 
CAS 

Google Scholar 
Langenberg, J., Feldmann, M. & Theuvsen, L. Alley cropping agroforestry systems: using Monte-Carlo simulation for a risk analysis in comparison with arable farming systems. German J. Agric. Econ. 67, 95–112 (2018).
Google Scholar 
Otter, V. & Langenberg, J. Willingness to pay for environmental effects of agroforestry systems: a PLS-model of the contingent evaluation from German taxpayers’ perspective. Agroforest Syst. 94, 811–829 (2020).Article 

Google Scholar 
Zhang, X. et al. Quantification of global and national nitrogen budgets for crop production. Nat. Food 2, 529–540 (2021).Article 

Google Scholar 
Markwitz, C., Knohl, A. & Siebicke, L. Evapotranspiration over agroforestry sites in Germany. Biogeosciences 17, 5183–5208 (2020).Article 

Google Scholar 
Pardon, P. et al. Trees increase soil organic carbon and nutrient availability in temperate agroforestry systems. Agric. Ecosyst. Environ. 247, 98–111 (2017).Article 
CAS 

Google Scholar 
European Commission. Commission regulation (EC) No 1120/2009 (EC, 2009).Piñeiro, V. et al. A scoping review on incentives for adoption of sustainable agricultural practices and their outcomes. Nat. Sustain. 3, 809–820 (2020).Article 

Google Scholar 
Kay, S. et al. Agroforestry is paying off – economic evaluation of ecosystem services in European landscapes with and without agroforestry systems. Ecosyst. Serv. 36, 100896 (2019).Article 

Google Scholar 
European Council. Council agrees its position on the next EU common agricultural policy. Press release. https://www.consilium.europa.eu/en/press/press-releases/2020/10/21/council-agrees-its-position-on-the-next-eu-common-agricultural-policy/ (2020).IUSS Working Group WRB. World Reference Base for Soil Resources 2014. International Soil Classification System for Naming Soils and Creating Legends for Soil Maps (FAO, 2014).Garland, G. et al. A closer look at the functions behind ecosystem multifunctionality: a review. J. Ecol. 109, 600–613 (2021).Article 

Google Scholar 
Naumann, C. & Bassler, R. Die Chemische Untersuchung von Futtermitteln 3. Auflage (Chemical Analysis of Feedstuff 3rd Edition) (VDLUFA-Verlag, 1976).Beule, L., Lehtsaar, E., Rathgeb, A. & Karlovsky, P. Crop diseases and mycotoxin accumulation in temperate agroforestry systems. Sustainability 11, 2925 (2019).Article 
CAS 

Google Scholar 
Verwijst, T. & Telenius, B. Biomass estimation procedures in short rotation forestry. For. Ecol. Manag. 121, 137–146 (1999).Article 

Google Scholar 
Harris, D., Horwáth, W. R. & van Kessel, C. Acid fumigation of soils to remove carbonates prior to total organic carbon or carbon-13 isotopic analysis. Soil Sci. Soc. Am. J. 65, 1853–1856 (2001).Article 
CAS 

Google Scholar 
Blake, G. & Hartge, K. in Methods of Soil Analysis: Part 1 – Physical and Mineralogical Methods 363–375 (Americal Society of Agronomy, Inc., 1995).Davidson, E. A., Hart, S. C., Shanks, C. A. & Firestone, M. K. Measuring gross nitrogen mineralization, and nitrification by 15N isotopic pool dilution in intact soil cores. J. Soil Sci. 42, 335–349 (1991).Article 
CAS 

Google Scholar 
Tiessen, H. & Moir, J. O. in Soil Sampling and Methods of Analysis Ch. 25 (CRC Press, 1993).Beule, L. et al. Poplar rows in temperate agroforestry croplands promote bacteria, fungi, and denitrification genes in soils. Front. Microbiol. 10, 3108 (2020).Article 

Google Scholar 
Ando, S. et al. Detection of nifH sequences in sugarcane (Saccharum officinarum L.) and pineapple (Ananas comosus [L.] Merr.). Soil Sci. Plant Nutr. 51, 303–308 (2005).Article 
CAS 

Google Scholar 
Singh, J., Singh, S. & Vig, A. P. Extraction of earthworm from soil by different sampling methods: a review. Environ. Dev. Sustain. 18, 1521–1539 (2016).Article 

Google Scholar 
Brookes, P. C., Landman, A., Pruden, G. & Jenkinson, D. S. Chloroform fumigation and the release of soil nitrogen: a rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biol. Biochem. 17, 837–842 (1985).Article 
CAS 

Google Scholar 
Shen, S. M., Pruden, G. & Jenkinson, D. S. Mineralization and immobilization of nitrogen in fumigated soil and the measurement of microbial biomass nitrogen. Soil Biol. Biochem. 16, 437–444 (1984).Article 
CAS 

Google Scholar 
Marx, M.-C., Wood, M. & Jarvis, S. C. A microplate fluorimetric assay for the study of enzyme diversity in soils. Soil Biol. Biochem. 33, 1633–1640 (2001).Article 
CAS 

Google Scholar 
Matson, A. L., Corre, M. D., Langs, K. & Veldkamp, E. Soil trace gas fluxes along orthogonal precipitation and soil fertility gradients in tropical lowland forests of Panama. Biogeosciences 14, 3509–3524 (2017).Article 
CAS 

Google Scholar 
Wen, Y., Corre, M. D., Schrell, W. & Veldkamp, E. Gross N2O emission and gross N2O uptake in soils under temperate spruce and beech forests. Soil Biol. Biochem. 112, 228–236 (2017).Article 
CAS 

Google Scholar 
McKenzie, N. J., Green, T. W. & Jacquier, D. W. in Soil Physical Measurement and Interpretation for Land Evaluation 150–162 (Csiro Publishing, 2002).Priesack, E. Expert-N model library documentation. https://expert-n.uni-hohenheim.de/en/documentation (2005).Formaglio, G., Veldkamp, E., Duan, X., Tjoa, A. & Corre, M. D. Herbicide weed control increases nutrient leaching compared to mechanical weeding in a large-scale oil palm plantation. Biogeosciences 17, 5243–5262 (2020).Article 
CAS 

Google Scholar 
Kroetsch, D. & Wang, C. in Soil sampling and methods of analysis (eds Angers, D. A. & Larney, F. J.) 713–725 (CRC Press, 2008).Kurniawan, S. et al. Conversion of tropical forests to smallholder rubber and oil palm plantations impacts nutrient leaching losses and nutrient retention efficiency in highly weathered soils. Biogeosciences 15, 5131–5154 (2018).Article 
CAS 

Google Scholar 
Markwitz, C. Micrometeorological Measurements and Numerical Simulations of Turbulence and Evapotranspiration over Agroforestry (University of Göttingen, 2021).Jarrah, M., Mayel, S., Tatarko, J., Funk, R. & Kuka, K. A review of wind erosion models: data requirements, processes, and validity. Catena 187, 104388 (2020).Article 

Google Scholar 
van Ramshorst, J. G. V. et al. Reducing wind erosion through agroforestry: a case study using large eddy simulations. Sustainability 14, 13372 (2022).Article 

Google Scholar 
Kanzler, M., Böhm, C., Mirck, J., Schmitt, D. & Veste, M. Microclimate effects on evaporation and winter wheat (Triticum aestivum L.) yield within a temperate agroforestry system. Agroforest Syst. 93, 1821–1841 (2019).Article 

Google Scholar 
Clough, Y. et al. Land-use choices follow profitability at the expense of ecological functions in Indonesian smallholder landscapes. Nat. Commun. 7, 13137 (2016).Article 
CAS 

Google Scholar  More

Sapolsky, R. M. The influence of social hierarchy on primate health. Science 308, 648–652 (2005).Article 
ADS 
CAS 

Google Scholar 
Snyder-Mackler, N. et al. Social status alters immune regulation and response to infection in macaques. Science 354, 1041–1045 (2016).Article 
ADS 
CAS 

Google Scholar 
Levy, E. J. et al. Higher dominance rank is associated with lower glucocorticoids in wild female baboons: A rank metric comparison. Horm. Behav. 125, 104826 (2020).Article 
CAS 

Google Scholar 
Sapolsky, R. M. Social status and health in humans and other animals. Annu. Rev. Anthropol. 33, 393–418 (2004).Article 

Google Scholar 
Goymann, W. & Wingfield, J. C. Allostatic load, social status and stress hormones: the costs of social status matter. Anim. Behav. 67, 591–602 (2004).Article 

Google Scholar 
Cavigelli, S. A. & Chaudhry, H. S. Social status, glucocorticoids, immune function, and health: Can animal studies help us understand human socioeconomic-status-related health disparities?. Horm. Behav. 62, 295–313 (2012).Article 
CAS 

Google Scholar 
Meyer, J. S. & Hamel, A. F. Models of stress in nonhuman primates and their relevance for human psychopathology and endocrine dysfunction. ILAR J. 55, 347–360 (2014).Article 
CAS 

Google Scholar 
Saltzman, W., Schultz-Darken, N. J., Scheffler, G., Wegner, F. H. & Abbott, D. H. Social and reproductive influences on plasma cortisol in female marmoset monkeys. Physiol. Behav. 56, 801–810 (1994).Article 
CAS 

Google Scholar 
Abbott, D. H. et al. Are subordinates always stressed? A comparative analysis of rank differences in cortisol levels among primates. Horm. Behav. 43, 67–82 (2003).Article 
CAS 

Google Scholar 
Sadoughi, B., Lacroix, L., Berbesque, C., Meunier, H. & Lehmann, J. Effects of social tolerance on stress: Hair cortisol concentrations in the tolerant Tonkean macaques (Macaca tonkeana) and the despotic long-tailed macaques (Macaca fascicularis). Stress 1, 1–9 (2021).
Google Scholar 
Kawachi, I. & Berkman, L. Social cohesion, social capital, and health. Social Epidemiol. 174, 290–314 (2000).
Google Scholar 
Dong, M. et al. Insights into causal pathways for ischemic heart disease: adverse childhood experiences study. Circulation 110, 1761–1766 (2004).Article 

Google Scholar 
Galobardes, B., Lynch, J. W. & Davey Smith, G. Childhood socioeconomic circumstances and cause-specific mortality in adulthood: Systematic review and interpretation. Epidemiol. Rev. 26, 7–21 (2004).Article 

Google Scholar 
Lockwood, K. G., John-Henderson, N. A. & Marsland, A. L. Early life socioeconomic status associates with interleukin-6 responses to acute laboratory stress in adulthood. Physiol. Behav. 188, 212–220 (2018).Article 
CAS 

Google Scholar 
Taylor, S. E. Mechanisms linking early life stress to adult health outcomes. Proc. Natl. Acad. Sci. 107, 8507–8512 (2010).Article 
ADS 
CAS 

Google Scholar 
Uchino, B. N. Social Support and Physical Health: Understanding the Health Consequences of Relationships (Yale University Press, 2004).Book 

Google Scholar 
Holt-Lunstad, J. & Uchino, B. N. Social support and health. Health Behav. Theory Res. Pract. 1, 183–204 (2015).
Google Scholar 
Gurven, M., Allen-Arave, W., Hill, K. & Hurtado, A. M. Reservation food sharing among the Ache of Paraguay. Hum. Nat. 12, 273–297 (2001).Article 
CAS 

Google Scholar 
Hill, K. & Hurtado, A. M. Ache Life History: The Ecology and Demography of a Foraging People (Routledge, 2017).Book 

Google Scholar 
Kraft, T. S., Venkataraman, V. V., Tacey, I., Dominy, N. J. & Endicott, K. M. Foraging performance, prosociality, and kin presence do not predict lifetime reproductive success in Batek hunter-gatherers. Hum. Nat. 30, 71–97 (2019).Article 

Google Scholar 
Venkataraman, V. V., Kraft, T. S., Dominy, N. J. & Endicott, K. M. Hunter-gatherer residential mobility and the marginal value of rainforest patches. Proc. Natl. Acad. Sci. 114, 3097–3102 (2017).Article 
ADS 
CAS 

Google Scholar 
Woodburn, J. Egalitarian societies. Man 1, 431–451 (1982).Article 

Google Scholar 
Marlowe, F. The Hadza: Hunter-Gatherers of Tanzania Vol. 3 (University of California Press, 2010).
Google Scholar 
Fedurek, P. et al. Status does not predict stress: Women in an egalitarian hunter–gatherer society. Evol. Hum. Sci. 2, 1–10 (2020).
Google Scholar 
Kornienko, O. & Santos, C. E. The effects of friendship network popularity on depressive symptoms during early adolescence: Moderation by fear of negative evaluation and gender. J. Youth Adolesc. 43, 541–553 (2014).Article 

Google Scholar 
Smelser, N. J. & Baltes, P. B. International Encyclopedia of the Social & Behavioral Sciences Vol. 11 (Elsevier, 2001).
Google Scholar 
Kim, D. A., Benjamin, E. J., Fowler, J. H. & Christakis, N. A. Social connectedness is associated with fibrinogen level in a human social network. Proc. R. Soc. B Biol. Sci. 283, 20160958 (2016).Article 

Google Scholar 
Kindermann, T. A. & Gest, S. D. Assessment of the peer group: Identifying naturally occurring social networks and capturing their effects. In Handbook of peer interactions, relationships, and groups, 100–117 (2009).Kornienko, O., Clemans, K. H., Out, D. & Granger, D. A. Friendship network position and salivary cortisol levels. Soc. Neurosci. 8, 385–396 (2013).Article 

Google Scholar 
La Greca, A. M. & Lopez, N. Social anxiety among adolescents: Linkages with peer relations and friendships. J. Abnorm. Child Psychol. 26, 83–94 (1998).Article 

Google Scholar 
Okamoto, J. et al. Social network status and depression among adolescents: An examination of social network influences and depressive symptoms in a Chinese sample. Res. Hum. Dev. 8, 67–88 (2011).Article 

Google Scholar 
Ulset, V. S. et al. Are unpopular children more likely to get sick? Longitudinal links between popularity and infectious diseases in early childhood. PLoS ONE 14, e0222222 (2019).Article 
CAS 

Google Scholar 
Hawkes, K. Showing off: Tests of an hypothesis about men’s foraging goals. Ethol. Sociobiol. 12, 29–54 (1991).Article 

Google Scholar 
Smith, E. A. Why do good hunters have higher reproductive success?. Hum. Nat. 15, 343–364 (2004).Article 

Google Scholar 
Apicella, C. L., Feinberg, D. R. & Marlowe, F. W. Voice pitch predicts reproductive success in male hunter-gatherers. Biol. Lett. 3, 682–684 (2007).Article 
CAS 

Google Scholar 
Apicella, C. L. Upper-body strength predicts hunting reputation and reproductive success in Hadza hunter–gatherers. Evol. Hum. Behav. 35, 508–518 (2014).Article 

Google Scholar 
Smith, K. M., Olkhov, Y. M., Puts, D. A. & Apicella, C. L. Hadza men with lower voice pitch have a better hunting reputation. Evol. Psychol. 15, 1474704917740466 (2017).Article 

Google Scholar 
MacDougall-Shackleton, S. A., Bonier, F., Romero, L. M. & Moore, I. T. Glucocorticoids and “stress” are not synonymous. Integr. Organ. Biol. 1, 017 (2019).
Google Scholar 
Ouellette, S. J. et al. Hair cortisol concentrations in higher-and lower-stress mother–daughter dyads: A pilot study of associations and moderators. Dev. Psychobiol. 57, 519–534 (2015).Article 
CAS 

Google Scholar 
Stalder, T. et al. Stress-related and basic determinants of hair cortisol in humans: A meta-analysis. Psychoneuroendocrinology 77, 261–274 (2017).Article 
CAS 

Google Scholar 
Heimbürge, S., Kanitz, E. & Otten, W. The use of hair cortisol for the assessment of stress in animals. Gen. Comp. Endocrinol. 270, 10–17 (2019).Article 

Google Scholar 
Fedurek, P. et al. Relationship between proximity and physiological stress levels in hunter-gatherers: The Hadza. Horm. Behav. 147, 105294 (2023).Article 

Google Scholar 
Bowers, K. et al. Maternal distress and hair cortisol in pregnancy among women with elevated adverse childhood experiences. Psychoneuroendocrinology 95, 145–148 (2018).Article 
CAS 

Google Scholar 
Wells, S. et al. Associations of hair cortisol concentration with self-reported measures of stress and mental health-related factors in a pooled database of diverse community samples. Stress 17, 334–342 (2014).Article 
CAS 

Google Scholar 
Faresjö, T. et al. Elevated levels of cortisol in hair precede acute myocardial infarction. Sci. Rep. 10, 1–8 (2020).Article 

Google Scholar 
Fuchs, A. et al. Link between children’s hair cortisol and psychopathology or quality of life moderated by childhood adversity risk. Psychoneuroendocrinology 90, 52–60 (2018).Article 
CAS 

Google Scholar 
Staufenbiel, S. M., Penninx, B. W., Spijker, A. T., Elzinga, B. M. & van Rossum, E. F. Hair cortisol, stress exposure, and mental health in humans: A systematic review. Psychoneuroendocrinology 38, 1220–1235 (2013).Article 
CAS 

Google Scholar 
Davison, B., Singh, G. R. & McFarlane, J. Hair cortisol and cortisone as markers of stress in Indigenous and non-Indigenous young adults. Stress 22, 210–220 (2019).Article 
CAS 

Google Scholar 
Kim, E., Bolkan, C., Crespi, E. & Madigan, J. The relationship between hair cortisol, chronic stress, and well-being among older adults with dementia. Innov. Aging 3, S468 (2019).Article 

Google Scholar 
Woodburn, J. Egalitarian societies revisited. Proper. Equal. 1, 18–31 (2005).
Google Scholar 
Berbesque, J. C., Wood, B. M., Crittenden, A. N., Mabulla, A. & Marlowe, F. W. Eat first, share later: Hadza hunter–gatherer men consume more while foraging than in central places. Evol. Hum. Behav. 37, 281–286 (2016).Article 

Google Scholar 
Marlowe, F. W. & Berbesque, J. C. Tubers as fallback foods and their impact on Hadza hunter-gatherers. Am. J. Phys. Anthropol. 140, 751–758 (2009).Article 

Google Scholar 
Berbesque, J. C. & Marlowe, F. W. Sex differences in food preferences of Hadza hunter-gatherers. Evol. Psychol. 7, 147470490900700400 (2009).Article 

Google Scholar 
Hawkes, K., O’Connell, J. F. & Blurton Jones, N. G. Hunting income patterns among the Hadza: Big game, common goods, foraging goals and the evolution of the human diet. Philos. Trans. R. Soc. Lond. B 334, 243–251 (1991).Article 
ADS 
CAS 

Google Scholar 
Hawkes, K. Hunting and the evolution of egalitarian societies: Lessons from the Hadza. Hierarch. Action Cui Bono 27, 1–10 (2000).
Google Scholar 
Stibbard-Hawkes, D. N., Attenborough, R. D. & Marlowe, F. W. A noisy signal: To what extent are Hadza hunting reputations predictive of actual hunting skills?. Evol. Hum. Behav. 39, 639–651 (2018).Article 

Google Scholar 
Smith, K. M. & Apicella, C. Partner choice in human evolution: The role of character, hunting ability, and reciprocity in Hadza campmate selection. (2019).Smith, K. M. & Apicella, C. L. Hadza hunter-gatherers disagree on perceptions of moral character. Soc. Psychol. Pers. Sci. 11, 616–625 (2020).Article 

Google Scholar 
Gurven, M., Allen-Arave, W., Hill, K. & Hurtado, M. “It’s a wonderful life”: Signaling generosity among the Ache of Paraguay. Evol. Hum. Behav. 21, 263–282 (2000).Article 
CAS 

Google Scholar 
Aktipis, A. et al. Cooperation in an uncertain world: For the Maasai of East Africa, need-based transfers outperform account-keeping in volatile environments. Hum. Ecol. 44, 353–364 (2016).Article 

Google Scholar 
Cronk, L. et al. Managing risk through cooperation: Need-based transfers and risk pooling among the societies of the Human Generosity Project. in Global Perspectives on Long Term Community Resource Management, 41–75 (Springer, 2019).Cronk, L. & Aktipis, A. Design principles for risk-pooling systems. Nat. Hum. Behav. 1, 1–9 (2021).
Google Scholar 
Jones, N. B. Demography and Evolutionary Ecology of Hadza Hunter-Gatherers Vol. 71 (Cambridge University Press, 2016).
Google Scholar 
Crittenden, A. N. et al. Oral health in transition: The Hadza foragers of Tanzania. PLoS ONE 12, e0172197 (2017).Article 

Google Scholar 
Bennett, F. J., Barnicot, N. A., Woodburn, J. C., Pereira, M. S. & Henderson, B. E. Studies on viral, bacterial, rickettsial and treponemal diseases in the Hadza of Tanzania and a note on injuries. Hum. Biol. 1, 243–272 (1973).
Google Scholar 
Ibar, C. et al. Evaluation of stress, burnout and hair cortisol levels in health workers at a University Hospital during COVID-19 pandemic. Psychoneuroendocrinology 128, 105213 (2021).Article 
CAS 

Google Scholar 
Rajcani, J., Vytykacova, S., Solarikova, P. & Brezina, I. Stress and hair cortisol concentrations in nurses during the first wave of the COVID-19 pandemic. Psychoneuroendocrinology 129, 105245 (2021).Article 
CAS 

Google Scholar 
Hill, K. R., Wood, B. M., Baggio, J., Hurtado, A. M. & Boyd, R. T. Hunter-gatherer inter-band interaction rates: Implications for cumulative culture. PLoS ONE 9, e102806 (2014).Article 
ADS 

Google Scholar 
Bird, D. W., Bird, R. B., Codding, B. F. & Zeanah, D. W. Variability in the organization and size of hunter-gatherer groups: Foragers do not live in small-scale societies. J. Hum. Evol. 131, 96–108 (2019).Article 

Google Scholar 
Fedurek, P. et al. Social status does not predict in-camp integration among egalitarian hunter-gatherer men. Behav. Ecol. 33, 65–76 (2022).Article 

Google Scholar 
Ponzi, D., Muehlenbein, M. P., Geary, D. C. & Flinn, M. V. Cortisol, salivary alpha-amylase and children’s perceptions of their social networks. Soc. Neurosci. 11, 164–174 (2016).Article 

Google Scholar 
Marlowe, F. W. Mate preferences among Hadza hunter-gatherers. Hum. Nat. 15, 365–376 (2004).Article 

Google Scholar 
Von Rueden, C. R. & Jaeggi, A. V. Men’s status and reproductive success in 33 nonindustrial societies: Effects of subsistence, marriage system, and reproductive strategy. Proc. Natl. Acad. Sci. 113, 10824–10829 (2016).Article 

Google Scholar 
Townsend, C. Egalitarianism, Evolution Of (Wiley, 2018).Book 

Google Scholar 
Winterhalder, B. Diet choice, risk, and food sharing in a stochastic environment. J. Anthropol. Archaeol. 5, 369–392 (1986).Article 

Google Scholar 
Cornell, T. & Allen, T. B. War and Games Vol. 3 (Boydell Press, 2002).
Google Scholar 
Smáradóttir, S. Health and Wellbeing in the Arctic: The Critical Issues of Food Insecurity and Suicide Among Indigenous People.Finkler, H. W. Violence and the administration of justice: A focus on inuit communities in Northern Canada. BC Third World LJ 4, 137 (1983).
Google Scholar 
Bowles, S. Did warfare among ancestral hunter-gatherers affect the evolution of human social behaviors?. Science 324, 1293–1298 (2009).Article 
ADS 
CAS 

Google Scholar 
Fry, D. P. & Söderberg, P. Lethal aggression in mobile forager bands and implications for the origins of war. Science 341, 270–273 (2013).Article 
ADS 
CAS 

Google Scholar 
Gat, A. Proving communal warfare among hunter-gatherers: The quasi-rousseauan error. Evol. Anthropol. 24, 111–126 (2015).Article 

Google Scholar 
Kreyszig, E. Bernstein polynomials and numerical integration. Int. J. Numer. Meth. Eng. 14, 292–295 (1979).Article 
MATH 

Google Scholar 
Meyer, D. et al. Misc functions of the department of statistics, probability theory group (formerly: E1071). Package e1071. TU Wien (2015).R Development Core. A Language ans Environment for Statistical Computing. (R Found Stat Comput Vienna, 2018).Wennig, R. Potential problems with the interpretation of hair analysis results. Forensic Sci. Int. 107, 5–12 (2000).Article 
CAS 

Google Scholar 
Kumari, M., Shipley, M., Stafford, M. & Kivimaki, M. Association of diurnal patterns in salivary cortisol with all-cause and cardiovascular mortality: Findings from the Whitehall II study. J. Clin. Endocrinol. Metab. 96, 1478–1485 (2011).Article 
CAS 

Google Scholar 
Marmot, M. G. & Sapolsky, R. Of baboons and men: Social circumstances, biology, and the social gradient in health. in Sociality, hierarchy, health: Comparative biodemography: Papers from a workshop (2014).Hoffman, M. C., Karban, L. V., Benitez, P., Goodteacher, A. & Laudenslager, M. L. Chemical processing and shampooing impact cortisol measured in human hair. Clin. Investig. Med. 37, E252 (2014).Article 
CAS 

Google Scholar 
Sauvé, B., Koren, G., Walsh, G., Tokmakejian, S. & Van Uum, S. H. Measurement of cortisol in human hair as a biomarker of systemic exposure. Clin. Investig. Med. 30, E183–E191 (2007).Article 

Google Scholar 
Slominski, R., Rovnaghi, C. R. & Anand, K. J. Methodological considerations for hair cortisol measurements in children. Ther. Drug Monit. 37, 812 (2015).Article 
CAS 

Google Scholar 
Xiang, L., Sunesara, I., Rehm, K. E. & Marshall, G. D. Jr. A modified and cost-effective method for hair cortisol analysis. Biomarkers 21, 200–203 (2016).Article 
CAS 

Google Scholar 
Tukey, J. Exploratory Data Analysis (Addison-Wesley, 1977).MATH 

Google Scholar 
Mangiafico, S. & Mangiafico, M. S. Package ‘rcompanion’. Cran Repos 1–71 (2017).Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. lmerTest package: Tests in linear mixed effects models. J. Stat. Softw. 82, 1–26 (2017).Article 

Google Scholar 
Bates, D. M. lme4: Mixed-Effects Modeling with R. (2010).Lüdecke, D. ggeffects: Tidy data frames of marginal effects from regression models. J. Stat. Softw. 3(26), 772. https://doi.org/10.21105/joss.00772 (2018).Article 

Google Scholar 
Nowok, B., Raab, G. M. & Dibben, C. synthpop: Bespoke creation of synthetic data in R. J. Stat. Softw. 74, 1–26 (2016).Article 

Google Scholar  More

At December’s United Nations Convention on Biological Diversity summit (COP15), an insidious threat emerged to national parks — even as scientists argued for expanding protected areas. The World Travel & Tourism Council wants commercial tourism to be allowed to build developments in national parks globally, without obligation to help finance park conservation (see go.nature.com/3x2fsi9). This would undermine existing private tourism developments that do support conservation.
Competing Interests
The authors declare no competing interests. More

The University of California, Los Angeles, suspended ecologist Priyanga Amarasekare without salary or benefits for one year, and will cut her salary by 20% for two more years.Credit: Al Seib/Los Angeles Times via Getty

In April of last year, the Ecological Society of America awarded Priyanga Amarasekare one of the highest honours in the field of ecology: the Robert H. MacArthur Award. A little over two months later, the University of California, Los Angeles (UCLA), placed Amarasekare on a one-year suspension without pay or benefits, and forbid her from accessing her laboratory, maintaining her insect colonies, managing her grants or contacting students. Now scientists from around the world, who call Amarasekare a “highly distinguished ecologist”, “a committed teacher and outstanding mentor” and a “tireless advocate for under-represented groups”, are calling for her reinstatement.
Scientists question Max Planck Society’s treatment of women leaders
The precise allegations that led to her suspension are unknown. UCLA has declined to release them, and barred Amarasekare from discussing the matter publicly. But long-standing tensions between Amarasekare and the university are no secret. A native of Sri Lanka and one of two women of colour who have tenure in the ecology and evolution department, she has previously accused the university of discrimination for repeatedly denying her promotions that were granted to colleagues. Former students and faculty members who are familiar with the situation think that Amarasekare’s suspension was retaliation for speaking out.Some 315 scientists raised concerns about her suspension in a petition that was delivered to the university on 23 January, arguing that Amarasekare “has long been denied significant advancement within her department, out of keeping with her contributions to the field”. Moreover, the sanctions levied against Amarasekare — including the one-year suspension and 20% salary reduction for an additional two years — represent “the kind of punishment normally applied only to the most egregious wrongdoings”, including scientific misconduct and sexual harassment violations, the petitioners write.In the absence of compelling evidence to the contrary, the scientists ask that UCLA rescind the disciplinary actions and fully compensate Amarasekare.Officials with UCLA say that the university “supports freedom of expression and does not condone retaliation of any sort”. They declined to discuss the accusations against or in support of Amarasekare, saying the university is “bound to respect the privacy of the numerous individuals involved in this matter”. Amarasekare also declined to comment.A confusing decisionColleagues told Nature that Amarasekare is the rare ecologist whose research spans the theoretical, computational and experimental realms. One project in her laboratory that touches on all of these areas focuses on the impact of climate change on insect communities. “She’s really several years ahead of everybody else,” says Andy Dobson, an ecologist at Princeton University in New Jersey who led the petition. Dobson has written letters to support Amarasekare’s various applications for promotion at UCLA and says he has been baffled by the university’s decisions. “She complained, and most of what’s happened seems to be a reaction against that,” he says.
Legal win for US scientist bolsters others caught in China crackdown
Nature spoke to several former students and faculty members who defended Amarasekare in administrative hearings in September 2021. Although none knew the specific details of the charges against her, they all thought she had been targeted for speaking out against what she saw as discrimination within the department. In particular, they said Amarasekare vented about her own experience at UCLA on a departmental e-mail listserve created to discuss issues of racism and discrimination in the aftermath of the killing of George Floyd, whose death in May 2020 sparked national protests.“That’s why she got into trouble. She ended up criticizing pretty much the entire department — with good reason,” says Marcel Vaz, an ecologist at Wilkes University in Wilkes-Barre, Pennsylvania, who was a graduate student in the department at the time. He and other students came forward to support her. “We demanded some explanation,” Vaz says, “but we never got any feedback.”Peter Kareiva, a former UCLA faculty member who spoke on Amarasekare’s behalf during the administrative proceedings, calls her a brilliant scientist as well as a terrific teacher and student mentor. Kareiva witnessed Amarasekare raise uncomfortable issues and challenge internal policies in faculty meetings. He says she might have made mistakes in terms of “facilitating harmony” among fellow faculty members, but that her goal was always to improve the department.
How a scandal in spider biology upended researchers’ lives
“I am still incredulous by the punishment levied,” says Kareiva, who now serves as president of the Aquarium of the Pacific in Long Beach, California.It is unclear what happens next, but scientists and former students and faculty members contacted by Nature are concerned about the impact on Amarasekare’s current students, the disruption of federally funded research and the potentially irretrievable loss of time-sensitive experiments that could provide insights regarding the ecological impacts of climate change.As the recipient of the MacArthur award, Amarasekare is expected to discuss this research when she delivers her keynote address at the Ecological Society of America’s annual meeting in Portland, Oregon, in August. More

There is a growing impetus to leverage our fundamental understanding of microbial community assembly towards applied problems. With microbes contributing to diverse physiological, biogeochemical, and agricultural processes, the potential to control and optimise microbial communities holds promise for interventions ranging from industrial and environmental remediation to human medicine and biofuel production [1, 2]. Realising this goal is contingent on high fidelity between theory, experiments, and the natural dynamics of target systems.Theoretical and experimental research in microbial community optimisation has largely proceeded along two parallel paths. Theoretical approaches leverage mathematical models and metabolic networks to predict which species combinations are stable and how they can optimise a given function (e.g., maximum biomass, waste degradation or host health) [3,4,5,6,7]. Experimental studies often take a combinatorial approach, iteratively assembling different species combinations in vitro and evaluating their stability and functional attributes [8,9,10,11]. Both theory and experiments are valuable but they are also susceptible to their own modus operandi that may limit their correspondence and their translation to real-world systems. On the one hand, theoretical approaches typically adopt the analytical tractability of steady state dynamics, where microbial consumers and the resources on which they depend are assumed to establish a stable equilibrium. On the other hand, experimental approaches almost exclusively embrace the high-throughput efficiency of serial-batch culture, where consumers and resources are made to fluctuate over several orders of magnitude with each serial passage. This raises an important question: should we expect unity in the composition of optimised communities emerging under continuous resource supply (e.g., chemostat) versus the discrete pulsed resource supply of, for example, serial-batch culture?To explore how microbial community composition varies under contrasting resource supply dynamics, we performed simulations of a classical resource-competition model:$$frac{{dN_i}}{{dt}} = N_ileft( {mathop {sum}limits_{j = 1}^n {mu _{ij}left( {R_j} right) – m} } right)$$
(1)
$$frac{{dR_j}}{{dt}} = {Psi}_jleft( {R_j} right) – mathop {sum }limits_{i = 1}^n Q_{ij}mu _{ij}left( {R_j} right)N_i,$$
(2)
where Ni is the population density of consumer i, Rj is the concentration of resource j, μij(Rj) is the per capita functional response of consumer i, m is the per capita mortality rate due to dilution, Ψj(Rj) is the resource supply function, and Qij is the resource quota of consumer i on resource j (amount of resource per unit consumer). The consumer functional response is given by the Monod function, (mu _{ij}(R_j) = mu _{max_{ij}}frac{{R_j}}{{K_{s_{ij}} + R_j}}) , where (mu _{max_{ij}}) is the maximum growth rate and (K_{s_{ij}}) is the half saturation constant for consumer i on resource j.To set up the simulations, we randomly sampled the parameters of the Monod growth functions, (μmax and Ks) for five species competing for five substitutable resources (essential resources are treated separately in the supplementary information, with similar findings). In one set of parametrisations (n = 100 unique competitor combinations) we used both random μmax and Ks, and in another set (n = 100) we imposed a trade-off in maximum growth rate and substrate affinity (( {frac{{mu _{max}}}{{K_s}}} )) (Fig. 1a). The rationale for imposing a trade-off is that metabolic theory predicts that organisms that invest energy into a high maximum growth rate will have lower substrate affinities and vice versa [12, 13]. To ensure reasonable growth rates relative to the time-scale of resource pulsing, we sampled μmax such that minimum doubling times spanned from 21 to 52 min (when all resources are non-limiting). For each of the random competitor combinations, we simulated resources under continuous or pulsed resource supply with resource replenishment every 1/2, 1, 2, 4, 12, or 24 h. Under pulsed resource supply, Ψj(Rj) and m are removed from Eq. (1) and (2) and replaced by discontinuous resource pulsing and cell transfer at fixed intervals. The total resource flux (and mortality) was held constant under all frequencies of resource supply i.e., less frequent replenishment corresponds to larger resource pulses (see Supplementary Information for full model/simulation specifications).Fig. 1: Quantifying compositional overlap between communities assembled under continuous vs. pulsed resource supply.a Per capita growth responses (Monod functions) from a single iteration of the model assuming a trade-off between maximum growth rate and resource affinity (colours correspond to individual consumers). b Time series of consumers in a under different resource supply regimes. Numbers above individual panels reflect pulsing intervals in hours. The amplitude of population fluctuations increases with longer intervals between pulses, with distinct phases of growth, saturation, and instantaneous mortality visible at a finer temporal resolution (Fig. S10). c Example measure of compositional overlap (Jaccard similarity index) between communities assembled under continuous resource supply (far left panel in b) vs. pulsing every two hours (centre panel in b).Full size imageAfter allowing the competitors to reach a steady state (time-averaged over 24 h under pulsed treatments), we quantified the correspondence between the continuous supply treatment and the pulsed treatments using the Jaccard similarity index, (Jleft( {A,B} right) = frac{{left| {A cap B} right|}}{{left| {A cup B} right|}}) (0 ≤ J(A,B) ≤ 1), where the numerator gives the number of species (max = 5) that persist under continuous (A) and pulsed (B) resource supply, and the denominator gives the number of species (max = 5) that persist under continuous or pulsed resource supply (Fig. 1b, c).Under both sets of simulations (with and without enforcing a trade-off between maximal growth rate and resource affinity), we observe that the similarity in final community composition between continuous and pulsed resource supply decays with increasingly large intervals between resource replenishment (Fig. 2a). When no trade-off is imposed between maximum growth rate and resource affinity (orange line in Fig. 2a) the mean compositional similarity is only 0.68 when resources are pulsed every 2 h and down to 0.41 when resources are pulsed every 24 h (typical of serial-batch culture). The rate of decay in the Jaccard index is more severe when a trade-off is imposed between maximum growth rate and substrate affinity, to the extent that once pulsing intervals reach four hours there is almost zero overlap in community composition (blue line in Fig. 2a).Fig. 2: Impact of resource supply regime on community composition and abundance weighted mean trait values.a Compositional overlap (Jaccard similarity) between communities under continuous versus pulsed resource supply. Orange lines, points and circles denote model parametrisations with random sampling of both μmax and Ks; blue lines, points and circles denote model parametrisations with a trade-off imposed between μmax and resource affinity (( {frac{{mu _{max}}}{{K_s}}} )). Simulation parameters provided in the Supplementary Information. b Mean trait values for affinity and μmax averaged for each consumer across the five resources and weighted by their final abundance at the end of a simulation (cont. = continuous). In both a and b, small points (jittered for clarity) give the result of an individual simulation; large circles indicate the corresponding mean.Full size imageEcological theory provides an intuitive explanation for these observations. When resources are more continuously supplied, the better competitor is the one that can sustain a positive growth rate at the lowest concentrations of a limiting resource (i.e., has a higher resource affinity or lower R* in the language of resource competition theory [14]). In contrast, under increasingly pulsed resource supply, the better competitor is the one that can grow rapidly at higher resource concentrations. Having a high resource affinity (low R*) is of little benefit if resource concentrations fluctuate over large amplitudes because it only confers an ephemeral competitive advantage in the brief period before the resource is completely depleted (ahead of the next resource pulse). Instead, a high maximum growth rate is optimal because it allows the consumer to grow rapidly and quickly deplete a shared limiting resource. This high maximum growth strategy is, however, sub-optimal under continuous resource supply because a low R* strategist can draw the resource down and hold it at a concentration at which the maximum growth strategist is unable to maintain a positive growth rate.Looking at the mean trait values for resource affinity and μmax weighted by each consumer’s final abundance, it is indeed apparent that consumers with a higher affinity (averaged across the five resources) are favoured under continuous resource supply, while consumers with high maximum growth rates are favoured under pulsing intervals of increasing length (Fig. 2b). Enforcing this trade-off, therefore, leads to the rapid decline in compositional similarity we observe under resource pulsing. Notably, it also leads to a richness peak at intermediate pulsing intervals, where these alternative strategies have a higher probability of coexisting [15] (Fig. S1). At the same time, we still observe a decline in compositional similarity when μmax and Ks are randomly sampled independently of each other simply because the trade-off between maximum growth and resource affinity will emerge occasionally by chance. Two experimental tests of microbial community composition under continuous versus pulsed resource supply are consistent with these observations [16, 17].To evaluate the sensitivity of these observations to different assumptions, we ran additional simulations under various alternative model parameterisations and formulations. In brief, comparable trends to those described above are observed when: i) maximum growth rates are faster or slower than those presented in the main text (Figs. S2, S3); ii) all resources are assumed to be essential to growth (following Liebig’s law of the minimum) (Fig. S4); iii) a weaker trade-off is imposed between maximum growth and affinity (Figs. S5, S6); or iv) mortality is continuous rather than intermittent (Figs. S7, S8). We also investigated the relationship between observed compositional overlap and the dynamical stability under continuous resource supply, anticipating that more stable communities would tend to be more resistant to compositional shifts under resource pulsing. The reality appears more nuanced, namely that weaker dynamical stability at the limit of constant resource supply is associated with higher variance in compositional overlap under continuous vs. pulsed conditions (Fig. S9). In other words, systems with weaker stability are less predictable. A wide range of other microbial traits and trade-offs may interact unpredictably with the relationship between resource supply and community composition. The potential modulating role of system instabilities generated by cross-feeding interactions, non-convex trade-off functions, and the evolution of specialist versus generalist strategies present several especially valuable lines of enquiry [18,19,20].Although these observations are germane to any consumer-resource system, our emphasis here is on the emerging field of microbial community optimisation, where the practical implications are especially timely and important; namely, the resource supply regime must be tailored to the community being optimised. For example, wastewater treatment might be more appropriately modelled under continuous resource supply [21], whereas fermented food and beverage production may be more closely allied to the pulsed resource dynamics observed in batch culture [22]. Resource supply might also be manipulated to favourably modify the competitive hierarchy in an existing community (e.g., by regulating the rate of nutrient supply to the gut through meal timing). Indeed, there is emerging evidence that feeding frequency can drive significant changes in gut microbiota composition [23, 24]. Thus, resource supply dynamics should be considered both a constraint in the design of novel microbial communities and as a tuning mechanism for the optimisation of preexisting communities like those found in the human gut. More