Global ecological predictors of the soil priming effect

Identifying the global drivers of soil priming is essential to understanding C cycling in terrestrial ecosystems. We conducted a survey of soils across 86 globally-distributed locations, spanning a wide range of climates, biotic communities, and soil conditions, and evaluated the apparent soil priming effect using 13C-glucose labeling. Here we show that the magnitude of the positive apparent priming effect (increase in CO2 release through accelerated microbial biomass turnover) was negatively associated with SOC content and microbial respiration rates. Our statistical modeling suggests that apparent priming effects tend to be negative in more mesic sites associated with higher SOC contents. In contrast, a single-input of labile C causes positive apparent priming effects in more arid locations with low SOC contents. Our results provide solid evidence that SOC content plays a critical role in regulating apparent priming effects, with important implications for the improvement of C cycling models under global change scenarios.

to distinguish, however, apparent priming tends to occur shortly after adding readily 102 availably substrates (first days and weeks), while real priming takes longer 7,9 . 103 Overall, soil priming is a complex phenomenon that is regulated by multiple mechanisms, 104 involving abiotic and biotic factors (including, but not limited to, nutrient availability, 105 catabolism of different organic matter pools) 6,7,10,11 . Soil priming has been postulated to 106 be a major determinant of the capacity of soils to function as sources or sinks of 107 atmospheric CO2 12 . Consequently, inputs of fresh organic matter to the soil can cause an 108 accelerated microbial biomass turnover (apparent priming). Alternatively, a negative 109 priming due to reduced SOC mineralization or attenuated microbial biomass turnover can 110 occur when labile C is added to soil 6 . Recent modelling developments suggest that soil 111 priming is a strong candidate for inclusion in models to predict global distributions of C 112 because of the important role of priming in determining the exchange of C between soils 113 and the atmosphere 5,13 . However, we lack a unifying ecological context and an integrative 114 approach to understanding soil priming effects globally, which would allow us to 115 determine how the direction of the priming effect varies across different ecosystems and 116 why this variation exists.

117
A growing body of literature has identified nutrient availability, climate, soil type, 118 or plant and microbial attributes 14-18 as potentially important drivers of priming 7 . For 119 example, soil texture has been demonstrated to be an important factor controlling the soil 120 priming effect, and plants, through the amount and composition of rhizodeposits, also 121 play a key role in mediating priming effects 4 . Furthermore, climatic factors such as mean 122 annual temperature are related to soil priming effects 11 . However, in spite of the elevated 123 amount of C within microbial biomass 2 , a comprehensive understanding of the drivers of 124 the apparent priming effect across major biomes and gradients at the global scale is 125 lacking. This knowledge will shed light on how environmental factors regulate the 126 microbial biomass turnover and its contribution to CO2 fluxes under global change 127 scenarios 19,20 . Moreover, a better understanding of the ecological predictors of priming 128 will improve our ability to predict how CO2 fluxes might shift in response to human and 129 global change factors that influence the quality and quantity of fresh C inputs to soil and 130 soil microbial responses 12 , such as afforestation 21 , changes in plant C allocation to soil 131 due to the elevated levels of atmospheric CO2 12 , the addition of organic amendments to 132 soil 22 , nitrogen (N) deposition 23 , warming 24 and changes in land use 25 .

133
Herein, we conducted a soil survey of 86 locations across six continents, spanning 134 multiple climates (tropical, temperate, polar, arid and continental) and ecosystem types 135 (e.g., forest, grasslands and croplands; SI Appendix, Fig. S1). We aimed to identify the 136 major global ecological predictors of the apparent soil priming effect. Apparent priming 137 was determined using a soil incubation of 16 days coupled with 13 C-labeled glucose.

138
Ecological predictors included wide environmental gradients of mean annual 139 temperature, aridity, vegetation types, plant cover, soil chemical and physical properties,  Given that SOC is widely correlated with microbial biomass 26 , we hypothesized 146 that the effect size and the direction of the apparent priming effect is regulated by SOC 147 content, which, in turn, is modulated by the environmental and ecological context of each 148 soil 27,28 . Thus, we hypothesized that soils with lower SOC content, including soils from 149 arid sites with sparse plant cover where microbial biomass is strongly limited by C 29 , will 150 be more responsive to the inputs of labile C, ultimately stimulating microbial turnover 151 and the resulting apparent priming-mediated CO2 release (positive priming) 7 . Conversely, 152 we expected that the apparent priming effect would be negative in soils from mesic 153 regions with greater plant cover and higher litter and root inputs to soil where microbial 154 biomass and soil microbial respiration are less limited by the availability of C.

156
Considering that incubation with 13 C-glucose lasted 16 days, our results mainly reflect 157 the patterns of the apparent priming effect 7,8 . It corresponds to changes in CO2 release as 158 a consequence of microbial biomass turnover shortly after adding fresh-available 159 substrates 7,8 . Our findings indicate that the apparent soil priming effect is a globally 160 ubiquitous phenomenon and provide new insight into its major ecological predictors, in 161 spite the extreme heterogeneity of soils and incubation limitations, as described below. 162 We found contrasting responses of apparent priming associated with different globally 163 distributed ecosystem types. In some soils, a single-pulse of labile C accelerated the 164 turnover of microbial biomass (positive apparent priming). Conversely, the addition of 165 labile C can lead to reductions in microbial turnover in other soils (negative apparent 166 priming; Fig. 1A-B). For instance, positive apparent priming effects were associated with 167 shrub-and forb-dominated ecosystems, croplands and cold forests (Fig. 1A). In some 168 ecosystems (i.e. croplands, forblands and shrublands), the release of CO2 due to positive 169 apparent priming represented more than 20% of the basal microbial respiration rate (Figs.   indicating that it represents a causal scenario consistent with the data. Strikingly, soil 234 microbial biomass (estimated using substrate-induced respiration, SIR), which has been 235 postulated to be a major ecological predictor of priming effects 7 , was not a significant 236 predictor of apparent priming in the wide variety of soils tested here (Fig. 2). In other 237 words, our results suggest that the initial content of SOC ultimately regulates the apparent 238 soil priming effect. Soils with greater C content (therefore, less limited by C) are more 239 likely to exhibit negative or minimal apparent priming. Importantly, the negative  Table S1). However, in our SEM, only the relative abundance of 252 Basidiomycota had significant direct effects on the apparent priming effect after 253 considering multiple environmental factors simultaneously (Fig. 2-4). Basidiomycota are  Similarly, physical factors such as soil texture, which has also been proposed as a factor 292 regulating soil priming effects 45 , was not a significant factor across the broad range of 293 soils tested here. Other soil properties such as pH, available soil P content and salinity did 294 not show any direct effect on the apparent soil priming, but these factors indirectly 295 affected soil microbes (Fig. 2), and salinity had a total negative significant effect on 296 priming 46,47 .

297
Our SEM provides an ecological context for apparent priming effects across a 298 wide range of soils. Soils with greater plant cover located in more mesic ecosystems had 299 higher soil C contents and basal microbial respiration rates that were associated with a 300 greater likelihood of negative apparent priming effects (Figs. 2-4). A priori, the microbial 301 community in these soils is expected to be adapted to greater C inputs from plants. In 302 these communities, inputs of fresh substrate could be used by microbes to support growth, 303 assimilating C in microbial biomass and thus limiting the release of CO2 to the 304 atmosphere, explaining the negative apparent priming effect in these soils. Conversely, 305 our results suggest that positive apparent priming is likely greater in soils under drier 306 climates (i.e. shrublands) and with land use (e.g., croplands) with low SOC contents 28,31 307 (Figs 1C and D, 2, and 3). A previous study using an herbaceous savannah soil, also 308 revealed that positive priming effects were more likely to be observed in nutrient-limited 309 soils 16 . The microbial community of these soils is likely not adapted to the input of fresh-  P was determined from bicarbonate extracts using colorimetric analyses as explained in 365 Olsen and Sommers (1982) 55 . SOC content ranged between 0.1 and 38%, available P 366 between 0.5 to 72 mg P kg -1 soil, pH between 3.8 to 9.1 and the % of clay + silt varied 367 between 0.3 and 86%, respectively.  The reason is that our global survey includes soils with wide ranges in SOC and microbial 382 biomass, but also in many other factors that can regulate the soil priming effect (i.e. clay  Incubations were maintained for more than two weeks because previous studies have 403 revealed that the major part of CO2 release from soil tends to occur a few days or weeks 404 after substrate addition 7, . Longer incubation time was not used as we want to avoid CO2 405 saturation in the vials of C-rich soils. We are aware that our incubation conditions were 406 outside the range for the mean temperature and water content of soils and, consequently,

Statistical analyses
448 PERMANOVA 449 We first tested for significant differences in priming effect across major ecosystem types 450 using one-way non-parametric Permutational ANalysis Of Variance (PERMANOVA   Table S1); thus only these taxa were included in our SEM. Of these taxa,  The complete dataset associated with this paper has been deposited in figshare: