The loss of plant diversity globally is a challenge facing humanity (Rockstrom et al., 2009). Yet, how multiple global change drivers, specifically drought and diversity loss, interact to effect productivity remails poorly understood. The implications of plant diversity loss on aboveground plant productivity are well established (Tilman et al., 1997; Hector et al., 1999; Reich et al., 2012; Tilman et al., 2014; Wagg et al., 2022), and generally exceed drought effects (Rockstrom et al., 2009; Tilman et al., 2014). For example, a reduction in plant species from 16 to 1 can result in a reduction in biomass that is four times greater than effects observed with drought. Yet, these patterns describe aboveground effects, representing only a portion of total plant production. A community’s resistance to and ability to recover from altered precipitation is thought to depend upon plant species richness and composition (van Ruijven and Berendse, 2010; Isbell et al., 2015; Skelton et al., 2015), but again these studies focus heavily on aboveground production. Thus, understanding when aboveground patterns are reflective of those occurring belowground is necessary for understanding the implications of interacting global change drivers, i.e. plant diversity loss and drought.

Diversity effects on root biomass likely depend on the identity of the plants that comprise communities, i.e. defining which plant functional groups are present. Plants utilize different rooting strategies often corresponding with their functional groups and the traits they exhibit to acquire resources. For instance, some plants root deeply, while others invest heavily in shallow root systems (Mueller et al., 2013; Carmona et al., 2021). Some plants utilize their rooting systems to form symbiotic relationships with microbes, exchanging C for access to limiting resources, e.g. phosphorus or nitrogen (Reynolds et al., 2003; Vitousek et al., 2013), while others stimulate the activity of free living rhizosphere microbes or directly scavenge soil resources via root exudation (Phillips et al., 2011; de Vries et al., 2019). Experiments exploring the role of plant species richness in determining belowground biomass suggest plant functional group diversity governs root production increases observed with richness (Gastine et al., 2003; Mueller et al., 2013; Prommer et al., 2020; Furey and Tilman, 2021). For example, synergistic effects on belowground biomass have been observed when communities include C₄ grasses and legumes (Fornara et al., 2009; Mueller et al., 2013; Yang et al., 2019). Grasses with C₄ photosynthetic strategies often invest heavily in fine roots (Carmona et al., 2021), while legumes are capable of increasing access to nitrogen (N) (Vitousek et al., 2013) by forming symbiotic relationships with rhizobia (Zahran, 2001). However, disentangling the effects of richness and community composition on belowground biomass has been challenging (Gastine et al., 2003; Fornara et al., 2009; Mueller et al., 2013; Yang et al., 2019; Prommer et al., 2020; Furey and Tilman, 2021), as the longest running biodiversity experiments do not independently manipulate both the number of species and plant composition.

Furthermore, how belowground biomass responds to altered precipitation is likely determined by plant community composition, as plant species have different responses to drought and overwatering. Across regional and global scales, belowground production tends to increase with precipitation (Wu et al., 2011; Kang et al., 2013; Du et al., 2020). However, these patterns are inconsistent, and we only sometimes observe in manipulative experiments increased root biomass with precipitation (Xu et al., 2012; Evans and Burke, 2013). In many cases belowground biomass does not change (Sindhøj et al., 2000; Byrne et al., 2013; Flanagan et al., 2013; Denton et al., 2017), or is higher in communities experiencing drought (Wilcox et al., 2016; Slette et al., 2022). Environmental conditions, e.g. depth to the water table (Fan et al., 2017), can alter the rooting strategies of plants. Additionally, plant functional group and community composition may likewise determine these outcomes. For example, the presence of deep-rooted plants can increase soil moisture higher in soil profiles, alleviating stressful conditions for neighbors generating increased belowground production with diversity in systems experiencing drought (Horton and Hart, 1998; Pang et al., 2013). This represents an unexplored potential mediator of plant responses to diversity. Manipulative experiments evaluating potential interactive effects of drought, plant diversity, and community composition on belowground biomass are rare. While few studies have explored the interaction between drought, plant community composition, and diversity on root biomass, it has been observed that less abundant species can generate increased root biomass with drought (Slette et al., 2022). This suggests that community composition could govern belowground biomass responses to diversity with altered precipitation.

While diversity can generate increased belowground biomass, it remains unclear how changes in environmental conditions, such as altered precipitation regimes, and plant community composition effect these outcomes. We sought to determine whether aboveground production benefits from plant diversity are similarly observed in belowground biomass. Specifically, we asked whether plant community composition and precipitation interact with each other and plant diversity to effect root standing crops three and five years after plant establishment. To address this, we established a grassland rainfall and diversity manipulation experiment, that independently alters both plant richness and community composition, by selecting species for mixtures from the same or multiple families. This experimental design can uniquely disentangle the role plant diversity and functional group may independently play in generating benefits from diversity. We might expect aboveground production benefits from richness will be replicated by root biomass belowground, and those patterns will be greatest in families including both legumes and C₄ grasses (Fornara et al., 2009; Mueller et al., 2013; Yang et al., 2019). Furthermore, we might expect root biomass will increase with precipitation, as has been observed in meta-analyses across regional and global scales (Wu et al., 2011; Kang et al., 2013; Du et al., 2020), but whether belowground root biomass responses to biodiversity increase with precipitation is unknown.

Root, aboveground, and combined aboveground and belowground biomass increased with plant diversity overall for both years ( Figure 2 ; Supplementary Figure S4 ). Plant richness significantly increased aboveground production ( Figure 2B ; Table 1 ), and the magnitude of the relationship significantly strengthened with time ( Figure 3 ; Table 1 ). Six species mixtures generated 255% more aboveground biomass than monocultures ( Figure 2B ). Similarly, we observed a significant positive relationship between plant species richness and root standing crop ( Figure 2C ; Table 2 ). Root standing crop increased 132% when species richness increased from one to six, and 232% from one to five species. Notably, the highest diversity level of single-family grass communities has five species and are not represented in six species richness level. Root, aboveground, and combined above and belowground biomass increased over time ( Table 1 ). While an interaction between richness and year was present in two of the candidate models explaining root standing crop ( Table 2 ) and in five out of the seven explaining combined above and belowground plant biomass ( Supplementary Table S5 ), the interaction was not significant in either model ( Table 2 ).

Plant family identity was a strong determinant of root standing crops and moderated the magnitude of the relationship between root standing crop and richness ( Figure 4B ; Supplementary Table S5 ). In three of the five candidate models explaining root standing crop, a significant main effect of plant family designations (CC4) was present ( Table 2 ). We observed higher mean root standing crop in grass (monoculture mean 463.9 ± 410.3g m⁻²), lower in legume (monoculture mean 212.2 ± 155.1g m⁻²), and mid-level in communities comprised of all asters (monoculture mean 332.9 ± 255.6g m⁻²), and across richness levels average differences between families held and significantly differed ( Table 2 ; Supplementary Figure S5 ; Supplementary Table S5 ). In the other two models, a significant interaction between plant community composition and richness was present ( Table 2 ). Here we observed a strong positive relationship between root standing crop and richness in communities composed of all grasses, a weak positive relationship for communities composed from multiple families and all asters and a weakly negative relationship in communities composed of all legumes ( Figure 4B ; Table 1 ). We observed a 173%, 136%, and 167%, increase in root standing crops from one to six species for asters, legumes, and grass communities respectively. Though the slope was steepest in grass communities, the percent increase in standing crop was highest in aster communities because root standing crops in grass monocultures were on average high compared aster and legume monocultures. We observed the lowest percent increase from one to six species in multiple family communities, 129%, which were compared to the average root standing crop of all monocultures ( Supplementary Table S6 ).

In contrast with root standing crops, plant family diversity (single vs multiple), but not plant family identity (grasses vs legumes vs asters), moderates aboveground production ( Supplementary Tables S5, S7, S8 ). We observed a significant main effect of plant community composition with two levels (CC2) on aboveground biomass ( Table 1 ). The relationship between plant richness and aboveground biomass was steeper in single family communities in comparison to multiple family communities ( Figure 4A ; Supplementary Table S7 see the CC2 _SINGLE:S). Plant community composition was not in the most probable model predicting combined above and belowground production ( Supplementary Tables S5, S9, S10 ). We observed a weak relationship between plant community composition with two levels (CC2) and richness in the candidate models predicting combined above and belowground biomass replicating the pattern observed in aboveground biomass ( Supplementary Table S9 ; Supplementary Figure S6 ).

The effect of precipitation on root standing crop changed with time, while the precipitation treatment did not predict changes in aboveground production ( Table 1 ). Root standing crop significantly increased overtime from 2020 to 2022 by 1.4-fold, three and five years after initial establishment ( Figure 5 ; Table 2 ; Supplementary Table S11 ). In all candidate models, we observed a significant interaction between year and precipitation treatment ( Table 2 ). In 2020 root biomass was significantly higher in 50% precipitation treatments generating 179% greater root standing crops than high water treatments ( Supplementary Table S11 ). In 2022, we observed no difference between mean root standing crop in high and low precipitation ( Figure 5 ). While rain was in models predicting aboveground production, it did not significantly affect aboveground biomass ( Supplementary Tables S7, S8 ). The interaction between year and precipitation on combined above and belowground biomass replicated patterns observed in belowground root biomass ( Supplementary Figure S8 ; Supplementary Tables S9, S10 ).

While it is well-established that plant diversity generates increased aboveground productivity (Tilman et al., 1997; Hector et al., 1999; Reich et al., 2012; Wagg et al., 2022), whether these patterns are replicated belowground is less certain. Moreover, whether plant diversity increases belowground biomass is dependent on plant community composition and precipitation, both of which are anticipated to change in the Anthropocene (Knapp et al., 2002; Nippert et al., 2006; Zeppel et al., 2014; Lee et al., 2021), and has broader consequences under future climate change scenarios. This critical gap needs to be resolved to understand how interactive effects of climate change and biodiversity loss, both significant global change drivers themselves (Rockstrom et al., 2009; Tilman et al., 2014), will affect total plant production with cascading effects on biodiversity and soil ecosystem function (Guerra et al., 2020). In a rainfall and grassland diversity manipulation experiment we observed strong evidence of plant diversity benefits on root standing crops ( Figure 2C ). Diversity benefits for root standing crops were greatest in communities composed of grasses and weakest in communities composed of all legumes ( Figure 4B ). We did not, however, observe increases in belowground standing crops when plots included both legumes and C₄ grasses in diverse communities, and instead our diverse multiple-family mixtures underperformed mixtures with only grasses. Consistent with our expectations, plant community composition had a strong influence on belowground standing crops ( Figure 4B ) and root biomass increased in the plots over time ( Figure 5 ). In year three, 2020, we observed higher root standing crops in low compared to high water treatments, while those differences dissipated two years later as root biomass increased in high precipitation treatments to levels comparable to 50% precipitation treatments. This contrasted with aboveground patterns, and our expectations, that increased precipitation generated greater yields.

Effects of precipitation on belowground root biomass changed over time. Generally, across regional to global scales root biomass increases with precipitation (Wu et al., 2011; Kang et al., 2013; Du et al., 2020). While we expected our findings to reflect this, as has been confirmed in some rainfall manipulation experiments (Evans and Burke, 2013; Xu et al., 2013), we observed patterns inconsistent with these expectations ( Figure 5 ). In some manipulative experiments no change in root biomass is observed between precipitation treatments (Sindhøj et al., 2000; Gill et al., 2002; Byrne et al., 2013; Flanagan et al., 2013; Denton et al., 2017), which is consistent with our 2022 results. In other instances, belowground biomass increases with low watering (Wilcox et al., 2016; Slette et al., 2022), consistent with our 2020 results. We suspect time may influence the relationship between belowground root biomass and precipitation. One study found that six years after establishment, plants in a water-limited grasslands had vertically redistributed their roots to optimize water uptake (Zhang et al., 2019). After exposure to reduced precipitation, root biomass redistributed to 10–30cm and 30–50cm depths, while in high precipitation treatments more root biomass was found only at 0–10cm depths. Root redistribution may explain why we observed strong differences between drought in year three that are no longer detectable in year five. Our samples are composites of 0–20cm depths, which could mask the differences between precipitation treatments over time as roots redistribute in response to long term exposure to altered precipitation. Though notably 80% of tallgrass prairie root biomass can be found in the top 25cm of soils (Nippert and Knapp, 2007), suggesting our sampling approach likely still captured the majority of roots.

Plant community composition may likewise play a role in how root standing crops respond to changes in precipitation. For instance one study found in comparison to the most abundant C₄ grass which generated similar root biomass in high and low water treatments, subdominant plants drove the overall increase in belowground biomass observed under drought, a pattern that was consistent across 0–10, 10–20, and 20–30cm depths (Slette et al., 2022). Though we did not observe an interaction between plant community composition and precipitation, it is possible these patterns will develop over time as plants change their rooting strategies in response to long-term exposure to altered precipitation.

Plant community composition was a strong determinant of belowground root biomass, while aboveground production consistently increased with diversity driving combined above and belowground biomass benefits associated with increasing diversity. A few caveats must be acknowledged combing these measures. First, where each year aboveground biomass is removed prior to the growing season, our sampling efforts belowground are unable to distinguish between live and dead roots, nor are they able to remove the previous year’s (or years’) growth. Thus, our measures aboveground represent annual growth, production, while belowground biomass represent a standing crop. Despite these discrepancies in method that are difficult to avoid within the context of the experimental design, combined the patterns still provide some insights into how plant biomass above and belowground together responds to changes in diversity precipitation. Second, combined above and belowground biomass might underestimate the contribution of deep-rooted plants to production, as roots were only sampled to 20cm. Given the sampling limitations, belowground patterns might be underrepresented in combined above and belowground biomass measures.

It is notable that communities containing all grasses had the highest belowground biomass, and their diversity had the greatest positive effect on root biomass, while legume diversity did not generate increases in belowground standing crops ( Figure 4B ). This could reflect well established trait differences between C₄ grasses and legumes. Warm season C₄ grasses, which characterizes most grass species in this experiment, invest heavily in the production of shallowly distributed fine roots. Legumes, in contrast, tend to root deeply, investing fewer resources into the generation of fine roots (Gastine et al., 2003; Prommer et al., 2020; Carmona et al., 2021; Furey and Tilman, 2021). We, unlike previous studies, observed no benefits of having both shallow and deep rooting strategies in a single community ( Figure 4B ). Instead, our multiple family mixtures represent essentially the average of the pool of species in the experiment, a pattern consistent with other plant diversity manipulation experiments sampled less than five years after establishment (Gastine et al., 2003), and thus in early stages of community development.

Importantly, in experiments sampled more than ten years after establishment, mixtures containing both grasses and legumes generated increased root yields (Mueller et al., 2013; Yang et al., 2019), and in some cases generated root productivity benefits in mixtures that were greater than expectations based on belowground monoculture yields (Fornara et al., 2009). This may reflect a delay in the accrual of belowground biomass from slow growing plants or it may likewise reflect delayed benefits associated with plants capable of forming symbioses with N fixers included in a community. Resource benefits may take time to accrue in plant communities as the initial disturbance required to setup the experiment may have resulted in a loss of accessible N and many plant diversity manipulation experiments are established on N deplete post-agricultural soils (Tilman et al., 1997; Hector et al., 1999). The reestablishment of N may develop in these systems over longer time periods, decadal scales, as N needs to accumulate in plant and microbial tissues and be decomposed to become available for plant use. These patterns could likewise be depth dependent, as the most diverse and productive plant communities had the deepest distribution of roots. This pattern has been attributed to the cooccurrence of C₄ grasses and legumes, which have differing rooting strategies from each other in terms of rooting depths and growth form (Mueller et al., 2013). It is possible by sampling only to 20cm we did not capture the range necessary to see the effects of these differing rooting strategies especially as communities develop, though this depth range is commonly used in grassland diversity experiments (Fornara et al., 2009; Lange et al., 2015; Yang et al., 2019).

Benefits from diversity developed belowground three to five years after establishment. These patterns are consistent with other findings suggesting belowground patterns replicate increases in biomass observed aboveground (Mueller et al., 2013; Ravenek et al., 2014; Eisenhauer et al., 2017; Yang et al., 2019). The timing of the emergence of belowground biomass increases observed with diversity was consistent with a similar grassland diversity manipulation experiment (Ravenek et al., 2014). The patterns observed in our study and by Ravenek, Bessler (Ravenek et al., 2014) emerged perhaps slightly faster than those observed by others (Gastine et al., 2003), though in the latter case the underperformance of a typically dominant plant drove those patterns, which may reflect a range of factors including unfavorable environmental conditions for that species. Diverse mixtures (16 species) can generate more roots than the best performing monocultures, suggesting belowground overyielding can develop (Fornara et al., 2009; Yang et al., 2019), though this pattern has not emerged in our experiment. However, the positive relationship we observed between richness and root standing crops suggests we might expect to see strong changes in microbial community composition and function with increased plant diversity five years after planting, as root inputs have been directly linked to increased microbial activity (Eisenhauer et al., 2017). Indeed, changes in soil microbiome composition in response to plant diversity and family composition have been observed in our experiment (Burrill et al., 2023). Since root inputs are more likely to form stable and persistent forms of soil C, these patterns could represent a more direct link between plant productivity and C sequestration.

These patterns confirm that aboveground production increases associated with diversity can generally be replicated in belowground biomass. However, in contrast with aboveground patterns where productivity–diversity relationships are found consistently across different community composition types, belowground whether root standing crops increased with richness is strongly dependent on plant community composition. Perhaps most surprisingly diversity within a single family can both rival (aboveground) and exceed (belowground) biomass increases observed in diverse multiple-family communities. Furthermore, whether belowground biomass increased with diversity occurred independent of altered precipitation regimes. Overall, these patterns clarify some of the biological (i.e. plant community composition) and environmental (i.e. precipitation) conditions under which diversity is associated with increased biomass, further disentangling when patterns aboveground production are linked with belowground biomass.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

LP: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Visualization, Writing – original draft, Writing – review & editing. MH: Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. JB: Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing.