View all hidden authors and organizations

Contributed by David Tilman, October 13, 2021 (submitted for review on June 18, 2021; reviewed by Andy Hector and G. Philip Robertson)

Plant biodiversity and soil fertility are declining. We found that restoring plant biodiversity on nutrient-poor, unfertilized soils resulted in more soil fertility increases than when these same plant species were grown in monoculture. The plant species in this biodiversity experiment are on a trade-off surface in terms of their nutrient content traits, preventing any one species or any type of species from significantly increasing soil fertility. Our results have implications for degraded agricultural ecosystems, suggesting that increasing plant functional biodiversity may help restore soil fertility. The creative application of our findings to pastures, cover crops and intercropping systems can provide greenhouse gas benefits from soil carbon storage and reduce the amount of fertilizer required to achieve optimal yields.

For ten thousand years, fertile soil has been an indispensable resource for mankind, but people have little knowledge of the ecological mechanisms involved in the creation and restoration of fertile soil, especially the role of plant diversity. Here, we use the results of long-term unfertilized plant biodiversity experiments to determine whether biodiversity, especially plant functional biodiversity, has affected the regeneration of fertility on degraded sandy soils. After 23 years, compared with the single cultivation of these same species, the soil nitrogen, potassium, calcium, magnesium, cation exchange capacity and carbon of the plots containing 16 species of perennial grassland increased by about 30% to 90%, and had about 30% to 90%. 150 to 370% more N, K, Ca and Mg in plant biomass. Our research results indicate that biodiversity may be combined with the increase in plant productivity caused by higher biodiversity, leading to increased soil fertility. In addition, for plots with high plant functional diversity, those plots containing grass, beans and weeds, the ratio of N, K, Ca and Mg accumulated in the total nutrient pool (plant biomass and soil) only contains this There are many more plots of one of the three functions. Group. Plant species in these functional groups trade-off between their tissue N content, tissue K content, and root quality, which shows why species from all three functional groups are essential for regenerating soil fertility. Our findings suggest that creative use of plant diversity may help restore soil carbon storage and soil fertility.

For the soil to be fertile, it must provide sufficient amounts of nutrients that may restrict plant growth, such as nitrogen (N), phosphorus (P), potassium (K), calcium (Ca) and magnesium (Mg), and have enough organic matter To retain water and nutrients (1, 2). Low levels of one or more of these factors can reduce plant productivity. In natural ecosystems, plants fix carbon (C) and N, release unavailable soil minerals through root chemistry and migrate them from deep soil to surface soil, and by absorbing and retaining nutrients, they contribute to the creation of fertile soil. Contribution (3⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ ⇓ –12). However, if plant species differ in their ability to release, capture, or retain certain restricted soil nutrients (5), then any single growing species may cause the soil to be relatively deficient in those hard-to-obtain nutrients. If plant species balance their ability to acquire different nutrients, each species is better at acquiring certain nutrients, but worse for others (13), then the diversity of plant species is important for long-term accumulation Multiple nutrients may be essential. Elements needed for soil fertility.

Here, we use long-term grassland biodiversity field trials to explore the potential roles that different perennial grassland plant species, plant traits, and plant biodiversity may play in generating and restoring soil fertility. Although greater plant biodiversity is associated with greater primary productivity and soil carbon accumulation (14⇓ ⇓ –17), increasing soil carbon alone does not make the soil more fertile. Higher fertility also requires the addition of all potentially restrictive nutrients such as N, P, K, Ca, and Mg, as well as an optimal soil pH and sufficient soil cation exchange capacity (CEC) (1, 2). Here, we tested the hypothesis that "the sustainability of soil nutrient cycles and the sustainability of soil fertility depend on biodiversity" (18).

Because higher plant species richness is related to more available soil nutrient absorption and greater plant biomass production (18), if the increased nutrients in plant biomass return to the soil as plant tissues decompose, higher plants Biodiversity may increase soil fertility (3, 4, 6, 7, 12, 19⇓ –21). The greater the plant diversity, the lower the leaching loss of these nutrients (22). This increase in soil fertility can increase biomass production and generate positive feedback, because more biomass input will add more nutrients to the soil (7, 19, 23). In particular, as the roots, leaves and other plant parts fall off, soil bacteria, fungi and invertebrates change and stabilize these organic matter inputs and release nutrients when they decompose plant tissues (9, 10, 21, 24⇓ ⇓ ⇓ ⇓ – 29). The nutrients released by decomposition can promote plant growth, thereby increasing the plant biomass that subsequently returns to the soil (7). In nutrient-poor soils, more plant diversity may lead to more accumulation of soil nutrients and organic matter, so compared with low-diversity ecosystems, plant productivity may increase more over time ( 21, 30, 31).

Our experiment was planted in the spring of 1994 to control the composition and diversity of perennial grassland plant species growing on sandy degraded soil. In August 1993, 6 to 8 cm of topsoil in the upper layer was removed from the abandoned farmland to eliminate the weed soil seed bank. Then the land was plowed many times, and it was bare soil from August 1993 to the spring of 1994. The 154 9 × 9 m plots established for this experiment were sown with 1, 2, 4, 8, or 16 perennial grassland species randomly selected from 18 species pools. Here, "plant diversity" refers to the number of species planted in a plot. We also calculated the "functional group diversity" of plants according to the functional groupings commonly used in grasslands, and classified plant species as grasses, legumes or weeds (32). The plots have never been fertilized and burned down in the early spring before greening each year, and large vertebrate herbivores are excluded by fences. Our off-glacial scouring plain soil in east-central Minnesota is agronomically classified as organic matter, with “very low” N and K content, but “very high” P content (SI Appendix, Table S1). Using archived soil samples collected from each plot before planting in 1994 and samples collected after 23 years of growth in 2017, we measured soil total nitrogen and carbon, exchangeable potassium, calcium and magnesium, soil CEC, soil pH, and The upper part of the extractable Brefite soil profile is 0 cm to 20 cm. In August 2017, the aboveground and underground (root; depth of 0 to 30 cm) plant biomass, as well as N, P, K, Ca, and Mg in the aboveground and root biomass were measured. We also measured the above-ground plant histochemistry of each individual plant species.

Higher levels of plant diversity lead to an increase in many factors that contribute to soil fertility. The comparison of soil in 1994 and 2017 in pretreatment showed that the soil N, K, Ca, Mg and C, CEC and soil pH value of the plots with higher plant diversity increased significantly (Figure 1 and SI appendix, Figure S1 And Figures S1 and S2 and Table S2). The soil phosphorus level was very high before planting, and it was still very high in 2017, and the impact of plant diversity was not detected (Figure 1H and SI appendix, Table S2). Although no plots have been fertilized, by 2017, the total soil N (0 to 20 cm depth) of the 16 treatments was 29% higher than the single cultivation average of these same species, and the soil K was 95% higher, 30%. The soil calcium content increased by %, the soil magnesium content increased by 29%, the total soil carbon content increased by 35%, the CEC increased by 34%, and the soil acidity decreased (the pH value of monoculture increased by 0.2) (Figure 1). Although the soil bulk density decreased with plant diversity, the average value ± SE of 1.46 g⋅cm−3 ± 0.015 in monoculture dropped to 1.37 g⋅cm−3 ± 0.018 in 16 species plots (F1,85 = 23, R2 = 0.21, P <0.001), the soil element levels expressed on the basis of concentration and area density are similar in nature (SI Appendix, Figure S3-S5 and Table S3).

Soil chemistry and plant diversity. Average value of green (diamond) in 1994 and 2017 (A) total carbon, (B) total nitrogen, (C) exchangeable orange (circle) soil chemistry (0 to 20 cm depth) before planting ±1 SE potassium, (D) Exchangeable calcium, (E) Exchangeable magnesium, (F) CEC, (G) soil pH, and (H) extractable briny phosphorus and the number of planting species (1, 2, 4, 8 or 16; right Number) relationship scale). The line is linear regression ± 1 SE (n = 154 plots). Use soil bulk density to calculate the amount of C, N, P, K, Ca, and Mg (grams per square meter). The sample size for each diversity level (1 to 16 species) is 1 species = 32 plots; 2 = 28; 4 = 29; 8 = 30; 16 = 35.

In surface soils with high plant diversity (0 to 20 cm depth), more accumulation of N, K, Ca, and Mg is accompanied by a greater percentage increase in the size of the above-ground and underground nutrient pools relative to single cultivation. Plants in 2017 Biomass (Figure 2). Linear regression shows that the tissue pools of N, K, Ca, and Mg in above-ground and underground biomass, relative to the average level of all single cultivations, are positively dependent on the logarithm of plant diversity (all P <0.001; Figure 2 and SI appendix, Figures S6 and S7 and Table S4).

Relative to the nutrient content of a single cultivation level. The 2017 biomass relative to shoot (blue; circle) and root (red; diamond) nutrient content of each diversity treatment is expressed as a percentage of the average nutrient content of all single cultivation combinations in 2017 (mean ± 1 SE). (A) Nitrogen, (B) Potassium, (C) Calcium and (D) Magnesium in the above-ground branch biomass and underground (0 cm to 30 cm) root biomass contained in the percentage change of the average of all single cultivation. The line is linear regression ± 1 SE (n = 154 plots). Shoots are dry above-ground biomass, and roots are dry underground biomass (0 cm to 30 cm). Multiply the biomass by the concentration of each element.

However, why might the production of larger plant biomass and the accumulation of soil nutrients depend on plant diversity? Due to the functional differences between species, diversity is thought to affect ecosystem processes (32). Since the perennial herbaceous species of the tall grass steppe are usually functionally classified as Poaceae, Fabaceae and weeds (excluding legumes; Compositae, Lamiaceae and Apocynaceae), we We tested whether the accumulation rate of specific nutrients in the ecosystem is related to the existence of these plant functional groups. To this end, we classify each plot according to the presence of grass (G), legumes (L) or weeds (F) species. This gives the composition of seven functional groups: G, L, F, G+F, L+F, G+L, and G+L+F. The ecosystem pool that accumulates N, K, Ca, and Mg (Figure 3) is calculated as the change in each plot of each element in the soil from 1994 to 2017 (in 0-20 -cm soil depth increments) and added to The total amount of each element (in grams per square meter) accumulated in the shoots and roots in 2017, because all plant biomass was removed the year before planting.

(A) Nitrogen, (B) Potassium, (C) Calcium, and (D) Magnesium changes in the total nutrient pool (black dots) of the ecosystem composed of each functional group. Each black dot represents the average value of the total ecosystem pool ± 1 SE. Pond is defined as the change in soil nutrient levels from 1994 to 2017 (0 to 20 cm depth increment) plus the amount of aboveground biomass and roots (0 cm to 30 cm) in 2017; the sum is expressed per square meter Nutrients in grams. The bar graph shows the value of each nutrient in above-ground biomass (gray), below-ground biomass (yellow), and soil (blue). The negative bar displayed below the zero line represents the decrease in elements from 1994 to 2017. Functional group composition: G = grass only, n = 22; F = non-legumes only, n = 10; L = legumes only, n = 11; FL = at least one weed and one legume, n = 5; GL = at least one grass and one legume, n = 23; GF = at least one grass and one grass, n = 14; GFL = at least one grass, one legume and one weed, n = 69 . The letter indicates whether the average value of a specific nutrient is different after Tukey's correction (P <0.05).

Compared to when there is only a single functional group, the presence of all three functional groups (G+F+L graph) is associated with the largest increase in the N, K, Ca, and Mg ecosystem pools (Figure 3). In particular, the accumulation of plots containing all three functional groups (G+L+F) in the soil and the plant biomass of each of the four nutrient elements (N, K, Ca, and Mg) are significant Higher than a plot with only a single functional group. Group (F, G or L diagram; Figure 3 and SI appendix, Table S5). This is not the case for any combination of only two functional groups (Figure 3). Compared with the G, F, or L plots, the N, K, Ca, or Mg ecosystem pools accumulated in the G+F and F+L plots did not increase significantly (Figure 3). The results of the G+L plot are intermediate, and there is no significant difference in the accumulation of Mg from F or L or the accumulation of Ca from L, but it has a greater N accumulation than the plots planted with a single functional group or G+F . Finally, although G+F+L and G+L have no difference in the ecosystem library of N, Ca or Mg, the K library of G+F+L is significantly higher than all other functional components except F+L , Indicating that there is an important reason for the large increase in K observed in high plant biodiversity.

In general, the plots planted with a single functional group are significantly lower than the G+F+L plots, and the ecosystem pool of most nutrients accumulated in the plots is significantly lower. The plots planted with two functional groups are only matched in a quarter The comparison significantly exceeds the single functional group (Fig. 3). A separate analysis of each plant, root, and soil nutrient pool shows that G+F+L produces more ground than plots planted with a single functional group And underground biomass, and accumulate more C, N, K, Ca, and Mg in 87% of the comparisons (39 out of 45 comparisons) (SI Appendix, Figure S8).

On a finer scale, for the amount of each of the four nutrients in the above-ground biomass, G+F+L is significantly greater than G+L, but never significantly greater than F+L (SI Appendix, Figure S8 ). For root nutrients, G+F+L in root K is significantly higher than F+L and G+L. The only functional group whose root K level is as high as G+F+L is F, only forb plots. For root Mg, there is no difference between F+L and G+F+L, but G+F+L has significantly more Mg than G+L. For root Ca, the opposite situation occurs: G+F+L is no different from G+L, but the Ca content is significantly higher than F+L. In summary, these results indicate that the co-existence of weeds and weeds and legumes are important contributors of K and Mg to the ecosystem pool, while the co-existence of legumes, legumes and grasses is more important for N and Ca Contributor.

Since not all of these nutrients may limit the production of plant biomass, we determined which soil variables are more closely related to the observed diversity-dependent changes in productivity, while taking into account the impact of plant diversity. Using linear multiple regression and multi-model inference, we found that the total biomass of plants is positively dependent on the logarithm of the number of species (156 ± 25.6 g⋅m−2 biomass per 1 loge (number of plant species), P <0.001), soil can be Exchange potassium (51.3 ± 8.8 g⋅m-2 biomass/g⋅m-2's potassium, P <0.001), soil total nitrogen (2.88 ± 0.67 g⋅m-2 biomass/g⋅m-2's N, P <0.001) and total soil carbon (0.23 ± 0.05 g⋅m-2 biomass/g⋅m-2 C, P <0.001) (SI Appendix, Table S6). This analysis shows that total soil carbon, total soil nitrogen, and exchangeable soil K are the soil variables with the strongest correlation with the plant biomass produced in this field experiment, which is consistent with our soil analysis, indicating that soil agronomy nitrogen Low content, K and organic matter at the beginning of our experiment (SI appendix, Table S1).

Finally, we determined how plant species in the three functional groups might differ in traits related to soil K, N, and C accumulation. Monoculture and 16 kinds of plots. For soil carbon accumulation, we use the average single cultivated root mass of each species, because the previous results of this experiment indicate that greater root biomass (in grams per square meter) is the variable most closely related to the increase in soil carbon content (14, 17).

The three measurement characteristics of the species (root mass, tissue %N, tissue %K) define a regression plane (Figure 4 and SI appendix) (F2,12 = 6.3, R2 = 0.51, P = 0.014). The species in each functional group tend to be similar to each other, which can be seen from their clustering trend (Figure 4). On the surface of this trade-off, the %N and %K of perennial C4 grass tissues are low, but the root biomass is the highest. Compared with C4 grasses, weeds and legumes with less root biomass are further differentiated: legumes have a higher %N, but a significantly lower %K. In contrast, Forbes' %K is higher, but %N is lower (Figure 4). Forbs and beans have similar %Ca and %Mg levels, and their levels are higher than grass (SI appendix, Figure S10).

The empirical trade-offs between the plant traits of the 15 perennial herbaceous species that persisted in the experiment. The regression plane (F2,12 = 6.3, R2 = 0.51, P = 0.014) fits the aboveground tissue potassium percentage (K) (x axis), aboveground tissue nitrogen percentage (N) (y axis) of a specific species, and the average monoculture Root biomass (grams per square meter; 0 to 30 cm depth; z-axis). Each point represents three of each of the 15 species (SI appendix, Table S7) classified as grass (dark green C4 grass and light green C3 grass), weeds (purple), and legumes (orange) Measure traits. %N and %K represent the average value of the single cultivation of each species and the biomass of each species in five 16 plots. Root biomass represents the average root mass (0 to 30 cm depth) of a single cultivation of each species. Remove the two C3 grasses (light green; below the plane), which are sub-dominant species in the ecosystem and grow poorly in monoculture. Increase the plane fit to F2, 10 = 15, R2 = 0.75, P = 0.001 (not shown)). The points of Andropogon gerardii (C4 grass) jitter slightly on the x and y axis to avoid overdrawing with Sorghastrum nutans (C4 grass).

In the early stages of the experiment, greater plant diversity was associated with more soil nitrate capture (18), and plots of 16 species were about 100% higher than the average yield of these species in monoculture. After 23 years, we found that greater plant diversity was associated with higher levels of soil C, N, K, Ca, and Mg, and the yield of 16 species plots was 200% higher than that of single cultivation. In this and other long-term biodiversity experiments (references 30 and 31, but see reference 33), over time, the primary productivity observed under higher diversity has gradually increased, as well as soil and plant organisms. More accumulation of multiple nutrients and carbon in the amount (figure and figure). 1 and 2) indicate the presence of the positive feedback effect of plant diversity on soil fertility (7, 23), increasing primary productivity over time .

We hypothesize that high plant diversity, especially the co-existence of grasses, legumes and weeds, will lead to more restrictions on the release and capture of soil nutrients (18). This in turn allows more production of plant biomass. We suggest that the nutrients and carbon content of the larger biomass (roots and shoots) produced by different mixtures of grasses, legumes and weeds are recovered during senescence, helping to create a more fertile soil. This more fertile soil will further increase plant biomass production and biomass nutrient pools, in a positive feedback loop that will continue until equilibrium is reached (7, 19, 20).

However, we have noticed that in our experiments, the early spring burning of each year may volatilize some C and N in the litter of wilted above-ground biomass, but may also deposit other elements in the ash that are available in bioavailable forms (such as , Ca, Mg, and K). Because the mass of roots in our grassland-like high-diversity plots is about 4 times that of above-ground biomass, the senescent biomass produced by root turnover may add C, N and other nutrients to the soil (6⇓ –8, 10, 11, 19, 34).

The three-way trade-off shown in Figure 4 illustrates why plant functional diversity may be critical to increasing soil fertility in our unfertilized experiments. This suggests that no single species and functional group itself can span the entire space of the root biomass-NK weighing surface, leading to an increase in soil C, N, and K that seems to limit productivity. In our experiments, these three increases are related to higher productivity, and therefore to the assumed feedback effect of higher productivity and its nutrient content on soil fertility. In contrast, it is usually necessary to increase the fertilization of N, P, and K to increase the productivity of a single agricultural crop, and this fertilization also leads to an increase in soil carbon content (35). However, our monoculture has never been as productive as our highly diverse plots (17). On our site, the agronomic level of soil P is very high before planting and 23 years after planting. Although there are differences in root biomass, N and K between functional groups, while soil C, soil N, and soil K increase with diversity, we found that soil P does not significantly depend on plant diversity or functional groups. There is no difference in P levels between functional groups when grown on this phosphorus-rich soil (Figure 1H and SI appendix, Figures S9 and S10).

The differences in traits between grasses, legumes and weeds (Figure 4) indicate that there is a mechanism link between the effects of plant traits and functional diversity on primary productivity and soil nutrient accumulation. The balance of the surface shows that each functional group should contribute more to the soil, including one of root biomass, N or K, but the other two contribute less. Because legumes have high %N and %Ca content, and weeds have high %K, %Ca and %Mg content, the presence of each of these functional groups should make the soil with a relatively high biomass concentration The elements are particularly rich (Figure 3 and SI appendix, Figures S8 and S10). The high root biomass of C4 grass may reduce nutrient loss through leaching and help increase soil organic carbon (14, 17, 18, 22). In addition, there is no functional group that grows alone that can increase soil carbon and nutrients like when all three groups are present (Figure 3 and SI appendix, Figure S8).

The increase in exchangeable K, Ca, and Mg in the surface soil may come from the root absorption of these elements in the deep soil. These elements concentrate on the surface and die and decompose as above-ground and shallow root tissues, or the early elements are deposited in the form of litter ash. Spring burns (6). For example, Ca tends to move unidirectionally from roots to shoots, and absorption is limited when tissues age (34, 36). If higher root uptake and nutrient recovery from ashes or root turnover are the mechanisms for the accumulation of soil fertility in high-diversity plots, then one would expect the accumulation of plants and soil ponds to be coupled (37). For example, soil K is highly dependent on plant diversity (R2 = 0.47). When 16 types of plots are compared with the average of all single cultivations, K is the nutrient with the largest increase (370%) in the aboveground plant pool (Figure 2). For K, Ca and Mg, compared with monoculture, the aboveground ponds of the 16 species plots increased by about 150% to 370%, and the root ponds increased by about 90% to 150%, which indicates higher Plant diversity leads to larger ecosystems to capture and retain the biomass of these cations (Figure 2). In addition, in these sandy soils, the increase in soil organic carbon is correlated with the increase in CEC (1994: R2 = 0.20, P <0.001; 2017: R2 = 0.52, P <0.001; SI appendix, Figure S11). Increase the retention of K, Ca and Mg in these soils.

Our long-term experiments have shown that soil fertility has seen an amazing increase in dependence on diversity. However, this magnitude may depend on our initial soil characteristics. The soil of our site is formed on a sandy plain deposited by glaciers and is classified as entisols in taxonomy, and its stratigraphic development is limited (38). At the beginning of this experiment, some of the topsoil and its organic matter were removed, and the soil was plowed and rounded, which tended to homogenize the remaining topsoil and deeper soil layers. In our starting soil, the initial levels of soil C and N are low, while the P level is high, which is a characteristic of the geologically young soil that is gradually developing (39). Therefore, if our degraded and abandoned agricultural soils accumulate C and nutrients in a logical manner (40), the increase rates of soil C, N, K, Ca, and Mg that we observe under high plant diversity may be greater than if our The soil initially contained higher levels of these elements.

Our results and their possible mechanism basis may provide insights into ways to restore soil carbon and increase agricultural ecosystems and manage restrictive soil macronutrients in forests. For instance, incorporating greater plant functional diversity via appropriate choice of the plant species used in crop rotations, intercropping, or cover crops may lead to long-term increases in soil fertility and subsequent reductions in the amount of fertilizer needed (41⇓ ⇓ –44 ). Because our results indicate that it is not just the number of plant species that is important, but a suitable set of complementary plant traits, it will be very important to determine whether only three such plant species may provide significant soil benefits relative to single cultivation. interesting.

In our research, the increased input of senescent plant biomass that occurs in higher diversity must be transformed and mineralized by soil microbes and invertebrate communities, which suggests that soil microbial biodiversity may also help explain Figure 1 The result of (45, 46), which is an interesting possibility (9, 16, 25, 27, 28). More accumulation of plant or soil pathogens in monoculture, or the increase of soil reciprocators, or the decrease of soil pathogens when plant diversity is high are other possible ways that microbial biodiversity may affect ecosystem functions over time (25, 47) .

In general, our results indicate that plant diversity, including plant functional diversity, can play an important role in the production of soil fertility, which may be a positive feedback of increased dependence on soil fertility diversity through nutrient capture and productivity effect. Our results raise an interesting possibility that the high plant diversity of most natural ecosystems may be an important factor in the formation of fertile soils around the world. Creative use of plant diversity may help increase soil carbon storage and degraded soil fertility.

In August 1993, the experimental field had been abandoned for more than 15 years when the herbicide glyphosate was applied and the once dead and dry surface vegetation was burned. Scrape the top 6 cm to 8 cm of soil to reduce the presence of weed annual plant seeds in the soil seed bank. This also reduces soil carbon and soil nutrient levels. The site was plowed twice and harrowed many times that year, and plowed again before planting in May 1994. These 168 plots were originally 13 mx 13 m, but were later reduced to 9 mx 9 m in the center, where 1, 2, 4, 8 or 16 perennial plant species randomly selected from 18 species libraries were planted. The pond is composed of perennial grassland species common in the regional tall grass prairie and two types of oak trees common in the nearby oak savanna. Herbs are functionally divided into C4 grasses, C3 grasses, legumes and non-leguminous weeds. There are four species in each functional group.

Due to poor establishment, 14 plots withdrew from the test, leaving 154 plots. In particular, these two oak tree species failed to survive due to annual burning. Two of the four C3 grasses, Agropyron smithii and Elymus canadensis, initially sprouted but failed to survive for a long time. Therefore, the final experimental design consisted of 154 plots, seeded with 1, 2, 4, 8, or 16 randomly selected perennial grassland species, with 32, 28, 29, 30, and 35 replicates for each diversity level. . A single culture is not intentionally copied. In contrast, monoculture processing is based on randomly selecting individual species from a species library. Most species are randomly assigned to two single cultivations. However, Poa pratensis and Panicum virgatum are monocultures; Liatris aspera, Lespedeza capitalata, Dalea purpureum and Schizachyrium scoparium have three monocultures; and Sorghastrum nutans have four. In addition, one hybrid Solidago failed to germinate in the first year and was planted with another hybrid Solidago same species in the spring of 1995. In the third year, hardwood flowers germinate and eventually mature and dominate its monoculture.

Every spring, before greening, the plots are burned, but no fertilization is applied. Each plot is manually weeded every year to remove unplanted species. The experiment was fenced to exclude white-tailed deer. More detailed information can be found on the website of the Long-Term Ecological Research (LTER) Project of the Cedar Creek Ecosystem Science Reserve. The experiment name is "e120: Biodiversity II: Effects of Plant Biodiversity on Population and Ecosystem Processes" ( https://www. cedarcreek.umn.edu/research/experiments/e120).

For each plot in the experiment, we determine whether it is planted with grasses ("G", grasses), legumes ("L", legumes) or weeds ("F", asteraceae, Lamiaceae and Apocynaceae). Then we classify each plot according to the functional group planted in it. This grouping gives seven functional group compositions: G, L, F, G+F, L+F, G+L, and G+L+F. The sample size is as follows: G = grass only, n = 22 plots; F = forb only, n = 10; L = beans only, n = 11; FL = at least one weed and one bean, n = 5; GL = at least one grass and one legume, n = 23; GF = at least one grass and one grass, n = 14; GFL = at least one grass, one legume and one weed, n = 69 .

In September 2017, a soil sampler with a diameter of 1.9 cm was used to collect nine soil cores from each plot at a depth of 60 cm and in increments of 20 cm in a uniformly spaced 3 × 3 sampling grid pattern. The nine cores of a plot are then combined, dried at 60°C, sieved into 2 mm sections, and then mixed thoroughly. In 1994, before planting, soil samples were similarly collected and processed. Those samples remain in the glass archive vials until analysis. The 2017 soil samples were also archived similarly.

In August, when the above-ground biomass was approaching its peak, two parallel 0.10m x 6m strips were cut on the soil surface of each plot every year to sample plant biomass. The strips are located in the middle half of each plot, about 1 m to 2 m apart, and their position should not clamp the area that has been cut off in the past ten years. One zone per plot is classified into species; the other is unclassified. The root biomass was then sampled in the pruning area, using a soil probe with a diameter of 5.1 cm to collect three cores (six per plot) on each strip with a depth of 30 cm. Clean the roots with a sieve to remove the soil. The root and above-ground biomass are dried in a dehumidifying and drying chamber at 60 °C until a constant quality is reached. Since 1996, the above-ground biomass has been sampled every year. The underground biomass was sampled in 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2006, 2010, 2015 and 2017.

The University of Minnesota Research and Analysis Laboratory analyzed the soil samples we collected in 2017 from 0 cm to 20 cm, 20 cm to 40 cm, 40 cm to 60 cm, and 1994 from 0 cm to 20 cm in depth increments for Exchange cations (calcium, magnesium, potassium, sodium) were extracted with pH 7 ammonium acetate, and aluminum was extracted with 1 M KCl, and then analyzed by inductively coupled argon plasma emission spectrometer (iCap 7600 Duo ICP-OES analyzer, Thermo Fisher Scientific) (48). Effective CEC is measured by the summation method of exchangeable Ca, Mg, K, Na, and Al (48). Use standard Bray-1 extract (49) (0.025 M HCl and 0.03 M NH4F) to measure extractable P, and perform colorimetric analysis on a Brinkmann PC 900 probe colorimeter (Thermo Fisher Scientific). In 2017, the Waypoint Analytical Commercial Laboratory analyzed soil samples of 0 cm to 20 cm, 20 cm to 40 cm, and 40 cm to 60 cm, as well as 0 cm to 20 cm, 20 cm to 40 cm, and 40 cm to 60 cm. Used in 1:1 soil in 1994: soil pH value in deionized water slurry. In 1994, 2015 and 2017, the total soil carbon and nitrogen of soil samples with depths of 0 cm~20 cm, 20 cm~40 cm, and 40 cm~60 cm were analyzed. Use the average of 2015 and 2017 to reduce sampling noise. Use dry combustion gas chromatography to analyze ground soil on an elemental analyzer (Costech ECS 4010 CHNSO Analyzer). Due to the lack of carbonate minerals in these soils (38), total C represents total organic C. In order to assess the agronomic status of our starting soil (SI Appendix, Table S1), Waypoint Analytical used losses in 1994 soil ignition.

The chemical composition of the dry above-ground and underground biomass sampled in August 2017 was analyzed. The homogeneous and unsorted dry biomass shear bars from each of the 154 plots were completely ground before sampling. In addition, for all single cultivation plots and 5 plots of 16 species, the total number of classifications (including leaves, stems, and inflorescences (if any)) of each of the 15 plants was ground to provide Estimation of individual characteristics. Since the legume lupin has a spring growth and seed shedding pattern, a separate spring biomass sampling is required to determine its tissue nutrients. Therefore, the L. perennis sample from June 2019 was also analyzed and used instead of the 2017 sample . Use Model 4 Wiley Mill (Thomas Scientific) to grind plant samples, and analyze histochemistry in a commercial laboratory Waypoint Analytical using EPA manual SW-846 Method 3050B. Specifically, each plant sample was digested in concentrated nitric acid and then heated at 95°C for 15 minutes. Then add 30% hydrogen peroxide until foaming is no longer observed, then add 5 mL of concentrated HCl and continue heating for 30 minutes. After filtering with Whatman #2 and analyzing by inductively coupled plasma emission spectroscopy (Perkin-Elmer Optima 8300), finally 5 to 10 ml of water is added to the sample.

Adjust the concentration of each measured soil variable to an area density amount (g/m2) with soil bulk density. In 2018, the AMS Inc. separate soil core sampler and removal jacks (part numbers 400.99, 403.41, 403.73, 211.05, and 211.06) were used to measure the bulk density at a depth of 60 cm (in 20 cm increments) in a subset of the plots. 87) The sample size is 26, 16, 15, 15 and 15, randomly selected on the number of 1, 2, 4, 8, and 16 species respectively. At the same time, each species includes two replicates in a single cultivation (if available ). For the unmeasured plots, we used a linear regression of the dependence of the bulk density (0 cm to 20 cm) on the soil carbon percentage measured in 2017 (0 cm to 20 cm) to estimate the bulk density. Bulk density was not measured in 1994. We estimate the regression fitting of soil carbon percentage (0 cm to 20 cm) measured in 1994 and 2017 soil carbon percentage (0 cm to 20 cm) to calculate the bulk density of each plot in 1994 (0 cm to 20 cm). 20 cm) value. The average bulk density predicted in 1994 was 1.45 g·cm-3 (0 cm to 20 cm), which is similar to the Nymore series soil bulk density measured in the field soil survey of 1.4 g·cm-3 (0 cm to 23 cm) (Reference 38 , Table 5, page 22). The processing average of the estimated bulk density from 0 cm to 20 cm in 1994 ± 1 SE is 1.45 g⋅cm−3 ± 0.01, 1.44 g⋅cm−3 ± 0.00, 1.45 g⋅cm−3 ± 0.01, 1.44 g⋅cm − 3 ± 0.01 and 1.44 g⋅cm−3 ± 0.01, which are 1, 2, 4, 8, and 16 species, respectively. The processed average of the bulk density measured at 0 cm to 20 cm in 2018 is ± 1 SE is 1.46 g⋅cm−3 ± 0.015, 1.43 g⋅cm−3 ± 0.015, 1.42 g⋅cm−3 ± 0.019, 1.36 g⋅cm−3 ± 0.019 and 1.37 g⋅cm−3 ± 0.018, which are 1, 2, 4, 8, and 16 species, respectively. Then, we use the equivalent soil quality method to adjust the sampling at 20 cm depth for all plots, although assuming the density has changed, increase the mass by increments from 20 to 40 cm depth or subtract the relative mass from 0 cm to 20 cm . The change in bulk density relative to the 1994 reference value (SI appendix) (50). The nutrient pool in the above-ground biomass is calculated using the percentage of each element in the biomass from unsorted cut strips multiplied by the dry biomass in each plot. Aboveground biomass is calculated as the average of the dry weight (grams per square meter) of the classified and unclassified strips, as has been done historically in this experiment. Plant litter, the dead biomass on the soil surface, is not included in this measurement, but given that the field is burned every year, the amount of litter is negligible. Underground biomass is calculated as the dry weight (g/m2) of each plot (0 cm to 30 cm). In order to reduce the sampling noise caused by interannual changes, the average of the above-ground and underground biomass measured in 2015 and 2017 is used for all statistical analysis and calculations. To improve readability, we refer to these as 2017 in the text as the year for chemical analysis of plant tissues. The changes in the ecosystem nutrient pools of N, K, Ca, and Mg are estimated to be the changes of each element in the soil from 1994 to 2017 for each plot (g/m2 of each element in the soil depth of 0-20 cm) Increment) plus the total amount (in grams/square meter) of each element measured in the above-ground and underground biomass (0-30 cm) in 2017.

Use R version 4.1.1 and JMP 14 Pro for analysis. Linear regression is used to test the dependence of soil and plant variables on experimental plant biodiversity using the natural logarithm of the number of plant species (1, 2, 4, 8, 16). For the analysis of the plant biomass pool, the percentage increase of a single cultivation average was used as the response variable. For each set of analyses, the false discovery rate (51) correction was applied to the P value of each regression. The regression results are robust to various transformations of the y variable and are presented on the untransformed y scale. Use a generalized least squares model with power to test the sum of above-ground and underground biomass (total biomass) against the logarithm of plant biodiversity and soil variables (C, N, Ca, Mg, K, P, p​​H) The variance structure (varPower) on the dependent fitted value (R package nlme). Multiple model reasoning (R package MuMIn) is used as a model selection method, the natural logarithm of each soil variable and plant biodiversity and the conditional average value using Bayesian information criteria. A generalized least squares model with variance structure was used to test the dependence of ecosystem nutrient pools (N, K, Ca, and Mg) on ​​different plant functional groups to explain unequal variances (varIdent, nlme). Use least squares means (R package emmeans) to compare the differences between means, and then use Satterthwaite's estimated degrees of freedom for Tukey correction. As a supplementary analysis, we conducted the same test on the effects of the existence of functional groups on soil C, N, K, Ca and Mg changes on above-ground and underground biomass (underground logarithmic transformation) and N and K pools. Above-ground and underground organisms The amount of Ca and Mg.

By using linear regression to analyze the dependence of underground biomass (0 cm to 30 cm) on aboveground N percentage and aboveground K percentage in monoculture, the surface of the trade-off of plant species in terms of their traits was tested. The time series average of the underground biomass of each species in a single cultivation is the response variable, and the average histochemistry (%N and %K) of each species measured in its single cultivation and five 16-species plots is the explanation variable. The R graphics package rgl is used to generate the regression plane with an aspect ratio of 1:1:1 (x:y:z) in Figure 4.

All data used in this article has been saved and is available from the Environmental Data Initiative. These data sets include the following: Soil C (https://doi.org/10.6073/pasta/9ed3740e181a3c41ec2cb787ef3a615b), Soil N (https://doi.org/10.6073/pasta/9b55ae5a418c59fb8e3d (https://doi.org/ 10.6073/pasta/9b55ae5a418c59fb8e3d) doi.org/10.6073/pasta/b37f4e38718784480259f3d92fb7a9d7), soil CEC, K, calcium, magnesium (https://doi.org/10.6073/pasta/0fa7f5b395fhttps://doi.org/10.6073/pasta/b37f4e38718784480259f3d92fb7a9d7), soil CEC, K, calcium, magnesium (https://doi.org/10.6073/pasta/0fa7f5b395fhttps://doi602. org/10.6073/pasta/810ebc0a46361bd6a4d1693f17b440fb), soil OM (https://doi.org/10.6073/pasta/6521113c6115b8fbeae7ef0ab6ebca9e), aboveground biomass (https://doi.org/10.6073/pasta/7ef2def75062dec735 https://doi.org/ 10.6073/pasta/0479da667672693c3cf2a6b2c8d14002), soil preparation plant histochemistry (% N, P, K, Ca, Mg), (https://doi.org/10.6073/pasta/f433650417672693c3cf2a6b2c8d14002), (https ://doi.org/10.6073/pasta/f43365041736da2da2dafsoil) .org/10.6073/pasta/270fbd77c7bd6d1cbb59647a3eccb639), individual plant chemical properties (% N, P, K, Ca, Mg) (https://doi.org/10.6073/ pasta/cea0d7152825cb639). All data are submitted by the Cedar Creek Ecosystem Science Reserve site of the LTER network. All research data is additionally included in supporting information. The data reproduced in the main text of the chart and analysis are contained in data set S1 (Figure 1-3) and data set S2 (Figure 4). SI appendix, figure. S1, S2, S6-S9 and S11 can be reproduced using data set S1. SI appendix, figure. S3-S5 can be reproduced using data set S3. Data set S4 provides data to reproduce the conversion of soil concentration to area density. Data set S5 provides data used to reproduce the conversion of tissue concentration to area density. Data set S6 provides the data used to reproduce the response variables derived in Figure 2. To reproduce the SI appendix, the data in Figure S10 is in the data set S7. The metadata of these data sets is provided in the SI appendix. The R code used to reproduce the analysis is archived on Zenodo and will be maintained by GNF (https://doi.org/10.5281/zenodo.5565171).

We are grateful to Troy Mielke and many interns and staff who have been managing experimental treatments and collecting data since 1994. NSF LTER grants DEB-9411972, DEB-0080382, DEB-0620652, DEB-1234162, and DEB-1831944 to fund this work, as did the Balzan Foundation’s award to DT

Author contributions: DT design research; GNF and DT conducted research; GNF and DT analysis data; GNF and DT wrote this paper.

Reviewers: AH, University of Oxford; and GPR, Michigan State University.

The author declares no competing interests.

View relevant content online, such as comments.

This article contains online support information at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2111321118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.