Diversity-induced plant history and soil history effects modulate plant responses to global change

Global change has dramatic impacts on grassland diversity. However, little is known about how fast species can adapt to these changes and how this affects their responses to global change. To close this gap, we performed a common garden experiment testing whether plant responses to global change are influenced by the selection history of the plants and the conditioning history of soil at different levels of plant diversity. Therefore, we collected seeds and took soil samples from 14-year old plant communities of a biodiversity experiment. Offspring of plants from low- and high-diversity communities were either grown in their own soil or in soil of a different community, and were either exposed to drought, increased nitrogen input, or a combination of both. Results show that, under nitrogen addition, offspring of plants selected at high diversity produced more biomass than those selected at low diversity, while drought neutralized differences in biomass production. Moreover, under the influence of global change drivers, mainly soil, and to a lesser extent plant history, influenced the expression of plant traits. Our results show that plant diversity modulates plant-soil interactions and growth strategies of plants, which feedback on the eco-evolutionary pathways of the plants and thus their responses to global change.


Introduction
in a freezer and allowed to germinate in spring 2017. After germination, soil samples were 294 collected from the plots and mixed with sterilized background soil (5% + 95%), filled in pots 295 and planted with two seedlings (12 pot replicates per plot). In four pots per plot, plant and soil 296 had the same plot origin (home soil); in four pots, species richness of plant and soil origin were 297 the same, but plant species composition was different (away-same soil) and in four pots, species 298 richness of plant and soil origin were different (= different origin of plant and soil; away-299 different soil; total Nrpots = 576). Plants were exposed to global change drivers: drought, 300 nitrogen input, or the combination of drought and nitrogen input, or were not treated (control).  biomass production as well as on plant traits (growth height ("Height"), shoot nitrogen 315 concentration ("Nshoot"), leaf greenness ("Greenness"), leaf dry matter content ("LDMC"), specific 316 leaf area ("SLA"), stomatal conductance ("gs"), mildew infestation ("Mildew"), root diameter 317 ("Dia"), specific root length ("SRL"), root length density ("RLD")) of the four study species. For 318 legacy effects, schematic illustrations of plants indicate differences in shoot and/or root biomass, 319 when originated from two-species ("2") or six-species ("6") communities (= plant history (PH)), 320 when grown in two-species ("2") or six-species ("6") community soil (= soil history (SH)), or 321 when grown in away ("a") or home ("h") soil (= soil treatment; "as" = away-same soil  four study species) originated from two-or six-species communities (plant history; a, d, g); 340 grown in soil originated from two-species or six-species communities (soil history; b, e, h); or 341 grown in home, away-same or away-different soil (soil treatment; c, f, i) and were either non-342 treated (control) or treated with drought, nitrogen input (N input) or a combination of both (D 343 + N). Bars show mean values (± 1 SE); stars above bars indicate significant differences (P < 344 0.05), stars in brackets indicate marginally significant differences (P < 0.1).  referred to as "highly-productive" species, while A. pratensis and P. trivialis are referred to as 553 "low-productive" species.

Preparation of background substrate and study plants 556
For the pot substrate, we used a sterilized sand-soil mix (= background substrate), which 557 was then inoculated with fresh living soil (5% of the total substrate by weight) from the selected 558 plots. This inoculation method is a common procedure to investigate plant-soil interactions and 559 has the advantage that only low amounts of living soil are needed and that potential abiotic  For the other three species, one seedling per pot plate cell was removed if more than two seeds 577 were germinated. can be found in Appendix S1). On 23 August 2017, treatments with the global change drivers 608 were started. For every treatment (control, drought, nitrogen input, combination of drought and 609 nitrogen input), we used three of the 12 pots per plot (one home, one away-same, and one away-610 different pot, respectively; Fig. 1).

611
(I) For control, pots were watered as before (380 ml; every other day) and were not 612 fertilized.

613
(II) Drought was simulated by reduced water saturation (= 30% water saturation = 225 ml; 614 calculation can be found in Appendix S1). Pots were still watered every other day but

621
(IV) For the combination of drought and nitrogen input, pots were watered with a reduced 622 amount (225 ml) and were fertilized once a week (in the same way as for the nitrogen 623 input treatment alone).

624
Once a month, all pots were weighted before watering. The measured weight per pot was 625 subtracted from dry soil weight plus the assigned amount of water (380 or 225 ml). The 626 difference revealed the amount of water which was then used to water the pot to keep the 627 anticipated levels of water saturation for the drought and control treatment.

648
LDMC was calculated as the ratio of dry weight to fresh weight (mgleaf gleaf -1 ) and SLA as the 649 ratio of leaf area to dry weight (mmleaf 2 mgleaf −1 ).

650
For biomass harvest, plants were cut at ground level, and roots were cleaned by rinsing 651 off all soil over a 0.5 mm sieve. The fresh root biomass was weighed and a subsample of around  shoot and root biomass production.

708
Because of multiple significant interactions between legacy treatments and global 709 change driver treatments (Appendix S2: Table S1-S10), we further analyzed the data for each 710 global change driver treatment separately. We used plant history, soil history, and soil treatment 711 as fixed effects for species-specific analysis, and for analyses across all four species, we 712 extended the models by fitting species identity first and all possible interactions between species 713 identity and legacy treatments in the end.

714
All models were fitted with maximum likelihood (        nitrogen input resulted in lower SLA in home than in away soil (Table S3). Fertilized D. glomerata 1120 plants had higher SLA in six-than in two-species community soil (  (Table S3). In D. glomerata, plants had higher gs when originated from six-species communities and 1128 treated with both global change drivers; however, this was also found in control plants (Table S4).

1129
Across all study species, root diameter, SRL and RLD were not influenced by legacy treatments 1130 when treated with global change drivers (Table S1). In A. elatius, root traits also did not differ, when  (Table S2). In A. pratensis, plants exposed to drought had higher SRL and RLD 1134 in two-than in six-species soil. When fertilized, we did not find an effect of legacy treatment, but the 1135 combination of both global change drivers led to higher SRL and lower root diameter when plants 1136 were grown in away-same than in away-different or home soil (Table S3). In D. glomerata, RLD of 1137 plants exposed to drought was higher when originated from six-species than from two-species 1138 communities. This positive diversity impact disappeared when fertilized (Table S4). In P. trivialis, fertilized, this difference disappeared (Table S5).

1141
Mildew infestation of D. glomerata plants exposed to drought was higher in home than in away soil, 1142 while this drought effect was cancelled out by nitrogen input (Table S6). Mildew infestation of P. 1143 trivialis plants was not significantly influenced by plant or soil history, neither with nor without 1144 global change drivers (Table S6).   N)). Shown are degrees of freedom (Df), Chi 2 and P-values (P). Significant effects (P < 1160 0.05) are given in bold, marginally significant effects (P < 0.1) in italics.

1162
Growth  and nitrogen input (D x N)). Shown are degrees of freedom (Df), Chi 2 and P-values (P). Significant 1205 effects (P < 0.05) are given in bold, marginally significant effects (P < 0.1) in italics.