Climate change and primary production: Forty years in a bunchgrass prairie

Our ANPP observations are counter to expectation given that NBR has become warmer and drier based on annual averages over the past 109 years, as reported elsewhere for western Montana [7,75]. This result indicates that specific seasonal, not annual, climate change trends may be more important [8,13,23–25] at local spatial scales [11–14], and questions U.S. grass- land projections made at large spatial scales (national and regional) based on annual average climate values [8,9,76,77]. Phenology changes. The Julian day on which peak “green” plant biomass (> 88% of ANPP) occurs declined by 23 days (16%) over our study. This large phenological change was not related to increasing temperature as often projected [7,75], but related to increasing grass relative abundance and the increasing ratio of late-May–June precipitation to temperature. As noted earlier, NBR grasses are almost exclusively (> 99%) “cool season” (C3), which grow rap- idly and more abundantly with wetter and cooler conditions. Therefore, the peak period of growth may be more affected by changing species composition and specific seasonal, not annual, climate trends. At another western Montana location, earlier flowering dates for dicots are reported, but this is correlated with different local seasonal climate trends [78] than pre- sented here. Summer senescence (% decline in “green” plant biomass from the peak) increased by 119% over our study, which correlated with increasing summer temperature and grass ANPP. Again, specific seasonal, not annual, climate change was more important as summer (July–

Aug) temperature increased more than the average annual value (2.3˚C vs. 0.7˚C). Further- more, senescence is expected to be more intense for the most abundant “cool season” grasses. Finally, increased senescence had no apparent effect on ANPP, as ANPP is largely determined by the ratio of late-May–June precipitation to temperature supporting “cool season” grass growth. Fall re-initiation of plant growth declined by 67.8% over our study, which correlated with increasing summer temperature, i.e., greater summer senescence. Fall plant production now is very low in most years, as it may be more difficult for plants to overcome severe summer senes- cence. This differs from the historic pattern of fall “greening” in this ecosystem [33]. For exam- ple, in 1805, Lewis and Clark mentioned this [79], and 19th and early 20th century settlers [80– 82] relied on this to provide livestock forage after summer senescence. Plant community and characteristic changes. Bluegrasses decreased by 89% and wheat- grasses increased by 143% over our study, as summer senescence intensified. The majority of bluegrass and wheatgrass species in our study are natives that form clumps (cespitose) and can spread vegetatively (rhizomatous). However, bluegrasses are more shallowly rooted than wheatgrasses (~25 vs. 40 cm) and would be more negatively affected by warm and dry condi- tions during summer (less drought tolerant, USDA plant database). The dramatic post-2005 increase in wheatgrasses may be due to a steady rise in summer temperatures (0.05˚C/yr:

P < 0.0003) and senescence (>1%/yr: P < 0.03) until 2005 and then summer temperature (21.3˚C: P < 0.45) and senescence (29%: P < 0.21) stabilize. Species richness increased by 51.7% over our study, as summer senescence intensified. Spe- cies richness is important in grasslands as it strongly affects ecosystem functioning [83], which has been experimentally demonstrated [84]. Cleland et al.’s [85] survey of grassland regions found that richness increased with low precipitation, but we found that it increased with sum- mer temperature, which may be explained if regional low precipitation and high summer tem- perature are positively correlated. However, while richness increased, the Shannon diversity index, which includes richness and evenness of abundance, only weakly increased (26.5%) over our study, which is counter to expectation [85]. Therefore, greater richness must be coun- tered by less evenness, if more drought resistant species are added and become more common, while less drought resistant species are not lost, but become rarer. PLOS ONE Bunchgrass prairie and climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0243496 December 23, 2020 14 / 21