Short-lived herbaceous plants provide a useful model to rapidly reveal how multiple generations of plants in natural plant communities of sensitive desert ecosystems will be affected by N deposition. We monitored dynamic responses of community structure, richness, evenness, density and biomass of herbaceous plants to experimental N addition (2:1 NH4+:NO3− added at 0, 0.5, 1, 3, 6 and 24 g N m− 2 a− 1) in three seasons in each of three years in the Gurbantunggut desert, a typical temperate desert of central Asia. We found clear rate-dependent and season-dependent effects of N deposition on each of these variables, in most cases becoming more obvious through time. N addition reduced plant richness, leading to a loss of about half of the species after three generations in the highest N application level. Evenness and density were relatively insensitive to all but the greatest levels of N addition for two generations, but negative effects emerged in the third generation. Biomass, both above and below ground, was non-linearly affected by N deposition. Low and intermediate levels of N deposition often increased biomass, whereas the highest level suppressed biomass. Stimulatory effects of intermediate N addition disappeared in the third generation. All of these responses are strongly interrelated in a cascade of changes. Notably, changes in biomass due to N deposition were mediated by declines in richness and evenness, and other changes in community structure, rather than solely being the direct outcome of release from limitation. The interrelationships between N deposition and the different plant community attributes change not only seasonally, but also progressively change through time. These temporal changes appear to be largely independent of interannual or seasonal climatic conditions.
Properties of annual plant communities subjected to N addition
In October 2008, sixty 8×8m plots were randomly placed across an experimental area of 5,400 m2 in the center of the Gurbantunggut desert (44.876 N, 87.823 E), with an average separation distance of about 10 m. The plots had similar plant community composition and structure before the N treatments. From year 2008 to 2011, five N concentrations plus one control (without N) were randomly applied on the plots, totaling 10 replicates of each of the six concentrations. The rates of six N treatments were 0, 0.5, 1, 3, 6 and 24 g N m-2 a-1 (hereafter denoted as N0, N0.5, N1, N3, N6 and N24, respectively). The N treatments were applied in two equal pulses per year in March after snow thaw and October before snowfall every year, coinciding reasonably with a fall-spring pulses associated with rainy seasons, and with fertilization (and thus deposition) pulses in agricultural areas of the region. Each applied N treatment consisted of 2:1 molar ratio of NH4+: NO3- (as NH4NO3 and NH4Cl), in 3 L of water per plot (about 0.037 mm of rainfall equivalent) applied using a spray. Controls received an equivalent amount of water only. The first treatment began in October 2008 and were repeated every year after that to simulate long-term effects of N deposition. In each plot, we established one 1× 1 m permanent quadrat for the investigation of community composition and structure. The richness (number of species per plot) and density (number of individuals per plot) were measured in mid-spring (April), late spring (May – June), and summer (July - August) in each year after N treatments. These seasons were chosen because various life forms of plants reach their peak biomass during these periods. We calculated the evenness index J’ based on the number of individuals per species (Tuomisto, 2012).
Peak aboveground biomass provides a good estimate of annual aboveground production in communities dominated by annual plants (Sala et al., 1988). We measured production three times each year, concurrent with community measures. We selected one 0.5×0.5 m quadrat (far from the permanent quadrat for community investigation) in each 8×8 m plot for the measurement of aboveground and belowground biomass. All plants from the quadrat were collected using spades. We separated shoot and root portions for each species in the lab (after washing adherent sand from roots using tap water), and measured the biomass after drying at 70 oC for 24 hours in the oven to obtain consistent weight. The aboveground biomass was the total shoot biomass of all annual plants (including ephemeroid plants). The belowground biomass was the total root biomass of all annual plants excluding ephemeroid plants. The ephemeroid plants were excluded because their roots were cloned together (two or more ramets fused together) and exist for several years. The root biomass of ephemeroids was significantly higher than other annual plants and significantly differed among quadrants, even before the start of the experiment. In order to use community structure as a variable in some analyses, data reduction was necessary. As a data reduction tool, we used nonmetric multidimensional scaling of total above and belowground biomass by species, based on the Bray-Curtis distance measure (McCune and Grace, 2002). Prior to ordination, we omitted any species that were present in fewer than 3 samples to reduce noise and omitted any samples that lacked any plant biomass because empty samples are incompatible with this distance measure. After these modifications, we performed a general relativization, rescaling the abundance of all species within a sample such that they summed to 1. We obtained a two-axes ordination and rotated it so that Axis 1 correlated with season, the apparent strongest driver of composition. We saved axis scores for each sample for use in our structural equation model.