Species Richness Interacts with Drought to Affect Litter Decomposition via its Effect on Litter Nitrogen Concentration

19 Biodiversity loss, exotic plant invasions and climatic change are currently the three major challenges to our 20 globe and can each affect various ecological processes, including litter composition. To gain a better 21 understanding of global change impacts on ecological processes, these three global change components 22 need to be considered simultaneously. Here we assembled experimental plant communities with species 23 richness levels (1, 2, 4, 8 or 16) and subjected them to drought (no, moderate or intensive drought) and 24 invasion (invasion by the exotic annual plant Symphyotrichum subulatum or not). We collected litter of the 25 native plant communities and let it decompose for nine months within the communities. Drought decreased 26 litter decomposition, while the exotic plant invasion had no impact. Increasing species richness decreased 27 litter decomposition under the mesic condition (no drought), but had little impact under moderate and 28 intensive drought. A structural equation model showed that drought and species richness affected litter 29 decomposition mainly via influencing litter nitrogen concentration, but not via altering the quantity and 30 diversity of soil meso-fauna or soil physio-chemical properties. The negative impact of species diversity on 31 litter decomposition under the mesic condition was mainly ascribed to a sampling effect, i.e. via 32 particularly low litter nitrogen concentrations in the two dominant species. Our results indicate that species 33 richness can interact with drought to affect litter decomposition via effect on litter nitrogen. We conclude that nitrogen-dependent litter decomposition should be a mechanism to predict integrated effects of plant 35 diversity loss, exotic plant invasions and climatic change on litter decomposition.


Introduction 38
Climatic change, biodiversity loss and biological invasions are currently the three major challenges to the 39 health of our globe (Venezia et  seedlings were retained and excess seedlings were removed. Thus, at the start of the experiment, plant 127 density was maintained at 32 seedlings per container, and each species was represented by the same 128 number of seedlings (e.g. for four-species mixtures, there were eight seedlings for each of the four species 129 in a container). The 32 seedlings were spatially evenly distributed in the container, and seedlings of the 130 same species were not adjacent. In each container, we also removed undesired seedlings, i.e. those not 131 belonging to the originally sown species. All the containers were randomly placed inside a plastic rain 132 shelter in Taizhou University, Zhejiang Province, China, which was open at the bottom sides to allow air to 133 be ventilated. 134

Drought manipulation 135
Using automatic drip irrigation systems, we set up three drought intensity treatments (no, moderate and intensive drought, the irrigated time was 50% (gravimetric soil water content ranging from 12.4-15.4%) 147 8 and 25% (gravimetric soil water content ranging from 10.0-12.6%) of that in the no drought treatment, 148 respectively. Depending on the weather conditions, plant communities in the containers were irrigated once 149 a day between May and September, once every other day between March and April and between October 150 and December, and once every week between January and February. 151 The six replicates of the 50 communities of different richness levels were randomly assigned to one of 152 the three drought treatments, so that each drought treatment had two replicates. The drought treatments 153 started on 12-Mar-2015. 154

Invasive species and invasion introduction 155
Symphyotrichum subulatum (Michx.) G. L. Nesom is an annual herb of Asteraceae. It is native to South 156 America, but now widespread in warm regions of the world (Zhuge et al. 2011). In China, S. subulatum is 157 listed as an invasive exotic plant species as it has invaded many areas (Zhuge et al. 2011). This species 158 often forms a mono-dominant community and displaces native plant species. It grows 16-150 cm tall with 159 erect stems. It reproduces sexually and each plant can produce profuse viable seeds. Seeds are dispersed by 160 wind. In Dec-2015, for half (150) of the experimental communities, 50 seeds of S. subulatum were evenly 161 sown in each container. For the other half, no seeds of S. subulatum were sown. 162

Litter decomposition experiment 163
In Nov-2015, about one month before the invasion treatment was started and when most leaves shed, we 164 collected freshly produced litter (mainly leaves) of each community (in each container). Then we 165 thoroughly mixed the litter collected from the two communities with exactly the same species composition 166 and subjected to the same drought treatment, i.e. one was later used for the invasion treatment and the other 167 for the corresponding control treatment (no invasion). The mixed litter (about 40 g) was treated as one litter 168 sample. The litter samples were cleaned and dried at ca. 40 o C in the oven. The litter samples were not 9 air-dried to avoid pretreatment decomposition as the air at this time was quite humid. 170 For each litter sample, a subsample of 3 g was ground and used for analyzing initial concentration of 171 total nitrogen (Autoanalyzer 3, BRAN+LUEBBE, Germany, measured with three replicates). Ten 172 subsamples, each about 2 g (range: 1.997-2.003 g), of each litter sample were placed in ten plastic 173 litterbags (6 cm × 8 cm) with a mesh size of 2 mm, which permits entry of both microfauna and mesofauna 174 (Smith and Bradford 2003). Larger leaf litter was cut into pieces of 2 cm × 2 cm before being placed into 175 the litterbags. 176 In Dec-2015, we placed the ten litterbags of each litter sample into the litter layers of the two 177 communities where it had been collected, with five litterbags in the community invaded by the exotic 178 species (A. subulatus) and the other five in the corresponding community not invaded by the exotic species. 179 Caution was taken not to cause excessive disturbance during litterbag placement. 180 Three litterbags in each community were collected three, six and nine months after the litterbags had 181 been placed into the communities. After cleaning, the litter from each litterbag was dried in the oven at 182   Each k value was based on litter mass of the four sampling times (0, 3, 6 and 9 months). 206 We used ANOVA to assess the effects of species richness, drought, invasion and their interactions on 207 k. In this model, species richness, drought and invasion were treated as fixed factors and species 208 composition was included as a random factor nested within species richness. We also used linear 209 regressions to explore the relationships of k with species richness in each of the six treatment combinations 210 of drought and invasion. 211 We employed linear regressions to explore the relationships between k and litter initial N 212 concentration in each of the six treatment combinations of drought and invasion. We also used linear 213 11 regressions to examine the relationships between litter initial N concentration and species richness in each 214 of the three drought treatments (invasion and un-invasion treatments were not considered separately as the 215 litter was a mixture from the invaded and uninvaded plot of the same species composition so that the initial 216 litter composition of these two plots were the same). 217 A functional diversity index was also calculated following the method of Rao (1982) using the 218 Euclidean distance based on litter N and P and abundance (aboveground biomass) of each species. 219 However, there were no significant relationships between the functional diversity index and litter 220 decomposition (Fig. S3).  (Table S4). Drought, species 226 richness and invasion were modeled as treatment variables, and number of mesofauna individuals, number 227 of mesofauna species, soil microbial activity, litter N, community aboveground biomass, soil temperature 228 and soil pH were modeled as response variables whose effects on k were mediated by these treatments. 229 Because data on soil microbial activity was only available for the monocultures and the 16-specis mixtures 230 (total n = 120) and random numbers that have the same mean and standard deviation of the measured 231  Results 245

Effects of drought, invasion and species richness on litter decomposition 246
Drought significantly affected litter decomposition rate (k; Table 1). Average across species richness and 247 invasion treatments, k decreased from 0.148 with no drought, via 0.103 under moderate drought, to 0.080 248 under intensive drought (Fig. 1). There was a significant interaction of drought × species richness on k 249 (Table 1): species richness showed a negative relationship with k under no-drought treatment, but not under 250 moderate or intensive drought treatments (Fig. 1). Invasion had no significant effect on k (Table 1). There 251 were no further two-way or three-way interactions on k. 252

Relationships of litter nitrogen with litter decomposition and species richness 253
Litter N concentration had a significant positive relationship with k value in all the six treatment 254 combinations of drought and invasion (r = 0.308 -0.683, all P < 0.05, n = 49; Fig. 2). Also, litter N 255 concentration was significantly negatively related to species richness under no drought (r = -0.411, n = 49, 256 P = 0.003; Fig. 2), but had no significant relationship with species richness under moderate or intensive 257 13 drought (Fig. 2). 258

Direct and indirect effects on litter decomposition 259
The structural equation model fitted the data well (GFI, P = 0.986; Maximum likelihood χ 2 = 22.664, P = 260 0.161; RMSEA = 0.034, P = 0.759; Fig. 3, Table S4). The parameters included in the model explained 41.1% 261 of the variation in litter decomposition (Fig. 3). 262 The strongest predictors of k were drought (negative) and litter nitrogen (positive; Fig. 3). Drought 263 could directly affect k and could also indirectly affect k via its direct effect on litter N concentration and soil 264 microbial activity (Fig. 3). While drought could also directly affect community aboveground biomass, 265 number of mesofauna individuals, number of mesofanua species, soil temperature and pH, it could not 266 indirectly affect k via influencing these biotic and abiotic factors (Fig. 3). 267 Species richness could not directly affect k, but could indirectly affect k via altering litter N 268 concentration (Fig. 3). Species richness could directly positively affect community aboveground biomass 269 and number of mesofauna species, but could not indirectly affect k via influencing these two factors (Fig. 3). 270 Invasion had neither direct nor indirect effect on litter decomposition, although it could directly affect 271 number of mesofauna species (Fig. 3). 272

Effects of dominant species on litter nitrogen 273
The presence of P. scabiosaefolia and/or A. migoana significantly decreased litter N concentration of the 274 whole communities under no drought (Fig. 4) and led to lower litter decomposition rate (Fig. 5A, D). As 275 these two species occurred more often in more diverse communities, litter decomposition rate of these more 276 diverse communities had lower values and lower variance (Fig. 5A, D), suggesting a significant sampling disappeared from moderate to intensive drought (Fig. 4). Accordingly, the presence of P. scabiosaefolia 281 and/or A. migoana had no effect (Fig. 5B, E, F) and even increased the litter decomposition rate somewhat 282 (Fig. 5C), which led to the negative relationship between species richness and litter decomposition rate 283 disappearing under moderate and intensive drought. 284

Discussion 285
We tested the effects of three major global change factors (plant species diversity, exotic plant invasion and 286 drought) on litter decomposition of native plant communities. Our results revealed that plant species 287 richness interacted with drought to affect litter decomposition via influencing litter N concentration, and 288 that drought could decrease litter decomposition also via its impact on litter quality (particularly litter N) 289 and soil microbial activity. However, we found no effect of exotic plant invasion on litter decomposition. increasing species richness decreased the decomposition rate of litter of native plant communities, but such 295 a diversity effect disappeared under moderate or intensive drought (Fig. 1). The diversity effect on litter 296 decomposition corresponded well to the diversity effect on litter initial N concentration (Fig. 2), and also 297 litter decomposition rate was significantly positively correlated with litter initial N concentration (Fig. 2). microenvironments. In our study, species richness was also found to promote soil fauna diversity (Fig.  336 3), but it influenced litter decomposition mainly via influencing litter nitrogen rather than modifying 337 microenvironments (promoting soil fauna diversity). However, it is not inconceivable that, as for the 338 litter quality effects reported above, a relatively small number of detritivorous species could have a 339 disproportionately large effect of decomposition. Such effects may not be captured by mesofauna 340 richness or abundance per se. directly decreased litter decomposition (Fig. 3, Table S4b), which may be related to the low soil 347 moisture and associated lower microbial activity in the drought treatment (Zhang and Zak 1995;Walse 348 et al. 1998). 349 Apart from the direct effect, drought also indirectly affected litter decomposition of native plant 350 communities mainly via its impact on litter N concentration and soil microbial activity (Fig. 3). Thus, 351 the drought-mediated changes in litter quality and soil microbial activity are also two of the 352 mechanisms underlying the drought effect on litter decomposition. We also found that drought 353 significantly decreased quantity and diversity of soil mesofauna. However, different from our 354 expectation (Table S4a)

Effects of exotic plant invasions on litter decomposition 364
Different from our expectation (Table S4a), we observed no effect of invasion by the exotic annual forb S. 365 subulatum on litter decomposition of native plant communities. In this study, we mimicked the initial stage 366

Conclusions 384
No matter to which combination of drought intensity and invasion treatments our experimental plant 385 communities were exposed, nitrogen-dependent litter decomposition was always found in plant 386 communities irrespective of species richness. The effect of plant richness on litter decomposition changed 387 with drought intensity, which could be ascribed to a change in the sampling effect on litter nitrogen in 388 response to the increase in drought intensity. While nitrogen was the main limiting element in our system, 389 and in most ecosystems in the world, we need to be aware that phosphorous is the main limiting element in 390 several other ecosystems such as some wetlands (Saaltink et  Funding (Not applicable) 403 Conflicts of interest (The authors declare that they have no conflict of interest.) 404

Ethics approval (Not applicable) 405
Consent to participate (Not applicable) 406

Consent for publication (Not applicable) 407
Availability of data and material (The datasets used and/or analysed during the current study are 408 available from the corresponding author on reasonable request) 409 Code availability (Not applicable) 410 Authors' contributions 411  Results of t-tests and the significance levels ( ** P < 0.01; ns P > 0.05) are also given. 665 blue, purple and white rectangles respectively denote the plots with P. scabiosaefolia, A. migoana, 668 both P. scabiosaefolia and A. migoana, and neither P. scabiosaefolia nor A. migoana, respectively. 669 The size gradient of scatter points denotes litter nitrogen level. Results of t-tests and the 670 significance levels (** P < 0.01; ns P > 0.05) are also given.      Litter nitrogen concentration in plots in the presence or absence of the dominant species (Patrinia scabiosaefolia and/or Artemisia migoana) under no, moderate and intensive drought. Results of t-tests and the signi cance levels (** P < 0.01; ns P > 0.05) are also given.

Figure 5
Litter decomposition rate (k) in plots uninvaded or invaded by the exotic species Symphyotrichum subulatum under (A, D) no, (B, E) moderate or (C, F) intensive drought. Red, blue, purple and white rectangles respectively denote the plots with P. scabiosaefolia, A. migoana, both P. scabiosaefolia and A. migoana, and neither P. scabiosaefolia nor A. migoana, respectively. The size gradient of scatter points denotes litter nitrogen level. Results of t-tests and the signi cance levels (** P < 0.01; ns P > 0.05) are also given.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. supplementarymaterials.pdf