Artificial light at night (ALAN) decreases plant diversity and performance in experimental grassland communities – Data on species biomass and traits

Artificial light at night (ALAN) affects many areas of the world and is increasing globally. To date, there has been limited and inconsistent evidence regarding the consequences of ALAN on plant communities as well as the fitness of their constituent species. ALAN could be beneficial for plants as they need light as an energy source, but they also need darkness for regeneration and growth. We created model communities composed of 16 plant species sown, exposed to a gradient of ALAN ranging from ‘moonlight only’ to conditions like situations typically found directly underneath a streetlamp. We measured plant community composition and its production (biomass), as well as functional traits of three plant species from different functional groups (grasses, herbs, legumes) in two separate harvests. We found that biomass was reduced by 33% in the highest ALAN treatment compared to the control, Shannon diversity decreased by 43% and Evenness by 34% in the first harvest. Some species failed to establish in the second harvest. Specific leaf area, leaf dry matter content and leaf hairiness responded to ALAN. These responses suggest that plant communities will be sensitive to increasing ALAN, and they flag a need for plant conservation activities that consider impending ALAN scenarios.

Methods
The EcoUnits were filled with 1.23 m3 of unsterilised, well-mixed soil from the vicinity of the EcoTron, as we also monitored soil communities in the same experimental setup (see Cesarz et al.). Plant communities comprising 16 plant species were sown into soil on 19 February 2020 (see Table S1). Because the soil was not sterilised, some of the local seed bank was also transferred into our experiment. Plant communities were harvested by clipping aboveground plant biomass (2 cm above topsoil) on June 11, July 3, and August 28 (establishment period), as well as on October 27 and December 8 (measurement period). This harvest regime mimics typical intensive grassland management in central Europe, with short growth phases in between harvest events. 

For this study, we analysed the last two harvests in detail to address temporal variations and accumulated effects of the ALAN treatment (see Table S1; hereafter referred to harvest 1 and 2, respectively). The harvests differed in length: harvest 1 encompassed a time for regrowth of 9 weeks, whereas harvest 2 only encompassed 6 weeks, as this was embedded in a bigger experimental setup. The biomass of one-eighth (0.19 m2) of each EcoUnit (subplot) was separated into species (both sown and not sown as well as ‘unknown’) and then dried to constant weight at 60°C for three days. Plant identification was sometimes not possible when the plants were not fully mature. These species were all clustered as ‘unknown species’, whereas for others only the genus could be determined. Dead biomass was also recorded. Plant functional trait data was collected for one species each per functional group of grasses (B. hordeaceus), non-legume forbs (P. lanceolata) and legumes (T. repens) just before the harvests in October and December. The species were selected based on their frequent occurrence in the EcoUnits. However, not all plant traits were measured on all species and in all EcoUnits. P. lanceolata was originally not sown into the communities but had become one of the dominant species in the EcoUnits by October and was thus selected for our experiment. It was not very abundant by the end of the experiment as it did not regenerate well after the harvest in October. All traits were collected and measured just before the harvest unless stated otherwise. Stretched plant height of three representative individuals per species and EcoUnit was measured using a ruler. Then, ten healthy leaves from at least three manually randomly selected individuals per species and EcoUnit were harvested and transported to the laboratory, where SLA, LDMC, toughness, hairiness and wettability were measured.

All ten leaves were scanned on an Epson Expression 11000 XL scanner and the resulting images were analysed using ImageJ to determine the leaf area. In the case of T. repens, only the lamina was scanned. Leaves were weighed and subsequently dried at 70°C for at least 48h, and dry weight was recorded to calculate SLA (leaf area of fresh leaf/ dry weight) and LDMC (dry weight/ fresh weight). All weights were measured using a precision scale (QUINTIX315_1S, Sartorius Lab Instruments GmbH & Co. KG, Goettingen, Germany). A few days afterwards, the chlorophyll fluorescence measurements and the SPAD values were determined on living plants in the EcoUnits, just before the harvests. The hairiness, or rather the density of trichomes, of the leaves was analysed by counting the hairs from an image taken at 400-fold magnification using a light microscope and focussing on the middle part of the leaf (Ocular 10x/22, Di-Li-2009, Distelkamp-Electronic, Kaiserslautern, Germany) in ImageJ. For that, four of the leaves used in SLA measurements were chosen at random. Hairs were counted on the upper and lower leaf side and then added to make a total for both leaf sites. T. repens did not show any hair on its lamina. The samples of this species were excluded from the subsequent analysis. The leaf thickness was measured with a digital caliper (WEZU Messwerkzeuge Remscheid GmbH, Remscheid, Germany) at the same spot as leaf toughness. For leaf toughness, the puncture resistance was measured using a surgical blade at a speed of 129 mm min-1 on an electric test stand (Sauter GmbH, Wutöschingen, Germany) and the force of the cut was measured with a power meter (FH 50, Sauter GmbH). The leaf toughness was than calculated as the quotient between the puncture resistance and the thickness. The leaf wettability was investigated via measuring the contact angle (CA) of a water droplet and the leaf, where high CA means low wettability. For that, a droplet of 5 µl distilled water was placed on a flat leaf surface for 90 seconds and then photographed (Nikon D5300 with a Sigma DC Objective, Chiyoda, Tokyo, Japan). The CA was then measured using ImageJ. Chlorophyll fluorescence was measured using a PocketPEA device (Hansatech, King’s Lynn, Norfolk, UK). We measured the parameters PIabs as well as plant stress via Fv/Fm after 30 min of dark adaption to ensure a full reduction of the photosystems on three replicate individuals for each EcoUnit and species. These measurements were not performed on P. lanceolata, as not many individuals were abundant after harvesting the leaves for the previous analysis. The SPAD value was measured using a SPAD 502 (Minolta Camera Co., Osaka, Japan) on the same individual. For each individual, three replicate measurements were performed as the values varied within individuals.

夜间人工光（Artificial Light at Night, ALAN）已在全球范围内广泛分布且强度持续攀升。迄今为止，关于ALAN对植物群落及其组成物种适合度的影响，相关研究证据有限且结论不尽一致。ALAN可为植物提供必要的光能作为能量来源，但植物同时也需要黑暗环境以完成更新与生长。我们构建了由16种植物组成的模拟群落，播种后将其暴露于梯度化的ALAN处理中，光照水平覆盖从“仅月光”到典型路灯正下方的环境区间。我们分别在两次采样中测定了植物群落组成、群落生产力（生物量），以及来自3个功能群（禾草类、非豆科草本、豆科植物）的3个物种植株的功能性状。
结果显示，最高强度ALAN处理组的生物量较对照组降低33%；第一次采样中，香农多样性指数下降43%，均匀度指数下降34%。第二次采样中部分物种未能成功定植。比叶面积、叶片干物质含量与叶片毛茸密度均对ALAN产生了响应。上述结果表明植物群落对ALAN强度的增加极为敏感，提示我们需针对未来ALAN扩张场景开展针对性植物保护工作。

## 方法
实验生态单元（EcoUnits）填充了1.23立方米未灭菌且充分混匀的土壤，采自EcoTron实验平台周边区域——我们在同一实验装置中同步监测了土壤群落（详见Cesarz等的研究）。2020年2月19日，我们将包含16个物种植株的植物群落播种至土壤中（详见附表S1）。由于土壤未经过灭菌处理，本地种子库中的部分物种也随之进入本实验。
我们分别于2020年6月11日、7月3日、8月28日（定植期）以及10月27日、12月8日（测定期）通过剪除距表层土壤2cm处的地上生物量的方式完成采样。该采样方案模拟了中欧典型的集约化草地管理模式，采样间隔期内植物生长周期较短。
本研究聚焦于最后两次采样（分别记为采样1和采样2），以分析ALAN处理的时间变异与累积效应（详见附表S1）。两次采样的间隔时长存在差异：采样1的植株再生周期为9周，而采样2仅为6周——这是因为本实验隶属于更大规模的实验框架。
我们从每个实验生态单元的1/8区域（0.19㎡，即小样方）中采集地上生物量，将其按物种分类（包括播种物种、非播种物种以及未知物种），随后置于60℃环境下烘干至恒重（耗时3天）。由于部分植株未完全成熟，无法准确鉴定物种，这类样本均被归为“未知物种”；另有部分样本仅能鉴定到属水平。此外我们还记录了枯落生物量。
我们分别于10月和12月的采样前，针对3个功能群各选取1个物种测定功能性状：禾草类的B. hordeaceus、非豆科草本的P. lanceolata与豆科的T. repens。选取标准为该物种在实验生态单元中出现频率较高。但并非所有功能性状都在所有物种与所有实验生态单元中完成测定。P. lanceolata最初并未被播种至群落中，但至10月已成为实验生态单元中的优势物种之一，因此被纳入本实验；但在10月采样后其再生效果不佳，至实验末期丰度已较低。除另有说明外，所有性状均在采样前完成测定。
我们使用直尺测量每个物种、每个实验生态单元内3株代表性植株的伸展株高。随后从每个物种、每个实验生态单元中至少3株随机选取的个体上采集10片健康叶片，送至实验室测定比叶面积（SLA）、叶片干物质含量（LDMC）、叶片韧性、毛茸密度与润湿性。
将全部10片叶片置于爱普生Expression 11000 XL扫描仪中扫描，使用ImageJ软件分析图像以获取叶面积。对于T. repens，仅扫描其叶身（lamina）。称量叶片鲜重后，将其置于70℃环境下烘干至少48小时，称量干重以计算SLA（鲜叶面积/干重）与LDMC（干重/鲜重）。所有重量测定均使用精密电子天平（QUINTIX315_1S，赛多利斯实验室仪器股份两合公司，德国哥廷根）。
采样前数日，我们在实验生态单元内的活体植株上完成叶绿素荧光测定与SPAD值测定。叶片毛茸密度通过光学显微镜放大400倍拍摄的图像进行计数，计数区域为叶片中部（目镜10x/22，Di-Li-2009，Distelkamp-Electronic，德国凯泽斯劳滕），使用ImageJ软件完成统计。我们随机选取用于SLA测定的4片叶片，分别计数叶片上下表面的毛茸数量并求和。T. repens的叶身未观察到毛茸，因此该物种的相关样本被排除在后续分析之外。我们使用数显卡尺（WEZU测量工具股份公司，德国雷姆沙伊德）在测定叶片韧性的同一位置测量叶片厚度。叶片韧性采用穿刺抗性表征：使用手术刀片以129 mm·min⁻¹的速度在电动测试台（Sauter股份公司，德国武托辛根）上进行穿刺，通过功率计（FH 50，Sauter股份公司）测定切割力，最终叶片韧性以穿刺抗性与叶片厚度的比值计算得到。
叶片润湿性通过测定水滴与叶片表面的接触角（CA）表征：接触角越高，润湿性越低。我们将5μL的蒸馏水滴于平整的叶片表面，静置90秒后使用相机（尼康D5300搭配Sigma DC镜头，日本东京千代田区）拍摄图像，随后使用ImageJ软件计算接触角。叶绿素荧光测定使用PocketPEA设备（Hansatech，英国诺福克郡金林）。我们对每个实验生态单元、每个物种的3株重复个体进行测定，在暗适应30分钟以确保光合系统完全还原后，测定PIabs参数与Fv/Fm以表征植株胁迫程度。由于P. lanceolata在为此前的性状分析采集叶片后个体丰度较低，因此未对其开展此项测定。我们使用SPAD 502叶绿素仪（美能达相机株式会社，日本大阪）对同一植株进行SPAD值测定，每株重复测定3次，因为植株内部的测定值存在差异。