The Restoration Potential of Fynbos Riparian Seed Banks after Alien Clearing

Riparian areas are highly complex systems with varying levels of disturbance that are highly susceptible to invasion by alien plants. Once invaded, riparian areas play a major role in the dispersal and spread of invasive alien plants (IAPs) through the river system and, in some cases, to neighbouring landscapes. Riparian areas have therefore been prioritized by many alien clearing initiatives in South Africa. Current practice for the restoration of cleared areas is minimal and relies mainly on the un-aided recovery of native species from residual individuals and soil stored seed banks. Little research, however, has been done on the effectiveness of this approach or the extent to which riparian seed banks contribute towards community restoration. This study is part of a national research initiative (Targets for Ecosystem Repair in Riparian Ecosystems in Fynbos, Grassland and Savanna Biomes) funded by Department of Water Affairs and Forestry, in collaboration with Working for Water, The Centre for Invasion Biology and the Universities of Cape Town, Stellenbosch, Rhodes and Witwatersrand. The initiative undertook to investigate different restoration techniques on various invaded sites for their cost-effectiveness, efficiency, practicality and conservation integrity. This study has three aims. The first is to determine the composition of seed banks in un-invaded riparian areas within the fynbos biome to be used as a benchmark for future research, restoration grading and other management requirements. The second aim is to determine the composition of seed banks in heavily invaded riparian areas, and thus to assess the impact of invasion on the integrity of the seed banks. The third aim is to evaluate the restoration potential of riparian seed banks following the clearing of invasive alien plants (IAPs). Study sites were selected within four river systems in the south-western part of the Western Cape Province in South Africa: the Berg, Eerste, Molenaars and Wit Rivers. Plots were selected in both invaded (>75% IAP canopy cover; considered "closed" alien stands) and un-invaded (also termed reference, with <25% IAP canopy cover) sections of the river. Replicate plots were established along varying gradients of elevation (mountain stream and foothill) and moisture regimes (dry, wet and transitional bank zones). Soil samples were collected together with above-ground vegetation surveys and comparisons were made. Results from this study confirm those of previous studies that seed banks offer little reference to current aboveground vegetation, but rather offer insight into past vegetation history as well as future vegetation assemblages. Worldwide, many of the species that characteristically form seed banks are early successional species. A community study was done for the seed bank based on the species that germinated and were identifiable at termination of the project (6 months after initiation). Three clusters of species could be identified. One group comprised 32 generalist species that occurred in both reference and invaded sections of the rivers. A second group comprised 39 species associated with invaded sites, and a third group of 40 species that was associated with reference sites. A few sub-community groups were found within both the "reference" and "invaded" community groups which were assumed to be habitat specific. Most species were "pioneer" or relatively-short lived, early-successional species which play a vital role in the initial post-disturbance vegetation cover, and facilitate establishment of later successional species. Seed banks are notoriously variable over space and time, and floristic representation is often biased as a result of differences among species in seed production, dispersal and longevity in the soil. The general consensus is that seeds have an irregular, clustered spatial distribution that is dictated by both biological and environmental factors. Within river systems, the irregular clustering can be exceptionally skewed with the influence of pockets of high sediment deposition along the bank. Environmental factors that were found to significantly skew germination results were the presence of fire, as well as the extent and intensity of invasion (duration and cover). The high level of diversity and abundance in reference Berg River mountain stream seed banks was perceived to be a direct result of a moderate fire frequency (between 8-15 years) and the relatively natural state of the vegetation (i.e. very little invasion). Also, diversity and richness of indigenous species from the Wit and Molenaars Rivers were substantially higher in the invaded samples than the reference samples, probably because both river systems have a long history of invasion and other anthropogenic disturbances which would have an effect on the samples from "reference" sections (i.e. even a 25% presence of IAPs seems adequate enough to alter the composition of the seed bank). Correspondence analyses showed that species had clear affinities towards different levels of "key" riparian environmental variables (fire, invasion and anthropogenic disturbance). Most species were associated with moderate levels of fire frequency, invasion history, and anthropogenic disturbance. Comparisons of seed bank species assemblages between the lateral and longitudinal variables of the rivers offered insights into the habitat requirements of certain fynbos and riparian species. Most significant were the results from bank zone comparisons which showed distinct species groupings along the different moisture bands. As could be expected, riparian species were best represented within the wet bank zones and fynbos species within the dry bank zone, while species characteristic of both zones occurred in the transitional zone, making this seed bank zone the richest in species. Mountain stream sections were richer and more diverse than foothill sections over both invaded and reference samples. This is hypothesised to be linked to lower levels of anthropogenic disturbance experienced in the mountain stream sections. The impact of invasion on the riparian seed bank was most clearly shown through the correspondence analyses for the 20 most frequently occurring species. The seed bank assembly patterns were clearly defined by the state of the river (reference or invaded). Interestingly, this pattern was evident at all three spatial scales; landscape (rivers), reach (mountain stream and foothill sections) and habitat (dry, wet and transitional zones). The reference seed bank assemblage was more tightly grouped, implying that the species were more closely associated with each other and less variable than those of the invaded seed bank assemblages. The species groupings within the invaded seed banks were influenced by variables such as reach and zone, whereas the reference seed bank assemblages seem relatively unaffected by these variables. This implies that the presence of invasive alien plants creates additional variation within the seed bank which alters the natural groupings. At a broad scale, the invaded seed banks were less species rich. This means that not only will the resulting seedling community be harder to predict, but it will also have fewer species. However although generally lower in species richness, the seed banks from almost all invaded rivers interestingly showed a higher diversity of indigenous species than their reference counterparts. This is very promising in terms of rehabilitation of post-cleared riparian sites, but more information is needed to understand the seed bank composition and determine how sustainable the seed banks are for rehabilitation in the long-term. All invaded sections had fewer herbaceous perennial species but more herbaceous annual species. Graminoids made up 50% or more of the seed bank regardless of state (reference or invaded), while woody species (shrubs/shrublets) were generally more prevalent in the reference samples. These results imply that following the removal of invasive alien plants, the vegetation to regenerate from the seed bank is likely to comprise of short-lived, herbaceous species that are not necessarily an accurate reflection of the indigenous riparian community. It is however important to note that this study investigated only the species that were able to germinate over the study period (6 month germination period). Many riparian species may not have been represented because they are either late germinators or may not be present in the soil seed bank at all. In order to gain a holistic understanding of riparian community recruitment, it is recommended that seed bank studies such as this one be included in a more broad scale, long term investigation which takes into account various reproductive strategies used by riparian species. Research of this nature is in its infancy worldwide and there are many challenges involved in measuring diversity and change in these systems. However, within the scope of this study, I suggest that these results shed light on previously unanswered and important questions regarding the ecology of seed banks in the riparian ecosystems of the Western Cape.河岸带（Riparian areas）是高度复杂的生态系统，受干扰程度各不相同，且极易遭受外来植物入侵。一旦被入侵，河岸带会在入侵外来植物（Invasive Alien Plants, IAPs）通过河流系统扩散、乃至蔓延至周边景观的过程中发挥关键作用。因此，南非的诸多外来植物清除计划均将河岸带列为优先整治区域。当前针对已清除入侵植物区域的修复实践极为有限，主要依赖本土物种通过残存个体与土壤种子库（soil stored seed banks）自然恢复。然而，关于该修复方法的有效性，以及河岸带土壤种子库对群落修复的贡献程度，相关研究仍寥寥无几。本研究属于一项全国性研究计划——“丰伯斯生物群落（Fynbos biome）、草原与稀树草原生物群落河岸生态系统修复目标（Targets for Ecosystem Repair in Riparian Ecosystems in Fynbos, Grassland and Savanna Biomes）”，该计划由水利与林业部资助，与“为水而工作”（Working for Water）、入侵生物学中心以及开普敦大学、斯泰伦博斯大学、罗德斯大学和金山大学合作开展。此项计划旨在针对不同入侵位点的多种修复技术，从成本效益、效率、实用性与保护完整性层面展开研究。本研究包含三大目标：其一，明确丰伯斯生物群落内未受入侵的河岸带土壤种子库组成，以此作为未来研究、修复分级及其他管理需求的基准；其二，确定重度入侵河岸带的土壤种子库组成，以此评估入侵对种子库完整性的影响；其三，评估清除入侵外来植物后，河岸带土壤种子库的修复潜力。研究位点选取于南非西开普省西南部的四条河流系统：贝尔赫河（Berg River）、埃尔斯特河（Eerste River）、莫伦阿尔斯河（Molenaars River）与威特河（Wit Rivers）。在河流的入侵（入侵外来植物冠层覆盖度＞75%，被视为“封闭”的外来植物群落）与未入侵（亦称为参照组，入侵外来植物冠层覆盖度＜25%）区域均设置样地。沿海拔梯度（山溪与山麓）与水分梯度（旱岸、湿岸与过渡岸带）设置重复样地。采集土壤样本的同时开展地上植被调查，并对二者进行对比分析。本研究结果证实了既往研究的结论：土壤种子库对当前地上植被的参考价值有限，却能为植被历史与未来植被组合提供见解。全球范围内，典型的土壤种子库形成物种多为早期演替物种。本研究基于项目启动后6个月的培养期内萌发并可鉴定的物种，开展了种子库群落研究，共划分出3个物种集群：第一组包含32种广适性物种，同时存在于河流的参照组与入侵组区域；第二组包含39种与入侵位点相关的物种；第三组包含40种与参照位点相关的物种。在“参照组”与“入侵组”群落中均发现了少量亚群落集群，推测其与生境特异性相关。多数物种为“先锋”或相对短命的早期演替物种，它们在受干扰后的初始植被覆盖阶段发挥关键作用，并可促进后续演替物种的定植。众所周知，土壤种子库的空间与时间变异性极强，且由于物种在种子产量、扩散能力与土壤种子寿命的差异，植物区系的代表性往往存在偏差。学界普遍认为，种子的空间分布呈不规则聚集模式，这一模式由生物与环境因素共同决定。在河流系统中，这种不规则聚集模式可能因河岸带高沉积物沉积区的影响而极度倾斜。研究发现，显著影响萌发结果的环境因素包括火灾发生情况，以及入侵的程度与强度（入侵持续时间与冠层覆盖度）。参照组贝尔赫河山溪种子库的高多样性与丰富度，被认为是中等火灾频率（8~15年一次）与植被相对自然状态（即几乎无入侵）的直接结果。此外，威特河与莫伦阿尔斯河的本土物种多样性与丰富度，在入侵组样本中显著高于参照组样本，这可能是因为这两条河流均存在长期入侵历史与其他人为干扰，进而影响了“参照组”区域的样本（即即便入侵外来植物冠层覆盖度仅为25%，似乎也足以改变种子库的组成）。对应分析（correspondence analyses）结果显示，物种与不同水平的“关键”河岸带环境变量（火灾、入侵与人为干扰）存在明确的关联性。多数物种与中等水平的火灾频率、入侵历史及人为干扰相关。对比河流横向与纵向变量下的种子库物种组合，可为部分丰伯斯物种与河岸物种的生境需求提供见解。其中最显著的结果来自岸带对比：不同水分梯度下存在明显的物种集群。正如预期，河岸物种在湿岸带的代表性最佳，丰伯斯物种则集中于旱岸带，而同时适应两类生境的物种分布于过渡岸带，使得该岸带的种子库物种丰富度最高。无论入侵组还是参照组样本，山溪区域的物种丰富度与多样性均高于山麓区域，推测这与山溪区域受到的人为干扰更少有关。入侵对河岸带种子库的影响，可通过20种最常见物种的对应分析结果得到清晰体现。种子库的组合模式由河流状态（参照组或入侵组）明确界定。有趣的是，这一模式在三个空间尺度上均成立：景观尺度（河流）、河段尺度（山溪与山麓区域）与生境尺度（旱岸、湿岸与过渡岸带）。参照组种子库的物种组合更为紧密，意味着其物种间的关联性更强，且变异性低于入侵组种子库。入侵组种子库内的物种集群受河段与岸带变量的影响，而参照组种子库的物种组合似乎不受这些变量的显著影响。这表明，入侵外来植物的存在会为种子库带来额外变异，进而改变自然群落组合。从大尺度来看，入侵组种子库的物种丰富度更低。这意味着不仅由此产生的幼苗群落更难预测，且其物种数量也更少。然而，尽管整体物种丰富度较低，几乎所有入侵组河流的种子库均展现出比参照组更高的本土物种多样性，这一点对于已清除入侵植物的河岸带修复而言极具前景，但仍需更多研究以理解种子库组成，并明确长期修复中种子库的可持续性。所有入侵区域的多年生草本物种占比更低，而一年生草本物种占比更高。无论参照组还是入侵组，禾本科植物（graminoids）均占种子库组成的50%及以上；而木本物种（woody species，灌木/小灌木）在参照组样本中更为常见。上述结果表明，清除入侵外来植物后，从种子库中萌发的植被大概率以短命的草本物种为主，未必能准确反映本土河岸群落的特征。但需注意的是，本研究仅调查了在培养期（6个月）内能够萌发的物种，诸多河岸物种可能因萌发较晚，或根本不存在于土壤种子库中，而未被本研究纳入统计。为全面理解河岸带群落的自然更新过程，建议将此类种子库研究纳入更具广度的长期调查中，同时考虑河岸物种的多种繁殖策略。此类研究在全球范围内尚处于起步阶段，对这类生态系统的多样性与变化进行监测也面临诸多挑战。但就本研究的范畴而言，本研究结果为西开普省河岸生态系统中种子库生态学领域此前未被解答的重要问题提供了新的见解。