1 INTRODUCTION

Understanding variation in the relative importance of deterministic and stochastic processes in community assembly is an ongoing focus of research in community ecology (Shipley et al., 2012; Weiher & Keddy, 1995). Deterministic processes are species interactions and ‘environmental filtering’ (selection imposed by the abiotic environment), whereas stochastic processes are dispersal, disturbances and births and deaths (Chase & Myers, 2011). It has been suggested that the relative importance of each process type on community assembly varies for different taxa and across spatial and temporal scales (Chase, 2014; Chu et al., 2007; Farjalla et al., 2012). Consequently, temporal community dynamics emerging from deterministic and stochastic processes are likely to be different and thus provide a signature of the underlying processes (e.g. Maren et al., 2018; Stegen et al., 2013).

Vascular epiphytes are sessile organisms living non-parasitically on three-dimensional dynamic patches (i.e. host plants, Zotz, 2016). From an epiphyte perspective, newly establishing trees represent a new substrate sensu Platt and Connell (2003) in the context of natural disturbance and community assembly. As hosts grow, changes in their characteristics influence epiphyte assemblages development (Taylor & Burns, 2015). The production and ageing of a new substrate increases the habitable area and habitat heterogeneity (Zotz, 2016), which will increase species richness (Buckley, 2011; Flores-Palacios & Garcia-Franco, 2006; Hortal et al., 2009; Taylor & Burns, 2015) and abundance (Ockinger & Nilsson, 2010; Snäll et al., 2004; Spruch et al., 2019). The environmental heterogeneity (Stein et al., 2014) caused by structural enrichment (Zotz & Vollrath, 2003) resembles that created by a founder or pioneer species in the context of primary succession in terrestrial groups, thus rendering the differentiation of the ecological processes behind the observed dynamics challenging. Directional replacement of species is expected on new substrates; that is, an early successional species is replaced by late-successional species (see references within Platt & Connell, 2003), as in primary succession. In the case of vascular epiphytes, it is contested whether the concept of typical primary succession appropriately captures the nature of the ecological processes behind epiphyte assemblage dynamics (Zotz, 2016). Instead, it may rather be conceptualised as a facilitation cascade via successive habitat formation (during host ontogeny) and modification (Thomsen et al., 2010, 2018).

In contrast to ground-rooted plants, there is limited understanding of the processes driving vascular epiphyte assemblages. The composition of vascular epiphyte assemblages is assumed to be mainly defined by abiotic factors, for example, vegetative growth and spatial distribution are mainly limited by water availability as the major cause of mortality (Ding et al., 2016; Laube & Zotz, 2003; Olaya-Arenas et al., 2011; Rascher et al., 2012; Zuleta et al., 2016). However, the relative role of deterministic and stochastic processes in vascular epiphyte assemblage composition has yet to be established.

The relative importance of the processes driving community assembly may differ between epiphytes and ground-rooted plants (Zotz, 2016). This conjecture is primarily based on the structural dependency of epiphytes; they are physically dependent on host plants. Host tree species provide patches with distinct conditions for the development of vascular epiphyte assemblages (Wagner et al., 2015). Hosts can affect epiphyte assemblages via differences in architecture (Cardelus & Chazdon, 2005; Einzmann et al., 2014), physical and chemical bark characteristics (Benzing & Renfrow, 1974; Callaway et al., 2002), growth rate (Flores-Palacios & Garcia-Franco, 2006; Zotz & Vollrath, 2003), and canopy dynamics (Sarmento Cabral et al., 2015; Wagner & Zotz, 2020). Moreover, fundamental characteristics of these ‘islands’, such as substrate dynamics (Fahrig, 1992; Spruch et al., 2019; Taylor & Burns, 2015), change with time (Fahrig, 1992; Snäll et al., 2005). Thus, host age affects species richness, composition, abundance (Lie et al., 2009; Wagner & Zotz, 2020) as well as dynamics (Fedrowitz et al., 2012) of vascular and non-vascular epiphytes.

The absence of a species from a particular habitat may be due to lack of dispersal, habitat unsuitability (environmental filtering) and negative biotic interactions (Cadotte & Tucker, 2017). In general, negative biotic interactions (competition, herbivore pressure) are thought to play a minimal role in vascular epiphyte assembly (Burns & Zotz, 2010; Taylor et al., 2016; Zotz, 2016). Kraft et al. (2015) pointed out that a strict definition of environmental filtering refers to cases where the abiotic environment prevents establishment or persistence in the absence of biotic interactions; therefore, epiphytes are interesting model organisms to understand the prevalence of environmental filtering in shaping community assembly.

Environmental filtering is most often inferred from the presence of non-random assemblage patterns (Cadotte & Tucker, 2017). For epiphytes, the increase in mean pair-wise compositional similarity over time for epiphyte assemblages on two different host species in lowland forests (Laube & Zotz, 2006a, 2007) provides tentative evidence for directional changes in assemblage composition (Mendieta-Leiva & Zotz, 2015). Using a unique new dataset, comprising consecutive epiphyte censuses from an old-growth lowland forest in Panama covering more than a decade, we investigate the relative importance of niche and neutral processes for assembly by examining species compositional changes, dispersal limitation, and the potential effect of host tree identity and host size. We use tree individuals as the sampling unit, treating them analogous to plots in terrestrial vegetation ecology. Trees are discrete habitable patches with delimited borders (Southwood & Kennedy, 1983) and can be conceptualised as habitat islands for epiphytes (Ellis, 2012; Taylor & Burns, 2015) with their own dynamics (Wagner & Zotz, 2020). Assuming dominance of neutral processes, we predict that (a) species composition of vascular epiphytes should not exhibit ‘directional’ change over time; (b) epiphyte species composition of epiphyte assemblages does not change with host tree species and host size; (c) differences in epiphyte species composition are related to the distances among trees, indicating the effects of dispersal limitation. In contrast, if niche-based mechanisms dominate, we predict directionality in changes of species composition and that differences in epiphyte species composition are related to differences in host plant properties but not to among host tree distance.

中文翻译：

---

热带维管附生植物组合物种组成随时间的方向变化

1 简介

了解群落组装中确定性和随机过程的相对重要性的变化是群落生态学研究的一个持续重点（Shipley 等人，  2012 年；Weiher & Keddy，  1995 年）。确定性过程是物种相互作用和“环境过滤”（非生物环境强加的选择），而随机过程是扩散、干扰和出生和死亡（Chase & Myers,  2011）。有人提出，每种过程类型对群落组装的相对重要性因不同的分类群和空间和时间尺度而异（Chase，  2014；Chu et al.，  2007；Farjalla et al.，  2012）。因此，确定性和随机过程产生的时间群落动态可能不同，因此提供了潜在过程的特征（例如 Maren 等人，  2018 年；Stegen 等人，  2013 年）。

维管附生植物是非寄生在三维动态斑块上的固着生物（即宿主植物，Zotz，  2016 年）。从附生植物的角度来看，在自然干扰和群落集结的背景下，新建立的树木代表了一种新的基质意义 Platt 和 Connell ( 2003 )。随着宿主的生长，其特征的变化会影响附生植物组合的发育（Taylor & Burns，  2015 年）。新基质的产生和老化增加了可居住面积和栖息地异质性（Zotz，  2016 年），这将增加物种丰富度（Buckley，  2011 年；Flores-Palacios & Garcia-Franco，  2006 年；Hortal 等人，  2009 年）; Taylor & Burns，  2015 年）和丰度（Ockinger & Nilsson，  2010 年；Snäll 等人，  2004 年；Spruch 等人，  2019 年）。由结构富集（Zotz & Vollrath,  2003 ）引起的环境异质性（Stein et al.,  2014 ）类似于在陆地群体的初级演替背景下由创始人或先驱物种创造的环境异质性，从而导致背后生态过程的分化观察到的动态具有挑战性。预计在新基板上进行物种的定向替换；也就是说，早期演替物种被晚期演替物种所取代（参见 Platt & Connell,  2003中的参考文献）)，如在主要继承中。在维管附生植物的情况下，典型的初级演替的概念是否恰当地捕捉了附生植物组合动力学背后的生态过程的性质存在争议（Zotz，  2016 年）。相反，它可能被概念化为通过连续的栖息地形成（在宿主个体发育期间）和修改（Thomsen et al.,  2010 , 2018）的促进级联。

In contrast to ground-rooted plants, there is limited understanding of the processes driving vascular epiphyte assemblages. The composition of vascular epiphyte assemblages is assumed to be mainly defined by abiotic factors, for example, vegetative growth and spatial distribution are mainly limited by water availability as the major cause of mortality (Ding et al., 2016; Laube & Zotz, 2003; Olaya-Arenas et al., 2011; Rascher et al., 2012; Zuleta et al., 2016). However, the relative role of deterministic and stochastic processes in vascular epiphyte assemblage composition has yet to be established.

驱动群落组装过程的相对重要性在附生植物和地根植物之间可能不同（Zotz，  2016）。这个猜想主要是基于附生植物的结构依赖性；它们在身体上依赖寄主植物。寄主树种为维管附生植物组合的发育提供了具有独特条件的斑块（Wagner et al.,  2015）。寄主可以通过结构差异（Cardelus & Chazdon,  2005 ; Einzmann et al.,  2014）、物理和化学树皮特征（Benzing & Renfrow,  1974 ; Callaway et al.,  2002）、生长速率（Flores-Palacios & 加西亚-佛朗哥，  2006; Zotz 和 Vollrath，  2003 年）和冠层动力学（Sarmento Cabral 等人，  2015 年；Wagner 和 Zotz，  2020 年）。此外，这些“岛屿”的基本特征，例如底物动力学 (Fahrig,  1992 ; Spruch et al.,  2019 ; Taylor & Burns,  2015 )，会随着时间而变化 (Fahrig,  1992 ; Snäll et al.,  2005 )。因此，宿主年龄会影响维管和非维管附生植物的物种丰富度、组成、丰度（Lie 等人，  2009 年；Wagner & Zotz，  2020 年）以及动态（Fedrowitz 等人，  2012 年）。

来自特定栖息地的物种的缺失可能是由于缺乏扩散、栖息地不适宜（环境过滤）和负面的生物相互作用（Cadotte & Tucker,  2017）。一般来说，负面的生物相互作用（竞争、食草动物压力）被认为在维管附生植物组装中起最小作用（Burns & Zotz,  2010 ; Taylor et al.,  2016 ; Zotz,  2016）。卡夫等人。( 2015) 指出环境过滤的严格定义是指非生物环境在没有生物相互作用的情况下阻止建立或持续存在的情况；因此，附生植物是有趣的模式生物，可以了解环境过滤在塑造群落组装中的普遍性。

环境过滤通常是从非随机组合模式的存在中推断出来的（Cadotte & Tucker,  2017）。对于附生植物，随着时间的推移，低地森林中两种不同宿主物种的附生植物组合的平均成对组成相似性的增加（Laube & Zotz，  2006a，2007）为组合组成的方向变化提供了初步证据（Mendieta-Leiva & Zotz，  2015）。使用一个独特的新数据集，包括来自巴拿马古老低地森林的连续附生植物普查，涵盖十多年，我们通过检查物种组成变化、扩散限制和潜在影响来研究生态位和中性过程对组装的相对重要性主机树身份和主机大小。我们使用树个体作为采样单元，将它们类似于陆地植被生态中的地块。树木是具有分隔边界的离散可居住斑块（Southwood 和 Kennedy，  1983 年），并且可以被概念化为附生植物的栖息地岛（Ellis，  2012 年；Taylor 和 Burns，  2015 年），具有自己的动态（Wagner 和 Zotz，  2020 年））。假设中性过程占主导地位，我们预测（a）维管附生植物的物种组成不应随时间表现出“定向”变化；(b) 附生植物组合的附生植物种类组成不随寄主树种和寄主大小而变化；(c) 附生植物种类组成的差异与树木之间的距离有关，表明了扩散限制的影响。相反，如果基于生态位的机制占主导地位，我们预测物种组成变化的方向性，并且附生植物物种组成的差异与寄主植物特性的差异有关，但与寄主树的距离无关。