Invasiveness and impact of invasive species on the Tibetan Plateau are inconsistent

Abstract

Managing invasions in the context of globalization is a challenge in part because of the difficulty of inferring invader impacts from their invasiveness (i.e. ability to invade ecosystems). Specifically, the relationship between invasiveness and impact may be context-dependent and it has not been explored in such a unique ecosystem as the Tibetan Plateau. Here, we investigated 32 invasive plant species on the Tibetan Plateau in terms of their distribution, abundance, per capita effects on natives and traits across a large geographic transect to test the relationship between invasiveness and impacts on native communities. We decomposed the components (range, R; local abundance, A; per capita effect, E) that drive the impacts, and investigated the relative contributions of plant traits to these components. The results showed that there was no correlation between invasiveness (R × A) and impacts (R × A × E) of invasive species on the Tibetan Plateau. Specifically, plant invasiveness per se did not indicate a serious threat of harmful impact. In this ecosystem, R and A together drove invasiveness, while R alone drove impacts. Fruit type significantly influenced E, and species bearing berry fruits had the most negative per capita effects. However, plant traits did not drive invasiveness or impact through R, A or E. Our results suggest that the mismatch between components driving invasiveness vs. impact prevent the prediction of impacts of invasive species from their invasiveness. Therefore, management actions directed against invasive plants should prioritize broadly distributed species or those with demonstrated high impacts on native species.

摘要

外来入侵物种在青藏高原的入侵性和生态影响不一致

在全球变化背景下，管控入侵物种是一项具有挑战性的工作。这主要由于从入侵物种的入侵性推断它们的生态危害存在很大的不确定性。更确切地说，入侵性和生态危害的关系可能依赖于所研究的生态系统，同时这种关系还没有在像青藏高原这样独特的高寒生态系统中被研究过。在这项研究中，我们在青藏高原东南部的一个样带上调查了32种入侵植物。我们将这些入侵物种的生态危害分解为分布范围(R)、局部丰度(A)和个体对本地植物群落的影响(E) 3个组分，并研究了植物性状对这些组分的相对贡献。研究结果表明，入侵植物的入侵性(R × A)和其所产生的危害(R × A × E)没有相关性。这表明这些入侵植物的入侵性不能反映它们对本地群落生态危害的严重性。在青藏高原，R和A共同驱动了入侵性，而R单独驱动了生态危害。植物性状并没有通过这3个组分驱动入侵性和生态危害，只有果实类型(主要是浆果)影响了E。我们的结果表明，驱动入侵性和生态危害的组分的不匹配导致了入侵植物的入侵性不能反映它们的生态危害。因此，关于入侵物种的管控应该优先考虑那些分布广泛的物种或者已经被证实会对本地物种产生负面影响的物种。

INTRODUCTION

Understanding and quantifying the ecological impacts of invasive species in the introduced range is an important task for biological invasion research (Barney et al. 2013;,Essl et al. 2016; Levine et al. 2003; Mayfield et al. 2021). Invasive species have a severe impact on native ecosystems by reducing biodiversity, homogenizing communities, changing geochemical cycling and modifying other processes (Castro-Diez et al. 2014; Davidson et al. 2018; Petsch et al. 2022; Vilà et al. 2011). In particular, invaders may have positive feedbacks on invasions by changing the nutrient conditions of native ecosystems, thus causing a progressive deterioration of ecosystem functions (Ren et al. 2021; Zhang et al. 2019; Zhou and Staver 2019). Moreover, the impact of invasive species may change over time (Blossey et al. 2021). With the continuous introduction of invasive species through global trade and the expansion of such species’ ranges due to climate change, the need for management based on risk assessments and management priorities has led to a growing interest in quantifying and predicting the impacts of specific invasive species in specific places (Feng et al. 2022; Laginhas and Bradley 2022; Walther et al. 2009; Wang et al. 2021; van Kleunen et al. 2015).

Assessments of the ecological impacts of invasive species require clear definitions and reliable impact measurements (Ricciardi et al. 2013). On a geographical scale, the impact of invasive species can be determined from three dimensions: (i) the range size of the species’ distribution (R), (ii) the mean abundance per unit area within the range (A) and (iii) the per capita or per unit biomass effect (E). Thus, impacts (I) of invasive species can be defined as the product of R, A and E (i.e. I = R × A × E; Parker et al. 1999), and then based on this approach, it is possible to compare different species or a given species among different ecosystems. This approach has received theoretical attention and has been used with empirical plot data to estimate the impacts of invasive species (Barney et al. 2013; Gurevitch et al. 2011; Latombe et al. 2022; Pearson et al. 2016; Thiele et al. 2009). However, there is still confusion in the current measurement of invader impacts, partly because most studies only use one of these three components to represent impact, while in fact, these three components are all independent parts of the overall impact estimate (Barney et al. 2013; Gooden et al. 2009; Parker et al. 1999). The most common method used is to estimate the impact of invasive species in terms of either their abundance or their range (Bradley et al. 2019; Cheney et al. 2020; Sofaer et al. 2018). However, these two components by themselves actually are an estimate of invasiveness (R × A), which is different from impact (Catford et al. 2016; Pearson et al. 2016; Ricciardi and Cohen 2006). Since the measurement of invader impact is often unknown as per capita effects are less well understood, invasiveness is often used as a proxy for impact on the receiving ecosystem (O’Loughlin et al. 2019). For example, Pearson et al. (2016) found that invasiveness and impacts had a consistent ranking in a study of invasive plants in 31 grasslands across the western and central Montana in the USA. This supported the use of invasiveness to infer invasive plant impacts in this system. However, other studies have shown that the invasiveness of invasive species did not reflect their impacts (Horackova et al. 2014; Ricciardi and Cohen 2006). So far, the reasons why this inference is true in some cases and not in others have not been addressed.

The potential cause is that the components driving invader impacts may vary among species or ecosystems. A given species in a certain place may have a large impact due to one of the multiple combinations of its R, A and E (Parker et al. 1999). One example is the study of Pearson et al. (2016) on grasslands in North America, which showed that A independently drove invader impacts. However, there is no reason to believe that the other two components would be unimportant in a different system. The variable contributions of plant traits to the different impact’s components may be the source for the uncertainty. For example, dispersal-related traits might be independent of invader impact, by which seeds that contribute to long-distance spread are likely to increase a species’ range and contribute to invasiveness (Nunez-Mir et al. 2019; Sofaer et al. 2018), but do not directly or necessarily contribute to the decline of native plant abundance. Liao et al. (2021) found species with a wide geographical range had traits that were related to resource acquisition strategies, while plants with high local abundance had traits that were related to resource conservation strategies. Therefore, it is very helpful to distinguish the traits that drive invasiveness from those that lead to high impacts of invasive plants to understand the relationship between the two (Helsen et al. 2021).

Given that traits driving plant performance are heavily context-dependent (i.e. traits that are favored in one environment may become unimportant in another; Catford et al. 2022; Kempel et al. 2020), the impact of invasive species may largely depend on the system in which it is measured (Pyšek et al. 2012; Strayer 2020; Thiele et al. 2009). That is, invader impact is determined jointly by the exotic species and the ecosystems it invades. Global warming has enabled exotic species to expand into regions in which they previously could not survive and reproduce (Walther et al. 2009). Especially given that invasive species often come from warmer native ranges, they may be thermal adaptive species and therefore benefit more from warmer climates (Géron et al. 2021; Liu et al. 2017; Walther et al. 2009). Thus, a cold system, such as an alpine ecosystem, may produce contrasting results with warmer study sites and provide valuable information for understanding and measuring the impact of invasive species. However, few studies on quantifying the invader impacts have addressed cold systems.

In the past half century, climate warming and land use change have jointly eroded ecosystem services on the Tibetan Plateau (Hopping et al. 2018). The degradation of the original vegetation and the increased propagule pressure of invasive species introduced by human activities have led to higher risk of invasion. Due to the ecological fragility of this area, biological invasions may have a serious ecological impact. Therefore, it is urgent to investigate the invasion status and ecological impacts of invasive plants on the Tibetan Plateau. In this study, we conducted a field survey on a sample transect across a large geographic range on the Tibetan Plateau, and we recorded the abundances of invasive and native plant species and measured a set of their traits. We tried to answer three questions: (i) are plants’ invasiveness and their impact related in an alpine ecosystem? (ii) do the same components drive invasiveness and impact? and (iii) do the relative contributions of different traits to the components of impact differ?

MATERIALS AND METHODS

Study area

The Tibetan Plateau is the highest plateau in the world, with an average altitude of more than 4000 m, known as the ‘Third Pole’. The climate on the Plateau is generally cold and dry with an annual mean temperature that decreases from 6.9°C in the southeast to −4.9°C in the northwest and an annual precipitation that decreases from 593 to 84 mm over the same reach (Ding et al. 2016). The vegetation on the plateau is mainly alpine grassland, and the dominant species are Stipa purpurea, Carex moorcroftii and Kobresia pygmaea among others (Yang et al. 2009; Zheng et al. 2021).

Field survey

To measure the impacts of invasive plant species on the Tibetan Plateau, we conducted a field survey on a sample transect spanning 12 sites in southeastern Tibetan in the 2021 growing season from 20 August to 17 September (Fig. 1). We surveyed from Mangkang in the east to Lhasa in the west, completing each site in an average of 2 days. Although surveying was not possible on some days due to rain, we completed the entire field survey within a month. In each site, five sample areas were set along a distance gradient (at 0, 1, 2, 4 and 8 km) moving out from the urban boundary, and six 1 m² plots were established within 50 m along the roadside of each sampled area. The experimental design comprised 360 plots in total, but due to the steep terrain, only 345 plots could actually be reached and sampled. In each plot, we recorded all plant species, and counted the number of each. The abundance of most species was easily determined, but two special methods were used for several species. First, for small and dense plants, such as Carex parvula, we randomly selected a subplot of 0.1 m² within the 1 m² plot, dug up the plants within the subplot, removed the soil from the roots, determined their abundance based on the number of independent roots then multiplied the number of individuals by 10 to estimate their abundance in the 1 m² plot. Second, for plants that propagate by stolons, such as Fragaria nubicola, we defined each root as an individual. Species were classified into invasive species or native species based on the Plants of the World Online (https://powo.science.kew.org/).

We measured plant height and aboveground fresh biomass for the largest individual of each species in each plot. We checked the lifecycle and fruit type of these species on CABI Digital Library (https://www.cabidigitallibrary.org) and iPlant (http://www.iplant.cn/). Specifically, plant lifecycles were divided into annuals and perennials. Facultative annual species (lifecycle is variable) were identified as annuals. Fruit types were divided into achene, berry, capsule, caryopsis and utricle. There were five types of fruit (silique, gourd, schizocarp, aggregate and legume) with fewer than three species recorded, so they were combined into a group named Other.

Data analysis

We used two terms, R (invaded range) and A (mean invader abundance per unit area occupied) to calculate the invasiveness of each invasive (i.e. non-native) species (Pearson et al. 2016). To calculate R, we counted the number of 1 m² plots occupied by each invasive species. For A, we calculated the mean abundance of each invasive species in the plots where it occurred. Finally, the product R × A was used as the invasiveness score of each species (i.e. the measure of its invasion success across our study area).

The term E was used to denote the per capita effect of each invasive species on native plant abundance. We used R-package lme4 (Bates et al. 2015) to run a generalized linear model with a Poisson error structure to evaluate the effects of each invader’s abundance on the overall abundance of the native plants, as well as tested their significance (P < 0.05). We used parameter estimates (i.e. slope; E) to represent the change of native abundance in relation to the abundance of each invasive species. Finally, we expressed the impact score (I) of each invasive species as: I = R × A × E (Parker et al. 1999; Pearson et al. 2016). A negative score indicates that native species abundance decreases with an increase of the target invasive species abundance, while a positive score indicates the opposite; the absolute value of the score indicates the intensity of the impact (i.e. a more negative score indicates a larger impact on native plants).

We examined the relationship between invasiveness and impact using a linear regression model. The model was tested for outliers by the check_outliers function in R-package performance (Lüdecke et al. 2021), and the method was cook. Thus, one species (Dysphania schraderiana) was excluded from the model. We used a Pearson correlation to examine the correlation among the components R, A and E.

We ran a structural equation model with the psem function in the R-package piecewiseSEM (Lefcheck and Freckleton 2015) to test the effects of plant traits on their invasiveness and impact via their invasion range (R), local abundance (A) and their per capita effect (E). First, we created an a priori model and assumed that R and A affected invasiveness, and R, A and E all affected impact. Plant traits, including height, biomass, lifecycle and fruit type, were hypothesized to affect R, A and E separately. When categorical variables (lifecycle or fruit type) were examined, multiple comparisons were performed using the R-package multcompView (Piepho 2004) to determine the significance level of each subgroup variable. Biomass was log transformed to meet model assumptions. The fit of the final model was evaluated using Fisher’s C statistic and its associated P value (Lefcheck and Freckleton 2015). The standardized parameter estimates, namely path coefficients, were used to represent the relative contributions of traits to impact’s components, and that of the components to invasiveness or impact. All statistical analyses were performed using R 4.2.2.

RESULTS

Impact components of invasive plant species

We observed a total of 32 invasive plant species (Supplementary Table S1). The range in their R values was from 1 to 139, and the range in their A values was from 1.0 to 340.5. Twenty-nine species had significant per capita effects on native plant abundance, including five species (Carduus nutans, Erigeron canadensis, Galinsoga parviflora, Lolium arundinaceum and Trifolium repens) that had positive effects (Table 1).

Relationship between invasiveness and impact

Overall, the ranking of invaders’ invasiveness scores and impact scores was inconsistent (Fig. 2). That is to say the species with high invasiveness scores did not fully reflect the direction and intensity of their impacts on native abundance. The top three most invasive species were D. schraderiana, G. parviflora and Fragaria × ananassa, and the top three most impactful species were Datura stramonium, D. schraderiana and Chloris virgata. The linear regression model showed that there was no correlation between invasiveness score and impact score (F_1,29 = 0.24, P = 0.628, Fig. 3). The components R, A and E (R&A r = −0.05, P = 0.803; R&E r = +0.17, P = 0.366; A&E r = +0.17, P = 0.361) were not correlated.

Drivers of invasiveness and impact

The results of the structural equation model showed that plant traits did not drive invasiveness and impact through R, A or E (Fig. 4). Although fruit type had a significant effect on E, E did not drive either invasiveness or impact. Plant species with a berry fruit had the most negative per capita effects on native plant abundance. For invasiveness and impact, R drove both invasiveness and impact; A independently drove invasiveness; while E drove neither invasiveness nor impact.

DISCUSSION

The measurement of invader impacts and the exploration of driving mechanisms will provide guidance for the scientific management of invasions. This study measured the impacts of 32 invasive plant species on native grassland plants in the Tibetan Plateau. We found that plant invasiveness per se did not indicate a serious threat of harmful impact. For example, G. parviflora, a species with the high invasiveness score in this study, even had a positive impact on native abundance. In contrast, the study of Pearson et al. (2016) on grasslands in Montana (USA) found that the plants’ invasiveness scores had a similar ranking with their impact scores. Variability of the components driving impact among different ecosystems may explain the inability of invasiveness alone to predict the impact of invasive species.

We found that the range (R) primarily drove invader impact in the Tibetan Plateau, contrary to the results of Pearson et al. (2016) who found that local abundance (A) primarily drove impact. We believe that there are many different mechanisms that drive impacts among ecosystems due to the diversity of field habitats. For example, in grasslands in North America, the seeds of invasive Tragopogon dubius can be easily spread by wind over wide areas. But the plant’s seeds are favored for consumption by rodents, so this establishment limit leads to the plant having locally low abundance (Pearson et al. 2012). Therefore, whether this species can seriously impact North American grasslands will depend on how it becomes abundant at the local scale. Another example is on isolated small islands, the impacts of invasive species were related to the endemism of species (Walsh et al. 2012). In the Tibetan Plateau, range size is useful for explaining invader impacts, but it may not help explain the inconsistency in invasiveness and impact because it also drives invasiveness. Since local abundance drives invasiveness but not impact, abundance-related traits may help explain why invasiveness did not predict impact. In this dry and cold ecosystem, plant growth and reproduction are limited by energy due to the short growing season, so whether plants can rapidly grow bigger that supports more seed production is crucial to their impacts. This reflects that a large number of seeds is the basis for their local abundance. However, we did not find any plant traits (height, individual biomass, lifecycle or fruit type) that contributed to local abundance. Of course, other traits that we did not measure may be useful predictors of impact components. It is also possible that an approach using biomass per species in a plot instead of abundance per species could have obtained different results. Nevertheless, in other ecosystems, range size and local abundance are often correlated, e.g. widely distributed species are often abundant at the local scale (Fristoe et al. 2021; Theuerkauf et al. 2017; Wen et al. 2018).

Invasiveness is determined by an invasive species range and abundance but the impact also is driven by their per capita effects on native species. The debate about whether invasiveness can be used to infer impact essentially is the debate about whether the individual number of invaders or biological characteristics (i.e. per capita effect) dominates the overall impact (Dick et al. 2017; Pearse et al. 2019). We found that fruit type significantly influenced the per capita effect. Fruit type plays a key role in determining the success of plant invasions (Aronson et al. 2007; Gosper and Vivian-Smith 2009). Different fruit types are adapted to different ecosystems, and non-fleshy fruits became more frequent toward the open and arid grasslands (Chen et al. 2017; Lorts et al. 2008; Lyu et al. 2021). In this study, species with berry fruits had the most negative per capita effects, which may be linked to this being uncommon in this ecosystem. However, our study did not find that per capita effect drove overall impacts. This may have been due to the fact that the competitive advantage of invasive species in the cold Tibetan Plateau environment may be locally reduced because invasive species often come from warmer native ranges and thus favor warmer climates (Géron et al. 2021; Liu et al. 2017). We even found that several invaders had positive per capita effects, i.e. as their abundance increased, so did the abundance of native plants. A potential explanation of such positive relationships is that plants may positively interact to drive community assembly in harsh environments as hypothesized by the stress-gradient hypothesis (Bertness and Callaway 1994; He et al. 2013). Combining our results with those of previous studies, we believe that the variation of components driving invader impact in different systems is a reason for the uncertainty linking invasiveness and impact, and this need to be verified in a wider range of systems.

The ecological impacts caused by biological invasions are complex, and it is difficult to fully understand and measure them (Mayfield et al. 2021; O’Loughlin et al. 2019). The apparent impacts of invasive species on native plant communities as quantified by native abundance, however, can often be easily and quickly measured. In fact, any change in plant abundance may further affect the structure and function of the entire ecosystem through cascading effects, including but not limited to population dynamics, genetic diversity and geochemical cycling (Ehrenfeld 2003; Lavoie 2017; Parker et al. 1999; Van Cleemput et al. 2020). For example, invasive plant species can affect the structure and dynamics of native plant communities by disrupting the long-term evolutionary stability of interactions between native plants and their pollinators (Charlebois and Sargent 2017; Parra-Tabla et al. 2020). These are topics worthy of further evaluation in future prospective studies. Still, it must be recognized that there is a risk that the accuracy, stability and certainty of proxies for impact are not well tested (O’Loughlin et al. 2019). In general, it is necessary to improve the precision of invader impact assessment required for management decisions, and the actions to infer the impact with invasiveness need to be discussed in a broader context.

This study generated valuable information when compared with the study of Pearson et al. (2016) on the grasslands of North America. In this comparison, we showed how the relationship between invasiveness and impact, and their drivers, can vary across ecosystems. In fact, any changes caused by ecosystem differences reflect the context dependence of ecological processes and mechanisms (Catford et al. 2022). For example, the ecological impact of invasive plants depends heavily on climatic conditions, as they are considered to be more suitable for warmer climates (Liu et al. 2017; Walther et al. 2009). Although invasive species are less common in cold regions, biological invasions are on the rise in these areas due to climate warming. Meanwhile, plant invasion in cold environments is particularly sensitive to disturbance (Lembrechts et al. 2016). Human activities on the Tibetan Plateau, such as tourism and grazing, have contributed to the spread of invasive plant species, which have largely contributed to their impact. Clearly, the unique ecosystem of the Tibetan Plateau provides an opportunity to gain novel insights.

Supplementary Material

Supplementary material is available at Journal of Plant Ecology online.

Table S1: Trait information of 32 invasive species on the Tibetan Plateau. Height and biomass are the average values for the largest plants of each species in each plot.

Table S2: The complete outputs from Structural Equation Modeling. Significant (P < 0.05) terms are indicated in bold.

Funding

This research was funded by the Central Government Guides Local Projects of China (XZ202101YD0016C), the open funding from Tibet Joint Key Laboratory of Ecological Security (STAQ-2021T-2), the open funding from Yunnan Key Laboratory of Plant Reproductive Adaption and Evolutionary Ecology, Yunnan University (2019DG056) and the National Natural Science Foundation of China (31822007 and 32071660).

Conflict of interest statement. The authors declare that they have no conflict of interest.

REFERENCES