Analysing spatio‐temporal patterns of non‐native fish in a biodiversity hotspot across decades

Analysing the spatio‐temporal patterns and dynamics of non‐native species is essential to understanding the mechanisms underlying successful invasions and developing effective management strategies. Yet, such analyses generally neglect the influence of receiving ecosystem types and non‐native species sources (i.e. alien species, non‐natives originating outside the concerned region; translocated species, non‐natives introduced to locations outside their historical range within the concerned region).


| INTRODUC TI ON
The expansion of international trade has resulted in the intentional and unintentional introductions of numerous species beyond their natural ranges and habitats (Hulme, 2009).Unfortunately, the global accumulation of non-native species has not yet reached a saturation point, and experts have predicted a continuing upward trend in the coming years (Bailey et al., 2020;Seebens et al., 2017Seebens et al., , 2021)).Increasing biological invasions can lead directly or indirectly to significant declines in biodiversity and incur enormous economic costs (Diagne et al., 2021;Du et al., 2023;Reid et al., 2019).For example, a global meta-analysis found that 30 invasive mammalian predators are important drivers of irreversible loss of phylogenetic diversity worldwide, causing negative impacts on 596 threatened and 142 extinct species (Doherty et al., 2016).The financial cost of biological invasions is staggering, with a global estimated cost of at least US$1.288 trillion and Asia alone bearing a total cost of US$432.6 billion (Diagne et al., 2021;Liu et al., 2021).Consequently, most countries and regions worldwide have enacted relevant laws and regulations, as well as developed evidence-based prevention and management measures to mitigate the ecological and economic impacts of biological invasions (Turbelin et al., 2017).Nevertheless, the effectiveness of implementing these management measures critically depends on a clear understanding of the spatio-temporal patterns and dynamics of non-native species (Azzurro et al., 2022;Daga et al., 2015;Muñoz-Mas et al., 2023;Vitule et al., 2012).This is because, based on this foundation, researchers can predict the spreading trend of non-native species and forecast changes in invasion situations for specific regions (Liu et al., 2019), while managers and policymakers can timely identify key non-native species and delineate priority areas for designing effective prevention and management strategies (Magalhães et al., 2020;Pyšek et al., 2020).
Freshwater ecosystems are particularly vulnerable to invasion by non-native animals and plants (Bando et al., 2023;Moorhouse & Macdonald, 2015).Because of economic and recreational purposes, fish species, in particular, are frequently introduced into freshwater ecosystems globally, resulting in their widespread establishments and often leading to freshwater biodiversity declines and, in the worst case, ecosystem collapses (Cucherousset & Olden, 2011;Muñoz-Mas et al., 2023;Su et al., 2021).To effectively conserve freshwater biodiversity and ecosystem functions, as well as inform management strategies, it is essential to unravel the mechanisms behind successful invasions through analysing the spatio-temporal patterns and understanding the dynamics of non-native fishes (Azzurro et al., 2022).However, existing analyses of the spatio-temporal dynamics of non-native species including fish generally pay little attention to the influence of receiving ecosystem types and non-native species sources (i.e.alien species, non-natives originating outside the concerned region; translocated species, non-natives introduced to locations outside their historical range within the concerned region; Comte et al., 2021).
There are several reasons why the different ecosystem types should be examined and contrasted in invasion studies.First, it has been shown that different ecosystem types differ significantly in their ability to resist invasions, even when they are in the same region suffering from similar frequency and intensity of introductions (Holle & Simberloff, 2005).The differences can be attributed mainly to the suitability of environmental conditions and degree of resistance by native species towards non-native species in receiving ecosystems (Bernery et al., 2022).For example, studies have shown that reservoirs may be more vulnerable to fish invasions than rivers or lakes due to their higher nutrient concentrations and altered flow regimes, which can create conditions that favour non-native species (Johnson et al., 2008;Loures & Pompeu, 2019;Zhang et al., 2022).
Second, due to the growing awareness of the negative impacts of non-native species invasions on receiving ecosystems, introductions crossing administrative regions (e.g.country, state or province) have been substantially restricted (Hulme, 2021).However, intraregional species translocations within an administrative region have not received sufficient attention before (Dawson et al., 2017;Vitule et al., 2019;Xiong et al., 2015).Spatially close areas within a region are generally characterized by similar environments and species in biological communities are relatively closely related, with different communities thus showing relatively small phylogenetic differences.
According to the mechanisms of successful invasions revealed in previous research, both characteristics mentioned above can facilitate successful invasions (Li et al., 2015;Xu et al., 2022).Nonnative species tend to exhibit more intense competition or frequent hybridization with their close relatives in receiving ecosystems, as they generally have similar functional traits, niches and genetic characteristics (Blackwell et al., 2021).Therefore, analysing the spatio-temporal patterns and dynamics of non-native fishes should carefully consider the receiving waterbody types and non-native species sources.
Biodiversity hotspots are regions with high species richness, including species that are both endemic and threatened, playing a vital role in global biodiversity conservation (Myers et al., 2000).
However, biodiversity hotspots are highly susceptible to biological invasions due to natural ecosystem features and anthropogenic disturbances, creating favourable conditions for the spread and establishment non-native species (Li et al., 2016).First, while biodiversity hotspots are important owing to their unique and fragile ecosystems, their heterogeneous environments can also provide potential ecological niches and abundant resources (e.g.habitats and food) for non-native species to establish and thrive (Montti et al., 2021).
biodiversity hotspot, biological invasion, historical trends, intra-regional species translocation, invasion hotspots, waterbody sensitivity Second, endemic species in biodiversity hotspots are often adapted to specific habitats, making them vulnerable to competition with non-native species that often have wider niches and may have a greater competitive advantage (Bellard et al., 2014).Third, anthropogenic activities, such as habitat loss (e.g.caused by deforestation), international trade (e.g.timber and pets), transportation and tourism, have a more significant impact on biodiversity hotspots, increasing the extinction risk of native species and the likelihood of successful non-native species introductions (Liu et al., 2017).Furthermore, previous studies have suggested that detected invasion patterns may have varied predictability in biodiversity hotspots with abundant native species (Bernery et al., 2022).Therefore, analysing the spatio-temporal dynamics of non-native species in these regions can enhance our understanding of the mechanisms underlying successful invasions.
Yunnan is a renowned biodiversity hotspot in China and considered important globally (Wang et al., 2018).This region has the highest diversity of freshwater fish in China (Chen et al., 1998).However, rich natural resources (e.g.unique landscape scenery and water resources) also make Yunnan the most attractive place for tourists and key areas for hydropower development and aquaculture in China (Liu et al., 2021;Zhang et al., 2019).Consequently, the intentional and unintentional introductions of non-native fish caused by the above human activities have been increasingly severe.The peak of non-native fish introductions in Yunnan occurred mainly in the late 1950s and late 1970s (Ding et al., 2017;Kang et al., 2022;Zhang et al., 2018).However, analyses on the spatio-temporal patterns and dynamics of non-native fishes in Yunnan, especially for long-term time series, remain unavailable.
The freshwater fishes of Yunnan provide a distinct opportunity to disentangle the influence of receiving ecosystem types and nonnative species sources from the underlying mechanisms behind successful invasions.In this study, we compiled the largest occurrence database for freshwater fish in Yunnan, including both native and non-native species.Our aim was to analyse the spatio-temporal patterns and dynamics of non-native fish in this biodiversity hotspot between 1950 and 2022.Our objectives are as follows: (1) to obtain a comprehensive inventory of species composition and distribution, geographical origin and introduction pathways of non-native fish and (2) to analyse the spatio-temporal patterns and dynamics of non-native fish richness and invadedness, with a particular focus on receiving waterbody types and non-native species sources (i.e.alien and translocated species).We hypothesized that lentic waterbodies (i.e.lakes and reservoirs) are more profoundly affected by both alien and translocated species than lotic waterbodies (i.e.rivers) and that the introduction of translocated species is a long-standing issue as alien species invasions, which were mainly neglected the past.The outcomes of our study will: (1) help to develop effective management strategies for controlling non-native species and preserving the ecological integrity of native fish communities in Yunnan and (2) provide insights into understanding the potential mechanisms behind successful invasions and developing effective prevention and management strategies, particularly for biodiversity hotspots.

| Study area
Yunnan (97°31′-106°11′E; 21°8′-29°15′N), which hosted the 15th UN Convention on Biological Diversity (CBD-COP15) in 2021, is located in southwestern China and is part of the Mountains of Southwest China ecoregion (Figure 1; Wang et al., 2018).It has a 394,100 km 2 area bordering Myanmar, Laos and Thailand.The topography of Yunnan is complex with large differences in elevation, being characterized by mountainous areas accounting for 85% of the total land area.Its topography exhibits diverse features across its four cardinal directions.To the west, mountain ranges and northsouth-flowing rivers dominate the landscape.In contrast, the eastern section is typified by karst plateau terrain, with non-navigable rivers flowing in deep mountain valleys.The northern region, meanwhile, forms part of the Yunnan-Guizhou Plateau.The southern section, on the contrary, is characterized by extensive tropical forests, marked by relatively low elevation with a warm and humid climate.
The complex topography plays a crucial role in shaping the spatial patterns of climate (Wang et al., 2018).The climate of Yunnan can be characterized as a subtropical highland monsoon climate with distinct vertical and seasonal variations, including four major climate zone types (i.e.cold, temperate, subtropical and tropical).The mean annual temperatures ranged from 4.3 to 24°C, and the mean annual precipitation ranged from 600 to 3000 mm (Fan et al., 2011;Li et al., 2015).
The varied topography, climate and geology have resulted in high levels of environmental heterogeneity, supporting a diverse range of ecosystems, including subtropical forests, temperate forests, alpine forests, grasslands, wetlands, rivers and lakes ( Liu et al., 2021).There are around 25,426 species of macrofungi, lichens, higher plants and vertebrates, accounting for about half of the species of China, including some of the world's most endangered species such as the Yunnan Golden Monkey, Black-necked Crane and the Chinese Giant Salamander.Yunnan is estimated to have over 600 freshwater fish species (accounting for around 40% of the total number of freshwater fish species in China), and most of them are endemic (Chen, 2013;Chu & Chen, 1989, 1990;Tao et al., 2023).The region is dotted with over 23 isolated highland lakes and 125 reservoirs (>1 km 2 ), covered by the watersheds of six large rivers, namely the Jinsha-Yangtze (watershed area in Yunnan: 109,382 km 2 ), Nanpan-Pearl (59,191 km 2 ), Yuan-Red (74,349 km 2 ), Lancang-Mekong (88,287 km 2 ), Nu-Salween (33,529 km 2 ) and Dulong-Irrawaddy (19,097 km 2 ; Figure 1; Wang et al., 2022).These watersheds encompass all main types of waterbodies, including rivers, lakes and reservoirs (Figure 1).

| Fish occurrence dataset
To analyse the spatio-temporal patterns and dynamics of nonnative fishes in Yunnan, we compiled and curated the occurrence data of both native and non-native freshwater fish by including our long-term field sampling records and systematically searching all possible sources (He et al., 2020;Tao et al., 2023;Yang et al., 2022;Zhang et al., 2019).We conducted a thorough literature search in multiple databases, such as the Web of Science (WoS), Scopus and the Chinese National Knowledge Infrastructure (CNKI), a comprehensive Chinese database that includes journal papers, theses, proceedings, newspapers, yearbooks and more.We also searched for books, online databases [e.g.FishBase and the Global Biodiversity Information Facility (GBIF)] and unpublished reports.Chinese sources were particularly considered because of their importance in research coverage, which allowed us to obtain a more comprehensive overview of non-native fishes in Yunnan (Tao et al., 2018).The search was initially conducted in July 2022 and updated in March 2023.Our search queries were based on the names of the target fish (e.g.scientific names, common names and order names) and the names and types of the waterbodies (e.g.rivers, lakes and reservoirs) in Yunnan (Chen, 2013;Chu & Chen, 1989, 1990;He et al., 2020;Tao et al., 2023).Data from the WoS and Scopus, based on titles, abstracts and keywords, were searched using the following English  S1).The Chinese search phrase in CNKI corresponded to the English version (Table S1).We then screened the We extracted occurrence data information, including species names, georeferenced locations and times, for both native and non-native species from the text, tables, figures and supporting information of the references.To extract data from figures, we used the WebPlotDigitizer (Version 4.4; Burda et al., 2017).For highly relevant references without openly shared data, we contacted the authors by email.To avoid including invalid species and synonyms, we verified the validity of each species according to FishBase (Froese & Pauly, 2022) and Eschmeyer's Catalogue of Fishes (Fricke et al., 2022).Based on our study objectives, we classified species into different categories, including native, non-native, alien and translocated species, based on their historical distribution status in each watershed.Native species refer to indigenous freshwater fishes in a specific watershed, while non-native species are those introduced to areas beyond their native ranges (Xiang et al., 2021;Xiong et al., 2015).Considering the impact of non-native species sources, we further classified them into alien and translocated species, referring to Yunnan.Alien species are non-natives originating outside Yunnan, while translocated species are non-natives introduced to locations outside their historical range within Yunnan.Species introduction pathways, preferred habitat and maximum total length of non-native species were derived from literature records (Li et al., 2016;Xiong et al., 2015) and FishBase.According to the maximum total lengths, species were classified into three body-size groups: large (>40 cm), medium (between 15 and 40 cm) and small (<15 cm; Reis-Filho et al., 2019).We then calculated the occupancy of non-native species across subdrainage basins and waterbody types in Yunnan.

| Analysing spatio-temporal patterns and dynamics of non-native fishes
To delineate the spatial distribution of both native and non-native fishes, we used expert knowledge to mitigate potential bias resulting from heterogeneous occurrence data, which proved to be an effective strategy in a previous study (Tao et al., 2023).We avoided species distribution modelling as it could overestimate species distribution range (Loiselle et al., 2003), potentially erasing differences between subdrainage basins and not aligning with our research objectives.Additionally, some species had insufficient occurrence data points for developing species distribution models.
The workflow for estimating distribution involved occurrence point buffering and subdrainage basin clipping, which was revised from the rigorous strategy developed in Tao et al., (2023).
We first snapped all species occurrence data points to the drainage system using the 'Near Analysis' toolbox in ArcGIS Pro (Environmental Systems Research Institute (ESRI), 2022).The snapping results were carefully screened, especially in areas where tributaries confluence with the main stem and adjacent areas of different watersheds.
Next, we created a buffering zone for each occurrence point of all species.The buffering distance of each occurrence point for a specific species was based on fish dispersal capacity and the river order it occurred using an empirical-data-developed method (Radinger & Wolter, 2014).Fish dispersal capacity-related functional traits (i.e. total length, TL and aspect ratio of the caudal fin, AR) of each species were extracted from FISHMORPH (Brosse et al., 2021) and FishBase (Froese & Pauly, 2022).The river order of each occurrence point was obtained from HydroRIVERS (Lehner & Grill, 2013).The buffering distance of each occurrence point was estimated using the R package 'fishmove' (Radinger & Wolter, 2014).The estimated dispersal distance was used as the diameter to construct the buffer zone for each occurrence point.Then, the buffer zones were overlapped with subdrainage basins (i.e.hydrounits) at Level 7 in HydroBASINS (Lehner & Grill, 2013).Considering that fish can only move among hydrologically connected basins through waterways, the overlaps were constrained within the same watershed where the fish occurrence point was presented.Level 7 was used because the coverage of occurrence points was estimated to be low at a finer scale (i.e.Levels 8-12) according to the database compiled.In addition, fish assemblages are more likely to differ at higher levels (i.e.Levels 1-7), based on estimates of fish dispersal capacity and existing studies (Radinger & Wolter, 2014;Tao et al., 2023).However, larger subdrainage basins (e.g.Level 6) are likely to include more complex waterbody types, which violates our research objective of differentiating the influence of receiving waterbody types on invasion success.Subdrainage basins with an area less than 50 km 2 , which are generally located at the borders of Yunnan, were combined into adjacent subdrainage basins within the same watershed (Oberdorff et al., 2019).The study area was divided into 202 subdrainage basins (Figure 1).According to the dominant waterbody type present, the subdrainage basins were allocated into four groups, including river type (subdrainage basins containing neither lakes nor reservoirs, N = 107), lake type (subdrainage basins containing lakes but no reservoirs, N = 11), reservoir type (subdrainage basins containing reservoirs but no lakes, N = 73) and lake-reservoir coexistence type (subdrainage basins containing both lakes and reservoirs, N = 11).As fish occurrence data for small lakes and reservoirs were largely unavailable, only those with an area over 1 km 2 were counted (Wang et al., 2022).For subdrainage basin clipping, we assigned the entire subdrainage basin where the occurrence point presented is the distribution range of the species.
The upstream or downstream subdrainage basins of the occurrence point but overlapped by the buffer zone were also assigned as the distribution range of the species.In addition, if the distance between two occurrence points exhibiting a longitudinal relationship is less than 10 times the dispersal distance of the species, the subdrainage basins situated between them are considered part of the species distribution range.
To examine the temporal patterns and dynamics of non-native fish invasions across Yunnan, the accumulative occurrence data records of non-native species were divided into four periods: P1960s (i.e.before 1960), P1980s (1980), P2000s (2000) and P2020s (2022).This period division considers the length of each time period and the history of aquaculture introduction in Yunnan Province (Ding et al., 2017).The peak of non-native fish introductions in Yunnan occurred mainly in the late 1950s and late 1970s.The occurrence data records of native species were undivided, and their distributions were used as the baseline for evaluating the severity of invasion of a corresponding area (e.g.subdrainage basins, watersheds and waterbody types).

| Data analysis
To examine temporal changes, we calculated the cumulative number of records and cumulative species richness of non-native fish (CSR NN ; Azzurro et al., 2022), which accounts for variations in sampling records across time due to changes in research efforts.To evaluate the severity of invasion in a specific area, we calculated an index of invadedness, which expresses the degree of invasion by non-native fish as a ratio of non-native species to native species richness (Comte et al., 2021).Additionally, we separately considered alien species and translocated species, when necessary, as well as waterbody types.
To demonstrate the phylogenetic differences between nonnative species (i.e.alien and translocated) and native species, we generated the phylogenetic tree for Yunnan fish species using the R packages 'FishPhyloMaker', 'ggtree' and 'ggtreeextra' (Nakamura et al., 2021;Xu et al., 2021;Yu et al., 2017).The statistical analysis of median value differences in records, richness and invadedness for non-native species across the four periods for the entire region, six watersheds and four waterbody types was conducted using the nonparametric Kruskal-Wallis H tests, along with multiple comparisons using the 'PMCMRplus' R package (Pohlert, 2022).The Wilcoxon signed-rank tests were used to compare the records, richness and invadedness of two species sources across periods and watersheds.
The mapping data visualizations were conducted using ArcGIS Pro.
All analyses were conducted using R software (version 4.2.1;R Core Team 2022).
In the phylogenetic tree, alien species exhibit a tendency to cluster together and separate from native species, whereas translocated species tend to disperse and display closer relatives with native species (Figure S2).According to the maximum total lengths, there were 44, 31 and 19 non-native species were classified as large, medium and small body-size fish, respectively.Thirty-four species prefer lentic habitats, three species prefer lotic habitats, and 28 species have no apparent preference.Regarding the source relative to Yunnan, 49 species are alien fishes, and 45 species are translocated fishes.
For waterbody types, the records, richness and invadedness of the different waterbody types did not differ statistically across periods and watersheds (Kruskal-Wallis H tests, p > .05 in all cases; Figure 4).However, records, richness and invadedness of the laketype basins are generally higher than those of the other three waterbody types (Figure 4a,c,e).Within different watersheds, the records, richness and invadedness were generally higher for lake-type and reservoir-type basins than for river-type basins (Figure 4b,d,f).For non-native species sources, the records, richness and invadedness F I G U R E 2 Phylogenetic tree of the 94 non-native fish species extracted from 3040 records showing the preferred habitat, maximum total length, species source, introduction pathway, first recorded period, species occupancy and preferred receiving waterbody type.Nodes show the order of species.Numbers in parentheses indicate the number of species. of translocated species did not differ significantly from those of alien species across periods and watersheds (Wilcoxon signed-rank tests, p > .05 in all cases; Figure 5).The richness and invadedness per subdrainage basin of non-native, alien and translocated species of the lake-type basin were generally higher than the other three type basins (Kruskal-Wallis H with paired post hoc tests, p < .05; Figure 6).

| DISCUSS ION
The current study utilized the largest and most comprehensive freshwater fish occurrence dataset to analyse the spatio-temporal patterns and dynamics of non-native fish introductions in Yunnan, an important biodiversity hotspot in a global perspective.The study resulted in comprehensive maps depicting the accumulation of non-native fish records, richness and invadedness.Our study highlights the previously overlooked issue of translocated species introductions and emphasizes the varying sensitivities of different waterbody types to non-native species, with lake-type subdrainage basins being more susceptible to non-native fish than rivers.It offers valuable insights for identifying gaps in intra-regional introductions and implementing prevention and management strategies for nonnative species in different waterbody types.

| Introductions of alien and translocated species
Enhancing our understanding of the ecological and economic impacts, as well as the underlying mechanisms facilitating successful establishment, necessitates a specific focus on non-native species source, whether they are introduced externally or internally (Comte et al., 2021;Liu et al., 2017).Regrettably, prior research has often neglected the significant matter of intra-regional species translocations (Dawson et al., 2017;Xiong et al., 2015).Our findings highlight a persistent challenge of intra-regional introductions, where both translocated and alien species have been introduced simultaneously across all time periods and watersheds (Figure 5).
Notably, there is a discernible trend revealing a higher frequency of translocated species records surpassing those of alien species (Figure 5a).This observation can be attributed to a historical bias in regional prevention and management efforts, which primarily focussed on managing alien species while neglecting translocated species (Turbelin et al., 2017;Vitule et al., 2019).This oversight allowed translocated species to go unnoticed until their occurrence was recorded retroactively.
Neglecting the inclusion of translocated species within a region may have detrimental effects on the prevention and management of non-native species.On the one hand, neglecting intra-regional translocated species introductions may underestimate the number of non-native species (Vitule et al., 2019).Our study represents a significant update to the existing database of non-native fish distribution in Yunnan, encompassing 94 species (Figures 2 and 3), a considerably larger number than previous studies (e.g.34 species in Chen, 2013, 51 species in Li et al., 2016, and40 species in Chen, 2021).This notable underestimation in earlier studies can be attributed to insufficient attention given to intra-regional introductions, specifically translocated species.For instance, recent studies in the Amazon Basin also support this view.Initial investigations identified fewer than 20 species, but a comprehensive analysis focussed on translocated species found a minimum of 41 non-native fish species, more than twice the previously reported number (da Costa Doria et al., 2021).This significant disparity can largely be attributed to the inclusion of translocated species, a factor that has often been overlooked or insufficiently characterized (Dawson et al., 2017).On the other hand, neglecting translocated species can lead to substantial underestimation of both invasion debt and extinction debt effects caused by non-native species introductions and native species extinctions in some mega biodiversity countries (e.g.Brazil and China; Ding et al., 2017;Liu et al., 2017;Vitule et al., 2012).
Compared with alien species, translocated species are also more likely to have a greater ecological impact and to establish populations successfully.For example, translocated species tend to be dispersed among native species in phylogenetic trees, while alien species tend to cluster together (Figure S2), which show that the phylogenetic differences between translocated and native species are often smaller in comparison with alien species.In southern Tanzania, Africa a notable example of biointeraction between a translocated fish and a native fish was observed with the introduction of Nile tilapia (Oreochromis niloticus) into various lakes (Blackwell et al., 2021).
The invasive Nile tilapia and the native Korogwe tilapia (Oreochromis korogwe) share a close genetic relationship and exhibit notable similarities in their functional and morphological characteristics.As a result, intense competition and frequent hybridization between the two species have occurred, leading to population decline and loss of unique and irreplaceable genetic resources of Korogwe tilapia (Blackwell et al., 2021).Moreover, it has also been observed that larger regions or countries typically harbour a higher number of translocated species (Brito et al., 2020;Liu et al., 2017;Vitule et al., 2012).This pattern underscores the essential role played by translocated species in driving the process of biotic homogenization and shaping the historical composition of freshwater fish faunas.Therefore, it is imperative to broaden the focus of future prevention and management efforts on both alien species and translocated species (Jiang et al., 2021;Vitule et al., 2019), particularly for hotspots like Yunnan with a high endemic biodiversity.

| Sensitivity of receiving waterbody types to non-native species
The resistance of different ecosystems to non-native species can vary significantly, even when subjected to similar levels of introduction or propagule pressure (Holle & Simberloff, 2005).
Different waterbodies exhibit different habitat and hydrological characteristics, with lakes and reservoirs representing lentic ecosystems, while rivers embody lotic ecosystems (Drakou et al., 2008).Of these, reservoirs are man-made and a mixture of features of lakes and rivers, thereby representing somewhat novel environments for fish.Our results indicate that the records, richness and invadedness of lake-type basins are significantly higher than those of river-type basins, and reservoir-type basins are slightly higher than those of river-type basins (Figures 4 and   6).This partly reflects that lentic ecosystems are more susceptible to non-native species, which is consistent with the results of previous studies, including those conducted in some mega biodiversity countries (Daga et al., 2015;Loures & Pompeu, 2019;Vitule et al., 2012;Zhang et al., 2022) water and frequent flood events may make it more challenging for non-native fish to establish and survive (Koutsikos et al., 2019).
In contrast, lentic ecosystems with slower flow rates and limited hydrological connectivity may provide more stable habitat for non-native species colonization (Loures & Pompeu, 2019;Zhang et al., 2019;Zhang et al., 2022).Furthermore, lakes and reservoirs are profoundly impacted by anthropogenic activities (e.g.aquaculture), thereby facing the challenges associated with the introduction of non-native species (Ding et al., 2017;Johnson et al., 2008).Lakes have hence experienced substantial declines and extinctions of native species over time due to multiple factors such as aquaculture activity introductions, degradation of water quality and overfishing (Ding et al., 2017).Similarly, reservoirs undergo transformations in their flow regimes, transitioning from lotic to lentic conditions.This alteration may create novel ecological niche opportunities exploited by non-native species that are more adaptable and tolerant (Chen et al., 2023;Zhang et al., 2019).As a result, both lakes and reservoirs are highly susceptible to the invasion of non-native species due to anthropogenic influences and altered ecological dynamics.
The sensitivity of lentic ecosystems varies among different waterbody types.Previous studies conducted in the Great Lakes region of the United States have indicated that reservoirs are more susceptible to invasion by non-native species than natural lakes (Johnson et al., 2008).However, our study showed that reservoirs typically exhibit a lower record, richness and invadedness of non-native species compared with lakes (Figures 4 and 6).It may be attributed to the historical anthropogenic introductions experienced by lakes, which have led to a substantial influx of nonnative fish species (Ding et al., 2017).In contrast, reservoirs have a relatively short construction history (mostly formed after P1980s), with lower frequency of non-native fish introductions comparing to other watersheds, such as the Yellow River (Jiang et al., 2021;Zhang et al., 2019).The richness of non-native species in reservoirs has not yet fully saturated.

| Prevention and management of non-native species
A comprehensive understanding of the spatio-temporal patterns and dynamics of non-native species empowers us to formulate effective strategies for prevention and management (Azzurro et al., 2022;Muñoz-Mas et al., 2023).Our findings undoubtedly offer important insights for preventing and managing non-native species, both in biodiversity hotspots such as Yunnan and elsewhere around the globe.We found that the introductions of translocated species were comparably as serious as those of alien species (Figures 2, 5 and 6), but the former were generally overlooked, including their potentially detrimental effects.This means that all existing prevention and control strategies and measures for non-native species must take into account translocated species, not just alien species (Vitule et al., 2019).We can prioritize the following areas of work.
First, enhancing monitoring and early detection systems for translocated species is imperative, as they are morphologically similar to native species and are at a higher risk of being overlooked (Blackwell et al., 2021;Xia et al., 2023).Second, it is necessary to strengthen communication to foster consensus among stakeholders, including scientists, government agencies, conservation organizations and local communities, that translocated species are also harmful non-native species (Jarić et al., 2021).This will enable their active engagement in resisting the introductions of translocated species.
Third, it is also urgent to study the ecological impacts of translocated species and evaluate their invasiveness using sophisticated technologies (e.g.Aquatic Species Invasiveness Screening Kit, Copp et al., 2016).Lastly, supplementing existing laws and regulations with provisions prohibiting the introduction of translocated species is also a must.
Regarding the sensitivities of waterbody types to non-native species, lakes and reservoirs bear the brunt of fish invasions, while rivers have fewer non-native species (Figures 4 and 6; Vitule et al., 2012;Ding et al., 2017;Zhang et al., 2019;Brito et al., 2020;Kang et al., 2022).It is likely that accumulations of non-native fish in lakes are close to saturation now (Figure 3; Seebens et al., 2017).
Therefore, eradication measures are no longer practical here.
Physical barriers, selective fishing and the sterile male technique should be used in future to enhance the management of non-native species and prevent their further spread (Bernery et al., 2022).
For reservoirs, most of them are currently unsaturated, so efforts should focus on reducing propagule pressure and preventing the introduction of new non-native species.This is also meaningful as reservoirs can act as 'stepping-stone' habitats for the continued spread of freshwater invaders (Johnson et al., 2008).Maintaining natural flow regimes in dam-impacted reaches as much as possible can also be used to prevent the further spread of non-native species while promoting the conservation of native species (Ding et al., 2023;Tao et al., 2023).Although the number of non-native species in rivers remains relatively low, there is a need to focus on preventing the introduction of flow-insensitive alien species and translocations of rheophilic non-natives from neighbouring watersheds.

| CON CLUS ION
In conclusion, the spatio-temporal patterns and dynamics of nonnative fish introductions of a biodiversity hotspot were successfully analysed in our study.It emphasizes the largely overlooked issue of translocated species introductions and their comparable or greater ecological impacts compared with alien species.The study further identifies subdrainage basins of lake and reservoir types, which are subject to significant human disturbance, as particularly vulnerable to non-native species invasion.These findings are crucial for recognizing the limitations of current non-native species prevention and management strategies and for implementing effective biodiversity conservation measures.Addressing the adverse impacts of non-native species on biodiversity hotspots and preserving their antiquity, naturalness and uniqueness necessitate collaborative efforts among government agencies, scientific institutions, conservation organizations and local people.Joint initiatives should prioritize the protection of these valuable ecosystems and the development of proactive management approaches to mitigate the negative effects of non-native species and safeguard biodiversity hotspots for future generations.
search phrase: ((fish* OR 'new species' OR carp OR '…' OR Acipenseriformes OR '…' OR 'Abbottina obtusirostris' OR '…') AND (Yunnan OR Yun-Gui Plateau OR Jinsha OR '…' OR Dianchi OR '…' OR 'Gongguoqiao' OR '…')) (Table titles and abstracts of the documents returned by the search and excluded records with explicit reasons (e.g.reviews without sampling data and studies with mismatched study taxa and regions; Figure S1).Primary research articles mentioned in review papers that potentially contain relevant data were also included to complement our reference pool.The searching, screening and filtering strictly F I G U R E 1 Study area and occurrence records of native and non-native fish species.(a) The geographical location of Yunnan; (b) six watersheds in Yunnan; (c) distribution of four different waterbody types with a total of 202 subdrainage basins; (d) 14,002 occurrence records of native fish; (e) 783 species of native fish and (f) 3040 occurrence records of non-native fish.followed the workflow of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis; Figure S1; Page et al., 2021).

F
I G U R E 3 Spatio-temporal accumulation of non-native fish in Yunnan.(a, c and e) Cumulative records, species richness and invadedness.(b, d and f) Cumulative records, species richness and invadedness of different watersheds in the P2020s.In boxplots, the grey dots represent individual subdrainage basins, the red diamonds in the centre represent the mean, the bars represent the median and the interquartile range, and the whiskers represent the 100% percentile range excluding outliers.The different lowercase letters indicate significant differences between median values (Kruskal-Wallis H with paired post hoc tests, p < .05).DI, Dulong-Irrawaddy; JY, Jinsha-Yangtze; LM, Lancang-Mekong; NP, Nanpan-Pearl; NS, Nu-Salween; YR, Yuan-Red.

F
I G U R E 4 Spatio-temporal accumulation of non-native fish in different waterbody types.(a, b) records, (c, d) species richness, (e, f) invadedness of non-native fish across four periods and six watersheds.The p values of Kruskal-Wallis H tests are shown.

.
Lotic ecosystems with varying flow patterns, substrate types, vegetation, fast-flowing F I G U R E 5 Spatio-temporal accumulation of alien and translocated fish species.(a, b) records, (c, d) species richness, (e, f) invadedness of alien and translocated fish species across four periods and six watersheds.The p values of Wilcoxon signed-rank tests are shown.

F
Cumulative records, species richness and invadedness of different waterbody types in the P2020s.The dots represent individual subdrainage basins, the red diamonds in the centre represent the mean, the bars represent the median and the interquartile range, and the whiskers represent the 100% percentile range excluding outliers.The different lowercase letters indicate significant differences between median values (Kruskal-Wallis H with paired post hoc tests, p < .05).