基于MaxEnt模式的白头叶猴栖息地评价——以广西崇左白头叶猴保护区为例

针对物种分布格局与其环境变量关系的研究,对于生态廊道规划与环境恢复研究具有重要意义。本文以白头叶猴（Trachypithecus leucocephalus）为研究对象,针对广西崇左白头叶猴国家级自然保护区51个白头叶猴分布点和11个环境变量数据,利用MaxEnt模型（maximum entropy modeling）对栖息地适宜性进行综合评价,通过模型生成栖息地适宜度指数（habitat suitability index,HSI）阈值,划定研究区域为低适生区、适生区和高适生区3种类型,其中低适生区面积为5 061.43 km²,适生区和高适生区面积分别为42.80 km²和20.63 km²。主要环境变量分析表明：年均降水量、土地利用分类、坡度和年平均气温的综合贡献值分别为54.6%、17.4%、11.8%和9.5%,4项环境变量累积贡献值达到93.3%,是影响白头叶猴分布的主要环境因素。为更有效保护白头叶猴及其栖息地,建议在保护区规划生态廊道,扩大恢复区内白头叶猴适宜栖息地面积,以促进白头叶猴种群间基因交流。     

广西扶绥白头叶猴对栖息地的选择与利用

为了解白头叶猴（Trachypithecus leucocephalus）的栖息地利用规律及其影响因素,2016年2月至2017年1月,采用瞬时扫描取样法对广西崇左白头叶猴国家级自然保护区一群白头叶猴的栖息地利用进行了研究。结果表明,白头叶猴对山体不同部位的利用存在显著性差异（χ² =39.467,df=3,P<0.001）,其中,对崖壁（56.75±9.55）%的利用比例最大,其次是对山坡（39.42±10.93）%和山顶（2.98±2.54）%的利用,而对山脚（0.84±1.47）%的利用频率最低。白头叶猴对不同微生境类型的利用存在差异（χ²=27.709,df=3,P<0.001）,其中对乔木（49.37±12.31）%的利用比例最大,其次是裸岩（24.05±13.61）%,随后依次为藤本（15.48±8.01）%和灌木（10.87±5.45）%。白头叶猴主要在山坡上觅食,利用崖壁移动、休息,进行社会活动;主要利用裸岩进行社会活动,觅食、移动、休息主要发生在乔木上。从整体来看,白头叶猴在雨季对乔木的利用频率显著大于旱季（Z=-2.680,n=12,P=0.007）;雨季在山坡觅食频率显著大于旱季（Z=-2.517,n=12,P=0.012）,而在崖壁觅食频率刚好相反（Z=-2.842,n=12,P=0.004）;白头叶猴雨季在乔木休息的频率显著大于旱季（Z=-2.355,n=12,P=0.019）。白头叶猴对栖息地的利用受到温度的影响。白头叶猴对乔木的总体利用频率随着平均温度的升高而增加（r=0.664,n=12,P=0.018）;觅食时,对崖壁、裸岩的利用频率均与平均温度成负相关关系（崖壁：r=-0.685,n=12,P=0.014;裸岩：r=-0.600,n=12,P=0.039）;休息时,对乔木的利用频率与平均温度呈正相关关系（r=0.650,n=12,P=0.022）。不同季节,白头叶猴对栖息地的利用方式不同。白头叶猴的栖息地利用模式可能是在觅食利益和捕食风险之间作出的权衡,并受到环境温度的影响。     

厦门湾中华白海豚潜在生态廊道识别及人类活动干扰评估

生态廊道具有维持或恢复生态连通性的功能，对于连接生物栖息地、保护物种多样性具有重要意义。现有的生态廊道研究主要集中于陆地，而海洋生态系统具有水体广泛连通、缺乏直观的景观斑块等特点，导致海洋生态廊道的研究成为长期以来的科学难题。以栖息地位于厦门湾的国家一级保护动物中华白海豚（Sousa chinensis）为对象，尝试基于物种分布模型和最小成本路径分析法建立海洋生态廊道的识别方法。研究采用物种分布模型识别厦门湾内中华白海豚的适宜生境分布区和节点，并利用模型产出的生境适宜性结果生成海洋中的阻力表面，模拟计算节点与节点间在阻力表面上的最小成本路径，从而生成物种扩散网络。研究结果显示，厦门湾中华白海豚的分布主要受到航道距离、到岸线距离和叶绿素浓度三项因素的影响，主要适宜生境位于西海域至九龙江口和大嶝海域。潜在的核心生态廊道面积93.19km²，次级生态廊道面积170.41km²，九龙江口-鼓浪屿南侧-黄厝-大小嶝岛沿线可能是厦门湾中华白海豚的主要迁移路线。在此基础上，从用海空间重叠和桥梁影响两方面开展了人类活动对廊道的干扰评估。评估结果显示旅游活动和航运活动可能是中华白海豚生态廊道的主要影响来源，其中核心廊道受旅游活动影响更大，次级廊道受航运活动的影响更大。此外，厦门岛北部的桥梁亦可能会影响中华白海豚生态廊道的连通性。研究创新性地提出了海洋生态廊道的空间定量化识别方法并成功应用于厦门湾中华白海豚研究，研究结果可为我国海洋生物多样性保护、海洋保护区的空间规划等领域提供新的科学工具。     

基于MaxEnt模型的二郎山廊道大熊猫栖息地适宜性评价

基于地理分布点和环境变量数据,利用MaxEnt模型(Maximum Entropy Modeling)对二郎山廊道大熊猫Ailuropoda melanoleuca栖息地适宜度进行综合评价:坡度、坡向和干扰距离的综合贡献值分别为53.00%、21.90%和18.00%,累积贡献率达92.90%,是影响该区域大熊猫分布的主要生态因子。使用自然断点法将大熊猫栖息地分为不适宜、潜在、适宜和最适宜栖息地4种类型:不适宜栖息地面积为168.45 km~2,占研究区域面积的39.02%;潜在栖息地面积为212.71 km~2,占研究区域面积的49.28%;适宜栖息地和最适宜栖息地面积分别为49.79 km~2和0.68 km~2,两者仅占研究区域面积的11.70%。建议在该廊道开展大熊猫栖息地人工修复项目,以增加该地大熊猫的适宜栖息地面积;同时尽可能减少人类活动对大熊猫栖息地的干扰,以便廊道在促进各种群之间的基因交流中更好地发挥作用。     

大兴安岭南段马鹿生境适宜性分析与生态廊道构建

构建生态廊道在缓解生境破碎化对生物多样性的影响、维持濒危物种的遗传多样性、维护自然生态系统结构完整与功能稳定方面具有重要作用。以内蒙古大兴安岭南段分布的马鹿(Cervus elaphus)种群为研究对象,利用MaxEnt模型对其生境适宜性进行分析,并利用最小累积阻力模型构建潜在生态扩散廊道,探讨大兴安岭南段区域隔离马鹿种群的栖息地连通方案。结果显示,马鹿栖息地呈破碎化状态,种群有明显的隔离分布趋势,现有适宜栖息地具有海拔较低(800—1200 m)、坡度较缓(<15°)、靠近水源、植被类型多为靠近山林的灌丛或草地等特点。所构建12条生态廊道具有经过河流浅水节段、远离村落等特点,便于落实栖息地生态恢复管理措施。研究从区域尺度综合分析了大兴安岭南段马鹿栖息地现状及连通性,有助于优化适宜栖息地格局,促进马鹿扩散和栖息地连通,为该物种隔离种群及其栖息地保护规划提供现实指导和基础资料。     

中国东北虎栖息地分析与潜在生态廊道构建

东北虎（Panthera tigris altaica）是现存5个虎亚种中体型最大者,其作为全球生物多样性保护的旗舰物种,在维持健康生态系统功能中占据不可替代的重要地位。近几十年来,由于东北虎栖息地受到人类活动强烈干扰,致使栖息地破碎化,主要栖息地孤立分布,呈现岛状,天然生态廊道消失殆尽,东北虎的保护面临巨大挑战。因此,确定东北虎关键栖息地,构建与恢复东北虎栖息地之间的生态廊道十分必要。本研究运用专家模型结合东北虎栖息地选择规律和栖息地特征,综合分析植被类型、国家级与省级自然保护区分布、地形因子以及人为干扰因子共7个主要影响因子;通过层次分析法(AHP)获得各影响因子的相对权重值,运用加权线性方程获得了东北虎潜在适宜栖息地,并确定了东北虎核心分布区以及分布区间的综合代价值。通过廊道设计模型（Linkage mapper）得到东北虎核心栖息地间的潜在生态廊道。结果得到了21条东北虎潜在生态廊道,对打通国内零星分布区,特别是张广才岭-完达山-老爷岭之间的迁移通道、扩大东北虎生存空间具有现实指导意义。     

道路网络对物种迁移及景观连通性的影响

道路对区域景观产生破碎化及阻碍效应,使得生态过程发生受阻、景观功能受损。科学评估道路的生态效应对指导道路选线、生态恢复工程及景观规划有重要价值。本文采用阻力面概念和最小耗费模型,实现了景观阻力、斑块质量、廊道畅通性和景观连通性的计算,定量分析了物种迁移过程中道路网的干扰程度。结果表明:道路引起生态廊道长度平均增加30%,廊道曲率平均增加0.19;道路也使廊道通畅性平均降低90%,影响程度与廊道耗费值呈线性相关,系数为0.76;整体景观连通性下降30.8%,扩散能力越弱的廊道所受道路阻碍效应越强。本文建立的模拟评价方法可用于确定景观尺度上特定生态过程的关键格局,同时也可为景观生态规划中生态廊道设计及景观连通性定量分析提供科学依据。     

干旱内陆河流域生态安全格局的构建及优化——以石羊河流域为例

生态安全格局识别及构建是保障干旱区生态安全、实现可持续发展的基本空间途径。石羊河流域是典型的干旱内陆河流域，其生态环境极为脆弱、敏感。基于石羊河流域生态本底特征，选取并定量评估水资源、生物多样性、水土保持和沙漠化4种生态系统服务，识别生态源地；以建筑物指数和植被净初级生产力为阻力因子，应用加权叠加法构建基本阻力面，并运用最小累积阻力模型及水文分析法识别生态廊道和生态功能节点，进行生态安全格局构建及优化。结果表明：（1）2005-2015年，石羊河流域生态源地增加，生态环境质量趋于好转，特别是下游民勤绿洲区及上游祁连山区源地面积增加明显。2005、2010、2015年生态源地面积占流域总面积的比重分别为16.7%、14.7%、19.8%；（2）2005-2015年，生态廊道明显增加，流域整体生态安全格局网络趋向复杂完善。2005年和2010年提取的生态节点都为36个，2015年的生态节点为35个，生态廊道从2005年的23条增加到2015年47条，部分潜在廊道发展演化为廊道；（3）基于2005-2015年生态安全格局分析，构建了以"二带区、三绿洲、五廊道、多中心"为核心的"绿洲廊道功能区"的优化格局模式，以期为石羊河流域生态环境治理与恢复以及区域可持续发展规划提供借鉴与决策依据。     

基于景观格局和连接度评价的生态网络方法优化与应用

景观生态学中常凭借最小累积阻力模型构建目标种生态网络,以提升破碎栖息地间的景观连接度,缓解生境破碎化负面影响.但传统最小累积阻力生态网络方法缺乏对生态网络的效用验证,对研究地的景观结构变化与生态过程的影响认识不足.本研究运用景观格局指数与连接度概率指数,定量评价生态网络构建前后的研究地景观结构与连接度特征,并以崇左白头叶猴栖息地生态网络为例,详尽叙述此生态网络方法的优化与应用过程.通过对白头叶猴栖息地斑块进行辨认、踏脚石斑块识别,对研究地用地类型进行划分并进行阻力赋值,运用最小累积阻力模型生成了20条白头叶猴栖息地生态网络廊道;然后利用景观结构指数与连接度概率指数结合的方法,对生成的生态网络结构和功能连接度进行评价.结果表明: 凭借最小累积阻力模型构建的目标种生态网络,能有效提升栖息地生境的完整性和连续性,降低总体破碎化水平,并改善生境质量.同时,该生态网络构建能提升生境景观的结构连接度与功能连接度,且两方面的连接度变化在结果上具有极显著的一致性(R²=98.3%,P<0.01).生态网络带来的景观结构方面变化与功能连接度的关联性不强,两种指数间的相互关系不如结构与功能的内在关系显著.     

整合电路理论的生态廊道及其重要性识别

景观连接度被认为是影响诸多生态过程的一个重要因素.基于最小累积阻力模型的最小成本路径识别方法可以有效识别异质性景观中的功能连接,已被广泛应用到景观的功能连接评价与生态廊道模拟的研究中.基于电路理论的连接度模型用电阻代替了图论中的边、用电阻距离代替成本距离,来衡量异质性景观的功能连接.本文以SIMMAP 2.0软件生成的模拟景观为对象,借助于Linkage Mapper工具和Circuitscape软件,探讨如何将最小累积阻力模型与基于电路理论的连接度模型相结合来识别生态廊道及其景观要素的相对重要性.结果表明: 两种模型在应用中各有优势,互为补充.最小成本路径方法可以有效识别栖息地之间的最小成本廊道,基于电路理论的连接度模型通过电流密度的计算可以有效识别对景观连接性有重要影响的景观要素和“夹点”地区,并且“夹点”的位置不受廊道宽度的影响,在廊道重要性识别研究中具有明显优势.该方法可为区域生态保护规划和生态廊道设计提供科学依据.     

Transient effects of corridors on polygyne fire ants over a decade

1. Although corridors are frequently regarded as a way to mitigate the negative effects of habitat fragmentation, concerns persist that corridors may facilitate the spread of invasive species to the detriment of native species.
2. The invasive fire ant, Solenopsis invicta, has two social forms. The polygyne form has limited dispersal abilities relative to the monogyne form. Our previous work in a large-scale corridor experiment showed that in landscapes dominated by the polygyne form, fire ant density was higher and native ant species richness was lower in habitat patches connected by corridors than in unconnected patches.
3. We expected that these observed corridor effects would be transient, that is, that fire ant density and native ant species richness differences between connected and unconnected patches would diminish over time as fire ants eventually fully established within patches. We tested this prediction by resampling the three landscapes dominated by polygyne fire ants 6 to 11 years after our original study.
4. Differences in fire ant density between connected and unconnected habitat patches in these landscapes decreased, as expected. Differences in native ant species richness were variable but lowest in the last 2 years of sampling.
5. These findings support our prediction of transient corridor effects on this invasive ant and stress the importance of temporal dynamics in assessing population and community impacts of habitat connectivity.

     

Identifying habitat linkages to maintain connectivity for corridor dwellers in a fragmented landscape

Anthropogenic habitat fragmentation typically precedes conservation planning; maintaining remaining linkages among core habitat areas can thus become a key conservation objective. Identifying linkages for dispersal and ensuring those linkages have long-term protection and management are challenging tasks for wildlife managers. These tasks can be especially daunting for smaller species with low mobility, termed corridor dwellers, which must maintain sustainable populations within corridors. Between May 2007 and June 2009, we collected occurrence locations for a corridor dweller, the Palm Springs pocket mouse (Perognathus longimembris bangsii), from museums, previous research, and our own field sampling. We used those data to model their suitable niche space and then identify suitable linkages between proposed conservation areas. We used a partitioned Mahalanobis D² statistic to create a spatially explicit niche model describing the distribution of a suitable niche space, and we validated the model statistically, with live trapping and with burrowing owl (Athene cunnicularia) diets. Our model identified soil characteristics, topographic ruggedness, and vegetation as variables delimiting Palm Springs pocket mouse habitat; sand content of the soils was an especially important characteristic. Our historic distribution model identified 120,000–90,000 ha as historically potential Palm Springs pocket mouse habitat; roughly 39% of that has been lost to more recent development. Most of the remaining suitable habitat occurred in the northwestern portion of the valley. We modeled habitat within core reserves as well as within proposed linkages between those reserves as having high similarity to known occupied habitats. Live trapping in areas with high (≥0.95) Habitat Suitability Index (HSI) values resulted in captures at 66% of those locations and, along with burrowing owl diets, refined a qualitative model as to what constituted a suitable Palm Springs pocket mouse corridor. While most corridor analyses have focused on mobile species which may traverse corridors in hours, days, or weeks, linkages for corridor dwellers must include habitat for sustaining multi-generational populations. This requires evaluating whether continuous suitable habitat exists within proposed corridors. Our research demonstrates how niche modeling can provide a landscape-scale view of the distribution of suitable habitat to evaluate conservation objectives for connectivity. © 2011 The Wildlife Society.     

Opportunities of habitat connectivity for tiger (Panthera tigris) between Kanha and Pench National Parks in Madhya Pradesh, India

The Tiger (Panthera tigris) population in India has undergone a sharp decline during the last few years. Of the number of factors attributed to this decline, habitat fragmentation has been the most worrisome. Wildlife corridors have long been a subject of discussion amongst wildlife biologists and conservationists with contrasting schools of thought arguing their merits and demerits. However, it is largely believed that wildlife corridors can help minimize genetic isolation, offset fragmentation problems, improve animal dispersal, restore ecological processes and reduce man animal conflict. This study attempted to evaluate the possibilities of identifying a suitable wildlife corridor between two very important wildlife areas of central India--the Kanha National Park and the Pench National Park--with tiger as the focal species. Geographic Information System (GIS) centric Least Cost Path modeling was used to identify likely routes for movement of tigers. Habitat suitability, perennial water bodies, road density, railway tracks, human settlement density and total forest edge were considered as key variables influencing tiger movement across the Kanha-Pench landscape. Each of these variables was weighted in terms of relative importance through an expert consultation process. Using different importance scenarios, three alternate corridor routes were generated of which one was identified as the most promising for tiger dispersal. Weak links--where cover and habitat conditions are currently sub-optimal--were flagged on the corridor route. Interventions aimed at augmenting the identified corridor route have been suggested using accepted wildlife corridor design principles. The involvement of local communities through initiatives such as ecotourism has been stressed as a crucial long term strategy for conservation of the Kanha-Pench wildlife corridor. The results of the study indicate that restoration of the identified wildlife corridors between the two protected areas is technically feasible.     

Landscape influences on dispersal behaviour: a theoretical model and empirical test using the fire salamander,Salamandra infraimmaculata

When populations reside within a heterogeneous landscape, isolation by distance may not be a good predictor of genetic divergence if dispersal behaviour and therefore gene flow depend on landscape features. Commonly used approaches linking landscape features to gene flow include the least cost path (LCP), random walk (RW), and isolation by resistance (IBR) models. However, none of these models is likely to be the most appropriate for all species and in all environments. We compared the performance of LCP, RW and IBR models of dispersal with the aid of simulations conducted on artificially generated landscapes. We also applied each model to empirical data on the landscape genetics of the endangered fire salamander, Salamandra infraimmaculata, in northern Israel, where conservation planning requires an understanding of the dispersal corridors. Our simulations demonstrate that wide dispersal corridors of the low-cost environment facilitate dispersal in the IBR model, but inhibit dispersal in the RW model. In our empirical study, IBR explained the genetic divergence better than the LCP and RW models (partial Mantel correlation 0.413 for IBR, compared to 0.212 for LCP, and 0.340 for RW). Overall dispersal cost in salamanders was also well predicted by landscape feature slope steepness (76 %), and elevation (24 %). We conclude that fire salamander dispersal is well characterised by IBR predictions. Together with our simulation findings, these results indicate that wide dispersal corridors facilitate, rather than hinder, salamander dispersal. Comparison of genetic data to dispersal model outputs can be a useful technique in inferring dispersal behaviour from population genetic data.     

Evaluating Landscape Options for Corridor Restoration between Giant Panda Reserves

The establishment of corridors can offset the negative effects of habitat fragmentation by connecting isolated habitat patches. However, the practical value of corridor planning is minimal if corridor identification is not based on reliable quantitative information about species-environment relationships. An example of this need for quantitative information is planning for giant panda conservation. Although the species has been the focus of intense conservation efforts for decades, most corridor projects remain hypothetical due to the lack of reliable quantitative researches at an appropriate spatial scale. In this paper, we evaluated a framework for giant panda forest corridor planning. We linked our field survey data with satellite imagery, and conducted species occupancy modelling to examine the habitat use of giant panda within the potential corridor area. We then conducted least-cost and circuit models to identify potential paths of dispersal across the landscape, and compared the predicted cost under current conditions and alternative conservation management options considered during corridor planning. We found that due to giant panda''s association with areas of low elevation and flat terrain, human infrastructures in the same area have resulted in corridor fragmentation. We then identified areas with high potential to function as movement corridors, and our analysis of alternative conservation scenarios showed that both forest/bamboo restoration and automobile tunnel construction would significantly improve the effectiveness of corridor, while residence relocation would not significantly improve corridor effectiveness in comparison with the current condition. The framework has general value in any conservation activities that anticipate improving habitat connectivity in human modified landscapes. Specifically, our study suggested that, in this landscape, automobile tunnels are the best means to remove current barriers to giant panda movements caused by anthropogenic interferences.     

Connectivity of priority areas for the conservation of large carnivores in northern Mexico

The loss and fragmentation of habitat has negative effects on populations of large carnivores, but ecological corridors that allow dispersal of individuals among habitat remnants mitigate these effects. Our objectives were to identify 1) priority areas for the conservation of three species of large carnivores in northern Mexico, 2) the corridors that can maintain connectivity between them, and 3) pinch points signifying habitat loss that threatens connectivity. We generated species distribution models using MaxEnt and GLM to obtain a consensus model for each species. We applied an inverse function to the probability gradient of the consensus models to calculate the resistance and identify the corridors between priority areas. With Linkage Mapper software, we generated the corridors, calculated their centrality and that of the priority areas, and identified the areas where the corridors are narrower (i.e., pinch points). Finally, we identified the main anthropic fragmentation elements in the most important corridors. We identified 6 priority areas for jaguar, 20 for puma and 21 for black bear, with 5 corridors for jaguar, 22 for puma and 29 for black bear. The pinch points were produced by agricultural fields, human settlements, roads, or combinations of these factors. Depending on the element of fragmentation in each corridor, we propose specific strategies at the pinch points, e.g., applying restoration programs, including wildlife crossings to mitigate road killed cases, promoting payment programs for environmental services or compensation in cases of conflict, to increase the support of local inhabitants for conservation.     

Do forest-dwelling plant species disperse along landscape corridors?

Woody corridors in fragmented landscapes have been proposed as alternative habitats for forest plants, but the great variation in species-specific responses blurs the overall assessment. The aim of this study was to estimate the dispersal success of forest-dwelling plants from a stand into and along an attached woody corridor, and to explain the observed patterns from the point of view of species’ dispersal traits and corridor properties. We sampled 47 forest–corridor transects in the agricultural landscapes of southeastern Estonia. Regionally common forest-dwelling species (observed in at least 10 % of seed-source forests) were classified on the basis of their ecological response profile—forest-restricted species (F-type) and forest-dwelling generalists (G-type). Species richness and the proportion of F-type species decreased sharply from the seed-source forest core to the forest edge and to the first 10–15 m of the corridor, while G-type species richness remained constant throughout the transect. Corridor structure had a species-specific effect—F species were promoted by old (≥50 years) and wide (≥10 m) corridors, while G species were supported by young and narrow corridors with ditch-related soil disturbances. Moderate shade (canopy cover <75 %) was optimal for all forest-dwelling species. Large dispersule weight, and not seed weight, dispersal vector or Ellenberg’s indicator values, was the trait that differentiated F species from G species. We conclude that most woody corridors are only dispersal stepping-stone habitats for habitat generalist species, and not for specialists. Only century old corridors can relieve the dispersal limitation of forest-restricted species.     

Prediction and error in multi‐stage models for spread of aquatic non‐indigenous species

Aim Predictions of spread of non‐indigenous species allow for greater efficiency in managing invasions by targeting areas for preventative measures. The invasion sequence is a useful concept in predictions of spread, as it allows us to test hypotheses about the transport and establishment of propagules in novel habitats. Our aims are twofold: (1) to develop and validate multi‐stage invasion models for the introduced fishhook waterflea, Cercopagis pengoi, and (2) to assess how variability in the transport patterns of the propagules influences the accuracy and spatial extent for predictions of spread. Location New York State, USA. Methods We developed a two‐stage model for the spread of C. pengoi. First, we developed a stochastic gravity model for dispersal based on surveys of recreational boat traffic in New York State as a proxy for propagule pressure. We then modelled the probability of establishment based on predicted levels of propagule pressure and measures of lakes’ physicochemistry. In addition, we used Monte Carlo simulations based on the gravity model to propagate variability in boater traffic through the establishment model to assess how uncertainty in dispersal influenced predictions of spread. Results The amount recreationalists were willing to spend, lake area and population size of the city nearest to the destination lake were significant factors affecting boater traffic. In turn, boater traffic, lake area, specific conductance and turbidity were significant predictors of establishment. The inclusion of stochastic dispersal reduced the rate of false positives (i.e. incorrect prediction of an invasion) in detecting invasions at the upper 95% prediction interval for the probability of establishment. Main conclusions Combinations of measures of propagule pressure, habitat suitability and stochastic dispersal allow for the most accurate predictions of spread. Further, multi‐stage spread models may overestimate the extent of spread if stochasticity in early stages of the models is not considered.