A Comparison of Supervised Classification Methods for the Prediction of Substrate Type Using Multibeam Acoustic and Legacy Grain-Size Data

Detailed seabed substrate maps are increasingly in demand for effective planning and management of marine ecosystems and resources. It has become common to use remotely sensed multibeam echosounder data in the form of bathymetry and acoustic backscatter in conjunction with ground-truth sampling data to inform the mapping of seabed substrates. Whilst, until recently, such data sets have typically been classified by expert interpretation, it is now obvious that more objective, faster and repeatable methods of seabed classification are required. This study compares the performances of a range of supervised classification techniques for predicting substrate type from multibeam echosounder data. The study area is located in the North Sea, off the north-east coast of England. A total of 258 ground-truth samples were classified into four substrate classes. Multibeam bathymetry and backscatter data, and a range of secondary features derived from these datasets were used in this study. Six supervised classification techniques were tested: Classification Trees, Support Vector Machines, k-Nearest Neighbour, Neural Networks, Random Forest and Naive Bayes. Each classifier was trained multiple times using different input features, including i) the two primary features of bathymetry and backscatter, ii) a subset of the features chosen by a feature selection process and iii) all of the input features. The predictive performances of the models were validated using a separate test set of ground-truth samples. The statistical significance of model performances relative to a simple baseline model (Nearest Neighbour predictions on bathymetry and backscatter) were tested to assess the benefits of using more sophisticated approaches. The best performing models were tree based methods and Naive Bayes which achieved accuracies of around 0.8 and kappa coefficients of up to 0.5 on the test set. The models that used all input features didn''t generally perform well, highlighting the need for some means of feature selection.     

The Influence of Data Resolution on Predicted Distribution and Estimates of Extent of Current Protection of Three ‘Listed’ Deep-Sea Habitats

Modelling approaches have the potential to significantly contribute to the spatial management of the deep-sea ecosystem in a cost effective manner. However, we currently have little understanding of the accuracy of such models, developed using limited data, of varying resolution. The aim of this study was to investigate the performance of predictive models constructed using non-simulated (real world) data of different resolution. Predicted distribution maps for three deep-sea habitats were constructed using MaxEnt modelling methods using high resolution multibeam bathymetric data and associated terrain derived variables as predictors. Model performance was evaluated using repeated 75/25 training/test data partitions using AUC and threshold-dependent assessment methods. The overall extent and distribution of each habitat, and the percentage contained within an existing MPA network were quantified and compared to results from low resolution GEBCO models. Predicted spatial extent for scleractinian coral reef and Syringammina fragilissima aggregations decreased with an increase in model resolution, whereas Pheronema carpenteri total suitable area increased. Distinct differences in predicted habitat distribution were observed for all three habitats. Estimates of habitat extent contained within the MPA network all increased when modelled at fine scale. High resolution models performed better than low resolution models according to threshold-dependent evaluation. We recommend the use of high resolution multibeam bathymetry data over low resolution bathymetry data for use in modelling approaches. We do not recommend the use of predictive models to produce absolute values of habitat extent, but likely areas of suitable habitat. Assessments of MPA network effectiveness based on calculations of percentage area protection (policy driven conservation targets) from low resolution models are likely to be fit for purpose.     

Prediction model of random forest for the risk of hyperuricemia in a Chinese basic health checkup test

Objectives: The present study aimed to develop a random forest (RF) based prediction model for hyperuricemia (HUA) and compare its performance with the conventional logistic regression (LR) model. Methods: This cross-sectional study recruited 91,690 participants (14,032 with HUA, 77,658 without HUA). We constructed a RF-based prediction model in the training sets and evaluated it in the validation sets. Performance of the RF model was compared with the LR model by receiver operating characteristic (ROC) curve analysis. Results: The sensitivity and specificity of the RF models were 0.702 and 0.650 in males, 0.767 and 0.721 in females. The positive predictive value (PPV) and negative predictive value (NPV) were 0.372 and 0.881 in males, 0.159 and 0.978 in females. AUC of the RF models was 0.739 (0.728–0.750) in males and 0.818 (0.799–0.837) in females. AUC of the LR models were 0.730 (0.718–0.741) for males and 0.815 (0.795–0.835) for females. The predictive power of RF was slightly higher than that of LR, but was not statistically significant in females (Delong tests, P=0.0015 for males, P=0.5415 for females). Conclusion: Compared with LR, the good performance in HUA status prediction and the tolerance of features associations or interactions showed great potential of RF in further application. A prospective cohort is necessary for HUA developing prediction. People with high risk factors should be encouraged to actively control to reduce the probability of developing HUA.     

Towards Quantitative Spatial Models of Seabed Sediment Composition

There is a need for fit-for-purpose maps for accurately depicting the types of seabed substrate and habitat and the properties of the seabed for the benefits of research, resource management, conservation and spatial planning. The aim of this study is to determine whether it is possible to predict substrate composition across a large area of seabed using legacy grain-size data and environmental predictors. The study area includes the North Sea up to approximately 58.44°N and the United Kingdom’s parts of the English Channel and the Celtic Seas. The analysis combines outputs from hydrodynamic models as well as optical remote sensing data from satellite platforms and bathymetric variables, which are mainly derived from acoustic remote sensing. We build a statistical regression model to make quantitative predictions of sediment composition (fractions of mud, sand and gravel) using the random forest algorithm. The compositional data is analysed on the additive log-ratio scale. An independent test set indicates that approximately 66% and 71% of the variability of the two log-ratio variables are explained by the predictive models. A EUNIS substrate model, derived from the predicted sediment composition, achieved an overall accuracy of 83% and a kappa coefficient of 0.60. We demonstrate that it is feasible to spatially predict the seabed sediment composition across a large area of continental shelf in a repeatable and validated way. We also highlight the potential for further improvements to the method.     

Predicted Deep-Sea Coral Habitat Suitability for the U.S. West Coast

Regional scale habitat suitability models provide finer scale resolution and more focused predictions of where organisms may occur. Previous modelling approaches have focused primarily on local and/or global scales, while regional scale models have been relatively few. In this study, regional scale predictive habitat models are presented for deep-sea corals for the U.S. West Coast (California, Oregon and Washington). Model results are intended to aid in future research or mapping efforts and to assess potential coral habitat suitability both within and outside existing bottom trawl closures (i.e. Essential Fish Habitat (EFH)) and identify suitable habitat within U.S. National Marine Sanctuaries (NMS). Deep-sea coral habitat suitability was modelled at 500 m×500 m spatial resolution using a range of physical, chemical and environmental variables known or thought to influence the distribution of deep-sea corals. Using a spatial partitioning cross-validation approach, maximum entropy models identified slope, temperature, salinity and depth as important predictors for most deep-sea coral taxa. Large areas of highly suitable deep-sea coral habitat were predicted both within and outside of existing bottom trawl closures and NMS boundaries. Predicted habitat suitability over regional scales are not currently able to identify coral areas with pin point accuracy and probably overpredict actual coral distribution due to model limitations and unincorporated variables (i.e. data on distribution of hard substrate) that are known to limit their distribution. Predicted habitat results should be used in conjunction with multibeam bathymetry, geological mapping and other tools to guide future research efforts to areas with the highest probability of harboring deep-sea corals. Field validation of predicted habitat is needed to quantify model accuracy, particularly in areas that have not been sampled.     

The Predictive Performance and Stability of Six Species Distribution Models

Background

Predicting species’ potential geographical range by species distribution models (SDMs) is central to understand their ecological requirements. However, the effects of using different modeling techniques need further investigation. In order to improve the prediction effect, we need to assess the predictive performance and stability of different SDMs.

Methodology

We collected the distribution data of five common tree species (Pinus massoniana, Betula platyphylla, Quercus wutaishanica, Quercus mongolica and Quercus variabilis) and simulated their potential distribution area using 13 environmental variables and six widely used SDMs: BIOCLIM, DOMAIN, MAHAL, RF, MAXENT, and SVM. Each model run was repeated 100 times (trials). We compared the predictive performance by testing the consistency between observations and simulated distributions and assessed the stability by the standard deviation, coefficient of variation, and the 99% confidence interval of Kappa and AUC values.

Results

The mean values of AUC and Kappa from MAHAL, RF, MAXENT, and SVM trials were similar and significantly higher than those from BIOCLIM and DOMAIN trials (p<0.05), while the associated standard deviations and coefficients of variation were larger for BIOCLIM and DOMAIN trials (p<0.05), and the 99% confidence intervals for AUC and Kappa values were narrower for MAHAL, RF, MAXENT, and SVM. Compared to BIOCLIM and DOMAIN, other SDMs (MAHAL, RF, MAXENT, and SVM) had higher prediction accuracy, smaller confidence intervals, and were more stable and less affected by the random variable (randomly selected pseudo-absence points).

Conclusions

According to the prediction performance and stability of SDMs, we can divide these six SDMs into two categories: a high performance and stability group including MAHAL, RF, MAXENT, and SVM, and a low performance and stability group consisting of BIOCLIM, and DOMAIN. We highlight that choosing appropriate SDMs to address a specific problem is an important part of the modeling process.     

Random forests, a novel approach for discrimination of fish populations using parasites as biological tags

Due to the complexity of host-parasite relationships, discrimination between fish populations using parasites as biological tags is difficult. This study introduces, to our knowledge for the first time, random forests (RF) as a new modelling technique in the application of parasite community data as biological markers for population assignment of fish. This novel approach is applied to a dataset with a complex structure comprising 763 parasite infracommunities in population samples of Atlantic cod, Gadus morhua, from the spawning/feeding areas in five regions in the North East Atlantic (Baltic, Celtic, Irish and North seas and Icelandic waters). The learning behaviour of RF is evaluated in comparison with two other algorithms applied to class assignment problems, the linear discriminant function analysis (LDA) and artificial neural networks (ANN). The three algorithms are used to develop predictive models applying three cross-validation procedures in a series of experiments (252 models in total). The comparative approach to RF, LDA and ANN algorithms applied to the same datasets demonstrates the competitive potential of RF for developing predictive models since RF exhibited better accuracy of prediction and outperformed LDA and ANN in the assignment of fish to their regions of sampling using parasite community data. The comparative analyses and the validation experiment with a 'blind' sample confirmed that RF models performed more effectively with a large and diverse training set and a large number of variables. The discrimination results obtained for a migratory fish species with largely overlapping parasite communities reflects the high potential of RF for developing predictive models using data that are both complex and noisy, and indicates that it is a promising tool for parasite tag studies. Our results suggest that parasite community data can be used successfully to discriminate individual cod from the five different regions of the North East Atlantic studied using RF.     

A novel index to aid in prioritizing habitats for site‐based conservation

Funding biodiversity conservation strategies are usually minimal, thus prioritizing habitats at high risk should be conducted. We developed and tested a conservation priority index (CPI) that ranks habitats to aid in prioritizing them for conservation. We tested the index using 1897 fish species from 273 African inland lakes and 34 countries. In the index, lake surface area, rarity, and their International Union for Conservation of Nature (IUCN) Red List status were incorporated. We retrieved data from the Global Biodiversity Information Facility (GBIF) and IUCN data repositories. Lake Nyasa had the highest species richness (424), followed by Tanganyika (391), Nokoué (246), Victoria (216), and Ahémé (216). However, lakes Otjikoto and Giunas had the highest CPI of 137.2 and 52.1, respectively. Lakes were grouped into high priority (CPI > 0.5; n = 56) and low priority (CPI < 0.5; n = 217). The median surface area between priority classes was significantly different (W = 11,768, p < .05, effect size = 0.65). Prediction accuracy of Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) for priority classes were 0.912 and 0.954, respectively. Both models exhibited lake surface area as the variable with the highest importance. CPI generally increased with a decrease in lake surface area. This was attributed to less ecological substitutability and higher exposure levels of anthropogenic stressors such as pollution to a species in smaller lakes. Also, the highest species richness per unit area was recorded for high‐priority lakes. Thus, smaller habitats or lakes may be prioritized for conservation although larger waterbodies or habitats should not be ignored. The index can be customized to local, regional, and international scales as well as marine and terrestrial habitats.     

Life History Traits and Niche Instability Impact Accuracy and Temporal Transferability for Historically Calibrated Distribution Models of North American Birds

A primary assumption of environmental niche models (ENMs) is that models are both accurate and transferable across geography or time; however, recent work has shown that models may be accurate but not highly transferable. While some of this is due to modeling technique, individual species ecologies may also underlie this phenomenon. Life history traits certainly influence the accuracy of predictive ENMs, but their impact on model transferability is less understood. This study investigated how life history traits influence the predictive accuracy and transferability of ENMs using historically calibrated models for birds. In this study I used historical occurrence and climate data (1950-1990s) to build models for a sample of birds, and then projected them forward to the ‘future’ (1960-1990s). The models were then validated against models generated from occurrence data at that ‘future’ time. Internal and external validation metrics, as well as metrics assessing transferability, and Generalized Linear Models were used to identify life history traits that were significant predictors of accuracy and transferability. This study found that the predictive ability of ENMs differs with regard to life history characteristics such as range, migration, and habitat, and that the rarity versus commonness of a species affects the predicted stability and overlap and hence the transferability of projected models. Projected ENMs with both high accuracy and transferability scores, still sometimes suffered from over- or under- predicted species ranges. Life history traits certainly influenced the accuracy of predictive ENMs for birds, but while aspects of geographic range impact model transferability, the mechanisms underlying this are less understood.     

随机森林算法基本思想及其在生态学中的应用——以云南松分布模拟为例

通常来讲,生态学者对于解释生态关系、描述格局和过程、进行空间或时间预测比较感兴趣。这些工作可以通过模拟输出值(响应)与一些特征值(即解释变量)的关系来实现。然而,生态数据模拟遇到了挑战,这是因为响应变量和预测变量可能是连续变量或离散变量。需要解释的生态关系通常是非线性的,并且解释变量之间具有复杂的相互作用关系。响应变量和解释变量存在缺失值并不是不常有的现象,奇异值也经常出现在生态数据中。此外,生态学者通常希望生态模型即要易于建立又易要于解释。通常是利用多种统计方法来分析处理各种各样情景中出现的独特的生态问题,这些模型包括(多元)逻辑回归、线性模型、生存模型、方差分析等等。随机森林是一个可以处理所有这些问题的有效方法。随机森林可以用来做分类、聚类、回归和生存分析、评估变量的重要性、检测数据中的奇异值、对缺失数据进行插补等。鉴于随机森林本身在算法上的优势,将就随机森林在生态学中的应用进行总结,对建模过程进行概述,并以云南松分布模拟研究为例,对其主要功能特点进行案例展示。通过对随机森林的一般术语、概念和建模思想进行介绍,有利于读者掌握本方法的应用本质,可以预见随机森林在生态学研究中将得到更多的应用和发展。     

Detection of hard and soft selective sweeps from Drosophila melanogaster population genomic data

Whether hard sweeps or soft sweeps dominate adaptation has been a matter of much debate. Recently, we developed haplotype homozygosity statistics that (i) can detect both hard and soft sweeps with similar power and (ii) can classify the detected sweeps as hard or soft. The application of our method to population genomic data from a natural population of Drosophila melanogaster (DGRP) allowed us to rediscover three known cases of adaptation at the loci Ace, Cyp6g1, and CHKov1 known to be driven by soft sweeps, and detected additional candidate loci for recent and strong sweeps. Surprisingly, all of the top 50 candidates showed patterns much more consistent with soft rather than hard sweeps. Recently, Harris et al. 2018 criticized this work, suggesting that all the candidate loci detected by our haplotype statistics, including the positive controls, are unlikely to be sweeps at all and that instead these haplotype patterns can be more easily explained by complex neutral demographic models. They also claim that these neutral non-sweeps are likely to be hard instead of soft sweeps. Here, we reanalyze the DGRP data using a range of complex admixture demographic models and reconfirm our original published results suggesting that the majority of recent and strong sweeps in D. melanogaster are first likely to be true sweeps, and second, that they do appear to be soft. Furthermore, we discuss ways to take this work forward given that most demographic models employed in such analyses are necessarily too simple to capture the full demographic complexity, while more realistic models are unlikely to be inferred correctly because they require a large number of free parameters.     

Lack of general associations between intertidal assemblages and rock hardness

Different assemblages of intertidal biota may be associated with the hardness of the rock type that comprises the seashore. However, very few published studies have investigated mobile and sessile assemblage differences between hardness classes. To remedy this, we investigated the physical attributes and biotic assemblages of 12 rock platforms across two bioregions, encompassing seven rock types categorised into either the soft (≤4 using Moh’s scale of scratch hardness) or hard (>4 on Moh’s scale) classes. Rock types from the soft versus hard classes differed physically, but the biotic assemblages showed few general differences between hardness classes. Instead, most biotic differences associated with rock hardness were specific to just one of the marine bioregions sampled, not both. Hardness-related assemblage differences were only weakly-to-moderately correlated with hardness differences in mineralogy or microhabitat density. The detection of bioregion-specific hardness differences for intertidal assemblages, rather than general hardness class trends, indicate that the type of rock comprising the seashore in each bioregion may be more strongly associated with the biotic patterns identified than hardness per se.     

Genetic variation in the hTAS2R38 taste receptor and food consumption among Finnish adults

Genetic variation in bitter taste receptors, such as hTAS2R38, may affect food preferences and intake. The aim of the present study was to investigate the association between bitter taste receptor haplotypes and the consumption of vegetables, fruits, berries and sweet foods among an adult Finnish population. A cross-sectional design utilizing data from the Cardiovascular Risk in Young Finns cohort from 2007, which consisted of 1,903 men and women who were 30–45 years of age from five different regions in Finland, was employed. DNA was extracted from blood samples, and hTAS2R38 polymorphisms were determined based on three SNPs (rs713598, rs1726866 and rs10246939). Food consumption was assessed with a validated food frequency questionnaire. The prevalence of the bitter taste-sensitive (PAV/PAV) haplotype was 11.3 % and that of the insensitive (AVI/AVI) haplotype was 39.5 % among this Finnish population. PAV homozygotic women consumed fewer vegetables than did the AVI homozygotic women, 269 g/day (SD 131) versus 301 g/day (SD 187), respectively, p = 0.03 (multivariate ANOVA). Furthermore, the intake of sweet foods was higher among the PAV homozygotes of both genders. Fruit and berry consumption did not differ significantly between the haplotypes in either gender. Individuals perceive foods differently, and this may influence their patterns of food consumption. This study showed that the hTAS2R38 taste receptor gene variation was associated with vegetable and sweet food consumption among adults in a Finnish population.     

Utility of Intraoperative Frozen Section in the Diagnosis of Periprosthetic Joint Infection

Purpose

Intraoperative frozen section (FS) is an effective diagnostic test for periprosthetic joint infection (PJI). We evaluated the diagnostic characteristics of single- and multiplex-site intraoperative FS, and evaluated the results of single-site FS combined with those of C-reactive protein (CRP) level and erythrocyte sedimentation rate (ESR) for assessing PJI.

Methods

We studied 156 painful joint arthroplasties in 152 consecutive patients presenting for revision total joint arthroplasty due to PJI. Receiver operating characteristic analysis was used to determine the optimal cutoff values for CRP level, ESR, and intraoperative FS histopathology. Sensitivity, specificity, positive and negative predictive values, and accuracy of the diagnostic tests were assessed using a 2×2 table.

Results

We investigated the diagnostic utility of polymorphonuclear leukocyte number (PMN) per high-power field (HPF) on FS. Our data showed that 5 PMNs per HPF is a suitable diagnostic threshold, with a high accuracy in single- and multiplex-site FS. Five PMNs in any 1 of 5 sites had the highest sensitivity of 0.86 and a specificity of 0.96. Five PMNs in every 1 of 5 sites had greater diagnostic utility, with a specificity of 1; however, the sensitivity of this measure fell to 0.62. Five PMNs in single-site FS had a sensitivity of 0.70 and a specificity of 0.94. Five PMNs in single-site FS or CRP level ≥15 mg/L increased the sensitivity to 0.92; however, the specificity decreased to 0.79.

Conclusion

Compared with single-site FS, any 1 positive site on multiplex-site FS may improve sensitivity, while every 1 positive site on multiplex-site FS may improve specificity. Five PMNs in any 1 of 5 sites on FS has excellent utility for the diagnosis of PJI. Additional systematic large-scale studies are needed to verify this result.     

Cost effective prediction of the eutrophication status of lakes and reservoirs

1. Eutrophication is a serious threat in many parts of the world, and identifying the environmental factors that determine the spatial distribution of eutrophicated waterbodies as well as the development of management tools is a challenge. 2. In this study, data from the Ile‐de‐France region were analysed to determine if catchment scale environmental variables could predict concentrations of chlorophyll a (used as a proxy for eutrophication status) of artificial lakes and reservoirs. 3. General additive models (GAM) and random forest models (RF) displayed greater predictive power than generalised linear models, indicating the importance of non‐monotonic relationships. Using RF modelling, very high predictive accuracy was achieved for both continuous and binomial (eutrophic or not) response variables (continuous: R² = 0.715; binomial: kappa = 0.764, 89% of waterbodies were accurately predicted). The better predictive power and robustness of RF versus GAM was attributed to the formers ability to better handle complex interactions between predictors and to account for threshold effects. 4. Our results confirmed the close link between the water quality of lakes and reservoirs and the characteristics of their catchments. Moreover, we also showed that (i) simple (e.g. linear and/or monotonic) relationships between catchment land use and water quality were only found for sub‐regional datasets, and (ii) land use needs to be considered in association with complementary environmental variables (hydromorphological variables) to best assess its impact on water quality.     

Pretreatment CT and PET radiomics predicting rectal cancer patients in response to neoadjuvant chemoradiotherapy

BackgroundThe purpose of this study was to characterize pre-treatment non-contrast computed tomography (CT) and ¹⁸F-fluorodeoxyglucose positron emission tomography (PET) based radiomics signatures predictive of pathological response and clinical outcomes in rectal cancer patients treated with neoadjuvant chemoradiotherapy (NACR T).Materials and methodsAn exploratory analysis was performed using pre-treatment non-contrast CT and PET imaging dataset. The association of tumor regression grade (TRG) and neoadjuvant rectal (NAR) score with pre-treatment CT and PET features was assessed using machine learning algorithms. Three separate predictive models were built for composite features from CT + PET.ResultsThe patterns of pathological response were TRG 0 (n = 13; 19.7%), 1 (n = 34; 51.5%), 2 (n = 16; 24.2%), and 3 (n = 3; 4.5%). There were 20 (30.3%) patients with low, 22 (33.3%) with intermediate and 24 (36.4%) with high NAR scores. Three separate predictive models were built for composite features from CT + PET and analyzed separately for clinical endpoints. Composite features with α = 0.2 resulted in the best predictive power using logistic regression. For pathological response prediction, the signature resulted in 88.1% accuracy in predicting TRG 0 vs. TRG 1–3; 91% accuracy in predicting TRG 0–1 vs. TRG 2–3. For the surrogate of DFS and OS, it resulted in 67.7% accuracy in predicting low vs. intermediate vs. high NAR scores.ConclusionThe pre-treatment composite radiomics signatures were highly predictive of pathological response in rectal cancer treated with NACR T. A larger cohort is warranted for further validation.     

100% Classification Accuracy Considered Harmful: The Normalized Information Transfer Factor Explains the Accuracy Paradox

The most widely spread measure of performance, accuracy, suffers from a paradox: predictive models with a given level of accuracy may have greater predictive power than models with higher accuracy. Despite optimizing classification error rate, high accuracy models may fail to capture crucial information transfer in the classification task. We present evidence of this behavior by means of a combinatorial analysis where every possible contingency matrix of 2, 3 and 4 classes classifiers are depicted on the entropy triangle, a more reliable information-theoretic tool for classification assessment.Motivated by this, we develop from first principles a measure of classification performance that takes into consideration the information learned by classifiers. We are then able to obtain the entropy-modulated accuracy (EMA), a pessimistic estimate of the expected accuracy with the influence of the input distribution factored out, and the normalized information transfer factor (NIT), a measure of how efficient is the transmission of information from the input to the output set of classes.The EMA is a more natural measure of classification performance than accuracy when the heuristic to maximize is the transfer of information through the classifier instead of classification error count. The NIT factor measures the effectiveness of the learning process in classifiers and also makes it harder for them to “cheat” using techniques like specialization, while also promoting the interpretability of results. Their use is demonstrated in a mind reading task competition that aims at decoding the identity of a video stimulus based on magnetoencephalography recordings. We show how the EMA and the NIT factor reject rankings based in accuracy, choosing more meaningful and interpretable classifiers.     

Downscaling Pest Risk Analyses: Identifying Current and Future Potentially Suitable Habitats for Parthenium hysterophorus with Particular Reference to Europe and North Africa

Pest Risk Assessments (PRAs) routinely employ climatic niche models to identify endangered areas. Typically, these models consider only climatic factors, ignoring the ‘Swiss Cheese’ nature of species ranges due to the interplay of climatic and habitat factors. As part of a PRA conducted for the European and Mediterranean Plant Protection Organization, we developed a climatic niche model for Parthenium hysterophorus, explicitly including the effects of irrigation where it was known to be practiced. We then downscaled the climatic risk model using two different methods to identify the suitable habitat types: expert opinion (following the EPPO PRA guidelines) and inferred from the global spatial distribution. The PRA revealed a substantial risk to the EPPO region and Central and Western Africa, highlighting the desirability of avoiding an invasion by P. hysterophorus. We also consider the effects of climate change on the modelled risks. The climate change scenario indicated the risk of substantial further spread of P. hysterophorus in temperate northern hemisphere regions (North America, Europe and the northern Middle East), and also high elevation equatorial regions (Western Brazil, Central Africa, and South East Asia) if minimum temperatures increase substantially. Downscaling the climate model using habitat factors resulted in substantial (approximately 22–53%) reductions in the areas estimated to be endangered. Applying expert assessments as to suitable habitat classes resulted in the greatest reduction in the estimated endangered area, whereas inferring suitable habitats factors from distribution data identified more land use classes and a larger endangered area. Despite some scaling issues with using a globally conformal Land Use Systems dataset, the inferential downscaling method shows promise as a routine addition to the PRA toolkit, as either a direct model component, or simply as a means of better informing an expert assessment of the suitable habitat types.