Accuracy of plant assemblage predictions from species distribution models varies along environmental gradients

Aim: Climatic niche modelling of species and community distributions implicitly assumes strong and constant climatic determinism across geographic space. This assumption had however never been tested so far. We tested it by assessing how stacked-species distribution models (S-SDMs) perform for predicting plant species assemblages along elevation. Location: Western Swiss Alps. Methods: Using robust presence-absence data, we first assessed the ability of topo-climatic S-SDMs to predict plant assemblages in a study area encompassing a 2800 m wide elevation gradient. We then assessed the relationships among several evaluation metrics and trait-based tests of community assembly rules. Results: The standard errors of individual SDMs decreased significantly towards higher elevations. Overall, the S-SDM overpredicted far more than they underpredicted richness and could not reproduce the humpback curve along elevation. Overprediction was greater at low and mid-range elevations in absolute values but greater at high elevations when standardised by the actual richness. Looking at species composition, the evaluation metrics accounting for both the presence and absence of species (overall prediction success and kappa) or focusing on correctly predicted absences (specificity) increased with increasing elevation, while the metrics focusing on correctly predicted presences (Jaccard index and sensitivity) decreased. The best overall evaluation - as driven by specificity - occurred at high elevation where species assemblages were shown to be under significant environmental filtering of small plants. In contrast, the decreased overall accuracy in the lowlands was associated with functional patterns representing any type of assembly rule (environmental filtering, limiting similarity or null assembly). Main Conclusions: Our study reveals interesting patterns of change in S-SDM errors with changes in assembly rules along elevation. Yet, significant levels of assemblage prediction errors occurred throughout the gradient, calling for further improvement of SDMs, e.g., by adding key environmental filters that act at fine scales and developing approaches to account for variations in the influence of predictors along environmental gradients.

Cellular distribution of Hsp70 expression in rat skeletal muscles. Effects of moderate exercise training and chronic hypoxia.

Rat hindlimb muscles constitutively express the inducible heat shock protein 72 (Hsp70), apparently in proportion to the slow myosin content. Since it remains controversial whether chronic Hsp70 expression reflects the overimposed stress, we investigated Hsp70 cellular distribution in fast muscles of the posterior rat hindlimb after (1) mild exercise training (up to 30 m/min treadmill run for 1 h/day), which induces a remodeling in fast fiber composition, or (2) prolonged exposure to normobaric hypoxia (10%O(2)), which does not affect fiber-type composition. Both conditions increased significantly protein Hsp70 levels in the skeletal muscle. Immunohistochemistry showed the labeling for Hsp70 in subsets of both slow/type 1 and fast/type 2A myofibers of control, sedentary, and normoxic rats. Endurance training increased about threefold the percentage of Hsp70-positive myofibers (P < 0.001), and changed the distribution of Hsp70 immunoreactivity, which involved a larger subset of both type 2A and intermediate type 2A/2X myofibers (P < 0.001) and vascular smooth muscle cells. Hypoxia induced Hsp70 immunoreactivity in smooth muscle cells of veins and did not increase the percentage of Hsp70-positive myofibers; however, sustained exposure to hypoxia affected the distribution of Hsp70 immunoreactivity, which appeared detectable in a very small subset of type 2A fibers, whereas it concentrated in type 1 myofibers (P < 0.05) together with the labeling for heme-oxygenase isoform 1, a marker of oxidative stress. Therefore, the chronic induction of Hsp70 expression in rat skeletal muscles is not obligatory related to the slow fiber phenotype but reveals the occurrence of a stress response.

Niche, distribution and global change : modeling insights into biogeography and conservation biology Olivier Broennimann

RESUME Les évidences montrant que les changements globaux affectent la biodiversité s'accumulent. Les facteurs les plus influant dans ce processus sont les changements et destructions d'habitat, l'expansion des espèces envahissantes et l'impact des changements climatiques. Une évaluation pertinente de la réponse des espèces face à ces changements est essentielle pour proposer des mesures permettant de réduire le déclin actuel de la biodiversité. La modélisation de la répartition d'espèces basée sur la niche (NBM) est l'un des rares outils permettant cette évaluation. Néanmoins, leur application dans le contexte des changements globaux repose sur des hypothèses restrictives et demande une interprétation critique. Ce travail présente une série d'études de cas investiguant les possibilités et limitations de cette approche pour prédire l'impact des changements globaux. Deux études traitant des menaces sur les espèces rares et en danger d'extinction sont présentées. Les caractéristiques éco-géographiques de 118 plantes avec un haut degré de priorité de conservation sont revues. La prévalence des types de rareté sont analysées en relation avec leur risque d'extinction UICN. La revue souligne l'importance de la conservation à l'échelle régionale. Une évaluation de la rareté à échelle globale peut être trompeuse pour certaine espèces car elle ne tient pas en compte des différents degrés de rareté que présente une espèce à différentes échelles spatiales. La deuxième étude test une approche pour améliorer l'échantillonnage d'espèces rares en incluant des phases itératives de modélisation et d'échantillonnage sur le terrain. L'application de l'approche en biologie de la conservation (illustrée ici par le cas du chardon bleu, Eryngium alpinum), permettrait de réduire le temps et les coûts d'échantillonnage. Deux études sur l'impact des changements climatiques sur la faune et la flore africaine sont présentées. La première étude évalue la sensibilité de 227 mammifères africains face aux climatiques d'ici 2050. Elle montre qu'un nombre important d'espèces pourrait être bientôt en danger d'extinction et que les parcs nationaux africains (principalement ceux situé en milieux xériques) pourraient ne pas remplir leur mandat de protection de la biodiversité dans le futur. La seconde étude modélise l'aire de répartition en 2050 de 975 espèces de plantes endémiques du sud de l'Afrique. L'étude propose l'inclusion de méthodes améliorant la prédiction des risques liés aux changements climatiques. Elle propose également une méthode pour estimer a priori la sensibilité d'une espèce aux changements climatiques à partir de ses propriétés écologiques et des caractéristiques de son aire de répartition. Trois études illustrent l'utilisation des modèles dans l'étude des invasions biologiques. Une première étude relate l'expansion de la laitue sáuvage (Lactuca serriola) vers le nord de l'Europe en lien avec les changements du climat depuis 250 ans. La deuxième étude analyse le potentiel d'invasion de la centaurée tachetée (Centaures maculosa), une mauvaise herbe importée en Amérique du nord vers 1890. L'étude apporte la preuve qu'une espèce envahissante peut occuper une niche climatique différente après introduction sur un autre continent. Les modèles basés sur l'aire native prédisent de manière incorrecte l'entier de l'aire envahie mais permettent de prévoir les aires d'introductions potentielles. Une méthode alternative, incluant la calibration du modèle à partir des deux aires où l'espèce est présente, est proposée pour améliorer les prédictions de l'invasion en Amérique du nord. Je présente finalement une revue de la littérature sur la dynamique de la niche écologique dans le temps et l'espace. Elle synthétise les récents développements théoriques concernant le conservatisme de la niche et propose des solutions pour améliorer la pertinence des prédictions d'impact des changements climatiques et des invasions biologiques. SUMMARY Evidences are accumulating that biodiversity is facing the effects of global change. The most influential drivers of change in ecosystems are land-use change, alien species invasions and climate change impacts. Accurate projections of species' responses to these changes are needed to propose mitigation measures to slow down the on-going erosion of biodiversity. Niche-based models (NBM) currently represent one of the only tools for such projections. However, their application in the context of global changes relies on restrictive assumptions, calling for cautious interpretations. In this thesis I aim to assess the effectiveness and shortcomings of niche-based models for the study of global change impacts on biodiversity through the investigation of specific, unsolved limitations and suggestion of new approaches. Two studies investigating threats to rare and endangered plants are presented. I review the ecogeographic characteristic of 118 endangered plants with high conservation priority in Switzerland. The prevalence of rarity types among plant species is analyzed in relation to IUCN extinction risks. The review underlines the importance of regional vs. global conservation and shows that a global assessment of rarity might be misleading for some species because it can fail to account for different degrees of rarity at a variety of spatial scales. The second study tests a modeling framework including iterative steps of modeling and field surveys to improve the sampling of rare species. The approach is illustrated with a rare alpine plant, Eryngium alpinum and shows promise for complementing conservation practices and reducing sampling costs. Two studies illustrate the impacts of climate change on African taxa. The first one assesses the sensitivity of 277 mammals at African scale to climate change by 2050 in terms of species richness and turnover. It shows that a substantial number of species could be critically endangered in the future. National parks situated in xeric ecosystems are not expected to meet their mandate of protecting current species diversity in the future. The second study model the distribution in 2050 of 975 endemic plant species in southern Africa. The study proposes the inclusion of new methodological insights improving the accuracy and ecological realism of predictions of global changes studies. It also investigates the possibility to estimate a priori the sensitivity of a species to climate change from the geographical distribution and ecological proprieties of the species. Three studies illustrate the application of NBM in the study of biological invasions. The first one investigates the Northwards expansion of Lactuca serriola L. in Europe during the last 250 years in relation with climate changes. In the last two decades, the species could not track climate change due to non climatic influences. A second study analyses the potential invasion extent of spotted knapweed, a European weed first introduced into North America in the 1890s. The study provides one of the first empirical evidence that an invasive species can occupy climatically distinct niche spaces following its introduction into a new area. Models fail to predict the current full extent of the invasion, but correctly predict areas of introduction. An alternative approach, involving the calibration of models with pooled data from both ranges, is proposed to improve predictions of the extent of invasion on models based solely on the native range. I finally present a review on the dynamic nature of ecological niches in space and time. It synthesizes the recent theoretical developments to the niche conservatism issues and proposes solutions to improve confidence in NBM predictions of the impacts of climate change and species invasions on species distributions.

The permeability P-glycoprotein: a focus on enantioselectivity and brain distribution.

IMPORTANCE OF THE FIELD: The permeability glycoprotein (P-gp) is an important protein transporter involved in the disposition of many drugs with different chemical structures, but few studies have examined a possible stereoselectivity in its activity. P-gp can have a major impact on the distribution of drugs in selected organs, including the brain. Polymorphisms of the ABCB1 gene, which encodes for P-gp, can influence the kinetics of several drugs. AREAS COVERED IN THIS REVIEW: A search including publications from 1990 up to 2009 was performed on P-gp stereoselectivity and on the impact of ABCB1 polymorphisms on enantiomer brain distribution. WHAT THE READER WILL GAIN: Despite stereoselectivity not being expected because of the large variability of chemical structures of P-gp substrates, structure-activity relationships suggest different P-gp-binding sites for enantiomers. Enantioselectivity in the activity of P-gp has been demonstrated by in vitro studies and in animal models (preferential transport of one enantiomer or different inhibitory potencies towards P-gp activity between enantiomers). There is also in vivo evidence of an enantioselective drug transport at the human blood-brain barrier. TAKE HOME MESSAGE: The significant enantioselective activity of P-gp might be clinically relevant and must be taken into account in future studies.

Imaging the ultrasmall-angle x-ray scattering distribution with grating interferometry.

X-ray imaging with grating interferometry has previously been regarded as a technique providing information only in direct space. It delivers absorption, phase, and dark-field contrast, which can be viewed as parameters of the underlying but unresolved scattering distribution. Here, we present a method that provides the ultrasmall-angle x-ray scattering distribution and, thus, allows simultaneous access to direct and reciprocal space information.

Very high-resolution environmental predictors in species distribution models: moving beyond topography?

Recent advances in remote sensing technologies have facilitated the generation of very high resolution (VHR) environmental data. Exploratory studies suggested that, if used in species distribution models (SDMs), these data should enable modelling species' micro-habitats and allow improving predictions for fine-scale biodiversity management. In the present study, we tested the influence, in SDMs, of predictors derived from a VHR digital elevation model (DEM) by comparing the predictive power of models for 239 plant species and their assemblages fitted at six different resolutions in the Swiss Alps. We also tested whether changes of the model quality for a species is related to its functional and ecological characteristics. Refining the resolution only contributed to slight improvement of the models for more than half of the examined species, with the best results obtained at 5 m, but no significant improvement was observed, on average, across all species. Contrary to our expectations, we could not consistently correlate the changes in model performance with species characteristics such as vegetation height. Temperature, the most important variable in the SDMs across the different resolutions, did not contribute any substantial improvement. Our results suggest that improving resolution of topographic data only is not sufficient to improve SDM predictions - and therefore local management - compared to previously used resolutions (here 25 and 100 m). More effort should be dedicated now to conduct finer-scale in-situ environmental measurements (e.g. for temperature, moisture, snow) to obtain improved environmental measurements for fine-scale species mapping and management.

Distribution of free and liposomal doxorubicin after isolated lung perfusion in a sarcoma model.

BACKGROUND: Isolated lung perfusion (ILP) with free and a novel liposomal-encapsulated doxorubicin (Liporubicin, CT Sciences SA, Lausanne, Switzerland) was compared with respect to drug uptake and distribution in rat lungs bearing a sarcomatous tumor. METHODS: A single sarcomatous tumor was generated in the left lung of 39 Fischer rats, followed 10 days later by left-sided ILP (n = 36) with free and equimolar-dosed liposomal doxorubicin at doses of 100 microg (n = 9) and 400 microg (n = 9) for each doxorubicin formulation. In each perfused lung, the drug concentration and distribution were assessed in the tumor and in three areas of normal lung parenchyma by high-performance liquid chromatography (n = 6) and fluorescence microscopy (n = 3). Histologic assessment and immunostaining with von Willebrand factor was performed in 3 animals with untreated tumors. RESULTS: The sarcomatous tumors in controls were well vascularized with fine branching capillaries present throughout the tumors. Isolated lung perfusion resulted in a heterogeneous drug distribution within the perfused lung and a consistently lower drug uptake in tumors than in lung parenchyma for both doxorubicin formulations and both drug doses applied. Isolated lung perfusion with free doxorubicin resulted in a significantly higher drug uptake than Liporubicin in both the tumor and lung tissue for both drug doses applied (p < 0.01). However, the tumor/normal tissue drug ratio was lower for free than for liposomal doxorubicin at a drug dose of 100 microg (0.27 +/- 0.1 vs 0.53 +/- 0.5; p = 0.225) and similar for both doxorubicin formulations at a drug dose of 400 microg (0.67 +/- 0.2 vs 0.54 +/- 0.2; p = 0.335). Both doxorubicin formulations resulted in fluorescence signaling emerging from all tissue compartments of normal lung parenchyma but only in weak and sporadic signaling from the tumors confined to the tumor periphery and vessels situated within the tumor for both drug doses assessed. CONCLUSIONS: Isolated lung perfusion with free and liposomal doxorubicin resulted in a heterogeneous drug distribution within the perfused lung and in a lower drug uptake in tumors than in lung tissue for both doxorubicin formulations and drug doses applied.

Comparison of the site distribution of melanoma in New Zealand and Canada

A comparison of the site distribution of cutaneous malignant melanoma in New Zealand and Canada was performed. This series deals with 41,331 incident cases registered between 1968 and 1990 and is the largest to date to evaluate the influence of age and gender on the site distribution of melanoma. Site-specific, age-standardized rates per unit surface area and relative tumour density were assessed by gender and country and differences compared with statistical techniques adapted to this context. The age-standardized rates for all sites were higher in New Zealand than in Canada, the ratio being 3.2 for men and 3.8 for women. Occurrence of melanoma was denser for chronically than intermittently exposed sites in both New Zealand and Canada. The highest incidence rate per unit area was for the ears in men which was more than 5 times the rate for the entire body in each country. For each gender, melanomas were relatively commoner on the trunk and the face in Canada, and on the lower limbs in New Zealand. The variations in the site distribution were similar in each country and consistent with the effect of differential patterns of sun exposure between genders. Our results show that the levels of risk of melanoma between phenotypically comparable populations exposed to different amount of UV radiation vary in a site-specific manner, especially for intermittently exposed sites. This suggests that both environmental conditions and lifestyle factors influence the site distribution of melanoma in these two populations.

Interhemispheric distribution of Alzheimer disease and vascular pathology in brain aging

BACKGROUND AND PURPOSE: Most of the neuropathological studies in brain aging were based on the assumption of a symmetrical right-left hemisphere distribution of both Alzheimer disease and vascular pathology. To explore the impact of asymmetrical lesion formation on cognition, we performed a clinicopathological analysis of 153 cases with mixed pathology except macroinfarcts. METHODS: Cognitive status was assessed prospectively using the Clinical Dementia Rating scale; neuropathological evaluation included assessment of Braak neurofibrillary tangle and Ass deposition staging, microvascular pathology, and lacunes. The right-left hemisphere differences in neuropathological scores were evaluated using the Wilcoxon signed rank test. The relationship between the interhemispheric distribution of lesions and Clinical Dementia Rating scores was assessed using ordered logistic regression. RESULTS: Unlike Braak neurofibrillary tangle and Ass deposition staging, vascular scores were significantly higher in the left hemisphere for all Clinical Dementia Rating scores. A negative relationship was found between Braak neurofibrillary tangle, but not Ass staging, and vascular scores in cases with moderate to severe dementia. In both hemispheres, Braak neurofibrillary tangle staging was the main determinant of cognitive decline followed by vascular scores and Ass deposition staging. The concomitant predominance of Alzheimer disease and vascular pathology in the right hemisphere was associated with significantly higher Clinical Dementia Rating scores. CONCLUSIONS: Our data show that the cognitive impact of Alzheimer disease and vascular lesions in mixed cases may be assessed unilaterally without major information loss. However, interhemispheric differences and, in particular, increased vascular and Alzheimer disease burden in the right hemisphere may increase the risk for dementia in this group.

Genome-wide association scan meta-analysis identifies three Loci influencing adiposity and fat distribution.

To identify genetic loci influencing central obesity and fat distribution, we performed a meta-analysis of 16 genome-wide association studies (GWAS, N = 38,580) informative for adult waist circumference (WC) and waist-hip ratio (WHR). We selected 26 SNPs for follow-up, for which the evidence of association with measures of central adiposity (WC and/or WHR) was strong and disproportionate to that for overall adiposity or height. Follow-up studies in a maximum of 70,689 individuals identified two loci strongly associated with measures of central adiposity; these map near TFAP2B (WC, P = 1.9x10(-11)) and MSRA (WC, P = 8.9x10(-9)). A third locus, near LYPLAL1, was associated with WHR in women only (P = 2.6x10(-8)). The variants near TFAP2B appear to influence central adiposity through an effect on overall obesity/fat-mass, whereas LYPLAL1 displays a strong female-only association with fat distribution. By focusing on anthropometric measures of central obesity and fat distribution, we have identified three loci implicated in the regulation of human adiposity.

Distribution of artifactual gas on post-mortem multidetector computed tomography (MDCT)

L'imagerie est de plus en plus utilisée en médecine forensique. Actuellement, les connaissances nécessaires pour interpréter les images post mortem sont faibles et surtout celles concernant les artéfacts post mortem. Le moyen radiologique le plus utilisé en médecine légale est la tomodensitométrie multi-coupes (TDMC). Un de ses avantages est la détection de gaz dans le corps. Cette technique est utile au diagnostic d'embolie gazeuse mais sa très grande sensibilité rend visible du gaz présent même en petite quantité. Les premières expériences montrent que presque tous les corps scannés présentent du gaz surtout dans le système vasculaire. Pour cette raison, le médecin légiste est confronté à un nouveau problème : la distinction entre du gaz d'origine post-mortem et une embolie gazeuse vraie. Pour parvenir à cette distinction, il est essentiel d'étudier la distribution de ces gaz en post mortem. Aucune étude systématique n'a encore été réalisée à ce jour sur ce sujet.¦Nous avons étudié l'incidence et la distribution des gaz présents en post mortem dans les vaisseaux, dans les os, dans les tissus sous-cutanés, dans l'espace sous-dural ainsi que dans les cavités crânienne, thoracique et abdominale (82 sites au total) de manière à identifier les facteurs qui pourraient distinguer le gaz post-mortem artéfactuel d'une embolie gazeuse¦Les données TDMC de 119 cadavres ont été étudiées rétrospectivement. Les critères d'inclusion des sujets sont l'absence de lésion corporelle permettant la contamination avec l'air extérieur, et, la documentation du délai entre le moment du décès et celui du CT-scan (p.ex. rapport de police, protocole de réanimation ou témoin). La présence de gaz a été évaluée semi-quantitativement par deux radiologues et codifiée. La codification est la suivante : grade 0 = pas de gaz, grade 1 = une à quelques bulles d'air, grade 2 = structure partiellement remplie d'air, grade 3 = structure complètement remplie d'air.¦Soixante-quatre des 119 cadavres présentent du gaz (62,2%), et 56 (75,7%) ont montré du gaz dans le coeur. Du gaz a été détecté le plus fréquemment dans le parenchyme hépatique (40%); le coeur droit (ventricule 38%, atrium 35%), la veine cave inférieure (infra-rénale 30%, supra-rénale 26%), les veines sus-hépatiques (gauche 26%, moyenne 29%, droite 22 %), et les espaces du porte (29%). Nous avons constaté qu'une grande quantité de gaz liée à la putréfaction présente dans le coeur droit (grade 3) est associée à des collections de gaz dans le parenchyme hépatique (sensibilité = 100%, spécificité = 89,7%). Pour décrire nos résultats, nous avons construit une séquence d'animation qui illustre le processus de putréfaction et l'apparition des gaz à la TDMC post-mortem.¦Cette étude est la première à montrer que l'apparition post-mortem des gaz suit un modèle de distribution spécifique. L'association entre la présence de gaz intracardiaque et dans le parenchyme hépatique pourrait permettre de distinguer du gaz artéfactuel d'origine post-mortem d'une embolie gazeuse vraie. Cette étude fournit une clé pour le diagnostic de la mort due à une embolie gazeuse cardiaque sur la base d'une TDMC post-mortem.¦Abstract¦Purpose: We investigated the incidence and distribution of post-mortem gas detected with multidetector computed tomography (MDCT) to identify factors that could distinguish artifactual gas from cardiac air embolism.¦Material and Methods: MDCT data of 119 cadavers were retrospectively examined. Gas was semiquantitatively assessed in selected blood vessels, organs and body spaces (82 total sites).¦Results: Seventy-four of the 119 cadavers displayed gas (62.2%; CI 95% 52.8 to 70.9), and 56 (75.7%) displayed gas in the heart. Most gas was detected in the hepatic parenchyma (40%); right heart (38% ventricle, 35% atrium), inferior vena cava (30% infrarenally, 26% suprarenally), hepatic veins (26% left, 29% middle, 22% right), and portal spaces (29%). Male cadavers displayed gas more frequently than female cadavers. Gas was detected 5-84 h after death; therefore, the post-mortem interval could not reliably predict gas distribution (rho=0.719, p<0.0001). We found that a large amount of putrefaction-generated gas in the right heart was associated with aggregated gas bubbles in the hepatic parenchyma (sensitivity = 100%, specificity = 89.7%). In contrast, gas in the left heart (sensitivity = 41.7%, specificity = 100%) or in peri-umbilical subcutaneous tissues (sensitivity = 50%, specificity = 96.3%) could not predict gas due to putrefaction.¦Conclusion: This study is the first to show that the appearance of post-mortem gas follows a specific distribution pattern. An association between intracardiac gas and hepatic parenchymal gas could distinguish between post- mortem-generated gas and vital air embolism. We propose that this finding provides a key for diagnosing death due to cardiac air embolism.

Anatomical distribution of areas of preserved cartilage in advanced femorotibial osteoarthritis using CT arthrography (Part 1).

OBJECTIVE: To determine subregions of normal and abnormal cartilage in advanced stages of femorotibial osteoarthritis (OA) by mapping the entire femorotibial joint in a cohort of pre-total knee replacement (TKR) OA knees. DESIGN: We defined an areal subdivision of the femorotibial articular cartilage surface on CT arthrography (CTA), allowing the division of the femorotibial articular surface into multiple (up to n = 204 per knee) subregions and the comparison of the same areas between different knees. Two readers independently classified each cartilage area as normal, abnormal or non-assessable in 41 consecutive pre-TKR OA knees. RESULTS: A total of 6447 cartilage areas (from 41 knees) were considered assessable by both readers. The average proportion of preserved cartilage was lower in the medial femorotibial joint than in the lateral femorotibial joint for both readers (32.0/69.8% and 33.9/68.5% (medial/lateral) for reader 1 and 2 respectively, all P < 0.001). High frequencies of normal cartilage were observed at the posterior aspect of the medial condyle (up to 89%), and the anterior aspect of the lateral femorotibial compartment (up to 100%). The posterior aspect of the medial condyle was the area that most frequently exhibited preserved cartilage in the medial femorotibial joint, contrasting with the high frequency of cartilage lesions in the rest of that compartment. CONCLUSIONS: Cartilage at the posterior aspect of the medial condyle, and at the anterior aspect of the lateral femorotibial compartment, may be frequently preserved in advanced grades of OA.

Evaluating effects of spectral training data distribution on continuous field mapping performance

Continuous field mapping has to address two conflicting remote sensing requirements when collecting training data. On one hand, continuous field mapping trains fractional land cover and thus favours mixed training pixels. On the other hand, the spectral signature has to be preferably distinct and thus favours pure training pixels. The aim of this study was to evaluate the sensitivity of training data distribution along fractional and spectral gradients on the resulting mapping performance. We derived four continuous fields (tree, shrubherb, bare, water) from aerial photographs as response variables and processed corresponding spectral signatures from multitemporal Landsat 5 TM data as explanatory variables. Subsequent controlled experiments along fractional cover gradients were then based on generalised linear models. Resulting fractional and spectral distribution differed between single continuous fields, but could be satisfactorily trained and mapped. Pixels with fractional or without respective cover were much more critical than pure full cover pixels. Error distribution of continuous field models was non-uniform with respect to horizontal and vertical spatial distribution of target fields. We conclude that a sampling for continuous field training data should be based on extent and densities in the fractional and spectral, rather than the real spatial space. Consequently, adequate training plots are most probably not systematically distributed in the real spatial space, but cover the gradient and covariate structure of the fractional and spectral space well. (C) 2009 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved.