TSTon H. Snelder

Ton H. Snelder

*This is an algorithmically generated author record based on publicly available information.

Ton H. Snelder has a total of 26 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • NATURAL FLOW REGIME CLASSIFICATIONS ARE SENSITIVE TO DEFINITION PROCEDURES

    by T. H. Snelder, D. J. Booker

    The correspondence and performance of six classifications of flow regimes of New Zealand rivers that were all mapped onto the same digital river network were assessed. Classification 1 was defined deductively, based on expert‐defined rules. Classifications 2 to 6 were defined inductively using hydrological indices calculated from 321 natural daily flow records. Classifications 2 to 4 were defined by first clustering the gauges based on the hydrological indices and then predicting the class of each segment of the network using a Random Forest classifier. Classifications 5 and 6 were defined by first predicting the indices for each segment of the network using Random Forest regression models. Cluster analysis was then used to group the network segments into classes. Further differences between classifications were due to differences in the standardisation of the hydrological indices and clustering algorithms. Correspondence (extent to which the patterns defined by the classifications were similar) was assessed formally using the adjusted Rand index and visually. The performance of the classifications was assessed using classification strength calculated using the hydrological indices and ANOVA calculated for individual indices. Correspondence between the classifications was low (adjusted Rand index range, 0.1–0.5). Classification strength and ANOVA statistics assessed using cross validation indicated that the inductive classifications performed better than the deductively defined classification and that there were some significant differences in performance between the inductive classifications. However, these differences were not large from a practical point of view. Our results indicate that there are many credible classifications of the flow regimes of a study region. When considering methods for defining flow regime classifications, aspects other than the predictive performance, such as flow data requirements and how easily the final classification can be explained, should be considered. Copyright © 2012 John Wiley & Sons, Ltd.


  • Can bottom-up procedures improve the performance of stream classifications?

    by Ton Snelder, José Barquín Ortiz, Doug Booker, Nicolas Lamouroux, Hervé Pella, Ude Shankar


  • Effect of Classification Procedure on the Performance of Numerically Defined Ecological Regions

    by Ton Snelder, Anthony Lehmann, Nicolas Lamouroux, John Leathwick, Karin Allenbach

    Ecological regionalizations define geographic regions exhibiting relative homogeneity in ecological (i.e., environmental and biotic) characteristics. Multivariate clustering methods have been used to define ecological regions based on subjectively chosen environmental variables. We developed and tested three procedures for defining ecological regions based on spatial modeling of a multivariate target pattern that is represented by compositional dissimilarities between locations (e.g., taxonomic dissimilarities). The procedures use a “training dataset” representing the target pattern and models this as a function of environmental variables. The model is then extrapolated to the entire domain of interest. Environmental data for our analysis were drawn from a 400 m grid covering all of Switzerland and consisted of 12 variables describing climate, topography and lithology. Our target patterns comprised land cover composition of each grid cell that was derived from interpretation of aerial photographs. For Regionalization 1 we used conventional cluster analysis of the environmental variables to define 60 hierarchically organized levels comprising from 5 to 300 regions. Regionalization 1 provided a base-case for comparison with the model-based regionalizations. Regionalization 2, 3 and 4 also comprised 60 hierarchically organized levels and were derived by modeling land cover composition for 4000 randomly selected “training” cells. Regionalization 2 was based on cluster analysis of environmental variables that were transformed based on a Generalized Dissimilarity Model (GDM). Regionalization 3 and 4 were defined by clustering the training cells based on their land cover composition followed by predictive modeling of the distribution of the land cover clusters using Classification and Regression Tree (CART) and Random Forest (RF) models. Independent test data (i.e. not used to train the models) were used to test the discrimination of land cover composition at all hierarchical levels of the regionalizations using the classification strength (CS) statistic. CS for all the model-based regionalizations was significantly higher than for Regionalization 1. Regionalization 3 and 4 performed significantly better than Regionalization 2 at finer hierarchical levels (many regions) and Regionalization 4 performed significantly better than Regionalization 3 for coarse levels of detail (few regions). Compositional modeling can significantly increase the performance of numerically defined ecological regionalizations. CART and RF-based models appear to produce stronger regionalizations because discriminating variables are able to change at each hierarchic level.


  • Co‐variation of fish assemblages, flow regimes and other habitat factors in French rivers

    by Ton H. Snelder, Nicolas Lamouroux

    Summary: 1. Using data from 270 minimally disturbed river sites distributed across France, we examined relationships between fish assemblages and four sets of explanatory variables representing climate, position in the river network, geomorphology and flow regime (temporal flow variation).2. Flow regimes, represented by 157 hydrological indices, were derived from 763 gauging stations representing unmodified flows. Principal components analysis was used to reduce the 157 indices to nine non‐redundant hydrological variables, and hydrological regionalisation was then used to transfer the hydrological variables to the sampling sites.3. Climate, position in the river network and geomorphology, collectively referred to as ‘habitat’ factors, and were represented by the variables mean January and July temperature, distance from source, mean flow and altitude and width, depth, slope, sediment size and % lotic habitat.4. Partial constrained ordination (pCCA) showed that hydrological variables individually explained 9.7% of the variability of fish assemblages, however the variation explained by the hydrological variables decreased to only 3.4% when the variation explained by the habitat variables was removed.5. Our study showed that co‐variation between hydrological and habitat variables need to be accounted for to avoid overestimating the importance of biological–hydrological relationships. Failing to do so may have important implications for the development of environmental flow prescriptions, as apparent biological responses to altered flow regimes are likely to be confounded by variation associated with habitat variability.


  • Strong Influence of Variable Treatment on the Performance of Numerically Defined Ecological Regions

    by Ton Snelder, Anthony Lehmann, Nicolas Lamouroux, John Leathwick, Karin Allenbach

    Numerical clustering has frequently been used to define hierarchically organized ecological regionalizations, but there has been little robust evaluation of their performance (i.e., the degree to which regions discriminate areas with similar ecological character). In this study we investigated the effect of the weighting and treatment of input variables on the performance of regionalizations defined by agglomerative clustering across a range of hierarchical levels. For this purpose, we developed three ecological regionalizations of Switzerland of increasing complexity using agglomerative clustering. Environmental data for our analysis were drawn from a 400 m grid and consisted of estimates of 11 environmental variables for each grid cell describing climate, topography and lithology. Regionalization 1 was defined from the environmental variables which were given equal weights. We used the same variables in Regionalization 2 but weighted and transformed them on the basis of a dissimilarity model that was fitted to land cover composition data derived for a random sample of cells from interpretation of aerial photographs. Regionalization 3 was a further two-stage development of Regionalization 2 where specific classifications, also weighted and transformed using dissimilarity models, were applied to 25 small scale “sub-domains” defined by Regionalization 2. Performance was assessed in terms of the discrimination of land cover composition for an independent set of sites using classification strength (CS), which measured the similarity of land cover composition within classes and the dissimilarity between classes. Regionalization 2 performed significantly better than Regionalization 1, but the largest gains in performance, compared to Regionalization 1, occurred at coarse hierarchical levels (i.e., CS did not increase significantly beyond the 25-region level). Regionalization 3 performed better than Regionalization 2 beyond the 25-region level and CS values continued to increase to the 95-region level. The results show that the performance of regionalizations defined by agglomerative clustering are sensitive to variable weighting and transformation. We conclude that large gains in performance can be achieved by training classifications using dissimilarity models. However, these gains are restricted to a narrow range of hierarchical levels because agglomerative clustering is unable to represent the variation in importance of variables at different spatial scales. We suggest that further advances in the numerical definition of hierarchically organized ecological regionalizations will be possible with techniques developed in the field of statistical modeling of the distribution of community composition.


  • Development of an Ecologic Marine Classification in the New Zealand Region

    by Ton H. Snelder, John R. Leathwick, Katie L. Dey, Ashley A. Rowden, Mark A. Weatherhead, Graham D. Fenwick, Malcolm P. Francis, Richard M. Gorman, Janet M. Grieve, Mark G. Hadfield, et al.

    We describe here the development of an ecosystem classification designed to underpin the conservation management of marine environments in the New Zealand region. The classification was defined using multivariate classification using explicit environmental layers chosen for their role in driving spatial variation in biologic patterns: depth, mean annual solar radiation, winter sea surface temperature, annual amplitude of sea surface temperature, spatial gradient of sea surface temperature, summer sea surface temperature anomaly, mean wave-induced orbital velocity at the seabed, tidal current velocity, and seabed slope. All variables were derived as gridded data layers at a resolution of 1 km. Variables were selected by assessing their degree of correlation with biologic distributions using separate data sets for demersal fish, benthic invertebrates, and chlorophyll-a. We developed a tuning procedure based on the Mantel test to refine the classification’s discrimination of variation in biologic character. This was achieved by increasing the weighting of variables that play a dominant role and/or by transforming variables where this increased their correlation with biologic differences. We assessed the classification’s ability to discriminate biologic variation using analysis of similarity. This indicated that the discrimination of biologic differences generally increased with increasing classification detail and varied for different taxonomic groups. Advantages of using a numeric approach compared with geographic-based (regionalisation) approaches include better representation of spatial patterns of variation and the ability to apply the classification at widely varying levels of detail. We expect this classification to provide a useful framework for a range of management applications, including providing frameworks for environmental monitoring and reporting and identifying representative areas for conservation.


  • Improved eco-hydrological classification of rivers

    by Ton H. Snelder, Barry J. F. Biggs, Ross A. Woods


  • Is the River Environment Classification an improved landscape-scale classification of rivers?

    by Ton H. Snelder, Franck Cattanéo, Alastair M. Suren, Barry J. F. Biggs

    Landscape classifications, such as ecoregions, are widely promoted as spatial frameworks for aquatic conservation and environmental management. The River Environment Classification (REC) is an alternative approach to landscape-scale river classification. The REC classifies individual sections of river networks according to the climate, topography, geology, and land cover of the unique upstream watershed of each section. The classification is hierarchical and is based on the assumption that climate, topography, geology, and land cover are dominant causes of spatial variation in physical and biological characteristics of rivers at 4 spatial scales. The ability of the REC to explain variation in invertebrate samples collected from 237 sites in New Zealand was compared to that of region-based classifications using classification strength tests. Classifications with differing numbers of classes were compared using relative classification strengths calculated by expressing absolute classification streng...


  • NUTRIENT CONCENTRATION CRITERIA AND CHARACTERIZATION OF PATTERNS IN TROPHIC STATE FOR RIVERS IN HETEROGENEOUS LANDSCAPES1

    by Ton H. Snelder, Barry J.F. Biggs, Mark A. Weatherhead

    A method is demonstrated for the development of nutrient concentration criteria and large scale assessment of trophic state in environmentally heterogeneous landscapes. The method uses the River Environment Classification (REC) as a spatial framework to partition rivers according to differences in processes that control the accrual and loss of algae biomass. The method is then applied to gravel bed rivers with natural flow regimes that drain hilly watersheds in New Zealand's South Island. An existing model is used to characterize trophic state (in terms of chlorophyll a as a measure of maximum biomass) using nutrient concentration, which controls the rate of biomass accrual, and flood frequency, which controls biomass loss. Variation in flood frequency was partitioned into three classes, and flow data measured at 68 sites was used to show that the classes differ with respect to flood frequency. Variation in nutrient concentration was partitioned at smaller spatial scales by subdivision of higher level classes into seven classes. The median of flood frequency in each of the three higher level classes was used as a control variable in the model to provide spatially explicit nutrient concentration criteria by setting maximum chlorophyll a to reflect a desired trophic state. The median of mean monthly soluble reactive phosphorus and soluble inorganic nitrogen measured at 68 water quality monitoring sites were then used to characterize the trophic state of each of the seven lower level classes. The method models biomass and therefore allows variation in this response variable to provide options for trophic state and the associated nutrient concentrations to achieve these. Thus it is less deterministic than using reference site water quality. The choice from among these options is a sociopolitical decision, which reflects the management objectives rather than purely technical considerations.