Wetland Land Cover Research Articles

Forest Habitat Change Dynamics in a Riparian Wetland

Our study aims at monitoring ecosystem dynamics in the Szigetköz Danube floodplain (Hungary), where the wetland communities have been seriously affected by the abrupt hydrological changes after the Danube-diversion in 1992. Wetland habitats were mapped in the last years for a chosen area, based on a comparable use of archived analogue and digital aerial images and available silvicultural and botanical inventories. During the object-based image analysis the classification was mainly based on texture measures. The resulting map time series show habitat borders for each year and present the changes separately for each class. The method is suitable for the characterization of spatial and temporal structure of wetland land cover in a realistic way.

Influence of wetland networks on bat activity in mixed-use landscapes

Parks and other protected lands can provide important source habitat and act as valuable dispersal corridors in urbanizing environments. However, most wetlands within protected areas are managed in isolation without consideration of the broader landscape connections. We studied the importance of wetland habitat connectivity and landscape context to bat activity in five National Parks along a gradient of increasing urbanization within the Mid-Atlantic United States. Ninety-six Anabat stations were set up throughout the parks, from which we derived the characteristic spatial scales at which bat activity was associated with wetlands. This information was used in a graph theoretic framework to construct network models of potential landscape connectivity for those species that had positive associations with wetland land cover. We found that the importance of wetlands as a predictor of bat activity varied on a species-by-species basis and increased when network measures were used that accounted for connected area in a broad spatial context. The results demonstrate that both area and connectivity of wetland foraging habitat may act as orthogonal variables to availability of roosting habitat in explaining the distribution of highly mobile species. We use the results to illustrate the value of network analysis to guide the coordinated management of two of the parks’ most valued natural resources – wetlands and bats.

Physical Habitat and Fish Assemblage Relationships with Landscape Variables at Multiple Spatial Scales in Wadeable Iowa Streams

AbstractLandscapes in Iowa and other midwestern states have been profoundly altered by conversion of native prairies to agriculture. We analyzed landscape data collected at multiple spatial scales to explore relationships with reach‐scale physical habitat and fish assemblage data from 93 randomly selected sites on second‐ through fifth‐order wadeable Iowa streams. Ordination of sites by physical habitat showed significant gradients of channel shape, habitat complexity, substrate composition, and stream size. Several landscape variables were significantly associated with the physical habitat ordination. Row crop land use was associated with fine substrates and steep bank angles, whereas wetland land cover and greater sinuosity and catchment land area were associated with complex channel and bank morphology and greater residual pool volume, woody debris, and canopy cover. Thirteen landscape variables were significant predictors of physical habitat variables in multiple linear regressions, with adjusted R2 values ranging from 0.07 to 0.74. Inclusion of landscape variables with physical habitat variables in multiple regression models predicting fish assemblage metrics and a fish index of biotic integrity resulted in negligible improvements over models based on only physical habitat variables. Physical habitat in wadeable Iowa streams is strongly associated with landscape characteristics. Results of this study and previous studies suggest that (1) landscape factors directly influence physical habitat, (2) physical habitat directly influences fish assemblages, and (3) the influence of landscape factors on fish assemblages is primarily indirect. Understanding how landscape factors, such as human land use, influence physical habitat and fish assemblages will help managers make more informed decisions for improving Iowa's wadeable streams.

Wetland mapping in the Congo Basin using optical and radar remotely sensed data and derived topographical indices

This study reports results of a classification tree approach to mapping the wetlands of the Congo Basin, focusing on the Cuvette Centrale of the Congo River watershed, an area of 1,176,000 km 2. Regional expert knowledge was used to train passive optical remotely sensed imagery of the Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) sensors, JERS-1 active radar L-band imagery, and topographical indices derived from 3 arc sec elevation data of the Shuttle Radar Topography Mission (SRTM). All data inputs were resampled to a common 57 m resolution grid. A classification tree bagging procedure was employed to produce a final map of per-grid cell wetland probability. Thirty bagged trees were ranked and the median result was selected to produce the final wetland probability map. Thresholding the probability map at < 0.5 yielded a proportion of wetland cover for the study area of 32%, equivalent to 360,000 km 2. Wetlands predominate in the CARPE Lake Tele–Lake Tumba landscape located in the western part of the Democratic Republic of the Congo and the south-eastern Republic of Congo, where they constitute 56% of the landscape. Local topography depicting relative elevation for sub-catchments proved to be the most valuable discriminator of wetland cover. However, all sources of information (i.e. optical, radar and topography) featured prominently in contributing to the classification tree procedure, reinforcing the idea that multi-source data are useful in the characterization of wetland land cover. The method employed freely available data and a fully automated process, except for training data collection. Comparisons to existing maps and in situ field observations indicate improvements compared to previous efforts.

Nitrogen stable isotopes in streams: effects of agricultural sources and transformations

The nitrogen stable isotope ratio of biological tissue has been proposed as an indicator of anthropogenic N inputs to aquatic ecosystems, but overlap in the isotopic signatures of various N sources and transformations make definitive attribution of processes difficult. We collected primary consumer invertebrates from streams in agricultural settings in Wisconsin, U.S.A., to evaluate the relative influence of animal manure, inorganic fertilizer, and denitrification on biotic delta15N. Variance in biotic delta15N was explained by inorganic fertilizer inputs and the percentage of wetland land cover in the watershed, but not by animal manure inputs. These results suggest that denitrification of inorganic fertilizer is the primary driver of delta15N variability among the study sites. Comparison with previously collected stream water NO3-N concentrations at the same sites supports the role of denitrification; for a given N application rate, streams with high biotic delta15N had low NO3-N concentrations. The lack of a manure signal in biotic delta15N may be due its high ammonia content, which can be dispersed outside the range of its application by volatilization. Based on our findings and on agricultural census data for the entire United States, inorganic fertilizer is more likely than manure to drive variability in biotic delta15N and to cause excessive nitrogen concentrations in streams.

An empirical evaluation of the nutrient‐color paradigm for lakes

We analyzed a diverse set of 1,646 north‐temperate lakes to evaluate the nutrient‐color paradigm that integrates total phosphorus (TP) and colored dissolved organic carbon to define lake trophic status. Our objectives were to quantify the combined influence of TP and color (Col) on lake trophic status, to determine if TP and Col had similar relationships with hydrogeomorphic (HGM) variables, and to examine how TP and Col affected the balance of heterotrophic and autotrophic processes. For the latter we examined the Col to chlorophyll a ratio (Col : Chl a), an index of allochthonous contributions of carbon to pelagic consumers, and deviations of lake pCO2 from atmospheric, an index of net heterotrophy. Both Col and TP had strong effects on Chl a (positive) and Secchi transparency (negative), suggesting that ignoring Col would lead to misinterpretation of these widely used trophic status indicators. Lakes with high TP and Col tended to be shallower with large catchment to lake area ratios. Negative correlations with water retention time (WRT) were stronger for Col than for TP. Both TP and Col were related to forest and wetland land cover, although the direction of the relationships were opposite. Only 29% of the lakes had relatively high allochthony according to their Col : Chl a ratios; these were predominately high color, oligotrophic or mesotrophic lakes with short WRT. Over 90% of a subset of 682 lakes were net heterotrophic, with pCO2 exceeding atmospheric levels. The positive relationship between pCO2 and Col : Chl a suggests that only in very heterotrophic systems was the transfer of allochthonous carbon to pelagic consumers appreciable. Our results provide strong empirical support for the nutrient‐color paradigm and highlight its importance both for management applications and for expanding our understanding of how lakes are influenced by terrestrial subsidies of carbon and nutrients.

Optimum RADARSAT-1 configurations for wetlands discrimination: a case study of the Mer Bleue peat bog

This study investigated the effect of the RADARSAT-1 incidence angle on the discrimination of the wetlands, including open bog, treed bog, and marsh, in the Mer Bleue bog conservation area in Ontario, Canada. The results demonstrated that RADARSAT-1 can penetrate tall herbaceous vegetation in marsh at all incidence angles and short shrub vegetation in open bog with standing water at low incidence angles only, which reveals that RADARSAT-1 backscatter double bounce from open wetlands (marsh and bog) with standing water is related to the vegetation types. The analysis of the RADARSAT-1 synthetic aperture radar (SAR) backscatter of wetland and non-wetland land cover types confirmed that the backscatter return of each land cover type decreases as the incidence angle increases. It was also found that the separations of marsh, treed bog, and open bog without standing water do not vary significantly with increasing incidence angles. However, the incidence angle does affect the discrimination of open bog with standing water. The results from separability measures of land cover classes demonstrate that the combination of a low incidence angle (<31°) RADARSAT-1 standard beam mode image with Landsat-7 data can provide good separation for all class pairs in the study area. The combination of RADARSAT-1 standard beam mode 1 (S1) and Landsat-7 images gave better classification results than the combination of RADARSAT-1 standard beam mode 5 (S5) and Landsat-7 images: the classification accuracies (kappa indices) of various wetlands increased in the range 0.02-0.13. For optimal spatial detail, fine beam mode data are desirable because of the higher spatial resolution (~9 m). However, in terms of the discrimination and classification accuracy of wetlands, fine beam mode data are not suggested because of the high incidence angle. Therefore, it is concluded that a lower incidence angle RADARSAT-1 image (e.g., S1) is the optimal beam mode when combining with Landsat images for the discrimination and classification of wetlands.

Development of Stream Macroinvertebrate Models That Predict Watershed and Local Stressors in Wisconsin

Clear, quantitative biological standards are needed for stream protection. Three hundred ninety-seven sites in Wisconsin were used to develop and test stream macroinvertebrate indices of biotic integrity (IBI) designed to provide such standards for the upper Midwestern United States. The IBI development approach was unique in 1) selecting base assemblage metrics a priori, 2) using watershed land cover and local habitat to quantify human influence, 3) weighting local- and water- shed-scale variables to characterize human influence in a biologically meaningful way, and 4) weight- ing macroinvertebrate metrics to characterize the assemblage in a way that detects human influence. Canonical correlation analysis provided IBI models by identifying the strongest relation between a set of macroinvertebrate metrics and a set of environmental condition measures. Metrics related to assemblage composition, structure, and function represented the macroinvertebrates, including or- ganic pollution tolerance, sediment tolerance, species richness, proportion or number of Ephemer- optera-Plecoptera-Trichoptera, Amphipoda, Isopoda, Diptera, and Chironomidae taxa, and propor- tion that feed by shredding, scraping, and gathering. Environmental condition measures included proportions of urban, forage crop, row crop, forest, open-water, and wetland land cover at the wa- tershed scale, and riparian condition, bed and bank condition, and habitat heterogeneity at the local scale. The Driftless Area and Northern Forest ecoregions were unique and warranted separate models, whereas the Central and Southeast ecoregions were merged into one model. Partial regressions iden- tified the relative influence of variables at the watershed and local scales. Model testing with inde- pendent data indicated the appropriateness of classification into the 3 regions, and that the IBIs could predict human influence well. The IBI models show promise for land-cover planning, pollution mon- itoring, and developing biological criteria for stream and watershed protection.

Comparisons of P-Yield, Riparian Buffer Strips, and Land Cover inSix Agricultural Watersheds

Riparian buffer strips may protect streams from phosphorus (P) pollution. We compared 2 years of daily P-yield (μg m−2 day−1) from six southeast Wisconsin watersheds with contrasting riparian buffer attributes. Of the variables measured, mean daily P-yield was most closely correlated with the variability in riparian patch size. Variability in P-yield was most closely correlated with characteristics of the riparian buffer, such as percent wetland land cover, riparian continuity, and stream sinuosity. During the most extreme events, mean P-yield was negatively correlated with the percentage of wetland land cover in the upland watershed. Correlations suggest that riparian continuity may influence P-loading in these watersheds. Our results corroborate the importance of continuity and uniformity of riparian buffers as moderators of P flow from upland agricultural lands into streams.

Mercury and methylmercury in water and sediment of the Sacramento River Basin, California

Mercury (Hg) and methylmercury (CH 3Hg +) concentrations in streambed sediment and water were determined at 27 locations throughout the Sacramento River Basin, CA. Mercury in sediment was elevated at locations downstream of either Hg mining or Au mining activities where Hg was used in the recovery of Au. Methylmercury in sediment was highest (2.84 ng/g) at a location with the greatest wetland land cover, in spite of lower total Hg at that site relative to other river sites. Mercury in unfiltered water was measured at 4 locations on the Sacramento River and at tributaries draining the mining regions, as well as agricultural regions. The highest levels of Hg in unfiltered water (2248 ng/l) were measured at a site downstream of a historic Hg mining area, and the highest levels at all sites were measured in samples collected during high streamflow when the levels of suspended sediment were also elevated. Mercury in unfiltered water exceeded the current federal and state recommended criterion for protection of aquatic life (50 ng/l as total Hg in unfiltered water) only during high streamflow conditions. The highest loading of Hg to the San Francisco Bay system was attributed to sources within the Cache Creek watershed, which are downstream of historic Hg mines, and to an unknown source or sources to the mainstem of the Sacramento River upstream of historic Au mining regions. That unknown source is possibly associated with a volcanic deposit. Methylmercury concentrations also were dependent on season and hydrologic conditions. The highest levels (1.98 ng/l) in the Sacramento River, during the period of study, were measured during a major flood event. The reactivity of Hg in unfiltered water was assessed by measuring the amount available for reaction by a strong reducing agent. Although most Hg was found to be nonreactive, the highest reactivity (7.8% of the total Hg in water) was measured in the sample collected from the same site with high CH 3Hg + in sediment, and during the time of year when that site was under continual flooded conditions. Although Hg concentrations in water downstream of the Hg mining operations were measured as high as 2248 ng/l during stormwater runoff events, the transported Hg was found to have a low potential for geochemical transformations, as indicated by the low reactivity to the reducing agent (0.0001% of the total), probably because most of the Hg in the unfiltered water sample was in the mercury sulfide form.

Mapping coastal ecosystems over a steep development gradient using C-CAP protocols

A nine category land cover map of over 5500 sq.km of Long Island, New York was produced from TM imagery using protocols established for the Coastwatch Change Analysis Program (C-CAP). The adjustments made to the protocol to conform to local environmental conditions and the methodology employed to map five upland and four wetland land cover types are described. Initial classification accuracy using only TM imagery fell below acceptable levels for some categories, with one wetland land cover being particularly troublesome. Spectral confusion among land cover was reduced by the incorporation of ancillary data sets, and an accuracy of over 90% was achieved. Three accuracy assessments through confusion matrices and kappa statistics provided data on individual points of spectral mix and the effect of area-weighting the accuracy evaluation.