基于遥感的湿地景观格局季相分析

以中国东北地区三江平原北部为研究区域,利用2012年多季相遥感影像作为数据源,结合野外调查数据,应用面向对象的分类方法,根据影像的物候、时相等特征,提取不同月份的湿地信息,进行景观格局季相分析。结果表明:(1)研究区湿地面积、类型格局在同一年不同季节不同月份会有不同幅度的变化,总体呈现缓增骤减的态势。湿地主要分布在低洼地区,主要湿地类型为草本沼泽,其次为河流,其他湿地占总面积比例较小。(2)研究区各阶段湿地都有转化,主要发生在湿地和非湿地之间,多数表现在草本沼泽和草地之间的转化。(3)湿地分布和湿地转化面积主要集中在低海拔区域和低坡度区域,其中海拔100 m和坡度5°以下范围内的湿地分布面积和湿地转化面积占湿地总面积及湿地转化面积的绝大部分。(4)年内季节性湿地转化与降水、温度和湿地植被物候关系密切。     

Characteristic community structure of Florida’s subtropical wetlands: the Florida wetland condition index for depressional marshes, depressional forested, and flowing water forested wetlands

Land use and land cover change has a marked affect on wetland condition, and different wetland types are affected differentially depending on many abiotic and biotic variables. To assess wetland condition, we have developed a Florida wetland condition index (FWCI) composed of indicators of community structure in the diatom, macrophyte, and macroinvertebrate assemblages for 216 wetlands (n = 74 depressional marsh, n = 118 depressional forested, n = 24 flowing water forested wetlands). Depressional wetlands located along a human disturbance gradient throughout Florida were sampled for each assemblage. Forested flowing water wetlands were sampled for macrophytes only. The landscape development intensity index (LDI) was used to quantify the human disturbance gradient. In general, human disturbance in adjacent areas had the greatest impact on depressional herbaceous wetlands, followed by depressional forested wetlands. Forested flowing water wetlands (i.e., forested strands and floodplain wetlands) were less affected by local conditions, with most of their changes in wetland condition correlated with alterations at the larger watershed scale. Strong correlations between the FWCIs and LDI index scores suggest that changes in community structure can be detected along a gradient of human land use activities adjacent to wetland ecosystems.     

Landscape diversity and forest edge density regulate stream water quality in agricultural catchments

It is well known that the composition of land cover within a watershed plays a large role in regulating stream water quality. However, there remains significant uncertainty regarding the effect of spatial configuration of different types of land cover on water quality. Using periphytic algae (diatoms) as indicators of stream trophic state, we investigated the relationship between landscape configuration and water quality in a large number of watersheds (590) at varying catchment scales in Eastern Canada. Variation partitioning analysis showed that landscape configuration explained 48% of the variation in water quality. However, since the physiographic setting constrains most agricultural activities, most of the variation was attributed to the shared influence of surficial deposits, land cover and landscape configuration (34%). The results from regression models showed that the geomorphological setting of watersheds (surficial deposits and slopes) and the proportion of different land cover types (mainly forests, wetlands, crops and urban areas) have a major impact on stream water quality. Nevertheless, a few configuration metrics emerged as important factors. Landscape diversity appeared to have a negative impact on water quality, whereas forest and wetland edge densities had a positive impact. Moreover, the influence of these lanscape metrics seems to occur at certain thresholds. In areas of intensive farming, streams with a forest area that covers at least 47% of the watershed have a better water quality. Below this threshold, eutrophic and meso-eutrophic conditions are more frequent in streams and rivers. The shape and location of forested patches were also found to be relevant. Woodlands and wetlands with an edge density higher than 36 m/ha and located along streams and gullies have a positive impact on water quality. For the same proportion of forest, complex patches will be more efficient filters than large regular patches. Forest edge density seems to control the extent of the interface with the agricultural sources and thus promotes the “sink” effect of forests on nutrients.     

The ecological condition of geographically isolated wetlands in the southeastern United States: The relationship between landscape level assessments and macrophyte assemblages

Geographically isolated wetlands (GIWs) are common features of the Dougherty Plain physiographic region in southwestern Georgia. Due to lack of protection at the state and federal levels, these wetlands are threatened by intensive agricultural and silvicultural land uses common in the region. Recently, the ecological condition of such GIWs was assessed for the southeastern United States using the Landscape Development Intensity Index (LDI), a practical assessment tool that relies on remotely sensed land use and land cover (LULC) data surrounding isolated wetlands to rapidly predict wetland condition. However, no assessments have been attempted for GIWs in the Dougherty Plain specifically. Our goal was to develop a framework to guide and refine remote assessment of wetland condition within this agriculturally intense region of the southeastern USA. In this study, we characterized human disturbances associated with isolated wetlands in the Dougherty Plain, and paired the rapid assessment of GIWs using LDI with an intensive assessment of wetland plant communities. Specifically, we: (1) examined how macrophyte assemblages and vegetation metrics vary across a human disturbance gradient in the Dougherty Plain; (2) compared multiple condition assessment outcomes using variations of the LDI method that differed in spatial extent and resolution of LULC categories; and (3) determined the predicted condition of GIWs in the Dougherty Plain as indexed by LDI and compared with region-wide assessments of GIWs of the southeastern USA. Generally, the relationship between wetland plant communities and surrounding land use supported the assumptions of the LDI index in that wetlands surrounded by agricultural land use classes featured distinct plant communities relative to those surrounded by forested land use classes. Our results indicated that finer spatial resolution of LULC data improved the predictive ability of LDI. However, based on incongruence between wetland vegetation composition and LDI scores in some forested landscapes, this study identified limitations of the LDI assessment method, particularly when applied in regions in which prescribed fire is an important ecological driver of vegetation and habitat. Thus, we conclude that LDI may be biased toward an overestimation of reference condition GIWs, even though the habitat may be functionally degraded by the absence of natural processes such as fire. Regardless, relative to the assessment of the entire southeastern US, a greater proportion of total GIWs of the Dougherty Plain were identified as impaired due to the intensity of irrigated agricultural land use.     

Immature mosquitoes in agricultural wetlands of the coastal plain of Georgia,U.S.A.: Effects of landscape and environmental habitat characteristics

This study is the first to report on the relationships between immature mosquitoes (larvae and pupae) and landscape and environmental habitat characteristics in wetlands associated with row crop agriculture. Indicator species analysis (ISA) was used to test for associations among mosquito species and groups of wetland sites with similar Landscape Development Intensity (LDI) values. Results indicated that Anopheles quadrimaculatus, Culex erraticus, and Psorophora columbiae were associated with agricultural wetlands (LDI > 2.0), whereas Anopheles crucians and Culex territans were associated with forested reference wetlands (LDI < 2.0) in both wet and dry years. The species fidelity to wetland type, regardless of the hydrologic regime, demonstrates these species are robust indicators of wetland condition. Data on immature mosquito assemblages were compared to selected landscape and environmental habitat variables using Akaike's Information Criterion (AIC_c) model selection. LDI indices, dissolved oxygen concentration, the proportion of emergent vegetation, and the proportion of bare ground in wetlands were important factors associated with the selected mosquito species. These results indicate that LDI indices are useful in predicting the distributions of disease vectors or other nuisance mosquito species across broad geographic areas. Additionally, these results suggest mosquitoes are valuable bioindicators of wetland condition that reflect land use and hydrologic variability.     

广西滨海区域景观格局分析及土地利用变化预测

研究滨海湿地景观动态及其发展趋势对于维持沿海地区可持续发展具有重要意义。利用移动窗口法分析广西滨海湿地景观格局时空变化,同时提取珍珠湾、茅尾海和丹兜海典型区进一步分析景观变化趋势,最后采用土地利用变化模型(Land Change Modeler,LCM)预测区域未来的景观演化趋势。结果表明,2000—2014年间,草地、乔木园地和常绿阔叶灌木林面积减少,而工业用地面积明显增加。研究区总体的景观异质性变化不大,破碎化趋势不明显。比较景观格局指数发现,边缘密度和多样性指数的空间分布特征相似,多样性指数较大的区域,边缘密度也大,并且多集中于珍珠湾、茅尾海和研究区的中东部。3个典型区的格局变化趋势相似,景观多样性指数降低而均匀度增加,其中,湿地斑块数目减少且所占面积比也减少。LCM模型结果表明,至2020年,草本湿地与森林灌丛湿地面积进一步降低,而工业用地和居民区则持续增长;其中,3个典型区土地利用变化均受到了工业发展的影响,以珍珠湾土地利用动态度最高,茅尾海其次,而丹兜海最小。当前,区域景观的变化明显降低了湿地植被的质量,加强湿地植被保护,限制工业区域的无序发展是维持区域生境完整和安全的有效途径。     

Rate of succession in restored wetlands and the role of site context

Question: Are changes in plant species composition, functional group composition and rates of species turnover consistent among early successional wetlands, and what is the role of landscape context in determining the rate of succession? Location: Twenty‐four restored wetlands in Illinois, USA. Methods: We use 4 years of vegetation sampling data from each site to describe successional trends and rates of species turnover in wetlands. We quantify: (1) the rate at which composition changes from early‐successional to late‐successional species and functional groups, as indicated by site movement in ordination space over time, and (2) the rate of change in the colonization and local extinction of individual species. We correlate the pace of succession to site area, isolation and surrounding land cover. Results: Some commonalities in successional trends were evident among sites. Annual species were replaced by clonal perennials, and colonization rates declined over time. However, differences among sites outweighed site age in determining species composition, and the pace of succession was influenced by a site's landscape setting. Rates of species turnover were higher in smaller wetlands. In addition, wetlands in agricultural landscapes underwent succession more rapidly, as indicated by a rapid increase in dominance by late‐successional plants. Conclusions: Although the outcome of plant community succession in restored wetlands was somewhat predictable, species composition and the pace of succession varied among sites. The ability of restoration practitioners to accelerate succession through active manipulation may be contingent upon landscape context.     

Land use changes and associated environmental impacts on the Mediterranean shallow Lake Stymfalia, Greece

Land cover and land use changes affect ecological landscape functions and processes. Land use changes mainly caused by human activities, is a common reason for wetlands degradation worldwide. Lake Stymfalia, located at Peloponnese, southern Greece, is an ancient wetland with a great ecological value. Lake Stymfalia has been severely degraded and transformed during the past 60 years due to agricultural activities in the surrounding areas and watercourses alterations. In this context, we investigated the land cover/ use changes and the role of the reed beds in the terrestrialization process of this shallow wetland. This particular effort utilized remotely sensed data and Geographical Information Systems (GIS) techniques to estimate land use alterations for the period 1945–1996. Patch related landscape indices were generated to analyze impacts on landscape features. Spatial and thematic information concerning the surface area and the major land cover types of the lake for years 1945, 1960, 1972, 1987, 1992, and 1996 was obtained from aerial photographs and land surveys of the area, and was stored in the GIS database. The 1996 map was ground verified, corrected and updated to 2004 conditions. From the spatio-temporal analysis of the stored data, a permanent decrease of the open water surface has been observed between the years 1945 and 1996. The results indicated that the reed beds expanded dramatically, increasing by 89.3%, and is the predominant aquatic vegetation of the whole wetland. Open water areas and wet meadows decreased by 53.7 and 96.5% respectively. Landscape analyses and, in particular, the use of selected landscape metrics, proved useful for detecting and quantitatively characterising dynamic ecological processes. As land cover/use analysis of the wetland has shown much serious environmental degradation, conservation measures should be undertaken urgently.     

Aquatic macroinvertebrate assemblages in mitigated and natural wetlands

Many wetlands have been constructed in West Virginia as mitigation for a variety of human disturbances, but no comprehensive evaluation on their success has been conducted. Macroinvertebrates are extremely valuable components of functioning wetland ecosystems. As such, benthic and water column invertebrate communities were chosen as surrogates for wetland function in the evaluation of 11 mitigation and 4 reference wetlands in West Virginia. Mitigation wetlands ranged in age from 4 to 21 years old. Overall familial richness, diversity, density and biomass were similar between mitigation and reference wetlands (p > 0.05). Within open water habitats, total benthic invertebrate density was higher in reference wetlands, but mass of common taxa from water column samples was higher in mitigation wetlands (p < 0.05) Planorbidae density from benthic samples in emergent habitats was higher in reference than mitigated wetlands. Benthic Oligochaeta density was higher across open water habitats in mitigation wetlands. All other benthic taxa were similar between wetland types. Among the most common water column orders, Isopoda density was higher in reference wetlands, but Physidae density was higher in mitigation wetlands. Within mitigation wetlands, emergent areas contained higher richness and diversity than open areas. These data indicate that mitigation and reference wetlands generally support similar invertebrate assemblages, especially among benthic populations. The few observed differences are likely attributable to differences in vegetative community composition and structure. Mitigation wetlands currently support abundant and productive invertebrate communities, and as such, provide quality habitat for wetland dependent wildlife species, especially waterbirds and anurans.     

Island biogeography of temporary wetland carabid beetle communities

Aim The study tests if island biogeography is applicable to invertebrate communities of habitat islands in the agricultural landscape that are not fragments of formerly larger habitats. Location Thirty temporary wetlands in the agricultural landscape of northeast Germany. Methods The composition and species richness of carabid beetle communities was analysed. Habitat area, isolation, the density of temporary wetlands in the landscape, land‐use intensity and the maximum duration of flooding were recorded as independent variables. Overall species richness and wetland species richness were studied in independent regression analyses. The community composition was analysed by means of a Canonical Correspondence Analysis (CCA). A partial CCA was used to analyse the effect of the distance to the edge of the field after removing impacts of other independent variables. Results The area of the habitats and various measures of isolation (mean distances = 81–240 m) did not influence species richness or wetland species richness. The community composition was mainly determined by the land‐use intensity, habitat area did not have significant effects, and the distance to the edge of the field was the only effective isolation parameter. Short‐winged species were more often affected by the distance to the edge of the field than full‐winged species. Main conclusion There is evidence that the distances between the wetlands do not provide an effective barrier to the species dispersal and, therefore, metapopulation structures including subpopulations of multiple temporary wetlands might counteract local area effects on subpopulations. Short‐winged species, however, might be more affected by isolation than full‐winged species. As carabid beetle community structure in most early successional habitats is similar, these results may be representative of many agricultural landscape habitats. Nature conservancy concepts that aim to increase habitat area and habitat connectivity have successfully been applied to fragmented late‐successional habitats. The present study indicates that such concepts do not necessarily result in higher diversity or larger populations in early successional habitats.     

Colony persistence in waterbirds is constrained by pond quality and land use

相似文献   

Identifying nascent wetland forest conversion in the Democratic Republic of the Congo

Wetlands cover large areas in the Democratic Republic of the Congo. However, their extent and distribution have not been accurately mapped. While wetland forests remain largely undisturbed, increasing threats by anthropogenic activities have been observed in areas with high population density per arable or exploitable land. The scarcity of terra firma forests in some territories of the Democratic Republic of the Congo has forced local communities to develop cropping methods that allow for cultivation in periodically flooded areas. Assessing wetlands extent and status is critical for long term conservation of these highly vulnerable ecosystems. In this study, we use multi-source and multi-resolution optical and radar remotely sensed data and elevation derived indices to map the wetlands of the Democratic Republic of the Congo. Results showed that wetlands are a significant part of the landscape in the country, covering an estimated 440,000 km² or 19.2 % of the total country area. By combining the wetlands map with a previously produced land cover depiction of the Democratic Republic of the Congo, a map including forested wetlands as a thematic class was derived. We investigated whether high terra firma population density and low percent remaining terra firma forest are related at the lowest administrative level (Sector); specifically, we tested these two variables as predictors of wetland forest cover loss. A polynomial regression relating population and primary terra firma forest to wetland forest cover loss yielded an r ² of 0.76, illustrating a nascent and significant land cover change dynamic. Areas most at risk for future wetland forest loss lie in the western Cuvette, and include (north–south) the Sud-Ubangi, Mongala, Equateur and Mai-Ndombe Districts. By quantifying available upland forest resources and overlaying with population density, it was possible to identify stressed areas inside of the forest domain (traditionally known for having generically high levels of forest resources). Results illustrate the need for addressing issues of wetland forest management and protection in the Democratic Republic of the Congo, especially where increasing populations are exhausting primary terra firma forest resources.     

Roadsides provide refuge for orchids: characteristic of the surrounding landscape

Seminatural habitats are declining throughout the world; thus, the role of small anthropogenic habitats in the preservation of plants is becoming increasingly appreciated. Here, we surveyed the orchid flora of roadside verges in five Central European countries (Austria, Hungary, Romania, Slovakia, and Slovenia) and tested how the surrounding landscape matrix affects the overall number of species and individuals, and also different functional groups of orchids. We found more than 2,000 individuals of 27 orchid species during our surveys. According to our results, the increasing coverage of agricultural and urban areas negatively affects both the number of orchid species and individuals on roadsides. Our study further suggests that differences in the surrounding habitats affect which species are found on roadsides, since the increasing coverage of grasslands or forested areas around orchid occurrences had a significant positive effect on the number of grassland or forest‐dwelling species and individuals, respectively. Most variance in orchid numerosity and diversity was explained by the cover of the suitable habitat types of the respective taxa in the surrounding landscape of the sampling points. This highlights the importance of roadsides acting as refugia for numerous species and valuable plant communities as well as in supporting biodiversity in general.     

Beaver‐created successional gradients increase β‐diversity of invertebrates by turnover in stream‐wetland complexes

相似文献   

Vegetation trajectories of restored agricultural wetlands following sediment removal

Recognition of wetland ecosystem services has led to substantial investment in wetland restoration in recent decades. Wetland restorations can be designed to meet numerous goals, among which reestablishing a diverse native wetland plant community is a common aim. In agricultural areas, where previously drained wetland basins can fill with eroded sediment from the surrounding landscape, restoration often includes excavation to expose buried seed banks. The extent to which excavation improves the diversity of wetland plant communities is unclear, particularly in terms of longer‐term outcomes. We examined plant species diversity and community composition in 24 restored agricultural wetlands across west‐central Minnesota, U.S.A. In all study wetlands, hydrology was restored by removing subsurface drainage and plugging drainage ditches, thus reestablishing groundwater connectivity and hydroperiod (“business as usual” treatment). In half of the wetlands, accumulated sediment was removed from the basin and redeposited on the surrounding landscape (“excavated” treatment). Initially, sediment removal significantly decreased invasive species cover, particularly of hybrid cattail (Typha × glauca) and reed canary grass (Phalaris arundinacea), and increased community diversity and evenness. Over time, the effects of sediment removal diminished, and eventually disappeared by approximately 6 years after restoration. While our results demonstrate that sediment removal improves initial restoration outcomes for plant communities, longer‐term benefits require sustained management, such as invasive species control or resetting of basins through additional excavation.     

Restoration of Wetlands from Abandoned Rice Fields for Nutrient Removal, and Biological Community and Landscape Diversity

A number of experimental freshwater wetlands (150 m long × 75 m wide) with different ages since they were abandoned as rice fields, were used to analyze the prospects of multipurpose wetland restoration for such degraded areas. Nitrogen and phosphorus removal rate of the wetlands were determined monthly during the flooding season to estimate their efficiency as filters to remove nutrients from agricultural sewage. The number of wetland birds was recorded regularly to identify their habitat preferences. Both the temporal dynamics and changes in the spatial pattern of land use cover during the last 20 years were determined from aerial photographs and field analysis. All the wetlands appeared to be very efficient in the removal of nitrogen and phosphorus exported from rice fields. Usually 50–98% of the nitrogen and less than 50% of the soluble phosphorus were removed by the wetlands at any stage of restoration. Wetland birds preferred wetlands with intermediate plant cover for resting and sleeping activities better than rice fields and either very open wetlands or very dense ones with tall vegetation. Apart from the improvement in water quality and the restoration of natural habitats, restoration of wetland belts around lagoons will increase spatial heterogeneity and diversity of the landscape.     

Forecasting climate change impacts on the distribution of wetland habitat in the Midwestern United states

Shifting precipitation patterns brought on by climate change threaten to alter the future distribution of wetlands. We developed a set of models to understand the role climate plays in determining wetland formation on a landscape scale and to forecast changes in wetland distribution for the Midwestern United States. These models combined 35 climate variables with 21 geographic and anthropogenic factors thought to encapsulate other major drivers of wetland distribution for the Midwest. All models successfully recreated a majority of the variation in current wetland area within the Midwest, and showed that wetland area was significantly associated with climate, even when controlling for landscape context. Inferential (linear) models identified a consistent negative association between wetland area and isothermality. This is likely the result of regular inundation in areas where precipitation accumulates as snow, then melts faster than drainage capacity. Moisture index seasonality was identified as a key factor distinguishing between emergent and forested wetland types, where forested wetland area at the landscape scale is associated with a greater seasonal variation in water table depth. Forecasting models (neural networks) predicted an increase in potential wetland area in the coming century, with areas conducive to forested wetland formation expanding more rapidly than areas conducive to emergent wetlands. Local cluster analyses identified Iowa and Northeastern Missouri as areas of anticipated wetland expansion, indicating both a risk to crop production within the Midwest Corn Belt and an opportunity for wetland conservation, while Northern Minnesota and Michigan are potentially at risk of wetland losses under a future climate.     

Extent, properties, and landscape setting of geographically isolated wetlands in urban southern New England watersheds

We assessed the extent and characteristics of geographically isolated wetlands (i.e., wetlands completely surrounded by upland) in a series of watersheds in the urban northeast US. We applied a previously developed index of urbanization to a sample of 10 watersheds selected at random from a set of 30 watersheds whose boundaries lay within the borders of Rhode Island, USA. The index of urbanization in our sample watersheds ranged over more than an order of magnitude and increased with increasing amount of urban land use in the watersheds (r ² = 0.51, F = 8.22, P = 0.02). The density of isolated wetlands in the watersheds averaged 1.93 ± 0.21 wetlands km⁻² and comprised 38.2 ± 1.77% of all wetlands. Isolated wetlands were smaller than those connected to other waters (non-isolated), and accounted for 6.01–16.5% of the total wetland area in the watersheds. The area of isolated wetlands as a percent of all wetland area significantly increased with increasing watershed urbanization (r ² = 0.62, F = 12.9, P = 0.007). Isolated wetlands were predominantly deciduous forested wetlands, and urban land cover in the 50 m buffer surrounding isolated wetlands was significantly higher than in the 50 m surrounding non-isolated wetlands. The proportion of urban land cover was greater in a 150 than a 50 m buffer surrounding the wetlands. Our results suggest that an increase in the index of urbanization of 50 will result in 7% of the watershed’s wetlands being lost from federal protection. These findings indicate that the process of urbanization, along with accompanying habitat fragmentation, may result in an increase in the vulnerability of wetlands to loss and degradation and therefore has implications for the management and conservation of geographically isolated wetlands.     

Palm Community Structure and Land Cover Changes in the San Juan Biological Corridor, Costa Rica

Biological corridors connect forest fragments, minimize the isolation of remaining habitats, and reduce the negative impact of fragmentation. Important studies in a corridor include documentation of land cover and the relationships between the plant community and land cover changes. In this study, density, species richness, and species diversity of palms of different life stages and growth forms were documented in forty 10 × 10 m census plots in the San Juan Biological Corridor (SJBC), Costa Rica. Changes in the percent cover of agricultural, partially forested, and forested lands from 1979 to 2001 were quantified based on existing land classifications. Associations between these palm community characteristics and the reduction in forest cover in 200-m and 1000-m annuli surrounding the census plots were investigated. These palm community characteristics were spatially interpolated to examine their distributions in SJBC. The community characteristics of palms of all growth forms varied among census plots. Forested lands in SJBC decreased from 88 percent to 26 percent during 1979–2001, while the percentage cover of agricultural and partially forested lands increased. The annual deforestation rate was 0.8 percent from 1979 to 1992, 4.3 percent from 1992 to 1997, and 7.5 percent from 1997 to 2001. Overall, palm community characteristics of small-canopy palms, dwarf palms, and cotyledonous seedlings had lower values with a lower percentage forest cover. The spatially interpolated palm community characteristics were consistently greatest within or adjacent to protected areas. This study demonstrates the complexity of the landscape and the effect of loss of adjacent forests on the palm community.     

Vegetation Similarity and Avifaunal Food Value of Restored and Natural Marshes in Northern New York

Measuring the success of wetland restoration efforts requires an assessment of the wetland plant community as it changes following restoration. But analyses of restored wetlands often include plant community data from only one time period. We studied the development of plant communities at 13 restored marshes in northern New York for 4 years, including 1 year prior to restoration and 3 years afterwards. Restored wetlands ranged in size from 0.23 to 1.70 ha. Four reference wetlands of similar basin morphology, soil type, and size (0.29–0.48 ha) that occurred naturally in the same area were studied as comparisons. Dike construction to restore hydrology disturbed the existing vegetation in some parts of the restored sites, and vegetation was monitored in both disturbed and undisturbed areas. Undisturbed areas within the restored sites, which were dominated by upland field grasses before restoration, developed wetland plant communities with lower wetland index values but comparable numbers of wetland plant species than the reference wetlands, and they lagged behind the reference sites in terms of total wetland plant cover. There were significantly more plant species valuable as food sources for wetland birds, and a significantly higher percent cover of these species, at the undisturbed areas of the restored sites than at the reference wetlands. Areas of the restored sites that were disturbed by dike construction, however, often developed dense, monospecific cattail stands. In general, the plant communities at restored sites became increasingly similar to those at the reference wetlands over time, but higher numbers of herbaceous plants developed at the restored sites, including food plants for waterfowl, rails, and songbirds. Differences in shrub cover will probably lessen as natural recolonization increases shrub cover at the restored sites. Natural recolonization appears to be an effective technique for restoring wetlands on abandoned agricultural fields with established plant cover, but it is less successful in areas where soil has been exposed by construction activity.