Rapid bioassessment techniques used for aquatic macroinvertebrate and plants in wetlands were tested in New South Wales, Australia. Wetlands surveyed ranged from coastal wetlands in the North and Central parts of the State, to tableland wetlands west of Sydney. Wetlands varied in dominant vegetation, hydrology, substrate and level of human impairment. Different options for sampling (mesh size, live-picking times, quadrat size, number of samples or quadrats) and analysis (taxonomic resolution, transformations, biotic indices, multivariate and univariate analyses) are compared to determine optimal sampling effort and evaluate the effects of errors or variability. Results show that, for wetlands of New South Wales, sampling procedures developed and tested in streams and other regions of Australia may not be the most efficient. Using the data from 21 wetlands in New South Wales, a number of analytical techniques were evaluated for the effects of errors. Results show that species-level multivariate analysis is more sensitive in detecting less obvious differences between wetlands (i.e., small effect sizes), while family-level analyses are more appropriate for large effect sizes. A modified waterplant index was developed that is simpler and has a wider application than the other Australian options available. Inherent problems in each index tested were addressed. The results show that the process of summarising a large amount of information into a single value will result in the loss of both information and variability between samples and this cumulative effect of error may effect the assessment of wetland condition. The practical outcome of this thesis is a set of standardised steps to assess wetland quality using biological assemblages. The results show that protocols and indices for rivers are not directly transferable to palustrine, vegetation dominated wetlands. I present protocols that are more appropriate to wetlands and recognise that each protocol would need to be adapted for each wetland type. Despite the need for flexible protocols, I promote the need for a standard approach to wetland sampling and the need for consideration of the effects of errors in sampling designs. This study highlights the need for more research on the response of specific stressors to wetlands flora and fauna. The results from this study also show that wetland macroinvertebrates and plant communities can be used as surrogates in multivariate analyses for detecting large differences between wetlands (wetland types) but that impact assessment requires more detailed investigations including species identification and careful consideration of the choice of reference and control sites. In conclusion I emphasise the need for scientific rigour in the use of biological indicators and consideration of the effects of errors and implications to sampling designs. / Doctor of Philosophy (PhD)

Contamination of recreational and shellfish waters with fecal pollution is a major water quality issue with associated economic impacts and human health risks. Reliable fecal source identification and rapid, quantitative analyses are essential components of risk assessment. Enteric bacteria that are endemic to specific hosts have a potential role as public health indicators of fecal pollution. Building on previous work to discriminate ruminant and human fecal contamination, we cloned class Bacteroidetes 16S rRNA genes from pig, elk, dog, cat, and seagull fecal DNAs. Unique restriction patterns were identified among clones from each of the host species using Terminal Restriction Fragment Length Polymorphisms (T-RFLP). Clones exhibiting unique patterns were sequenced and analyzed phylogenetically, along with human, horse, and cattle sequences recovered from previous work. The analysis revealed both endemic and cosmopolitan (global) host distributions. The sequence data were used to identify host-specific genetic markers for pig and horse feces, and to design PCR primers that identify these sources of fecal pollution in water. There was a high degree of sequence overlap among the fecal Bacteroidetes of wild and domestic ruminants, and among human, domestic pet, and seagull Bacteroidetes. We compared fecal Bacteroidetes rRNA genes from these hosts using subtractive hybridization, a method that identifies differences between closely related genomes or gene sequences. A Bacteroidetes rDNA marker that distinguishes elk and cow feces was identified, as well as a host-specific marker for dog fecal Bacteroidetes. The four newly designed PCR primers were tested for specificity and sensitivity, and the dog primer was successfully used, along with the human and ruminant-specific primers, in a collaborative study comparing fecal source tracking methods. We also developed a real time Taq nuclease assay for quantification of fecal Bacteroidetes 16S rDNA, and compared it with an EPA-approved enumeration method for the current standard public health indicator, Escherichia coli, in serial dilutions of sewage primary influent. There was a strong, positive correlation between the methods, and the Taq nuclease assay was sensitive and much more rapid than the E. coli assay. PCR source identification and enumeration of fecal Bacteroidetes 16S rDNA show promise for application in a health risk-based analysis of fecal pollution. / Graduation date: 2005

The focus of this research was to develop bacterial community indicators of stream sanitary and ecological condition. The first study compared substrate utilization patterns between centrifuged and uncentrifuged split samples. We found a shift in the relative proportion of each group of bacteria following centrifugation, with a marked increased in the fecal coliform group and relatively fewer heterotrophic and total coliform bacteria. Centrifuged samples consistently responded faster and oxidized more substrate than did their uncentrifuged counterparts. Substrate utilization patterns of centrifuged sub-samples from 19 sites showed better separation between Willamette Valley and Cascade ecoregions than did the uncentrifuged sub-samples in ordination space. We recommend developing microtiter plates with substrates specific types of environmental stress. The second study determined the minimum volume of water needed and the maximum time and temperature that bacteriological water samples captured on a membrane filter can be held in guanidine isothiocyanate buffer (GITC) prior to DNA extraction for community fingerprint analysis. We found 100 ml water samples yielded more information than the 50 ml or the 250 ml water samples and observed a marked decrease in information for samples that were held at room temperature for more than 24 hours. We concluded that 100 ml samples were optimal for bacterial community DNA fingerprint analysis. Furthermore, we recommended transporting filtered water samples held in GITC on ice and keeping the samples frozen until DNA is extracted for further analysis. The third study addressed questions of sampling error and response variability of two PCR-based indicators, bacterial community-level Terminal-Restriction Fragment Length Polymorphisms and Bacteroidetes ruminant and human specific fecal source tracking markers. We found the T-RPLP and Bacteroidetes markers to show very little sampling error, and suggested collecting a single 1-liter water sample. A high turbidity scenario resulting in higher fecal pollution and lower bacterial species richness explained why decreased TRF richness was strongly associated with high fecal coliform density, turbidity, and human Bacteroidetes detection. We propose that in times of increased turbidity, a disturbance in the bacterial community occurs, reducing bacterial richness and increasing a few types of stress-resistant fecal bacteria. / Graduation date: 2004

Traditional public health bacterial indicators of water quality and the Biolog�� system were evaluated to compare their response to other indicators of stream condition with the state of Oregon and between ecoregions (Coast Range, Willamette Valley, Cascades, and Eastern Oregon). Forty-three randomly selected Oregon rivers were sampled during the summer low flow period in 1997 and 1998. Testing included heterotrophic plate counts (HPC), total coliforms, fecal coliforms, E. coli, and Biolog�� GN plates. Statewide, HPC correlated strongly with physical habitat and chemistry indicators while fecal coliforms and E. coli were highly correlated only with the river chemistry indicators. Total coliform bacteria did not correlate with either of the above environmental indicators. Dividing the sites by ecoregion, Eastern Oregon was characterized by high HPC, fecal coliforms, E. coli, nutrient loads, and indices of human disturbance, whereas the Cascades ecoregion had correspondingly low counts of these indicators. The Coast Range reflected statewide results and the Willamette Valley presented no consistent indicator pattern. Attempts to separate ecoregions with the Biolog system were not successful nor did a statistical pattern emerge between the first five principle components and the other environmental indicators. Our research has shown that traditional public health microbial indicators may, however, be useful in measuring the effects of anthropogenic stress over large spatial scales. / Graduation date: 2000

The Environmental Monitoring and Assessment Program (EMAP) site selection protocol was used to generate a random sample of streams throughout the state of Oregon. One hundred and forty-six selected streams were sampled during the summer, low-flow period of 1997. Traditional microbial public health indicators, including heterotrophic plate counts (HPC), total coliforms (TC), fecal coliforms (FC) and E. coli, were enumerated using the membrane filtration technique. Nearly 17% (3.4-23.6%, 95% C.I.) of streams exceeded the current state of Oregon standard for water-contact recreation. Levels of bacteria were also shown to differ significantly by ecoregion. The Cascades ecoregion had low levels of all types of bacteria. The Willamette Valley, Columbia Plateau and Snake River Basin had high levels of one or more groups of bacteria measured. Twenty-six sites were resampled and FC and E. coli estimates were not significantly different for the different sampling dates. Biolog GN plates were used to provide a measure of the functional diversity of microbial communities for the same streams as above. Two groups were formed based on inoculum density and Biolog GN plates were analyzed using principal component analysis (PCA). The first few principal components explained nearly half of the variation of the data in both groups. Principal components were correlated with the average carbon source utilization, levels of coliform bacteria, and ecoregions. These results indicate that patterns produced by Biolog GN plates may be useful in the assessment of ecological condition of freshwater streams. Subsequent publications will explore the relationships between the pattern of substrate utilization of Biolog GN plates with other indicators of ecological function. / Graduation date: 1999

Thesis (M. A.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

Thesis (M.S.)--North Carolina State University, 2003. / Title from PDF title page (viewed on Jan. 13, 2005). Includes vita. Includes bibliographical references (p. 76-86).

published_or_final_version / Botany / Master / Master of Philosophy

Eighteen-month-old adult female Peking Ducks (Anas platyrhynchos) were placed in the Ottawa, and St. Lawrence rivers for periods varying between 14 and 72 days during the summers of 1987, 1988 and 1989. Two stations located on Lake St. Francis were selected for establishing a time-curve for the accumulation of organochlorine pesticides (OC) and PCB in the liver. The ducks rapidly lost weight during their first 20 days in the natural environment but stabilised after about 40 days. This weight loss was caused by the stress of habitat, change, the energetic costs in searching for food and moulting. / There was a rapid increase in the number and concentration of contaminants detected in the livers of ducks exposed to pollutants in the natural environment. The concentrations found were low (99.9% of OC-PCB are under 1 mg/kg wet weight basis). These individuals were generally from 10 to 1000 times more contaminated than the control ducks for 40 days of exposure. / The use of domestic ducks as bioindicators has several advantages. They can be obtained readily and cheaply, and the sex and age of the individuals can be controlled. Since they are not very mobile and therefore are easy to locate, they provide information on the contamination (metals or OC-PCB) of specific sites. The use of Peking ducks has some limitations. There is a need for appropriate habitats to ensure their survival, and there is considerable stress caused by transferring the ducks from the breeding farm to the natural environment, thereby obliging them to search for food and exposing them to predators and poachers. (Abstract shortened by UMI.)

The degradation of wetlands and loss of their associated ecosystem services is widely recognised in South Africa, however, at present there is no standard method of biologically assessing wetland health in this country. Internationally, particularly in the U.S.A and Australia, wetland bioassessment techniques using macroinvertebrates are well established. A number of these wetland bioassessment protocols have been derived from local river biomonitoring techniques, as there is a belief that river and wetland ecology and macroinvertebrate assemblages at family level are similar. However, some authors consider wetland macroinvertebrate assemblages and ecological processes to differ greatly from those found in rivers, and believe that such techniques are not transferable. South Africa has a well established macroinvertebrate biomonitoring protocol for rivers called SASS5 (South African Scoring System Version 5). This study is a preliminary investigation into the extent to which the SASS5 scoring system is applicable to the assessment of nutrient enriched wetland water quality. Macroinvertebrates are particularly suitable as biomonitoring tools: they respond to a variety of stressors, have life cycles that allow for integrated responses to episodic pollution, and are relatively easy to identify to family level. When selecting wetlands for the development of a biomonitoring protocol, wetlands should all be of the same; classification (Le. palustrine), geomorphological and climate setting, hydrological regime and dominant vegetation class. Sampling was restricted to sedge-dominated palustrine wetlands in the midlands of KwaZulu-Natal, with similar hydro-geomorphological settings. Due to wetlands and rivers having different biotopes (e.g. no riffles present in wetlands), the SASS5 sampling protocol could not be used, thus a pilot investigation was undertaken to derive a suitable sampling technique for \ collecting a representative and diagnostic sample of aquatic macroinvertebrates from a wetland. This technique was developed based on published methods. Both sweep net and activity trap sampling were conducted, and each evaluated for their effectiveness at macroinvertebrate collection. Sweep net sampling was tested over a range of sweep intensities (2-6 sweeps), and activity traps were placed at four different depths: at the water surface, just below the surface, 0.10.15m below surface and on the substrate. A total of 32 taxa identified to family level were identified in the samples. Taxon diversity and composition did not differ in the activity traps placed at the four depth locations. Taxon diversity did not differ significantly between different sweep intensities; however there was a significant difference in taxon composition between the different sweep intensities and between activity trap and sweep net samples (p<0.05). Sixty-eight percent of taxa appeared more frequently in sweep net sampling compared to activity trap sampling. Six taxa were found exclusively in sweep net samples, and two taxa were recorded exclusively in activity traps. There was no trend in either method collecting more or missing any unique trophic group. In conclusion, activity traps are not required to supplement sweep net data, and a technique using a sweep net with a sweep intensity of five would be suitable to collect a representative sample of wetland macroinvertebrates. Using the derived technique, four reference and three wetlands impacted by dairy effluent were sampled. Six macroinvertebrate samples were collected from each of the seven wetland, together with data for selected physico-chemical variables, macrohabitat condition, biotope suitability and organism detectability. For each sample, the macroinvertebrates were identified and assigned a predetermined SASS5 tolerance score between 1 and 15, with higher scores indicating increased sensitivity to poor water quality. 11 A total of 39 taxa, identified to family level, were collected during sampling. SASS5 scores ranged from 15-82. Five of the wetlands had mean SASS5 scores of between 46 and 59. Five of the wetlands had an intra-wetland SASS5 score range greater than 30. ASPT values ranged from 3.3 to 5.5, and few high scoring (~8) taxa were collected. There was no significant difference in SASS5 scores between samples collected above, at and downstream of an effluent discharge point within the same impacted wetland. SASS5 scores for reference wetlands were also not significantly higher than those recorded for impacted wetlands. Comparison of ranked SASS5 scores and environmental data did suggest a relationship between the variables, but was not significant. Based on the SASS5 score water quality guidelines, all sampled wetlands were considered to have impacted water quality; however, this was not supported by the macrohabitat and physico-chemical results. Possible reasons for the low SASS5 scores include: the lack of 'stones in/out current' biotopes in wetlands, lower levels of dissolved oxygen present compared to rivers, and the limited detectability of organisms due to large amounts of substrate in the samples. A wetland adaptation of SASS5 would require the reassignment of modified scores to certain taxa based on their distribution in wetlands of varying water quality. The SASS5 score level of 100 and the ASPT value of 6 (as specified in the SASS5 score water quality guidelines) were found to be inappropriate for wetlands. It is suggested that, either the range of taxa tolerance scores be increased (1 to >15), or the score level of 100 be lowered. The ASPT value should also be reduced. Although SASS5 appears unsuitable for assessing wetlands, variations in taxon composition between sampled wetlands, identified through CA analysis, suggests that macroinvertebrates are responsive to changes in wetland condition, and thus have potential as indicators of wetland water quality. Nine taxa responsive to the presence of nitrogen have been identified as being potentially good indicators. iii Further research should focus on the testing of SASS5 throughout the year, in a range of wetland types, and in wetlands moderately to severely impacted by pollutants other than dairy effluent. It is recommended that a habitat or biotope index be developed and used in conjunction with any future wetland macroinvertebrate bioassessment protocols. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2005.