Upper Midwest Environmental Sciences Center
Summary
of Monitoring Findings for Fiscal Year 2001
Monitoring Activities and Highlights
Daily discharge data were collected at St. Paul, Minnesota; Winona, Minnesota; Keokuk, Iowa; and St. Louis, Missouri, for the Upper Mississippi River and at Kingston Mines, Illinois, for the Illinois River (Table 1). The period of record for the five stations ranged from 63 to 123 years. Mean daily discharge (m3/sec) was calculated for each station for each water year (defined as the period from October through the following September). Compared to historic means, mean discharge for the 2001 water year was slightly below normal on the Illinois River but well above normal for all of the stations on the Mississippi River.
Stations in Minnesota, St. Paul (Figure 2),
and Winona (Figure 2) showed peak flows
between the 50- and 100-year recurrence intervals. Mississipi River
discharges resulted from higher than normal precipitation in fall
2000, a delayed snow melt, and record rains in April 2001 (Mitton 2002).The
impact of these record discharges is also seen, although less dramatically,
at the more downstream stations of Keokuk (Figure
2) and St. Louis (Figure 2).
Water surface elevations (in feet above mean sea level) for the Long Term Resource Monitoring Program (LTRMP) study areas are shown in Figures 3–8.
Year 2000 Land Cover/Land Use and Aquatic Areas GIS Database
The Upper Midwest Environmental Sciences Center (UMESC) Geospatial Services Laboratory used 1:15,000-scale true color and 1:24,000-scale color infrared aerial photography acquired in late summer 2000 to create detailed land cover/land use (LCU) data and georeferenced photo mosaics for the Upper Mississippi River System (UMRS) trend pools. The color infrared photography and a generalized 31-class vegetation classification system were used to categorize more than 800,000 acres of floodplain LCU using a minimum mapping unit of 1 ha (approximately 2.5 acres). This is the first year of a 3-year effort to complete a 2000 LCU data set for the entire UMRS floodplain. Once it is complete, this data set will provide an 11-year time step since the 1989 systemic coverage, allowing examination of changes including those resulting from the 1993 flood and Environmental Management Program (EMP) habitat projects. In addition to the 2000 LCU data set, systemic, or near-systemic, LCU data sets will exist for the 1890s, 1975, and 1989.
The true color photography was scanned at 200 dots per inch and referenced to the Earth using the Universal Transverse Mercator Zone 15 and 16 projections and the NAD27 and NAD83 datums. The photographs were then compiled and compressed into 2 m/pixel .sid images using the MrSID Geospatial Encoder. This format is readable by most GIS programs and can compress large images into a much smaller file. The UMRS Pool 13 (Figure 9) was converted from 1.5 GB of TIFF images into an easily downloadable 22 MB mosaic with little loss of resolution. These mosaics are designed primarily as background reference information on which similarly referenced GIS data, such as LCU or aquatic vegetation sampling sites, can be overlaid (Figure 9).
Both the 2000 LCU and 2000 photo mosaics are served over the Internet via the UMESC Web site.
Investigation of Remote Sensing Technology for Land Cover Acquisition
The UMESC began mapping floodplain vegetation in 1989 by interpreting 1:15,000-scale color infrared aerial photography to a 150-class, genus-based scheme. While this method generated land cover maps with unprecedented detail, it was time-consuming and expensive to do for the entire system. A reassessment of the mapping in 1998 determined that the classification scheme could be simplified to 31 classes, while still meeting most analysis needs of UMRS resource managers.
In this study, we are investigating whether efficiency can be further improved by changing the type of remotely sensed images used to produce floodplain maps. We are evaluating three types of images: aerial photography at a scale of 1:24,000; high spatial resolution (4-m) data from a new, commercial satellite (IKONOS); and data from an airborne, hyperspectral sensor (AISA). Hyperspectral sensors record energy over narrow, but contiguous, ranges of the electromagnetic spectrum. When particular cover types have unique energy reflectance characteristics, computer algorithms can be used to help distinguish between them, allowing a more automated mapping process. The IKONOS and AISA sensors are designed to help discriminate between cover types that are either too limited in area or too similar in their reflectance characteristics to have been mapped by previous sensors.
We selected three representative areas within the UMRS floodplain, each approximately 100 km2, as test sites for this study. These sites are portions of Pools 8 and 22 of the Mississippi River and a portion of La Grange Pool of the Illinois River. We acquired aerial photography at both 1:15,000 and 1:24,000 scales and IKONOS images for all three study sites in late summer or early fall 2000. The interpretation of the 1:15,000 photography was used as reference data against which the 1:24,000- and IKONOS-based maps could be compared.
Initial results indicated that the map based on 1:24,000 scale photography agreed with the reference data 55% of the time. The reference data was interpreted using a 1-acre minimum mapping unit (MMU), while an MMU of about 2 acres was used when interpreting the smaller-scale photography. Agreement between the two maps increased significantly (to 71% overall) when reference polygons smaller than 2 acres were excluded. The IKONOS classification agreed with the reference map 34% of the time. Accuracy, or levels of agreement, such as these are coarse metrics and should be viewed as relative measures. Reference data are rarely 100% accurate. In addition, inherent differences between photo-interpretation and automated classification make the two processes difficult to compare. However, these early results indicate that smaller scale aerial photography is more suitable than high-resolution satellite data for mapping the present LCU types.
Hyperspectral and reference data (1:15,000 scale aerial photography)
were acquired for the study area in Pool 8 in August 2001. Classification
of the hyperspectral data will be done by a private contractor and assessed
by the UMESC in FY2002. A final report
incorporating all phases of this study will also be completed in FY2002.
A second set of surveys was completed by the UMESC to provide direct support to Habitat Rehabiltation and Enhancement Project (HREP) planning and evaluation. Surveys were selected based on input from managers associated with HREPs. These included Indian Slough and Peterson Lakes, Pool 4; Lake Onalaska dredge cuts, Pool 7; Cold Springs, Pool 9; portions of lower Pool 11; selected backwaters in lower Pool 12; and Browns Lake dredge cuts, Pool 13. Standard GIS data sets were created for the survey data. Although these surveys were not selected to meet the short-term objective of the systemic mapping (e.g., complete pools with existing data), the data were included in the systemic database when appropriate.
Sedimentation is a major concern for resource managers of the Upper Mississippi River. This process will largely determine the fate of off-channel areas in the Upper Mississippi River (Fremling and Claflin 1984; Nielsen et al. 1984) and the fate of the biota that rely on those habitat conditions. Backwater lakes are of particular interest because they periodically provide a lentic environment in a lotic system. Most backwaters lakes are typically shallow (<1 m) during the low discharge conditions; thus, additional sediment deposition could lead to the loss of these habitats. Studies to determine sedimentation rates and processes were among the recommendations developed by the Sediment Transport and Geomorphology Work Group convened to define informational needs for the Upper Mississippi River (Gaugush and Wilcox 2002).
In 1997, we redesigned the LTRMP short-term sedimentation monitoring study conducted by Rogala and Boma (1996) to overcome problems in the original design. We modified the original protocol to include random sampling and extended the length of transects to include terrestrial banks adjacent to backwaters. Transects in Pools 4, 8, and 13 were surveyed annually between 1997 and 2001. Objectives of this monitoring study included estimating sedimentation rates during a 5-year period beginning in 1997, determining variability in sedimentation rates along transects and among years, and examining spatial and temporal variability to identify variables of interest for future modeling efforts. Overall, we addressed spatial and temporal variability in sedimentation rates at finer scales than previously investigated.
Water quality monitoring within the Program has focused on factors in the Upper Mississippi River and its major tributaries that significantly affect aquatic habitats, including physicochemical features, suspended sediment, and major plant nutrients. Sampling in 2001 followed the design that was established in 1993 and significantly modified (reduced by about 30%) in 1999. The design combined quarterly sampling episodes (stratified random sampling) with 4-week fixed-site sampling conducted at tributaries, major inflows and outflows from the study reaches, and other locations of specific interest. Stratified random sampling (about 100 sites per study reach) provides unbiased, seasonal information on water quality across broad areas, such as entire navigation pools. Fixed-site sampling (about 20 sites per study reach) provides more continuous information at specific locations.
Water quality monitoring in the LTRMP has steadily evolved and adapted to gain greater efficiencies. For example, starting in spring 2001, electronic data sheet software was phased into service for water quality field operations (Figure 10). Deployment was completed by August 2001. The field-rugged laptop computers used to run this software acquire data directly from the field monitoring instruments and thus eliminate manual transcription. The software also includes extensive error checking of data as it is entered manually or recorded directly from the field instruments. This allows on the spot detection and correction of nearly all data errors. This approach eliminated the use of a separate data entry contractor and has dramatically reduced the resources needed to detect and correct errors in the LTRMP database.
A primary goal of the LTRMP water quality component is to track changes in the physical and chemical conditions that define suitable habitat with the UMRS, such as temperature, turbidity, dissolved oxygen, and nutrient levels that occur naturally and in response to management actions (e.g., Pool 8 drawdown). For example, dissolved oxygen beneath the ice is a major determinant of suitable wintering habitat and a continuing concern of resources managers. Several HREP projects are intended to improve winter oxygen in Upper Mississippi River backwaters and LTRMP data show the typical levels and long-term variations in backwater dissolved oxygen as it responds to snow and ice cover that are mostly beyond management control (Figure 11). This allows for assessment and prediction of low oxygen conditions and helps place the potential or effect of management action into a meaningful context.
The status and trends of major plant nutrients, particularly nitrogen, and suspended sediment in the Mississippi River basin, continue to receive national attention. The LTRMP water quality monitors delivery of these materials to the river from the major tributaries and is the only source of this information for most of the Upper Mississippi River. The UMRS receives large inputs of the major plant nutrients (nitrogen and phosphorus) that produce harmful algal blooms in lakes and reservoirs. Although less well documented, large rivers also respond to these nutrient inputs. An example is provided by Pool 26 where suspended chlorophyll a (a measure of algal biomass) in the main channel during stratified random sampling in summer has shown an increase since 1996 (Figure 12) and is typically in the "nuisance bloom" range (>30 µg/L). High concentrations of suspended algae transported into the pool from the Illinois River contribute to this pattern (Figure 13).
Long-term monitoring of the UMRS is needed to place year-to-year changes
in the system into an appropriate perspective. In a large river, the long-term
cycle of flood and drought has major influence on ecological structure
and functioning. The period of monitoring by LTRMP from 1988 to 2001 has
covered only about one of these long-term cycles (Figure
14). The Program has not yet observed an extreme drought (as seen
in 1988), but has recorded the effects of two major floods (1993 and 2001).
The important role of large floods in transporting nitrogen (primarily
nitrate) out of the system and into the Gulf of Mexico is clearly shown
by LTRMP data (Figure 15).
Submersed aquatic plants are an important component of the Upper Mississippi River ecosystem. They provide food for migratory waterfowl, improve water quality by stabilizing sediments and assimilating nutrients, provide spawning and nursery areas for fish, and support invertebrate populations by providing structure and surface area.
The 2001 sampling of submersed aquatic vegetation was conducted at Pools 4, 8, 13, 26, and La Grange Pool in the UMRS using a stratified random sampling protocol described in Yin et al. (2000). Sampling points were distributed randomly in shallow aquatic areas that would be 2.5 m or less deep at flat-pool conditions. Areas deeper than the cutoff depths were not sampled and presumed to support no submersed aquatic vegetation.
The estimated percent frequencies of submersed aquatic vegetation in the shallow water areas in Pools 4, 8, 13, 26, and La Grange Pool were 36.9, 47.5, 41.7, 0.3, and 0%, respectively. The longitudinal pattern of the five pools is the same as revealed in the previous 3 years from 1998 to 2000 (Figure 16). This pattern is also consistent with the longitudinal pattern displayed in the aerial photographs of 1989 that submersed aquatic vegetation was abundant in the Upper Mississippi River reaches upstream of Lock and Dam 13, but rare or negligible elsewhere in the UMRS (Rogers and Theiling 1999). A deviation from this longitudinal pattern was observed following the 1987-1989 drought and in 1993 after an unusually high flood disturbance, when little submersed aquatic vegetation occurred in the entire UMRS. We did not sample the entire UMRS in 2001, but we have no reason to suspect a deviation from the normal pattern occurred in 2001.
The within-pool distribution patterns of submersed aquatic vegetation were highly heterogeneous but remained little changed since 1998. Submersed aquatic vegetation was sparse and species-poor in upper Pool 4 above Lake Pepin compared with the lower Pool 4 below Lake Pepin. Submersed aquatic vegetation distributed widely throughout Pool 8 except in the lower end where water depth generally exceeded 1 m. A considerable amount of submersed aquatic vegetation was recorded in Pool 13, most of which occurred in the contiguous backwaters and impounded areas at the lower half of the pool. An insignificant amount of submersed aquatic vegetation was found in Pool 26, in the isolated backwater areas of the Illinois River. In La Grange Pool, submersed aquatic vegetation was found to exist in the lakes on the Illinois River floodplain and was absent in the river's backwater areas. Modeling efforts determined that the distribution of submersed aquatic vegetation appears to be correlated with the physical parameters of water depth, current velocity, and fetch. Other factors not included in the model, such as sediment, also can affect submersed aquatic vegetation.
The spatial extents of submersed aquatic vegetation in the five key pools
have remained stable since 1998 based on the frequencies of the sites
sampled that supported submersed aquatic vegetation (Figure
16). However, the amount of vegetation per site has displayed a trend
of steady decline in Pool 8 and a trend of a steady increase in Pool 13
since 1998 (Figure 17). Factors responsible
for the two opposite trends are yet to be determined.
Fish monitoring within the Program measures long-term trends in species abundance, community composition, and community structure within multiple habitat classes in each of the six study areas using standardized sampling protocols (Gutreuter et al. 1995). The sampling design is based on the overall hypothesis that fish are influenced by physical, chemical, and hydrological changes to the river system such as sedimentation, nutrient loading, water level manipulation, and biotic invasion by exotic species. Detection of changes in Upper Mississippi River fish populations will provide evidence for forming and testing hypotheses directed at determining the factors causing the observed changes.
In 2001, fish were sampled at stratified random and fixed-site locations with 10 types of active and passive sampling gear in each of the six study areas (Gutreuter et al. 1995; Burkhardt et al. 2000). A total of 2,520 samples were collected across the six study areas, yielding nearly 410,000 fish and 63-75 species per study area. No major deviations from the established sampling design occurred in 2001. Details on minor effort reallocation changes and missed samples per study area can be found within the 2001 LTRMP Annual Status Report for Fish (http://www.umesc.usgs.gov/reports_publications/ltrmp/fish/2001/fish-srs.html).
In 2001, fish monitoring continued to provide valuable detection data on several state-listed threatened and endangered species. Species collected included blue sucker (Cycleptus elonatus; Pool 8 and Open River), lake sturgeon (Acipenser fulvescens; Pool 26), river redhorse (Moxostoma carinatum; Pool 8), bluntnose darter (Etheostoma chlorosomum; Pool 13), western sand darter (Ammocrypta clara; Pools 13), freckled madtom (Noturus nocturnes; Pool 13), pugnose minnow (Opsopoeodus emiliae; Open River), paddlefish (Polyodon spathula; Pool 4, Open River), mooneye (Hiodon tergisus; Open River), Mississippi silvery minnow (Hybognathus nuchalis; Open River), silver chub (M. Storeriana; Open River), speckled chub (Macrhybopsis aestivalis; Pool 8), and river darter (Percina shumardi; Open River).
Several exotic fish species have entered the UMRS since LTRMP monitoring was initiated in 1989. Fish monitoring continues to provide important information on the relative abundance, habitat associations, and spread of these species through the UMRS. In 2001, all exotic species collected were limited to Pool 26, Open River, and La Grange Pool. Species collected include silver carp (Hypothalmichthys molitrix), bighead carp (H. nobolis), grass carp (Ctenopharyngodon idella), goldfish (Carassius auratus; Pool 26 and La Grange Pool), and white perch (Morone americana; La Grange Pool). No notable increases in abundance from 2000 monitoring efforts were noted for any of these species.
In 2001, analyses on data collected from a spatially expanded monitoring
effort in 2000 were completed. The objective of this effort was to assess
whether fish community data collected in the six study areas can be used
to make inferences on fish community characteristics across the entire
UMRS. Standardized LTRMP electrofishing methods were used sample
fish communities in a spatially expanded set of Upper Mississippi
River navigation pools (n = 14)
from June to October 2000. Multivariate statistical methods were used
to assess the degree to which pools were similar. Results indicated two
major groupings of UMRS pools based on similarities in community composition
(Figure 18) and community structure (Figure
19). Additional structure was also evident beyond these two coarse
spatial groupings. In general, pools were most similar to nearby pools
and most dissimilar to distant pools. This result lends support to the
premise that LTRMP fisheries data could potentially be used to make inferences
about community composition and structure to the entire UMRS because
the six current study areas are evenly distributed within the major pool
groupings identified in this study.
In 2001, we also completed an assessment of sampling efficiencies within
the fisheries monitoring program. Results and recommendations can be found in Ickes and Burkhardt 2002.
Fishery Resources of Deep Channels of the UMRS
The fishery resources of the major channels of large rivers are poorly known (Baker et al. 1991), mainly because sampling is difficult (Casselman et al. 1990). As a result, historic perceptions of patterns of abundance of fishes in large rivers are shaped mainly by observations from more easily sampled shallow aquatic areas. However, the U.S. Geological Survey (USGS) UMESC has developed unique capabilities to sample fishes of deep river channels by quantitative trawling and is producing entirely new knowledge about these fishery resources (Gutreuter et al. 1999; Dettmers et al. 2001a, b).
The main channels of large river-floodplain systems typically have been considered less productive and less hospitable to many organisms than floodplain backwaters. They have been described mainly as highways used by fishes to move among other suitable habitats (Junk et al. 1989), although it is recognized that some fishes do spend at least some of their time in the main channel (Nielsen et al. 1986; Fremling et al. 1989). In earlier work, we used trawling to identify spatial and temporal patterns in the abundance of adult fishes in the main channels of the Upper Mississippi and Illinois Rivers (Dettmers et al. 2001a) and identified persistent residents of deep channels including shovelnose sturgeon (Scaphirhynchus platorynchus) and blue catfish (Ictalurus furcatus). Further, Galat and Zweinmüller (2001) found that the majority of threatened or endangered riverine fish species in large North American and European rivers require access to channels for at least one life stage. Clearly, the deep-channel complexes are a poorly known but critical component of habitat for many fishes of large rivers.
Our goal is to gain a better understanding of the factors that affect production of fishes in the deep channels of large rivers. Our research in FY 2001 focuses on elucidation of chronic effects of commercial navigation traffic on the abundance of fishes. Chronic denial of habitat by disturbance will be measured as any difference in abundance between matched paired segments of navigation channel and large secondary channel that is proportional to the volume of navigation traffic. In outlying years, we also propose to (a) measure the energetic cost of disturbance using electromyogram tags; (b) estimate trawl efficiency to correct abundance estimates for avoidance; (c) model spatial and temporal patterns in abundance and biomass as functions of channel morphology, bed slope, and other factors that may explain major modes of variation; (d) estimate growth and total annual mortality of key fishes to enable estimation of biological production; and (e) model responses of biological production to potential changes in river management. This program of research is feasible only because of ongoing collaboration and contribution of USGS-base funds and funds from the U.S. Army Corps of Engineers. All of this work is enhanced by or enhances other ongoing research initiatives including a study of nutrient dynamics, productivity pathways, and landscape patterns in the abundance of unionid mussels.
A total of 116 trawl samples were collected from matched paired segments of navigation and secondary channel in Pools 10, 11, and 26 of the Mississippi River during FY 2001. Catches were dominated by gizzard shad (Dorosoma cepedianum), channel catfish (Ictalurus punctatus), freshwater drum (Aplodinotus grunniens), and smallmouth buffalo (Ictiobus bubalus).
Two letter reports were submitted to the U.S. Army Corps of Engineers. A letter report on Feasibility of Methods for Determining Efficiency of Bottom Trawls in Rivers by Steven J. Zigler identified alternatives for estimation of the fraction of fish collected by the trawl. Estimates of this efficiency are needed to convert catches per area swept to abundance per unit area of bottom. The letter report concluded that a combination of aimed trawl experiments and acoustic sensing would be needed. A letter report on Alternatives for Probability-based Sampling Designs for Systemwide Trawl Surveys by Steve Gutreuter described alternatives for inclusion of bottom trawling in routine monitoring conducted by the LTRMP.
Integrated Analysis of Fish Monitoring Data
Environmental monitoring programs are frequently designed to track changes in key physical, chemical, and biological features of an ecosystem. As such, these programs provide critical information for detecting changes in system state, investigating causal mechanisms of the observed changes, and making resource management decisions. Because monitoring programs require significant investments of time, money, and human resources to implement and maintain, periodic evaluations of monitoring programs are necessary to determine if the sampling design adequately addresses program goals and objectives. Periodic evaluations also permit assessment of a program's ability to provide adequate and useful information for changing management and science needs.
We evaluated the LTRMP sampling design for fisheries by analyzing data from stratified random samples collected in 1993-1999 in six trend analysis areas (TAAs). Specifically, we investigated whether the sampling design could provide similar information with fewer sampling gears. Presently, the fish monitoring design uses 10 gear types. Our goals were to identify and quantify information provided by each gear used to monitor fishes in the LTRMP, develop alternative sampling design scenarios based on our analyses and expert opinion, and engage program partners in a discussion on the relative value of each gear within the present sampling design.
Community composition (presence or absence), community structure (relative abundance within the full community) and detection of annual changes in single-species catch-per-unit-effort (Lubinski et al. 2001) are the key metrics provided from fish monitoring. We performed a wide array of analyses and data summaries among TAAs, gear types, sampling strata, seasonal sampling period, and fish size for these key metrics. Preliminary results were made available to program partners that permitted post hoc simulations of alternative reduced-gear sampling designs. Program partners, using expert opinion and preliminary results, developed a suite of potential options for component refinement. These options were then evaluated for their ability to eliminate duplicative effort while maintaining the Program's ability to characterize the composition and structure of fish communities, and to track single-species trends in abundance at levels of statistical power comparable to the existing design.
Based on expert opinion and preliminary results, an option for developing the best pool-specific combinations of gears was considered most viable by the program partners. Detailed analyses of catch by gear resulted in identical conclusions across TAAs regarding which gears produced the most useful information. This result suggests that some gears are somewhat redundant with one another, probably due to similarities in their selectivity for species and size classes (e.g., comparable net mesh sizes) (Figure 20). A nonparametric ordination method known as Non-Metric Multidimensional Scaling was used to avoid assumptions of normally distributed data. For each gear, total catch for every species collected in the Program's history was calculated and partitioned into two size classes (<= 125 and >125 mm). Catch data were standardized due to unknown differences in selectivity among gears. The ordination is based on Bray-Curtis Similarity among gears for standardized size-based community structure data. Gears are placed in ordination space based on similarities in measured community structure. Three patterns are evident. First, all gears form a rough circle, indicating that the 10 gears presently fished do a good job at characterizing the larger UMRS fish community. Second, there is a strong gradient in size selectivity among the gears (selectivity for small-sized fish on the left, medium-sized fish in the middle, and large-sized fish on the right). Third, several gears are almost entirely redundant, indicated by nearly overlapping symbols on the plot.
Once redundant gear types were identified, we evaluated the potential impacts of removing each gear from the Program on spatial and temporal continuity of the long-term data set. Based on these evaluations, we proposed eliminating four gears—seines, tandem fyke nets, tandem mini fyke nets, and night electrofishing—from the LTRMP sampling design for fishes.
The proposed sampling design has the following characteristics relative to the present sampling protocol:
- Trend analyses within and among TAAs require an uninterrupted time series of data, but such analyses must typically be gear specific due to unknown differences in gear selectivity. Temporal data continuity required for analysis of trends will remain unchanged for the retained gears, within and among TAAs;
- Understanding how fishery characteristics vary among TAAs require consistent spatial sampling. The gears we propose to eliminate are optional and inconsistently fished within and across TAAs. Also, strata unique to the tandem gears are not common among all TAAs. Therefore, systemic spatial data continuity remain unchanged;
- In each TAA, 91-100% of the species observed under the present sampling protocol would still be detected. All of the species that would no longer be detected are rare (i.e., <10 individuals collected from 1993 to 1999);
- The probability of detecting a particular species in any one year will be minimally affected. Less than 12 species per TAA will be detected less frequently under the proposed design than under the present design;
- Of those species for which the present design provided >50% power to detect interannual changes in abundance, 85-100% of species will retain at least 50% power at the stratum scale and 97-100% at the TAA scale;
- The above performance can be achieved with a 3-62% reduction in total catch by TAA. The vast majority of "lost catch" is fish less than 120 mm in total length from eight highly abundant and ubiquitous species. These taxa are collected in sufficient numbers to characterize their placement in the overall community and to detect changes in abundance when sufficient power is present, by gears retained under the proposed sampling design.
- All strata will continue to be sampled except impounded offshore and backwater contiguous offshore. However, these strata do not contribute unique species or power to detect interannual change and are statistically similar in community composition and structure to their nearshore strata counterparts.
Recommendations made from this work were ratified by the LTRMP Analysis Team and the Environmental Management Program Coordinating Committee in fall 2001 and will be implemented in 2002.
Macroinvertebrates are an important food for a variety of fish and waterfowl within the river system (Hoopes 1960; Jude 1968; Ranthum 1969; Thompson 1973). Collectively, the river's macroinvertebrate fauna performs an important ecological function by digesting organic material and recycling nutrients. Macroinvertebrates also can be useful as biological indicators of water and sediment quality (Myslinski and Ginsburg 1977; Rosenberg and Resh 1992). An indicator species can be defined as a species that has particular requirements to a known set of physical or chemical parameters.
Mayflies (Hexagenia spp.), fingernail clams (Sphaeriidae), and midges (Chironomidae), part of the soft-sediment substrate fauna, were chosen as target organisms for the LTRMP because of their important ecological role in the UMRS. Asiatic clams and zebra mussels (Dreissena polymorpha) were chosen for sampling because of their potential adverse effects on the economy and biology of the UMRS (Tucker 1995a,b).
Annual monitoring of macroinvertebrates began in 1992. Samples were collected with standard operating procedures (Thiel and Sauer 1999) using Ponar grab samplers. All mayflies, fingernail clams, midges, Asiatic clams, and zebra mussels were counted and the presence or absence of several other macroinvertebrates (Odonata, Plecoptera, Trichoptera, Diptera, Bivalvia, Oligochaeta, Decapoda, Amphipoda, and Gastropoda) was recorded.
In 2001, samples were collected at both stratified random and fixed-site sampling locations in five LTRMP study areas yielding a total of 617 samples. The Open River Reach was dropped from the macroinvertebrate monitoring design because the previous 9 years of sampling have consistently indicated low densities of most taxa, especially mayflies and fingernail clams. River crests were the second or third highest on record for many areas of the Upper Mississippi River. Fighting high water and strong currents, the LTRMP field stations were able to complete all sampling.
Macroinvertebrate stratified random sampling in 2001 produced a total of 1,610 mayflies, 3,289 fingernail clams, 1,773 midges, 5 Asiatic clams, and 7,838 zebra mussels from 557 total samples. This is a 47% decline in the total number of mayflies and a 32% decline in midges from 2000, yet these numbers are still within the range of numbers seen since sampling began in 1992. The total number of fingernail clams collected in 2001 increased 22% from 2000, whereas zebra mussel numbers increased by 64%. This is the highest total number of zebra mussels collected since sampling of zebra mussels began in 1995.
The poolside estimated mean densities of mayflies decreased between 2000 and 2001 in all study areas (Figure 21). Pools 4 and 8 had the highest estimated mean density of mayflies, whereas in Pool 13 mean densities reached their lowest recorded level at 77 m-2. Changes in densities of fingernail clams and midges were variable among study areas with no consistent patterns (Figures 22 and 23). However, in Pool 13, density of midges dropped precipitously from 574 m-2 in 2000 (the highest level ever recorded by LTRMP) to 28 m-2 in 2001 (Figure 23).
Based on within-pool distributions, we concluded that Lake Pepin (a natural lake in Pool 4) and the impounded areas in Pools 8, 13, and 26 supported the highest densities of mayflies and fingernail clams. The impounded and backwater contiguous aquatic areas have consistently been the most productive areas for mayflies and fingernail clams over the years. Midge densities were highest within backwater contiguous areas in Pools 8, 13, and 26 and La Grange Pool. In Pool 4, the highest density of midges was found in Lake Pepin. In Pool 26, midge densities reached their highest recorded level of 267 m-2 in backwater contiguous areas compared to all other years and aquatic areas. Zebra mussel densities were highest in impounded areas in Pools 4, 8, and 13. Extremely low numbers of zebra mussels were found in Pool 26 (33 individuals collected) and La Grange Pool (1 individual collected). In general, fine-grained substrates, composed largely of silt and clay, supported the highest numbers of mayflies, fingernail clams, and midges. In previous years, zebra mussels were found predominantly on the gravel rock substrates; however, in 2001 it seems that zebra mussels were finding ways to colonize the softer substrates.
Overall, the high spring flood of 2001 did not affect LTRMP macroinvertebrate sampling. The flood's immediate effect on invertebrate abundance is unclear, but high flows with associated sediment scouring and deposition could have contributed to reduced densities of mayflies and midges. However, sampling in 2002 will provide the first opportunity to assess the long-term effects of the flood and of the experimental reduction in water levels conducted in Pool 8 in summer 2001.
Analyses of mayfly abundance data in Pool 13 of the Upper Mississippi River has provided important insights into intra- and interannual variation in abundances, the role of substrate composition, and whether some sampled habitats preclude the presence of mayflies in some years. Interannual variation in mayfly abundances in Pool 13 appears to represent a small proportion (<1%) of the total observed variation within that pool. If management was interested in intraannual variation, then this small proportion would suggest designs with relatively dense sampling in selected years and sparser sampling in intervening years.
In addition, approximately 80% of the interannual variation seen in mayfly abundances in Pool 13 may be explained by changes in sampled substrate composition. These changes may reflect actual substrate changes in the pool or over- or undersampling, by chance, of substrate categories. The former explanation is supported by noting that higher spring maximums tend to coincide with a shift in the average of the categorical substrate predictor towards higher, sandier values. If management is not interested in sediment-related effects on mayflies, then annual averages should be adjusted for substrate effects.
Substrate was also found to be the most important predictor of mayfly abundance within years. This finding led to a collaborative research effort with the Department of Statistics, Iowa State University, to generate maps of substrate categories and associated predictive errors. These maps would use 10 years of accumulated substrate information from the macroinvertebrate and other components. The predicted substrate surfaces are expected to lead to more precise mayfly abundance predictions.
Last, mayfly abundances may be a function of both presence or absence and count processes. The former describes whether habitat is suitable for mayflies and, if so, the latter then determines the resulting abundance. Such a "mixture" of processes greatly complicates the modeling process. Regardless, modeling both processes promises inferences into two different and important components of the mayfly population process in the UMRS.
Over-target Projects
Bathymetric Mapping of the UMRS
This is a multiyear project, with the long-term objective being the completion of a systemic GIS coverage of bathymetry for the UMRS. In addition, intermediate products provide data for use in planning and evaluating HREP, as well as data for a variety of ecological modeling projects. In 2001, both systemic mapping and surveys specifically for HREP were completed. As part of the systemic mapping to be conducted through contract, an inventory was updated and GIS coverages of data gaps in selected pools were generated. A contract was developed through the U.S. Army Corps of Engineers Rock Island District for surveys in the selected pools of the UMRS. Additional funds were made available to the St. Louis District for cost-sharing for Middle Mississippi River surveys. Pools were selected for survey based on existing data by targeting pools that were nearly complete, or selecting pools in areas with large data gaps in the system.
Surveys were performed in Pools 5, 10, 11, 17, and 18 on the UMRS. Poolwide coverages were not completed in 2001 for any of these pools because of conditions limiting the survey of problematic areas (e.g., shallow vegetated areas). Pools 5 and 10 are the nearest to completion (>80% complete), with the other pools being from 50% to 80% complete. Surveys were also completed in the Middle Mississippi River, with about 80% of that reach being completed. The data were processed into geographic information system data sets at UMESC. Data and maps for the pools with near poolwide coverage (Pools 5 and 10) are available on the Internet through the LTRMP bathymetry page at the UMESC site (http://www.umesc.usgs.gov/aquatic/bathymetry.html). Other data sets for partial GIS coverages can be requested from UMESC. Updates on project status and viewable maps were made to the bathymetry web pages.
Scoping Plan for the Acquisition
of Precise Elevation Data for the Upper Mississippi River
Since its inception in 1986, the LTRMP has compiled an extensive database of spatially referenced information for the UMRS, including bathymetry, invertebrate, fish, and waterfowl surveys, and land cover/land use (LCU) maps. The acquisition of high-resolution floodplain elevation data would allow resource managers to maximize the utility of this monitoring and research information. High-resolution elevation data are needed in water level management studies to predict the extent of inundation; in reforestation studies to determine favorable sites for mast tree species; in nutrient and species distribution studies to map connectivity of channels with other floodplain water bodies; and in various modeling efforts where an understanding of seasonal water extent is vital.
A new remote sensing technology called Light Detection and Ranging (LIDAR) has emerged in the last decade. The LIDAR uses an airborne laser system to generate georeferenced, terrestrial elevation points. The digital elevation model (DEM) produced from these points can then be combined with bathymetric data into a seamless floodplain map (Figure 24).
As part of a scoping plan for the acquisition of LIDAR for the UMRS, we provided background information on the technology; described current applications; estimated costs for acquisition for the UMRS; and compiled relevant contact information (contract preparation; LIDAR contractors; managers of current projects; and liaisons for identifying potential cost-sharing efforts). We also provided a template "LIDAR Scope of Work" that can be tailored for specific UMRS projects.
Cost for acquiring LIDAR data for the UMRS is approximately $1,000/mi2, depending in part on the size and shape of each study area and the level of processing provided by the contractor. The LIDAR is more cost-effective for larger areas (e.g., a single pool or larger), although a pilot study encompassing a single pool or less is recommended to fine-tune the acquisition methodology.
The Federal Emergency Management Agency recently completed guidelines for flood hazard mapping. Their accuracy standards require that DEMs for floodplains have a vertical Root Mean Square Error of 15 cm or less, which is equivalent to a National Standard for Spatial Data Accuracy accuracy of about 30 cm. (Accuracy terminology is explained more fully in an appendix to the scoping document.) The LIDAR is the most cost-effective method for acquiring this level of detail for large areas.
We recommend working closely with individuals of the USGS National Mapping Program who serve as liaisons for specific states. They are generally able to provide information on current LIDAR mapping efforts in their particular states, as well as help coordinate work with potential partners. Coordination and product standardization will be particularly important for the UMRS, since the river serves as a boundary for many government entities.
Development of a Georeferenced Database on Unionid Mussels in the Upper Mississippi River
Biologists know little about the natural and anthropogenic factors that control distribution and abundance of unionid mussels, or the spatial scales at which these factors may operate. Malacological Consultants, a private consulting firm in La Crosse, Wisconsin, has unique data on mussels in the Upper Mississippi River derived at different spatial scales ranging from broad-scale surveys to detailed, fine-scale descriptions within mussel beds. We purchased some of these data, entered them into a GIS framework, and produced a georeferenced map that documents the locations and species composition of mussel beds within the studied reach. The mussel database contains 52 variables related to the sampling effort including georeferenced coordinates, discharge and elevation at the time of sampling, live and dead density of mussels, species richness, relative abundance, and conservation status. A summary of these data are shown in Table 2; Figure 25 shows the location and sampling methods used to obtain these data.
Data from Malacological Consultants have been combined with additional mussel data into a GIS-based framework that is being used in additional analyses to assess whether the spatial distribution of mussels can be predicted from a variety of features in the surrounding landscape and selected biotic and abiotic variables that are not typically measured in biological assessments (i.e., shear stress, sediment stability, distance from tributary mouths). The combined database can also be used to identify metrics that will be useful in determining an effective inventory and monitoring approach at various spatial scales. This framework should be broadly applicable throughout the UMRS and will provide valuable input to decisions regarding mussel management and LTRMP planning. This work leverages current work at the UMESC on physical and chemical predictor variables and compliments U.S. Army Corps of Engineers' work on an Upper Mississippi River mussel database.
This study produced a variety of digital and paper products that can be used by LTRMP managers in their decision making process. The products that were delivered to the U.S. Army Corps of Engineers include a georeferenced searchable database on CD-ROM media, GIS-based maps that document the spatial distribution of mussel beds in the selected reach, technical support for all data and applications developed during this project, technical support and technology for demonstrations of the ArcView project, and archival of digital products.
Implementation of Open River Field
Station Experimental Trawl (Missouri Trawl)
Trawling is used to collect fishes for commercial and scientific purposes
in coastal marine systems (e.g., Matsushita and Shida 2001), reservoirs
(e.g., Michaletz et al. 1995), and rivers (e.g., Dettmers et al. 2001a).
Trawl configurations, mesh size and shape, and environmental factors may
influence catch (Glass and Wardle 1989; Kunjipalu et al. 1992; Chopin
and Arimoto 1994; Kim and Wardle 1997; Godo and Walsh 1998; Dahm 2000;
Ryer and Olla 1999; Matsushita and Shida 2001). Many factors affect the
species and size classes that enter the cod end of the trawl, such as
sea-state and towing speed (Lowry et al. 1998). Some researchers have
noted that the mesh size and shape of a trawl allows escapement (Dremiere
et al. 1999; Polet 2000), resulting in high catch variability in the cod
end. Trawl selection varies greatly among researchers, and different approaches
are used to maximize target organism catch (Van Den Avyle et al. 1995;
Pine 2000).
In 1997, Open River Field Station biologists developed an experimental trawl (Missouri trawl). The Missouri trawl is a modified LTRMP trawl that incorporates a dual-mesh design. Fish that pass through a larger inner mesh are caught in a small outer mesh panel (i.e., a pass-through technique).
The objectives of this study were to (1) implement the Missouri trawl in the upper portion of the Middle Mississippi River, Upper Mississippi River Pool 26, and the lower Illinois River to assess applicability to other systems; (2) compare and contrast Missouri trawl, LTRMP trawl, and 10.2-m four-seam bottom trawl samples from Pool 26 and the lower Illinois River; (3) summarize results and observations to make recommendations for future use; and (4) publish a description of the Missouri trawl detailing its efficacy in large rivers.
The Missouri trawl was constructed using a two-seam 19.05-mm bar mesh slingshot balloon trawl (LTRMP trawl; Gutreuter et al. 1995) completely covered with a 4.76-mm diameter delta heavy mesh (i.e., outer trawl). Both trawl bodies used a 3.18-mm mesh size cod end to retain catch.
The 10.2-m four-seam bottom trawl was used to sample fish in Upper Mississippi River Pool 26 and the lower Illinois River. This trawl was constructed with 2.54-cm bar mesh. The cod end was approximately 2.4 m long and located 10.7 m from the wings.
Data were analyzed from river reaches where the LTRMP and Missouri trawls were used and included Upper Mississippi River Navigation Pools 4, 13, and 26 and Open River Reach. Fish were identified to species, measured to the nearest millimeter, and enumerated.
Species richness per trawl haul was compared between the LTRMP and Missouri trawl. We used the default stepwise multiple regression procedure (SAS Institute 1988) to determine the effects of reach and gear on species richness. Independent variables that were significant to the 0.15 level entered the regression model and included trawl type, reach, and the interactions between reach and trawl type. The final variable represents an interaction between trawl type and reach.
We estimated the rate of species caught by randomly selecting 100 observations in the Open River Reach data set. Each bar code was assigned a random number then we sorted the data set in ascending order. The first bar code listed was "sample" 1. We continued "sampling" bar codes until 100 observations were reached. A logarithmic trend line was used to plot each sample for both trawl types.
Data collected from Pool 26 using the 10.2-m four-seam, LTRMP, and Missouri trawls were compared using the minimum, maximum, and median lengths of coincidentally caught fish species from the three trawl types.
Fifty-eight species were caught in 331 trawl hauls using the Missouri trawl and 46 species were caught in 528 hauls using the LTRMP trawl (Table 3). Approximately 26% of the variation in species richness was explained by three independent variables that entered the stepwise multiple regression model (F = 181.42; df = 3; P < 0.0001; Table 4). Species richness per sample was significantly lowered when the LTRMP trawl was used in Pools 4, 13, and 26 and Open River Reach. The positive interaction term between Pool 4 and the Missouri trawl suggests the Missouri trawl was especially effective in this reach. The negative interaction term between Pool 4 and the LTRMP trawl suggests the LTRMP trawl does not perform as well in this pool when compared to other reaches.
In the Open River Reach, the Missouri trawl caught more species than the LTRMP trawl at any sampling interval using the randomized "sampling" method. In 100 samples, the LTRMP trawl caught 25 species and the Missouri trawl caught 40 species (Figure 26).
Thirty fish species were caught using the 10.2-m four-seam trawl in Pool 26 (Table 5). Of the 30 species, only five were caught coincidentally in the three trawl types. However, only four species were caught in sufficient numbers to compare length ranges: blue catfish, channel catfish, white bass (Morone chrysops), and freshwater drum. There were differences in length ranges among caught species (Figure 27) in the three trawl types. The 10.2-m trawl caught larger individuals. The widest length range of blue catfish was obtained using the Missouri trawl in the Open River Reach.
Trawling is a feasible method for sampling fish communities in the Upper Mississippi River. The advantages of the Missouri trawl include low equipment cost, simple operation, and improved fish richness and abundance information. The LTRMP trawl catch is variable and does not detect as many species as the Missouri trawl. The 10.2-m trawl appears to collect large fish. The Missouri trawl may be used in tandem with the 10.2-m trawl to obtain improved estimates of fish species richness and relative abundance in the navigation channel and channel borders of the Mississippi River. The Missouri trawl seems to be particularly useful to sample small, rare fishes using deep, swift channel border habitats; thus, this gear would be a valuable addition to a gear array in detecting long-term trends of ecologically significant species.
Reconnaissance and Remapping of the Wing Dams (Dikes) in the Open River Reach
Wing dikes are the most numerous of the humanmade structures in the unimpounded Mississippi River, and there seems to be scattered information at best about wing dike locations and dimensions in the Middle Mississippi River. There have been many changes to the river in the last several years (i.e., new dikes and modifications to existing dikes). The flood of 1993 and subsequent large floods have changed the bank line of the river in many areas. Scour holes behind some wing dikes were filled by sedimentation while others increased in size. There are no recent GIS coverages of the Middle Mississippi River (but see D'Erchia 1993) that includes updated information about wing dike locations and dimensions. All of these factors make it difficult to accurately locate sampling sites on the river.
To better understand the efficacy of the existing monitoring protocol in the Open River reach, an analysis of water quality data and the use of wing dikes by fish communities was conducted. This information may be used to develop new monitoring protocols in the Open River reach. Part of the analyses included the remapping and reconnaissance of wing dikes between Upper Mississippi River river miles (RM) 30-80. In previous years, only 5-10 dikes were surveyed annually.
Location and depth data were collected using a Garmin GPSMap 168 Sounder with a GBR 21 beacon receiver. The accuracy of this unit averaged approximately 1.6 m and two sounder features were used to collect data. The first feature was the waypoint function. Each time the function was invoked, the unit recorded the time, date, depth, and Universal Transverse Mercator (UTMs) coordinates. The second feature was the tracklog function, which was useful for mapping the wing dikes. We set the user-defined interval to record data every 2 sec (i.e., time, date, depth, and UTMs). After the dike and the scour hole were mapped, the tracklog was saved on a laptop computer. The sounder's memory had a maximum capacity of 500 waypoints and 10 tracklogs.
Mapping the scour hole and taking the physical dimensions of the wing dike were completed in one process. Typically, the nose of the boat was placed on the tip of the dike. A waypoint was entered at that position and the tracklog memory was cleared. The total length of the dike was measured with a Bushnell Yardage Pro 1000 range finder, which had an accuracy of +/- 1 m at 914 m. The length was taken from the tip of the dike to where the dike was keyed into the top of the bank. The operator then motored to the landward end of the dike to place a waypoint where the bank line and the dike met. If the dike was notched, the operator would maneuver the boat onto the bank and a waypoint was recorded at each side of the notch. These points provided the physical dimensions of each dike. Transects (parallel with the wing dike) were then run at regular intervals across the scour hole until the scour and associated area were mapped. Waypoints were entered at regular intervals while running these transects. One transect followed the contour of the bank line features associated with the scour hole (i.e., sand bars, islands, and mudflats). The crew leader used their judgment on the number of transects and waypoints mapping the wing dikes and scours. Smaller dikes often required fewer transects and waypoints. After mapping, a waypoint was placed on the upstream side of the wing dike where the bank line and the dike met (Figure 28).
From August 13 to 24, 2001, we identified and surveyed 211 wing dikes between RM 30 to 80. Straight dikes comprised 93.8% of all dikes, while the remaining L-shaped dikes made up the other 6.2% (Table 6). The wing dikes were classified as one of six types based on features (i.e., straight, L-shaped, notched, L-shaped with a notch, pile dikes, and straight dikes with trailing dike). Straight dikes were typically straight or slightly bent or curved. Straight dikes were the most numerous (67.7%) in the study reach. Trail dikes were defined as dikes that had a leg extending downstream anywhere from the bank line out to the middle of the dike. Trail dikes had the greatest mean depth (6.78 m) of all the dike types. L-shaped dikes were constructed similar to straight dikes, but had with a leg extending downstream from the tip of the dike. L-shaped dikes had the second deepest mean depth (5.75 m); however, there were only 10 (4.7% of all dikes) from RM 30 to 80. Notched dikes comprised 22.5% of all dikes in Open River reach. The notches ranged from 5 to 94 m. There were only three (1.4% of all dikes) L-shaped dikes with a notch in the Open River reach. The maximum depth behind this type of dike was 20.6 m. Pile dikes had the shallowest mean depth (2.03 m) and the shallowest maximum depth (15.8 m) of all the dike types.
Data collected were used to create an ArcView overlay showing the waypoint and tracklog points that were collected. These data were then queried to visualize areas with depths greater than 6.1, 9.1, 12.2, and 15.2 m. Each query was converted into a shape file and added to the ArcView coverage to produce a reasonably accurate map of the wing dikes and depths of their scour holes. Each depth range was identified using a specific color (Figure 29). One hundred sixty-six dikes had water greater than 6.1 m, 129 dikes had water greater than 9.1 m, 63 dikes had water greater than 12.2 m, and 32 dikes had water greater than 15.2 m deep. These data can transformed into a bathymetric coverage. Figure 30A shows an example at wing dike 32.2, where we have caught pallid sturgeon. The original data of more than 100 waypoints were plotted using scatter plots (Figure 30B). The points were then interpolated along each UTM transect to fill in the missing area (Figure 30C). Finally, the plots were rotated and tilted to help visualize the bottom topography of the river (Figure 30D).
The method we developed to obtain wing dike dimension and scour elevation
data is quick and inexpensive to collect. In applications where an extensive
bathymetry survey is not needed, this method provides a reasonably accurate
image of wing dike dimensions and scour hole bathymetry. The images that
may be created from these data can provide general depths behind the scour
holes, but can also be transformed to display a topographic image. The
depth maps created from this survey have been used in several other projects
for locating sampling areas (e.g., pallid sturgeon and overwintering surveys).
The number of waypoints that could be taken limited the resolution of
the data. A bathymetry survey needs to be done on all the wing dike scour
holes to provide an accurate and detailed image of the bed topography.
This can be used to track changes bank in elevation and topography between
dikes, which may provide important habitat use information by rare species.
This summary provides a cursory view of the major highlights from the Long Term Resource Monitoring Program in 2001. For more detailed information, we highly recommend the reader review the individual annual reports for 2001 (fisheries, vegetation, and macroinvertebrates).
Page Last Modified: April 17, 2018