SIGNIFICANCE OF RIVERINE HYPOXIA FOR FISH: THE CASE OF … Sunflower Jawra.pdf · SIGNIFICANCE OF...

SIGNIFICANCE OF RIVERINE HYPOXIA FOR FISH: THE CASEOF THE BIG SUNFLOWER RIVER, MISSISSIPPI1

F. Douglas Shields, Jr. and Scott S. Knight2

ABSTRACT: Degraded streams draining low-relief, intensively cultivated watersheds may experience periods ofhypoxia or anoxia. A three-year study of water chemistry, fish, and physical habitat in the Big Sunflower Riverin northwestern Mississippi coupled with continuously logged physicochemical and hydrology data provided byothers showed prolonged periods of hypoxia associated with higher flows. Fish species richness was directlyrelated to dissolved oxygen (DO) concentration (r2 = 0.35, p = 0.00004), and ordination using nonmetric multidi-mensional scaling (NMS) indicated strong association between fish community structure and DO. Low-headweirs supported relatively dense and diverse fish communities and thus provided local habitat enhancement,but may create stagnant zones upstream due to backwater effects that exacerbate low DO problems. Althoughhypoxia has been reported for some lightly degraded rivers and floodplains, our observations suggest hypoxia inBig Sunflower River and similar systems alters fish species composition and should be remediated. Cost-effectiveremediation will require better understanding of autotrophic and heterotrophic processes that control DO andthe relationship of these processes to discharge.

(KEY TERMS: rivers; fish; habitat; hypoxia; water quality; weirs; dissolved oxygen; biodiversity.)

Shields, F. Douglas, Jr. and Scott S. Knight, 2011. Significance of Riverine Hypoxia for Fish: The Case ofthe Big Sunflower River, Mississippi. Journal of the American Water Resources Association (JAWRA) 48(1):170–186. DOI: 10.1111 ⁄ j.1752-1688.2011.00606.x

INTRODUCTION

Restoration of river ecosystems presents many vex-ing problems (Gore and Shields, 1995). Among themost difficult are those associated with degradedstreams draining low-relief, intensively cultivatedfloodplains. Such streams are usually channelized forflood control and drainage and offer little stablegravel or stone substrate for aquatic organisms.Water physicochemistry is often characterized by ele-vated sediment loads, turbidities, and nutrient and

pesticide concentrations (Cooper et al., 1982; Killgoreet al., 2008; Stephens et al., 2008). Perhaps mostimportant, long (>10 km) reaches may experienceperiods of hypoxia or anoxia. Hypoxia, or the condi-tion of aquatic systems where dissolved oxygen (DO)concentrations are below either 2 mg ⁄ l or 30% of sat-uration, is widely recognized as detrimental to mostforms of aquatic biota and many ecosystem services.Except for very brief (hours) exposure, anoxia (totalDO depletion) is fatal for most aquatic biota, and DOdepletion is widely recognized as a key indicator ofaquatic ecosystem degradation (Paerl et al., 1998).

1Paper No. JAWRA-10-0164-P of the Journal of the American Water Resources Association (JAWRA). Received September 30, 2010;accepted August 08, 2011. ª 2011 American Water Resources Association. This article is a U.S. Government work and is in the publicdomain in the USA. Discussions are open until six months from print publication.

2Respectively, Research Hydraulic Engineer and Research Ecologist, USDA-ARS National Sedimentation Laboratory, PO Box 1157,Oxford, Mississippi 38655 (E-Mail ⁄ Sheilds: [email protected]).

JAWRA 170 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

Vol. 48, No. 1 AMERICAN WATER RESOURCES ASSOCIATION February 2012

Popular and scientific literature teem with accountsand information on marine (e.g., Rabouille et al.,2008; Turner et al., 2008; Zhang et al., 2009), estua-rine (e.g., Paerl et al., 1998; Kemp et al., 2005), andlacustrine (review in Garvey et al., 2007) hypoxia,but hypoxia in nonestuarine river waters is less com-monly mentioned (Garvey et al., 2007; Diaz and Bre-itburg, 2009). Standard texts in environmental andwater resources engineering usually contain descrip-tions of ‘‘DO sag curves’’ that develop in rivers down-stream from point sources of biochemical oxygendemand (BOD) such as wastewater treatment works(e.g., Viessman and Hammer, 1993; Davis and Corn-well, 1998), but discussions of DO depletion absentsuch point sources are not commonly presented. How-ever, streams that are 303(d) listed for DO depletionare often impacted by diffuse nonpoint source pollu-tion, so the significance of such pollution is wellknown (Cooper, 1993).

The DO level in any river reach is the result of abalance between sources and sinks of DO. Sourcesinclude reaeration and photosynthesis whereas sinksare BOD from substances in either the water columnor in bed sediments including biological aerobic respi-ration. BOD is associated with allochthonous organicmatter (OM) inputs from point or nonpointwatershed sources, or due to autochonous primaryproduction on the stream bed or within the water col-umn. Reported riverine DO problems arise fromdiverse environments, so a variety of source ⁄ sinkconditions are likely involved. For example, acuteevents producing fish kills on regional scales havebeen reported for low-gradient streams in Mississippi(Killgore et al., 2008; Pennington, n.d.) and Louisiana(Ice and Sugden, 2003) and in the Neuse estuary,North Carolina (Paerl et al., 1998), following the

passage of hurricanes. In contrast to these acute,extraordinary events, chronically low DO values havebeen reported in many intensively cultivated water-sheds (Table 1).

Hypoxia in streams of the Mississippi River flood-plain have been attributed to a combination of natu-ral and anthropogenic factors, including pollution,low-stream reaeration rates due to low channel gradi-ents, abundant riparian vegetation and swamps thatcontribute organic material, and high temperatures(Holt and Harp, 1993; Kleiss et al., 2000). Significantquantities of dissolved organic carbon may be leachedfrom crop residues (Schreiber, 1985; Schreiber andMcDowell, 1985), and oxygen demand by forest littermay be great enough to deplete oxygen in floodwaterscovering forested floodplains (Whittemore, 1982).

Typically, state standards require long-term aver-age DO levels >5.0 mg ⁄ l and instantaneous levels>3.0 or 4.0 mg ⁄ l for the protection of aquatic life.These standards are based on early work by fisheriesbiologists (e.g., Moore, 1942; Doudoroff and Shum-way, 1970). More recent reviews emphasize the varia-tion of critical DO levels for sublethal effects amongfish species (Richards et al., 2009). Biotic response tohypoxic (but not anoxic) conditions can be complex,with invertebrate communities exhibiting wideranges of sensitivity (Connolly et al., 2004; Kallerand Kelso, 2007). Fish responses tend to be morestraightforward, with communities in streams drain-ing highly developed landscapes often exhibiting sig-nificant, positive associations with DO (Miltner andRankin, 1998; Killgore and Hoover, 2001; Smileyet al., 2009). However, hypoxic conditions have beenreported for lightly developed ‘‘reference’’ watershedsin the Southeastern Coastal Plain (Ice and Sugden,2003; Kaller and Kelso, 2007; Mason et al., 2007) and

TABLE 1. Reported Examples of Riverine Hypoxia in Streams Draining Intensively Cultivated, Low-Relief Landscapes.

Location DatesNumberof Sites

ReportedDO (mg ⁄ l) Reference

Northwestern Mississippi 1971-1972 20 0.4 to 17.4 Parker and Robinson (1972)Bear Creek system, northwesternMississippi

Summers in 1977-1979 2 1.3 to 19.0 Cooper et al. (1982)

Mississippi, Arkansas 2002 50 1.5 to 16 Rebich et al. (2004)Northwestern Mississippi May-June 2006 52 1.6 to 13.6;

Median = 5.5Bryson et al. (2007)

Eastern Arkansas and northwesternMississippi

Summer 2001 17 0.3 to 14.6 Stephens et al. (2008)

Northern and central Illinois Summers 2005 and 2006 17 0 to 20 Garvey et al. (2007)Cedar Creek watershed, Indiana andUpper Big Walnut Creek watershed,Ohio

2005-2007 21 0.3 to 26.3 Smiley et al. (2009)

Northwestern Mississippi Continuous 48-hcollections duringSeptember or October2007

28 17 Sites hadminimum DO <4

Hicks and Stocks (2010)

SIGNIFICANCE OF RIVERINE HYPOXIA FOR FISH: THE CASE OF THE BIG SUNFLOWER RIVER, MISSISSIPPI

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 171 JAWRA

in blackwater rivers in southern Georgia (Vellidiset al., 2003; Todd et al., 2007, 2009). As such rela-tively unimpacted systems exhibit hypoxia, and asmany of the fish species native to these systems exhi-bit adaptations to low DO (Hoover and Killgore,1998), some observers argue that current water-qual-ity criteria for DO levels are either unnatural orunrealistic (Ice and Sugden, 2003; Kaller and Kelso,2007; Mason et al., 2007; Justus and Wallace, 2009;Todd et al., 2009). Is environmental hypoxia an indi-cator of ecological impairment in low-gradient rivers?Below, we examine associations among DO, fishes,and physical habitat conditions in the Big SunflowerRiver, Mississippi, a chronically hypoxic stream thatdrains part of the Mississippi River Alluvial Plainecoregion. Our goal is to explore the relationshipsamong riverine hypoxia, streamflow, and fish commu-nities in order to contribute to the discussion regard-ing the ecological significance of hypoxia.

THE BIG SUNFLOWER RIVER

The Big Sunflower River originates near MoonLake in Coahoma County in northwestern Mississippiand flows south through intensively cultivated landsfor about 200 km to enter the Yazoo River about70 km upstream from its confluence with the Missis-sippi River at Vicksburg. The river drains about8,000 km2 and is not regulated by large reservoirs,although at least two weirs produce significant back-water effects at base flow (Figure 1). One of theweirs, Lock and Dam 1, was constructed at river kilo-meter (RKM) 88 in the 1900s to allow commercialnavigation, but has been abandoned for several dec-ades and is gradually deteriorating as it is no longermaintained. The other weir, Holly Bluff Cutoff weir(RKM 30.7), was constructed to maintain minimumwater depths in a bypass channel to prevent terres-trial plant invasion (Miller, 1995). Depletion of lowflows in the river due to groundwater extraction foragriculture has been a major issue since ca. 1970(Pennington, 1993). During dry periods within thegrowing season, flows in the river are comprised ofirrigation returns, wastewater discharges, andaugmentation from groundwater (Pennington, 1993;Ervin and Tietjen, 2008). Watershed land use isabout 78% agriculture, 12% forest, and 4% wetland(MDEQ, 2003).

At the time of this study, the reach downstreamfrom Lock and Dam 1 had nearly trapezoidal cross-sections, minimal woody riparian vegetation, andinstream large wood and no aquatic macrophytes(Figure 1). Substrates were sand and silt; gravel was

absent. Banks were steep, and point bars were absentin the study reaches. Gradients and thus velocitieswere extremely low (<0.5 m ⁄ s). Temperature and tur-bidity tended to be high (summer temperatures>25�C and winter turbidity >200 nephelometric tur-bidity units (NTU)). Prior to channelization com-pleted in the 1960s, the river provided a viablerecreational fishery and was used by local residentsfor boating and fishing. Wide, sandy point bars werean important feature of the system (J. Oliver, 2009,personal communication). The 1960s channelizationproject included 76.5 km of channel clearing, 34.6 kmof channel cleanout, 69.4 km of channel enlargement,channel cutoffs aggregating 25.6 km in length, andconstruction of a weir at the downstream end of HollyBluff Cutoff to control low water levels (Miller, 1995).However, at the time of this study (2007-2009), studyreach channel width, channel position, and sinuositywere similar to that depicted on air photographstaken in 1956-1957 and 1962.

Despite the size of the river (Q1.5 yr = 273 m3 ⁄ s)(MDEQ, 2003), relatively little scientific literature isavailable regarding its water physicochemistry orbiota. Justus (2003) developed an index of ecologicalintegrity for streams in this region and rated sites atBig Sunflower RKM 173.2 and 63.4 as most degraded(34th of 36 sites) and moderately degraded (29th of36 sites), respectively. Major water-quality issues arerelated to agricultural land use; significant levels ofhistoric and current-use pesticides have been foundin water and fish tissue (Kleiss and Justus, 1997;Pennington, 1997; Zimmerman et al., 2000). Turbid-ity and suspended sediment concentrations tend to behigh (>100 NTU and >100 mg ⁄ l), especially duringhigher flows (Martin et al., 2003; Rebich et al., 2004).Chronically low DO levels (<2 mg ⁄ l) were reportedfor reaches downstream from RKM 230 during an18-month period ending December 2007 using in situwater physicochemistry loggers at 13 sites betweenRKM 136 and 336 (Ervin and Tietjen, 2008). Thishypoxia was attributed to ‘‘diffuse inputs [of oxygen-demanding substances] from agricultural activities inthe region.’’

Despite water physicochemical degradation, avail-able data indicate that the Big Sunflower River sup-ports a diverse assemblage of fishes, with 57 speciescaptured in 114 collections (Killgore et al., 2001).Available evidence suggests that fish populationsincreased and became more speciose between 1972and 1992 due to reductions in DDT and municipalwastewater loads (Yarborough et al., 1994; Lucas andPennington, 2001). George et al. (1995) examinedpaddlefish (Polyodon spathula) in a 60-km reach ofthe Big Sunflower immediately upstream of our studyarea and found a stable or increasing populationin good condition relative to other major rivers.

SHIELDS AND KNIGHT


Shephard and Jackson (2005) found smaller size andyounger age at sexual maturity for channel catfish(Ictalurus punctatus) in the Big Sunflower River com-pared with other similar-sized Mississippi rivers. Fas-ter growth rates (Shephard and Jackson, 2006) andhigher hoopnet catch per unit of effort (Shephard andJackson, 2009) for this species in the Big Sunflowerwere also reported. These studies related catfish suc-cess to higher soil fertility in the Big Sunflowerwatershed and suggested that the high turbidity ofthe Big Sunflower may have favored ictalurids oversight-feeding species and also provided cover fromsome predators. Miller and Payne (2004) describedseveral densely populated native mussel beds inreaches downstream of Lock and Dam 1, and Caskey

et al. (2002) reported 46 invertebrate taxa collectedfrom RKM 63.4 in 1997.

METHODS

Two segments of the Big Sunflower River wereselected for study: a 6-km long reach immediatelydownstream from Lock and Dam 1 (RKM 82 to 88)and an 8-km reach of the old river bypassed by theHolly Bluff Cutoff, a 10.6-km canal that bypasses a22.5-km river section and empties into the main riverchannel at RKM 31.3. The airline distance between

BS1

BS3

BS2

BS4

HB4HB1

HB3

HB2

BS1

BS2, 3, 4

HB1

HB2, 3, 4

FIGURE 1. Study Site Locations Along the Big Sunflower River in Northern Mississippi. BS1 was immediatelydownstream from abandoned Lock and Dam 1, whereas HB1 was immediately downstream from Holly Bluff Cutoff weir.

Each sampling reach was 200 m long. Reaches BS1 and HB1 were classified as ‘‘below weir,’’ reaches BS2, BS3, andBS4 were classified as ‘‘channelized,’’ and reaches HB2, HB3, and HB4 were classified as ‘‘old river.’’



the two segments was about 35 km and the river dis-tance was about 50 km. A low weir in the lower endof the Cutoff channel forces all flow through the oldriver section at extreme low stage, and allows someflow in both the cutoff and the old river section athigher stages. In addition to the old river, the secondstudy segment included the reach immediately down-stream from the Holly Bluff Cutoff weir (Figure 1).Within each segment, one 200-m long sampling reachwas established immediately downstream from theweir structure along with three, 200-m long reachesmore distant from the structure at �2 km intervals.Each reach was sampled biannually (May-June andNovember) for three years (2007-2009).

Measurements

Physical habitat conditions were assessed usinga boat-mounted acoustic Doppler current profiler(ADCP). On some occasions, the ADCP data weresupplemented with echosounder readings coupledwith GPS. In the second (middle) year of the study(2008), bed samples were collected from one or twopoints within each reach using a petite Ponar dredgeand returned to the laboratory for grain-size analysisusing standard procedures. Data from the ADCPwere correlated with bottom materials using theapproach described by Shields (2010). Subreach-meanvalues of water depth and velocity were used to com-pute the reaeration coefficient using an empiricalformula (Bennett and Rathbun, 1972; Aristegi et al.,2009).

Continuous measurements of stage and basic waterphysicochemistry were available from a gage at RKM63.4 (USGS 07288700). Additional stage data wereavailable from gages at RKM 100 and 30 (http://www.rivergages.com, accessed September 16, 2011).Discharge measurements were available from anupstream location (RKM 173.2, USGS 0728500). Inaddition to these continuous measurements, DO,temperature, pH, and conductivity were measuredat the surface using handheld meters while sam-pling fish and habitat conditions. At the same time,surface water grab samples were collected, stored onice, and then returned to the laboratory and ana-lyzed for total and dissolved solids, nitrate, nitrite,total Kjeldahl nitrogen, total phosphorus, and ortho-phosphate. Sampling occurred between 10:00 and16:40 h Central Time.

Fish-sampling methods generally conformed bystandard guidelines (Reynolds, 1983). Fish were col-lected within each 200-m sampling reach using aboat-mounted electroshocker to apply pulsed DC cur-rent at voltages adjusted to provide maximum catchefficiency depending upon the specific conductance of

the water at a given site. Sites were sampled forapproximately 30 min and in such a manner as tocover the major habitats within each reach. Shockingtime was recorded to the nearest second so that catchper unit of effort could be calculated. Stunned fishwere collected using dip nets (42 cm · 42 cm opening,45 cm deep, 75 cm mesh). Fish were identified byspecies and measured for length. Weights wereestimated based on length-weight relationships forfreshwater fishes in Mississippi and Alabama (Swin-gle, 1972). All fish that could be identified on sitewere released after measurements were taken. Thoseindividuals that could not be readily identified in thefield were labeled and preserved in 10% formalinsolution and returned to the National SedimentationLaboratory for identification. Number of species,numerical catch per unit of effort, and biomass catchper unit of effort were calculated for each collection.

Analysis

Box and whisker plots were prepared usingmonthly statistics of mean daily discharge (gageUSGS 0728500) and once-daily continuously mea-sured water physicochemistry variables (USGS07288700) for the three years of the study (2007-2009). The grab sample water chemistry data and thenumber of individuals, biomass, and number of spe-cies of fish per collection were subjected to two-wayANOVA with season and reach classification (belowweir, channelized, old river) as independent vari-ables. As collections may have violated the assump-tions of ANOVA by not being truly independentsamples due to their streamwise proximity, we alsoran ANOVA using location (water distance in kmfrom Lock and Dam 1) as a factor instead of reachclassification. Simple Spearman rank-order correla-tions were also computed between each of the waterand fish variables and location to assess the amountof variance explained by location along the river cor-ridor rather than season or reach type. For purposesof ordination analysis, a matrix containing fish spe-cies abundances for each collection was created. NMSand ancillary graphical and correlation analyses(McCune and Grace, 2002; Peck, 2010) were runusing PC-ORD 4.41 software (McCune and Mefford,1999) to assess species composition patterns. NMSanalysis was limited to collections containing morethan three fish (35 collections of the 41 total) and the25 most common species (species found in three ormore collections). Due to extreme variation in theabundances of some species between collections, allentries in the 35 collections · 25 species matrix werelog10(x + 1)-transformed prior to ordination. Sorensondistances were constructed with NMS from the

SHIELDS AND KNIGHT


35 · 25 data matrix with a random starting configu-ration and 50 runs with real data using the autopilot‘‘slow and thorough’’ mode in PC-ORD. The strengthof patterns revealed by NMS ordination was assessedwith Monte Carlo tests of the probability that a simi-lar final stress could be obtained by chance. TheMonte Carlo test randomized the data by randomlyreallocating elements within columns of the 35 · 25data matrix and repeatedly running the NMS. Thenthe stress obtained from NMS of real (unshuffled)data (McCune and Grace, 2002) was compared withthe Monte Carlo outputs. To pass the Monte Carlotest, the final stress had to be lower than that for95% of the randomized runs. After ordination, axeswere rotated to maximize correlation between the DOconcentration measured manually concurrent withfish collection and Axis 1. In order to detect responseof the fish community to physical and chemical gradi-ents, a second matrix was constructed giving meansfor habitat variables measured concurrently witheach fish collection and sample site location (asdefined above). We examined graphical overlays andbiplots of the physical habitat data onto the influen-tial ordination axes and computed Pearson correla-tions between habitat variables and NMS Axis 1 and2 scores to examine relationships between fish speciescomposition and habitat.

RESULTS

Our data show wide variations in streamflow andwater physicochemistry, but more modest variationin other physical habitat attributes. Streamflow andwater physicochemistry were dominated by seasonalvariations (Figure 2). Streamflows were highest inMay and lowest in November, coinciding with oursampling trips. Stages measured at RKM 63.4 ondates when sampling occurred varied by more than7 m, and sampling date discharges measured at theupstream gage (RKM 173.2) varied from <1 to�150 m3 ⁄ s. Summer (June-August) flows were con-sistently low, and January-May flows were consis-tently high (Figure 2). Monthly median dischargeswere <17 m3 ⁄ s except for May and December, whichhad median flows ‡55 m3 ⁄ s. Highest levels of con-ductance, pH, and temperature coincided with sum-mer periods of low turbidity and low streamflow.Turbidity levels were consistently lower in June-November than in Winter and Spring, with monthlymeans ranging from several hundred NTU in Winterto 28 NTU in July. Location (streamwise distancefrom upstream end of study reach) explained <15%of the variation in any water physicochemistry

variable and <3% of the variation in descriptors offish collections.

Seasonal DO patterns were more complex than forthe other continuously monitored parameters(Figure 2). For a given month, DO was positively cor-related with discharge for flows >�30 m3 ⁄ s (RKM173.2), but for a given discharge, cooler monthstended to have DO levels nearly twice as high as war-mer months. Monthly mean DO concentrations fellbelow the state standard of 5.0 mg ⁄ l (MDEQ, 2003)for April-May and July-September, reaching a mini-mum in the month of September (2.0 mg ⁄ l, med-ian = 1.6 mg ⁄ l, SD = 1.5 mg ⁄ l). DO levels were higherin June, July, and August than in the higher flowmonths of May and September. For the years 2007-2009, mean DO was <5.0 mg ⁄ l for the months of Aprilthrough November, inclusive.

Consistent with the continuous data for May-Juneand November, grab sample water chemistry on dayswhen Spring samples were collected featured lowertemperatures, pH values, and DO concentrations buthigher turbidity (Tables 2 and 3). Mean grab samplewater temperatures on Fall sampling dates were 11degrees cooler than for Spring; temperature differ-ences between reach types were slight but significantwith higher temperature at the unshaded channeli-zed sites (Table 3). Conductivity did not vary withreach or season, but mean DO for Spring samplingdates was only 2.9 mg ⁄ l, which was less than half ofthe Fall mean. DO difference based on reach typewere significant, with highest levels at the channeli-zed sites (Tables 2 and 3). DO was weakly correlated(r = 0.45, p = 0.003) with computed reaeration coeffi-cients, and both DO and computed reaeration coeffi-cient tended to be lower for higher discharges(Figure 3). DO was more tightly related to meandepth and temperature (r < )0.53, p < 0.0003 forboth) than reaeration. Spring pH (6.7) was almost afull unit lower than for Fall (7.6). Highest levels oftotal solids occurred in Spring in the channelizedreaches downstream from Lock and Dam 1, and low-est levels were found in Spring in the old riverreaches. Median Spring turbidity levels in theupstream (BS1-BS4) sites were 110-150 NTUwhereas those downstream (HB1-HB4) were only44-90 NTU. Turbidity tended to be considerably lowerin Fall (mean: 61 NTU) than Spring (mean: 175NTU) and also much lower in the old river reaches(mean: 61 NTU) than for the others (mean: �150NTU). Interactions between season and reach typewere significant with respect to turbidity and totalsolids concentrations (Table 3). In summary, ourstudy reaches tended to be warm, turbid, enrichedwaters. Using the index proposed by Bryson et al.(2007) based on total Kjeldahl nitrogen concentration,turbidity, and DO, all samples were rated as



eutrophic or mesotrophic with total productivityscores ranging from 6 to 9 with a mean of 7.2 (meso-trophic = 4-6, eutrophic = 7-9).

Physical habitat conditions tended to be uniform,with most flow velocities <0.2 m ⁄ s and most depths>3.0 m, producing computed reaeration coefficientsthat averaged 4.6 ⁄ day and ranged from 0.7 to18.8 ⁄ day. Greatest diversity in habitat conditionsoccurred downstream from Lock and Dam 1(Figure 4), where rapid flow over the stone from thecrumbling structure poured into a deep (�10 m) scourhole, and eddies formed over lateral bars along thechannel margins that were covered with musselshells. The reach downstream from Holly Bluff Cutoffweir was more uniform than the reach below Lockand Dam 1 as the weir was lower relative to normalstages and did not pass flow at low stage. In addition,

the weir was comprised of sheet piling with a concretecap and was much shorter in the streamwise direc-tion than the crumbling Lock and Dam 1 structureand its stilling basin. Nevertheless, both structuresprovided a wider range of flow conditions than thechannelized or old river reaches (Figure 4). Bed mate-rial sizes exhibited levels of diversity that mirroredhydraulic conditions (Table 4). Equipment failure pre-vented fish sample collection in four of the eightstudy reaches (BS2 and BS3 and HB2 and HB3,Figure 1) in Fall 2008. Three samples contained nofish (BS4 in Spring 2009 and BS3 and BS4 in Fall2009), so instead of the planned 48 collections, therewere 41 electrofishing samples. These samples con-tained a total of 1,836 fish representing 35 species.Gizzard and threadfin shad (Dorosoma cepedianumand Dorosoma petenense) comprised 54 and 11%,

FIGURE 2. Box and Whisker Plots for Big Sunflower River Once-Daily Physicochemical Data Measured at 08:00 h Local Timeat RKM 63.4 and Discharges Measured at RKM 173.2 in 2007-2009. Horizontal lines bisecting boxes represent median values.

Boxes enclosed by rectangles are based on 13 to 44 daily values; all others are based on 45 to 93 daily values.

SHIELDS AND KNIGHT


respectively, of the numerical catch whereas small-mouth and bigmouth buffalo (Ictiobus bubalus andIctiobus cyprinellus) comprised 39 and 25%,respectively, of the fish biomass. Only 1 of the 35captured species, Cyprinella camura, was ranked asintolerant of water-quality degradation by Jesteret al. (1992). Twenty-eight of the fish species werefound below Lock and Dam 1, and 21 were foundbelow Holly Bluff Cutoff weir. The three channelizedreaches downstream from Lock and Dam 1 produceda total of 25 species whereas the three old riverreaches produced 27 species. Fall collections con-tained an average of almost twice as many speciesper collection than Spring (8.4 compared with 4.9),and the median number of species per collectionbelow weirs was 7, more than twice as great as themedians for the old river and channelized reaches(both = 3). Differences in fish species richness andbiomass catch per unit of effort by season werestatistically significant (Tables 5 and 6), but location

was not a significant factor and explained <3% of thevariation in fish species richness and catch per unitof effort. Two-way ANOVA using location and seasonas factors instead of reach type and season yieldedsimilar findings for the significance of differences inmeans for different locations (p > 0.49). Fish pre-ferred the physically diverse conditions just belowLock and Dam 1. Median biomass per unit effort forthis location (31.5 kg ⁄ h) was more than twice asgreat as for any other reach, and this site produced26 species on just one sampling date (November2007). Catches and catch per unit of effort recordedfor this site may underestimate biomass and abun-dance, as many fish escaped capture when electro-shocking produced stunned fish faster than theycould be netted. It should also be noted that severalof the fish at this site were too large to capture withthe dip nets used (42 cm · 42 cm opening, 45 cmdeep, .75 cm mesh), resulting in slight under esti-mates of the catch.

TABLE 2. Means (standard error of the mean) of Grab-Sampled Water-Quality Variables for the Big Sunflower River, Mississippi.

Variable

Reach Type

Below Weir Channelized Old River

Spring Fall Spring Fall Spring Fall

Temperature, �C 26.5 (0.5) 16.5 (0.5) 28.4 (0.5) 16.8 (0.5) 27.1 (0.5) 15.8 (0.5)Conductivity (lS ⁄ cm) 177 (30) 183 (30) 155 (30) 162 (28) 139 (28) 167 (28)pH 6.6 (0.2) 7.6 (0.2) 6.5 (0.2) 7.7 (0.2) 6.8 (0.2) 7.4 (0.2)DO (mg ⁄ l) 2.6 (0.7) 7.8 (0.7) 3.6 (0.7) 8.3 (0.7) 2.5 (0.6) 6.0 (0.7)Turbidity (NTU) 234 (26) 49 (26) 251 (24) 53 (24) 43 (24) 80 (24)Total solids (mg ⁄ l) 272 (27) 194 (27) 281 (25) 176 (25) 126 (25) 212 (25)Total Kjeldahl nitrogen (mg ⁄ l) 1.4 (0.3) 1.6 (0.3) 1.7 (0.3) 2.0 (0.3) 1.6 (0.3) 1.6 (0.3)Total phosphorus (mg ⁄ l) 1.14 (0.13) 0.76 (0.13) 1.11 (0.120 0.66 (0.12) 1.30 (0.12) 1.03 (0.12)Chlorophyll a (mg ⁄ l) 17.2 (5.7) 21.1 (4.7) 30.9 (4.7) 22.9 (4.3) 21.6 (4.7) 17.1 (4.3)

Note: Data and samples were collected concurrently with fish and physical habitat data.

TABLE 3. F (p) Values Resulting From Two-Way Analyses of Variance of Grab-Sample Water Quality for Big Sunflower River,Mississippi on Dates When Fish and Aquatic Habitat Samples Were Collected.

Variable

Source of Variation

Season (d.f. = 1)Reach Type

(d.f. = 2)Season 3 Reach Type

(d.f. = 2)

Temperature 748.6 (<0.001) 3.53 (0.040) 1.32 (0.280)Conductivity 0.342 (0.563) 0.461 (0.634) 0.0896 (0.915)pH 50.64 (<0.001) 0.00787 (0.992) 1.718 (0.195)DO 65.6 (<0.001) 22.2 (0.041) 0.90 (0.415)Turbidity* 23.93 (<0.001) 11.52 (<0.001) 27.35 (<0.001)Total solids 2.353 (0.134) 3.95 (0.029) 8.417 (0.001)Total Kjeldahl nitrogen 0.480 (0.493) 0.803 (0.456) 0.119 (0.888)Total phosphorus 12.64 (0.001) 2.851 (0.072) 0.268 (0.766)Chlorophyll a 0.540 (0.468) 1.808 (0.181) 0.752 (0.480)

Notes: Seasons were Spring or Fall, and reach types (see Figure 1) were below weir (BS1 and HB1), channelized (BS2, BS3, and BS4), or oldriver (HB2, HB3, and HB4). Bold values indicate p < 0.05.

*Log10-transformed.



Fish collections were made across a wide gradientof DO levels and fish numbers; species richness andspecies composition were tightly coupled to DO. Atleast 17 of the 41 fish collections occurred when con-currently measured DO levels at 0.2 m depth were<4.0 mg ⁄ l, and at least three collections occurredwhen DO <1.0 mg ⁄ l. Thus, some fish were able tosurvive oxygen concentrations that are generally con-sidered lethal. However, many of the captured indi-viduals were severely stressed as indicated byvasodilation of blood vessels just below the skin sur-face along the sides and belly. The number of fishspecies in a given collection was significantly corre-lated with the measured DO (r2 = 0.35, p = 0.00004).Furthermore, species capture frequency (number ofindividuals of a species in a given collection dividedby the total catch of that species across all collections)displayed strikingly different patterns for variousspecies. For example, 100% of the Hypophthalmich-thys molitrex captures and 87% of the D. cepedianumcaptures occurred when DO <5 mg ⁄ l, whereas a nearlyopposite trend was exhibited by the Centrarchid,

Lepomis macrochirus, with no captures occurring forDO <6.7 mg ⁄ l (Figure 5). Other species, for example,Lepisosteus platostomus and I. bubalus, exhibitedintermediate sensitivity to DO.

Biological gradients among the sample units weredetected by the NMS ordinations. NMS ordinationusing the 25 most common species produced threeaxes that together explained 80% of the variation inthe sample space with a final stress of 15.18 and afinal instability of 10)7 at 146 iterations. These axeswere rotated to maximize the correlation betweenAxis 1 and DO. NMS Axis 1 was significantly posi-tively correlated with conductivity, DO, and pH, andinversely with temperature, discharge, and meanwater depth (Table 7). Axis 2 was inversely propor-tional to water velocity whereas Axis 3 was weaklyproportional to discharge (Table 7). A plot of fish col-lections in the NMS plane showing DO levels bybroad categories (Figure 6a) indicates sites withhigher DO levels shifted to the right on Axis 1 andsomewhat lower on Axis 2. Abundances of manyfishes were related to the NMS axes (Figure 6b and

FIGURE 3. Scatter Plots Showing Relationships Among River Discharge, Dissolved Oxygen Concentration, andComputed Reaeration Coefficients for Big Sunflower River.

SHIELDS AND KNIGHT


Table 8). Only the highly tolerant Gambusia affinisand the invasive exotic H. molitrex plotted to theextreme left on Axis 1 (Figure 6b). The game fishes,

I. punctatus, L. macrochirus, Micropterus salmoides,Morone chrysops, and Pomoxis annularis; the loticspecies, Aplodinotus grunniens, Carpiodes carpio,

Beelow Lock and

C

Dam 1

Channelized

Below HolCutoff w

Old River

Bed elevatioscale, m

Velocity vectoscale, m/s

lly Bluffeir

n m

ors s

FIGURE 4. Bed Elevations and Current Velocities for Reaches Below Weirs, Channelized Reaches,and Old River Reaches of the Big Sunflower River, May 2007.

TABLE 4. Bed Material Conditions in Sampled Subreaches of Big Sunflower River, 2008.

Subreach(Figure 1) Bed Material Observations

Median Bed Material Size,D50 (mm)

BS1 Large (�1 m) broken concrete rubble, sand bar coveredwith mussel shells, hard clay

No sample

BS2 Soupy mixture of fines 0.005BS3 Sand 0.407BS4 Soft mud 0.008HB1 Large riprap near weir, hard clay elsewhere Sample contained only a few

crumbs of hard clayHB2 Repeated casts of Ponar dredge yielded no sample No sampleHB3 Soft mixture of sand and fines 0.008HB4 Repeated casts of Ponar dredge yielded no sample No sample



I. bubalus, and I. cyprinellus; the forage species,D. cepedianum, Menidia beryllina, and Notropis athe-rinoides; and the tolerants, Cyprinella venusta, Cypri-nus carpio, Lepisosteus oculatus, and Pimephalesvigilax were positively correlated with Axis 1, reflect-ing DO influences (r > 0.334, p < 0.05). The gamespecies I. punctatus, I. bubalus, Lepomis megalotis,

and P. annularis and the predators Amia calva andL. platostomus were positively associated with Axis 2(r ‡ 0.34, p < 0.05) (Figure 6b and Table 8).

DISCUSSION

The study area of the Big Sunflower River exhib-ited hypoxia during cooler, higher flows in the Spring,although simulation modeling by Martin et al. (2003)suggested that low DO events would coincide withlower flows (Summer and Fall). Even when not hyp-oxic, DO levels were depressed (means <5.0 mg ⁄ l) forthe months of April through November, inclusive.This finding is consistent with earlier observationspublished by other workers (Killgore et al., 2001,2008).

TABLE 5. Means (standard error of the mean) of 41 Fish Collections From the Big Sunflower River, Mississippi.

Variable

Reach Type

Below Weir Channelized Old River

Spring Fall Spring Fall Spring Fall

Number of fish species per sample 6 (2) 11 (2) 3 (2) 8 (2) 6 (2) 7 (2)Number of fish per unit of effort (no. ⁄ h) 27 (77) 133 (77) 15 (63) 43 (71) 223 (63) 50 (71)Biomass catch per unit of effort (kg ⁄ h) 16 (8) 32.3 (8) 6 (6) 18 (7) 15 (6) 23 (7)

TABLE 6. F(p) Values Resulting From Two-Way Analyses of Variance of Key Descriptors of 41 Fish Collections for Big Sunflower River.

Variable

Source of Variation

Season (d.f. = 1) Reach Type (d.f. = 2) Season 3 Reach Type (d.f. = 2)

No. species 5.199 (0.028) 1.286 (0.288) 0.991 (0.381)Numerical catch per unit of effort 0.030 (0.863) 1.181 (0.318) 2.17 (0.128)Biomass catch per unit of effort 4.776 (0.035) 1.457 (0.246) 0.208 (0.813)

Notes: Seasons were Spring or Fall whereas reach types (see Figure 1) were below weir (BS1 and HB1), channelized (BS2, BS3, and BS4), orold river (HB2, HB3, and HB4). Bold values indicate p < 0.05.

TABLE 7. Pearson Product-Moment CorrelationBetween NMS Axes and Habitat Variables Measured

Concurrently With Electrofishing.

Habitat Variable Axis 1 Axis 2

Water temperature )0.435 0.042Conductivity 0.443 0.100Dissolved oxygen concentration 0.691 )0.017pH 0.452 0.021Discharge )0.472 )0.204Mean water velocity )0.274 )0.435Mean water depth )0.385 0.005

Note: Bold values indicate p < 0.05.

Dissolved oxygen concentration, mg/L

0 2 4 6 8 10 12

Cum

ulat

ive

perc

enta

ge o

f cap

ture

s

0

20

40

60

80

100

Hypophthalmichthys molitrex Dorosoma cepedianum Ictiobus bubalus Lepisosteus platostomus Lepomis macrochirus

FIGURE 5. Cumulative Number of Captures vs. Dissolved OxygenConcentration (DO) Measured Concurrently With Electrofishing.For example, 100% of the H. molitrex captures occurred when DO<5 mg ⁄ l whereas only 7% of the L. macrochirus captures occurredat DO <6.7 mg ⁄ l.

SHIELDS AND KNIGHT


Martin et al. (2003) found that DO depletion in thereaches below RKM 100 were partially caused bybackwater conditions upstream from two weirs – theHolly Bluff Cutoff weir (RKM 31) and Lock and Dam 1(RKM 88). Impoundment by these weirs decreases rea-eration rates and increases retention time, allowinggreater impact of sediment oxygen demand. Killgoreet al. (K.J. Killgore, J.J. Hoover, C.E. Murphy, S.G.George, and D.R. Johnson, U.S. Army Corps of Engi-neers Engineer Research and Development Center,Vicksburg, Mississippi, 2011, unpublished data) notedthat low DO generally occurs in Delta streams at lowflows when streams become stagnant. Evidently, the

observed hypoxia apparently reflects a complexresponse to deep, slow flow, organic loads from diffusesources upstream (e.g., forest litter and crop residues)and perhaps mortality of algal biomass that developsin the shallower, nutrient-rich upstream reaches(Martin et al., 2003). In a study of the riverine swampsand associated channels of the Atchafalaya River insouthern Louisiana, Sabo et al. (1999) observedchronic hypoxia (18.2% of 1,445 measurements) inlow-energy channels where circulation patternswere weak, and flows became stagnant. Although pre-impact data were absent, they hypothesized thatlarge-scale anthropogenic changes in the system flow

velocity

velocity

Temperature, discharge, depth DO, conductance, pH

FIGURE 6. NMS Ordination of Fish Collections, Big Sunflower River Showing Plotting Positions for (a) Each CollectionWith DO Concentration Measured Concurrent With Fish Collection, and (b) Each Species.



patterns were the cause of these stagnation zones. Insimilar fashion, water physicochemistry and fish datafor the Big Sunflower River prior to weir constructionare lacking, but backwater effects and greater depthsupstream from these structures undoubtedly reducereaeration and may trap carbonaceous materials fromupstream (Martin et al., 2003). These effects may bemost severe during prolonged high flows when BODinputs and flow depths are greater. High turbidity lev-els and higher velocities associated with high flowsmay limit primary production. Data published by oth-ers as well as our own indicate an inverse relationshipbetween DO and turbidity (Figure 7a). Highest chloro-phyll a values and thus primary production levelsoccur at intermediate levels (20-100 NTU) of turbidity(Figure 7b), and fall as turbidity levels increase aboveabout 200 NTU. A study of three oxbow lakes in thisregion showed chlorophyll concentrations were inver-sely proportional to phosphorus when suspended sedi-ment concentrations exceeded 150 mg ⁄ l (equivalent toabout 160 NTU for Big Sunflower River data in thisstudy) (Knight et al., 2008).

Data presented above indicate that the fish commu-nity of the Big Sunflower River is somehow adapted tospatially and temporally extensive hypoxia. Fairlydiverse communities of admittedly tolerant fish specieshave been found in streams with very low DO (Ste-phens et al., 2008), and behavioral adaptations to

hypoxia (gulping air or breathing surface film andanaerobic metabolism) have been reported (Gee et al.,1978; Hoover and Killgore, 1998). Both short-term andevolutionary changes have been observed in speciesexposed to chronic hypoxia. Air breathing, piping, andbuccal and opercular pumping are generally acceptedevolutionary adaptations to niches associated with lowDO concentrations. Several authors have reportedincreased gill surface area in species that are exposedto chronically hypoxic waters (Chapman and Chap-man, 1998; Chapman et al., 2002; Sollid et al., 2003).In some species, individuals found in low oxygen condi-tions have longer gill filaments than the same speciesin more oxygen-rich environments (Ebeling and Weed,1963; Kobayshie, 1973). These changes are consideredadaptive, short term, and reversible, but have not beenreported for fishes in temperate lowland rivers.

Despite such adaptation, it would not be correct toassume that the fish and other components of the BigSunflower River ecosystem are unimpaired by waterphysicochemistry degradation. Indeed, the sensitivityof many fish species to low DO is manifest in theirdisappearance from our collections during hypoxicperiods. Furthermore, the physiological stress weobserved, although sublethal, may ultimately contrib-ute to mortality via disease (Wedemeyer et al., 1976)or adversely impact populations in other ways(Garvey et al., 2007). Evidently, sensitive fish popula-tions recover following hypoxic episodes in our studyreaches by re-emerging from local refugia or recolon-ization from shallower, more normoxic environmentsupstream or downstream. A regional study of fish inlarge, unregulated rivers such as the Big Sunflowershowed that catch per unit of effort (with seines)was positively correlated (linear quantile regression)with DO, but interestingly, species richnesswas inversely proportional to DO (K.J. Killgore, J.J.Hoover, C.E. Murphy, S.G. George, and D.R. John-son, U.S. Army Corps of Engineers EngineerResearch and Development Center, Vicksburg, Mis-sissippi, 2011, unpublished data).

Effective approaches for remediation of hypoxia inthe Big Sunflower River and similar low-gradient sys-tems are not obvious. Instream aeration is prohibi-tively expensive, and installation of weirs and otherhabitat structures (Miller, 1995; Miller et al., 1995)may produce local improvement at the expense ofreaches upstream and downstream. Our ANOVA andNMS analyses did not detect differences between fishcollections from subreaches immediately downstreamfrom weirs and those further away, in contrast withwork by others on warmwater rivers impounded bylow head dams (Gillette et al., 2005). Ecological resto-ration of this and similar systems will require betterunderstanding of biotic communities that controlinstream photosynthesis and respiration and the

TABLE 8. Pearson Product-Moment Correlation Between the 25Most Abundant Fish Species Numbers and NMS Axes.

Fish Species Axis 1 Axis 2

Amia calva 0.228 0.343Aplodinotus grunniens 0.688 0.248Carpiodes carpio 0.556 0.272Cyprinella venusta 0.393 )0.076Cyprinus carpio 0.605 0.300Dorosoma cepedianum 0.356 )0.201Dorosoma petenense 0.281 )0.265Gambusia affinis )0.096 0.060Hypophthalmichthys molitrex )0.044 0.115Ictalurus punctatus 0.416 0.401Ictiobus bubalus 0.392 0.427Ictiobus cyprinellus 0.534 )0.143Lepisosteus oculatus 0.577 0.046Lepisosteus osseus 0.299 )0.150Lepisosteus platostomus 0.041 0.555Lepomis humilis 0.282 0.128Lepomis macrochirus 0.626 0.311Lepomis megalotis 0.316 0.484Menidia beryllina 0.386 )0.009Micropterus salmoides 0.476 0.266Morone chrysops 0.429 0.106Notropis atherinoides 0.423 0.010Pimephales vigilax 0.456 0.132Pomoxis annularis 0.571 0.335Pylodictis olivaris 0.114 0.240

Note: Bold values indicate p < 0.05.

SHIELDS AND KNIGHT


levels of turbulence, light penetration, and nutrientsthat control them (Acuna et al., 2011). As theextremely low gradient of these systems limits thepotential for increasing reaeration rates, controls oninputs of nutrients and oxygen-demanding substancesmay be required (e.g., Sullivan et al., 2010). Further-more, remediation efforts must consider water quan-tity as well as water-quality issues, which willrequire tradeoffs among competing interests. At anyrate, removal of existing weir structures and low-flowaugmentation to maintain depths in the absence ofweirs should be seriously considered to remediate

upstream reaches adversely impacted by near-stag-nant backwater conditions.

SUMMARY AND CONCLUSIONS

Some degraded streams draining low-relief, inten-sively cultivated floodplains exhibit chronic environ-mental hypoxia that violates water-quality standards.These conditions have received little attention rela-

(a)

(b)

FIGURE 7. Turbidity vs. (a) DO and (b) Chlorophyll a for Two Extensive Studies of Mississippi Delta StreamsCompared With Similar Data From This Study.



tive to coastal and marine hypoxia and riverine pointsource impacts. A case study of two stretches of theBig Sunflower River documented chronic depressedDO associated with higher flows when depths weregreater. Fish species richness was correlated with DO(r2 = 0.35, p = 0.00004), and ordination analysisshowed that the community structure was definitelyrelated to DO or associated factors. Although hypoxiahas been reported for some lightly degraded riversand floodplains, our observations suggest hypoxia inBig Sunflower River and similar systems is not natu-ral and should be remediated. Cost-effective remedia-tion will require better understanding of microbialand algal systems that control DO.

ACKNOWLEDGMENTS

Assistance with field data collection was provided by John Mas-sey, Calvin Vick, Terry Welch, Duane Shaw, Jimmy Oliver, andMichael Ursic. Justin Murdock, James Martin, Michael Schwar,and Billy Justus reviewed an earlier version of this article andmade many helpful comments.

LITERATURE CITED

Acuna, V., V.C. Vilches, and A. Giorgi, 2011. As Productive and Slowas a Stream Can Be – The Metabolism on a Pampean Stream.Journal of the North American Benthological Society 30(1):71-83.

Aristegi, L., I. Oihana, and A. Elosegi, 2009. Comparison of SeveralMethods to Calculate Reaeration in Streams, and Their Effectson Estimation of Metabolism. Hydrobiologia 635:113-124, doi:10.1007/s10750-009-9904-8.

Bennett, J.P. and R.E. Rathbun, 1972. Reaeration in Open ChannelFlow. U.S. Geological Survey Professional Paper 737, Washing-ton D.C.

Bryson, J.R., R.H. Coupe, and M.A. Manning, 2007. Characteriza-tion of Water Quality in Unmonitored Streams in the Missis-sippi Alluvial Plain, Northwestern Mississippi, May-June 2006.U.S. Geological Survey Scientific Investigations Report 2007-5116, Reston, Virginia.

Caskey, B.J., B.G. Justus, and H. Zappia, 2002. Occurrence ofInvertebrates at 38 Stream Sites in the Mississippi EmbaymentStudy Unit, 1996-99. U.S. Geological Survey Open-File Report02-190, Pearl, Mississippi.

Chapman, C.A., F.G. Nordlie, and A.W. Seifert, 2002. RespiratoryOxygen Consumption Among Groups of PseudocrenilabrusMulticolor Victoriae Subjected to Different Oxygen Concentra-tions During Development. Journal of Fish Biology 61:242-251.

Chapman, L.J. and C.A. Chapman, 1998. Hypoxia Tolerance of theMormyrid Petrocephalus Catostoma: Implications for Persis-tence in Swamp Refugia. Copeia 1998:762-768.

Connolly, N.M., M.R. Crossland, and R.G. Pearson, 2004. Effect ofLow Dissolved Oxygen on Survival, Emergence and Drift inTropical Stream Macroinvertebrate Communities. Journal ofthe North American Benthological Society 23(2):251-270.

Cooper, C.M., 1993. Biological Effects of Agriculturally DerivedSurface Water Pollutants on Aquatic Systems – A Review. Jour-nal of Environmental Quality 22:402-408.

Cooper, C.M., L.A. Knight, and J. Herring, 1982. Fish Compositionin a Sediment-Laden Mississippi Delta Stream System. Journalof the Mississippi Academy of Sciences 27:163-175.

Davis, M.L. and D.A. Cornwell, 1998. Introduction to Environmen-tal Engineering (Third Edition). McGraw-Hill, Boston.

Diaz, R.J. and D.L. Breitburg, 2009. The Hypoxic Environment. In:Hypoxia, J.G. Richards, A.P. Farrell, and C.J. Brauner (Editors).Academic Press, London, United Kingdom, pp. 1-25.

Doudoroff, P. and D.L. Shumway, 1970. Dissolved Oxygen Require-ments of Freshwater Fishes. FAO Fisheries Technical Paper No.86, Food and Agriculture Organization of the United Nations,Rome, Italy.

Ebeling, A.L. and W.H. Weed, III, 1963. Melamphaidae III. System-atics and Distribution of the Species in the Bathypelagic FishGenus Scopelogadus Vaillant. DANA Report, 60, Copenhagen,Denmark, pp. 1-58.

Ervin, G. and T. Tietjen, 2008. Assessing the Effectiveness ofStreamflow Augmentation in the Sunflower River to MaintainWater Quality and Wetland Integrity. Report as of FY2007 for2006MS44B, Mississippi Water Resources Research Institute,Starkville.

Garvey, J.E., M.R. Whiles, and D. Streicher, 2007. A HierarchicalModel for Oxygen Dynamics in Streams. Canadian Journal ofFisheries and Aquatic Sciences 64:1816-1827.

Gee, J.H., F.T. Ross, and H.J. Smart, 1978. Reactions of SomeGreat Plains Fishes to Progressive Hypoxia. Canadian Journalof Zoology 56:1962-1966.

George, S.G., J.J. Hoover, K.J. Killgore, and W.E. Lancaster, 1995.Biology of Paddlefish in a Mississippi Delta River. Proceedings,Mississippi Water Resources Research Conference. http://www.wrri.msstate.edu/results.asp?Author=george&submitbutton=Search, accessed September 16, 2011.

Gillette, D.P., J.S. Tiemann, D.R. Edds, and M.L. Wildhaber, 2005.Spatiotemporal Patterns of Fish Assemblage Structure in aRiver Impounded by Low-Head Dams. Copeia 3:539-549.

Gore, J.A. and F.D. Shields, Jr., 1995. Can Large Rivers BeRestored? BioScience 45(3):142-152.

Hicks, M.B. and S.J. Stocks, 2010. Water-Quality Data for SelectedStreams in the Mississippi Alluvial Plain Ecoregion, Northwest-ern Mississippi, September-October 2007. U.S. Geological Sur-vey Data Series Report 493, Reston, Virginia, 118 pp.

Holt, A. and G.L. Harp, 1993. Ichthyofauna of the Village CreekSystem. Journal of the Arkansas Academy of Science 47:54-60.http://libinfo.uark.edu/aas/issues/1993v47/v47a12.pdf, accessedSeptember 16, 2011.

Hoover, J.J. and K.J. Killgore, 1998. Fish Communities. In:Southern Forested Wetlands, M.G. Messina and W. H. Conner(Editors). Lewis Publishers, Boca Raton, Florida, pp. 237-260.

Ice, G. and B. Sugden, 2003. Summer Dissolved Oxygen Concentra-tions in Forested Streams of Northern Louisiana. SouthernJournal of Applied Forestry 27(2):92-99.

Jester, D.B., A.A. Echelle, W.J. Matthews, J. Pigg, C.M. Scott, andK.D. Collins, 1992. The Fishes of Oklahoma, Their Gross Habi-tats, and Their Tolerance of Degradation in Water Quality andHabitat. Proceedings of the Oklahoma Academy of Sciences72:7-19.

Justus, B.G., 2003. An Index of Ecological Integrity for the Missis-sippi Alluvial Plain Ecoregion: Index Development and Rela-tions to Selected Landscape Variables. U.S. Geological SurveyWater-Resources Investigations Report 03-4110, Little Rock,Arkansas, 32 pp.

Justus, B.G. and J.E. Wallace, 2009. Fish and Invertebrate Assem-blage Relations to Dissolved Oxygen at 35 Sites in SouthernLouisiana. Abstract in: 39th Annual Mississippi WaterResources Conference, Mississippi Water Resources ResearchInstitute, Starkville, Mississippi. http://www.wrri.msstate.edu/

SHIELDS AND KNIGHT


results.asp?Author=justus&submitbutton=Search, accessed Sep-tember 16, 2011.

Kaller, M.D. and W.E. Kelso, 2007. Association of Macroinverte-brate Assemblages with Dissolved Oxygen Concentration andWood Surface Area in Selected Subtropical Streams of theSoutheastern USA. Aquatic Ecology 41(1):95-110.

Kemp, W.M., W.R. Boynton, J.E. Adolf, D.F. Boesch, W.C. Boicourt,G. Brush, J.C. Cornwell, T.R. Fisher, P.M. Glibert, J.D. Hagy,L.W. Harding, E.D. Houde, D.G. Kimmel, W.D. Miller, R.I.E.Newell, M.R. Roman, E.M. Smith, and J.C. Stevenson, 2005.Eutrophication of Chesapeake Bay: Historical Trends and Eco-logical Interactions. Marine Ecology Progress Series 303:1-29.

Killgore, K.J. and J.J. Hoover, 2001. Effects of Hypoxia on FishAssemblages in a Vegetated Waterbody. Journal of AquaticPlant Management 39:40-44.

Killgore, K.J., J.J. Hoover, and S.G. George, 2001. Effects of LowFlows on the Fish Assemblages in the Big Sunflower River, Mis-sissippi. Abstract in: 27th Annual Meeting of the MississippiChapter American Fisheries Society, Jackson, Mississippi.http://mississippiafs.org/pdfs/2001MSAFSProgram(27th).pdf,accessed December 20, 2009.

Killgore, K.J., J.J. Hoover, C.E. Murphy, K.D. Parrish, D.R. John-son, and K.F. Myers, 2008. Restoration of Delta Streams: ACase History and Conceptual Model, EMRRP-ER-08. U.S. ArmyEngineer Research and Development Center, Vicksburg,Mississippi.

Kleiss, B.A., R.H. Coupe, G.J. Gonthier, and B.G. Justus, 2000.Water Quality in the Mississippi Embayment, Mississippi, Loui-siana, Arkansas, Missouri, Tennessee, and Kentucky, 1995-98.U.S. Geological Survey Circular 1208, Reston, Virginia.

Kleiss, B.A. and B.G. Justus, 1997. Preliminary Results of Fish Tis-sue Analysis in the Mississippi Embayment NAWQA Program.Abstract in Mississippi Water Resources Research Conference.http://www.wrri.msstate.edu/pdf/Kleiss97.pdf, accessed Decem-ber 31, 2009.

Knight, S.S., R.F. Cullum, C.M. Cooper, and R.E. Lizotte, Jr., 2008.Effects of Suspended Sediments on the Chlorophyll – Phospho-rus Relationship in Oxbow Lakes. International Journal of Ecol-ogy and Environmental Sciences 34(1):1-6.

Kobayshie, B.N., 1973. Systematics, Zoogeography, and Aspects ofthe Biology of the Bathypelagic Fish Genus Cycothone in thePacific Ocean. Ph.D. Thesis, University of California San Diego,Scripps Institute of Oceanography, San Diego, California.

Lucas, G. and D.A. Pennington, 2001. The 1992 Fish PopulationSurvey of the Upper Sunflower River. 27th Annual Meeting ofthe Mississippi Chapter American Fisheries Society, Jackson,Mississippi. http://mississippiafs.org/pdfs/2001MSAFSProgram(27th).pdf, accessed December 20, 2009.

Martin, J.L., W.L. Kingery, V.J. Alarcon, and W.H. McAnally,2003. Modeling the Big Sunflower River for TMDL DevelopmentPhase 1. Mississippi Department of Environmental Quality Sur-face Water Division TMDL Program, Mississippi State, Missis-sippi.

Mason, A., Y.J. Xu, P. Saksa, A. Vioska, J.M. Grace, J. Beebe, andR. Stich, 2007. Streamflow and Nutrient Dependence of Temper-ature Effects on Dissolved Oxygen in Low-Order ForestStreams. In: Proceedings of the 10-14 March 2007 Fourth Con-ference Forest Service Personnel, A. McFarland and A. Saleh(Editors). American Society of Agricultural and Biological Engi-neers, St. Joseph, Michigan, pp. 374-380.

McCune, B. and J.B. Grace, 2002. Analysis of Ecological Communi-ties. MjM Software Design, Gleneden Beach, Oregon. http://www.pcord.com.

McCune, B. and M.J. Mefford, 1999. PC-ORD. Multivariate Analy-sis of Ecological Data, Version 4. MjM Software Design, Glene-den Beach, Oregon.

MDEQ (Mississippi Department of Environmental Quality), 2003.TMDL for Organic Enrichment, Nutrients and Sediment for theBig Sunflower River Yazoo River Basin Sunflower, Coahoma,Washington, Humphreys, and Sharkey Counties, Mississippi.MDEQ Office of Pollution Control Water Quality AssessmentTMDL ⁄ WLA Section and Mississippi State University CivilEngineering Department, Jackson, Mississippi, http://www.deq.state.ms.us/Mdeq.nsf/pdf/TWB_BigSunDOandSedimentJun03/$File/YazooRBBigSunflowerDOSedimentJun03.pdf?OpenElement,accessed January 8, 2010.

Miller, A.C., 1995. A Habitat Improvement Plan for the Big SunflowerRiver, Mississippi. Technical Report EL 95-29, U.S. Army Engi-neer Waterways Experiment Station, Vicksburg, Mississippi.

Miller, A.C., J. Harris, and K. Quackenbush, 1995. A HabitatImprovement Plan for the Big Sunflower River, Mississippi. In:Proceedings of the First International Conference on WaterResources Engineering, W.H. Espey, Jr. and P.G. Combs(Editors). American Society of Civil Engineers, New York, pp.726-730.

Miller, A.C. and B.S. Payne, 2004. Reducing Risks of MaintenanceDredging on Freshwater Mussels (Unionidae) in the Big Sun-flower River, Mississippi. Journal of Environmental Manage-ment 73:147-154.

Miltner, R.J. and E.T. Rankin, 1998. Primary Nutrients and theBiotic Integrity of Rivers and Streams. Freshwater Biology40:145-158.

Moore, W.G., 1942. Field Studies on the Oxygen Requirements ofCertain Fresh-Water Fishes. Ecology 23(3):319-329.

Paerl, H.W., J.L. Pinckney, J.M. Fear, and B.L. Peierls, 1998. Eco-system Responses to Internal and Watershed Organic MatterLoading: Consequences for Hypoxia in the Eutrophying NeuseRiver Estuary, North Carolina, USA. Marine Ecology ProgressSeries 166:17-25.

Parker, W. and D. Robinson, 1972. Survey of Sources of StreamPollution and Its Effect Upon Bottom Organisms, Fishes andChemical Qualities of the Water. Mississippi Museum of Natu-ral Science Project F-9-14, Jackson, Mississippi.

Peck, J.E., 2010. Multivariate Analysis for Community Ecologists:Step-By-Step Using PC-ORD. MjM Software Design, GlenedenBeach, Oregon.

Pennington, D., n.d. Dissolved Oxygen and Fish Kills in DeltaStreams 2005. Yazoo-Mississippi Delta Joint Water Manage-ment District, Stoneville, Mississippi.

Pennington, D.A., 1993. Augmentation of Low Flows of the UpperSunflower River. Mississippi Water Resources ResearchConference. http://www.wrri.msstate.edu/pdf/Pennington93.pdf,accessed January 4, 2010.

Pennington, K, 1997. An Analysis of Fish and Sediment SamplesFrom Mississippi Delta Rivers. Mississippi Water ResourcesResearch Conference. http://www.wrri.msstate.edu/pdf/Pennington97.pdf, accessed December 31, 2009.

Rabouille, C., D.J. Conley, M.H. Dai, W.-J. Cai, C.T.A. Chen, B. Lan-sard, R. Green, K. Yin, P.J. Harrison, M. Dagg, and B. McKee,2008. Comparison of Hypoxia Among Four River-DominatedOcean Margins: The Changjiang (Yangtze), Mississippi, Pearl,and Rhone Rivers. Continental Shelf Research 28(12):1527-1537,ISSN 0278-4343, doi: 10.1016/j.csr.2008.01.020.

Rebich, R.A., H.L. Welch, and R.H. Coupe, 2004. Evaluation ofWater-Quality and Habitat Assessment Data to DetermineRanges in Stream Conditions in the Mississippi River AlluvialPlain of Northwestern Mississippi and Eastern Arkansas. U.S.Geological Survey Water-Resources Investigations Report 03-4251, Reston, Virginia, 47 pp.

Reynolds, J.B, 1983. Electrofishing. In: Fisheries Techniques, L.A.Neilsen and D.L. Johnson (Editors). The American FisheriesSociety, Bethesda, Maryland, pp. 147-164.



Richards, J.G., A.P. Farrell, and C.J. Brauner, 2009. Hypoxia. FishPhysiology, Volume 27. Academic Press, London, United King-dom, ISBN: 978-0-12-374632-0.

Sabo, M.J., C.F. Bryan, W.E. Kelso, and D.A. Rutherford, 1999.Hydrology and Aquatic Habitat Characteristics of a RiverineSwamp: II. Hydrology and the Occurrence of Chronic Hypoxia.Regulated Rivers: Research & Management 15:525-542.

Schreiber, J.D., 1985. Leaching of Nitrogen, Phosphorus, andOrganic Carbon From Wheat Straw Residues: II. Loading Rate.Journal Environmental Quality 14:256-260.

Schreiber, J.D. and L.L. McDowell, 1985. Leaching of Nitrogen,Phosphorus, and Organic Carbon From Wheat Straw Residues:I. Rainfall Intensity. Journal Environmental Quality 14:251-256.

Shephard, S. and D.C. Jackson, 2005. Channel Catfish Maturationin Mississippi Streams. North American Journal of FisheriesManagement 25:1467-1475.

Shephard, S. and D.C. Jackson, 2006. Difference in Channel Cat-fish Growth Among Mississippi Stream Basins. Transactions ofthe American Fisheries Society 135:1224-1229.

Shephard, S. and D.C. Jackson, 2009. Density-Independent Growthof Floodplain River Channel Catfish Ictalurus Punctatus. Jour-nal of Fish Biology 74(10):2409-2414, doi: 10.1111/j.1095-8649.2009.02239.x.

Shields, F.D., Jr., 2010. Aquatic Habitat Bottom ClassificationUsing ADCP. Journal of Hydraulic Engineering 136(5):336-342.

Smiley, P.C., Jr., R.B. Gillespie, K.W. King, and C. Huang, 2009.Management Implications of the Relationships Between WaterChemistry and Fishes Within Channelized Headwater Streamsin the Midwestern United States. Ecohydrology 2(3):294-302.

Sollid, J., P.D. Angelis, K. Gundersen, and G.E. Nilsson, 2003.Hypoxia Induces Adaptive and Reversible Gross MorphologicalChanges in Crucian Carp Gills. Journal of Experimental Biology206:3667-3673.

Stephens, W.W., M.T. Moore, J.L. Farris, J.L. Bouldin, and C.M.Cooper, 2008. Considerations for Assessments of WadableDrainage Systems in the Agriculturally Dominated Deltas ofArkansas and Mississippi. Archives of Environmental Contami-nation and Toxicology 55(3):432-441.

Sullivan, A.B., D.M. Snyder, and S.A. Rounds, 2010. Controls onBiochemical Oxygen Demand in the Upper Klamath River, Ore-gon. Chemical Geology 269:12-21.

Swingle, W.E., 1972. Length-Weight Relationships of AlabamaFishes. Fisheries and Allied Aquaculture Departmental SeriesNo. 1. Agricultural Experiment Station, Auburn University,Auburn, Alabama, 89 pp.

Todd, M.J., G. Vellidis, R.R. Lowrance, and C.M. Pringle, 2007.Measurement of Water Residence Time, Flowpath and SedimentOxygen Demand in Seasonally Inundated Floodplain Swamps ofthe Georgia Coastal Plain. In: Proceedings of the 10-14 March2007 Fourth Conference, A. McFarland and A. Saleh (Editors).American Society of Agricultural and Biological Engineers, St.Joseph, Michigan, pp. 140-147.

Todd, M.J., G. Vellidis, R.R. Lowrance, and C.M. Pringle, 2009.High Sediment Oxygen Demand Within an Instream Swamp inSouthern Georgia: Implications for Low Dissolved Oxygen Lev-els in Coastal Blackwater Streams. Journal of the AmericanWater Association 45(6):1493-1507, doi: 10.1111/j.1752-1688.2009.00380.x.

Turner, R.E., D. Justic, and N.N. Rabalais, 2008. Gulf of MexicoHypoxia: Alternate States and a Legacy. Environmental Scienceand Technology 42(7):2323-2327, doi: 10.1021/es071617k.

Vellidis, G., R. Lowrance, M.C. Smith, and A.W. Milton, 2003.Establishing the Natural Range of Dissolved Oxygen Levels inStreams of the Southern Coastal Plain of Georgia. In: Proceed-ings of the 8-12 November 2003 Conference. American Societyof Agricultural Engineers, St. Joseph, Michigan, pp. 414-419.

Viessman, W., Jr. and M.J. Hammer, 1993. Water Supply and Pol-lution Control (Fifth Edition). HarperCollins College Publishers,New York.

Wedemeyer, G.A., F.P. Meyer, and L.S. Smith, 1976. Diseases ofFish Book 5: Environmental Stress and Fish Diseases. TFHPublications, Inc., Neptune City, New Jersey, 181 pp.

Whittemore, R.C., 1982. An Examination of the Effect of Forest Lit-ter on Water Quality. Technical Bulletin No. 376, NationalCouncil for Air and Stream Improvement, Inc., Research Trian-gle Park, North Carolina.

Yarborough, G.M., D.A. Pennington, and G. Lucas, 1994. TheEffects of Low Flows on the Flathead Catfish (Pylodictis Oliv-aris) in the Upper Sunflower River. Mississippi WaterResources Research Conference. http://www.wrri.msstate.edu/pdf/Yarborough94.pdf, accessed December 31, 2009.

Zhang, H., S.A. Ludsin, D.M. Mason, A.T. Adamack, S.B. Brandt,and X. Zhang, 2009. Hypoxia-Driven Changes in the Behaviorand Spatial Distribution of Pelagic Fish and Mesozooplanktonin the Northern Gulf of Mexico. Journal of Experimental Mar-ine Biology and Ecology 381(Suppl. 1):S80-S91.

Zimmerman, L.R., E.M. Thurman, and K.C. Bastain, 2000. Detec-tion of Persistent Organic Pollutants in the Mississippi DeltaUsing Semipermeable Membrane Devices. The Science of theTotal Environment 248(2-3):169-179.

SHIELDS AND KNIGHT


SIGNIFICANCE OF RIVERINE HYPOXIA FOR FISH: THE CASE OF … Sunflower Jawra.pdf · SIGNIFICANCE OF...

Documents

Transcript of SIGNIFICANCE OF RIVERINE HYPOXIA FOR FISH: THE CASE OF … Sunflower Jawra.pdf · SIGNIFICANCE OF...