J. N. Am. Benthol. Soc., 1988, 7(4)387-409 e) 1988 by The North American Benthological Society North American prairie as systems for ecological study*

WILLIAM J. MATTHEWS Biological Station and Department of Zoology, University of Oklahoma, Kingston, Oklahoma 73439 USA

Abstract. The Great Plains and Osage Plains of interior North America included vast prairie regions before settlement by western man. Prairie streams that exist today are ecologically interesting for their unstable flow regimes and harsh fluctuations in environmental conditions. How much extant prairie streams have changed from their pre-settlement conditions is unknown, but writings of early explorers suggest that then, as now, mainstream flows were highly variable. Although historically, some prairie were turbid, the siltiness of others has increased with agricultural expansions, and some mainstream biota () are known to have disappeared. Small streams in these regions have probably changed even more; although they were clear 100 years ago, many are now highly turbid. Differences in ecological conditions among streams of the prairie region may be as great as those between prairie and nearby upland streams. For example, streams of the southern plains are characterized by irregular flow and substrates of small size, whereas streams of the northern prairies are more consistent in flow and many have cobble substrates. For some prairie streams, published studies exist of many ecological features, such as hydrologic regimes, produc- tivity, respiration, organic matter processing, and composition of the biota; however, such basic features are unknown for most streams of the interior plains. Limited measurements suggest that typical central or southern prairie streams differ from streams of nearby uplands or northern forested regions in functional properties and biotic composition. Community structure and ecological func- tioning of many prairie streams appear strongly influenced by physicochemical limitations resulting from irregular flow regimes and environmental harshness, such as major disturbances due to drought and . Despite the above conditions, numerous examples suggest that prairie faunas are also influenced by biotic interactions, including those due to multi- effects. A relatively new approach in stream ecology applies the analysis of complex hydraulic parameters to questions concerning distributions or adaptations of the biota. This approach may be profitable in prairie streams, but needs modification to include the sometimes lengthy periods of low or no flow in these systems. As a result of the discussions giving rise to this paper, numerous specific topics are suggested for future investigations. These can be generalized in four categories: (1) basic description and comparison of biotas, processes, or rates; (2) biotic adaptations; (3) controlling mechanisms; and (4) comparisons of prairie stream with those of other kinds of temperate streams in North America. Streams of the Mississippi Embayment, Interior Highlands, the Rocky and Appalachian mountains, and the prairies of our inland plains provide a contrasting array of study sites at a similar latitude for consideration by ecologists desiring a broad comparative or experimental ap- proach to questions in stream ecology. Key words: prairie streams, historical changes, hydrologic regime, , microbial pro- cessing, organic matter dynamics, biotic interactions, disturbance, physicochemical limitations, stream biota.

The region the Rocky Mountains (Fig. 1), included the wid- est reach of unbroken prairie that existed in The Great Plains and the Osage Plains, a vast North America two centuries ago. Geologically, expanse of relatively flat land reaching from the the Osage Plains are part of the Central Low- Ozark and Ouachita uplands to the foothills of land which covers about 1,680,000 km' in the central United States and Canada (Hunt 1974). * Paper resulting from discussions at a workshop The Osage Plains rise toward the west to an on "Prairie streams" (University of Oklahoma Bio- logical Station, Kingston, Oklahoma, 1-2 May 1987) altitude of approximately 600 m at the 100th and at a symposium on "Community structure and meridian, which is an approximate boundary function in temperate and tropical streams" held 24- between the Central Lowland and the Great 28 April 1987 at Flathead Lake Biological Station, Uni- Plains Physiographic Province. The Great Plains versity of Montana, Poison, USA. Province, including the Alberta Plain of Canada 387 388 W. J. MATTHEWS [Volume 7

exception of stream borders, more open grass- land than forest. The forested reaches that do exist, like the "crosstimbers" of southern Kan- sas, central Oklahoma, and north Texas, consist of trees that seem dwarfed compared with their eastern counterparts. The central plains region is dissected by ma- jor rivers like the Missouri, Platte, Kansas, Ar- kansas, Cimarron, Canadian, Washita, and Red that drain generally from west to east (Fig. 2). The larger of these systems have headwaters in the eastern slopes of the Rocky Mountains or the high plains of eastern Colorado or North Texas, and roughly parallel one another en route to the Mississippi or the Gulf of Mexico. The southern Great Plains is drained by streams of the Texas gulf coast, including the Trinity, Brazos, Colorado, and Pecos Rivers (Fig. 2). In 0 1300 km Canada, the upper Great Plains are drained by FIG. 1. Location of the Great Plains and the Osage the vast McKenzie, Peace, and Saskatchewan Plains of North America, and of the Interior High- River systems. lands (Ozark and Ouachita uplands). It is relatively easy with standard references to delineate geographically this interior plains (Fig. 1), includes approximately 1,490,000 km2 or prairie region and the major drainages there- of semiarid land (Hunt 1974), rising from the in. Much more difficult is the task of defining east to an altitude of about 1675 m at the base a "prairie stream" as a unique type that can be of the Rocky Mountains. While there are details discussed in the context of modern aquatic ecol- of difference in structural geology of the Great ogy. For example, cool and warm water rivers Plains and the Central Lowland, Hunt (1974) of the northern Great Plains may have higher showed that these two formations are separated gradient, coarser substrates, and more persis- mostly by the 100th meridian, the 600-m con- tent flow than similar-sized streams farther tour, the 50-cm rainfall line, a boundary be- south in Kansas, Oklahoma, or Texas (J. A. Gore, tween tall and short grasses, and the eastern personal communication). Because of the diffi- limit of Tertiary formations that contain sedi- culty of generalizing about prairie streams from ments eroded from the Rocky Mountains and Canada to the southern United States, much of washed onto the plains. this paper focuses upon streams of the central These interior plains are not all homoge- and southern plains with which I am more fa- neous geologically or biologically. For example, miliar, and which were the subject of most dis- in many places they are not flat. The Flint Hills cussion in the Prairie Streams Workshop. of eastern Kansas have relatively sharp relief, with flat caprock mesas grading downward Existing conditions through grassy slopes to cottonwood-lined streams similar to streams of the Ozark Another difficulty is knowing whether or not upland. In Oklahoma, the Arbuckle and Wich- the prairie streams that we perceive today are ita Mountains contrast sharply with the prairie, reasonable facsimiles of those a traveler would and generally the topography is more undu- have found 200 years ago, or if they are mere lating than the words "plain" or "prairie" sug- remnants of the former systems, having been gest. Large salt flat or sand areas exist, and ravaged by pump, plow, and pollution. Prairie the Black Mesa in the Oklahoma panhandle streams of the vast interior plains range in size provides sharp vertical relief in sandstone. These from the large river mainstreams to tens of local phenomena aside, however, the interior thousands of kilometers of intermittent or plains are united collectively by relatively low ephemeral streams that flow only during part relief, highly variable rainfall, and, with the of the year. A few years ago, I collected min- 1988] NORTH AMERICAN PRAIRIE STREAMS 389 flows in streams of west Kansas, taking large numbers of in the Smoky Hill and other small rivers in the area. Shortly thereafter I at- tempted another collecting trip on a north to south axis some 40 kilometers west of my pre- vious route. The trip was a loss, as the same streams slightly farther west had very little or no water. Even the mainstreams can be elusive. I have collected fish in the Arkansas River at Great Bend, Kansas, where the water was knee to waist deep, 50 m or so wide, and with sub- stantial flow although there had been no recent rains. By traveling upstream a few kilometers, I have stood with one foot on each bank of this same Arkansas River mainstream, where there was scarcely enough water for mosquitofish and a few red shiner minnows. In a similar fashion, other southwestern rivers such as the Rio Grande and the Canadian wax and wane through their course downstream, responding not only to va- garies of rainfall but to withdrawal of irrigation or municipal water from the river channels or porous aquifers. Despite tremendous variation among main- FIG. 2. Major drainages of the interior plains of streams of the prairies, creeks can be the United States. even more unpredictable. In many prairie creeks "you haven't been there until you've been there a whole year", and this probably should be stat- Many prairie mainstreams, defying the popular ed as "five years" or "ten years" or more. Un- concept that they are muddy, are actually rather certainty among days, seasons, and years seems clear except immediately after rains. Rivers to be normal rather than unusual in these flowing over sand, like the Ninnesecah or the streams. "Stream" research begun in boldly Chikaskia Rivers of southern Kansas, can be flowing streams in southern Oklahoma can be very clear, and even the North and South Ca- frustrated in late summer as flow ceases and nadian Rivers of Oklahoma are relatively clear moisture disappears from desiccated streambeds. at low flow. However, a few kilometers away, In Marshall County, Oklahoma, numerous e.g., in the Walnut River at Winfield, Kansas, creeks seem fairly similar most of the year: steep or the Washita River in southern Oklahoma, soil-gravel banks, beds of sand and gravel over turbidity levels are very high. The Washita Riv- coarse fragmented limestone, and relatively er, which drains erosible red soils of south- clear water with substantial flow. However, in western Oklahoma, is the most turbid river I the entire county only Brier Creek, with its nu- have encountered. Similar differences in tur- merous springs, flows in late summer bidity are apparent in northern prairie rivers (Power and Matthews 1983) while all other in Montana, Wyoming, and South Dakota: the creeks lose surface flow and may dry complete- Yellowstone, Big Horn, and Tongue Rivers are ly. clear, whereas the Powder and Cheyenne Rivers Despite the wide variations in streams of the are highly turbid with quicksand bottoms (Gore, central and southern prairies, the region has personal communication). some generally definable stream types, from Prairie mainstreams and creeks do not differ largest mainstreams to smallest creeks. All have merely in transparency. Dissolved solids may in common a potential lack of water. Beyond differ dramatically across short distances. The limitations in amounts of water, prairie streams Red River flows in western Oklahoma across seem most differentiated by geology, which de- deep Permian salt deposits from which ions are termines the substratum over which they flow. leached to the surface by numerous springs and 390 W. J. MATTHEWS [Volume 7 small creeks. As a result, its salinity is higher sas River and its larger have long than that of neighboring rivers, with conduc- been subject to fluctuation in amount of dis- tivity approaching that of sea water. In those charge. There seems also to have been consid- conditions only two native fish species are able erable fluctuation in the past, as now, in the to survive (Echelle et al. 1972). Near the town degree of turbidity, especially in the larger of Oscar, Oklahoma, a small creek flows with streams". Metcalf noted that observations by clear water about 4 km from its salt- source the naturalist 0. P. Hay in 1885 suggested that to the Red River. The salinity of this beautiful stream was notable only in eastern parts little creek exceeds that of sea water much of of streams of northwest Kansas, but that in the the year, and the ichthyofauna is dominated by 1900s siltation progressed farther upstream, salt-tolerant pupfish. When spates or high waters owing to attributed to agriculture. Met- dilute the stream, fishes temporarily invade from calf (1966) noted depletion of numerous fish the river, but disappear from the creek as salt species since the time of Hay, and that certain concentrations return to normal (C. Hubbs, gastropods have become extinct in the Kansas University of Texas and D. Edds, B. Wagner, River system in this century. Cross and Moss Oklahoma State University, unpublished data). (1987) documented negative changes in fish faunas of the Kansas prairie streams, noting that Historical perspective small-stream fish faunas disappeared or suf- fered declines first, and that in the last 30 yr To have a frame of reference for considering even fishes generally tolerant of "big-river" prairie streams as natural, potentially coevolved conditions have declined markedly. systems, we need to know how much these Isaac Cooper was a member of the Fremont streams now differ from their pristine state. We expedition as it traveled across Kansas to Col- know that vast differences in the prairies fol- orado, south through the present Raton Pass of lowed settlement: fires no longer held back en- New Mexico, then eastward along the South croachment of trees in some areas (Leopold Canadian River to Ft. Gibson, Oklahoma (Mow- 1949); agricultural activities deeply overturned er and Russell 1972). Of streams in the Kansas the soils and introduced many artificial chem- River basin (June 1845) he wrote: "we found icals into watersheds; cattle replaced the native pretty good water—there exist some fine springs bison as dominant grazers; and gallery forests along these streams and the finest water I ever were cut to provide firewood or lumber. In the drank was out of the spring called Big John. face of such alterations, how different are prai- For the most part however, travellers have to rie streams in 1988 compared with those in 1788? drink rain water . . . though this stream (Ar- Metcalf (1966), reviewing studies by histori- kansas River) be a mountain torrent and flows ans and anthropologists, suggested that from snow covered peaks, yet owing to its wide droughts sufficient to depopulate the plains oc- channel at this place and loose, sandy bed, there curred repeatedly in prehistoric times. As early was barely sufficient water in it to flow. The as 1800, considerable variation existed in cur- sun's rays had full control over it and it was rent or depth of streams in the Kansas River warmer than fresh milk during the day". One basin, as suggested by journals of Lewis and cannot help being impressed that this descrip- Clark (Metcalf 1966). Turbidity appears to have tion fits much of the mainstream Arkansas River been quite variable; Thomas Say reported the today, upriver of its conversion to the Kerr- gallery forest of the Kansas River "about a half- McClellan navigation channel by the U.S. Army mile wide, but not entirely uninterrupted". Fre- Engineers. mont found the lower Blue River in Kansas to The Smoky Hill River (Kansas) was described be a "clear and handsome stream . . . running by Cooper as a "broad channel filled with mud- with rapid current" (Metcalf 1966). Fremont also dy and thick water . .", and, "its water is of a found some creeks in the region had clear water dusky and smokey colour being impreg- and sandy beds, but that others were dry. Met- nated with clay & sand" (Mower and Russell calf (1966) noted that none of the early explor- 1972). In contrast, in field notes of 14 June 1978 ers in the Kansas River basin reported silty bot- I described the Smoky Hill River as "clear to toms, although several mentioned sand or slightly turbid", over a bottom of sand and small quicksand. Metcalf summarized that "the Kan- gravel. Thus, in 1845, before plowing or grazing 1988] NORTH AMERICAN PRAIRIE STREAMS 391 in the western plains, some streams like the Co. (west edge of Oklahoma) not a "noble Smoky Hill were apparently highly turbid or stream", but flowing in various small channels silt-laden during some periods. over a bed about 150 m wide, and "red with In the upper South Canadian River of mud". the present-day Texas panhandle, Cooper found Cpt. R. B. Marcy visited the Little Wichita "several small creeks of good water", and "a River (Clay Co., Texas) in early May 1852 (Fore- beautiful spring". The waters of the Canadian man 1937), stating that "The stream at fifteen River he found "impregnated . . . strongly with miles above its with Red River is red clay & sand" (Mower and Russell 1972), twenty feet wide and ten inches deep, with a much as it exists today. He also wrote: "the rapid current, the water clear and sweet". In course of the Rojo (=Canadian) is too variable; June 1981, I described the same stream as mud- alternately winding like a serpent from one side dy, with high turbidity, and a bottom of slimy to the other & when elevated by freshets . . . mud. On 9 May 1852 March found the Big Wich- overrunning the whole basin". This description ita River (Clay Co., Texas) deep, sluggish, about of the river channel applies very well now to 50 m wide, and the water at high stage very the South Canadian River. At another location, turbid, being "heavily charged with red sedi- Cooper reported the South Canadian River mentary matter". The Red River was 120 m wide, channel "upwards of a mile in width, and over 2 m deep at high stage, "highly charged with the whole of this space, there was flowing but a dull-red sedimentary matter, and slightly a small quantity of water in the southern side, brackish to the taste". In west Oklahoma, near not deeper than 2 or 3 inches". He described North Fork of Red River, Marcy's party often the "peculiarity" of the Canadian and most of found clear springs of "cold, limpid water" the mountain streams that flow across the prai- (Foreman 1937), which are rare or absent in that rie was that the bed was well-supplied with region now. water from the mountains, but that upon en- The writings of the early explorers, describ- tering vast beds of sand (east Texas panhandle ing streams in Kansas, Oklahoma, or Texas be- and west Oklahoma) it is "swallowed up" fore any substantial settlement, provide us with (Mower and Russell 1972). information on physical characteristics of the On 3 August 1853, Lt. A. W. Whipple found prairie streams before the advance of agricul- the Canadian River, 40 km above its junction ture and watershed modification. My overall with Arkansas, was about 120 m broad (Fore- impression is that the pre-settlement main- man 1941). "The water flows sluggishly; is of stream prairie rivers, like the Kansas, Canadian, whitish color, nearly clear, and less than knee- or Red were in many characteristics much as deep". Smaller streams described by Whipple they are now, with irregular or braided flow on 16 August 1853, were "but a thread, winding over wide beds, and periods of high turbidity. through a gravelly bed thirty feet in width . . . Silt apparently increased with advent of agri- showed water only in pools", much like many culture (Menzel et al. 1984, Metcalf 1966) and prairie creeks today. Slightly southwest of Pur- there is no doubt that changes in river main- cell, Oklahoma, Whipple found numerous streams have negatively affected fish faunas springs (none of which are now known to me (Cross and Moss 1987). However, we should to exist). On 25 August 1853, Whipple described also consider the possibility that large-scale tributaries of Walnut Creek south of Purcell, spates in prairie rivers with low relief and un- Oklahoma, as "Many rivulets, with crystal consolidated substrates may have frequently waters dancing in the sunlight . . . several changed patterns, bedload, erosion, branches, whose clear depths afforded new va- and conditions, both in historic rieties of fishes . . .". None of Walnut Creek to- and pre-historic time. Even a single 100+ -yr day is as clear as the stream Whipple described. event might result in significant changes in On 27 August 1853, Whipple found "the lovely channel geometry, with resulting differences in valley of Deer Creek, which bears the clear sweet suspended load or water chemistry. waters of numerous tributaries to the Canadi- The greatest discrepancies between the an". In 1978 I recorded this creek as "highly streams described by the early explorers and turbid, red". On 4 September 1853 Whipple those of today seem to be in the tributaries. In found the South Canadian River in Roger Mills the journals of Marcy, Whipple, and others, 392 W. J. MATTHEWS [Volume 7 many creeks described in the 1850s as clear or snow, and released to swell streams with spring free-flowing are today turbid, intermittent thaw. In the desert Southwest, stream flows are streams. The prairie streams probably show more highly seasonal; some exist only as flash overall impact of post-settlement alterations (spates) during the late summer "monsoon" (e.g., than do streams of surrounding uplands. How- in southern Arizona). In the Interior Highlands ever, even in rugged parts of the Ozark uplift, (Robison 1986), streams fluctuate less than those streams have not been immune to changes in of the prairie, but most Ozark or Ouachita basic hydraulic regimes due to cultural modi- streams show some seasonality, with high flows fications of watersheds. Black (1954) vividly de- in spring-early summer. scribed how clearcutting and conversion of Streams of the southern prairies have season- Ozark hillsides to pasture can change clear pe- al features that are somewhat like those of low rennial streams to ones flowing only half the tropics of both hemispheres. In lowland tropics year, and subject to extremely high of Panama (Zaret and Rand 1971) and southeast after rains. Asia approximately half the year is extremely We can only speculate about the degree to wet and half very dry. In the southern plains which the biota and ecological processes in of Vietnam slightly north of the Mekong delta, prairie streams now resemble those of the last a dry period from approximately August to Jan- century, but should be alert to the possibility uary desiccates small streams. Early in the new that "adaptations" of biota to prairie streams year rains begin, typically with heavy down- may have evolved in systems with character- pours late each afternoon. The previously dry istics or flow schedules somewhat different from countryside becomes a quagmire, and streams those of today. If, for example, one used sea- run full (Matthews, personal observations). In sonality of flow or physicochemical conditions spite of the generality of tropical wet-dry sea- to explain insect or fish life cycles in an extant sons, streams of the low tropics have flow pat- prairie stream, how might conclusions be al- terns that vary considerably and unpredictably tered if we knew that two centuries ago the among years, thus it may be difficult for stream physical characteristics or the hydrologic re- to adapt to any specific temporal pat- gime of the stream were vastly different from tern (B. Statzner, personal communication). Does present? similar low predictability apply for streams and organisms of the American plains? Streams of the central and southern prairies Ecological characteristics of of North America have a distinct wet-dry cycle, prairie streams with heavy rains in spring and early summer. Hydrologic regimes After mid-summer, evaporation is rapid and many prairie streams are subject to annual des- One important feature of streams throughout iccation from late summer through winter. Much the central and southern plains is their general of the prairie receives rain in late summer that seasonality of flow due to common patterns in on average almost equals that of spring or early climate and geology. Although the Osage Plains summer. In Tulsa, Oklahoma (96°W latitude) all to the east are wetter than the more western months from March to October average 7.6 cm Great Plains, this entire prairie region is sea- of rainfall (U.S. Department of Agriculture 1941). sonal in rainfall and evapotranspiration and, Weatherford, Oklahoma (99°W latitude) aver- therefore, in stream flow. Northern Great Plains ages >5 cm monthly of precipitation April to rivers probably have a very different hydro- October. For all of Oklahoma, mean precipita- graph, and flow more continually than do prai- tion from 1886 to 1938 averaged >7.6 cm rie rivers in Kansas or Oklahoma (Gore, per- monthly from April through September, with sonal communication). Many other parts of only November through February having <5 North America have seasonally flowing streams. cm (U.S. Department of Agriculture 1941). Des- The in the west (and the northern iccation of prairie streams during late summer Great Plains rivers) is strongly correlated to probably relates more to transpiration and snowmelt and depth of the montane snowpack. evaporation due to summer heating and inso- In the north, winter precipitation is retained as lation than to actual lack of precipitation. By 1988] NORTH AMERICAN PRAIRIE STREAMS 393 late summer, prairie soils are very dry, and little east-west location and overall stream size, but of the rain that does fall actually reaches a stream all show substantial increases in discharge in bed. May-July, and decreases from August through G. R. Marzolf (personal communication) sug- winter. gested that understanding movement and dis- In contrast, note flow patterns (Table 1) at tribution of water is the requisite first step in three stations on the Arkansas River in west any synthesis of a prairie . He envi- Kansas from Lakin (westernmost) to Great Bend sioned the driving variables as (1) the mid-con- (easternmost). At Lakin, flow is modest in the tinent hydrologic regime, (2) solution processes stream channel year-round. At Dodge City, the in soil and groundwater, and (3) riparian vege- Arkansas River is essentially dry, but at Great tation. Elucidation of these complicated inter- Bend a few kilometers downstream with inputs relationships is one of the goals of the Long from small tributaries it becomes a substantial Term Ecological Research (LTER) studies at stream with year-round regulated flow. The U.S. Konza Prairie (Kansas). Marzolf noted that the Geological Survey notes that for the Arkansas location of tallgrass prairie coincides with the River at Dodge City, "natural flow of stream dividing line between positive and negative affected by transmountain diversions, storage mean annual precipitation-to-evaporation val- reservoirs, power developments, ground-water ues. To the east, in tallgrass prairie, there is withdrawals and diversions for irrigation, and enough water to support tree growth, but pre- return flow from irrigated areas". Clearly, in historic prairie wildfire was sufficiently fre- planning any studies of prairie mainstreams, quent to prevent establishment of trees. investigators must take into account a myriad In the prairie studied by Marzolf most streams of cultural factors in addition to natural climate are intermittent, and water flux depends upon and geomorphology. lateral versus downward movement of ground- water. Konza prairie has an annual water bud- Substrate-water interactions and deep get of 820 mm, with approximate outputs of 180 interstitial biota mm lost to groundwater, 450 mm evapotrans- piration, and 190 mm for streamflow (Marzolf, Southern and central prairie streams may dif- personal communication). In the Konza Prairie fer from stony-bottomed upland streams in the four gauging weirs and many precipitation extent of soil-water interactions, or the degree gauges facilitate an ongoing study of water flux to which flowing water is exchanged with deep in unburned prairies, versus those burned at substrate. Many prairie streams can be separat- varying intervals, and will soon permit assess- ed into two classes: those with a primary sand ment of effects of native bison (Marzolf, per- substratum versus those with clay soil as the sonal communication). bed. Both types of substratum differ markedly Apart from local vagaries in climate and from that typical of stony upland streams, al- streamflow, the central and southern plains are though in some uplands, e.g., the Ozarks, characterized by streams that are water-limited streams may tend toward sandy bottoms. (Many at least part of the year. The biota must be able northern Great Plains rivers also have cobbled to survive potentially harsh conditions related substrata over most of their length, and thus to highly seasonal cycles in flow, punctuated may have more water exchange with the stream by occasional brief but extreme spates that can bed.) In clay-bottomed prairie streams, water occur at any time of the year. Although many has long residence time in individual pools as prairie mainstreams now have flow partially stream flow decreases in late summer. In iso- regulated by impoundments, there is still much lated pools or in pools minimally connected by evidence of strong seasonality of flows. In Table small trickles, a long residence time probably 1, streams relatively unaffected by any obvious allows water and soil to come to chemical equi- disturbances include the Smoky Hill River, librium. Such pools become in effect small Walnut Creek (Kansas), Walnut Creek (Okla- , with sharp thermal differences as much homa), Beaver River in the Oklahoma panhan- as 9-10°C from surface to bottom (Matthews, dle, and the Canadian River. Flow differs mark- personal observation) and perhaps more lentic edly among these selected streams owing to their than lotic characteristics. Soil-water interaction to s.0 &

TABLE 1. Total monthly discharge (cubic feet per second) for prairie streams, October 1981 to September 1982. Information from U.S. Geological Survey "Water year" reports for the states.

Stream (Town, State) Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Smokey Hill

(McAllaster, Kansas) 0 0 0 0 0 5 9 589 670 741 0.11 0

Walnut Creek .Ak

1 . (Rush Center, Kansas) 0 0 0 0 16 747 0.7 233 380 287 0 0 ' Arkansas River (Lakin, Kansas) 95 199 346 344 779 557 129 365 112 3899 101 2980 Arkansas River (Dodge City, Kansas) 0 0 0 0 0 0 0 0 0 11 2 0 Arkansas River SMaRLIVjAl (Great Bend, Kansas) 1046 1441 1303 1073 1158 3355 1267 2220 1057 3520 554 183 Walnut Creek (Purcell, Oklahoma) 931 1352 371 477 487 743 927 15,001 2901 1040 357 414 Beaver River (Beaver, Oklahoma) 0 17 3 4 6 27 26 1499 3185 987 802 0 Canadian River (Purcell, Oklahoma) 88 358 4002 736 598 3493 1143 2265 12,016 12,429 4257 1498 1988] NORTH AMERICAN PRAIRIE STREAMS 395

is also increased as many prairie streams flow year, etc. complicate interpretations. However, over highly erodable soils, and at elevated flow prairie streams for which net or gross primary carry large loads of soil particles. Within the productivity measurements exist appear to be , these particles can interact with relatively highly productive systems, or to have dissolved ions, and also adsorb algal cells and high rates of community metabolism. Bott et al. increase their precipitation from the water col- (1985) found high rates of primary production umn. Thus, I postulate that a clay-bottomed in streams of the southern Great Plains, ap- prairie stream is a system in which a relatively proaching that of desert streams. Hill and Gard- impermeable substratum minimizes water ex- ner (1987a) found two prairie streams of north change to deep . If the water in these Texas with summer gross primary productivity -2 -1 streams is effectively sealed from exchange with (GPP) ranging from 0.7 to 7.1 g 02 m c1 , while deep sediments, there is less potential impor- 24-hr community respiration was 0.6-5.3 g 02 tance of water and biotic exchange with the m-2 d-', and commented that these rates of com- deeper strata. The active biofilms that may be munity metabolism approached those for high- important deep in substrata of stony streams ly productive desert streams. Neel (1985) found (Bretschko and Klemens 1986) may be of min- in a northern prairie stream that occasions of imal importance or virtually nonexistent in clay- respiration dominance (negative balance) were bottomed streams of the prairies, but I know of rare. In a Kansas prairie stream, Gelroth and no studies addressing this question. Marzolf (1978) found the GPP and 24-hr res- In prairie streams with shifting sand substrata piration ranged 0.5-1.2 and 0.5-1.8 g 02m-2 d-1, or in stony northern prairie streams, water might respectively. My extrapolation of values from be exchanged extensively through the substra- Stewart (1987) (by multiplying by 14 assumed tum, and an active deep psammon could exist. hours of daylight) suggests net primary pro- Alternatively, if sand particles are small and duction on algal-colonized tiles in Brier Creek, -2 -1 well-packed, or if the substratum is character- Oklahoma of 0.7-1.93 g 02 m d in midsum- istically a sand-mud mixture, as in the Canadian mer, in a reach of stream where our earlier stud- River, water might flow only minimally through ies (Power et al. 1985) had suggested rapid in- the substratum for long distances. One other creases in standing crop of attached algae. opportunity for water-substrate interaction in Frances Gelwick (University of Oklahoma, per- small prairie streams is the situation in which sonal communication) found an average net -2 - water flowing from one perched pool to another primary productivity of 0.4 g 02 m hr ' at passes mostly through gravel or gravel-sand midday in full sunlight in April, for Brier Creek substrata of , affording time for chemical cobbles covered by dense growths of Rhizoclo- reactions with those substrata. Interaction of nium. water with gravel, particularly in riffles, may I am aware of little work documenting pri- be an important part of complex interactions mary productivity in upland streams of the In- that, depending upon temperature, can be im- terior Highlands. In one Ozark stream (Baron portant in regulating stream alkalinity (Stewart Fork, Oklahoma), Stewart (in Matthews et al. 1988). Brian H. Hill (personal communication) 1987) reported that stream bottoms dominated tested Stewart's hypothesis that gravel can by "felts" of blue-green algae were highly pro- -2 - abiotically remove significant quantities of ductive (0.6 g 02 m hr '), with productivity phosphorous from hard-water prairie streams, likely enhanced by the grazing of fishes. Very but permitted live microorganisms on the grav- high productivity values were found in Blue el (i.e., without drying, thus allowing microbes River, Oklahoma, an upland stream of the Ar- to live). The gravel removed up to 42% of phos- buckle Mountains, but these high values re- phorus from water with a higher percent re- ported by Duffer and Dorris (1966) may have moval attributable to microorganisms in peren- been very patchy. Overall, it is premature to nial than in intermittent streams. generalize about the productivity or rates of community metabolism in prairie streams rel- ative to that of streams at the same latitude in Productivity and nutrients the Ozark or Ouachita uplands, or the Rocky Comparing productivity among systems is dif- Mountains. Carefully coordinated simulta- ficult, as differences in methods, time of day or neous comparisons of this basic property of 396 W. J. MATTHEWS [Volume 7 stream ecosystems are needed to provide a start- farming to the edge of stream banks. Secondly, ing point for evaluation of similarities and dif- for prairie streams that depend on allochtho- ferences between and within such systems. nous inputs for energy subsidies, the material Factors limiting primary productivity in prai- often originates from grasses rather than trees. rie stream systems are not well-understood. Gurtz et al. (1982) showed that grasses com- Stewart (1987) found that nutrient enrichment posed 57% of the direct litterf all in open prairie (N + P + K) markedly increased primary produc- (although total particulate organic matter (POM) tivity in Brier Creek, Oklahoma, and that graz- input was much greater in stream reaches with ing by fishes (Campostoma anomalum) increased gallery forest). Many prairie streams are in close -specific primary productivity of pe- association with the tall or short grass species riphyton (but reduced standing crop and pro- of original prairie, or with pasture grasses. ductivity on an areal basis). In central prairie Grasslands rapidly accumulate detritus if they streams (Iowa), Burkholder-Crecco and Bach- are protected from fires, and this detritus can mann (1979) found that nutrient concentration affect nutrients and POM carried to the stream was not the factor limiting densities of sus- by overland flow during rainstorms. pended algae. In turbid prairie streams insuf- Small prairie streams are probably autoch- ficient light may limit algal growth, and the thonous in many cases. In Brier Creek and other shifting of fine-textured substrata (sands, small streams of south Oklahoma I have often silts) in some mainstreams probably inhibits observed large standing crops of Spirogyra that growth of attached algae, but I know of no doc- assume a floating "rope" growth form, with in- umentation of the latter. At low flow in the terwoven strands sometimes accumulating to South Canadian River (central Oklahoma), I lengths of a meter or more. Such growths are found dense mats of blue-green algae covering usually associated with shallows, attached to the sand substratum, but this typically occurred cobbles or gravel of shallow pools or riffles, and in autumn when the stream was relatively clear. are most common in late summer or autumn I know of no measurements of productivity from when stream flows are low. Agricultural activ- such mats in prairie rivers. Interestingly, ities such as fertilizer additions may stimulate Cam postoma anomalum, an herbivorous minnow this condition. However, I have also seen sim- whose diet is predominantly algae, thrives in ilar massive standing crops of attached algae in some turbid prairie streams. For example, I have Pennington Creek, Bryan County, Oklahoma, collected large numbers of this species from the which is a clear-water upland stream of the rug- Solomons River near Logan, Kansas, but I know ged Tishomingo Granite formation where ag- of no one who has determined the diet of in- ricultural inputs are low relative to row-crop dividuals from such turbid waters. areas. Autochthonous production coupled with The (Naiman et al. input of POM from grasses could form the basis 1987, Vannote et al. 1980) depicts many streams for a in these streams, and might min- as allochthonous in canopied headwaters, with imize the importance of leaf litter relative to its coarse particulate organic matter (CPOM) in- role in forested upland areas of North America. puts largely as leaf fall. Farther downstream, Prairie streams may differ from wooded those streams become more autochthonous, as streams in fundamental ways with respect to they widen and the canopy recedes, then still nutrient inputs or recycling. Nitrogen-fixation farther downstream the systems become het- by cyanobacteria, for example, is energy-inten- erotrophic if turbidity or depth inhibits pho- sive and so could proceed more rapidly in well- tosynthesis. Prairie streams likely differ from lighted streams than in streams where light is this picture in at least two ways. First, many less intense owing to canopy cover. Cultural small headwater prairie streams are sunlit due nutrient subsidies should not be overlooked in to a lack of forest, while farther downstream assessing nutrient budgets for prairie streams, where flow is more commonly perennial gal- and such subsidies probably exceed those of lery forests and closed canopy prevail (or pre- more upland regions. Vast areas of former prai- vailed, before settlement by western man). rie are now irrigated and fertilized extensively. However, even now most prairie streams have Additionally, a large percentage of native prai- at least some gallery forest along their lower rie is now pasture, and pasturelands are often reaches, largely because of the impracticality of fertilized to increase growth of forage. The cat- 1988] NORTH AMERICAN PRAIRIE STREAMS 397 tie themselves are a fact of life for ecologists per kilometer, and total seston concentration studying many prairie streams. Is it logical for was significantly related to discharge. Hill and us to seek pasture streams with cattle excluded Gardner (1987b) hypothesized that a lack of re- from study sites, when the typical range of nu- tention devices distinguishes prairie streams trient inputs in a large percentage of prairie from forested upland streams which have more systems now includes inputs from cattle near debris retention dams and filter-feeding mac- or within the stream? roinvertebrates, and might account for an over- all higher level of seston transport in prairie than in forested streams. Hooker and Microbial processing and organic Marzolf and Tate and found shred- matter dynamics (1987) Gurtz (1986) der insects low in abundance in prairie stream Microbial processing of materials may also be leaf-pack experiments, and that elm leaves de- uniquely adapted in grasslands (Marzolf, per- cayed faster in a perennial than in intermittent sonal communication). The research group at prairie streams. Brown and Ricker (1982) also the Konza Prairie reserve in Kansas found bac- found low abundance of shredders in leaf-packs teria of streams specialized for the kinds of in an upland Ozark stream that is near the plains. substrata they most often find in nature. In lab- Seston in the Texas and Kansas prairie streams oratory experiments (Marzolf, personal com- was dominated by ultrafine POM (Gurtz et al. munication, McArthur et al. 1985), bacteria iso- 1982, Hill and Gardner 1987b). Hill (personal lated from streams within gallery forest grew communication) concluded that prairie streams well on leachate from bur oak leaves, and also appear overall to be dominated by POM of on leachates from big bluestem grass. In con- smaller size classes than POM in forested streams trast, bacteria isolated from grassland streams of upland regions. grew well on big bluestem leachates, but poorly Presumably, much of the processing in in- upon bur oak leachates. Marzolf concluded that termittent streams in the presence of fewer bacteria from forested stream reaches are ex- shredders is microbial, leading to small partic- posed not only to tree leachates, but also to grass ulate size. The dominance of fine particulate leachates, because grass dominates the slopes organic matter (FPOM) in prairie streams may above the gallery forest. However, bacteria from also result from FPOM inputs by bank erosion, streams on grassy upland slopes would have no wind , and overland runoff (Gurtz et occasion to be adapted to grow on products from al. 1982, Hill and Gardner 1987b). Hill and the oak trees. Furthermore, components of some Gardner (1987b) also reported that periphyton bur oak leachates were toxic to some grassland is a dominant source of POM in some prairie bacteria. The bacteria in the grassy uplands may streams. Hill (1988) found in small tributaries be adapted to quickly process monosaccharides, of a fourth-order, intermittent Texas stream that which may be most available for short periods litter from grasses and forbs dominated the of time just after stormflows (Marzolf, personal POM, whereas farther downstream tree leaves communication). This line of investigation sug- were the primary POM source. However, these gests that microbial populations of prairie allochthonous sources influenced POM dynam- streams may offer opportunities for research on ics rather briefly (4 mo) after which periphyton highly specialized adaptations. production supported the stream ecosystem. Hill Little is known about functional responses of et al. (1988) also examined rates of leaf decom- streams to intermittency, which is common in position in Texas prairie streams, finding slight- Great Plains (Hill and Gardner 1987b). Theo- ly lower breakdown rates at intermittent than retically, seston dynamics (e.g., amounts, trans- at perennial sites. Also, within the perennial ports, rates of processing of CPOM) will differ stream, breakdown rates were higher at third- in intermittent streams, in which shredder or- and fourth-order sites than at a second-order ganisms are likely less numerous than in per- site. Hill (1988) experimented with models de- manent streams (Hill, personal communica- rived from classic stream hydraulic parameters, tion). Recently, Hill and co-workers compared designed to predict organic matter dynamics for seston dynamics in two intermittent Texas prai- a prairie stream; based on various studies by his rie streams. For both streams, retention was group in prairie streams, he concludes (personal slight, as there was less than one debris dam communication) that intermittent streams have 398 W. J. MATTHEWS [Volume 7 a lower ecological stability than perennial suggested that these spawning individuals were streams. young-of-year (Matthews, personal observa- tion). Why were some C. lutrensis in this stream Biota of prairie streams only 14 mm long (barely post-larval) in January of 1976? Were they in fact spawned in late au- The biota of prairie streams is relatively well tumn, just before cold weather? Little is known described, but the degree of coverage varies with about life-history tactics of most prairie fishes. taxonomic group and among stream types. Fish- Some of the biggest gaps in knowledge about es are likely the best-explored and best-known prairie-stream fishes are in the large rivers. The group. Earliest detailed investigations of fishes fish fauna of many large, deep prairie river in prairie streams were the U.S. Army "railroad mainstreams needs better quantification, and the surveys" conducted in the 1850s to seek routes effect of impoundments upon those faunas are from Missouri or Arkansas to the west coast, poorly understood. which led to major taxonomic advances (and The invertebrates of many prairie streams some false paths) in the 1860s. For example, the have been described in detail (Davis 1980, Gore expedition led by Capt. Robert Marcy crossed 1980, Henry 1986, Morris and Madden 1978), west Oklahoma and Texas, en route to the west but the degree to which a typical prairie stream coast in 1853. Surgeons attached to these ex- fauna differs from, say, that of upland Ozark peditions collected large numbers of plants, streams (Brown and Ricker 1982, Cather and fishes, and vertebrates which were later de- Harp 1975), awaits better detail on more streams scribed and classified by scientists at the U.S. in both regions. Neel (1985) provides a highly National Museum. Excellent summaries of zoo- detailed account of seasonal and annual varia- geography of prairie fishes are by Cross et al. tion in invertebrates of a small northern prairie (1986) and Conner and Suttkus (1986), and his- stream. He found lower diversity of inverte- toric alterations of the fish fauna of the central brates (133 taxa) than exists in many streams at prairies are documented by Cross and Moss lower latitudes. Prairie streams, particularly (1987) and Pflieger and Grace (1987). Matthews those that are intermittent, may have limited (1985) offers multivariate comparisons of hab- faunas. Brian Hill (personal com- itats of numerous common prairie-stream fish- munication) found only about 1-2% the abun- es, and most states in the region have a good dance of benthic invertebrates in an intermit- to excellent book on fishes by a skilled ichthy- tent prairie stream relative to the number that ologist. could be expected in a perennial upland stream Fishes generally are more diverse in the east- of the same size. The same factors that limit ern prairie, with greatest species richness in species richness of fish communities likely limit streams with both char- richness of invertebrate communities: lack of acteristics, stony bottoms in part, and complex water, unpredictable flows, homogeneous sub- structural features. Richness of fish faunas is strata, and (possibly even more critical to in- low in the western plains, where salinity vertebrates) a preponderance of mud or sand (Echelle et al. 1972), lack of water (Cross 1967), bottoms in many prairie streams. Conversely a or other physicochemical harshness (Matthews relatively rich may be found in riffles 1987) may exclude many fish species. Although of prairie streams. Finally, the invertebrate fau- the fishes of the prairies have been well ex- na of a special class of prairie streams—springs plored and described taxonomically, there is a or spring runs—can be as rich in species as those dearth of ecological knowledge for many prai- of more upland, mesic, or forested regions. In rie stream fish communities. Some common a two-summer survey of 50 springs through prairie fishes offer interesting enigmas. Do No- Oklahoma, J. J. Hoover and W. B. Milstead found tropis girardi really only in the midst of substantial invertebrate faunas associated with spates, as suggested by Moore (1944)? Are some springs in prairie areas (Matthews et al. 1983). of the small prairie fishes annuals, hatching in Although invertebrates in prairie streams early summer and themselves spawning in late have in general been reasonably well-docu- summer or early autumn?Cyprinella lutrensis, for mented, many areas remain unexplored. Few example, appears to breed successfully in late studies have focused on deep infauna of prairie summer, and general patterns in length-fre- streams, or interstitial biota (psammon) at any quency of populations in the Canadian River depth. If, as postulated above, clay substrata in- 1988] NORTH AMERICAN PRAIRIE STREAMS 399 hibit water-substratum exchange, in-fauna in inate primarily from the substratum, with the many prairie streams may be less than in stony diatom Nitzschia dominating both benthic and upland streams (Williams and Hynes 1974), but planktonic assemblages, but that there is much streams with unstable sand beds might be fruit- exchange between the two. Neel (1985) found ful sites for investigations of a micro-infauna. a substantial periphyton assemblage dom- Whitman (1979) investigated the psammon inated by Cladophora and diatoms in a northern community of a Texas creek, and Whitman and prairie stream. Even primary descriptions of the Clark (1982) found 2-3 ppm dissolved oxygen algal assemblage seem to be lacking in many as deep as 30 cm in the sand substratum of the prairie streams, and temporal-spatial dynamics throughout most of the year. of the flora of these streams is very poorly may be abundant in prairie riv- known. er mainstreams. Repsys and Rogers (1982) found relatively high densities of microcrustaceans in Biotic interactions the Missouri River from late autumn to spring. While drift-netting for fish eggs in the main Peckarsky (1983), Statzner (1987), and nu- channel of the Red River above Lake Texoma merous others have contrasted the importance one April, I found large numbers of cladocerans of abiotic and biotic dominance in stream com- in samples. However, these collections were munity regulation. Statzner et al. (1988) suggest made during high discharge, when some zoo- that biotic phenomena may dominate in pool found in river mainstreams may ac- environments where hydraulic stress is low, tually be produced in ponds or backwaters and whereas abiotic factors may play a more im- washed into the mainstream. William B. Rich- portant role in riffle (high hydraulic stress) en- ardson and Bret C. Harvey (University of Okla- vironments. As important as abiotic phenom- homa, personal communication) found substan- ena clearly are in many prairie streams, there tial numbers of microcrustaceans in pools of is ample evidence that biotic interactions like small prairie streams, including several taxa of algivory, predator-prey interactions, or com- benthic microcrustaceans such as Eucyclops and petition influence structure of prairie stream Bosmina that were formerly considered to be communities, and strong multi-level cascading obligate plankton. A benthic microcrustacean effects among components at various trophic assemblage can be common both in prairie levels have been documented. Power and Mat- streams (Brier Creek) and in nearby streams in thews (1983), Power et al. (1985), and Matthews the Arbuckle or Tishomingo granite uplifts (W. et al. (1987) showed in Brier Creek that algae- Richardson and B. C. Harvey, University of grazing minnows (Campostoma) could regulate Oklahoma, personal communications). location and amount of attached algae in stream Attached algae of prairie streams are histor- pools. Further, piscivorous basses ( Micropterus) ically less well known than are fish or inver- strongly influenced use by Cam postoma, tebrates. For example, in one study of periph- and thus directly affected dynamics of the stream yton of eastern Oklahoma streams, Pfiester et algae. Stewart (1987) showed that grazing by al. (1979) reported 344 taxa, 115 of which were Cam postoma increased the rate of primary pro- new records for Oklahoma. Power et al. (1985) duction by algae per unit biomass, although found that Spirogyra and Rhizoclonium dominat- cropping of the algae by these fishes lowered ed the periphyton of a south Oklahoma stream, net productivity per unit area. Stewart (1987) with attached diatoms and bluegreen algae al- also showed that even when algal growth was most ubiquitous as an understory. Power and enhanced by fertilizer additions, grazing by Stewart (1987) described the resistance to scour Cam postoma played a major role in regulating and recovery of the periphyton community from algae production. Few studies, if any, of grazing spates. In this stream (Brier Creek), algae largely effects due to stream invertebrates have been recovered in three to four weeks after a major carried out in prairie streams, but I have in scouring spate. Matthews (personal observa- progress a three-year study that will include tions) found that a similar period of time was this comparison. Further, this study will ex- required for re-establishment of blue-green al- amine effects of algae-grazing minnows not only gae felts on rocks in an Ozark upland stream on algae, but indirectly upon benthic inverte- after flood scour. Roeder (1977) found in the brates, and upon a variety of system-level pro- Skunk River, Iowa, that planktonic algae orig- cesses or phenomena such as productivity, 400 W. J. MATTHEWS [Volume 7 transport of particulate organic matter, and ni- N. bairdi continued its invasion (Pigg 1987). The trogen fixation in prairie versus upland streams. decline of N. girardi has been so drastic that the We have already found (Matthews et al. 1987) fish has now been listed by Oklahoma as a that some of the effects of algae-grazing min- "species of special concern", and it is rare in nows differ between Brier Creek and upland some parts of Kansas. While such circumstantial streams of the Ozark Mountains, and suspect evidence does not "prove" that competition is that differences in the dominant piscivore are rampant, and N. girardi is disappearing in some involved (Harvey et al. 1988). areas not yet invaded by N. bairdi (F. B. Cross, Harvey (1987) showed another example of personal communication), it suggests investi- cascading effects among multi-trophic levels in gations of common resource use by these species Brier Creek. In observational and manipulative would be useful. For all the "toughness" that experimental studies, he found that many larval we attribute to prairie stream organisms (phys- fish in the stream are eaten by "medium-sized" icochemical tolerance, ability to find food in fish: minnows and juvenile sunfish. However, harsh environment, etc.), perhaps many of them largemouth bass in the creek prey upon the are really "delicate" organisms, barely existing medium-sized fish. Harvey (1987) showed that in harmony with an environment to which they in pools where large bass kept the medium- are adapted, and highly vulnerable to change. sized fish at bay (i.e., forced them to inhabit shallow stream pool margins), larval fish were Physicochemical limitations provided refuges in which they survived in on the biota greatest numbers. The larger bass protected the larval fish by eliminating the threat from the The biota and biotic processes of many prairie fish that preyed on them. Algivory and pred- streams are regulated at least in part by phys- ator-prey control are most important in Brier icochemical stress. Water limitations and des- Creek during relatively stable environmental iccation set absolute limits of existence for fish, conditions, and become less important during but some invertebrate taxa and many algae and or immediately after floods or other distur- microbes can survive periods of drying of the bances. The Working Group at Flathead Lake stream bed. Matthews (1987) found that fish discussing biotic-abiotic interactions pointed recolonized a rewatered prairie stream within out the importance of time-scale considerations the spring or early summer of one year by where biotic interactions are concerned. How movement from permanent pools, and that pos- do effects of spates or droughts compare or in- itive correlation existed between oxygen tol- teract with biotic interactions to determine the erance and ability of species to colonize. Gore ultimate community structure or dynamics of (1982) followed colonization of fish and benthic community structure of prairie streams? invertebrates in a new channel of the Tongue I know of no clear demonstrations of com- River, and Smith and Distler (1981) evaluated petition in controlled experiments in prairie time of recovery of benthos after a chemical streams. However, there is circumstantial evi- discharge in a sandy plains stream. Many aquat- dence that interspecific competition can and may ic invertebrates arrive rapidly by flight, and al- have played a substantial role in producing the gae and microbes may reappear in a rewatered biotic communities we find in prairie creeks or stream by virtue of air transport as well as hy- rivers. Historically, the Arkansas River Shiner, dration of resistant structures. Notropis girardi, was restricted to the Arkansas Prairie streams may also be stressful environ- River drainage, and the very closely related Red ments with respect to temperatures and dis- River Shiner, Notropis bairdi, was restricted to solved oxygen concentrations (Matthews 1987, the drainage whose name it bears. In the mid- Matthews and Hill 1980). In the South Cana- 1970s, N. bairdi first appeared, apparently by dian River mainstream near Norman, Okla- accidental transfer, in the native range of N. homa, I found midsummer water temperatures girardi. During the next decade, N. bairdi spread up to 38°C, which appeared to restrict fish to rapidly through much of the range of the for- cooler microhabitats. Matthews and Maness merly abundant N. girardi, which virtually dis- (1979) showed a direct relationship between appeared from major portions of the system as thermal and oxygen tolerance of four minnow 1988] NORTH AMERICAN PRAIRIE STREAMS 401 species and their late-summer abundance in the In prairie streams of the central and southern Canadian River. Matthews (1987) showed more plains, physicochemical extremes are generally broadly that numerous fish species of the genus predictable over annual cycles based on mete- Notropis from prairie streams were more toler- orological data. If hot, low-oxygen conditions ant of temperature and oxygen stress than were occur every year in drying stream channels, the congeneric species of environmentally benign organisms might become adapted so that these upland Ozark streams. Overall, for minnows of apparently stressful events are not really a dis- the genus Notropis, Matthews (1987) gave evi- turbance. On a shorter time scale, a predictable dence that their success in harsh environments diel cycle of stress in many prairie mainstreams of prairie streams is related to physicochemical and tributaries results in lowest oxygen in early tolerance, acuity of selectivity of microhabitats, morning hours, and highest temperatures in or both. midafternoon. Again, aquatic organisms are In Brier Creek, intermittent headwaters are more likely to adapt to such predictable epi- harsh in summer relative to a perennially flow- sodes than to events that occur erratically. Resh ing mid- and lower section. During cessation et al. (1988) stressed that aquatic biota of the of flow in Brier Creek, we found temperatures prairies are likely adapted to allow for a certain of 39°C in shallow pools, and documented di- variability about some mean measure of poten- rect heat death of fish in the stream at that time tial stress. Events that fall within the expected (Matthews et al. 1982). In August 1982, contin- range would not constitute a disturbance, but uous temperature recorders showed average diel events outside the expected range (perhaps de- fluctuations of 9-10°C in Brier Creek headwa- termined by mean ± 1 SD, or some such con- ters, while temperatures fluctuated only 1-2°C vention) might constitute a disturbance. in the midreach (Matthews 1987). At this same Duration as well as intensity plays an im- time, early-morning oxygen concentration in 10 portant role in determining whether or not a of 11 isolated pools in the headwaters was 0.4- given event is a "disturbance". Overall, distur- 2.0 ppm, which is low enough to stress most bance would be an event that alters community . Within Brier Creek, longitu- organization or function, and allows for restruc- dinal distribution of the common fish species turing of the community. Parameters important was positively correlated with their tolerance in evaluation of a potential disturbance include to low oxygen conditions. intensity, frequency, duration, predictability, Another way to deal with stress, if the en- season, and the geomorphological setting. Mea- vironment is predictable, is to synchronize life surement of the disturbance would be accom- cycles to avoid exposure of vulnerable life stages plished by evaluating recovery time for stabil- to stress. In many parts of the world aquatic ity in levels of productivity, or stability in (or insects adapt, even among populations, to uni- similarity to) diversity of the previous com- voltine versus multivoltine reproductive pat- munity. Although numerous authors, e.g., Ross terns, depending upon seasonality of the en- et al. (1985) and Power and Stewart (1987) have vironment. Bernhard Statzner suggested evaluated particular disturbances in prairie (personal communication) that investigation of streams, no single study has exhaustively ex- life cycles of invertebrates in these harsh prairie amined all of the parameters above with respect streams, and comparing similar taxa among to a given disturbance event in a prairie stream. prairie, upland, etc. would be a fruitful area of Ross et al. (1985) did find that recovery of a investigation. stream fish community (Brier Creek, Oklahoma) was rapid (within a year) following extreme Disturbance drought. Droughts may be a more drastic disturbance Environments like prairie streams that in prairie streams than floods, if the droughts undergo stress from climatic extremes or from are prolonged. In the 1950s three to four years abrupt changes in water chemistry could be of extreme drought throughout the southern thought of as subject to frequent disturbance. Great Plains were coincident with changes in However, caution must be exercised in defining some fish communities (Hubbs and Hettler "disturbance" (Resh et al. 1988—see this issue). 1958). Matthews (1987) noted the destruction 402 W. J. MATTHEWS [Volume 7 of a headwater fish community in two succes- turbances?" However, my overall impression is sive years due to drought. Spates, in contrast, that although high discharge displaces adult seem to be have somewhat short-lived effects fish, destroys algae and invertebrate commu- in prairie streams. I have documented (unpub- nities, and harms immature fishes, recoloniza- lished) the composition of the Brier Creek fish tion proceeds rapidly when peak discharge assemblage by snorkeling a 1-km reach eight passes. Drought, however, kills more individ- times, year-round. Within days following a se- uals (at least of fish) and even after dewatering, vere spate in June 1983, the distribution of adult reestablishment of a biota must await coloni- fishes in this stream reach was similar to that zation or regrowth from dehydrated propa- before the spate, despite the physical severity gules. Further, drought or dewatering seems of the event (which was documented by Power more likely to eliminate critical components of and Stewart 1987). Harvey (1988) made a de- the system, such as microbes, than does high tailed study of washout of larval fish during discharge. However, after even extremely se- high discharge in Brier Creek in early summer. vere spates in which the entire substratum While flooding virtually eliminated minnow moved, I have found viable algae upon gravel and sunfish larvae <10 mm long, adults re- (e.g., within rugosities) which can rapidly re- sumed reproductive activity rapidly after the establish (within weeks) an active flora. spate. In a small upland stream not far from the Other disturbances that need consideration prairies, Gelwick (1987) found adult fish little in prairie stream systems are human distur- affected by a severe autumn spate, and that re- bances related to impoundments, agriculture, covery of fish assemblages from high-discharge lumbering, urbanization, mining, and so forth. effects was rapid. The may be The impacts of agriculture (Menzel et al. 1984) more severely disrupted by spates. Neel (1985) are likely to be the single most prominent fea- found macrobenthos of a prairie stream largely ture of the western culture in prairies. removed by spring spates, resulting in low pop- Slizeski et al. (1982) describe the wide variety ulation densities in late spring. The critical vari- of hydrological control measures that have been ables are likely event duration and mobility of applied to one major prairie river mainstream organisms. Fish may find short-term hydraulic (Missouri River), certainly representing major refugia during spates, but have no analogous disturbance from the pristine state. Effects of refuge in drought. Invertebrates, less mobile impoundments and other cultural perturba- than fish, may lack the ability to move to hy- tions upon prairie stream invertebrates are be- draulic refugia during substratum-moving ginning to be understood (Gore 1977, 1980, 1982, spates. Clearly, adaptation of life cycles of in- Gore and Bryant 1986, Morris and Madden 1978), vertebrates and fish and their life-history tactics but in many cases changes in a given system in intermittent prairie streams need to be better cannot be known precisely owing to lack of pre- understood with respect to flood and drought alteration surveys. Prairie stream impound- and the physical and chemical changes during ments may have less effect upon streams than these events. reservoirs in montane regions. Where there are Not all phenomena that constitute distur- steep valleys, dams are tall, reservoirs are deep, bance are predictable. During one late summer and epilimnetic releases are cold. Cold-water episode of dewatering, I marked for identifi- releases in many regions have altered tren- cation more than 700 sunfish, minnows, and chantly the biota of the tailwater streams (Craig bass that were crowded into two deep pools in and Kemper 1987, Hoffman and Kilambi 1971, a headwaters reach of Brier Creek. Unfortu- Lillehammer and Saltveit 1984, Ward and Stan- nately, rains were delayed, the pools dried up, ford 1979). This phenomenon has occurred in and all the fish died. If rain had fallen a week some prairie regions, with effects on the biota. or two earlier than it did, many of these fish For example, cold-water releases from Possum would have survived and been potential colo- Kingdom Dam in north Texas have clearly al- nists for that reach of stream. Perhaps the term tered the nature of the typically warm Brazos "stochastic" applies to some aspects of ecology River, and have created riffle below the of prairie streams, and perhaps not. Overall, it dam not unlike riffles in rivers of the Ozark may be difficult in prairie streams to answer a uplands. Interestingly, thermal tolerance and question like, "are droughts or spates worse dis- the enzyme systems of fishes below this dam 1988] NORTH AMERICAN PRAIRIE STREAMS 403 have actually changed in the 40 years since con- in winters with frequent thaw and refreezing. struction of the dam (King et al. 1985, Zim- J. A. Gore (personal communication) and stu- merman and Richmond 1981). In the northern dents have recorded ice as thick as 0.75 m on a Great Plains, numerous reservoirs with deep Wyoming prairie creek, and developed models hypolimnial releases have had dramatic effects to predict habitat availability under those con- upon tailwater biota (Gore 1977, 1980). How- ditions. ever, many reservoirs on southern prairie In all the disturbances in prairie streams, the streams are not typically so deep, and thus may perennial versus intermittent nature of flow in not markedly alter thermal characteristics of a the system needs to be taken into account, as river. Additionally, although prairie reservoirs well as the fact that many large prairie streams may regulate flow for some distance down- (like the Arkansas River) arise in mountains. stream, many steams (the South Canadian River These montane-to-prairie systems might be- for example) do not show much influence of have differently from systems that arise on the the impoundment far downstream from the prairie. The geomorphology of the stream chan- dam. Southwestern prairie rivers are so influ- nel and the point along a stream at which an enced by of water into their sandy event occurs may also determine the degree of stream beds, by evaporation, by local agricul- effect or whether an event can be called a dis- tural withdrawals and deposits of water, that turbance. Finally, it may be extremely impor- 50 km downstream from a typical southern prai- tant to know whether disturbances in prairie rie stream reservoir the mainstream probably streams act as reset mechanisms, or if they main- has little impact from the impoundment. If this tain prairie stream communities in a more-or- hypothesis is true, it contrasts with conditions less perpetual state of disclimax. in upland rivers like the White River (Arkansas) where flow is regular, and there are detectable Hydraulic approaches differences in the river for many miles down- stream from dams in the Ozark region, or with Until recently, hydraulic approaches were rivers of the more northern prairies where flow given very little emphasis in studies of stream is persistent (Gore, personal communication). organisms and their functional responses Fire is a disturbance to streams that may be (Statzner and Higler 1985, 1986). Statzner et al. more frequent, or was more frequent prehis- (1988—see this issue) summarize many recent torically in prairie streams, than in many up- advances in application of hydraulics in stream lands. Prairie fires were once frequent enough ecology; thus, details need not be repeated here. to prevent encroachment of forests onto the How well do hydraulic-based approaches to the prairie. Burning of prairie grass alters canopy study of stream processes or biota relate to prai- interception of rainfall (Gilliam et al. 1987), re- rie streams of the southern Great Plains? Much moves accumulated plant litter, enhances of the applicability of hydraulics to ecology of warming of soil by insolation, stimulates bac- stream organisms must assume that at least dur- teria, releases soil nitrogen, and increases pri- ing much of the year, or in critical periods of mary productivity on the burned site (Marzolf, their life cycles, flow is available and poten- personal communication). All the results above tially influential. In a typical small prairie stream affect the inputs of water, nutrients, leachates, with only a very small percent of total area and CPOM into prairie streams. Hence, fires consisting of riffles, many organisms might be perturb prairie stream ecosystems, but the per- adapted to factors other than hydraulic phe- turbation has both positive and negative effects. nomena. On the other hand, flow can be pe- In severe winters, ice is probably a distur- riodically dramatic in these streams (flash floods bance even in southern prairie stream systems. [spates], rapid stage rises), such that to persist, In shallow, intermittent prairie streams, com- organisms may need to be as highly flow-adapt- plete freezing eliminates fish. Even if the freeze ed as organisms are in perennial streams. is incomplete, ice cover can cause substantial Statzner suggested (personal communica- mortality of fish or invertebrates as oxygen is tion) that the hydraulic patterns that are critical depleted under the ice (Johnson et al. 1982, Mat- to invertebrates (or fish) in streams with sub- thews, personal observations). Neel (1985) not- stantial flow might be of less importance in many ed thicker ice cover on northern prairie streams streams of the North American prairies, de- 404 W. J. MATTHEWS [Volume 7 pending upon whether they are permanent, in- termittent streams that flow 20 to 80% of termittent (but flowing more than 20% of the the time, and in ephemeral or "interrupt- time), or ephemeral (with flow rare, i.e., less ed" streams that flow less than 20% of the than 20% of the time). In streams with only time, and during much of the year are dry occasional flow, or very low flow rates like many or exist as a series of pools? central or southern prairie streams, hydraulic (2) How do washout rates and retention times patterns may be critical during relatively little differ in upland and typical prairie streams, of the year. In many prairie streams, particu- where debris dams are largely lacking? larly in pools, other hydraulic patterns and/or Does the long retention time of water in non-hydraulic phenomena may regulate stand- prairie stream pools (by virtue of low flows ing crops of invertebrates and fish or their use or cessation of flow) play a major role in of microhabitats. nutrient transfer and all organic matter I suspect that "minimum flow" may not apply processing? in small streams of the prairie that normally (3) How variable are nutrient inputs from cease to flow for a significant portion of the rainwater (e.g., nitrate) throughout the year. A critical factor may be length of time Great Plains? Significant concentrations of between flows, as pools shrink and animals are nitrate seem to be in free-falling rainwater forced together spatially, or as quality of habitat at the Konza Prairie site, but concentra- fails: temperatures rise and oxygen declines. tions are less in parts of Oklahoma; com- What about animals normally in riffles, like parative studies seem needed. darters, forced into pools with predators? We (4) Can hydraulic parameters, geomorpholo- have no idea how this affects them. We also gy, and physicochemical measurements be have no idea how decreased flow, zero flow, incorporated into a useful hierarchy of drought, etc. affect phenomena such as com- prairie stream classification that includes petition between species. For example, in Brier variables like slope, channel morphome- Creek how do interactions among minnow or try, stream density, etc., to facilitate broad sunfish species change from early spring when comparisons within and among regions? flow is substantial to late summer when much Can expanded use of some of the U.S. Fish less water is available? and Wildlife, Environmental Protection What kinds of models make best predictions Agency, or U.S. Geological Survey systems about hydraulic effects in prairie streams? Pre- be helpful? How effective is a classification dictors for benthos include drift distance models based solely on mean annual discharge per (McLay 1970), benthic density-hydraulic en- unit of drainage area, perhaps with infor- vironment models like the Gore-Judy habitat mation on geology or rock type included? models (Gore and Judy 1981), and Statzner's How should hydraulic parameters be in- hydraulic system models (Statzner 1981). For cluded in stream classification schemes? fish, at least two approaches include optimum (5) How do precipitation-dominated versus swim speed models and stream position-net en- rock-dominated systems (e.g., in Rocky ergy gain models (Fausch 1984, Fausch and Mountains, with nutrient limitations) dif- White 1981, Trump and Leggett 1980). The de- fer? gree to which these models apply to intermit- (6) How do biogeographic phenomena con- tent prairie streams remains unknown. trol taxonomic and functional composition of prairie stream communities? Do biogeo- graphic influences upon distributions need Areas for investigation to be incorporated in schemes that seek to Some of the specific questions about prairie classify stream communities? (7) How important are considerations of streams that were raised during the workshop species diversity in prairie steams, not only are outlined in the preceding sections. Other taxonomically, but from the perspective of questions that were suggested by workshop functional groups? What are appropriate participants included the following: categories of "functional groups" for prai- (1) How different are stream biota and pro- rie stream invertebrates, or fishes? cesses in perennial prairie streams, in in- (8) Do droughts increase heterogeneity in 1988] NORTH AMERICAN PRAIRIE STREAMS 405

prairie stream systems (e.g., as pools be- studies could be devoted to basic ecological de- come isolated), whereas high discharge in- scription and comparison among prairie streams, tegrates or homogenizes the whole wa- or between prairie and nearby upland streams. tershed? For example, how do ecological rates and pro- (9) How do bacteria make a living in prairie cesses, or generation times of insects differ from streams? What regulates microbial metab- sand-bottomed streams of the southern Great olism in prairie streams? Plains, to streams of the Ozark upland, to cold (10) What are the dynamics of the rivers of the northern prairies? side of metabolism in prairie streams? What As studies in prairie streams progress beyond are the overall dynamics of the dissolved description or documentation, an increased fo- organic carbon (DOC) component in prai- cus will likely be upon mechanisms. In many rie streams? cases mechanisms will not be immediately ap- (11) How do root exudates contribute to nu- parent from comparative studies, no matter how trient inputs? How do dissolved materials detailed, and experimental approaches will help from roots of streamside vegetation get into clarify mechanisms that underlie community stream waters? structure or system processes. There will clearly (12) To what extent do animals and biofilms be room both for controlled laboratory research penetrate into sediments of prairie streams, and for well-designed manipulations in the and how important are they ecologically field. The strongest experiments will be those in these systems? Do spates that scour that permit clarification or quantification across stream beds and churn sediments result in several treatment levels, or that detect thresh- destruction of algae and bacterial films? olds, interactions, or indirect effects. Finally, although much information on prai- The preceding view of prairie stream ecology rie streams is lacking, the information that does and the Prairie Streams Workshop suggest many exist supports the suggestion that the high de- questions about these systems or about their gree of diversity of stream types within north- role in studies of stream ecology that can be temperate latitudes rivals that across latitudes. summarized in four categories: (1) basic descrip- Teams of investigators working at carefully se- tion and comparison of biotas, processes, or rates; lected field sites within a given temperate lat- (2) adaptations of the biota to prairie stream itude could profitably address many of the large conditions; (3) controlling mechanisms; and (4) questions in stream ecology on a comparative comparative or experimental studies of stream measurement or experimentation basis. Grant- ecosystems within the same latitude, including ed, studies restricted to one latitude have some streams of prairies, lowlands, and uplands. potential limitations. For example, regardless of Especially at lower trophic levels, the biota the elevations of sites that are chosen, locations of prairie streams is poorly known. Many basic at one latitude will have many similarities in rate measurements such as input and output of light regime, and studies that require simulta- nutrients or POM, productivity, or materials neous differences in day length would not be processing are lacking for most prairie stream possible. However, for other studies in stream systems. Therefore, within North American ecology investigators might find it desirable to prairie streams much research has yet to be done hold effects of day length as nearly similar as at the level of initial ecological exploration. possible, while varying elevation, rainfall re- Careful documentation of community compo- gimes, canopy cover, , stream sition and of system-level rates and processes discharge, and the like. Such studies could in- is needed if prairie streams are to be compared clude a series of field sites at one latitude in rigorously among themselves or with streams North America: streams of the coastal plains, of forested uplands. Little is known about ad- upland streams in the Appalachians and the aptations of the biota to the harsh, fluctuating Ozarks, low-gradient streams of the Mississippi conditions that characterize many prairie delta, very high-gradient streams of the Rocky streams. Numerous phenomena such as adap- Mountains or the Sierra Nevada, and streams tation of insect life cycles or life-history traits of the prairies and plains of the North American of many fishes are virtually unstudied in prairie Midwest. An array of stream types, in which streams. I suggest, therefore, that many future investigators could attack major questions in 406 W. J. MATTHEWS [Volume 7 stream ecology, would be available and valu- 1979. Potential productivity of able. three Iowa streams. Proceedings of the Iowa Academy of Science 86:22-25. CATHER, M. R., AND G. L. HARP. 1975. The aquatic Acknowledgements macroinvertebrate fauna of an Ozark and a del- taic stream. Arkansas Academy of Science Pro- I am grateful to all participants at the Prairie ceedings 20:30-35. Streams Workshop for many of the ideas that CONNER, J. V., AND R. D. SUTTKUS. 1986. Zoogeog- are summarized in this paper. In particular, I raphy of freshwater fishes of the western Gulf thank the following for their keynote addresses Slope. Pages 413-456 in C. H. Hocutt and E. 0. and critical reading of the manuscript: G. Rich- Wiley (editors). The zoogeography of North ard Marzolf, Kansas State University; Brian Hill, American freshwater fishes. John Wiley and Sons, University of Texas at Dallas; Bernhard Statz- New York. ner, University of Karlsruhe, FRG; and James CRAIG, J. F., AND J. B. KEMPER (editors). 1987. Reg- A. Gore, University of Tulsa. Jim Gore was par- ulated streams: advances in ecology. Plenum ticularly helpful with comments on differences Press, New York. between northern and southern prairie streams, CROSS, F. B. 1967. Handbook of the fishes of Kansas. Miscellaneous Publication Number 45, Univer- and Brian Hill provided access to unpublished sity of Kansas Museum of Natural History. data. Arthur Brown provided information on CROSS, F. B., R. L. MAYDEN, AND J. D. STEWART. 1986. upland stream systems. Several versions of the Fishes in the western Mississippi drainage. Pages manuscript were also critically reviewed by F. 363-412 in C. H. Hocutt and E. 0. Wiley (editors). P. Gelwick, A. P. Covich, M. E. Power, A. J. The zoogeography of North American freshwa- Stewart, and B. C. Harvey; and W. Richardson ter fishes. John Wiley and Sons, New York. and S. Wickham made helpful suggestions. It CROSS, F. B., AND R. E. Moss. 1987. Historic changes is now impossible to identify the specific in- in communities and aquatic habitats of plains tellectual contributions of all the individuals streams in Kansas. Pages 155-165 in W. J. Mat- who were involved in the development of this thews and D. C. Heins (editors). Community and evolutionary ecology of North American stream paper, but for any flaws in facts or conclusions fishes. University of Oklahoma Press, Norman. I take responsibility. I thank Loren G. Hill, Di- DAVIS, J. R. 1980. Species composition and diversity rector of the University of Oklahoma Biological of benthic macroinvertebrate populations of the Station, for his assistance and many courtesies Pecos River, Texas. Southwestern Naturalist 25: associated with the workshop; Pam Farris for 241-256. typing the manuscript; and Coral McCallister DUFFER, W. R., AND T. C. DORRIS, 1966. Primary pro- for illustrations. This paper is based in part on ductivity in a southern Great Plains stream. Lim- research under NSF grants BSR-8307014 and nology and Oceanography 11:143-151. BSR-8706046. ECHELLE, A. A., A. F. ECHELLE, AND L. G. HILL. 1972. Interspecific interactions and limiting factors of Literature Cited abundance and distribution in the Red River pupfish, Cyprinodon rubrofluviatilis. American BLACK, J. D. 1954. Biological conservation-with Midland Naturalist 88:109-130. particular emphasis on wildlife. McGraw-Hill, FAUSCH, K. D. 1984. Profitable stream positions for New York. salmonids: relating specific growth rate to net Barr, T. L., J. T. BROCK, C. S. DUNN, R. J. NAIMAN, R. energy gain. Canadian Journal of Zoology 62: W. OVINK, AND R. C. PETERSEN. 1985. Benthic 441-451. metabolism in four temperate stream systems: an FAUSCH, K. D., AND R. J. WkirrE. 1981. Competition interbiome comparison and evaluation of the riv- between brook trout (Salvelinus fontinalis) and er continuum concept. Hydrobiologia 123:3-45. brown trout (Salmo trutta) for positions in a Mich- BRETSCHKO, G., AND W. E. KLEMENS. 1986. Quanti- igan stream. Canadian Journal of Fisheries and tative methods and aspects in the study of the Aquatic Sciences 38:1220-1227. interstitial fauna of running waters. Stygologia FOREMAN, G. 1937. Adventure on Red River-report 2:297-316. on the exploration of the Red River by Captain BROWN, A. V., AND J. P. RICKER. 1982. Macroinver- Randolph B. Marcy and Captain G. B. McClellan. tebrate utilization of leaf detritus in a riffle of the University of Oklahoma Press, Norman. Illinois River, Arkansas. Arkansas Academy of FOREMAN, G. 1941. A pathfinder in the Southwest- Science Proceedings 26:10-13. the itinerary of Lieutenant A. W. Whipple during BURKHOLDER-CRECCO, J. M., AND R. W. BACHMANN. his explorations for a railway route from Fort 1988] NORTH AMERICAN PRAIRIE STREAMS 407

Smith to Los Angeles in the years 1853-1854. HILL, B. H., AND T. J. GARDNER. 1987a. Benthic me- University of Oklahoma Press, Norman. tabolism in a perennial and an intermittent Texas GELROTH, J. V., AND G. R. MARZOLF. 1978. Primary prairie stream. Southwestern Naturalist 32:305- production and leaf-litter decomposition in nat- 311. ural and channelized portions of a Kansas stream. HILL, B. H., AND T. J. GARDNER. 1987b. Seston dy- American Midland Naturalist 99:238-243. namics of two Texas prairie streams. American GELWICK, F. P. 1987. Longitudinal and temporal pat- Midland Naturalist 118:85-93. terns of riffle and pool fish assemblages in an HILL, B. H., T. J. GARDNER, AND 0. F. EIUSOLA. 1988. Ozark stream, Delaware County, Oklahoma. M.S. Breakdown of gallery forest leaf litter in inter- Thesis, University of Tulsa, Tulsa, Oklahoma. mittent and perennial Texas prairie streams. GILLIAM, F. S., T. R. SEASTEDT, AND A. K. KNAPP. 1987. Southwestern Naturalist 33:323-331. Canopy rainfall interception and throughfall in HOFFMAN, C. E., AND R. V. KELAmm. 1971. Environ- burned and unburned tallgrass prairie. South- mental changes produced by cold-water outlets western Naturalist 32:267-271. from three Arkansas reservoirs. Water Resources GORE, J. A. 1977. Reservoir manipulations and ben- Center Publication Number 5, Fayetteville, Ar- thic macroinvertebrates in a prairie river. Hy- kansas. drobiologia 55:113-123. HOOICER, K. L., AND G. R. MARZOLF. 1987. Differential GORE, J. A. 1980. Ordinational analysis of benthic decomposition of leaves in grassland and gallery communities upstream and downstream of a prai- forest reaches of Kings Creek. Transactions of the rie reservoir. Hydrobiologia 69:33-44. Kansas Academy of Science 90:17-24. GORE, J. A. 1982. Benthic invertebrate colonization: HUBBS, C., AND W. F. HETTLER. 1958. Fluctuations of source distance effects on community composi- some central Texas fish populations. Southwest- tion. Hydrobiologia 94:183-193. ern Naturalist 3:13-16. GORE, J. A., AND R. M. BRYANT. 1986. Changes in HUNT, C. B. 1974. Natural regions of the United fish and benthic macroinvertebrate assemblages States and Canada. W. H. Freeman, San Francisco. along the impounded Arkansas River. Journal of JOHNSON, L. S., D. L. WICHERS, T. A. WESCHE, AND J. Freshwater Ecology 3:333-345. A. GORE. 1982. Instream salmonid habitat ex- GORE, J. A., AND R. D. JUDY. 1981. Predictive models clusion by icecover. Wyoming Water Resources of benthic macroinvertebrate density for use in Research Institute, Water Research Series Num- instream flow studies and regulated flow man- ber 84. agement. Canadian Journal of Fisheries and KING, T. L., E. G. ZIMMERMAN, AND T. L. BEITINGER. Aquatic Sciences 38:1363-1370. 1985. Concordant variation in thermal tolerance GURTZ, M. E., G. R. MARZOLF, K. T. KILLINGBECK, D. L. and allozymes of the red shiner, Notropis lutrensis, SMITH, AND J. V. MCARTHUR. 1982. Organic mat- inhabiting tailwater sections of the Brazos River, ter loading and processing in a pristine stream Texas. Environmental Biology of Fishes 13:49-57. draining a tallgrass prairie riparian forest wa- LEOPOLD, A. 1949. A sand county almanac. Oxford tershed. Report to Kansas Water Resources Re- University Press, New York. search Institute, Manhattan. LILLEHAMMER, A., AND S. J. SALTVEIT (editors). 1984. HARVEY, B. C. 1987. Larval stream fish mortality and Regulated rivers. Universitetsforlaget As, Oslo. multi-trophic level interactions among stream MATTHEWS, W. J. 1985. Distribution of midwestern fishes. Ph.D. Thesis, University of Oklahoma, fishes on multivariate environmental gradients, Norman. with emphasis upon Notropis lutrensis. American HARVEY, B. C. 1988. Susceptibility of young-of-the- Midland Naturalist 113:225-237. year fishes to downstream displacement by flood- MArniEws, W. J. 1987. Physicochemical tolerance ing. Transactions of the American Fisheries So- and selectivity of stream fishes as related to their ciety 116:851-855. geographic ranges and local distributions. Pages HARVEY, B. C., R. C. CASHNER, AND W. J. MATTHEWS. 111-120 in W. J. Matthews and D. C. Heins (ed- 1988. Differential effects of largemouth and itors). Community and evolutionary ecology of smallmouth bass on habitat use by stoneroller North American stream fishes. University of minnows in stream pools. Journal of Fish Biology Oklahoma Press, Norman. 33:481-487. MATTHEWS, W. J., AND L. G. HILL. 1980. Habitat par- HENRY, B. C. 1986. Mayflies (Ephemeroptera) of the titioning in the fish community of a southwest- Concho River, Texas. Southwestern Naturalist 31: ern river. Southwestern Naturalist 25:51-66. 15-21. MATTHEWS, W. J., J. J. HOOVER, AND W. B. MILSTEAD. HILL, B. H. 1988. Organic matter dynamics in prairie 1983. The biota of Oklahoma springs: natural streams: model predictions and comparisons with biological monitoring of ground water quality. forested streams. Proceedings 10th North Amer- Technical Completion Report. Oklahoma Water ican Prairie Conference (in press). Resources Research Institute, Stillwater. 408 W. J. MATTHEWS [Volume 7

MATTHEWS, W. J., AND J. MANESS. 1979. Critical ther- ture. Pages 303-323 in T. D. Fontaine and S. M. mal maxima, oxygen tolerances, and population Bartell (editors). Dynamics of lotic ecosystems. Ann fluctuations in southwestern stream fishes. Arbor Science Publishers, Ann Arbor, Michigan. American Midland Naturalist 102:374-377. PFIESTER, L. A., R. LYNCH, AND T. L. WRIGHT. 1979. MATTHEWS, W. J., A. J. STEWART, AND M. E. POWER. Species composition and diversity of periphyton 1987. Grazing fishes as components of North in the Grand River Dam area, Mayes County, American stream ecosystems. Pages 128-135 in Oklahoma. Southwestern Naturalist 24:149-164. W. J. Matthews and D. C. Heins (editors). Com- PFLIEGER, W. L., AND T. B. GRACE. 1987. Changes in munity and evolutionary ecology of North Amer- the fish fauna of the lower Missouri River, 1940- ican stream fishes. University of Oklahoma Press, 1983. Pages 166-177 in W. J. Matthews and D. C. Norman. Heins (editors). Community and evolutionary MATTHEWS, W. J., E. SURAT, AND L. G. HILL. 1982. ecology of North American stream fishes, Uni- Heat death of Etheostoma spectabile (orangethroat versity of Oklahoma Press, Norman. darter) in a as related to Pico, J. 1987. Survey of fishes in the Oklahoma pan- their critical thermal maxima. Southwestern Nat- handle and Harper county, Northwestern Okla- uralist 27:216-217. homa. Proceedings of the Oklahoma Academy of MCARTHUR, J. V., G. R. MARZOLF, AND J. E. URBAN. Science 67:45-59. 1985. Response of bacteria isolated from a pris- POWER, M. E. AND W. J. MATTHEWS. 1983. Algae- tine stream to concentration and source of solu- grazing minnows (Campostoma anomalum), pis- ble organic carbon. Applied Environmental Mi- civorous bass ( Micropterus spp.), and the distri- crobiology 49:238-241. bution of attached algae in a small prairie-margin McLAY, C. 1970. A theory concerning the distance stream. Oecologia 60:328-332. travelled by animals entering the drift of a stream. POWER, M. E., W. J. MATTHEWS, AND A. J. STEWART. Journal of the Fisheries Research Board of Canada 1985. Grazing minnows, piscivorous bass and 27:359-370. stream algae: dynamics of a strong interaction. MENZEL, B. W., J. B. BARNUM, AND L. M. ANTOSCH. Ecology 66:1448-1456. 1984. Ecological alterations of Iowa prairie-ag- POWER, M. E., AND A. J. STEWART. 1987. Disturbance ricultural streams. Iowa State Journal of Research and recovery of an algal assemblage following 59:5-30. flooding in an Oklahoma stream. American Mid- METCALF, A. L. 1966. Fishes of the Kansas River sys- land Naturalist 117:333-345. tem in relation to zoogeography of the Great REPSYS, A. J., AND G. D. ROGERS. 1982. Zooplankton Plains. University of Kansas Publications, Mu- studies in the channelized Missouri River. Pages seum of Naturalist History 17:23-189. 125-145 in L. W. Hesse, G. L. Hergenrader, H. W. MOORE, G. A. 1944. Notes on the early life history Lewis, S. D. Reetz, and A. B. Schlesinger (editors). of Notropis girardi. Copeia 1944:209-214. The middle Missouri River. The Missouri River Moms, K. W., AND M. P. MADDEN. 1978. Benthic Study Group, Norfork, Nebraska. macroinvertebrate communities aid water qual- RESH, V. H., A. V. BROWN, A. P. Covicx, M. E. GURTZ, ity evaluation of the Washita River. Proceedings H. W. Li, G. W. MINSHALL, S. R. REICE, A. L. of the Oklahoma Academy of Science 58:93-97. SHELDON, J. B. WALLACE, AND R. WISSMAR. 1988. MOWER, N. A., AND D. RUSSEL (editors). 1972. The The role of disturbance theory in stream ecology. plains-being no less than a collection of vera- Journal of the North American Benthological So- cious memoranda taken during the expedition of ciety 7:433-455. exploration in the year 1845, from the western ROBISON, H. W. 1986. Zoogeographic implications settlements of Missouri to the Mexican border, of the Mississippi River basin. Pages 267-285 in and from Bent's Fort on the Arkansas to Fort C. H. Hocutt and E. 0. Wiley (editors). The zoo- Gibson, via South Fork of Canadian-North Mex- geography of North American freshwater fishes. ico and North-western Texas, by Francois des John Wiley and Sons, New York. Montainges. University of Oklahoma Press, Nor- ROEDER, D. R. 1977. Relationships between phyto- man. plankton and periphyton communities in a cen- NAIMAN, R. J., J. M. MELILLO, M. A. LOCK, T. E. FORD, tral Iowa stream. Hydrobiologia 56:145-151. AND S. E. REICE. 1987. Longitudinal patterns of Ross, S. T., W. J. MATTHEWS, AND A. A. ECHELLE. 1985. ecosystem processes and community structure in Persistence of stream fish assemblages: effects of a subarctic river continuum. Ecology 68:1139- environmental change. American Naturalist 126: 1156. 24-40. NEEL, J. K. 1985. A northern prairie stream. Uni- SLIZESKI, J. J., J. L. ANDERSON, AND W. G. DOROUGH. versity of North Dakota Press, Grand Forks. 1982. Hydrologic setting, system operation, PECKARSKY, B. L. 1983. Biotic interactions or abiotic present and future stresses. Pages 15-37 in L. W. limitations? A model of lotic community struc- Hesse, G. L. Hergenrader, H. S. Lewis, S. D. Reetz, 1988] NORTH AMERICAN PRAIRIE STREAMS 409

and A. B. Schlesinger (editors). The middle Mis- mass loss, nutrients, and invertebrates associated souri River. Missouri River Study Group, Nor- with elm leaf litter decomposition in perennial fork, Nebraska. and intermittent reaches of tallgrass prairie SMITH, D. L., AND D. A. DsTiEE. 1981. Recovery of streams. Southwestern Naturalist 31:511-520. a benthic macroinvertebrate community follow- TRUMP, C. L., AND W. C. LEGGETT. 1980. Optimum ing a toxic chemical discharge in a sandy plains swimming speeds in fish: the problem of cur- stream. Southwestern Naturalist 25:547-551. rents. Canadian Journal of Fisheries and Aquatic STATZNER, B. 1981. The relation between "hydraulic Sciences 37:1086-1092. stress" and microdistribution of benthic inver- U.S. DEPARTMENT OF AGIUCULTURE. 1941. Climate and tebrates in a lowland running water system, the man. U.S. Government Printing Office, Washing- Schierenseebrooks (North Germany). Archiv fiir ton, D.C. Hydrobiologie 91:192-218. VANNOTE, R. L., G. W. Mimi-LAIL, K. W. Cummilvs, J. STATZNER, B. 1987. Characteristics of lotic ecosys- R. SEDALL, AND C. E. CUSHING. 1980. The river tems and consequences for future research di- continuum concept. Canadian Journal of Fish- rections. Pages 365-390 in E.-D. Schulze and H. eries and Aquatic Sciences 37:130-137. Zwolfer (editors). Ecological Studies 61. Spring- WARD, J. V., AND J. A. STANFORD (editors). 1979. The er-Verlag, Berlin. ecology of regulated streams. Plenum Press, New STATZNER, B., J. A. GORE, AND V. H. RESH. 1988. Hy- York. draulic stream ecology: observed patterns and WHITMAN, R. L. 1979. Ecology of the sand dwelling potential applications. Journal of the North community of an east Texas stream. Ph.D. Thesis, American Benthological Society 7:307-360. Texas A&M University, College Station. STATZNER, B., AND B. HIGLER. 1985. Questions and WHITMAN, R. L., AND W. J. CLARK. 1982. Availability comments on the river continuum concept. Ca- of dissolved oxygen in interstitial waters of a nadian Journal of Fisheries and Aquatic Sciences sandy creek. Hydrobiologia 92:651-658. 42:1038-1044. Williams, D. D., AND H. B. N. HYNES. 1974. The STATZNER, B., AND B. HIGLER. 1986. Stream hydrau- occurrence of benthos deep in the substratum of lics as a major determinant of invertebrate zo- a stream. 4:233-251. nation patterns. Freshwater Biology 16:127-139. ZAREr, T. M., AND A. S. RAND. 1971. Competition in STEWART, A. J. 1987. Responses of stream algae to tropical stream fishes: support for the competi- grazing minnows and nutrients: a field test for tive exclusion principle. Ecology 52:336-342. interactions. Oecologia 72:1-7. ZIMMERMAN, E. G., AND C. M. RICHMOND. 1981. In- ST'EWART, A. J. 1988. Alkalinity dynamics in a hard- creased heterozygosity at the MDH-B locus in water prairie-margin stream. Archiv fiir Hydro- fish inhabiting a rapidly fluctuating thermal en- biologie 112:335-350. vironment. Transactions of the American Fish- TATE, C. M., AND M. E. GURTZ. 1986. Comparison of eries Society 110:410-416.