David S. Leigh

Habitat-specific responses of stream insects to land cover disturbance: biological consequences and monitoring implications

Changes in catchment land cover can impact stream ecosystems through altered hydrology and subsequent increases in sedimentation and nonpoint-source pollutants. These stressors can affect habitat suitability and water quality for aquatic invertebrates. We studied the impact of a range of physical and chemical stressors on aquatic insects, and tested whether the effects of these stressors differed in 3 habitat types: riffles, pools, and banks. Our study was conducted in Piedmont streams in Georgia (USA) where catchment development pressure and the potential for aquatic biodiversity loss are high. We sampled 3 replicates of riffle, pool, and bank habitats within a 100-m reach of 29 streams (11–126 km2) that varied in catchment land cover. Correlations between environmental variables and aquatic insects (both richness and density) within habitat types indicated that riffle habitats (vs pool and bank habitats) exhibited the strongest relations with environmental variables. Riffle assemblages were negatively affected by both physical (e.g., bed mobility) and chemical (e.g., specific conductance, nutrient concentrations) variables. The density of aquatic insects in pools was also correlated to physical and chemical variables, but there were few relationships with pool or bank richness or bank density. Because of greater relative impacts of disturbance in riffles versus banks, we found greater differences between riffle and bank richness in streams with greater sediment disturbance. The proportion of bank richness (bank richness/bank + riffle richness) increased with finer bed sediment (r2 = 0.43) and increased bed mobility (r2 = 0.35). We compared richness of facultative taxa (found in multiple habitats) between sites we characterized as minimally impacted and sediment-impacted. In riffles, richness of facultative taxa was lower in sediment-impacted vs minimally impacted sites (11.0 vs 20.2, p = 0.002, t-test), but was similar for both disturbance groups in banks (20.1 vs 22.7, p > 0.05, t-test). Our results suggest that taxa richness may be retained in bank habitats when riffle quality is poor and banks may serve as a refuge in highly disturbed systems. Such shifts in the distribution of benthos may be an early warning indicator of biotic impairment and have implications for biomonitoring and maintenance of habitat.

Paleochannels indicating wet climate and lack of response to lower sea level, southeast Georgia

Large meandering paleochannels on the flood plain of the middle Ogeechee River in southeast Georgia represent wet paleoclimate during late Pleistocene and early Holocene time and show no indication of downcutting in response to eustatically lower sea level. Radiocarbon dates indicate that the paleomeanders were active at ∼31–28 ka and ∼8.5–4.5 ka. Bivariate regression models that correlate modern channel dimensions to the discharge of low-magnitude floods (1–5 yr recurrence interval) indicate that paleodischarge of floods during those times was at least double that of modern floods, thus suggesting a wetter paleoclimate. These data corroborate independent studies of pollen and paleoclimate simulations that indicate wet early to middle Holocene (9–3 ka) conditions that were characterized by intensified monsoonal circulation. Paleoclimate conditions at ∼31–28 ka are less well known. Our analysis of the middle Ogeechee River flood plain indicates the absence of base-level response (downcutting) to eustatic sea-level lowering because the beds of the paleochannels are at approximately the same elevation as the bed of the modern channel. This supports recent arguments that the geomorphic response of coastal-plain streams to sea-level lowering is most apparent in deltaic and shelf environments and may not be recognized very far upstream from the coast.

Roxana silt of the Upper Mississippi Valley: Lithology, source, and paleoenvironment

Significant lithological differences exist between the middle Wisconsinan Roxana Silt (loess) of the Upper Mississippi Valley (the valley north of 41.5°) and that of the type area near St. Louis, Missouri. Primary differences are that the Upper Mississippi Valley Roxana Silt is much thinner, finer textured, and more weathered than that of the type area. Extrapolation of radiocarbon dates brackets sedimentation of Upper Mississippi Valley Roxana Silt between 55 and 27 ka, which is in close agreement with dates from the type area. Radiocarbon dates and paleobotanical data indicate that the Upper Mississippi Valley Roxana Silt accumulated at a maximum long-term-average sedimentation rate of about 7 cm/1,000 yr within a boreal forest environment, whereas the type area Roxana Silt accumulated at a maximum long-term-average sedimentation rate of about 60 cm/1,000 yr within a prairie-like environment. Multiple regression analysis shows that more than 75% of the variance in thickness and texture between the Upper Mississippi Valley Roxana Silt and that of the type area is explained by parameters of source-valley width, source-valley azimuth, and distance of the depositional site from the source-valley. This indicates that although the lithologic characteristics and sedimentation rates between the two regions are quite different, they are consistent with a unified loess sedimentation sys- tem linked to the ancient Mississippi River valley. A proglacial source for the Roxana Silt and a middle Wisconsinan advance of the Lauren-tide Ice Sheet is a topic of scientific debate because recent findings show a lack of Roxana-correlative till deposits in the Midwest. Alternative hypotheses include a glaciolacustrine source, a regional slope erosion source, and a cold-climate desert dust source. Tests of these hypotheses support a proglacial origin for the Roxana Silt. Multiple regression analysis of loess thickness and texture, along with paleobotanical evidence, indicates that the Roxana Silt was blown from the ancient Mississippi Valley, rather than from a cold-climate desert source or from flood plains linked to a glaciolacustrine source. Geochemical and mineralogical composition of the Upper Mississippi Valley Roxana Silt, compared to out-wash-derived Peoria Loess, indicates that detritus from regional hillslope erosion is not a unique component of the Roxana Silt. Combined data indicate that the midcontinental Roxana Silt was derived as loess blown from flood plains of middle Wisconsinan proglacial rivers between 55 and 27 ka.

Evaluating artifact burial by eolian versus bioturbation processes, South Carolina Sandhills, USA

Artifacts are commonly buried by approximately 50 cm of sediment at prehistoric archeological sites (early Archaic through Mississippian) on uplands of the Sandhills of the upper Coastal Plain of the southeastern United States. Bioturbation, eolian sedimentation, and colluviation are the primary processes that can explain artifact burial because of the upland position of the sites in an erosional landscape setting. Colluvial sedimentation is discounted at most of the sites because they occur on interstream divides and upper hillslope positions. Thus, the focus is on eolian sedimentation versus bioturbation as burial agents. Six sites in the midst of the Sandhills region along the corridor of South Carolina Highway 151 in Chesterfield County provide the data. The Sandhills consist primarily of Cretaceous and Tertiary marine, fluvial, and eolian sediments that are highly dissected and overlie crystalline rocks in the deep subsurface. Two of the sites are on high fluvial terrace remnants that predate 12 ka and serve as controls where bioturbation is the only reasonable burial process. Hillslope positions of the sites are on erosional elements of the landscape (crests, shoulder slopes, and upper backslopes) where sediment transfer operates (colluvial and overland flow), but where deposition is minimal. The sites occur on very sandy soils having a texture of loamy sand to sand. In some instances, a fine textured cover sand, which is about 1.5 m thick, overlies a clayey subsoil or Bt horizon. This cover sand has been interpreted by some as an eolian sand sheet that buries a second parent material and paleosol, but standard particle size and heavy mineral data indicate that it is simply a thick E horizon over a Bt horizon. Standard particle size fractionation at whole phi intervals, and particle size analysis of the heavy mineral fraction, indicate that eolian sedimentation is unlikely at five of the six sites. Heavy minerals were analyzed with respect to the sedimentological principle of hydraulic equivalence, which provides clear separation of eolian versus water-laid sediment. Results of particle size analysis suggest that the cover sands are water-laid (probably fluvial) at five of the six sites, which favors the bioturbation process of artifact burial. Heavy mineral analysis corroborates the standard particle size data, indicating that only one site, 38CT16, possibly is composed of eolian sediment. Soil profile development suggests that the age of the sediment at site 38CT16 probably is older than 12 ka and was in place prior to human occupation. Therefore, possible eolian sedimentation at that site is not relevant to artifact burial, which also suggests bioturbation is the primary process of artifact burial. Additional evidence favoring bioturbation as a vigorous artifact burial process in the Sandhills comes from the two sites on high elevation sandy fluvial terraces (38CT34, 38CT17) where artifacts are also buried. At these terraced sites bioturbation is the only possible burial process. Overall results suggest that bioturbation best explains the occurrence of buried artifacts and that eolian sedimentation processes are not readily apparent, and are not required, in explaining artifact burial.

Soil chronosequence of brasstown creek, Blue Ridge mountains, USA

Abstract A chronosequence of soils on alluvial terraces in the Appalachian highlands of the southeastern United States is identified by simple methods of quantitative soil morphology and chemical analysis. The chronosequence is from Brasstown Creek, a 5th order stream draining 65 km 2 of the Blue Ridge Mountains of north Georgia. The portion of the valley studied is about 1 km wide and contains five Quaternary terrace remnants. Radiocarbon dates show that the floodplain (T0) is late Holocene and includes historical (post ca. 1830 A.D.) sediment, and the four older terraces (T1–T5) are pre-Holocene. The floodplain soils are either Entisols in historical sediment or buried Inceptisols with weak Bw horizons in late Holocene to protohistoric sediment. The lowest terrace (T1) is terminal Pleistocene in age, and its soils have Bt horizons that are about 0.9 m thick, 10YR 4/5 to 10YR 5/4 in colour, and contain 25 to 35% clay. Soils on terraces older than T1 become progressively redder and contain more clay with increasing age, culminating with 2.5YR hues and 40 to 52% clay in the Bt horizon of T5. Percent clay in Bt horizons, rubification indices, iron oxide contents, and element ratios progressively change with age of the terraces, but other morphological properties, pH, and individual element concentrations are of less use as chronosequence variables.

Morphology and Channel Evolution of Small Streams in the Southern Blue Ridge Mountains of Western North Carolina

Small streams are understudied in the Southern Blue Ridge Mountains, yet they constitute a huge portion of the drainage network and are relevant with respect to human impact on the landscape and stream restoration efforts. Morphologies of 44 streams (0.01 to 20 km2 watersheds) from western North Carolina are characterized and couched in the context of historical channel evolution and human impacts. Topographic cross-sections and longitudinal gradients, channel and floodplain widths, and bed particle sizes are the basic data. Regression equations describe channel form. New optically stimulated luminescence, radiocarbon, and cesium-137 dates resolve channel evolution. Results indicate that channels behave in a predictable linear fashion of changing in response to increasing watershed size. However, forested reaches have much wider channels than pastured/grassland reaches, indicating more than 50 percent loss of instream habitat related to conversion of riparian forest to pasture/grassland in some cases. Floodplain widths conservatively indicate long-term (century-scale) lateral erosion rates of about 0.5 to 5.0 cm/yr. Colluvial inputs are probable drivers of channel form and particle size on the smallest headwater streams. The modern floodplain appears to have established itself in the early 1900s and it is still evolving.

Beyond the urban gradient: barriers and opportunities for timely studies of urbanization effects on aquatic ecosystems

Abstract Many studies have shown that streams degrade in response to urbanization in the watershed. These studies often are based on use of biotic and abiotic variables to measure stream health across a gradient of land cover/land use. The results of these studies can be applied to other urban systems, but often fail to provide a mechanistic understanding of the urban impact, in part, because of the nature of the experimental design. We analyzed the advantages and disadvantages of using environmental gradient studies to further understanding of urban stream systems. We also evaluated alternative experimental design approaches, including best management practice monitoring, long-term watershed studies, paired-watershed studies, and before–after control–impact studies, which could be used to complement the gradient approach. We illustrate these theoretical discussions with an urban paired-watershed case study in the Etowah watershed in northern Georgia. Our goal is to move experimental designs in a direction that will further our mechanistic understanding of the effects of existing urbanization on aquatic ecosystems and will provide opportunities to evaluate stream responses to environmentally sensitive urban land cover.

Late-Holocene paleofloods in the Upper Little Tennessee River valley, Southern Blue Ridge Mountains, USA

We derive a paleoflood chronology for the past 2000 years from three stratigraphic sections of overbank sediments with dates from radiocarbon, luminescence, 137Cs techniques, and historical records. Particle sizes were measured in 6–15 year intervals in post-1870 sediments and in 45–170 year intervals in pre-1870 sediments using an automatic laser analyzer. The sedimentological characteristics of ad 1948–2009 deposits were compared with gaging records, demonstrating that fine sand content and sorting discern time intervals of large floods, but flood magnitudes are not well resolved. This modern analog was applied to pre-1870 sediments and revealed two periods in the last 2000 years with large floods during ad 650–850 and ad 1100–1350, which are times when the regional tree-ring record showed extreme wetness and no severe or extreme droughts. Our findings indicate flood-prone phases of transitional climate at the beginning and end of the ‘Medieval Warm Period’ (MWP), and relatively subdued flooding during the ‘Little Ice Age’ (LIA), possibly correlated with rearrangement of macro-scale atmospheric circulation patterns between the MWP and the LIA.

Multi-millennial record of erosion and fires in the southern Blue Ridge Mountains, USA In: Greenberg, CH and BS Collins (eds.)

Bottomland sediments from the southern Blue Ridge Mountains provide a coarse-resolution, multi-millennial stratigraphic record of past regional forest disturbance (soil erosion). This record is represented by 12 separate vertical accretion stratigraphic profiles that have been dated by radiocarbon, luminescence, cesium-137, and correlation methods continuously spanning the past 3,000 years of pre-settlement (pre-dating widespread European American settlement) and post-settlement strata. Post-settlement vertical accretion began in the late 1800s, appears to be about an order of magnitude faster than pre-settlement rates, and is attributable to widespread deforestation for timber harvest, farming, housing development, and other erosive activities of people. Natural, climate-driven, or non-anthropic forest disturbance is subtle and difficult to recognize in pre-settlement deposits. There is no indication that pre-settlement Mississippian and Cherokee agricultural activities accelerated erosion and sedimentation in the region. A continuous 11,244 years before present (BP) vertical accretion record from a meander scar in the Upper Little Tennessee River valley indicates abundant charcoal (prevalent fires) at the very beginning of the Holocene (11,244–10,900 years BP). In contrast, moderate to very low levels of charcoal are apparent over the remaining Holocene until about 2,400 years BP when charcoal influx registers a pronounced increase. These data are consistent with the idea that Native Americans used fire extensively to manage forests and to expanded agricultural activities during Woodland and later cultural periods over the past 3000 years. However, there is no indication that prehistoric intentional use of fire and agriculture caused accelerated erosion and sedimentation.

Modeling stream-bank erosion in the Southern Blue Ridge Mountains

Deforestation, followed by soil erosion and subsequent deposition of alluvium in valleys, played a critical role in the formation of historical terraces in much of the Southern Blue Ridge Mountains. Such terraces add a significant amount of sediment to the tributaries of the region as streams laterally erode the terrace banks. This study examined the contribution of total sediment yield derived solely from eroded stream banks in small watersheds (<20 km²), using floodplain widths as proxies for long-term lateral erosion rates. The raw data were derived from watersheds with different land covers (Coweeta Creek and Skeenah Creek watersheds in the Upper Little Tennessee River basin). Bank-derived sediment yield estimates were modeled in a Geographic Information System, using linear regression to relate floodplain widths and erodible terrace bank heights. We found total stream length to be a good predictor of both lateral erosion rates and erodible bank heights. Land cover, basin/network morphometrics, and reach-scale stream conditions were not good predictors. Modeled lateral migration and sediment yield results compare favorably to empirical measurements from five independent watersheds in the region. Modeled estimates fall within ±50% or better of the observed values, at 16.33 to 25.02 t km−2 yr−1.