Variable Hydrological Conditions Research Articles

Continuous recruitment underpins fish persistence in the arid rivers of far-western Queensland, Australia

Fish living in highly variable and unpredictable environments need to possess life-history strategies that enable them to survive environmental extremes such as floods and drought. We used the length–frequency distributions of multiple fish species in multiple seasons and highly variable hydrological conditions to infer antecedent breeding behaviour in rivers of far-western Queensland, Australia. Hypotheses tested were as follows: (1) recruitment of some or all species of fish would occur within waterholes during no-flow periods; (2) there would be seasonal recruitment responses in some fish species; (3) recruitment of some species would be enhanced by channel flows and/or flooding. Hydrology and the incidence of flooding were highly variable across the study area during 2006–2008. Flood-influenced recruitment was evident for Hyrtl's tandan, Barcoo grunter and Welch's grunter. Silver tandan, golden goby, Cooper Creek catfish and Australian smelt showed evidence of seasonal recruitment unrelated to antecedent hydrology. However, most species demonstrated continual recruitment in isolated waterholes, irrespective of antecedent flow conditions and season. Continual and seasonal recruitment capabilities have obvious advantages over flood-pulse recruitment in rivers with highly unpredictable flood regimes and underpin the persistence of many fish species in arid and semiarid rivers.

National scale assessment of climate change impacts on flooding in Finland

This paper provides a general overview of changes in flooding caused by climate change in Finland for the periods 2010-2039 and 2070-2099. Changes in flooding were evaluated at 67 sites in Finland with variable sizes of runoff areas using a conceptual hydrological model and 20 climate scenarios from both global and regional climate models with the delta change approach. Floods with a 100-year return period were estimated with frequency analysis using the Gumbel distribution. At four study sites depicting different watershed types and hydrology, the inundation areas of the 100-year floods were simulated with a 2D hydraulic model. The results demonstrate that the impacts of climate change are not uniform within Finland due to regional differences in climatic conditions and watershed properties. In snowmelt-flood dominated areas, annual floods decreased or remained unchanged due to decreasing snow accumulation. On the other hand, increased precipitation resulted in growing floods in major central lakes and their outflow rivers. The changes in flood inundation did not linearly follow the changes in 100-year discharges, due to varying characteristics of river channels and floodplains. The results highlight the importance of comprehensive climatological and hydrological knowledge and the use of several climate scenarios in estimation of climate change impacts on flooding. Generalisations based on only a few case studies, or large scale flood assessments using only a few climate scenarios should be avoided in countries with variable hydrological conditions.

Stochastic modeling of nutrient losses in streams: Interactions of climatic, hydrologic, and biogeochemical controls

We present an analytical, stochastic approach for quantifying intra‐annual fluctuations of in‐stream nutrient losses induced by naturally variable hydrologic conditions. The relevance of the problem we address lies in the growing concern for the major environmental impacts of increasing nutrient loads from watersheds to freshwater bodies and coastal waters. Here we express the first‐order nutrient loss rate constant, ke, as a function of key biogeochemical and hydrologic controls, in particular the stream depth (h). The stage h modulates the impact of natural streamflow temporal fluctuations (induced by intermittent rainfall forcings) on the underlying biogeochemical processes and thus represents the major driver of at‐a‐site fluctuations of ke. Novel expressions for the probability distribution function (pdf) of h and ke are derived as a function of a few eco‐hydrologic, morphologic and biogeochemical parameters. The shape of such pdf's chiefly depends on the following attributes: (1) the average frequency of streamflow‐producing rainfall events, λ; (2) the inverse of mean catchment residence time, k; and (3) a stream channel shape factor, identified through the discharge rating curve exponent b. For λ/(kb) > 1, h and ke have lower intra‐annual variability and lower sensitivity to climatic and morphologic controls, leading to improved predictability and ease of measurement of these attributes. Moment analyses suggest that the variability of ke, relative to that of h, is attenuated for λ/(kb) > 1. Thus, the interplay between climate‐landscape parameters and the stream shape factor b controls the temporal variability induced by stochastic rainfall forcings on stream stages and nutrient removal rates.

A scale-dependent approach to study pollution control processes in wetland soils using three different techniques

The Elbe River, Germany, has received heavy metals and arsenic from the discharge of urban industrial, and agricultural effluent. During periods of inundation, these contaminants were transported with water into floodplain ecosystems, where they settled and accumulated predominantly in depressions and low-lying terraces. Markedly elevated arsenic concentration in soil solution during floods exceeded the inspection value of 10 μg L −1 of the German soil protection ordinance. Highly variable hydrological conditions in floodplains can affect the dynamics of pollutants. The study of processes controlling the dynamics of pollutants is challenging because the results are required to answer both scientific and practical questions regarding protection of groundwater and plants, sustainable management of floodplains or explain the fate of environmentally harmful substances. Our experiments in small groundwater lysimeter and biogeochemical microcosms tended to yield similar results regarding the functional relationships among the investigated site parameters. But the results of the field experiments, carried out at a floodplain site of the middle course of the Elbe River, Germany, are often characterized by complex and varying factors. Whereas arsenic tended to be mobilized during flooding due to decreasing redox potential ( E H), chromium showed the opposite trend, with peak concentrations at the highest E H values. Our approach at three different spatiotemporally scale levels, ranging from 23 days (microcosms) to two-and-a-half years (field soil hydrological facility) allows us to overcome process interferences observed in field studies.

Application of pseudorandom signals for underwater distance measurement on a shelf

In this paper, the results of experiments conducted on the Sea of Japan shelf are applied and discussed. The aim of these experiments was to study regularities in the formation of impulse characteristics for wave conductors under variable hydrological conditions. It has been shown that, in order to determine the distance and coordinates of underwater devices precisely when working on a shelf, one should identify the first and last impulses in impulse characteristics for the wave conductor, measure their arrival times, and multiply them to calculate the values of effective sonic velocity along the propagation routes of these impulses.

Effects of soil moisture and water depth on ERS SAR backscatter measurements from an Alaskan wetland complex

We conducted a preliminary investigation of the response of ERS C-band SAR backscatter to variations in soil moisture and surface inundation in wetlands of interior Alaska. Data were collected from 5 wetlands over a three-week period in 2007. Results showed a positive correlation between backscatter and soil moisture in sites dominated by herbaceous vegetation cover ( r = 0.74, p < 0.04). ERS SAR backscatter was negatively correlated to water depth in all open (non-forested) wetlands when water table levels were more than 6 cm above the wetland surface ( r = − 0.82, p < 0.001). There was no relationship between backscatter and soil moisture in the forested (black spruce-dominated) wetland site. Our preliminary results show that ERS SAR data can be used to monitor variations in hydrologic conditions in high northern latitude wetlands (including peatlands), particularly sites with sparse tree cover.

Role of abiotic and biotic factors in structuring the metazoan plankton community in a lowland river

AbstractWater and quantitative zooplankton samples were collected at approximately monthly intervals from June 2000 to June 2002 in the middle section of the Po River (Northern Italy). Highly variable hydrological conditions were observed in the 2 years, with river flow varying approximately between 400 and 9300 m3 s−1. Dissolved inorganic nitrogen (DIN) usually decreased at higher river discharges. Autotrophic production was limited by phosphorus availability during the growing period (May–October); no significant adverse effect of zooplankton grazing upon algal biomass was detected. A total of 161 zooplankton taxa were found. The records of the rotifers Lecane incospicua, L. cornuta, Proales sigmoidea, Cephalodella limosa, and of the cladoceran Phrixura leei were new for Italy. Rotifers strikingly were the most abundant group over the entire sampling period. Zooplankton abundance was positively correlated to water temperature and chlorophyll‐a concentration, and only in warmer months inversely correlated to river flow. Species of the genus Brachionus were responsible for the two density peaks (&gt;1000 ind L−1) of the rotifer taxocoenosis observed in July 2000 and August 2001, coinciding with chlorophyll‐a peaks higher than 50 µg L−1 and minima of Shannon equitability index. Water temperature, chlorophyll‐a, and soluble reactive phosphorus (SRP) were found to be the most important variables in explaining the observed rotifer assemblages by Canonical Correspondence Analysis. A simple descriptive model is proposed to represent the combined influence of abiotic constraints and biotic interactions on the zooplankton abundance and diversity in the middle reach of the Po River. Copyright © 2008 John Wiley &amp; Sons, Ltd.

Assessment of hydrological and seasonal controls over the nitrate flushing from a forested watershed using a data mining technique

Abstract. A data mining, regression tree algorithm M5 was used to review the role of mutual hydrological and seasonal settings which control the streamwater nitrate flushing during hydrological events within a forested watershed in the southwestern part of Slovenia, characterized by distinctive flushing, almost torrential hydrological regime. The basis for the research was an extensive dataset of continuous, high frequency measurements of seasonal meteorological conditions, watershed hydrological responses and streamwater nitrate concentrations. The dataset contained 16 recorded hydrographs occurring in different seasonal and hydrological conditions. Based on predefined regression tree pruning criteria, a comprehensible regression tree model was obtained in the sense of the domain knowledge, which was able to adequately describe most of the streamwater nitrate concentration variations (RMSE=1.02 mg/l-N; r=0.91). The attributes which were found to be the most descriptive in the sense of streamwater nitrate concentrations were the antecedent precipitation index (API) and air temperatures in the preceding periods. The model was most successful in describing streamwater concentrations in the range 1–4 mg/l-N, covering large proportion of the dataset. The model performance was little worse in the periods of high streamwater nitrate concentration peaks during the summer hydrographs (up to 7 mg/l-N) but poor during the autumn hydrograph (up to 14 mg/l-N) related to highly variable hydrological conditions, which would require a less robust regression tree model based on the extended dataset.

APPLICATION OF TREE-RING ISOTOPIC ANALYSES TO RECONSTRUCT HISTORICAL WATER USE OF RIPARIAN TREES

Historical patterns of water source use by trees inferred from long-term records of tree-ring stable isotopic content could assist in evaluating the impact of human alterations to natural stream flow regimes (e.g., water impoundments, stream flow diversions, and groundwater extraction). Our objective was to assess the utility of the hydrogen stable isotopic composition (SD) of tree rings as an index of historical water source use by riparian trees. We investigated the influence of site conditions that varied in climate and hydrology on the relationship between deltaD of Populus xylem water (deltaD(xyl)) and tree-ring cellulose (deltaD(cell)). deltaD(xyl) and deltaD(cell) were strongly correlated across sites (r2 = 0.89). However, the slope of this relationship was less than 1, indicating that factors other than deltaD(xyl) influenced deltaD(cell). Inverse modeling with an isotopic fractionation model for tree-ring cellulose suggested that the lack of one-to-one correspondence between deltaD(xyl) and deltaD(cell) was due to the influence of the hydrogen isotopic content of the atmospheric water vapor (deltaD(atm)). Empirically measured values of deltaD(cell) were typically within the seasonal range of deltaD(cell) predicted from the fractionation model. Sensitivity analyses showed that changes in deltaD(xyl) generally had a greater influence at high-elevation montane sites, whereas deltaD(xyl) and deltaD(atm) had about equal influence on deltaD(cell) at low-elevation desert sites. The intrasite relationship between deltaD(cell) and deltaD(xyl) among individual trees was poor, perhaps because of the within-site spatial variation in hydrologic conditions and associated tree physiological responses. Our study suggests that historical variation in deltaD(cell) of Populus provides information on historical variation in both time-integrated water source use and atmospheric conditions; and that the influence of atmospheric conditions is not consistent over sites with large differences in temperature and humidity. Reconstruction of xylem water sources of Populus in riparian ecosystems from deltaD(cell) will be more direct at higher elevation mountain sites than at low-elevation desert sites.

Quantifying the Hydroregime of a Temporary Pool Habitat: A Modelling Approach for Ephemeral Rock Pools in SE Botswana

Ecological and evolutionary processes in temporary rock pools operate within constraints imposed by their hydrologic regimes. These shallow pools flood when seasonal rains accumulate on impermeable substrates. Despite the ecological importance of hydrologic conditions for these ecosystems, we typically lack tools and empirical data required to understand the implications of hydrologic variability and climate change for biotic populations and communities in these habitats. In this study, we developed a hydrologic model to simulate rock pool hydrologic regimes based on rainfall, evapotranspiration, and basin geometry. The model was used to investigate long-term patterns of seasonal and inter-annual variation in hydroregime. In addition, hydrologic conditions associated with potential climate change scenarios were simulated and evaluated with respect to the biological requirements of the anostracan Branchipodopsis wolfi. The model’s output for daily inundation matched with field observations with an overall accuracy of 85% and correctly estimated complete hydroperiods with an overall accuracy of 70%. Simulations indicate large variation in individual hydroperiods (76–115%) as well as in the number of hydroperiods per year (19–23%). Furthermore, this study suggests that climate change may significantly alter the rock pool hydroregime. These findings confirm the hydrologic sensitivity of these ephemeral habitats to precipitation patterns, and their potential sensitivity to future climate change. Modelling indicates that the suitability of average inundation conditions for B. wolfi deteriorates significantly under future climate predictions. High levels of spatial and temporal variation in hydrologic conditions are dominant features of these habitats and an essential consideration for understanding population and community-level ecological processes.

Potential Roles of Fish, Birds, and Water in Swamp Privet (Forestiera acuminata) Seed Dispersal

Forestiera acuminata (swamp privet) is a common wetland shrub/small tree native to the southeastern United States. We examined several possible dispersal avenues for the plant. We tested germination of seeds exposed to various treatments, including passage through Ictalurus punctatus (Channel Catfish) guts, and conducted other tests and observations to infer seed-dispersal pathways. Channel Catfish consumed swamp privet drupes and defecated viable seeds, confirming that they are seed dispersers. Bombycilla cedrorum (Cedar Waxwings) ate the carbohydrate-rich drupes, and we predict that they disperse the seeds. We also inferred passive seed dispersal by water. Diverse dispersal pathways may allow for effective seed dispersal under a wide range of environmental conditions. Growing in wetlands and riparian areas, the plant experiences extreme annual variation in hydrologic conditions, which should influence the importance of the various dispersal pathways among years.

Factors influencing mineralizable carbon in a landscape with variable topography

Management-induced erosion has substantially increased the spatial variability in soil organic carbon (OC) stocks in landscapes with variable topography, but its impact on C dynamics is not well understood. The variability in cumulative C mineralization of samples from five positions in a landscape was examined in an incubation study and the effects of water content and depth within the A horizon were assessed. Mineralization was generally at a maximum at water contents between 80 and 95% water-filled pore space. The amount of C mineralized increased with OC content, but the proportion of the mineralizable OC decreased with increasing OC content. Few significant differences in C mineralization existed between surface and subsurface layers of the A horizon. The spatial patterns in maximum C mineralization were generally indicative of the patterns obtained when mineralization was calculated using spatial patterns of actual seasonal average water contents. This suggests the spatial pattern of maximum mineralization was not strongly influenced by variable hydrologic conditions on this site and reflects the spatial patterns in annual mineralization. The proportion of OC mineralized was not related to the proportion of the OC in particulate organic carbon (POC) or in the silt + clay fractions, but was significantly negatively related to the saturation of the clay + silt fraction with C. We speculate that the decline in the proportion of mineralized OC with increasing OC content reflects an increased proportion of the capacity of the silt + clay fraction to retain and physically protect OC within microaggregates. Key words: Erosion, deposition, mineralizable organic carbon, spatial variability

A mass‐balance approach to estimate in‐stream processes in a large river

AbstractA mass‐balance approach was used to estimate in‐stream processes related to inorganic nitrogen species (NH4+, NO2− and NO3−) in a large river characterized by highly variable hydrological conditions, the Garonne River (south‐west France). Studies were conducted in two consecutive reaches of 30 km located downstream of the Toulouse agglomeration (population 760 000, seventh order), impacted by modification of discharge regime and high nitrogen concentrations. The mass‐balance was calculated by two methods: the first is based on a variable residence time (VRT) simulated by a one‐dimensional (1‐D) hydraulic model; the second is a based on a calculation using constant residence time (CRT) evaluated according to hydrographic peaks. In the context of the study, removal of dissolved inorganic nitrogen (DIN) for a reach of 30 km is underestimated by 11% with the CRT method. In sub‐reaches, the discrepancy between the two methods led to a 50% overestimation of DIN removal in the upper reach (13 km) and a 43% underestimation in the lower reach (17 km) using the CRT method. The study highlights the importance of residence time determination when using modelling approaches in the assessment of whole stream processes in short‐duration mass‐balance for a large river under variable hydrological conditions. Copyright © 2007 John Wiley &amp; Sons, Ltd.

Annual and spatial variation for macroinvertebrates in the Upper Mississippi River near Cape Girardeau, Missouri

We sampled macroinvertebrates at four sites in the Open River section of the Upper Mississippi River near Cape Girardeau, Missouri (Rkm 106-114). In 1999, 2000, and 2002 - 2004, macroinvertebrates were sampled in the fine sediments downstream of four wing dikes with a Ponar dredge. In 1999 and 2001, macroinver- tebrates on rocks were sampled on the upstream face of those same wing dikes with rock baskets. We identified 68 taxa in the fine sediments and 50 taxa on the rocks, with only 27 taxa being found in both. In both habitats, annual variability in population densities and community structure exceeded spatial variability. Macroinvertebrate densities in the fine sediments ranged from approximately 3700 to 11,700 individuals/m 2 with oligochaetes com- prising 77- 95 % of total density. Non-metric Multidimensional Scaling ordination suggests that annual differ- ences in the fine sediment assemblage reflect variation in hydrological conditions (high spring flow and low an- nual flow). Macroinvertebrate density on the rocks ranged from approximately 57,800 to 163,000 individuals/m 2 with hydropsychid caddisflies (Hydropsyche bidens/orris and Potamyia flava) comprising 82 - 97 % of total den- sity. This dominance of oligochaetes in fine bottom substrates and caddisflies on hard substrates is consistent with many other studies on big rivers in the Mississippi River Basin (e. g., Upper and Lower Mississippi, Ohio, Il- linois, and Missouri Rivers), and may be the current reference condition for these big rivers. This makes tradi- tional water quality monitoring and assessment efforts using macroinvertebrates more difficult because increased dominance by more pollution tolerant taxa is perceived as evidence of water or habitat degradation. However, macroinvertebrate assemblages in our study reach seem less stressed than other big rivers known to be impaired. Specifically, densities of the pollution-sensitive mayfly Hexagenia indicate that conditions in our study reach have improved markedly since the 1960's.

Evolution de la méthode du gradex : approche par type de temps et modélisation hydrologique

Click to increase image sizeClick to decrease image sizeThe physico-statistical approach of the flood prediction by the GRADEX method seems very effective : it’s very straightforward and makes it possible to obtain completely coherent results by a robust and very readable step. To assess the quality of the approximation carried out by this model, some improvements have been tested on two significant points, leading to the SCHADEX method. SCHADEX is based on the study of the risk of precipitation related to each type of atmospheric circulation. This typology highlights the weather configuration of stronger rainfall risk and therefore increases the probabilities of extreme rains. The runoffs resulting from these precipitations, depending on highly variable hydrological conditions, are estimated by the MORDOR hydrological model. The results show a fast increasing of the rainfall efficiency going from common to rare floods. But this increase is smoother than with the initial GRADEX approach, which moderates the impact of heavier extreme rainfall.Valeurs rares et extrêmes de précipitations et de débits

Late Miocene development of the western Pacific warm pool: Planktonic foraminifer and oxygen isotopic evidence

The disappearance at ∼10 Ma of the deep dwelling planktonic foraminifer Globoquadrina dehiscens from the western Pacific including the South China Sea was about 3 Myr earlier than its final extinction elsewhere. Accompanying this event at ∼10 Ma was a series of faunal turnover characterized by increase in mixed layer, warm-water species and decrease to a minimum in deepwater species. Paleobiological and isotopic evidence indicates sea surface warming and a deepened local thermocline that we interpret as related to the development of an early western Pacific warm pool. The stepwise decline of G. dehiscens and other deep dwelling species from the NW and SW Pacific suggests more intensive warm water pileup than equatorial localities where surface bypass flow through the narrowing Indonesia seaway appears to remain efficient during the late Miocene. Planktonic δ 18O values from the South China Sea consistently lighter than the tropical western Pacific during the Miocene also suggest, similar to today, more variable hydrologic conditions along the periphery than in the core of the warm pool. Stronger hydrologic variability affected mainly by monsoons and increased thermal gradient along the western margin of the late Miocene warm pool may have contributed to the decline of deep dwelling planktonic species including the early extinction of G. dehiscens from the South China Sea region. The late Miocene warm pool became influential and paleobiologically detectable from ∼10 Ma, but the modern warm pool did not appear until about 4 Ma, in the middle Pliocene.

Progression of methanogenic degradation of crude oil in the subsurface

Our results show that subsurface crude-oil degradation rates at a long-term research site were strongly influenced by small-scale variations in hydrologic conditions. The site is a shallow glacial outwash aquifer located near Bemidji in northern Minnesota that became contaminated when oil spilled from a broken pipeline in August 1979. In the study area, separate-phase oil forms a subsurface oil body extending from land surface to about 1 m (3.3 ft) below the 6–8-m (20–26 ft)-deep water table. Oil saturation in the sediments ranges from 10–20% in the vadose zone to 30–70% near the water table. At depths below 2 m (6.6 ft), degradation of the separate-phase crude oil occurs under methanogenic conditions. The sequence of methanogenic alkane degradation depletes the longer chain n-alkanes before the shorter chain n-alkanes, which is opposite to the better known aerobic sequence. The rates of degradation vary significantly with location in the subsurface. Oil-coated soils within 1.5 m (5 ft) of land surface have experienced little degradation where soil water saturation is less than 20%. Oil located 2–8 m (6.6–26 ft) below land surface in areas of higher recharge has been substantially degraded. The best explanation for the association between recharge and enhanced degradation seems to be increased downward transport of microbial growth nutrients to the oil body. This is supported by observations of greater microbial numbers at higher elevations in the oil body and significant decreases with depth in nutrient concentrations, especially phosphorus. Our results suggest that environmental effects may cause widely diverging degradation rates in the same spill, calling into question dating methods based on degradation state.

Soil moisture content estimation using ground-penetrating radar reflection data

Ground-penetrating radar (GPR) reflection travel time data were used to estimate changes in soil water content under a range of soil saturation conditions throughout the growing season at a California winery. Data were collected during three data acquisition campaigns over an 80 by 180 m area using 100 MHz surface GPR antennas. GPR reflections were associated with a thin, low permeability clay layer located 0.8–1.3 m below the ground surface that was identified from borehole information and mapped across the study area. Field infiltration tests and neutron probe logs suggest that the thin clay layer inhibited vertical water flow, and was coincident with high volumetric water content (VWC) values. The GPR reflection two-way travel time and the depth of the reflector at the borehole locations were used to calculate an average dielectric constant for soils above the reflector. A site-specific relationship between the dielectric constant and VWC was then used to estimate the depth-averaged VWC of the soils above the reflector. Compared to average VWC measurements from calibrated neutron probe logs over the same depth interval, the average VWC estimates obtained from GPR reflections had an RMS error of 0.018 m 3 m −3. These results suggested that the two-way travel time to a GPR reflection associated with a geological surface could be used under natural conditions to obtain estimates of average water content when borehole control is available and the reflection strength is sufficient. The GPR reflection method therefore, has potential for monitoring soil water content over large areas and under variable hydrological conditions.

Comparison of the hydrological and hydrochemical evolution of Lake Naivasha (Kenya) during three highstands between 175 and 60 kyr BP

Three diatomite beds exposed in the Ol Njorowa Gorge south of Lake Naivasha, Central Kenya Rift, document three major lake-level highstands between 175 and 60 kyr BP. Diatom transfer-function estimates of hydrological and hydrochemical parameters suggest that a deep and large freshwater lake existed during the highstands at ∼135 and ∼80 kyr BP. In contrast, a shallower but more expanded freshwater lake existed at ∼110 kyr BP. The best analog for the most extreme highstand at ∼135 kyr BP is the highstand during the Early Holocene humid period from 10 to 6 kyr BP. The environmental conditions as reconstructed from diatom assemblages suggest long-lasting episodes of increased humidity during the high lake periods. This contrasts to the modern situation with a relatively shallow Lake Naivasha characterized by rapid water level fluctuations within a few decades. The most likely cause for the variable hydrological conditions since 175 kyr BP is orbitally driven insolation changes on the equator and increased lateral moisture transport from the ocean.

Comparison of the hydrological and hydrochemical evolution of Lake Naivasha (Kenya) during three highstands between 175 and 60 kyr BP

Three diatomite beds exposed in the Ol Njorowa Gorge south of Lake Naivasha, Central Kenya Rift, document three major lake-level highstands between 175 and 60 kyr BP. Diatom transfer-function estimates of hydrological and hydrochemical parameters suggest that a deep and large freshwater lake existed during the highstands at ∼135 and ∼80 kyr BP. In contrast, a shallower but more expanded freshwater lake existed at ∼110 kyr BP. The best analog for the most extreme highstand at ∼135 kyr BP is the highstand during the Early Holocene humid period from 10 to 6 kyr BP. The environmental conditions as reconstructed from diatom assemblages suggest long-lasting episodes of increased humidity during the high lake periods. This contrasts to the modern situation with a relatively shallow Lake Naivasha characterized by rapid water level fluctuations within a few decades. The most likely cause for the variable hydrological conditions since 175 kyr BP is orbitally driven insolation changes on the equator and increased lateral moisture transport from the ocean.