LAGUNITAS CREEK SEDIMENT AND STREAMBED MONITORING PLAN TECHNICAL COMPLETION REPORT STREAM CONDITIONS 2012 THROUGH 2017

Transcription

1 LAGUNITAS CREEK SEDIMENT AND STREAMBED MONITORING PLAN TECHNICAL COMPLETION REPORT STREAM CONDITIONS THROUGH 217 Prepared by: O Connor Environmental, Inc. P.O. Box 79 Healdsburg, CA 95 In association with Jack Lewis Statistical Consultant US Forest Service (Ret.) Prepared for: Marin Municipal Water District 22 Nellen Avenue Corte Madera, CA 9925 March, 219

2 Contents Executive Summary Introduction Background Information Overview of the Monitoring Plan Monitoring Goals Hypotheses Monitoring Plan Overview Channel Reach Framework for Monitoring Sites Systematic Streambed Surface Sampling Systematic Streambed Subsurface Sampling Sediment Diameter Thresholds for Analysis of Sediment Size Distributions. 2 Implementation Overview Systematic Random Sampling Overview of Data Analyses Monitoring Results Size Distribution of Surface Sediment Proportion of the Stream Bed Occupied by Fine Sediment Distribution of Habitat Types and Surface Sediment Distribution of Habitat Types by Reach Distribution of Surface Sediment Size Distribution by Habitat Type Distribution of Facies Distribution of Facies by Reach Depth of Fine Sediment Facies Sediment Depth as a Proportion of the Water Column-The V* Index Sub-surface Sediment Size Distributions-McNeil Samples Large Woody Debris... 67

3 List of Tables Table 1. Summary of reach characteristics and sampled area Table 2. Fine sediment sizes of significance to spawning habitat Table 3. Summary of monitoring sites sampled by reach and year Table. Dimensions of sample sites by reach Table 5. Summary of change in sediment diameter for percentiles of size distribution Table 6. Summary of proportion of fine sediment facies by reach for and Table 7. Estimated mean depth of fine sediment in fine sediment facies by reach for and Table. Estimated mean depth of fine sediment in all facies by reach for and Table 9. Mean depth of fine sediment facies by year and site Table 1. Mean V* ratio in fine sediment facies by reach, Table 11. Mean V* ratio in fine sediment facies in pools by reach, List of Figures Figure 1. Map of the Lagunitas Creek Watershed Sediment and Streambed Monitoring reaches Figure 2. Schematic representation of streambed sampling Figure 3. Location of sample sites in Lagunitas Creek watershed; annual monitoring sites are circled in yellow Figure. Cumulative surface sediment size distributions by monitoring reach Figure 5. Cumulative surface sediment size distributions by site Figure 6. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach M2 comparing to Figure 7. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach M3 comparing to Figure. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach T2 comparing /13 to Figure 9. Cumulative sediment size distributions for annual monitoring sites Figure 1. Confidence intervals for percentiles of cumulative sediment size distributions Figure 11. Mean and 95% confidence interval for proportion of stream bed sediment < mm by site and reach comparing and Figure 12. Mean and 95% confidence interval for proportion of stream bed sediment < mm Figure 13. Frequency distribution of habitat types by reach for and Figure 1. Frequency distribution of habitat types by site for and Figure 15. Variation in frequency of bars, riffles, runs, glides and pools by monitoring site, Figure. Frequency distribution of all habitat types by site to Figure 17. Surface sediment size distributions by habitat type and reach Figure 1. Frequency distribution of facies by reach for and Figure 19. Frequency distribution of facies by site for and Figure 2. Proportion of fine sediment facies for and Figure 21. Proportion of fine sediment facies by site, Figure 22. Proportion of fine sediment facies with 95% confidence intervals Figure 23. Histograms of depth of fine sediment in fine sediment facies pooled for all reaches for and Figure 2. Histograms of depth of fine sediment in fine sediment facies by reach for and Figure 25. Mean depth of fine sediment in fine sediment facies for and Figure 26. Histograms of depth of fine sediment in all facies pooled for all reaches for and Figure 27. Histograms of depth of fine sediment in all facies by reach for and Figure 2. Mean depth of fine sediment in all facies for and Figure 29. Mean depth of fine sediment in fine facies with 95% confidence interval, Figure 3. Distribution of V* ratio in fine sediment facies for and

4 Figure 31. Distribution of V* ratio in pool habitat for and Figure 32. Distribution of V* ratio for fine sediment facies by reach, to Figure 33. Distribution of V* ratio for fine sediment facies in pool habitat by reach, Figure 3. Mean and 95% confidence interval of V* ratio for fine sediment facies, Figure 35. Mean and 95% confidence interval of V* ratio for fine sediment facies in pool habitat, Figure 36. Box and whisker plots of distribution of V* ratio in submerged fine sediment facies Figure 37. Mean V* ratio in fine sediment facies with 95% confidence interval, Figure 3. Box and whisker plots of distribution of V* ratio in pool habitat in fine sediment facies Figure 39. Mean V* ratio in fine facies in pool habitat with 95% confidence interval, Figure. McNeil sample mean particle diameter with 95% confidence intervals for percentiles or diameters of interest by reach for and Figure 1. System-wide particle size percentiles and proportions of fine sediment with 95% confidence intervals comparing and Figure 2. Mean grain diameter or proportion of size distribution with 95% confidence intervals comparing pooled data from 215 and Figure 3. Mean grain diameter or percentage of size distribution for percentiles and diameters of interest for pooled data from reaches M2, M3 and T2, Figure. Mean diameter of sediment percentiles by site and reach, Figure 5. Percentage of sediment distribution for sediment diameters of interest, Figure 6. LWD load by site and reach with 95% confidence interval of the mean, Figure 7. LWD load in annual monitoring sites,

5 Executive Summary This report summarizes the results of Marin Municipal Water District s Sediment and Streambed Monitoring Plan for Lagunitas Creek (the Plan) implemented beginning in. Data collected through 217 and summarized and analyzed in this report. The Plan comprises a field investigation of sediment size distributions and channel morphology in Lagunitas Creek, Devils Gulch, and San Geronimo Creek motivated by potential impacts of fine sediment on anadromous fish habitat as described in State Water Resources Control Board Order WR Under the Plan, repeatable field protocols are used to sample and describe streambed sediment size distributions, channel morphological patterns, depth of fine sediment deposits, and large woody debris abundance, and to characterize their variation from year-to-year and reach-to-reach. In, data were collected at 1 of 2 monitoring sites in five larger-scale stream reaches. Two additional sites were sampled in 213 to complete the baseline data for the Plan drawn from 2 sample sites distributed evenly among the five reaches comprising the study area. Annual sampling at one site in each of the five reaches was conducted in 213, 21, 215 and 217 to investigate annual change across the watershed. In 2, twelve sites comprising three of the five sampling reaches were surveyed to provide a more robust evaluation of change relative to baseline conditions. The 2 surveys included three of the five sites monitored annually; there are three monitoring sites that have been surveyed each year from through 217. Data analyses summarized in this report focus primarily on two different comparisons: 1) temporal differences in sediment conditions between three reaches (each comprised of four monitoring sites) based on the baseline data set collected in and a repeat survey in 2 of three reaches, and 2) temporal differences between years at five monitoring sites surveyed annually from through 217; two of the five sites were not surveyed in 2. A prior report submitted in 215 based on analysis of data collected -21 provides quantitative evidence of systematic spatial differences in sediment conditions between monitoring reaches M1 (lower Lagunitas Creek downstream of Tocaloma Bridge), M2 (middle Lagunitas Creek between Tocaloma and Devil s Gulch), M3 (upper Lagunitas Creek between Devil s Gulch and San Geronimo Creek), T1 (Devil s Gulch) and T2 (San Geronimo Creek) as shown in Figure 1. Baseline data from indicated surface sediment compared between reaches can be described using the mathematical representation: (T1 T2) > (M3 = M2) > M1. Data from 2 indicated a shift in size distributions such that (T2 = M3) > M2, that is, sediment size became coarser in M3 and similar to T2. No data were collected in M1 and T1 in 2. 1

6 Baseline data from indicated that the proportion of the stream bed surface occupied by fine sediment (defined to be < mm diameter in this study) by reach can be described using the mathematical represenation: M1 > (M2 = M3) > (T1 = T2). These data demonstrate the spatial scale of the downstream fining of sediment in stream channels in the Lagunitas Creek watershed: the principal tributaries have the coarsest sediment, the middle reaches of Lagunitas Creek mainstem have intermediatesize sediment, and the downstream-most reach below Tocaloma has the finest sediment. Data from 2 indicated a shift in fine sediment abundance such that M2 and M3 are statistically indistinguishable (similar to ), M3 and T2 are likewise indistinguishable (dissimilar to ), and M2 has more abundant fine sediment than T2 (similar to ). In other words, in 2, fine sediment was reduced in M3 and became more like T2 and less like M2 in relation to conditions. The pattern of sediment deposited on the stream bed is patchy with distinct size distributions defining different patches. Characteristic patches (also referred to as sediment facies) of stream bed sediment in the study area that have substantial fine sediment (< mm diameter) are described in this study as: well-sorted sand (s), mixed fine gravel and sand (fgs), and gravel with pockets of sand (gs). In, fine sediment facies comprised about half of the streambed in reaches M1 and M2, about % of the streambed in M3, and less than one-fourth of the streambed in reaches T1 and T2. In 2, fine sediment facies had become less abundant and comprised about 25% of the stream bed in M2, and about 2% or less in M3 and T2. Samples of the fine sediment facies collected in baseline surveys indicated that areas of the streambed occupied by fine sediment facies should be regarded as poor-quality habitat for spawning. Spawning habitat quality was evaluated by collecting bulk sediment samples from the streambed using the McNeil method, determining the sediment size distribution for each site by sediment size analysis, and comparing the percentages of fine sediment < 1 mm and < 3.3 mm against threshold values. One McNeil sample was collected from a known or likely spawning site in each of the twenty sampling sites; these sites were typically in gravel facies. Sediment size distributions compared between reaches revealed no significant differences between reaches, in part owing to relatively wide confidence bands for these data with small sample size of four sites per reach. Consequently, the data were pooled to provide the narrowest confidence intervals for sediment sizes of interest. In, sediment < 1 mm diameter comprised about 12% of sediment with the 95% confidence interval ranging from about 9 to 1%. Sediment < 3.3 mm diameter comprised about 26% of sediment with 95% confidence interval ranging from about 23 to 29%. These data indicate good quality spawning habitat overall, however, these fine sediment concentrations are near the threshold values suggesting sensitivity of habitat to increases in fine sediment. 2

7 McNeil samples from 2 in reaches M2, M3 and T2 provide the best basis for evaluating recent changes in spawning habitat quality. Data from these reaches in were compared to the 2 data and no statistically significant changes were found. Nevertheless, the trend was a decline in fine sediment. Sediment <1 mm in comprised 1.7% of the bed material (95% confidence interval:. to 12.7%); in 2, it was.7% (95% confidence interval: 6.7 to 1.7%). Sediment < 3.3 mm in comprised 2.3% of the bed material (95% confidence interval: 2. to 27.9%); in 2 it was 21.% (95% confidence interval: 1. to 2.7%). Sediment data from McNeil samples provide some additional perspective on changes in spawning habitat quality over time. Prior studies conducted in 193- by other researchers sampled sediment in spawning gravels in reaches M2 and M3; the authors of this study collected similar data in reaches M2 and M3 in 2-5. Geometric mean diameter of spawning gravels in 2-5 (about 1 to 15 mm) were found to be unchanged from In -13, the 95% confidence interval of the geometric mean diameter in M2 was 5. to 9.6 mm and in M3 it was 2.9 to 11. mm with a mean of about 7.3 mm in both M2 and M3, a substantial decrease relative to 193- and 2-5. The geometric mean diameter in 2 in M2 increased to.3 mm and the confidence interval expanded (3.9 to 12.6 mm). The geometric mean diameter in M3 increased to 1. mm (confidence interval 5.7 to 15. mm). This suggests recent increases in fine sediment in spawning gravel indicated by the geometric mean diameter between 2-5 and -217 may be associated with changes in watershed erosion and channel conditions associated with the December 25 flood event which is the flood of record in the 35-year period of record for the Samuel P. Taylor State Park USGS stream gauge. Prior streambed monitoring data collected for the District for the period reflected increased proportions of sand on the streambed in 26 in many locations, and relatively high rates of bedload transport in Water Year 26. The depth of fine sediment deposits in fine sediment facies was measured to help quantify fine sediment stored in the streambed. In, reaches M1 and T2 had the greatest mean depth of fine sediment (25.1 and 21.7 cm respectively), T1 and M3 had the smallest mean depth of fine sediment (12.5 and 12.9 cm respectively), and M2 had intermediate fine sediment depth (mean of 19.9 cm). High values of fine sediment depth in M1 and T2 are unsurprising in that M1 is a depositional environment and the San Geronimo Creek watershed (drained by reach T2) is believed to be a significant source of sediment in the Lagunitas Creek watershed. Low values of fine sediment depth in T1 and M3 are unsurprising in that these channels have relatively high channel gradient and stream energy that would reduce potential for fine sediment deposition, and the sediment from much of M3 s natural drainage area is trapped 1 km upstream behind Peters Dam. These circumstances suggest that M3 might be expected to have a relatively coarse sediment size distribution. Intermediate fine sediment depth in M2 is consistent with its position in the watershed where Lagunitas Creek transitions from higher gradient/energy conditions in M3 to low gradient depositional conditions in M1. Data for depth of fine sediment in fine sediment facies pooled by reach comparing and 2 found significant differences in M2 and M3 and marginally significant 3

8 differences in T2. Depth of fine sediment increased about cm to 23.7 cm in M2, 17. cm in M3 and 26. cm in T2. Note that this consistent increase in depth of fine sediment is paired with a consistent reduction in proportion of fine sediment facies, suggesting a trend toward concentration of fine sediment in a smaller area of the streambed, perhaps without a significant change in overall quantity of fine sediment on (or in) the streambed. Considering all facies, there was a statistically significant decline of fine sediment depth of about 1 cm in M2 and M3. Depth of fine sediment on the streambed in 2 averaged for all facies was 9.2 cm in M2,. cm in M3, and 5. cm in T2. Depth of fine sediment was also measured to estimate the proportion of the water column occupied by fine sediment. These data have been used in some northern California streams to evaluate the degree to which pool habitat is occupied fine sediment deposits as an index of sedimentation effects on aquatic habitat. The index (termed V* or v star ) is calculated as the ratio of fine sediment depth to the sum of water depth and fine sediment depth. In this study, we expanded the concept to include all wetted habitat units where fine sediment facies were found; this conceptualization of the V* index is unique to Lagunitas Creek. Reach mean values of V* in ranged from about.35 to.55, with the highest values found in T2 (.55), M1 and T1 (about.3) and the lowest values in M2 and M3 (.3 and.35 respectively). Comparison of V* for in M2, M3 and T2 in and 2 found a marginally significant increase in V* in M2 from.36 to.3. In prior studies, V* measurements were restricted to pool habitat, consequently the V* index was also calculated for pools in the study area. In, pool V* ranged from about.33 to. in reaches M1, M2, M3 and T1; these values are statistically equivalent. Only in reach T2 were pool V* values significantly higher (about.52) than in other reaches, which is consistent with high sediment supply in San Geronimo Creek. V* in pools in 2 were not significantly different from when pooled by reach; there were some significant changes (both increases and decreases) at individual monitoring sites. These V* values are not directly comparable to pool V* from other regional studies; although similar in concept, a different data collection protocol was used for this study. Large woody debris (LWD) is an important element of natural stream habitat that helps create deep pools with shelter favored by juvenile coho salmon. The quantity of LWD expressed as the mass per unit stream bed area (LWD loading) can be used as a measure of habitat condition as set forth by the San Francisco Bay Regional Water Quality Control Board in the Lagunitas Creek TMDL. In stream reaches bordered by redwood forest stands the target value for LWD loading is 3 m 3 /ha. The target for reaches bordered by hardwood forest stands is 1 m 3 /ha. Mean LWD loading in in reaches adjacent to hardwood stands in reaches M1 and M2 ranged from 2 to 1 m 3 /ha, with one of seven sites exceeding the hardwood target. In 2, two sites in M2 recruited enough LWD such that the mean estimate of LWD load exceeds the hardwood target. In 217, the annual monitoring site in M1 also recruited enough LWD to exceed the hardwood target. Mean LWD loading in Lagunitas Creek reaches adjacent to redwood stands (tentatively identified as the four M3 sites plus M2-1) ranged from 12 to 13 m 3 /ha; these monitoring sites include LWD structures previously installed to enhance habitat conditions. One of the monitoring sites in M3 recruited significant

9 additional LWD as of 2; there were declines at other sites in M3. Forest stand types in Devil s Gulch (T1) and San Geronimo Creek (T2) include a mixture of redwood and hardwood stands. In, mean LWD loading in T2 was very low ( to 1 m 3 /ha) and low in T1 ( to 51 m 3 /ha) with the exception of one site where mean loading was 66 m 3 /ha. The annual monitoring sites in both T1 and T2 experienced significant increases in LWD loading through 217 relative to baseline conditions. Three sites appear to meet LWD target conditions. Unprecedented winter runoff in Water Year 217 produced an estimated 2-fold increase in bedload transport (based on an empirical bedload transport relation to stream discharge developed using data from Balance Hydrologics). There was dramatic coarsening of the streambed sediment size distribution in the sites that were measured that year in M3, T1, and T2. Changes were greatest in the most mobile size classes: less than mm up to D5. In the sites that are remeasured annually, mean abundance of fine facies excluding M1 declined from 33% of the stream bed in to 1% in

10 1 Introduction The Sediment and Streambed Monitoring Plan for Lagunitas Creek (the Plan) is an element of the Marin Municipal Water District s (the District) fisheries and riparian management plan (the Lagunitas Creek Stewardship Plan). Implementation of the Plan began in September with field sampling according to protocols detailed in the Plan. This report summarizes the results of the multi-year field sampling program concluded in October 217. The Plan comprises a framework for monitoring sediment in stream channels in the Lagunitas Creek watershed over time using repeatable protocols and data analyses to determine the size distribution of sediment on the streambed and characteristics of the smaller-diameter sediment sizes classes (i.e. fine sediment ). Because sediment data are often highly variable when making comparisons at the same location at different times or between different locations at the same time, data analyses utilize statistical techniques to confirm or reject apparent differences between sediment size distributions. The statistical analyses are critical to appropriate interpretation of the data. The Plan provides for repeated surveys of monitoring sites over time. Four monitoring sites were randomly selected within five monitoring reaches shown in Figure 1: M1 (lower Lagunitas Creek), M2 (middle Lagunitas Creek), M3 (upper Lagunitas Creek), T1 (Devil s Gulch) and T2 (San Geronimo Creek). This report considers data collected during the sixyear monitoring period In, data were collected at four monitoring sites within each of four monitoring reaches and at two sites in a fifth monitoring reach. In 213, two additional sites were surveyed to provide the full complement of four monitoring sites in all five reaches. The full array of twenty sample sites distributed evenly among five monitoring reaches is intended to serve as the monitoring baseline for the Plan. Annual sampling at one site in each of the five reaches was conducted in 213, 21, 215 and 217. In 2, three of the five monitoring reaches (M2, M3 and T2) were resampled for comparison with baseline data. This report quantitatively describes sediment conditions as a function of time in two separate sets of analyses: change over time in three sample reaches (M2, M3 and T2) comparing data from four monitoring sites in each reach collected in and 2, and change over time among five monitoring sites sampled annually from through 217; one site within each reach is included in the annual sampling and the same monitoring site has been sampled each year except 2, when reaches M1 and T1 were not included. 6

11 2 Background Information Lagunitas Creek drains much of west central Marin County and is the largest watershed in the county, encompassing 19 square miles of drainage area (see Figure 1). Lagunitas Creek is an important stream for spawning and rearing coho salmon (Oncorhynchus kisutch), which is federally listed as endangered, and steelhead trout (O. mykiss), which is federally listed as threatened. The creek also supports an important population of California freshwater shrimp (Syncaris pacifica), another listed endangered species. Extensive and long-term monitoring of the populations of coho, steelhead, and shrimp have been conducted in the watershed, along with repeated habitat type surveys, streambed monitoring, and targeted sediment studies. In 21, the San Francisco Bay Reqional Water Quality Control Board (SFRWQCB) adopted a TMDL for sediment for the Lagunitas Creek watershed. Lagunitas Creek originates on Mt. Tamalpais and flows eight miles through four reservoirs operated by the Marin Municipal Water District (District, MMWD). Kent Lake is the fourth reservoir along the main stem of Lagunitas Creek. From Kent Lake, Lagunitas Creek flows about 12 miles before emptying into Tomales Bay. The largest tributary to Lagunitas Creek is Nicasio Creek; MMWD also operates Nicasio Reservoir in this tributary, with about one mile of stream that flows from the dam of Nicasio Reservoir to Lagunitas Creek. Olema Creek is the second largest tributary to Lagunitas Creek and it supports a significant portion of the coho and steelhead populations of the watershed. Lagunitas Creek and Olema Creek meet in the estuary, at the newly restored Giacomini Wetlands, where a vast area of former dairy pasture has been re-opened to tidal action and restores vital estuary habitat. Other major tributaries to Lagunitas Creek include: San Geronimo Creek, Devil s Gulch, Cheda Creek, and McIsaac Creek, all of which also support salmonids. Like Olema Creek, in some years San Geronimo Creek and Devil s Gulch can also support significant portions of the coho and steelhead populations. The District diverts water from the Lagunitas Creek basin to provide water supply for over 19, residents in southern and central Marin County. The District operates seven water supply reservoirs in Marin County, five of which are within the Lagunitas Creek watershed. Kent Lake, which is formed by Peters Dam, is the most downstream reservoir on the mainstem of Lagunitas Creek. The District s diversions are regulated by the State Water Resources Control Board (SWRCB). In its Decision WR95-17, the SWRCB set out an order to MMWD to implement mitigation measures to address the impacts of MMWD water diversions at Kent Lake on Lagunitas Creek and the subsequent deleterious effects to the fishery resources of the creek. Among other elements, the Order instituted a streamflow schedule for Lagunitas Creek that the District must ensure is maintained through releases from Kent Lake. This flow schedule was meant to last for the life of the Kent Lake project. One element of Order WR95-17 required MMWD to develop and implement a ten-year sediment and riparian management plan with the following goals: 7

12 Reduce sedimentation and provide an appreciable improvement in the fishery habitat within the Lagunitas Creek watershed; and Improve the riparian vegetation and woody debris within the Lagunitas Creek watershed in order to improve habitat for fishery resources. Implement a monitoring program to track the status and trend of the coho salmon, steelhead, and freshwater shrimp populations in Lagunitas Creek. In response to Order WR95-17, the District developed the Lagunitas Creek Sediment and Riparian Management Plan. That plan included (as appendices) the Streambed Monitoring Protocol and the Aquatic Resources Monitoring Workplan for the Lagunitas Creek Drainage. That plan was approved by the District Board and SWRCB in 1997 and was established as a ten-year plan. The ten-year milestone was reached in September of 27. While MMWD s role and responsibility for aquatic resource management in Lagunitas Creek did not end in 27, it marked a time for MMWD to re-establish its actions into the future. Accordingly, the District developed the Lagunitas Creek Stewardship Plan, which was approved in June of 211. The Stewardship Plan lays out those actions to be taken by the District to manage the habitat of Lagunitas Creek for the benefit of the aquatic resource populations of coho salmon, steelhead, and California freshwater shrimp. The Lagunitas Stewardship Plan includes a monitoring component that consists of: Sediment and streambed monitoring; Fish population monitoring (focused on coho and steelhead life histories); California freshwater shrimp population monitoring; Stream habitat monitoring; Water quality monitoring (including water temperature); Streamflow monitoring; and Project site monitoring Implementation of the Sediment and Streambed Monitoring Plan 1 for Lagunitas Creek began in. This report summarizes findings from six years of monitoring through O Connor Environmental, Inc.. Sediment and Streambed Monitoring Plan for Lagunitas Creek. Prepared by O Connor Environmental and Jack Lewis for Marin Municipal Water District. 35p.

13 3 Overview of the Monitoring Plan The Plan describes sediment monitoring goals and how they relate to District fisheries and riparian management plans. Prior monitoring methods are reviewed in relation to proposed future monitoring. The Plan describes monitoring parameters and methods, including details pertaining to sampling methods, sample size and analytical methods. The Plan is suitable for implementation; however, it may be adapted over time. 3.1 Monitoring Goals The Plan is intended to provide data and analytical methods that can achieve the following goals. Document sediment and streambed conditions in Lagunitas Creek, including its major tributaries San Geronimo Creek and Devils Gulch. Provide a means to evaluate the efficacy of sediment management efforts implemented within the Lagunitas Creek watershed. Integrate hydrologic and geomorphologic characteristics of Lagunitas Creek with its biological components in an attempt to reveal how streamflow, sediment and streambed conditions influence fish and shrimp populations. The monitoring goals above are related to District fisheries management goals in Lagunitas Creek: Reduce the quantity of fine sediment that enters Lagunitas Creek and enhance the streambed habitat conditions of the creek, for the benefit of coho, steelhead, and California freshwater shrimp. Improve and enhance rearing habitat for salmonids and enhance the condition of the riparian corridor to benefit all fishery resources of the Lagunitas Creek watershed. 3.2 Hypotheses Monitoring data will be used to conduct formal hypothesis testing using appropriate sample statistics as described in this plan. The principal hypotheses to be tested are: Sediment size distributions are uniform throughout Lagunitas Creek; alternatively, sediment size distributions vary between stream reaches. Channel gradient and sediment supply, among other factors, are expected to influence sediment size distributions, hence there is reason to believe that sediment size distributions will differ among stream reaches. A prior report completed in 215 addressed this hypothesis through analysis of the baseline data set from and found that the hypothesis is false: there are differences in sediment size distributions among reaches. The median diameter (D5) of surface sediment compared between reaches conforms to the relationship (T1 T2) > (M3 = M2) > M1. Data from 2 indicated a shift in size distributions such that (T2 = M3) > M2, that 9

14 is, sediment size became coarser in M3 and similar to T2. No data were collected in M1 and T1 in 2. The proportion of the stream bed surface occupied by fine sediment (< mm diameter) by reach in was consistent with the mathematical representation: M1 > (M2 = M3) > (T1 = T2). Data from 2 indicated a shift in fine sediment abundance such that M2 and M3 are statistically indistinguishable (similar to ), M3 and T2 are likewise indistinguishable (dissimilar to ), and M2 has more abundant fine sediment than T2 (similar to ). In other words, in 2, fine sediment was reduced in M3 and became more like T2 and less like M2 in relation to conditions. These data demonstrate the spatial scale of the downstream fining of sediment in stream channels in the Lagunitas Creek watershed: the principal tributaries tend to have the coarsest sediment, the middle reaches of Lagunitas Creek mainstem tend to have intermediate-size sediment, and the downstream-most reach below Tocaloma has the finest sediment. Sediment size distributions are uniform over time; alternatively, sediment size distributions vary over time. Prior monitoring observations suggest that rainstorm and runoff history and proximity to watershed sediment sources affect temporal patterns of change in sediment size. This plan will test whether there are statistically significant time trends in sediment size distribution. This report focusses on the hypothesis that sediment size distributions are uniform over time. This study design is based on prior sampling studies (described below) that focused on surface sediment size distributions. Data from prior studies enabled the development of a sampling design with specified sample size and expected sampling error for surface sediment size distributions. Sediment size distributions will be statistically evaluated in terms of percentiles of the size distribution (e.g. median diameter-d5, D, D, etc.) and the proportion of the size distribution considered to be fine sediment (< mm diameter for purposes of the Plan). Hypothesis testing will be followed by examination of confidence intervals around percentiles and proportions of interest in the sediment size distribution. Additional data relating to size distributions of subsurface sediment, sedimentation processes, large woody debris and fish habitat will be evaluated using descriptive statistics and confidence intervals as well as statistical tests presented in the Plan. These analyses may guide development of additional hypotheses and determination of appropriate sample sizes. Modifications to monitoring specified in the Plan may be considered based on statistical power, sample size requirements, sampling efficiency, access to monitoring sites, and evolving objectives of Stewardship Plan. 1

15 3.3 Monitoring Plan Overview The Plan includes several discrete monitoring parameters distributed among broadly defined stream reaches in the Lagunitas Creek watershed as summarized below (Figure 1). It is designed to evaluate variation in sediment conditions over time and space (e.g. changes in size from year to year at different locations in the watershed) using sampling methods and analytical techniques that can distinguish between statistically-verified trends and random variation. The Plan will also provide quantitative and qualitative data regarding the quality and quantity of aquatic habitat in support of fish habitat monitoring by District biologists. The overall sampling design and sampling methods are summarized in this sub-section Channel Reach Framework for Monitoring Sites Monitoring sites are distributed within distinct reaches of the Lagunitas Creek mainstem and its tributaries. Three mainstem reaches have been identified (Figure 1) based on data from prior studies (Table 1) pertaining to sediment size, geomorphology and channel slope: Hwy 1 to Tocaloma Bridge (reach M1), Tocaloma Bridge to Devils Gulch (reach M2), and Devils Gulch to Shafter Bridge (reach M3). Two tributaries will be monitored: Devils Gulch (reach T1) and San Geronimo Creek (reach T2). 11

16 Figure 1. Map of the Lagunitas Creek Watershed Sediment and Streambed Monitoring reaches. 12

17 Monitoring data were collected and analyzed within each of these five reaches because they represent distinctive portions of Lagunitas Creek that have significantly different characteristics such as channel slope and width (Table 1), as well as differences in streamflow and sediment supply controlled by both natural conditions and the effects of Kent Lake and Lake Nicasio 2. Time trend analysis of channel conditions will be made more effective by collecting and analyzing monitoring data in this spatial framework because inherent variability between reaches will be distinguishable from change over time at a particular location caused by variation in sediment supply and/or streamflow. Table 1 describes the length and mean width of sampling sites distributed among five reaches; additional details regarding sampling sites are provided in subsequent sections of the report. Reach Table 1. Summary of reach characteristics and sampled area. Length a (miles) Mean Slope (ft/ft) Typical Bankfull Width (ft) f Reach Sampling Program Mean Reach Percent Width of Length of Sample Sampled Reach Transects (ft) Length (ft) Sampled M1: Tocaloma Bridge to Hwy c M2: Devils Gl. to Tocaloma Br c M3: Shafter Bridge to Devils Gl b T1: Devils Gulch 1..2 d T2: San Geronimo Creek g.6.7 b Notes: a. Reach lengths from MMWD (2) Lagunitas Creek Habitat Typing Survey 26 Analysis, Table 2. b. Slope estimated from longitudinal profile surveyed by SFBRWQCB. c. Slope estimated from preliminary analysis of 29 LiDAR data. d. Slope estimated from USGS topographic data. e. Sample length refers to the systematic sampling reaches. f. Representative widths determined from prior measurements in 2 and 25. See Figure 1 for reach locations. g. Only two sites in San Geronimo Creek were sampled in owing to access issues on private land; two additional sites were sampled in 213 to provide the full complement of four monitoring sites specified in the Plan. 2 O Connor Environmental, Inc. 26. Lagunitas Creek Fine Sediment Investigation. p

18 3.3.2 Systematic Streambed Surface Sampling A systematic sampling framework was used for data collection. The systematic sampling approach is used to determine the streambed surface sediment size distribution and other characteristics of interest, such as distribution of habitat types and sediment patches (facies), depth of fine sediment deposits, and large woody debris. A systematic random grid such as that depicted in Figure 2 was established within the lateral limits of the bankfull channel to conduct this sampling procedure. Data to be collected using the systematic sampling grid include: the size distribution of sediment on the surface of the streambed, the size distribution of sediment in the subsurface of the streambed (subsampled on the grid), the proportion of the channel bed occupied by fine sediment including characteristic sediment mixtures distributed in patches (sediment facies), proportions of habitat types (pool, run, glide, riffle and cascade) depth and distribution of fine sediment deposits, and volume and origin of large woody debris (measured on sediment transects) 15 3 Figure 2. Schematic representation of streambed sampling. 1

19 Habitat Types. The distribution of habitat types was systematically sampled on the grid. Observation and monitoring of habitat types may provide insights regarding sedimentation processes as they relate to fish habitat. Aquatic habitat types previously used for Lagunitas Creek include pool, glide, run, riffle and cascade. These habitat types are defined with respect to the wetted channel. Sampling transects span the channel from bank to bank and include portions of the channel that lie above and outside the wetted channel perimeter. These locations were categorized as bank (points located on the lower bank of the stream), or bar (points located on gravel bars lying in the channel cross-section). Sediment Facies. In some stream channels, the size distribution of sediment may be essentially homogeneous. More commonly in gravel-bed channels, patches of sediment with different characteristic sediment sizes may be found distributed across the stream bed reflecting local differences in hydraulics and sediment transport. These features have been investigated in Lagunitas Creek tributaries 3 as cited in the sediment TMDL staff report prepared by SFRWQCB. The organization of the streambed sediment in these distinctive patches has been studied as a morphological characteristic of streambeds induced by cross-channel differences in sediment transport dynamics and in relation to variations in watershed sediment supply over time 5,6. In this study, we refer to these patches as facies with distinct sediment size distributions that manifest distinct local depositional environments. Observation and monitoring of sediment facies provides insights to sedimentation processes, and facies distribution will be evaluated in relation to other sedimentation and habitat monitoring parameters. The relative abundance of fine sediment facies (sediment patches with substantial quantities of sediment < mm diameter visible on the surface of the streambed) on the bed of Lagunitas Creek, which is otherwise populated by gravel and cobble-sized material, produces a bimodal surface sediment distribution (see Figure 5). Fine sediment facies are found distributed throughout Lagunitas Creek and occur in various positions and abundance in relation to the gravel-cobble framework of the streambed. Fine sediment may become concentrated in relatively low-lying topographic positions with distinctive hydraulic characteristics such as pools. In many locations, the fine sediment facies completely bury the underlying gravel-cobble framework, and in other areas occupy 3 Cover, M. () Linkages between sediment delivery and streambed conditions in the Lagunitas Creek watershed, Marin County, California. Prepared for the San Francisco Bay Regional Water Quality Control Board, Oakland, CA. California State University, Stanislaus, Turlock, California. 36 p. CRWQCB, San Francisco Bay Region (21) Lagunitas Creek Watershed Fine Sediment Reduction and Habitat Enhancement Plan. 5 Lisle, T.E. and Madej, M.A Spatial variation in armoring in a channel with high sediment supply. IN P. Billi et.al. (Eds) Dynamics of Gravel Bed Rivers, John Wiley & Sons Ltd. 6 Lisle, T.E Particle size variations between bed load and bed material in natural gravel bed channels. Water Resources Research, 31()

20 a smaller portion of the channel bed such that the fine sediment facies appear as islands or peninsulas amid a sea of gravel and cobble. Our conceptual model accounting for the persistence and distribution of fine sediment facies in Lagunitas Creek is that these facies represent the readily available and transportable bed load sediment stored in stream channels and entrained during peak flows that typically occur a few to several times each winter. Fine sediment enters the stream channel network from various erosion sources in the watershed as summarized in the sediment TMDL, and coarser sediment, primarily sand and gravel, is deposited in stream channels throughout the watershed. The suspended sediment load (or wash load) is comprised of small diameter material (primarily clay and fine silt) that travels downstream at a velocity comparable to that of the stream flow. The suspended load is largely evacuated from the stream bed where it is found in low concentrations but may be deposited in significant quantity on floodplain surfaces or discharged to Tomales Bay. Sand, gravel and larger sediment is stored on the bed and transported during periods of peak stream flow as bed load sediment. The streambed sediment comprises a sediment reservoir that is reworked during periods of high streamflow that entrain sediment stored on the streambed. During these significant transport events, bed load sediment typically travels a length of channel proportional to a few to several multiples of the channel width, and therefore require a period of a few to several decades to traverse the channel network to Tomales Bay. Sediment sorting processes occur due to irregularities in channel form that produce variation in cross-channel and longitudinal flow velocity, which contributes to formation of patches of finer and coarser sediment on the stream bed. During lower magnitude peak flows, much or most of the gravel-cobble framework of the streambed may not be entrained (or may be only marginally entrained), however patches of finer sediment may be more readily entrained and significant transport of this finer fraction of sediment may occur. In this conceptual model, it can be said that coarser portions of the channel are armored with respect to low magnitude peak flows, while transport of sediment comprising the fine sediment facies may occur. During periods of higher, less frequent peak flows (e.g. associated with spill events from Kent Lake and/or intense storm runoff), the gravel-cobble bed is more widely entrained and significant transport of both the gravel-cobble framework and fine sediment facies occurs. During these relatively infrequent flows fine sediment stored in the subsurface beneath the gravel-cobble armor becomes available for transport, and maybe redistributed in new bar deposits and in fine sediment facies distributed on the channel surface. Sediment facies were systematically sampled on the grid described above. Sediment facies previously described in the study area include well sorted sand < 2 mm diameter and including some clasts of fine gravel up to about mm diameter, fine gravel and sand (well mixed, average median diameter about to 5 mm in and 2), gravel with pockets of sand (well-sorted gravel with pockets of much finer sediment, average median

21 diameter about 13 mm in and ranging from < mm to about mm in 2), gravel dominant (well sorted gravel up to 6 mm diameter, average median diameter about 2 mm in and about 22 mm in 2), and cobble dominant (well sorted cobbles > 6 mm diameter, average median diameter about 5 mm in ). Three facies represent the group of fine sediment facies: sand, fine gravel and sand, and gravel with pockets of sand. In the process of collecting data in, additional categories to describe the facies (or patch) setting of some transect points were identified. These categories describe the characteristic hydraulic/stream environment where the more typical types described above were not present. These were boulder, bedrock, roots and wood. These facies types result from the systematic random sampling design of this study, and do not generally represent a sediment facies or patch as would be described in studies focused on sediment patches. The boulder and bedrock facies were found only in reaches M3, T1 and T2; presumably these were absent in M1 and M2 because of low slope gradient and deeper alluvial valley fill. The roots and wood facies were found in all reaches associated with stream banks and LWD but were a very small proportion of the sample. Depth of Fine Sediment. The depth of fine sediment facies was sampled on the grid. Observation and monitoring of the depth of fine sediment facies will permit estimation of the quantity of fine sediment, its size distribution, and its distribution in Lagunitas Creek. These data also provide quantitative estimates of the depth and volume of fine sediment stored on the channel bed as an indication of sedimentation impacts on fish habitat as well as trend analysis related to effectiveness of watershed management to control erosion. This approach to monitoring fine sediment is based on prior studies of fine sediment in gravel bed streams in northern California. 7, The depth of fine sediment facies was measured by pushing a thin metal rod (about 5 mm diameter) into the stream bed until strong resistance was encountered, presumably indicating the depth of an underlying layer of the gravel-cobble framework of the streambed. This measurement was taken at all grid points within fine sediment facies. The approach to monitoring depth of fine sediment deposits is similar to the methodology for determining V* to systematically monitor the volume of pools occupied by fine sediment. 6 In this monitoring protocol, however, data on water depth from locations where fine sediment facies are not present is not collected, so the V* calculation for pool habitat cannot be determined. Consequently, V* values from other studies are not directly applicable to interpretation of these data. The quasi-v* data collected and analyzed in this study is more useful for comparisons over time and for reach-scale comparisons. The data provide a quantitative measure of the proportion of the volume of space below 7 Lisle T and Hilton S (1999) Fine bed material in pools of natural gravel bed channels. Water Resources Research 35(): Hilton, S. and Lisle, T.E. (1993) Measuring the fraction of pool volume filled with fine sediment. Research Note PSW-RN-1, 11 pp. USDA For. Serv., Albany, Calif. 17

22 the water surface occupied by sediment that might otherwise be occupied by aquatic organisms including salmonids. Large Woody Debris (LWD). Systematic random streambed sampling includes measurements of LWD (defined as woody material at least 1 cm in diameter and 2 m in length) on sample transects established for the sampling grid (Figure 2). A line transect technique was used wherein only the diameter of LWD encountered on sample transects is measured along with a description of the origin of the LWD; these data can be used to estimate LWD load (volume per unit area). 9,1,11 The origin of LWD was described as management for LWD in structures previously installed in the stream, live for living down woody material (typically willow or alder on stream banks or bars), and natural for other down LWD. The sediment TMDL specifies target LWD loading for desired future conditions; these data may be used to evaluate LWD abundance in accord with the TMDL and can be efficiently incorporated in the Plan. LWD is a component of fish habitat that contributes cover and may interact with streamflow to create pools and depositional features. Observations and monitoring of LWD can be used to characterize the role of LWD in forming habitat and permit evaluation of the effect of LWD on sediment size distributions. Prior studies of Lagunitas Creek suggested that streambed sediment sizes tend to be finer in the vicinity of LWD. 12 Sampling Limitations. It was expected that proposed methods would require modification for the Tocaloma to Hwy. 1 reach (M1) owing to the prevailing depth of water greater than about three feet, which we have found to be the practical limit of sampling techniques used in the Plan. In addition, sample sites within the San Geronimo Creek reach (T2) were restricted due to private property access limitations. These issues were manifest in the sampling program. Sampling methods were retained, but site selection methods in M1 and T2 reaches were modified. The sampling protocol is implemented by observers wading the stream channel. When water depths exceed about 3 feet, sampling becomes difficult and is impossible when water depths exceed about 3.5 feet. When transect locations overlapped with such deeper water areas, the protocol is to skip to the next upstream transect that can be sampled. Generally, this has affected only a few transects out of 3 transects at each site. However, at one site (reach M2, site 6, denoted M2-6), streambed changes observed in 217 have deepened and/or enlarged pools over time, resulting in a substantial shift in the overall sample site such that substantial lengths of streambed have 9 Swanson F et al. (19) Organic debris in small streams, Prince of Wales Island, southeast Alaska. Gen. Tech. Report. PNW-6. USDA Forest Service PNW Forest and Range Experiment Station, 11 p. 1 Van Wagner, C. (192) Practical Aspects of the Line Intersect Method, Petawawa National Forestry Institute, Canadian Forestry Service, Chalk River, Ontario. Information Report PI-X-12, 11 p. 11 O Connor, M. and Ziemer, R. (199) Coarse woody debris ecology in a second-growth Sequoia sempervirens forest stream. USDA Forest Service Gen. Tech. Report PSW-11, pp O Connor Environmental, Inc. (26) Lagunitas Creek Fine Sediment Investigation. Prepared for Marin Municipal Water District, pp

23 been sampled outside the original sample site. The comparability of data from this site has thus become questionable Systematic Streambed Subsurface Sampling Sampling and analysis of size distributions of the sediment underlying the streambed surface where salmonids are likely to spawn will provide data describing spawning habitat quality and the sedimentation status of the bed. These size distributions are also representative of the bed material of the Lagunitas Creek and are useful for evaluations of bed load transport. In addition, fine sediment facies were sampled to describe in greater detail the size distribution of fine sediment in fine sediment facies where the description of the surface size is limited (i.e. < mm). Spawning Gravel Condition. One spawning habitat site within each sample site was randomly selected from transects where prior spawning occurred based on prior redd surveys. A bulk sample of the bed material was collected by inserting an open-bottomed five-gallon bucket into the stream bed and excavating the enclosed sediment to a depth of about twenty to thirty centimeters. This produces a sample comparable to that of a McNeil sampler. The size distribution of this sediment was determined by a geotechnical laboratory; these distributions are presumed to be representative of the size distribution of spawning gravel and are used to characterize spawning habitat quality and spatial variability. These data are also compared to earlier data sets to evaluate changes over time. Sediment Supply and Transport Capacity The q* Index. As described in our 215 report on baseline conditions, sediment size distributions representative of spawning sites (McNeil samples) were used to calculate the theoretically-based geomorphic index q* representing the relationship between sediment transport capacity and sediment supply. 13 To compute q* the boundary shear stress on the channel bed for the prior winter s peak stream flow must be estimated along with the boundary shear stress at the threshold of entrainment for both the surface layer and the subsurface sediment (considered representative of the sediment load). The size distribution of the streambed surface at the sampling location was measured using a systematic method at the McNeil sample sites. The subsurface sediment size distribution was determined from McNeil samples. To estimate bed shear stress, we measure a representative flow depth and slope in the field at the McNeil sample site; bed shear stress is proportional to the product of flow depth and slope. This estimate has been derived from stream channel topographic surveying at McNeil sample sites in October of each year of the channel cross-section identifying a representative flow depth and the channel bed slope from the nearest 13 Dietrich, W.E. et.al Sediment supply and the development of the coarse surface layer in gravelbedded rivers. Nature 3:

24 identified channel slope controls upstream and downstream from the McNeil sample site. In several instances, these data did not yield realistic estimates of shear stress. We believe this is due to multiple factors, including the low slope gradient in many locations in Lagunitas Creek, backwater effects during peak flows that reduce the effective local stream gradient, and complexities in local flow resistance that cause deviations of flow velocity field and local bed shear stress. We have concluded that different methods are required to directly measure or accurately estimate boundary shear stress and that q* analysis, although potentially useful, should be discontinued unless an alternative method is developed to determine boundary shear stress. Fine Sediment Facies. As described in our 215 report on baseline conditions, the three fine sediment facies (sand, fine gravel and sand, gravel with pockets of sand) were randomly sampled in each reach in proportion to their occurrence using a sediment coring device (at sites in the wetted channel) or a trowel (at sites outside the wetted channel). These samples were dried and sieved for sediment size distribution using mesh sizes,, 3 and 6, corresponding to sediment diameters.76, 1.19,,595 and.25 mm respectively. The sample collection methodology for these data was judged to be inconsistent with uncertain repeatability. Although the data developed were useful and provided valuable perspective on the size distribution with fine sediment facies, this part of the monitoring plan was discontinued because we were not confident that the data would be reliable for purposes of trend analysis Sediment Diameter Thresholds for Analysis of Sediment Size Distributions Sediment with a diameter less than mm is considered fine for purposes of this study. The threshold diameter of mm is based largely on the practical constraint that it is increasingly difficult to implement systematic random sampling on smaller and smaller sediment grains simply because they are difficult to select from the bed. We selected mm as the size limit for our sampling protocol based on prior sampling in Lagunitas Creek. With respect to the effects of fine sediment on salmonid fish habitat, there is an extensive scientific literature. Fine sediment has been shown to be deleterious to salmonids at different life stages on the basis of laboratory and field studies. A meta-analysis of prior studies identified specific sediment diameters and thresholds of their concentration in spawning gravels corresponding to fifty percent mortality (Table 2) and suggested that these concentrations represent thresholds of concern for habitat quality. 1 Sediment finer than about.5 to 1. mm has been shown to significantly inhibit the flow of water through salmonids redds, thereby reducing the flow of oxygenated water to salmonids eggs and alevins and reducing the removal of waste products from the redd. Reduced inter-gravel water flow in redds has been shown to increase mortality of 1 Kondolf, G.M. 2. Assessing salmonid spawning gravel quality. Transactions of the American Fisheries Society, 129:

25 salmonids. Based on literature values (Kondolf 2), we identified a threshold range for salmonid survival to emergence from redds of 12 to 1% for sediment finer than 1 mm. Fine sediment can also prevent emergence of juvenile salmonids from redds by filling in the crevices and channels between larger sediment grains comprising redds. For coho salmon, the critical diameter of sediment contributing to entombment is about 3.3 mm; for steelhead, it is about 6.3 mm. The concentration of these sediment sizes in redds that correspond to a fifty percent mortality are in the range of 3 to 33% for coho and 3% for steelhead. Analysis of size distributions of sediment from McNeil sampling and from fine sediment facies includes determination of the percentage of sediment finer than these biologically significant sediment sizes: 1 mm, 3.3 mm and 6.3 mm. The latter size is considered only in McNeil sample analysis. The size distributions of McNeil samples are also expressed with reference to diameter of sediment for selected percentiles of the size distribution (, 25, 5, 75 and ) for comparison to other samples. The fine sediment facies are described by the percentages of sediment finer than selected sediment diameters (.25 mm,.595 mm, 1 mm and 3.3 mm). The latter two diameters were selected to reference biologically significant thresholds, while the former two diameters correspond to the sieve screen sizes used to analyze the samples and represent sediment diameters ranging to the upper size limit of fine sand. The abundance of these finer size classes of sand provide interpretative value with respect to sediment transport mechanisms and the likelihood of transport in suspension in the water column. Table 2. Fine sediment sizes of significance to spawning habitat. Particle Diameter 1 mm 3.3 mm Particle Description Coarse sand Very gravel fine mm Fine gravel 6.3 mm Fine gravel 22 mm Coarse gravel Significance of size class for Lagunitas Creek Representative diameter for spawning gravel quality; > 12-1% by weight of spawning gravel significantly reduces survival in redds (coho and steelhead) Representative diameter for spawning gravel quality; > 3-33% by weight of spawning gravel significantly reduces emergence from redds (coho) Size boundary selected to define fine sediment in surveys of Lagunitas Creek; smaller grains classified as < mm in field surveys and larger sediment grains measured to determine diameter size class Representative diameter for spawning gravel quality; > 3% by weight of spawning gravel significantly reduces emergence from redds (steelhead) Largest median diameter in any sample reach in Lagunitas Creek (Reach T1-Devils Gulch); representative of upper boundary of sediment size classes that are mobile throughout Lagunitas Creek study reaches and capable of being transported through the channel network to lower Lagunitas Creek (Reach M1). 21

26 Implementation.1 Overview The location of sampling sites distributed among the five designated reaches in the Lagunitas Creek watershed is displayed in Figure 3. The Plan specifies four sampling sites to be spaced evenly within each reach according to a systematic random sampling design. Limited access to the stream channel on private property in confined sampling in San Geronimo Creek (reach T2) to two reaches located on District property (Sites T2:12 and T2:2). Two additional sampling reaches in San Geronimo Creek were added and sampled in 213 (T2:9 and T2:31). Site selection in San Geronimo Creek (monitoring reach T2) was not strictly according to the sampling design owing to limited access to private property. The distribution of sites in T2 is well-spaced except for the relatively close proximity of sites T2:9 and T2:12. For purposes of data analysis and interpretation, all sites are considered randomly selected. Selected sites in M1 below Nicasio Creek could not be sampled owing to typical water depths greater than about three feet, and local water depths exceeding five feet. Field reconnaissance was conducted to identify potential sample reaches where typical water depths were less than about three feet on private property where access to Lagunitas Creek was granted. The first two sites deemed suitable based on field reconnaissance were selected for sampling. The significance of this sampling problem is that substantial portions of reach M1 below Nicasio Creek are comprised of deep pools that are not typical of Lagunitas Creek further upstream, and these were excluded from sampling. Consequently, data for reach M1 do not characterize the full range of stream channel environments found in the reach. The impact of this data limitation is mitigated in that the primary objective of the Plan is to detect changes in conditions over time by resampling the same sample sites. Data collection at monitoring sites summarized by reach and year is shown in Table 3. Table 3. Summary of monitoring sites sampled by reach and year. Year M1 M2 M3 T1 T2 m m m m m* m 1m 1m 1m 1m 2 m m m 217 1m 1m 1m 1m 1m Table 3 notes: Subscript m indicates McNeil samples collected; *denotes two sites sampled in 213 that were analyzed as part of the baseline data. 22

27 Figure 3. Location of sample sites in Lagunitas Creek watershed; annual monitoring sites are circled in yellow. 23

28 .2 Systematic Random Sampling Sample sites (four per reach) were selected based on a systematic random sampling design within each reach. Each reach was segmented into potential sample sites of the specified length (equivalent to twenty bankfull widths). Each segment was numbered sequentially. The number of sampling segments was divided by four into quartiles; one monitoring site was to be selected in each quartile to ensure broad distribution of sites in the reach. A random number between zero and one was generated and multiplied by the number of sampling segments in each quartile; this product was rounded to the nearest whole number, identifying the numbered sampling segment that became the downstreammost monitoring site. The other three monitoring sites in the reach were then identified by successively adding the number of sampling segments in the reach quartiles. For example, in reach M2, there were sixteen sampling segments, with four segments in each quartile. Segment two was randomly selected in the first quartile, and every fourth segment was selected thereafter; hence the complete sample consisted of segments 2, 6, 1, and 1. Systematic random sampling within sample sites was implemented per the concept illustrated in Figure 2. A representative channel cross section was surveyed near the beginning (downstream end) of the sample reach to determine bankfull width. Transects perpendicular to the stream channel alignment were spaced at intervals of one-half the bankfull width along the channel centerline. Transects were sampled by suspending a flexible measuring tape across the transect. The starting point on the transect began at the base of the left stream bank (oriented facing downstream) and ended at the base of the right bank. Sample points on each transect were determined as follows. First, the measured transect length (measured to the nearest tenth of a foot) was divided by 11 to determine the spacing between transect points. A random number between and 1 was generated and multiplied by the spacing between points to determine the random location of the first point. The remaining systematic random sample points were located at equally-spaced points across each transect. This sampling grid yields 3 data points from thirty transects at each sample site; this design provides relatively high accuracy while limiting the extent (and cost) of sampling 15. The downstream end of each monitoring reach was mapped and monumented in the field to provide certainty in relocating the site for subsequent resampling. Photographs were taken from the right bank across sample transects to provide additional physical references allowing for accurate relocation of each of the thirty transects surveyed at each site. Steel re-bar driven into the ground with safety caps have been added to several 15 The study plan called for forty transects distributed over twenty bankfull channel widths and data points per site; after initial field trials in September it was determined that it was necessary to reduce the size of sample sites by 25% so that sampling effort was better matched to the resources available to complete the data collection phase of monitoring. 2

29 monitoring sites to help monument and relocate monitoring sites. Flexible tapes are fixed to the re-bar to facilitate efficient re-location of sampling transects. Summary data describing the sampling of monitoring reaches are presented in Table. These data illustrate the relationship between bankfull width and transect spacing as well as the relative dimensions of sampled monitoring reaches. Table 1 also provides perspective on the relative proportion of reaches sampled. Sampled proportions of M2, M3 and T1 range from 15-1% of total length; substantially smaller portions of the much longer M1 and T2 reaches were sampled (7.6% in each case)..2 Overview of Data Analyses As described in Section 3.1, one of the principle objectives of the Plan is to document sediment and streambed conditions in Lagunitas Creek (monitoring reaches M1, M2 and M3) and its major tributaries that contain significant fish habitat, Devils Gulch (monitoring reach T1) and San Geronimo Creek (monitoring reach T2), and to compare conditions in these reaches. This objective was fulfilled in a report prepared in 215 drawing on analyses of data collected in supplemented by 213 data collected at sites T2:9 and T2:31 as described above. The second major hypothesis of the Plan is that sediment size distributions are uniform over time; the alternative hypothesis is that sediment size distributions vary over time. Prior monitoring observations suggest that storm history, proximity to watershed sediment sources, and the influence of Peters Dam on streamflow and sediment supply (erosion processes) could affect temporal patterns of variation in sediment size. Subsequent to the baseline data collection at 2 monitoring sites in /13, annual sampling at one site in each of the five reaches was conducted in 213, 21, 215 and 217. In 2, three of the five monitoring reaches (M2, M3 and T2) were re-sampled for comparison with baseline data. This report quantitatively describes sediment conditions as a function of time in two separate sets of analyses: change over time in three sample reaches (M2, M3 and T2) comparing data from four monitoring sites in each reach collected in and 2, and change over time among five monitoring sites sampled annually from through 217; one site within each reach is included in the annual sampling and the same monitoring site is sampled each year except in reaches M1 and T1, which were not sampled in 2. 25

30 Table. Dimensions of sample sites by reach. Reach Site Mean transect length (ft) Representactive bankfull width (ft) Transect spacing (ft) Channel length surveyed (ft) Channel area surveyed (sq-ft) M ,9 M ,3 M ,55 M ,72 Mean transect Total length & area 3.5 length-m1 surveyed-m1 2,3,73 M , M , M ,6 M ,672 Mean transect Total length & area 35.7 length-m2 surveyed-m2 2, 77,112 M ,35 M ,26 M , M , Mean transect Total length & area 3.1 length-m3 surveyed-m3 2,6 113,7 T ,2 T ,7 T1 17* ,13 T ,23 T ,321 Mean transect Total length & area 1.1 length-t1 surveyed-t1 1,35 2,35 T ,1 T ,195 T , T ,236 Mean transect length-t Total length & area surveyed-t2,67 * Site inadvertently sampled and included for monitoring data; site not intended for future monitoring. 26

31 5 Monitoring Results A wide variety of statistical analyses of the monitoring data were performed in exploration and preliminary analyses, not all of which are presented here. The statistical tests ultimately selected for various data are consistent with those specified in the Plan, but alternative approaches were utilized as appropriate to the data. The rationale for selection of statistical methods applied is discussed, and the statistical methods utilized are specified. It is anticipated that metrics of streambed condition and sedimentation will be further specified and refined as implementation of the Plan progresses over time. Results are presented in this section in the following order, grouped by types of data: 1. Size distribution of surface sediment. 2. Proportion of the stream bed occupied by fine sediment (< mm diameter). 3. Distribution of habitat types.. Distribution of sediment facies. 5. Depth and distribution of fine sediment facies. 6. Sediment depth as a proportion of the water column (the V* index) 7. Sediment size distribution of sub-surface sediment at spawning sites (McNeil samples).. Distribution and volume of large woody debris. In this section of this report, the results of statistical analyses of these data are reported for the monitoring parameters described above. For each monitoring parameter, the results of the to 2 comparison of 12 monitoring sites in reaches M2, M3 and T2 are reported first, followed by the results of annual sampling -217 of 5 monitoring sites. Annual sampling was restricted to a single site within each reach and results may not be generalizable to the larger reaches. 27

32 5.1 Size Distribution of Surface Sediment Reach Comparison to 2 The size distribution of sediment on the surface of pooled data from four monitoring sites in each of the reaches M2, M3 and T2 is presented in Figure. The cumulative frequency distribution of the sediment on the surface of the stream bed is plotted with the sediment diameter class on the x-axis and the percentage of sediment finer than the size class plotted on the y-axis. The largest sediment size class found in the reach corresponds to 1 percent of the cumulative frequency distribution. The median sediment diameter in each reach is at 5 percent of the cumulative frequency distribution (also denoted as D5). To test whether sediment size distributions differ from one another, the Χ 2 (chisquare) statistic was used to test for differences among proportions by contingency table analysis. Size classes with fewer than five occurences (all at the coarse tail of the distribution) in a reach were eliminated from the comparison. These data contradict the null hypothesis that sediment size distributions do not change over time. Sediment size distributions did not change significantly in M2 or T2; however, sediment size distribution of the bed surface did change significantly in M3 (p <.). Figure indicates that the cumulative frequency distribution up to the th percentile is coarser in 2 relative to ; the frequency of sediment > 5 mm diameter is unchanged. Cumulative Frequency (Pct Fin 6 1 D5 Reach M2 2 Cumulative Frequency (Pct Fin 6 1 D5 Reach M3 2 Cumulative Frequency (Pct Fin D5 Reach T Particle size (mm) Particle size (mm) Particle size (mm) Figure. Cumulative surface sediment size distributions by monitoring reach. Sediment size distributions for each site (pooled by reach in Figure ) is shown in Figure 5. In this set of comparisons, M2-2 expresses marginally significant change (p=.1) and M3-1 and M3-1 express significant change (p=.23 and p=.13, respectively) between and 2. Unlike sites in M2 and T2, the size distributions in all sites in reach M3 indicated coarsening in some portion of the cumulative distribution. 2

33 Cumulative Frequency (P 6 1 Reach M2 : Site Cumulative Frequency (P 6 1 Reach M2 : Site Cumulative Frequency (P 6 1 Reach M2 : Site Cumulative Frequency (P 6 1 Reach M2 : Site Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Cumulative Frequency (P 6 1 Reach M3 : Site Cumulative Frequency (P 6 1 Reach M3 : Site Cumulative Frequency (P 6 1 Reach M3 : Site Cumulative Frequency (P 6 1 Reach M3 : Site Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Cumulative Frequency (P Reach T2 : Site Cumulative Frequency (P Reach T2 : Site Cumulative Frequency (P Reach T2 : Site Cumulative Frequency (P Reach T2 : Site Particle size (mm) Particle size (mm) Particle size (mm) Figure 5. Cumulative surface sediment size distributions by site. Particle size (mm) Differences in sediment size distributions were also evaluated using bootstrap confidence intervals around the mean for selected percentiles of the distribution (5,, 25, 5, 75,, 95). With the bootstrap technique each data set was sampled with replacement 5 times, percentiles were computed from each sample, and a 95% confidence interval for each percentile was estimated from its bootstrap distribution of 5 values.. The mean diameter and 95% confidence intervals for percentiles for each site are plotted in groups for each sample reach in Figures 6, 7 and for reaches M2, M3 and T2, respectively. Data for percentile 5 and are not plotted because these percentiles are predominantly < mm diameter which is the lower limit of field measurements of sediment diameter. Bootstrap bias-corrected accelerated interval, or BCa. 29

34 Figure 6. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach M2 comparing to 2. Figure 7. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach M3 comparing to 2. 3

35 D25 (mm) D5 (mm) D75 (mm) T2:9 T2:12 T2:31 T2:2 T2:9 T2:12 T2:31 T2:2 T2:9 T2:12 T2:31 T2:2 D (mm) D95 (mm) T2:9 T2:12 T2:31 T2:2 Figure. Confidence intervals for selected percentiles of the size distribution of surface sediment for monitoring sites in reach T2 comparing /13 to 2. Using Friedman s test comparing pooled data for each reach for /13 and 2, marginally-significant differences were found for D25 in M2 and M3 (p=.55), for D5 in M3 (p=.55), and for all sites pooled for D25 (p=.29). Annual Site Comparison -217 T2:9 T2:12 T2:31 T2:2 Cumulative frequency distributions of streambed sediment size at the annual monitoring sites are provided in Figure 9. Data for each site was analyzed to test for changes over time using the chi-square statistic. Significant differences were found for M3, T1 and T2 (p<.). At M3, the largest change occurred from 2 to 217. At T1 and T2, the largest changes occurred from 215 to 217. A matched grid position test, similar to a matched pairs t-test, was used to compare multiple years of data using Friedman s test. This test also found significant changes in sediment size distribution at M3 (p=.17), T1 (p=.39), and T2 (p<.). Inspection of the cumulative frequency distributions in Figure 9 reveal that the sediment size distributions coarsened in 217 at the sites in M3, T1 and T2. 31

36 Cumulative Frequency (Pct Finer Reach M1: Site Cumulative Frequency (Pct Finer Reach M2: Site Cumulative Frequency (Pct Finer Reach M3: Site Particle size (mm) Particle size (mm) Particle size (mm) Cumulative Frequency (Pct Finer Reach T1: Site Cumulative Frequency (Pct Finer Reach T2: Site Particle size (mm) Particle size (mm) Figure 9. Cumulative sediment size distributions for annual monitoring sites

37 D25 (mm) D5 (mm) M1:35 M2:6 M3:6 T1:12 T2:12 M1:35 M2:6 M3:6 T1:12 T2:12 D75 (mm) D (mm) M1:35 M2:6 M3:6 T1:12 T2:12 M1:35 M2:6 M3:6 T1:12 T2:12 D95 (mm) M1:35 M2:6 M3:6 T1:12 T2:12 Figure 1. Confidence intervals for percentiles of cumulative sediment size distributions

38 Confidence intervals for the mean diameter of sediment frequency percentiles D5, D, D25, D5, D75, D and D95 are shown in Figure 1; these were generated using a bootstrapping method. Inter-annual variation was analyzed by examining the degree of overlap in the range of sediment diameters in the confidence intervals, which generally indicated absence of detectable year-to-year differences prior to 217. The data for 217 showed unprecedented inter-annual coarsening at M3, T1 and T2 for the D25 and D5 percentiles (Figure 1). Coarsening of D75 also occurred at M3 and T1. The diameter of percentiles D5 and D are not described here because these percentiles are almost entirely < mm (the smallest size class measured). To test for differences between baseline conditions and 217 conditions, a 95% confidence interval for the difference in diameters for percentiles from D25 to D95 was computed from the individual confidence intervals using the method of Zou and Donner (2). Differences were considered significant (p<.5) when the confidence interval did not include zero. The results of this analysis are summarized in Table 5. Decline in sediment diameter (fining) is indicated by - in the table; increases (coarsening) are indicated by +. Absence of detectable change is indicated by. This analysis indicates fining of the stream bed in M1 for D5-D95 and coarsening of the stream bed in relatively mobile sediment size classes (D25 and/or D5) in M2, M3 and T2. Table 5. Summary of change in sediment diameter for percentiles of size distribution Percentile M1 M2 M3 T1 T2 D D D75 - D - - D Proportion of the Stream Bed Occupied by Fine Sediment Reach Comparison to 2 Fine sediment for purposes of this study is defined to be sediment < mm diameter. As shown in Figure 11, the percentage of the stream bed comprised of fine sediment in reaches M2, M3 and T2 ranges from about 13% to 36%. Statistical analyses using contingency tables and the Χ 2 (chi-square) statistic were used to evaluate changes over time. Marginally-significant declines in fine sediment were found for M2-2 (p=.69) and T2-31 (p=.261); fine sediment declined significantly in M3-1 (p=.9). Friedman s test was used for inference at the reach-scale; M3 had a marginally-significant decline in fine sediment (p=.55) as did the three reaches pooled together (p=.292). 3

39 Proportion less than mm Proportion less than mm Proportion less than mm M2:2 M2:6 M2:1 M2:1 M3:2 M3:6 M3:1 M3:1 T2:9 T2:12 T2:31 T2:2 Reach:site Figure 11. Mean and 95% confidence interval for proportion of stream bed sediment < mm by site and reach comparing and 2. Annual Site Comparison -217 Reach:site Reach:site The mean proportion of the streambed with fine sediment (< mm) at annual monitoring sites is shown in Figure 12. Chi-square tests found significant changes (declines) in fine sediment at the sites in M3 and T2 (p<.). Figure 12. Mean and 95% confidence interval for proportion of stream bed sediment < mm

40 5.3 Distribution of Habitat Types and Surface Sediment Aquatic habitat types used to describe stream conditions for fish were identified at each observation point along each transect at monitoring sites. These aquatic habitat types are used by District biologists for their surveys. The habitat types distinguish differences in the wetted channel under base flow conditions based on the appearance and slope of the water surface. Pools are deep with a smooth water surface, glides have a smooth water surface with perceptible increase in velocity relative to pools. Runs become shallower with noticeable increase in velocity. Riffles are shallow and have a turbulent surface caused by gravel and cobble grains protruding through the water surface. Cascades are steeper with flow sometimes divided into chutes by large cobbles, boulders and bedrock protruding from the streambed. The categories bank and bar are unique to this survey, which includes portions of the bankfull channel width lying above the water surface. Changes in the distribution of habitat types over time is evaluated below along with variation in surface size sediment by habitat type over time Distribution of Habitat Types by Reach Reach Comparison to 2 The frequency of habitat types within each reach is displayed in Figure 13. The graphs display the count of habitat types associated with each touch of the stream bed along transects to sample sediment size distribution. These data are pooled for each reach, creating a samples size of about 12 for each reach. Cascades are absent in the lower slope gradient M2 reach. To test whether these reach-scale distributions of habitat type have changed over time, the Χ 2 (chi-square) statistic was used to test for differences among proportions by contingency table analysis. Habitat distributions are significantly different in all comparisons (p<.). The most consistent changes appear to be increased proportion of runs and decreased pools and riffles. 36

41 Habitat Types M2 Habitat Types M3 Habitat Types T bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Figure 13. Frequency distribution of habitat types by reach for and 2. The frequency of habitat types within each sample site is displayed in Figure 1. Each site has a sample size of about 3. The Χ 2 (chi-square) statistic was used to test for differences among proportions by contingency table analysis. There were significant differences in the distributions for /13 compared to 2 (p<.) for all sites except M3-2 and T2-31 where differences were marginally significant (p=.2 and p=.513, respectively). Habitat Types M2 Habitat Types M2 Habitat Types M2 Habitat Types M bank 2 bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Habitat Types M3 Habitat Types M3 Habitat Types M3 Habitat Types M bank 2 bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Habitat Types T2 s Habitat Types T2 s Habitat Types T2 s Habitat Types T2 s bank 2 bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Figure 1. Frequency distribution of habitat types by site for and 2. 37

42 Annual Site Comparison -217 The frequency of habitat types within each site is displayed in Figures 15 and. Figure 15 shows the shifting frequency of bars, riffles, runs, glides and pool over time in a line graph format. Figure provides these data, also included banks and cascades (not shown in Figure 15) in bar graphs showing change in frequency over time. These data are derived from sample size of about 3 stream bed touches per site. Cascades are absent in the lower slope gradient reaches (M2 and M1) and from T1, surprisingly, given its relatively high stream gradient. To test whether these reach-scale distributions of habitat type have changed over time, the Χ 2 (chi-square) statistic was used to test for differences by contingency table analysis. Habitat distributions are significantly different in all comparisons (p<.). There appear to be discernable trends of shifting proportions of habitat type at some sites. A brief description of apparent changes at each site from to 217 derived from interpretation of the distributions in Figure 15 and follows. Site M1-35 tends to have low slope gradient and a finer sediment size distribution. There is no data for 2. Riffle frequency has declined and glide frequency has increased. Pool frequency increased to 215 as frequency of runs decrease to 215, but both returned to frequencies similar to initial conditions in 217. Site M2-6 tends to have a low slope gradient and a finer sediment size distribution. Bars and riffles declined to their lowest frequency in 217. Runs decreased through the middle of the period and increased significantly to the end of the period. Glides increased through the middle of the period and declined but remained higher that initial frequency at the end of the period. Pool frequency declined from initial maximum to much lower frequency in 217; however, substantial increases in pool habitats too deep to sample were observed in 217, so it is likely that pool frequency has not decreased much or may have increased. Site M3-6 tends to have a high slope gradient and a coarser sediment size distribution. Frequency of bars was quite consistent, and riffle frequency varied little from to 217 despite sharp increase in 215 and sharp decline in 2. Runs increased substantially in 2 and 217. Glides varied considerably but declined substantially by 217. Pools declined from an initial maximum to low frequency in 2 and substantial recovery in 217 to about half the initial frequency. Site T1-12 tends to have a high slope gradient and a coarser sediment size distribution. Climate-driven annual fluctuations in water depth may have affected the frequency of habitats in the dry years There is no data for 2. Bars are the dominant type, indicative of the relatively narrow width of wetted channel relative to the bankfull channel and did not vary much over time. Riffles became more abundant in 217 relative to. Runs varied over time and became more abundant in 217. Glides declined to the end of the period, while pool frequency increased somewhat. 3

43 Site T2-12 tends to have a high slope gradient and mixed sediment sizes. Climate-driven annual fluctuations in water depth may have affected the frequency of habitats in the dry years Bars, cascades and riffles increased in frequency to 217. Runs declined to the middle of the period and rebounded at the end of the period. Glides were abundant through the middle of the period and vanished after 21. Pool frequency increased significantly in 215 but declined to near initial frequency in 217 Figure 15. Variation in frequency of bars, riffles, runs, glides and pools by monitoring site,

44 Habitat Types: M1 site 35 Habitat Types: M2 site bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Habitat Types: M3 site 6 Habitat Types: T1 site bank bar cascade riffle run glide pool bank bar cascade riffle run glide pool Habitat Types: T2 site bank bar cascade riffle run glide pool Figure. Frequency distribution of all habitat types by site to 217.

45 5.3.2 Distribution of Surface Sediment Size Distribution by Habitat Type The surface sediment size distribution sorted by reach and by habitat type (excluding banks ) is displayed graphically in Figure 17. The Χ 2 (chi-square) statistic was used to test for differences among proportions by contingency table analysis. Significant differences between and 2 were found only in reach M3 for runs (p=.1) and glides (p=.2). Although not found to be statistically-significant, the data plots suggest the following about surface sediment size distributions in the habitat units excluding bank and cascade : Bars appear to have coarser sediment in 2 compared to. Riffles appear to have coarser sediment in M2 and M3 in 2 compared to. Runs appear to have coarser sediment in M3 and T2. Glides appear to have coarser sediment in M2 and M3, and finer sediment in T2. Pools appear to have slightly coarser sediment in M3 and slightly finer sediment in T2. These data have not been analyzed with respect to annual variations

46 M2:bar 5.7 Cumulative Frequency (Pct Fin Cumulative Frequency (Pct Fin M2:riffle Cumulative Frequency (Pct Fin M2:run 5.7 Cumulative Frequency (Pct Fin M2:glide Cumulative Frequency (Pct Fin M2:pool Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Cumulative Frequency (Pct Fine M3:bar Cumulative Frequency (Pct Fine M3:riffle Cumulative Frequency (Pct Fine M3:run Cumulative Frequency (Pct Fine M3:glide Cumulative Frequency (Pct Fine M3:pool Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Cumulative Frequency (Pct Fin T2:bar 5.7 Cumulative Frequency (Pct Fin T2:riffle Cumulative Frequency (Pct Fin T2:run 5.7 Cumulative Frequency (Pct Fin T2:glide 11.3 Cumulative Frequency (Pct Fin T2:pool Particle size (mm) Particle size (mm) Particle size (mm) Particle size (mm) Figure 17. Surface sediment size distributions by habitat type and reach. Particle size (mm) 2

47 5. Distribution of Facies Sediment facies (distinctive patches of sediment on the streambed differentiated by characteristic size distributions) are used in this monitoring plan to characterize zones of fine sediment deposition on the bed of Lagunitas Creek. In particular, facies containing higher abundance of fine sediment (< mm diameter) are of interest because of their potential deleterious effect on fish habitat (Table 2) and their role in the storage and routing of fine sediment in Lagunitas Creek. The fine sediment facies are sand, fine gravel with sand, and gravel with pockets of sand; these are abbreviated as s, fgs and gs, respectively. Facies without noticeable deposits of fine sediment on the surface of the streambed are gravel-dominant (g) and cobbledominant (c). The foregoing five facies were utilized to describe facies in previous surveys of Lagunitas Creek. 17 Additional facies were encountered during field surveys, including boulder-dominant (bould), bedrock-dominant (b), roots (r) and wood (w) as discussed in Section 3.2. These patch types were adopted to provide a characterization of the local stream bed at each sample location. The distribution of facies by reach and by site is analyzed below comparing and 2. The variation in combined proportion of fine sediment facies (s, fgs and gs) comparing and 2 is also analyzed Distribution of Facies by Reach Reach Comparison to 2 The frequency of facies by monitoring reach comparing and 2 is displayed in Figure 1. The graphs display the count of facies associated with each touch of the stream bed along transects to sample sediment size distribution. These data are pooled for each reach, creating a samples size of about 12 for each reach. Unlike habitat types, facies are determined without reference to the wetted channel. The frequency of the wood and roots facies is very low in all reaches. Boulder and bedrock facies are rarely found in M2. The most common facies in all reaches in both and 2 was gravel. To test whether these reach-scale distributions of facies differ from one another, the Χ 2 (chi-square) statistic was used to test for differences among proportions by contingency table analysis. The proportions of facies was signficantly different in 2 compared to (p<.). 17 O Connor Environmental, Inc. 26. Lagunitas Creek Fine Sediment Investigation. Prepared for Marin Municipal Water District. 7 p. 3

48 Facies M2 Facies M3 Facies T b bould c fgs g gs root s wood b bould c fgs g gs root s wood b bould c fgs g gs root s wood Figure 1. Frequency distribution of facies by reach for and 2. Note: Fine sediment facies s, fgs, and gs are within the blue boxes; in black and 2 in red. Facies M2 site 2 Facies M2 site 6 Facies M2 site 1 Facies M2 site b bould g gs fgs root wood c s b bould g gs fgs root wood c s b bould g gs fgs root wood c s b bould g gs fgs root wood b bould c g gs fgs s c s Facies M3 site 2 Facies M3 site 6 Facies M3 site 1 Facies M3 site root wood b bould g gs fgs root wood c s b bould g gs fgs root wood c s b bould g gs fgs root wood b bould g gs fgs root wood c s c s Facies T2 site 9 Facies T2 site 12 Facies T2 site 31 Facies T2 site b bould g gs fgs root wood c s b bould g gs fgs root wood c s b bould g gs fgs root wood c s Figure 19. Frequency distribution of facies by site for and 2. Note: Fine sediment facies s, fgs, and gs are within the blue boxes; in black and 2 in red.

49 The frequency of facies for each monitoring site comparing and 2 is displayed in Figure 19. The proportions of facies was significantly different in 2 compared to (p<.3) for all sites except T2-9 (marginally significant difference, p=.22) and M2-1 and T2-12, where there was no difference. The predominant shifts were increases in gravel facies and decrease in fine gravel with sand facies. Changes in the proportion of the channel occupied by fine sediment facies combined (s+gs+fgs) from to 2 are displayed graphically in Figure 2. In reaches M2 and M3, three of the four sites had significant decreases in fine sediment facies (p<.35 and p<.6, respectively). In T2, two of four sites had marginally-significant decreases in fine sediment facies (p<.62). Inferences at the reach scale were evaluated using Friedman s test, which found a marginally-significant decline in fine sediment facies in M3 (p=.55), despite very low statistical power with n=. Proportion of bed in fine fa M2 2 M3 T2 2 2 Figure 2. Proportion of fine sediment facies for and 2. Note: Different-colored open circles represent different monitoring sites in each reach; the connecting line of the same color emphasizes the time-trend for each site. The proportions of the streambed with fine sediment facies in each reach is summarized in Table 6. Although statistical inferences are limited to M3, the overall trend appears to be a decline in fine sediment facies from to 2. Table 6. Summary of proportion of fine sediment facies by reach for and 2. Year M2 M3 T2 All Pooled % change -22.1% -1.% -19.% -2.6% 5

50 Annual Site Comparison -217 There were significant changes in the frequency of the fine sediment facies (s, fgs, gs) at all five sites over the period (chi-square test, p<.); observed proportions are summarized in Figure 21. The mean and 95% confidence intervals for the proportion of fine sediment facies for each year and site is plotted in Figure 22; the separation between envelopes of the range for the 95% confidence intervals reveals significant changes over time at each site. M1-35 was the only site where fine sediment facies had increased by the end of the period from.53 to.66; the increase occurred primarily in 213. At M2-6, fine sediment facies declined dramatically from.72 to.5; this apparent change should be interpreted with caution owing to a substantial change in the sample area caused by an increased proportion of deeper pools that could not be sampled. At M3-6, T1-12 and T2-12, the sites with steeper stream gradients, the proportion of fine sediment facies never exceeded.35. M3 declined from.1 to., T1 declined from.31 to.15, and T2 declined from.12 to.3. Declines in fine sediment facies occurred predominantly after 21. Excluding M1-35, overall decline in fine sediment facies as a proportion of the streambed for the other four sites combined was from.33 to.1. Figure 21. Proportion of fine sediment facies by site,

51 Figure 22. Proportion of fine sediment facies with 95% confidence intervals Depth of Fine Sediment Facies Fine sediment on the streambed is believed to have deleterious effects on the freshwater habitat of coho salmon and steelhead trout with respect to reproductive success (Table 2). In addition, fine sediment deposits on the streambed may occupy space in the water column that could otherwise be available to juvenile salmonids or other aquatic organisms. Consequently, the depth of fine sediment deposits was sampled in fine sediment facies (s, fgs and gs) as described in Section In addition to providing a direct measure of the potential effect of fine sediment deposits on the quantity of habitat available to juvenile salmonids, these data provide the basis for estimating the quantity of fine sediment deposited in fine sediment facies in monitoring reaches. Efforts by the District to reduce erosion in the watershed are intended to contribute to improved fish habitat by reducing fine sediment in the streambed. Estimates of the quantity of fine sediment stored on the stream bed surface of Lagunitas Creek will provide a direct measure of the impact of fine sediment on aquatic habitat. At all sample sites in fine sediment facies, the depth of fine sediment was estimated by inserting a thin metal rod 1.1 meter in length and 5 mm diameter into the fine sediment deposit as far as possible. Depth of penetration was recorded to the nearest centimeter as the estimate of fine sediment depth. The water depth was also recorded where present. The ratio of fine sediment depth to the sum of water depth and fine sediment 7

52 depth provides a measure of the volume of potential habitat in the water column occupied by sediment. This ratio is referred to as V* (see Section 3.3.2). In the gs facies, the surface grain size at the measurement station was used to determine whether the specific location was a patch of fine sediment with a sediment depth to be probed, or a patch of gravel where concentrated fine sediment was not present. In the two other fine sediment facies, this was not an issue because these facies were more homogeneous, and the extent of fine sediment deposits was not in question. In the facies gs (gravel with pockets of sand), the penetration test was conducted only if the surface particle size recorded at the station was finer than mm (sediment particles catching on the 11 mm sieve mesh diameter). This sampling rule was introduced to limit the data to fine sediment pockets in the facies. Subsequent analyses revealed that the average median surface diameter in the gs facies was < mm in M1, M2 and M3, and <2 mm in T1 and T2, indicating that the sampling rule was reasonable. Surface sediment size distribution in the gs facies was more heterogeneous than in the other fine sediment facies; mean D75 in the gs facies ranged from about 2 to 35 mm and mean D from about 25 to 55 mm. Reach Comparison to 2 Sediment depth data have been analyzed considering only the observations within fine sediment facies where penetration tests were made. The data are reported directly as the depth of fine sediment facies; this ignores the majority of the stream bed where fine sediment deposits (facies) are not present. Pooled data for all reaches combined comparing and 2 are displayed in Figure 23. The distributions of fine sediment depth in and 2 are different, but with marginal significance (Friedman s test, p=.292).

53 2 3 2 Percent of Total Depth of fines (cm) in facies s, Figure 23. Histograms of depth of fine sediment in fine sediment facies pooled for all reaches for and 2. Data pooled by reach comparing and 2 are presented in Figure 2. The Wilcoxon test for a change in depth of fine sediment by reach found significant differences in M2 and M3 (p=.36 and p<. respectively), and marginally significant differences in T2 (p=.172). The estimated mean depth of fine sediment in and 2 is summarized by reach in Table 7; depths of fine sediment in all reaches increased about cm. Note that this consistent increase in depth of fine sediment is offset by the consistent reduction in proportion of fine sediment facies (Table 5) such that the total volume of fine sediment is not well-represented by depth alone. 9

54 M2 2 M3 2 T2 3 Percent of Total M2 M3 T Depth of fines (cm) in facies s, fgs, and gs Figure 2. Histograms of depth of fine sediment in fine sediment facies by reach for and 2. Table 7. Estimated mean depth of fine sediment in fine sediment facies by reach for and 2. Year M2 (cm) M3 (cm) T2 (cm) % change +17.9% +36.9% +19.% The depth of fines for each site and reach sampled in and 2 is displayed in Figure 25. The Wilcoxon test was used to evaluate changes of depth with time. Two sites in M2 had significant changes in depth of 6.9 cm and 9.9 cm (p<.191). Two sites in M3 had significant changes in depth of 7.6 cm and 11.7 cm (p<.3). One site in T2 had a marginally significant change in depth of 6.6 cm (p=.51). Nevertheless, Friedman s test (used for proper inference to reach) found only a marginally significant difference only in T2 (p=.55) because an increase in depth occurred at all four sites in T2. 5

55 Depth of fines in fine facies M2 2 M3 T Figure 25. Mean depth of fine sediment in fine sediment facies for and 2. Note: Different-colored open circles represent different monitoring sites in each reach; the connecting line of the same color emphasizes the time-trend for each site. The depth of fine sediment on the channel bed surface for facies other than the fine sediment faces (s, fgs and gs) is zero. The preceding analysis of the depth of fine sediment on the bed surface in fine sediment facies can be extended to represent the depth of fine sediment on the entire stream bed. This analysis better represents overall fine sediment conditions on the bed because it accounts for the areal extent of fine sediment, not merely the depth of fine sediment as represented above. Figure 26 shows the pooled data for depth of fines across all facies for and 2. Pooled data for all sites and reaches for and 2 were not significantly different (Friedman s test). Histograms of fine sediment depth for all facies by reach are presented in Figure 27. The Wilcoxon test for a change in depth of fine sediment by reach found significant differences in M2 and M3 (p=.21 and p<. respectively) but not for T2. The estimated mean depth of fine sediment in and 2 is summarized by reach in Table ; depths of fine sediment decreased about 1 cm in M2 and M3. Table. Estimated mean depth of fine sediment in all facies by reach for and 2. Year M2 (cm) M3 (cm) T2 (cm) % change -.9% -2.% -2.% 51

56 2 6 Percent of Total Depth of fines (cm) in ALL fac Figure 26. Histograms of depth of fine sediment in all facies pooled for all reaches for and M2 2 M3 2 T2 6 Percent of Total M2 M3 T Depth of fines (cm) in ALL facies Figure 27. Histograms of depth of fine sediment in all facies by reach for and 2. 52

57 The mean depth of fine sediment for all facies by reach and site is represented in Figure 2. The Wilcoxon test was used to evaluate changes of depth with time. One site in M2 has a significant decline in fine sediment depth (-.1 cm, p<.), and two sites had marginally significant changes, one increase and one decrease in depth (1.9 cm, p=.253 and -1.5 cm, p=.7). Three sites in M3 had significant changes in depth, one increasing (.5 cm, p=.35) and two decreasing (-3.9 cm, p=.2 and -. cm, p<.). Two sites in T2 had a marginally significant change in depth, decreases of -.6 cm and -1. cm (p=.322 and p=.6333, respectively). Depth of fines in ALL facie 15 1 M2 2 M3 T Figure 2. Mean depth of fine sediment in all facies for and 2. Note: Different-colored open circles represent different monitoring sites in each reach; the connecting line of the same color emphasizes the time-trend for each site. Annual Site Comparison -217 We discuss mean depth of fine sediment in only the fine sediment facies in the evaluation of year-to-year changes. The mean depth of fine sediment is summarized in Table 9 and analyzed using the bootstrap technique to resample the data with replacement and generate 95% confidence intervals for the mean depth of fines (Figure 29). Year-by-year comparisons were evaluated using the Kruskal-Wallis test. Significant changes occurred in M1-35 and M2-6 (p<.1); a marginally significant change occurred in T1-12 (p=.59). Post-hoc comparisons were made for fine facies depth in M1 and M2 using the method of Siegel and Castellan (19) using the R package pgirmess. This found that for M1, fine sediment depth was significantly different (p<.5) in 21 (the minimum for this site) compared to and 217. At M2, fine sediment depth was significantly different (p<.5) in 215 and 2 (the maximum for this site) compared to and 213. These tests confirm the impressions conveyed by Figure 3. 53

58 Table 9. Mean depth of fine sediment facies by year and site. Year M1-35 M2-6 M3-6 T1-12 T Figure 29. Mean depth of fine sediment in fine facies with 95% confidence interval, Sediment Depth as a Proportion of the Water Column-The V* Index As discussed above, fine sediment deposits (facies) occupy space in the wetted channel that would otherwise be available habitat for fish and other aquatic organisms. This section analyzes the sediment depth data described in Section in terms of the ratio of fine sediment depth measured in the fine sediment facies to the sum of fine sediment depth and water depth. This ratio has been termed V* (the * denotes a dimensionless quantity, in this case depth divided by depth). The V* index has previously been determined for pool habitats, in part because this is where fine sediment facies are most evident and where reduced aquatic habitat volume for salmonid rearing is readily apparent. In addition, the morphology of pools allows for 5

59 a simple determination of a depth that does not vary with stream stage, allowing the index to be determined independent of stream stage. In Lagunitas Creek, summer base flows are nearly constant year to year owing to controlled releases from Kent Lake that maintain minimum flows required by the State. This constant flow and near constant stage allows us to apply the V* index across all habitat units (not only pools) in Lagunitas Creek below Peters Dam and Kent Lake (i.e. reaches M1, M2 and M3) with a reasonable expectation that the computed index from year to year would not be biased by variation in flow. V* data for reaches T1 and T2 may potentially vary from year to year owing to variability in streamflow. Both and 2 were approximately normal water years with about average rainfall; water years 213, 21 and 215 were very dry. We have computed our modified version of V* for all wetted habitat with fine sediment facies (Figure 3) as well as for pool habitat with fine sediment facies (Figure 31). Pool habitat is analyzed separately because this habitat is particularly important for summer rearing habitat used by juvenile coho salmon. Again, V* for pools calculated here is not comparable to that in most other similar studies because we did not measure water depth except at locations where fine sediment facies were present. Inclusion of data from pools from observation points not in fine sediment facies would substantially reduce the V* index because sample points with zero fine sediment depth would lower the mean of the computed index. Figure 3. Distribution of V* ratio in fine sediment facies for and 2. 55

60 Figure 31. Distribution of V* ratio in pool habitat for and 2. Pooling the data for all sites and reaches, we found no significant differences in V* ratio for all wetted habitat or pool habitat between and 2 (Friedman s test). Histograms of the V* data pooled by reach for all fine sediment facies and for fine sediment facies in pools are displayed in Figure 32 and Figure 33, respectively. A marginally significant increase in V* in fine sediment facies in M2 was found using the Wilcoxon test (p=.2219); the mean increase in the V* ratio for the reach was.3. Mean V* with 95% confidence intervals are shown in Figure 3. There were no reach differences detected in V* in pool habitats; however, there were marginally significant differences at some sites as indicated in Figure 35 displaying mean pool V* with 95% confidence intervals. Site M2-1 and M3-1 had marginally significant increases in V* ratio in pools (Wilcoxon test, p=.7 and p=.3571, respectively). Data for V* in pools in M3 include two sites with few or no data. Friedman s test also indicated a marginally significant (p=.55) change in V* ratio for all fine sediment facies in M2 in 2 compared with. 56

61 Figure 32. Distribution of V* ratio for fine sediment facies by reach, to 2. Figure 33. Distribution of V* ratio for fine sediment facies in pool habitat by reach,