Effectiveness of Experimental Riparian Buffers on Perennial Non-fish-bearing Streams on Competent Lithologies in Western Washington – Phase 2 (Nine Years after Harvest)

Headwater streams are largely understudied relative to their frequency in the landscape, constituting approximately 65% of the total stream length on forestlands in western Washington. We evaluated the effectiveness of riparian forest management prescriptions in maintaining key aquatic conditions and processes affected by Forest Practices for small non-fish-bearing (Type N) headwater stream basins underlain by competent, “hard rock” lithologies (i.e., volcanic or igneous rocks) in western Washington (see Chapter 1 – Introduction in this report). We compared current prescriptions to two alternatives, one with longer riparian leave-tree buffers and one with no buffers. We looked at the magnitude, direction (positive or negative), and duration of change for riparian-related inputs and response of instream and downstream components. We evaluated riparian processes affecting in-channel wood recruitment and loading, stream temperature and shade, discharge, suspended sediment export, nutrient export, channel characteristics, and stable isotopes. To evaluate biological response, we selected stream-associated amphibians as a key response variable because they are one of the important biotic resources for protection in non-fish-bearing streams. The results of this study are intended to inform the efficacy of current Forest Practices (FP) rules, including how landowners can continue harvesting wood resources while protecting important headwater habitats and associated species, and meeting resource objectives outlined in the FP Habitat Conservation Plan (FP HCP; Schedule L-1, Appendix N).

We used a Before-After Control-Impact (BACI) study design with blocking to examine how key aquatic resources, conditions and processes responded to riparian buffer treatments. We collected a minimum of two years of pre-harvest data from 2006 until harvest began in 2008, and post-harvest data from 2009 (one year post-harvest) through 2016 or 2017, depending on the response variable (i.e., up to nine years post-harvest; see Chapter 2 – Study Design in this report). Study sites included 17 Type N stream basins located in managed second-growth conifer forests across western Washington in three physiographic regions (Olympic Mountains, Willapa Hills and Southern Cascades). Sites were restricted to Type N basins ranging from 12 to 54 ha (30 to 133 ac) underlain by relatively competent lithologies, primarily volcanic flow rocks and breccias, and that were known to support Coastal Tailed Frog (Ascaphus truei) and Olympic, Columbia, or Cascade Torrent Salamanders (Rhyacotriton olympicus, R. kezeri, or R. cascadae).

We evaluated four experimental treatments, including an unharvested Reference (i.e., withheld from harvest; n = 6) and three alternative riparian buffer treatments with clearcut harvest: 100% treatment (a two-sided 50-ft [15.2-m] riparian buffer along the entire Riparian Management Zone [RMZ; n = 4]); FP treatment (a two-sided 50-ft [15.2-m] riparian buffer along at least 50% of the RMZ, consistent with the current Forest Practices buffer prescription for Type N streams [n = 3]); and 0% treatment (clearcut harvest throughout the entire RMZ [n = 4]). The timber harvests and associated riparian buffer treatments were implemented between October 2008 and August 2009.

In-channel wood plays an important functional role in Pacific Northwest streams, influencing channel morphology and hydraulics, storage and routing of sediment and organic matter, aquatic habitat, aquatic communities, and food resources. We found that harvest of timber in and adjacent to streamside riparian forests directly affected tree mortality, tree fall rates, wood recruitment to streams, and in-channel wood loading (pieces/m; see Chapter 3 – Stand Structure, Tree Mortality, Wood Recruitment and Loading in this report). The greatest post-harvest change in stand structure occurred in the 0% treatment and the unbuffered portions of the FP treatment where all riparian trees were removed during harvest. A pulse of large wood (≥10 cm [4 in] diameter) was recruited to streams adjacent to unbuffered reaches during harvest, with very little additional large wood recruitment in the following eight years. Windthrow was the dominant tree mortality agent in riparian buffers, with the highest mortality in the first two years post-harvest. The highest tree mortality rates and greatest reductions in density and basal area occurred in the FP treatment, where cumulative mortality eight years post-harvest was 51% and 56% of initial basal area and 50% and 68% of initial density for the RMZ and PIP buffers, respectively, compared to a cumulative mortality of 16% and 9% of basal area and 20% and 15% of initial density for the RMZ and PIP in the reference. Windthrow-associated tree fall in riparian buffers increased large wood recruitment to channels in the 100% and FP treatments. However, we found that most recruited trees (>80%) were suspended above the active stream channel. From two to eight years post-harvest, in-channel large wood loading (mean pieces per linear stream meter) continued to increase in the FP treatment, remained relatively stable in the 100% treatment, and decreased in the 0% treatment. The greatest increase in in-channel small wood loading (<10 cm [4 in] diameter) was in the 0% treatment in the two years post-harvest, and was comprised largely of logging slash from timber harvest of the streamside trees. Small wood loading continued to increase in the FP and 0% treatments through five years post-harvest and then declined in all buffer treatments.

Riparian vegetation is an important source of organic matter and macroinvertebrates, nutrients, and cool water to downstream reaches. The shading that this vegetation provides is a dominant control on stream temperature, which in turn is a critical environmental condition for many aquatic organisms and biological processes. Riparian shade decreased and water temperature increased in all buffer treatments after harvest (see Chapter 4 – Stream Temperature and Cover in this report). Canopy closure decreased by less than 10% in the 100% treatment but declined 32% in the FP treatment and 87% in the 0% treatment by three years post-harvest. After nine years, canopy closure returned to pre-harvest levels in the 100% treatment, but remained 15% and 27% below pre-harvest values at the FP and 0% treatments, respectively. The seven-day average temperature response increased in the 100% treatment by 1.1°C in the year immediately following harvest but returned to pre-harvest temperatures in the three years post-harvest. In the FP treatment, the temperature response ranged from +0.5 to +1.2°C and changed little throughout the post-harvest period, possibly from the ongoing loss of buffer trees to windthrow. In the 0% treatment, the temperature response was nearly 4°C in the first year after harvest but then steadily declined to a 0.8°C increase by nine years post-harvest. The greatest change in temperature occurred during the July–August period, but temperatures were also elevated in spring and fall at most locations. Substantial (>1.0°C) temperature responses within the harvest unit were attenuated downstream of the harvest unit where the stream had flowed through 100 m or more of unharvested forest or buffers wider than 50 ft. The primary driver of post-harvest temperature increases appeared to be loss of riparian cover. However, there was evidence that basin aspect may have influenced the magnitude of change; and in one locality, hyporheic flow may have mitigated higher temperature within a well-shaded downstream reach.

Changes in stream discharge, and in the sediment loads that are carried by those flows, have been long recognized as common, but highly variable, responses to timber harvest. We measured discharge and suspended sediment export in eight of the study sites, with four sites each (one of each buffer treatment) in the Olympic and Willapa Hills physiographic regions. Total water yield increased in all buffer treatment sites, but treatment effects varied with buffer treatment and climate, with sites receiving the most rainfall (i.e., Olympic block) and the greatest proportion of watershed area harvested (FP and 0% treatments) exhibiting the largest increases (see Chapter 5 – Stream Discharge, Turbidity, and Suspended Sediment Export in this report). On average, this study affirms prior literature reports that show discharges increasing 1 to 18 mm/yr for each percent of the watershed harvested, albeit with much variability as a function of buffer treatment, climate, and precipitation. Relative changes in flow were greatest for baseflows and median discharge, but specific discharges increased for all flows up to the 30-day recurrence interval in the FP and 0% treatment sites. Late summer discharge decreased in both 100% treatment sites through eight years post-harvest, presumably because of increased evapotranspiration rates in the residual vegetation during periods of little rainfall. In contrast, harvest did not change the magnitude of suspended sediment export regardless of buffer treatment. Over 11 years of the study, turbidity readings were very low over 95% of the time. Both turbidity and suspended sediment concentration increased with increasing discharge, typically during late fall and early winter storm events, and then rapidly declined. The basins appear to be supply-limited, with the quantity of exported sediment restricted to the quantity of sediment delivered to the stream from the adjacent uplands, and so additional flow (especially non-peak flow) has little ability to affect sediment transport. While discharge increased in all treatments after harvest, suspended sediment export events were episodic, poorly correlated with discharge, and not synchronized across all sites, suggesting that export magnitudes are unrelated to harvest.

Nutrients exported to downstream receiving waters may increase primary productivity, leading to decreases in instream dissolved oxygen from decomposition. Because the watersheds of western Washington drain to the sensitive, confined marine waters of Puget Sound, Grays Harbor, and Willapa Bay, nutrient loading is a potential environmental concern. We measured mean total nitrogen (N) and nitrate-N concentrations for nine years post-harvest in the same eight sites used in our discharge and suspended sediment export components of the study (see Chapter 6 – Nitrogen Export in this report). Nitrogen export increased in all treatment sites immediately after harvest, with variable increases among sites ranging from less than 10% to more than three-fold. The estimated change was greatest in the 0% treatment, intermediate in the FP treatment, and lowest in the 100% treatment. Controlling for treatment type, the increases corresponded to the proportion of the watershed harvested. Only the 0% treatment differed statistically from the other treatments. At seven and eight years post-harvest, the eight sites displayed no consistent response in nitrogen concentration or export to the buffer treatments or to the proportion of the watershed harvested: total-N export had declined from their immediate post-harvest levels at three sites and increased slightly at three sites, while nitrate-N export declined from post-harvest levels at four sites and increased slightly at two sites. Only one site, however, had recovered to pre-harvest export rates by eight years post-harvest.

Changes to wood recruitment and loading, stream flow and sediment transport from timber harvest can result in changes to physical stream channel characteristics, particularly channel dimensions and substrate sediment materials. Measurements in the 17 study basins in the years immediately pre-harvest, and again one, two, five and eight years post-harvest, showed some systematic patterns (see Chapter 7 – Stream Channel Characteristics in this report). In the two years post-harvest, we estimated a decline in stream wetted and bankfull widths in the 0% treatment compared to the pre-harvest period after controlling for temporal changes in the reference. This pattern continued through eight years post-harvest. We also measured a post-harvest increase in the proportion of the stream channel dominated by fine sediment substrates in the 0% treatment in the two years post-harvest, which was still evident eight years post-harvest. A similar increase was also estimated for the FP treatment eight years post-harvest, but not in the other sample years. Finally, we estimated an increase in the proportion of the channel rise attributed to in-channel steps in the 0% treatment in all post-harvest sampling years. We suspect that post-harvest changes in stream channel characteristics, primarily in the 0% treatment, can be attributed at least in part to post-harvest increases in in-channel small and large wood recruitment and loading. Wood recruitment and loading increased in all buffer treatments as logging slash and windthrown trees from unharvested RMZs entered the stream channel during and after timber harvest. This may explain the decrease we observed in stream wetted and bankfull widths, despite the observed increase in flows that would normally encourage increases in channel width.

Stable isotope ratios are especially useful for identifying shifts in trophic system organization due to canopy modification, which other researchers have associated with an increase in the contribution of algae to the trophic support of streams. Samples of biofilm, litterfall, instream detritus, macroinvertebrates, and amphibian tissue were collected during the pre-harvest period and one, two and eight years post-harvest (see Chapter 8 – Stable Isotopes in this report). Not every group was sampled for all sites in all years (see Tables 8-1 and 8-2 for sample sizes). We found limited and inconsistent differences in carbon (¹³C) and nitrogen (¹⁵N) isotopic signatures among treatments; however, stable isotope signals suggested that the organic matter sources supporting stream biofilm did not appreciably change in response to the buffer treatments. Though we did not detect a notable difference in the biofilm isotopic values between the pre- and post-harvest period, we estimated a decrease in mean δ¹³C for giant salamander larvae in the FP treatment and an increase in the 100% treatment in the two years post-harvest. Over this same period, we estimated a decrease in mean δ¹⁵N for gatherer invertebrates in the FP and 0% treatments. We found no evidence of an increase in algal content in the biofilm, thus challenging the hypothesis that canopy modification might increase trophic support from autotrophic sources. Our δ¹³C versus δ¹⁵N comparison of stable isotope data were also used to characterize stream-associated amphibian diet. Results indicated that Coastal Tailed Frog larvae were ingesting primarily biofilm. The post-metamorphic Coastal Tailed Frogs, torrent salamanders and giant salamanders, however, all exhibited stable isotope values that suggested a diet of aquatic predators and shredders, and terrestrial spiders.

Amphibians have experienced declines in local abundance and range contractions as a result of habitat loss and degradation, disease, and competition with introduced species. Stream-associated amphibians are frequently the dominant vertebrates in and along non-fish-bearing headwater streams. We observed the largest post-harvest response for Coastal Tailed Frog (see Chapter 9 – Stream-associated Amphibians in this report). In the two years post-harvest we estimated an increase in larval Coastal Tailed Frog density in the FP treatment compared to the pre-harvest period, after controlling for temporal changes in the reference; however, by eight years post-harvest we estimated substantial declines in larval density in all buffer treatments. In the two years post-harvest, post-metamorphic tailed frog density declined in the 100% treatment but increased in the 0% treatment. However, by eight years post-harvest we again estimated substantial declines in density in the 100% and FP treatments, whereas the change in density in the 0% treatment no longer differed from that of the reference. We estimated an increase in torrent salamander density in the 0% treatment in the two years post-harvest; by eight years post-harvest this increase was no longer evident in the 0% treatment although we estimated a decline in the FP treatment. Finally, for giant salamanders we estimated an initial decline in density in the FP treatment in the two years post-harvest, however, by eight years post-harvest we had no evidence of a difference for any treatment. Our study was designed to evaluate treatment effects, not the mechanisms behind potential changes in amphibian abundance. However, stream temperature, overstory canopy, wood loading, sediment retention, flow dynamics, stream morphology, and nutrients all have been associated with amphibian abundance, and changes documented in these metrics following timber harvest likely explain some or all of the changes in amphibian abundances.

In summary, the greatest effects of alternative buffer treatments were observed in riparian stand condition, large wood recruitment and in-channel wood loading, stream shade and temperature, stream channel characteristics, and stream-associated amphibian densities (see Appendix A in this report). The 100% treatment was generally the most effective in minimizing changes from pre-harvest conditions, the FP was intermediate, and the 0% treatment was least effective. The collective effects of timber harvest were most apparent in the 0% treatment in the two years immediately post-harvest. For many metrics, the magnitude of harvest-related change observed for a given treatment diminished over time. The one clear exception to this generality was for the stream-associated amphibians. For these species, treatment effects were largely not evident in the two years post-harvest except for declines in giant salamander density in the FP treatment. However, substantial negative declines were recorded for Coastal Tailed Frog density in the eight years post-harvest, including for larvae in all buffer treatments and post-metamorphic individuals in the 100% and FP treatments. We also estimated a decline in torrent salamander density in the FP treatment in the eight years post-harvest. Continued monitoring of the amphibian response to treatment is strongly recommended to expand on our understanding of the long-term impacts of timber harvest and variable length buffers on stream-associated amphibians.