Forest Harvest Operations -- Results And Discussion
Results And Discussion

Monitoring began in the summer of 1995, with the installation of the Ryan Tempmentors in June of 1995, followed by installation of the ISCO flowmeter and sediment sampler equipment in September of 1995. The first benthic survey was performed in November of 1995. Baseline conditions were established for both watersheds. The calibration period ended on September 27, 1996 when the installation of roads and other access features began. The harvest/BMP period continued from September 27, 1996 until October 31, 1997. The logging was completed mid-October , 1997 and the treatment areas were subsequently stabilized prior to October 31, 1997. Monitoring continued until July 1999, although drought conditions and lack of flow hampered monitoring during the final year of the study.

Best Management Practices
The costs for the installation of the roads, trails, landings and stream crossings to BMP standards were greater than anticipated. This was due, in part, to extremely wet conditions during installation. Additional equipment time and materials were necessary to complete the work. During BMP installation, it also became obvious that even the most detailed pre-harvest planning may require revisions. The “Best” in Best Management Practices is a relative term. Ultimate control over BMP installation lies not with the forester or landowner, but in the hands of the equipment operator charged with the task of following prescribed guidelines.

Evaluation of some individual BMPs:

Temporary bridge. A portable timber bridge was used for the principal stream crossing on the treatment watershed. This was a 20 ft. long by 12 ft. wide bridge, made in three 4 ft. wide sections of 12 inch thick oak timber bolted with 1.25 inch steel threaded rods, based on the same pattern as crane mats. The bridge cost $1800, and was made by a Maryland company. It was supported on each end by a 6 inch by 10 inch sill timber placed on the stream bank. Total installation time, including clearing and grading, took 4 hours, but could be done in less time with experience. Removal took less than 1 hour. Except for minor disturbance when equipment forded the stream to pull the bridge across, there was no visible disturbance or sediment input during installation, use, or removal. It proved to be sufficiently strong to handle fully loaded log trucks and tandem axle dump trucks carrying 20+ tons of stone. Depending on expected usage, this could probably be made with 8 - 10 inch thick timbers to reduce weight, or preservative treated to enhance longevity. Bridges such as this are reusable, and this bridge has since been used on another timber sale with favorable comment.

Stabilization by seeding. Portions of roads, trails, and landings having a slope greater than 10% and stream crossing sites were seeded, limed, fertilized, and mulched following harvesting. The seed used was a mixture which included 35 lbs. of tall fescue (“Forager” endophyte-free pasture type tall fescue was used, not K-31), 35 lbs. of creeping red fescue, and 2 lbs. of medium red clover per acre. This is the primary seed mix recommended by Maryland BMP specifications, and was fairly easy to obtain. The mix gave good results as long as soil and moisture conditions were satisfactory. Also applied per acre, as the required minimum, was 600 lbs. of 10-10-10 fertilizer and 1 ˝ tons of lime (applied as pelletized lime to minimize dust and facilitate spreading), and about 1 ton of straw mulch. The seed, fertilizer, and most lime was spread with hand-cranked spinner type spreaders, and the mulch was spread by hand. While the seed was easy to apply, and the fertilizer fairly easy, the lime and mulch were more difficult. The large amount of lime and straw that needed to be transported to sometimes steep and remote skid trails, which had already had waterbars installed, proved to be a lot of work. The spreading of lime at the designated rate was very time consuming, and the rate seemed excessive, especially where it could not be incorporated into the soil. However, in some areas which were seeded but not limed, the establishment of planted or volunteer cover was noticeably less successful on these acidic soils. These are noteworthy considerations, since failure to properly seed and mulch required areas has been shown to be problem on some logging jobs in Maryland (Koehn and Grizzel 1995).

Streamside forest buffer. The selective harvest of the 25 acre stand along Furnace Branch near the outlet of the treatment watershed required that a streamside forest buffer be left to provide protection for the stream. Buffer width varied from 75 to 150 feet based on slope. The length of the buffered stream was 1070 feet, with a total buffer length of 1485 feet (two sides of stream, one side partially out of sale area), covering an area of 2.9 acres. The buffer boundaries were marked, and basal areas measured. The shorter, eastern bank buffer was designated to remain uncut, due to a low initial basal area averaging only 48 sq.ft./acre in trees over 6 inches d.b.h. attributed to gypsy moth (Lymantria dispar) mortality which occurred approximately 10 years previously. On the western bank, an area of 1.8 acres, initial basal areas averaged 84.3 sq.ft./acre, with some locations as high as 130 sq.ft./acre. In order to reduce the basal area to the minimum allowable of 60 sq.ft./acre, 36 sawtimber trees in the buffer on the western bank were marked, containing a volume of 5,516 board feet, or 3,064 bd. ft. per acre. Logging equipment was kept out of the buffer except at the stream crossing. During harvest a few of the trees marked for cutting were left uncut, either by error or choice, and several unmarked trees were knocked down by falling timber. The post-harvest basal area in the harvested western bank buffer was 62 sq.ft./acre, with the average for the total buffer on both banks being reduced from 72 sq.ft. before harvesting to 58 square feet after harvesting. On-site inspections gave no indication of overland flow through the buffer during or after the harvest period. Stream temperatures were not significantly elevated by passage through this harvest area, and benthic macroinvertebrate populations at the lower end of this stream reach were not significantly different from pre-harvest conditions, as described below.

Wet area crossings. Access to the harvested areas required crossing a wetland area with a new road, and the need to stabilize some pre-existing roads and skid trails which had wet sections. The general prescription was to provide drainage outlets when possible, place geotextile (both woven and non-woven were used, moderate to heavy duty) in the roadbed, and cover with 6 or more inches of crushed stone aggregate (typically 4+ inches of 2 inch diameter stone on bottom, with 2+ inches of 3/4 inch stone on top). Installation proved to be very difficult due to the unusually wet conditions, with the alluvial silt and muck soils having the consistency and appearance of chocolate pudding. During the placement of hte stone, the wettest sections would not support construction equipment without the fabric sliding, buckling and rutting, and additional of stone and equipment time were needed. Shortly after installation, logging trucks and log skidders began using some of these areas, under continued wet conditions. Some of the moderately wet sections held up satisfactorily, but the wettest areas were completely churned up. Also, any sections used for skidding quickly lost their stone surfacing and had the geotextile torn. On the southern sale area, which was logged after the soils had dried somewhat and the roadbed had settled, the roadbed held up very well. Where this method is to be used, the construction should be done during the driest part of the year, and left to stabilize, if possible, before heavy winter/spring use. On the wettest areas, 12 or more inches of stone should be used. On this project, stone aggregate was relatively cheap and readily available due to the proximity of several quarries, but this is not always the case. Addition of a stiffener such as wire or plastic mesh placed under the geotextile may improve performance. Alternative methods such as plank or timber mats, and corduroy pole sections should be considered in the wettest areas, especially where roads are intended to be temporary. After harvesting, two of these failed wet sections were crowned and ditched on both sides, and 12 inch schedule 40 PVC pipe installed for cross drainage at frequent intervals, which seems to be working very well.

[Wet area in road being stablizd with geotextile and stone.] [Broad-based dip being constructed on truck haul road.]

Road and trail drainage. At slope-based intervals, excavated drainage structures such as dips and water bars were constructed into roads and skid trails, primarily using small bulldozers. Broad-based dips and rolling dips were used on truck roads; and waterbars were used to provide drainage for skid trails before and after use, along with out-sloping and grade breaks. At times there was some difficulty in getting the various equipment operators unfamiliar with forestry BMPs to understand and follow the standards. All these practices worked as designed, though a few needed to be repaired after disturbance from logging equipment, vehicles, horses and mountain bikes during wet weather.

[Figure 1. Daily Average Flow in the Treatment and Control Watersheds]

Daily average flow in ft3/sec for the treatment and control watersheds are presented in Figure 1. Flows were similar in both watersheds during the control and treatment periods. There were periods in 1997 and again in 1998 where there was no flow, particularly in the treatment watershed. Base flow levels were also low in the treatment watershed during 1996. There were several large flows triggered by storm events, two in 1996 and two in 1998. It is interesting to note that the magnitude of the response in each of the watersheds was different for each storm.

Total Suspended Solids
Mean total suspended sediment concentrations measured at the outlets of the control and treatment watersheds are presented in Table 2. Concentrations range from 1.3 mg/1 to 1235.7 mg/1 in the control watershed and 1.4 mg/1 to 1971.2 mg/1 in the treatment watershed. There were significant differences in mean TSS concentrations both between watersheds and between sampling periods.

Table 2. Mean TSS concentrations in treatment and control watershed during the calibration and treatment periods.
Watershed Calibration Period n Treatment Period n
Control 140.60 mg/I 100 62.93 mg/I 114
Treatment 241.22 mg/I 100 121.77 mg/I 114

The difference in mean TSS concentrations between the watersheds is primarily a function of the county road in the treatment watershed (Map 1). The road is a moderately used gravel road with a culvert that carries the stream under the roadway during low-flow periods, but crosses over the road during high-flow periods. The road surface generates additional runoff and sediment during storm events (Reid and Dunne 1984). The extra runoff and sediment is captured in roadside ditches and carried to the stream, increasing flow and TSS concentrations in the stream. The culvert under the road constricts flow and increases the velocity of the water as it passes through the culvert. The higher velocities cause stream bank and stream bed erosion on the downstream side of the road that also adds to the TSS concentrations in the stream.

The difference in mean TSS concentrations between the periods in each watersheds is a function of the difference in flows during the calibration and treatment periods. TSS concentrations are driven by flow. The extended periods of low flow and low TSS concentrations measured during the treatment period have a dramatic impact on the average TSS concentrations for the period.

To put the sediment data in context with other watersheds, comparisons with other forested watersheds and mixed land use watersheds are presented in Table 3. Because existing data from forested watersheds in Maryland are limited, two examples from forested watersheds in the Mid-Atlantic region are also presented. The examples used from Maryland are a predominately agricultural Piedmont watershed and an urbanizing watershed at the edge of the Piedmont as it falls to the Coastal Plain. The range of TSS concentrations measured at the outlet of the treatment and control watersheds were similar to concentrations measured in both forested and agricultural watersheds.

Table 3. Total Suspended Solids Yields from the treatment, control and comparison watersheds.
Watershed Physiographic Region Size (acres) % Forst % Ag % Other Average Yield (lbs/ac/yr)**
Treatment Piedmont/Blue Ridge* 330 100 0 0 161
Control Piedmont/Blue Ridge* 280 100 0 0 84.5
Ponds Branch MD1 Piedmont 95 100 0 0 10
Smith Creek NC2 Piedmont <6,400 74.6 22.4 3 148
Young Woman Creek PA3 Appalachian Plateau 25,568 95.9 4.4 0 260
Piney Creek MD4 Piedmont 20,032 13.6 75.3 11.1 1,216
White Marsh Run MD5 Piedmont 1,747 41 17 42 7,808

* while Sugarloaf Mt. is in the Piedmont physiographic province, it has characteristics more typical of the nearby Blue Ridge province.
** minumum of three years of data

    1. Cleaves et al. 1970.
    2. Lenat and Crawford 1989.
    3. Langland 1995.
    4. McCoy 1999.
    5. MD DNR unpublished data.

Sediment Loads and Yields
Monthly TSS load estimates are presented in Figure 2. Estimates of monthly loads generated from the control watershed range from 0.99 tons in December of 1998 to 19.31 tons in December of 1996. Estimates of monthly loads generated from the treatment watershed range from 0.78 tons in September 1998 to 103.22 tons in January 1996.

Since load estimates are a function of flow and concentration, the variability in the load estimates reflects the variability in flow and concentrations. The differences in the characteristics of the control and treatment watersheds that affected the TSS concentrations and flows also affect the TSS load estimates.

[Figure 2. Total Suspended Solids Loads from the treatment and control watersheds]

To account for the difference in size between the control and treatment watersheds and make comparisons with other watersheds, the load estimates for each watershed are divided by the acreage in the watershed to produce a TSS yield estimate expressed in lbs/acre (Table 3). It is interesting to note that there is variation yields from the forested watersheds. All of the yields were under 1000 lbs/acre/year. By comparison the yields from the agricultural or urbanizing watersheds are an order of magnitude or two higher than the yields from the forested watersheds. The characteristics that prevent the detachment of soil particles during rain events like vegetative cover, infiltration rates, water storage capacity and canopy interception are all much more prevalent throughout the forested watersheds (Patric 1976).

Maximum temperatures ranged from 22.73 - 23.35 degrees (Figure 12) during the summers of 1995, 1997 and 1998 when the watersheds suffered from severe drought conditions. Both the control and treatment streams were completely dry for more than two weeks, with water remaining only in pools in 1995. Maximum stream temperatures for the summer of 1996 ranged from 15.57-17.34 degrees centigrade. This lower 1996 maximum temperature was the result of continuous water flow in the streams through the summer. During drought conditions the State maximum for Natural Trout Waters (20 degrees centigrade) was exceeded in both the control and treatment watersheds. Streams such as these are extremely sensitive to changes in stream temperature, and even small increases can adversely affect existing fish populations.

Mean summer water temperatures during the calibration and treatment periods were significantly different for both watersheds. Mean summer temperature in the control watershed was 16.63oC during the calibration period and 17.69oC during the treatment period. Mean summer temperature in the treatment watershed was 17.74oC during the calibration period and 18.54oC during the treatment period. These differences demonstrate the effect of groundwater contributions. The cooler temperatures during the calibration period are the result of higher baseflow levels in 1996.

Paired Data
Regression equations were developed to describe the relationships between TSS concentrations in the control and treatment watersheds, flows in the control and treatment watersheds, and water temperature in the control and treatment watersheds during the calibration and treatment periods (Figures 3,4 and 5). The analysis of covariance between the calibration and treatment period relationships for each of these parameters indicates that there are no significant differences between these relationships (Tables 5,6 and 7). This indicates that the timber harvest, with associated BMPs, in the treatment watershed did not cause a change in the relationship between TSS concentrations in the two watersheds, flows being discharged from the two watersheds, and water temperature in the two watersheds. Based on the study design, the results indicate that the suite of BMPs employed and the harvest methods applied did not cause a change in TSS concentrations being discharged from the treatment watershed nor a change in water temperature in the treatment watershed. Although the results indicate that there was no significant change in the relationship between flows from the treatment and control watersheds during the two periods, the data suggest that there was some change in this relationship (Figure 4). The change in the alignment of the low flow data from the calibration period to the treatment period suggests that the treatment watershed generated less run-off during smaller storm events during the treatment period. There could be several explanations for this apparent difference. Small differences in soil types between the control and treatment watersheds may have had an effect on the hydrology. The treatment period had longer dry periods than the calibration period and differences in soil water storage may have affected flow. A second explanation could be the upgrading of the roads in the treatment watershed and the installation of BMPs designed to retain sediment and water. These changes in the treatment watershed may have increased water storage in the watershed and reduced run-off. The difference may also be the result of the random spacial variability in rain fall. The watersheds are adjacent, but may have received different amounts of rainfall.

[Figure 3. Paired TSS Concentrations and Regression Equations for the Calibration and Treatment Periods.]

[Figure 4. Daily Average Flow Calibration and Treatment Period Regression Models]

[Figure 5. Daily Average Temperature Calibration and Treatment Period Regression Models]

Table 4. Log 10 of TSS Concentrations in the Treatment and Control Watersheds General Linear Models Procedure

Dependent Variable: Log 10 of TREATMENT TSS Concentrations

Source                DF	Sum of Squares	Mean Square	F Value    Pr > F
Model                   2	92.30948648	46.15474324	210.24	0.0001
Error                 211	46.32241162	 0.21953750
Corrected Total       213	138.63189810

                  R-Square                   C.V.                  Root MSE              LTREATME Mean
                  0.665860                  28.91223               0.4685482             1.62058831

                                               	T for H0:        	 Pr > |T|        	Std Error of
Parameter                       Estimate      	Parameter=0                        	Estimate

INTERCEPT                   0.2080817827 B              2.72         	 0.0070          	0.07646357
LCONTROL                    0.9318494643               19.46    	 0.0001          	0.04789193
PERIOD    	cal            -.0012162547 B              -0.02     		 0.9857         	0.06767387
          	tre            0.0000000000 B               .             .               .

Table 5. Flow - Treatment and Control Watersheds General Linear Models Procedure


Source	DF	Sum of Squares 	Mean Square	F Value 	Pr > F
Model	3	874.97143907	291.65714636   	462.60	0.0001
Error	1107	697.93831363	0.63047725
Corrected Total	1110		1572.90975271

         R-Square              C.V.           Root MSE          DISCHGTF Mean
        0.556276            90.21707         0.7940259           0.88012828

                                             T for H0:  	 Pr > |T|       	Std Error of
Parameter            Estimate        	  Parameter=0                      	Estimate
INTERCEPT        0.2226175188 B              5.70           	0.0001          	0.03903460
DISCHGC CONTROL  0.6617962608 B             17.73          	0.0001          	0.03732699
PERIOD    CAL    0.0921460942 B              1.38               0.1684               0.06685375
	TRT   0.0000000000 B               .             .               .
	CAL    0.7974329515 B             13.16          	0.0001          	0.06060307
           TRT   0.0000000000 B               .             .               .

Table 6. Daily Average Temperature in the Treatment and Control Watersheds General Linear Models Procedure

Dependent Variable: AVETMPT Treatment Watershed

Source		DF	Sum of Squares	Mean Square	F Value	Pr > F
Model		2	864.42171981	432.21085990	905.23	0.0001
Error		326	155.65207958	0.47746037
Corrected Total	328	1020.07379939

         R-Square                      C.V.            Root MSE          AVETMPT Mean
         0.847411                  3.749779            0.6909850         18.42735562

			T for H0:         	Pr > |T|		Std Error of
Parameter 	Estimate		Parameter=0			Estimate
INTERCEPT 	3.834030754 B	10.37          	0.0001		0.36964676
AVETMPC 	0.843159553	41.28          	0.0001		0.02042559
PERIOD cal	-0.014414434 B	-0.18         	0.8565 		0.07965628
       tre 0.000000000 B          .             .               .

Benthic Macroinvertebrates
Sampling for macroinvertebrates had been planned for August and April of each year of the study. Drought conditions during the summer and early fall of 1995 eliminated flow from the treatment watershed stream, Furnace Branch, from August through early October, and forced sampling to begin in November of 1995. Severe drought and no flow conditions also prevailed in the summer and fall of 1998 forcing the fall sampling to be postponed until early December of that year. Although flow never completely stopped in the control watersheds during these drought periods, they were at minimal levels. All other sampling was accomplished during the August/April time frame as originally planned.

[Fifure 6. Habitat as a % of Maximum Score] [Figure 7. EPT Taxa Richness (Genus Level)]

The timber harvest activity had no impact on the habitat scores (Figure 6). Both streams had habitat scores above 75% of the total possible score and are considered equal to a reference or undisturbed condition. There are slight seasonal variations that reflect minimal scoring differences due to the presence or absence of leaves on the trees. The lower scores of Furnace Branch are due to a sandier substrate with higher levels of imbeddedness (primary habitat characteristics), and a higher percentage of cut banks (secondary habitat characteristic). The presence of a dirt/gravel county road through the upper portion of this watershed contributing sediment and flashy flows to the stream is considered the cause of these impairments.

The macroinvertebrate communities for Furnace and Bear Branches are consistent with those found in other similarly classed streams in the Monocacy watershed (Butler, pers. com.1999). A combined taxa list for these streams is provided in Appendix A. The suite of metrics calculated for the macroinvertebrate communities of these streams indicate no discernable impacts due to the forest harvest activity. The lower habitat quality and no flow drought conditions in Furnace Branch are considered the major factors creating the differences in macroinvertebrate community metric scores. The taxa richness metric scores are almost parallel across pre and post harvest periods except for the drought periods of 1995 and 1998 (Figure7). This pattern is repeated in the biotic index (Figure 8) and the EPT taxa richness (Figure 9). Although sampling in Furnace Branch was done a minimum of six weeks after flow returned (a standard interval for sampling after catastrophic events), the no flow conditions produced small sample sizes that tended to bias the biotic index score. In these streams, low sample sizes with lower overall taxa richness are generally dominated by the more pollution intolerant EPT (Figure 9). This EPT dominance, with their lower (better) biotic number, produces the better biotic scores. Alternatively, Bear Branch, with continuous although low flow, had a lower percentage of EPT and higher proportion of more tolerant taxa that created poorer biotic index scores (Figures 9, 10 and 11).

An overall benthic Rapid Bioassessment Protocol score is produced by scoring a comparison of the individual metrics to a reference condition or a bench mark to produce a percent of reference value. The reference used in this instance was the Bear Branch sample from April of 1997. This sample had the highest metric scores of any period during the study. Figure 11 shows the results of this comparison. As with the individual metrics, there is little change in the relationship between the two macroinvertebrate communities over the course of the study.

The data indicate clear differences in the macroinvertebrate communities of the treatment and control watersheds. These differences are present from the beginning of the calibration period and continue basically unchanged through the end of the treatment period. Differences in watershed characteristics, i.e. the presence of the dirt/gravel road in the treatment watershed and climatic changes, are judged to be the controlling factors. The road contributes sediment and concentrated storm flows into the headwaters of the treatment watershed. The sediment loads and increased bank erosion from storm flows create a sandier substrate than found in the control stream. The sandier substrate in the treatment watershed fosters a moderately different macroinvertebrate community through microhabitat differences and being prone to de-watering during drought conditions.

The variations in the individual and overall macroinvertebrate community RBP III metrics can not be attributed to the forest harvest activity. Any impacts to the macroinvertebrate community of Furnace Branch that may have been caused by the forest harvest activity were of such minor nature that they were masked by the magnitude of natural variability resulting from ambient conditions.

Photographic Log
Photographic documentation and visual observation detected several instances of BMPs for road drainage and stabilization negatively impacted by use of trucks and skidders during very wet conditions, but holding up well during moderately wet or dry conditions, and working very well for post-harvest stabilization. At no time was overland flow of storm water detected moving very far from the logging roads, landings, etc. When streams were walked during and immediately after storm events, there was no detectable overland flow of storm water from logged sites reaching streams. Photographs were able to clearly document the success of such practices as the portable bridge and post-harvest vegetative stabilization.

Stream crossing site, showing bridge being installed and used, and post-harvest stabilization, from photographic log.

[Loggers discuss BMP planning at a Master Logger Workshop.]

The educational and demonstration opportunities of this project have been well utilized. Many people have toured the project, including loggers, landowners, foresters, sediment control inspectors, municipal water supply managers, school teachers, college and high school students, local government officials, and officials from state and federal agencies such as the Environmental Protection Agency, USDA - Forest Service, USDA - Natural Resources Conservation Service and other agencies within Maryland DNR. There have also been several articles on the project in local and forestry newsletters, and formal presentations given to several professional and academic groups. Information on the project is available on our own web site at , which is also accessible through the Maryland DNR - Forest Service home page at (Programs, Chesapeake Bay & Water Quality Programs, Paired Watershed Study).

Some of the Educational Programs at the BMP Effectiveness Study Area
Date		Activity					Participants
10/7/95	Master Logger, Certification Workshop			28 loggers
9/28/96	Master Logger, Advanced BMP’s Workshop			9 loggers
10/11/96	Tour for Fred. Co. Planners & Sed. Crtl. Inspectors		8 agency staff
6/30/97	Chesapeake Bay Foundation Workshop/Tour			11 teachers
7/22/97	Regional Forest Service Meeting and Tour			16 foresters
9/6/97	Erosion and Sediment Control Certification			4 loggers
10/23/97	Maryland/Delaware Society of American Foresters		29 foresters
10/30/97	Regional Forestry Board Workshop			18 members
5/16/98	Forest Landowners Field Day				30 landowners
6/2/98	Tour for New York City Watershed Group			9 agency staff
6/23/99	Workshop/Tour for SCD’s & NRCS				32 agency staff

Back to Contents

This study was funded through a Clean Water Act Section 319(h) Grant from the U.S. Environmental Protection Agency

Maryland Department of Natural Resources
Forest Service and Chesapeake & Coastal Watershed Service
Annapolis, Maryland / April 2000 / FWHS-FS-00-01

The Maryland Department of Natural Resources Forest Service wants to know if this service is helpful
for you or your company, please send comments and/or suggestions to Maryland DNR.

Forestry Links
Forestry Home
Forestry Acronyms
Email DNR

Other DNR info
DNR home page
Bays & Streams
Wildlife & Heritage