C OMPROMISO DEL AUTOR CONSIGO MISMO : UNA MIRADA

General Technical Report GTR-694

Puget Sound region in Washington, Scotch broom germin- ated more abundantly when the cryptogamic layer (biolo- gical crust) of glacial outwash soils was intact versus removed (Parker 2002). Vertebrate herbivory decreased bio- mass of Scotch broom seedlings along the northern coast of California (Bossard and Rejmanek 1994). At a Sierra Nevada foothill site in California, a greater number of seedlings established on disturbed- versus intact-soil micro- sites; whereas at a coastal site, treatment effects were not significant, in part, because of confounding differences due to seed dispersal by ants (Bossard 1991).

To investigate the potential of Scotch broom to germi- nate and grow in forest understories, two cohorts of seeds were sown in Douglas-fir stands that had been thinned to several overstory retention levels. In each stand, seeds were sown on both mineral and organic seedbeds to determine if germination substrate affected seedling establishment. Seedling abundance was observed for up to two years after sowing seeds. Several microclimatic variables were meas- ured periodically to potentially explain differences in seedling abundance.

METHODS

Sites and Overstory Treatments

The research was conducted in the Overstory Density Study of Capitol State Forest near Olympia, WA (Harrington 2006). At each of three sites supporting 40- to 70-year old Douglas-fir, plots at least 1.9 ha in area (137 m x 137 m) were randomly assigned various overstory retention levels. Stands were thinned in summer 2000 to leave a relatively uniform spacing of well-formed trees. Advanced conifer regeneration and large hardwoods or shrubs also were cut. Three plots at each site were selected to represent overstory retention levels hereafter designated as clearcuts (absence of overstory trees), shelterwoods (35–54 trees ha-1

retained), or thinned stands (84–146 trees ha-1_retained).

The research was conducted within the outer 30.5-m buffer area of each plot to avoid disturbing ongoing genetic and species-mixture studies underway in the interior of each plot. Although growing conditions probably differed some- what between the exterior and interior of each plot, the

three overstory levels selected per site represented a broad range of light environments.

Cohort #1

In September 2003, Scotch broom seeds were collected from several sites in the Olympia area. Seeds were air dried for 60 days and stored at room temperature in a glass jar. A total of 18 samples of 40 seeds each were counted and placed in separate plastic bags. At each of the three plots per study site, understory shrub and herbaceous species were cleared within a 2- x 2-m area. The upper organic layer of the soil was removed and the soil was mixed to a depth of 30 cm. Approximately 2.5 cm of decomposed litter (O layer) was applied to a randomly located 0.25-m2₍₅₀

cm x 50 cm) subplot within the cleared area. An adjacent subplot of exposed mineral soil was left uncovered. One sample of seeds was scattered on the surface of each sub- plot in November 2003. The number of living seedlings in each subplot was counted monthly from April to November 2004, and in April and September 2005 and May 2006.

Cohort #2

In July 2004, seeds were collected from the same locations, air dried, and stored as described previously for Cohort #1. A total of 18 samples of 100 seeds each were placed in sep- arate plastic bags. Within each of the three plots per study site, a new 2- x 2-m area was cleared of all understory shrubs and herbs. In November 2004, the two seedbed types were created and seeds were sown as described for Cohort #1. Seedling abundance of Cohort #2 was deter- mined in April and September 2005 and May 2006. All seedlings from each cohort were removed at the final count to prevent spreading the species.

Microclimate

Overstory cover was measured at randomly selected points throughout each plot in July 2004 with a vertical densito- meter (Geographic Resource Solutions, Arcata, CA). Photosynthetically-active radiation (PAR) was measured concurrently with an Accupar®_{ceptometer (Decagon}

Devices, Pullman, WA) (see Harrington 2006 for details). Volumetric soil water content at 0-12 cm depth was meas- ured monthly from June to October 2004 and in March, 38

April, and September 2005 with a Hydrosense®_(Campbell

Scientific, Inc., Logan, UT). Soil water measurements were calibrated to the soils at each site (Harrington 2006) and included readings for each of the organic and mineral seedbeds per subplot. Soil temperature at 2-cm depth was measured with an I-button®_{sensor (Dallas Semiconductor}

Corp., Dallas, TX.) placed near the center of each subplot, which logged data every two hours from April 2005 to May 2006.

Statistical Analysis

All statistical analyses were performed with SAS (SAS Institute, Inc. 1999). The experimental design was a ran- domized complete block with three replications (sites) of six treatments. The six treatments were arranged as a split plot with overstory retention level (clearcut, shelterwood, or thinned stands) as the main-plot treatment and seedbed type (mineral versus organic) as the split-plot treatment. Periodic values of seedling abundance by subplot were expressed as proportions of the total number of seed sown and the data were normalized with an arc-sine, square-root transforma- tion (Sokal and Rohlf 1981). The same transformation was applied to normalize measurements of soil water content. Mean, minimum, and maximum daily values of soil tem- perature were averaged by month. Data for seedling abun- dance, soil water, and soil temperature were subjected to a repeated-measures analysis of variance (ANOVA) in SAS Procedure MIXED to determine the significance (α = 0.05) of overstory level, seedbed type, and their interaction, while accounting for random effects of individual subplots. The analysis of seedling abundance was conducted separately for each cohort on the pooled spring (April or May) and fall (September) measurements from the study to provide a relatively even spacing in time among measurements (i.e., an assumption of repeated-measures analysis). The same approach was used for analysis of soil water and tempera- ture data. Analyses were conducted with maximum likeli- hood estimation assuming an autoregressive covariance structure to account for time series trends. For each ANOVA, residuals were plotted against predicted values of the dependent variable to verify that residual variances were relatively homogeneous. If an interaction was detected

among experimental factors, pairwise comparisons of least- squares adjusted means were conducted with Bonferroni adjusted probabilities (Sokal and Rohlf 1981). The follow- ing orthogonal contrasts were performed to partition effects of overstory level (contrasts (1) and (2)) and its interaction with seedbed type (contrasts (3) and (4)):

(1) Absence versus presence of overstory trees (clearcuts versus the mean of shelterwoods and thinned stands).

(2) Shelterwoods versus thinned stands. (3) Effects of seedbed type differ between

absence and presence of overstory trees. (4) Effects of seedbed type differ between

shelterwood and thinned stands.

RESULTS

Herbivory from unknown sources appeared to be the pri- mary cause of reductions in seedling abundance over time, especially in shelterwoods and thinned stands where leaves of seedlings were commonly browsed in the late summer and fall, resulting in their subsequent mortality. Drought may have been responsible for mortality of some seedlings, particularly in thinned stands where wilted seedlings were observed in late summer. Seedling height after one growing season ranged from 5 to 20 cm except for one seedling that attained a height of 60 cm in a clearcut plot.

Average abundance of Scotch broom seedlings in Cohort #1 was less in clearcuts than in shelterwoods and thinned stands (orthogonal contrast (1): P = 0.016) (fig. 1). None of the treatment-by-timing interactions were signifi- cant (P ≥ 0.163). Organic seedbeds had higher seedling abundance than mineral seedbeds in Cohort #1, but only in shelterwoods and thinned stands (orthogonal contrast (3): P = 0.048) (fig. 2). Seedling abundance from Cohort #2 never exceeded 1 percent. Although the interaction of overstory level and timing was significant for Cohort #2 (P = 0.023), pairwise comparisons between overstory levels within each timing failed to detect significant differences in seedling abundance.

Average overstory cover was 0, 31, and 62 percent in clearcuts, shelterwoods, and thinned stands, respectively,

General Technical Report GTR-694

and average PAR was 100, 62, and 34 percent of full sun- light, respectively (Harrington 2006). Average soil water content was greater in shelterwoods than in thinned stands (orthogonal contrast (2): P = 0.023) (fig. 3a). Soil water

content did not differ significantly (P = 0.358) between organic and mineral seedbeds, and none of the treatment- by-timing interactions was significant (P ≥ 0.095). Mean daily soil temperature was 1.5°C greater in shelterwoods 40

Figure 1—Average abundance of two cohorts of Scotch broom seedlings during two growing seasons after sowing seeds in various overstory retention levels of Douglas-fir. Cohort #1 was sown in November 2003 (standard error = 0.657) and cohort #2 was sown in November 2004 (standard error = 0.302).

Figure 2—Average abundance of Scotch broom seedlings from Cohort #1 as affected by the interaction of Douglas-fir overstory retention level and seedbed type (standard error = 0.602).

than in thinned stands (orthogonal contrast (2): P = 0.028) (fig. 3b). Average values of soil water and mean daily soil temperature in clearcuts were intermediate to those in shel- terwoods and thinned stands. Minimum daily soil tempera- ture in September 2005 was lower in clearcuts (7.8°C) than in shelterwoods (9.9°C) or thinned stands (9.9°C) (data not shown). Maximum daily soil temperature did not vary sig- nificantly with overstory level or its interaction with timing (P ≥ 0.244).

DISCUSSION AND CONCLUSIONS