To test my hypothesis that undisturbed crescentic gouges can
facilitate biomat and vascular plant colonization, in September
2021, I quantified biotic cover in crescentic gouges on four
mid-elevation granite balds in New England, USA: Black Mountain
(Dummerston, VT, 390 m, 42.9229, -72.6054), Bald Mountain
(Woodstock, ME, 516 m, 44.9548, -70.7906), Speckled Mountain (Peru,
ME, 677 m, 44.29112, -70.95499), and Whitecap Mountain (Rumford, ME,
675 m, 44.57159, -70.66082). All sites are popular with hikers and
have open summits with well-developed but discontinuous biomat and
vegetation assemblages (except Black Mountain, where vegetation
cover is more continuous). On each granite bald, I located all
visible crescentic gouges, randomly determined (coin toss) whether
to sample each gouge I encountered, and then visually estimated the
percent cover of all biota (biomats and vascular plants) in 102
gouges – 39 on Black Mountain, 20 on Bald Mountain, 7 on Speckled
Mountain, and 36 on Whitecap Mountain. At each sampled gouge, I
measured gouge size (l x w x d, cm), proximity (m) to the nearest
upslope, well-developed biomat and vascular plant assemblage (i.e.,
source of propagules), and proximity (m) to the established hiking
trail (i.e., disturbance). For simplicity, I calculated an index of
crescentic gouge area (cm2) using the
formula for a semicircle ((πr2)/2), with
r=half the gouge length measured tip to tip.
To evaluate the effects of proximity to presumed propagule sources
(nearest upslope, well-developed assemblage), disturbance (hiking
trail), and gouge size on biotic cover in gouges, I used
DISTance-based Linear regression Models (DISTLMv.5; Anderson 2001)
performed by site (999 permutations). DISTLM partitions the
variances among treatment groups by computing the distance matrices
of raw data, and the resulting pseudo F-statistic is not bound by
the assumptions of the F distribution of traditional parametric
techniques (Legendre and Anderson 1999). Statistical significance
was determined at p≤0.05.
