1. Ridges,
Coves,
& Trees
Topography as control
on forest growth in
El Yunque Forest,
Puerto Rico
ABSTRACT
Tropical forests are important terrestrial carbon sinks under current
trends of rising anthropogenic CO2 (Oren et al., 2001). However,
many other environmental factors affect forest growth in a tropical
landscape. Recently, a LiDAR study at El Yunque National Forest,
Puerto Rico revealed a close correlation between landscape relief and
forest canopy height. Canopy is highest in the coves and lowest on
the ridges (Wolf et al.). We hypothesize that rates of landscape ero-
sion is the primary driver behind topographic control on canopy
height in El Yunque Forest.
We collected surface soil samples from upper Rio Blanco catch-
ment along four hillslope transects, at elevations between 600 and
700m. We recorded forest species compositions at each sampling
site, and measured soil mineral compositions using X-ray Powder
Diffraction (XRD) techniques at University of Pennsylvania. Re-
sults support our prediction that rock-derived minerals are depleted
on slow-eroding ridges, but are replenished in fast-eroding coves
and slopes. We speculate that the same topographic mechanism is
responsible for governing canopy height differences at a larger land-
scape scale across knickpoints. This study furthers our understand-
ing of the behavior of tropical hotspots (Taylor et al., 2015). Tropi-
cal forests' responses to rising atmospheric CO2 and to environ-
mental engineering schemes will likely exhibit tremendous topogra-
phy-induced variability.
INTRODUCTION
Rising atmospheric CO2 levels instigate responses from various natural systems. Trop-
ical forests are important sources of carbon sink, balancing a significant amount of the
yearly anthropogenic CO2 emissions (Beedlow et al, 2004).
Existing environmental controls over forest growth, however, limit the expres-
sion of expected balancing feedback. Above all, soil nutrient limitation is likely to be
the most important factor that prevents forest response from keeping pace with rise of
anthropogenic CO2 in the atmosphere (Beedlow et al, 2004).
BACKGROUND
Location: “The Rio Blanco Platform”
El Yunque Forest, Luquillo Mountains, eastern Puerto Rico.
Geologic History: The ancient shore platform of a relict island – El Yunque Island –
which encompassed the area of Luquillo Mountains today (Brocard et al., 2015). 4.2
Million Years ago, a sudden uplift event produced an erosional disequilibrium.
Substrate: Quartz Diorite (QD) bedrock.
Weatherable Minerals:
[Fig.1] Upstream
regions of Rio Blanco
– the “Rio Blanco
Platform” – is dissect-
ed by water channels
producing ridges
(yellow) and coves
(blue). (Adopted from
presentation slides by
Jane Willenbring)
[Fig.2a,b] (a, above) The mineral stoi-
chiometry and abundances (wt%) of
quartz diorite (QD) bedrock, saprolite,
and soil, which underlie the Rio Blanco
platform. (b, right) Changes n mineral
composition with depth in a typical QD
bedrock-saprolite-soil profile: minerals
are present in the soil-saprolite column
in relatively constant proportions, but
concentrations of these minerals fluctu-
ate in the surface soil
(white et al., 1997)
RESEARCH QUESTION
There is a close correlation between landscape relief and forest
canopy height. Canopy is highest in the coves and lowest on the
ridges. What governs this correlation?
HYPOTHESIS
Topography control canopy height by regulating rates of erosion,
nutrient distribution, and nutrient rejuvenation.
METHODS
Field Methods: We collected soil samples in January 2016. The sampling process was
designed to capture ridge-cove sequences in the landscape. We selected 4 different
transects (SABLL, SABHT, TWT, & ICG) for sampling. Top 15cm of soil is sampled
at each site. We collected O-layers where there is a significant amount.
Lab Methods: Soil samples are dried, homogenized, and ball-milled to <10µm
powder. From the homogenized samples we analyzed the mineral composition using
X-ray Powder Diffraction (XRD) technique at the University of Pennsylvania lab.
[Fig.3] Hillslope nutrient distribution
conceptual model proposed by Chad-
wick & Asner. Hydrological redistribu-
tion of subsurface soil, combined with
soil creep and overland flow which
redistribute surface soil, act in cohesion
to transport mass downslope towards
active stream channels. Plants produce
small-scale, in-situ nutrient cycling.
(Chadwick & Asner, 2015)
[Fig.4] There is a close correlation between landscape relief and forest canopy
height (middle map: green=higher canopy). Sampling transects are located in
the Rio Blanco platform; they’re selected to (1) capture hillslope gradient, (2)
maximize canopy change, and (3) avoid human disturbances.
RESULTS
The Hillslope Gradient: Field observations along selected hillslopes match our predictions based on the LiDAR data.
Canopy heights of dominant woody plants decrease with relief. Trees on steep slope and cove floor grow tall and straight, with few
branches. Trees on the ridge top grow low, and branch close to the ground.
Maximum density of the herbaceous layer occurs at the slope shoulder, where canopies are low or fragmentary. Maximum densi-
ty of woody plants occurs at the cove bottom, where tall-growing trees share canopy area efficiently.
Species compositions often change abruptly across the hillslope gradient. Well-defined boundaries separate ridge-top palo colo-
rado groves from the slope-cove sierran palm-tabonuco forests.
Soil Mineral Characteristics: Results from laboratory XRD analyses meaningfully reinforce our confidence in the original hypothesis:
(1) spectrums show a gradual upslope decrease in secondary mineral abundance, concurring with gradual changes in rates of erosion,
soil depth, and weathering states; (2) there is a lack of amorphous humps at the base of secondary mineral peaks from ridge samples,
indicating secondary minerals aren't actively forming at ridge sites, but instead are gradually lost from the system.
[Fig.5a,b,c] (a,left) slope and cove forest; (b, center) the shoulder slope - SABLL clearing; (c, right) ridge-top palo colorado grove.
[Fig.6a,b] From cove sites to ridge sites along SABLL (a, left) as well as ICG (b, right), we observe an overall increase in percent quartz content
and a corresponding decrease in kaolinite presence. Observed trend is less pronounced along ICG, a result of ICG cove’s less developed state.
[Fig.7a,b] (a, left) Ridge site (gold) pand cove site (green) comparison: cove (green) graph is characterized by an assortment of smaller peaks
representing secondary minerals. (b, right) Gradient of sampling sites from cove (SABLL-1, green) to ridge (SABLL-8, orange) shows gradual
stages of transition from site 1’s wide range of secondary mineral peaks towards site 8’s flat spectrum indicating secondary mineral depletion.
The Cove
Higher rates of erosion: ~ 100-150 m/My
Thin soil profile.
Dominated by tabonuco and palm trees.
Trees grow tall without branching.
The Ridge
Lower rates of erosion: ~ 40-60 m/My
Thick soil profile.
Dominated by palo colorado trees
Trees are shorter and branch close to the ground.
Soil - Saprolite
Corestones
Water Channel
(source of erosion)
Bedrock
(Quartz Diorite)
SABLL Clearing
(Deep soil swales & no canopy)
The Steep Slope
The Shoulder Slope
Bowen Chang - bowen@sas.upenn.edu - April 2016
Advisors: Dr. Jane Willenbring, Emma Harrison - Thesis instructor: Dr. Jane Dmochowski