Land-use history augments environment–plant
community relationship strength in a Puerto Rican wet
forest
James Aaron Hogan1*, Jess K. Zimmerman1, Mar
ıa Uriarte2, Benjamin L. Turner3 and
Jill Thompson1,4
1Department of Environmental Sciences, University of Puerto Rico R
ıo Piedras, San Juan, PR 00925, USA;
2Department of Ecology, Evolution and Environmental Biology, Columbia University, New York, NY 10027, USA;
3Smithsonian Tropical Research Institute, Apartado 0843-03092, Balboa, Ancon, Panama; and 4Centre for Ecology &
Hydrology (Edinburgh), Midlothian EH26 0QB, UK
Summary
1. Environmental heterogeneity influences the species composition of tropical forests, with implica-
tions for patterns of diversity and species coexistence in these hyperdiverse communities. Many
studies have examined how variability in soil nutrients and topography influence plant community
composition, with differing results. None have quantified the relative contribution of environmental
heterogeneity versus endogenous processes to variability in forest community composition over time
and with respect to successional recovery.
2. Using five consecutive trees censuses of a forest plot in Puerto Rico, conducted between 1990
and 2011, we evaluated the influence of edaphic and topographic variability on community composi-
tion. The plot has a well-documented land-use history and is subject to periodic hurricane distur-
bance. Using multiple canonical distance-based redundancy analyses, we studied how spatial
heterogeneity in soil nutrients and topography structure community composition over time, as the
forest recovers from long-term land-use effects and two major hurricanes in 1989 and 1998.
3. For the entire plot, spatial variables (principle coordinates of neighbourhood matrices), represent-
ing the autocorrelation of tree species in the community, explained the majority (49–57%) of the
variability in tree community composition. The explanatory power of spatial variables decreased
over time, as forest structure recovered from hurricane damage and the stems in the understorey
died. Soil nutrients and topography, collectively, explained a moderate portion (33–37%) of the spe-
cies compositional variation and were slightly more robust in explaining compositional differences
in areas of more intense past land use.
4. Areas of less-intense past land use showed weaker community–environmental trends overall,
illustrating a tendency for stronger resource competition (i.e. light, water and soil nutrients) between
species in these areas. This illustrates how environmental–plant community interactions are strength-
ened by the lasting effects of human land-use legacies, which persist for decades to centuries.
5. Synthesis. Our findings confirm past land use to be a fundamental driver of the structure and
composition of secondary forests through its impacts on the tree community, the abiotic terrestrial
environment and their interaction. Since the extent of second-growth tropical forests continues to
increase, our findings highlight the importance of understanding the processes that determine the rate
and nature of their succession.
Key-words: beta-diversity maps, land-use legacies, Luquillo, plant–soil (below-ground) interac-
tions, soil resources, spatial autocorrelation, terrain ruggedness, topography, tropical forest, variance
partitioning
*Correspondence author: E-mail: jamesaaronhogan@gmail.com
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society
Journal of Ecology 2016, 104, 1466–1477 doi: 10.1111/1365-2745.12608
Introduction
A number of theories exist regarding how such high levels of
tree species diversity are maintained in tropical forests, from
negative density-dependent mechanisms (e.g. Janzen–Connell
processes; Janzen 1970; Harms et al. 2000; Jansen et al.
2014) to dispersal-limited stochasticity (Hubbell 2001; see
Wright 2002 for a complete review of theories). Studies also
support local resource variation (i.e. soil nutrients, light and
water), and other niche-related processes as key in maintain-
ing diversity, especially in highly diverse tropical ecosystems
(Hutchinson 1957, 1978; Ricklefs 1977; Booth et al. 1988;
Paoli, Curran & Zak 2006). These processes shape observed
forest communities by operating at multiple scales, over vary-
ing time periods, and in conjunction with natural or anthro-
pogenic disturbance regimes.
Natural disturbances (e.g. hurricanes, floods and fire) and
human land use (e.g. logging) induce changes in species com-
position and abiotic resource availability, accelerating ecosys-
tem transformation (White 1979; Sousa 1984; Legendre &
Fortin 1989). A number of studies have demonstrated that
forest responses to human and natural disturbance depend on
species-specific life-history trade-offs. Differences in survival
and reproductive strategies among tropical tree species,
specifically the initial competitive advantage and subsequent
high mortality of early-successional pioneer species as the for-
est canopy recovers, determine the trajectory and rates of for-
est succession (Pacala & Rees 1998; Comita et al. 2010;
Wright et al. 2010; Uriarte et al. 2012). As human activities
continue to simplify and modify ecosystems (Chapin et al.
1997; Hautier et al. 2015), it is crucial to understand how
land-use legacies influence the abiotic environment of tropical
forests and how the impacts of these legacies play out over
time.
Time since land-use abandonment, the land-use intensity
and its duration are the main factors affecting soils and subse-
quent vegetation recovery rate and quality (e.g. forest struc-
ture, canopy height and biomass). For example, disturbance
intensity was found to be more important than environmental
variation in soil resources in determining structure and com-
position of regenerating forests in Hainan Island, China (Ding
et al. 2012). Human land use reduced total organic carbon
and nitrogen concentrations of soils in Mexican forests, with
increases after land-use abandonment, but at different rates
(Cram et al. 2015). Nitrogen and phosphorus have long been
thought to trade-off as limiting nutrients in terrestrial systems
and together may be used as indicators of soil fertility in the
tropics (Tateno & Chapin 1997). The recovery of soil fertility
and the organic humus layer in tropical forest, clay-derived
soils (e.g. Oxisols vs. Ultisols) that have been strongly
affected by past land use are likely to take decades to recover,
illustrating the long-term effects of soil-based alteration via
land use on vegetation composition and structure (Zimmer-
mann, Papritz & Elsenbeer 2010).
Environmental heterogeneity in resources reflects the com-
plex interaction of many biotic (e.g. herbivory and vegetation
turnover) and abiotic processes (e.g. decomposition, lithology
and climate) that operate at a variety of spatial and temporal
scales (Canham et al. 1994; Zimmerman et al. 1996; Wilson
2000; Tylianakis et al. 2008). Specifically, highly diverse
tropical forests are known for their large spatial variability in
soil chemical and texture, soil moisture, elevation, topography
and temporal dynamism in soil moisture (related to rainfall
and soil characteristics) and light conditions (e.g. light
dynamics associated with canopy structure and gap creation
or hurricane disturbance and recovery; Terborgh 1992; Hol-
dridge et al. 1971; Brokaw & Busing 2000; Davies et al.
1998; Cleveland et al. 2011).
Soil properties can influence tree canopy height, productiv-
ity and the rate of recovery of tropical forests following dis-
turbance, especially in terms of species community richness
(Ellis & Pennington 1992). High soil C: N, sand: clay, Al
and P concentrations were associated with a more complex
forest structure and increased diversity associated across a
successional gradient in Brazil (Martins et al. 2015). Simi-
larly, higher soil organic matter, higher nitrogen availability
and lower bulk density of soils were the main factors facilitat-
ing forest regrowth and succession in highly degraded areas
of China (Duan et al. 2008). Despite a number of studies on
this topic, results vary greatly across the tropics, and it
remains unclear how soil resources influence plant community
composition in forests recovering from land use or other
disturbance.
Although plant community diversity is generally greater at
high precipitation and lower elevation, at any particular time,
differences in physical environment can only weakly explain
patterns of alpha-diversity within tropical forest communities
(Lalibert
e, Zemunik & Turner 2014; Lalibert
e et al. 2013).
Therefore, to understand environmental–community relation-
ships, we must first understand community composition in
terms of beta-diversity or compositional differences between
forest areas. Globally, beta-diversity in tree species increases
with increasing topographic variation (De C
aceres et al.
2012). It is equally useful to evaluate the how topographies
influence community composition at more localized scales
within tropical forests. In a 20-ha permanent forest dynamics
plot (FDP) in south-western China, topography was shown to
play a significant role in controlling distributions of a number
of tree species, especially at early life stages (e.g. saplings
and immature trees; Lan et al. 2011). Other studies have
documented the strong effects of topographic-related habitat-
filtering of juvenile trees, showing a trend of decreasing
strength of filtering as trees mature (Kanagaraj et al. 2011;
Punchi-Manage et al. 2013; Hu et al. 2012; Zemunik et al.
2016).
Here, we examine the role of topography and soil resources
in structuring a Neotropical plant community that was
affected by land use in the past of different intensities within
a permanent 16-ha forest dynamics plot in Puerto Rico. In
this study, however, we do not seek to differentiate the direct
or indirect effects of land use on plant community–soil inter-
actions, but rather examine how the interactions change over
time and with the recovery of the forest from the land use
and the compound effect of two hurricane disturbances
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1467
(Hurricane Hugo in 1989 and Hurricane Georges in 1998).
Using multivariate redundancy analysis to partition the vari-
ability in plant community beta-diversity into spatial, edaphic
and topographic portions, we ask:
1 How well does variation in soil resources and topography
explain patterns of plant diversity composition with respect to
land-use legacies? We predicted soil resource and topographic
variables would have less explanatory power in structuring
plant communities with more intense land use (i.e. weaker
habitat association). The explanation could be that either the
land use has homogenized soil resource distributions, or plant
species distributions have been affected by the effects of past
land use. Alternatively, weak species–environment associa-
tions could reflect a compound effect of the two processes,
resulting in a decoupling of the relationships between the
abiotic environment and forest community composition.
2 To what degree do environmental–plant community rela-
tionships change over time and in response to hurricane dis-
turbances and with respect to land-use history? We expect the
strength of plant community–environmental relationships to
increase with time following hurricane disturbances as the
effect of the hurricane damage diminishes and the plant com-
munity reflects the environment. We hypothesize environmen-
tal relationships to be strongest in less-disturbed areas,
because there was less of a decoupling effect associated with
past low-intensity land-use pressure in these areas.
Materials and methods
SITE DESCRIPT ION AND TREE CENSUS METHODS
The Luquillo Forest Dynamics Plot (LFDP) [18°200 N, 62°490 W] is
a permanent forest monitoring plot located in the montane wet forest
of north-eastern Puerto Rico. The forest type is described as tabonuco
forest, named after a dominant species Dacryodes excelsa Vahl.
(Wadsworth 1951). The forest canopy is uniform, lacking canopy
emergents, in primary tabonuco forest with an average canopy height
of about 30 m and with few trees exceeding 2 m diameter at breast
height (dbh, diameter at 1.3 m above the ground surface; Brokaw
et al. 2004). Topographic variability within the LFDP is large with
elevation ranging between 333 and 428 masl. and slopes averaging
17%, but ranging from 3% to 60% (Weaver 2000; Thompson et al.
2002; Harris et al. 2012). Mean annual rainfall is 3685 mm (1975 to
present) and temperature ranges from 20.5 to 22.8 °C (Ram
ırez &
Melendez-Colom 2003; Thompson et al. 2004).
Soils in the LFDP are clay, comprised of deeply weathered Oxisols
and Ultisols developed in marine sedimentary lithology of volcanic
origin. These soils orders are typical of lowland tropical forests
world-wide; however, soil fertility in the LFDP is higher than much
of the continental lowland tropics (Soil Survey Soil Staff 1995;
Beinroth 2010). The plot has been censused five times at approxi-
mately 5-year intervals between 1990 and 2011. Due to Puerto Rico
being a medium-sized island located at a moderate distance from con-
tinental land masses, species richness is relatively low when com-
pared to other tropical forests, ranging from 43 to 54 speciesha1
(Lawrence 1996). In the most recent (2011) tree census, 123 species
were recorded in the whole LFDP (Hogan 2015).
In the 16-ha (320 m 9 500 m) LFDP, all free-standing wood
stems ≥1 cm dbh are mapped, measured and identified to species.
Tree measurement protocols follow those employed by the Center for
Tropical Forestry Science (CTFS) and large forest dynamics plots
across the globe (Condit 1998; Anderson-Teixeira et al. 2015). The
LFDP was established in 1990, soon after Hurricane Hugo, which
heavily damaged the forest (Scatena & Larsen 1991; Zimmerman
et al. 1994). A second hurricane, Georges, passed over the forest in
1998, but caused much less damage. Land-use legacies resulting from
settlement, coppicing and charcoal production prior to the 20th cen-
tury have emerged as a key driver of forest succession within the
northern LFDP. The LFDP is made up of three areas of secondary
forest of differing past land-use intensity (all three areas ≤80% forest
cover in 1936 aerial photographs; Foster, Fluet & Boose 1999), which
were combined to form one heterogeneous patch of secondary forest
that we refer to as the high land-use intensity area in the north of the
plot (Thompson et al. 2002; Uriarte et al. 2009). This area was com-
pletely cut over at some point prior to 1934 (Thompson et al. 2002).
The southern third of the plot contains relatively well-conserved area
of tabonuco forest, which harboured > 80% canopy cover in histori-
cal aerial photographs, and was only subjected to minor selective log-
ging in the 1940s for tests of release thinning around individual tress
(Frank Wadsworth, personal communication). Following major hurri-
canes in 1928 and 1932, it is believed that farming died out in the
area and stopped entirely in 1934 when the forest service bought the
land (Gerhart 1934; Weaver 2012), creating a land-use legacy gradi-
ent, of four distinct forest patches, that directly relates to forest
successional stage across the plot (Fig. 1).
QUANTIFY ING ENVIRONMENTAL VARIABLES WITHIN
THE LFDP
Field soil collection took place in 2011. Soil samples were taken from
0 to 10 cm depth in the mineral soil across a 40-m grid within the
LFDP following methodology in John et al. (2007), with additional
samples taken at 2 m and 8 m in random directions from selected
regular sample points to estimate local variability in soil chemistry.
Soil pH was determined in both deionized water and 10 mM CaCl2 in
a 1:2 soil to solution ratio using a glass electrode. Concentrations of
ammonium and nitrate were determined by extraction of fresh soils in
the field in 2 M KCl, with detection by automated colorimetry on a
Lachat Quikchem 8500 (Hach Ltd, Loveland, CO, USA). Plant-avail-
able phosphorus was estimated by extraction of air-dried and sieved
(< 2 mm) samples by Bray-1 solution, with detection of phosphate by
automated molybdate colorimetry. Exchangeable cations, including
aluminium, calcium, iron, magnesium, manganese and sodium, were
extracted from air-dried soils in 0.1 M BaCl2 (2 h, 1:30 soil to solu-
tion ratio), with detection by ICP–OES spectrometry on an Optima
7300 DV (PerkinElmer Inc, Shelton, CT, USA; Hendershot, Lalande
& Duquette 2008). Total exchangeable bases (TEB) were calculated
as the sum of calcium (Ca), potassium (K), magnesium (Mg) and
sodium (Na); effective cation exchange capacity (ECEC) was calcu-
lated as the sum of aluminium (Al), Ca, iron (Fe), K, Mg, manganese
(Mn) and Na; base saturation was calculated by (TEB/ECEC) 9 100.
Values were kriged to obtain soil cation estimates at the 20 9 20-m
scale.
Topographic variation within the LFDP is more variable than in
most other permanent forest dynamics plots (Losos & Leigh 2004).
The CTFS method for measuring topographic variability in large
FDPs is at the 20-m, or quadrat, scale (Condit 2012). Using previ-
ously surveyed elevations for quadrat corners within the LFDP (16
columns 9 25 rows; see Thompson et al. 2002), finer-scale topo-
graphic variation was measured at the 5-m scale using clinometers to
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
1468 J. A. Hogan et al.
measure the elevation of 5 9 5 m grid points delineating the subplots
and was used to calculate topographic variation. Elevation throughout
the LFDP is shown in Fig. 1, with the most elevated areas towards
the south-eastern plot boundary and the lowest elevations in the
north-west corner.
Topographic variables included slope, convexity, aspect and terrain
ruggedness. We incorporated a unique topographic variable, terrain
ruggedness, which is a quantifiable measure of terrain heterogeneity.
Slope was calculated using the CTFS method, where 4 planes are
drawn through all possible, unique combinations of 3 quadrat corners
at 5-m intervals within a 20 m 9 20 m plot and subsequently aver-
aged (Condit 2012). Convexity was calculated using the mean eleva-
tion of each quadrat relative to the mean of its eight orthogonal
neighbours (Condit 2012). Aspect was decomposed into east–west
and north–south orientations using sine and cosine, respectively (Lin
et al. 2013). The terrain ruggedness index, quantified for at the 20-m
scale, was implemented as defined by Riley (1999) and calculated in
GRASS (Grass Development Team 2012) using the 5-m scale topo-
graphic variation within the LFDP. We refer to environmental vari-
ables, hereafter, as the complete of set of the kriged soil nutrient
estimates and topographic variables, excluding spatial variables.
BETA-D IVERSITY MAPS
To visually examine variation within the forest community across the
LFDP, species diversity maps were generated using floral gradient
analysis (Thessler et al. 2005). Using Bray–Curtis dissimilarities,
non-metric multidimensional scaling (NMS) ordinations for each cen-
sus at the quadrat scale (20 9 20 m) were constrained to three axes.
Ordination axes scores, or their position in the 3-dimensional ordina-
tion space, were translated to hexadecimal colour values for red, blue
and green, respectively. The resulting maps display diversity differ-
ences in community composition chromatically (Baldeck et al. 2013).
Quadrats with colours of similar shades between quadrats represent
similar species composition, whereas colours of different shades show
areas where tree species composition is more different. Within each
census, redder shades were standardized within each census to show
areas where compositional make-up was most similar. NMS and col-
our standardization was done separately for each tree census; there-
fore, maps created using this method cannot be interpreted in relation
to one another (i.e. across censuses).
REDUNDANCY ANALYSIS /VARIANCE PARTIT IONING
We used canonical distance-based redundancy analysis (db-RDA;
Legendre & Anderson 1999), to examine the amount of variance
explained by environmental differences related to soil resources,
topography and space (Fig. 3; Peres-Neto et al. 2006). Redundancy
analysis, an extension of canonical correspondence analysis, is a
method of multivariate linear predictive models for a combination of
response variables (in our case, community composition) from a set
of predictor variables (in our case, space, topographic and soil nutri-
ent variables; Quinn & Keough 2002; McCune, Grace & Urban
2002). We chose to use RDA to test how well environmental predic-
tor variables, including space, could predict community composition
as a direct test for plant community–environment relationships. Hel-
linger-transformed species abundance counts within quadrats made up
the set of response variables (Legendre & Gallagher 2001; Dray,
Legendre & Peres-Neto 2006).
Due to the autocorrelated nature of tree species distributions,
space must be explicitly incorporated into the RDA (Legendre
1993). Principle coordinates of neighbourhood matrices (PCNM) is
currently the most appropriate tool for this (Borcard & Legendre
2002; Dray, Legendre & Peres-Neto 2006), because it incorporates
the spatial structure in species response data at the most relevant
species–environmental scale. PCNM decomposes the spatial structure
present in the community data into distance-based Moran’s Eigen-
vector Maps (db-MEM; Borcard et al. 2004; Borcard & Legendre
2002). The db-MEM comprised the set of all spatial variables. The
PCNM was computed in the R statistical environment (ver. 3.1.1; R
Development Core Team 2014) with the ‘PCNM’ package (Legen-
dre et al. 2010).
We expanded the set of environmental variables to permit for non-
linear relationships and increase model flexibility, by calculating sec-
ond- and third-order polynomials of all variables, except for aspect,
prior to their forward selection (Legendre et al. 2009). Forward selec-
tion of variables is one of the simplest data-driven model building
approaches, where explanatory variables are added to a model sepa-
rately and successively until no variables can be further added to
improve model fit or flexibility. Prior to variance partitioning,
explanatory variables were standardized and selected at the 95% con-
fidence interval using forward selection in the ‘packfor’ package
(Dray, Legendre & Blanchet 2007; Blanchet, Legendre & Borcard
2008). Three-way variance partitioning of response variables with
Fig. 1. Historical land-use demarcations within the LFDP (Thompson
et al. 2002), based on the amount of forest canopy in 1936 aerial
photographs (Foster, Fluet & Boose 1999). Per cent forest cover in
1936, in order from lightest to darkest: 0–20%, 20–50%, 50–80% and
80–100%. The low land-use intensity area is shown in the darkest
shade, and the three lightest shaded areas, together, make up the high
land-use intensity portion of the plot.
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1469
respect to spatial, topographic and edaphic variables was carried out
using the ‘vegan’ package (Oksanen et al. 2008).
PERMUTATIONAL ANALYSIS OF VARIANCE AND NON-
METRIC MULT ID IMENSIONAL SCALING BI -PLOTS
Differences in of the community beta-diversity with respect to past
land use were further explored by quantifying the magnitude of dif-
ference between quadrat groups of varying land-use intensity across
the LFDP (as determined by 1936 aerial photographs, Fig. 1) using
permutational multivariate analysis of variance (PERMANOVA) and
ordinating them using non-metric multidimensional scaling (NMS).
PERMANOVA was performed in the ‘vegan’ R package using the com-
munity species counts from the most recent (2011) tree census and
the a priori land-use classifications (Thompson et al. 2002). PER-
MANOVA P-values were Bonferroni-adjusted to correct for multiple
comparisons. For the NMS, forward-selected explanatory variables
from RDA models were overlain as environmental vectors on the
ordination, using 0.100 as the coefficient of determination cut-off to
allow for weak to moderate environmental variable relationships
(McCune, Grace & Urban 2002). Axis 1 in the ordination was
rotated with respect to land-use intensity and therefore can be
directly interpreted as an explanatory axis for land use. Visually,
this allows for the display of environmental–community relation-
ships across the entire plot. NMS ordinations were completed in
PC-ORD 6 (McCune, Grace & Urban 2002; McCune & Mefford
2011).
Results
BETA-DIVERSITY MAPS
Quadrats of similar species composition consistently grouped
together for each census in relation to past land-use intensity.
Quadrats in the northern portion of the LFDP, subject to more
intense land use, were clearly similar in community composi-
tion, as were quadrats in the southern portion where land use
was less intense. A second area of high community composi-
tional similarity was on the ridge at the western edge of plot
(~x = 100 m, y = 200 m), and compositional differences
were captured along the elevation gradient from the upper left
corner (north-west LFDP) to the bottom right corner of the
maps (south-east LFDP; Fig. 2; see Appendix S1 of Support-
ing Information). Most notably, the dominant colours in the
beta-diversity maps changed greatly for census 2 (1995). Red-
der shades (lowest compositional variability between quad-
rats) became less prominent in the high land-use intensity
portion (north) of the plot in 1995 due to the widespread
recruitment of understory pioneer species in the high-intensity
past land-use secondary forests following damage caused by
Hurricane Hugo and then switched back to more-red colours
and more compositional variability, recorded in the 2000 cen-
sus, as mortality in the understory occurs as the canopy
recovers (Hogan 2015).
REDUNDANCY ANALYSIS /VARIANCE PARTIT IONING
Across the entire LFDP, over half of the variability in tree
species community composition was related to spatial
(db-MEM) and environmental variables (Fig. 3; see
Appendix S2). Although environmental fractions (i.e. those
related to either soil or topography) were explanatory to some
degree, the portion of variability explained by spatial vari-
ables accounted for most of the variance explained by RDA
models. Generally, there was a decreasing trend in explana-
tory power over time for all RDA models, both at the whole-
plot level (Fig. 3) and with respect to past land-use intensity
(i.e. north vs. south; Figs. 4 and 5). Additionally, RDA mod-
els had greater explanatory power in areas of more intensive
past land use (the northern part of the LFDP; see Figure S1
of Supporting Information). For the high land-use intensity
part of the plot, environmental and spatial variables explained
from 57% to 44% of community composition over time
(Fig. 4). In contrast, the low land-use intensity area of tabo-
nuco forest (i.e. the southern part of the LFDP) had weaker
environmental–community relationships, with models explain-
ing only 38–44% of variation in community composition
(Fig. 5).
In all models, spatial variables accounted for the majority
of variation in community composition, explaining between
38% and 57% of the variability in community composition.
For the entire LFDP, soil cation concentrations and topo-
graphic variables only explained 30–33% and 11–14% of
variation, respectively (Fig. 3). The explanatory power of
environmental fractions was similar in magnitude, but more
variable for the high land-use intensity areas of the LFDP
(Fig. 4). However, in forest areas with low levels of previous
land-use, soil nutrients and topographic variation explained
less of the difference in community composition than in the
high land-use intensity areas, ranging between 23% and 27%
and between 14% and 16%, respectively (Fig. 5).
PERMANOVA AND NMS BI -PLOT
Community composition differed across the land-use gradi-
ent present in the LFDP, as shown in the beta-diversity
maps (Fig. 2). These differences have persisted over the
21 year study period and were found to be statistically sig-
nificant (PERMANOVA, P < 0.01) for both the high versus low
land-use intensity comparison and also when comparing all
four land-use intensities separately (Table 1; see
Appendix S3). In the NMS of the most recent tree census
(Fig. 6; 3-dimensional solution based on 125 iterations with
a final stress of 14.89 using Bray–Curtis distance measure),
a slight separation of the communities can be seen, with the
high land-use intensity quadrats mainly clustering out more
to the left hand side of axis 1 (land-use intensity), and with
the lower land-use intensity quadrats tending towards the
right side of the ordination. Convex hulls, or a polygons
containing all the sites within each land-use category, were
overlain on the ordination to display a tendency for more
compositional variability in forest community across the area
of more intense past land use in the LFDP. Axis 2 can be
interpreted as a catena effect (convexity) from high (ridge)
to low (valley) going from bottom to top of the axis. Base
saturation of soil (BS) decreased along the catena (axis 2).
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
1470 J. A. Hogan et al.
In 2011, all soil nutrient concentrations, and topographic
convexity, were key in explaining differences in environ-
mental–forest community relationships with respect to past
land-use intensity, and were positively correlated with
amount of canopy cover present in the 1936 historical aerial
photographs (i.e. past land-use intensity). For example, con-
centrations of exchangeable K, Na, Ca, Mg, Fe and Mn
were greater in less-disturbed sites. Total exchangeable bases
(TEB) and effective cation exchange capacity (ECEC) were
also greater in these areas. Aluminium, soil bulk density and
soil pH were the main notable exceptions, which appeared
on NMS bi-plots perpendicular to axis 1 (land use), meaning
that they had no noticeable relationships with past land-use
intensity. Similarly, topographic variables were neutrally or
weakly negatively related to past land use when examined
across the entire LFDP (Fig. 6).
Fig. 3. Variance partitioning diagram (upper left panel) and results for the entire LFDP by tree census. Census 1 (top-middle panel) took place
from 1990 to 1992, and subsequent censuses occurred at approximately 5-year intervals with census 5 (lower right panel) occurring in 2011.
Numerical values represent the per cent of variance explained by distance-based RDA models for their respective fractions (values <0 not shown).
Spatial variables (principle coordinates of neighbourhood matrix (Dray, Legendre & Peres-Neto 2006)) are presented in red, edaphic variables in
green and topographic variables in blue. Fraction a = the pure spatial component (space | environment). Fractions b + d = the proportion
explained by soil after accounting for topography (soil | topography). Fractions c + f = the proportion explained by topography after accounting
for soil (topography | soil). Fractions e + g = the topographically structured soil component (soil and topography). Fractions d + f + g = the spa-
tially structured environmental component (space and environment). Fractions a + d + f + g = the proportion of variation explained by spatial
variables (PCNM) alone (space). Fractions b + d + e + g = the proportion explained by only soil nutrients. Fractions c + e + f + g = the propor-
tion explained by topographic variables solely. Fractions b + c + d + e + f + g = the proportion explained by the environmental variables (soil
resources and topography). Total explanatory power of db-RDA models = the percentage of variability explained by all spatial and environmental
variables combined (a + b + c + d + e + f + g, or 100  residuals (h)).
Fig. 2. Beta-diversity maps for the LFDP, created using floral gradient analysis (Thessler et al. 2005) for each of the five tree censuses. Colour
represents the degree of compositional similarity (raw interpretation of Bray–Curtis dissimilarities) of tree species community composition
between 20-m quadrats; quadrats of similar colour represent communities with similar species composition. Redder shades were standardized to
areas of greatest community compositional similarity for that census. Notably, maps cannot be compared across censuses as the method compares
compositional similarity between 20-m quadrats for each census separately. Red vertical lines show the hurricanes occurrences (Hugo in Septem-
ber 1989 and Georges in September 1998).
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1471
Discussion
To examine the direct effect of past land-use intensity in envi-
ronmental niche structuring of plant communities, the vari-
ance partitioning results from the LFDP can be compared to
those of five other larger FDPs; BCI in Panama, Huai Kha
Khaeng in Thailand, Korup in Cameroon, Pasoh in Malaysia
and Yasuni in Ecuador (Baldeck et al. 2013). In these plots,
both soil resources and topography explained about the same
amount of variation in plant community composition (9–34%
and 5–29%, respectively), and soil resources and topography
together explained 13–39% of compositional variation (Bal-
deck et al. 2013). In addition, these large FDPs (≥ 50 ha
each), which all largely lack previous well-documented land
use, were divided into half to aid in comparison with smaller
FDPs, such as the LFDP. Differences between the north and
south of the LFPD were greater than the other plots, espe-
cially following hurricane disturbances (i.e. census 1 in 1990
following Hurricane Hugo and census 3 in 2000 following
Hurricane Georges). Difference between the total explanatory
power of RDA models for five FDP halves in the Baldeck
et al. (2013) study averaged 0.04. Differences in RDA total
model explanatory power for the northern and southern halves
of the LFDP were 0.12, 0.08, 0.10, 0.04 and 0.06, for tree
census 1 through 5 respectively, indicating that differences
between plant communities in the LFDP tended to be greater
than those in other large plots largely lacking the effects of
past land use and hurricane damage.
The beta-diversity maps clearly showed the northern por-
tion of the LFDP to be distinct in species composition. This
area is the high land-use intensity portion of the plot, with
three distinct areas of differing canopy cover, all less than
80% in the historical 1936 aerial photographs (Foster, Fluet
& Boose 1999; Thompson et al. 2002). Historical land-use
intensity varies across the three distinct areas, with some areas
having been largely cut over, and others having been domi-
nated by either agricultural practices or small subsistent
human settlements (Thompson et al. 2002; Thompson, Lugo
Fig. 4. Variance partitioning diagram (upper
left panel) and results for the higher land-use
intensity portion (northern) LFDP by tree
census. Numerical values represent the per
cent of variance explained by distance-based
RDA models for their respective fractions.
Figure panels and their partitioned fractions
are identically labelled with respect to
Fig. 3a.
Fig. 5. Variance partitioning diagram (upper
left panel) and results for the low land-use
intensity portion (southern) LFDP by tree
census. Numerical values represent the per
cent of variance explained by distance-based
RDA models for their respective fractions
(Values <0 not shown). Figure panels and
their partitioned fractions are identically
labelled with respect to Fig. 3a.
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
1472 J. A. Hogan et al.
& Thomlinson 2007). Land-use legacies are reflected in a
structurally younger forest where secondary forest species
with lower wood density are more abundant (Thompson et al.
2002; Uriarte et al. 2009) and canopy structure allows more
light through (Comita et al. 2010; Hogan 2015). These areas
were more affected by hurricanes (Zimmerman et al. 1995;
Canham et al. 2010; Uriarte et al. 2012) and have resulted in
secondary forests that have faster dynamics driven by greater
light availability in the understorey after hurricane damage.
The beta-diversity maps likely capture distance-restricted dis-
persal and recruitment within high-intensity land-use areas,
leading to greater similarity in community composition at
census 2, with community differences increasing over time as
shrub and pioneer species densities thin under a recovering
canopy (Hogan 2015). These results are similar to responses
measured in nearby USDA Forest Service plots, in terms of
similarity of community composition along successional tra-
jectories in time after hurricane disturbance (Heartsill Scalley
et al. 2010).
RDA models decreased in explanatory power over time.
Despite these changes, variance partitioning results were
stable over time, when compared to changes in stem densities,
stem diameter size distributions (i.e. forest structure) or spe-
cies diversity (i.e. compositional change) within the forest
(Hogan 2015). The better explanatory power of RDA models
in areas of more intense past land use in the LFDP points to
the large magnitude of change, and persisting effects of
anthropogenic land-use disturbance, within forest communi-
ties. This trend is well documented in vegetation succession
of tropical forest systems (Aide et al. 1995; Foster, Motzkin
& Slater 1998; Chazdon 2003; Holz, Placci & Quintana
2009). Clear differences between RDA models for areas of
high and low land-use intensity were observed. RDA models
had greater explanatory power in areas more intense past
land-use history within the LFDP, most likely due to stronger
successional responses to hurricane damage in these areas. In
other words, areas more impacted by previous land use had
greater plant community–environmental relationships, because
they respond disproportionately to hurricanes, at least with
respect to recruitment of understorey pioneers (Hogan 2015).
Such pioneer species often have considerably shorter life
spans than their primary forest counterpart species and, there-
fore, can more closely track successional changes related to
spatial environmental variation (i.e. light availability; Uriarte
et al. 2009, 2012), providing an explanation for why RDA
models had increased explanatory power in areas of more
intensive past land use.
Table 1. Results for permutational multivariate analysis of variance (permanova) for Bray–Curtis dissimilarities for the 2011 (census 5) tree com-
munity within the LFDP in relation to the land-use intensity (i.e. the amount of canopy cover in 1936 aerial photographs). DF = degrees of free-
dom, SS = sum of squares, Pseudo-F = F value by permutation. Bold face indicates statistical significance (P < 0.05); P-values are based on 999
permutations (i.e. the lowest possible P-value is 0.001).
Land-use Intensity
Comparison
(percentage
canopy cover in 1936) DF SS Pseudo-F R2 P-value
Bonferroni-adjusted
P-value
80–100% vs. 50–80% 1 8.916 64.887 0.194 0.001 0.006
80–100% vs. 20–50% 1 6.289 45.782 0.167 0.001 0.006
80–100% vs. 0–20% 1 4.536 35.099 0.182 0.001 0.006
50–80% vs. 20–50% 1 2.195 15.502 0.061 0.001 0.006
50–80% vs. 0–20% 1 1.643 12.108 0.067 0.001 0.006
20–50% vs. 0–20% 1 0.929 6.877 0.052 0.001 0.006
< 80% vs. ≥ 80% 1 10.367 71.687 0.152 0.001 0.001
Fig. 6. Non-metric multidimensional scaling (NMS) ordination of tree
communities of the LFDP for the most recent tree census (2011).
Black circles indicate low-intensity land-use quadrats (80–100% forest
cover in 1936), while grey triangles show quadrats in the high-inten-
sity land-use portion of the LFDP. A three-dimensional solution was
achieved with an average plot stress of 14.89 after 125 iterations;
axes 1 and 2 are shown. Overlain vectors for forward-selected
explanatory environmental variables (either topographic or edaphic)
from RDA models with an r2 value ≥ 0.100 are shown. The length of
the vector is proportional to the strength of the environmental–com-
munity relationship for that variable. Axis 1 was oriented with respect
to land-use strength from high intensity to low intensity going from
left to right.
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1473
Separation of quadrats within the LFDP with respect to
past land use and as the forest recovers from Hurricane Hugo,
which affected the forest in 1989, and to a much lesser degree
Hurricane Georges which passed over the LFDP in 1998, can
still be seen via community ordination. There is a greater
clustering of quadrats, as community composition between
areas of differing past land-use pressure converge in commu-
nity composition, with the widespread recruitment of pioneers
across the whole plot. In community ordinations for all five
LFDP tree censuses, compositional differences attributed to
land-use legacies peaked at census 3 in 2000, with land-use
legacies explaining 58.6% of the plot-wise compositional
variation. This trend is congruent with the pattern of the mag-
nitude of land-use legacies within the LFDP, explored using
statistical techniques that quantify the contribution of past
land use to changes in community composition over time
(Hogan 2015).
Environmental–community relationships within tropical for-
ests are difficult to detect and are strongest in well-conserved
areas that exhibit high levels of environmental heterogeneity
(Paoli, Curran & Zak 2006). In this study, we investigated the
interactive effects of past land use and hurricane disturbance
in altering environmental relationships. The results supported
our hypothesis that associations between plant community
compositional and the environment would be stronger follow-
ing hurricane damages to the forest (Introduction – Question
2). Despite this, environmental niche partitioning (i.e. topo-
graphic and soil niche) was greatest in the areas of more
intense past land-use pressure, recorded in the census immedi-
ately following Hurricane Hugo, providing evidence that land
use alters abiotic and biotic variables within the forest. Alter-
ation of the abiotic environment by land use selects for spe-
cies suited for these conditions (e.g. secondary successional
and pioneer species) to allow for competitive advantages for
anthropogenically associated species (e.g. species commonly
associated with human settlement or other land-use activity).
Such altered forest communities can persist for decades or
longer following land-use abandonment, as shown via the
lasting land-use legacies within the LFDP. Although these
findings are not new (Foster, Motzkin & Slater 1998; Thomp-
son et al. 2002), the possible explanatory mechanisms deserve
further research.
We clearly were incorrect in expecting weaker plant com-
munity–environmental relationships in areas of higher past
land-use pressure (Introduction – Question 1). Contrary to our
expectation, environmental–plant community structuring was
greater in areas of more intense past land use, probably due
to decreased competition in these areas when compared to
old-growth tabonuco forest. A similar study (Bachelot et al.
2016) found below-ground biotic factors (e.g. soil fungi and
microbes) to be more important in affecting community com-
position in the high land-use areas of the LFDP, suggesting
that land-use legacies not only alter soil fertility and structure,
but also biotic communities within them, providing an insight
into mechanisms for the lasting effects of land-use legacies
on forest species composition via soils. This research is
important because a widespread forest regrowth continues to
take place in the tropics as land-use patterns change (Mather
1992; Rudel, Bates & Machinguiashi 2002; Meyfroidt &
Lambin 2009; Chazdon 2014). Regenerating tropical forests
may resemble their old-growth counterparts in some metrics
(e.g. structure), but it may take centuries or longer to recover
old-growth species composition and interactions (Chazdon
2003; Lugo & Helmer 2004; Wright 2005; Lindenmayer,
Laurance & Franklin 2012).
In complex tropical forests, identifying environmental dri-
vers that organize plant communities is extremely difficult.
Many signals tend to be weak and confounded with other fac-
tors, such as disturbance. Despite this, when comparing two
areas of forest with differing previous land-use histories
within the LFDP, environmental relationships were stronger
in areas of more intense past land-use pressure. These find-
ings provide a mechanistic understanding of the possible role
that abiotic conditions associated with land-use legacies play
in altering the successional dynamics of recovering secondary
tropical forests. That mechanism is through the direct anthro-
pogenic alteration of forest composition and structure via
human land-use practices, likely providing competitive advan-
tages to secondary forest, or oftentimes anthropogenically
selected species due to the more rapid turnover and stronger
spatial interactions present in second-growth tropical forests.
Acknowledgements
This research was funded by grants BSR-8811902, DEB-9411973, DEB-
9705814, DEB-0080538, DEB-0218039, DEB-0516066 and DEB-0620910 from
the National Science Foundation to the International Institute for Tropical For-
estry, USDA Forest Service, as part of the Luquillo Long-Term Ecological
Research Program. Additional direct support was provided by the University of
Puerto Rico – Rıo Piedras (UPR–RP), the USDA Forest Service, the Andrew W.
Mellon Foundation and the Center for Tropical Forest Science. Financial support
to JAH was provided by NSF Grant HRD#1139888 through The Resource Cen-
ter for Science and Engineering at UPR–RP. We are grateful to Benjamin Bra-
noff for his help with the generation of some topographic variables. We thank
Tania Romero, Chris Nytch, James Dalling and Claire Baldeck for their contri-
bution to soil nutrient mapping. Lastly, we acknowledge the > 100 volunteers
and staff that have assisted in the five complete tree censuses of the LFDP.
Data accessibility
Tree census data for the Luquillo Forest Dynamics Plot are available for down-
load through the Smithsonian Tropical Research Institute ForestGEO (formerly
The Center for Tropical Forest Studies) data portal: http://ctfs.si.edu/Public/plot
dataaccess/index.php.
References
Aide, T.M., Zimmerman, J.K., Herrera, L., Rosario, M. & Serrano, M. (1995)
Forest recovery in abandoned tropical pastures in Puerto Rico. Forest Ecol-
ogy and Management, 77, 77–86.
Anderson-Teixeira, K.J., Davies, S.J., Bennett, A.C., Gonzalez-Akre, E.B.,
Muller-Landau, H.C., Joseph Wright, S., Abu Salim, K., Almeyda Zambrano,
A.M., Alonso, A. & Baltzer, J.L. (2015) CTFS-ForestGEO: a worldwide net-
work monitoring forests in an era of global change. Global Change Biology,
21, 528–549.
Bachelot, B., Uriarte, M., Zimmerman, J.K., Thompson, J., Leff, J.W., Asiaii,
A., Koshner, J. & McGuire, K. (2016) Long-lasting effects of land use his-
tory on soil fungal communities in second-growth tropical rain forests. Eco-
logical Applications. doi: 10.1890/15-1397.1.
Baldeck, C.A., Harms, K.E., Yavitt, J.B., John, R., Turner, B.L., Valencia, R.
et al. (2013) Soil resources and topography shape local tree community
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
1474 J. A. Hogan et al.
structure in tropical forests. Proceedings of the Royal Society B: Biological
Sciences, 280, 20122532.
Beinroth, F.H. (2010) Updated Taxonomic Classification of the Soils of Puerto
Rico, 2002. University of Puerto Rico-Mayaguez, Agricultural Research Sta-
tion, Rio Piedras, Puerto Rico.
Blanchet, F.G., Legendre, P. & Borcard, D. (2008) Forward selection of
explanatory variables. Ecology, 89, 2623–2632.
Booth, T.H., Nix, H.A., Hutchinson, M.F. & Jovanic, T. (1988) Niche analysis
and tree species introduction. Forest Ecology and Management, 23, 47–59.
Borcard, D. & Legendre, P. (2002) All-scale spatial analysis of ecological data
by means of principal coordinates of neighbour matrices. Ecological Model-
ling, 153, 51–68.
Borcard, D., Legendre, P., Avois-Jacquet, C. & Tuomisto, H. (2004) Dissecting
the spatial structure of ecological data at multiple scales. Ecology, 85, 1826–
1832.
Brokaw, N. & Busing, R.T. (2000) Niche versus chance and tree diversity in
forest gaps. Trends in Ecology & Evolution, 15, 183–188.
Brokaw, N., Fraver, S., Grear, J., Thompson, J., Zimmerman, J., Waide, R.,
Everham, E. III, Hubbell, S. & Foster, R. (2004) Disturbance and canopy
structure in two tropical forests. Tropical Forest Diversity and Dynamism:
Findings from a Large-scale Plot Network (eds E.C. Losos & E.G. Leigh),
pp. 177–194. University of Chicago Press, Chicago, IL, USA.
Canham, C.D., Finzi, A.C., Pacala, S.W. & Burbank, D.H. (1994) Causes and
consequences of resource heterogeneity in forests: interspecific variation in
light transmission by canopy trees. Canadian Journal of Forest Research,
24, 337–349.
Canham, C.D., Thompson, J., Zimmerman, J.K. & Uriarte, M. (2010) Variation
in susceptibility to hurricane damage as a function of storm intensity in
Puerto Rican tree species. Biotropica, 42, 87–94.
Chapin, F.S., Walker, B.H., Hobbs, R.J., Hooper, D.U., Lawton, J.H., Sala,
O.E. & Tilman, D. (1997) Biotic control over the functioning of ecosystems.
Science, 277, 500–504.
Chazdon, R.L. (2003) Tropical forest recovery: legacies of human impact and
natural disturbances. Perspectives in Plant Ecology, Evolution and Systemat-
ics, 6, 51–71.
Chazdon, R.L. (2014) Second Growth: The Promise of Tropical Forest Regenera-
tion in an Age of Deforestation. University of Chicago Press, Chicago, IL, USA.
Cleveland, C.C., Townsend, A.R., Taylor, P., Alvarez-Clare, S., Bustamante, M.,
Chuyong, G., Dobrowski, S.Z., Grierson, P., Harms, K.E. & Houlton, B.Z.
(2011) Relationships among net primary productivity, nutrients and climate in
tropical rain forest: a pan-tropical analysis. Ecology Letters, 14, 939–947.
Comita, L.S., Thompson, J., Uriarte, M., Jonckheere, I., Canham, C.D. & Zim-
merman, J.K. (2010) Interactive effects of land use history and natural distur-
bance on seedling dynamics in a subtropical forest. Ecological Applications,
20, 1270–1284.
Condit, R. (1998) Tropical Forest Census Plots: Methods and Results from
Barro Colorado Island, Panama and a Comparison with Other Plots.
Springer, Georgetown, TX, USA.
Condit, R. (2012) CTFS R Package. Center for Tropical Forest Science http://
ctfs.si.edu/Public/CTFSRPackage/.
Cram, S., Sommer, I., Fern
andez, P., Galicia, L., R
ıos, C. & Barois, I. (2015)
Soil natural capital modification through landuse and cover change in a tropi-
cal forest landscape: implications for management. Journal of Tropical For-
est Science, 27, 189–201.
Davies, S.J., Palmiotto, P.A., Ashton, P.S., Lee, H.S. & Lafrankie, J.V. (1998)
Comparative ecology of 11 sympatric species of Macaranga in Borneo: tree
distribution in relation to horizontal and vertical resource heterogeneity. Jour-
nal of Ecology, 86, 662–673.
De C
aceres, M., Legendre, P., Valencia, R., Cao, M., Chang, L.W., Chuyong,
G., Condit, R., Hao, Z., Hsieh, C.F. & Hubbell, S. (2012) The variation of
tree beta diversity across a global network of forest plots. Global Ecology
and Biogeography, 21, 1191–1202.
R Development Core Team (2014) R: A Language and Environment for Statis-
tical Computing. R Foundation for Statistical Computing, Vienna, Austria.
Ding, Y., Zang, R., Liu, S., He, F. & Letcher, S.G. (2012) Recovery of woody
plant diversity in tropical rain forests in southern China after logging and
shifting cultivation. Biological Conservation, 145, 225–233.
Dray, S., Legendre, P. & Blanchet, F. (2007) Packfor: forward selection with
permutation. R package version 0.0–7.
Dray, S., Legendre, P. & Peres-Neto, P.R. (2006) Spatial modelling: a compre-
hensive framework for principal coordinate analysis of neighbour matrices
(PCNM). Ecological Modelling, 196, 483–493.
Duan, W.J., Ren, H., Fu, S.L., Guo, Q.F. & Wang, J. (2008) Pathways and
determinants of early spontaneous vegetation succession in degraded lowland
of South China. Journal of Integrative Plant Biology, 50, 147–156.
Ellis, R. & Pennington, P. (1992) Factors affecting the growth of Eucalyptus
delegatensis seedlings in inhibitory forest and grassland soils. Plant and Soil,
145, 93–105.
Foster, D., Fluet, M. & Boose, E. (1999) Human or natural disturbance: land-
scape-scale dynamics of the tropical forests of Puerto Rico. Ecological Appli-
cations, 9, 555–572.
Foster, D.R., Motzkin, G. & Slater, B. (1998) Land-use history as long-term
broad-scale disturbance: regional forest dynamics in central New England.
Ecosystems, 1, 96–119.
Gerhart, G. (1934) Tract 11. Land Acquisition Supplementary Report, Luquillo
Purchase Unit. Caribbean National Forest. USDA Forest Service. Open File
Report. International Institute of Tropical Forestry, Catalina Service Station,
Palmer, Puerto Rico.
Grass Development Team (2012) Geographic Resources Analysis Support Sys-
tem (GRASS GIS) Software. Open Source Geospatial Foundation, USA.
Available at http://grass.osgeo.org.
Harms, K.E., Wright, S.J., Calder
on, O., Hern
andez, A. & Herre, E.A. (2000)
Pervasive density-dependent recruitment enhances seedling diversity in a
tropical forest. Nature, 404, 493–495.
Harris, N. L., Lugo, A. E., Brown, S. & Heartsill Scalley, T. (2012) Distur-
bances. Luquillo Experimental Forest: Research History and Opportunities.
(eds N.L. Harris, A.E. Lugo, S. Brown & T. Heartsill Scalley), pp. 152.
U.S.D.A., Washington, DC, USA.
Hautier, Y., Tilman, D., Isbell, F., Seabloom, E.W., Borer, E.T. & Reich, P.B.
(2015) Anthropogenic environmental changes affect ecosystem stability via
biodiversity. Science, 348, 336–340.
Heartsill Scalley, T., Scatena, F.N., Lugo, A.E., Moya, S. & Estrada Ruiz, C.R.
(2010) Changes in Structure, Composition, and Nutrients During 15 Yr of
Hurricane-Induced Succession in a Subtropical Wet Forest in Puerto Rico.
Biotropica, 42, 455–463.
Hendershot, W.H., Lalande, H. & Duquette, M. (2008) Ion exchange and
exchangeable cations. Soil Sampling and Methods of Analysis (eds M.R. Car-
ter & E. Gregorich), pp. 173–178. Canadian Society of Soil Science and
CRC Press, Boca Raton, FL, USA.
Hogan, J. A. (2015) Revisiting the Relative Roles of Land-Use and the Envi-
ronment in Subtropical Wet Forest: 21-years of Dynamics from the Luquillo
Forest Dynamics Plot, Puerto Rico. MSc Thesis. University of Pierto Rio,
Rio Piedras (Puerto Rico).
Holdridge, L.R., Grenke, W., Hatheway, W.H., Liang, T. & Tosi Junior, J.A.
(1971) Forest Environments in Tropical Life Zones: A Pilot Study. Oxford
Pergamon Press, New York, NY, USA.
Holz, S., Placci, G. & Quintana, R.D. (2009) Effects of history of use
on secondary forest regeneration in the Upper Parana Atlantic Forest
(Misiones, Argentina). Forest Ecology and Management, 258, 1629–
1642.
Hu, Y.-H., Lan, G.-Y., Sha, L.-Q., Cao, M., Tang, Y. & Xu, D.-P. (2012)
Strong neutral spatial effects shape tree species distributions across life stages
at multiple scales. PLoS ONE, 7, e38247.
Hubbell, S. P. (2001) The Unified Neutral Theory of Biodiversity and Biogeog-
raphy. Princeton University Press, Princeton, NJ, USA.
Hutchinson, G. (1957) Concluding remarks: Cold Spring Harbor Symposia on
Quantitative Biology. Quantitative Biology, 22, 415–427.
Hutchinson, G.E. (1978) An introduction to Population Ecology. Yale Univer-
sity Press, New Haven, CT, USA.
Jansen, P.A., Visser, M.D., Wright, S.J., Rutten, G. & Muller-Landau, H.C.
(2014) Negative density dependence of seed dispersal and seedling recruit-
ment in a Neotropical palm. Ecology Letters, 17, 1111–1120.
Janzen, D.H. (1970) Herbivores and the number of tree species in tropical for-
ests. The American Naturalist, 104, 501–528.
John, R., Dalling, J.W., Harms, K.E., Yavitt, J.B., Stallard, R.F., Mirabello, M.,
Hubbell, S.P., Valencia, R., Navarrete, H. & Vallejo, M. (2007) Soil nutri-
ents influence spatial distributions of tropical tree species. Proceedings of the
National Academy of Sciences, 104, 864–869.
Kanagaraj, R., Wiegand, T., Comita, L.S. & Huth, A. (2011) Tropical tree spe-
cies assemblages in topographical habitats change in time and with life stage.
Journal of Ecology, 99, 1441–1452.
Lalibert
e, E., Zemunik, G. & Turner, B.L. (2014) Environmental filtering
explains variation in plant diversity along resource gradients. Science, 345,
1602–1605.
Lalibert
e, E., Grace, J.B., Huston, M.A., Lambers, H., Teste, F.P., Turner, B.L.
& Wardle, D.A. (2013) How does pedogenesis drive plant diversity? Trends
in Ecology & Evolution, 28, 331–340.
Lan, G., Hu, Y., Cao, M. & Zhu, H. (2011) Topography related spatial distri-
bution of dominant tree species in a tropical seasonal rain forest in China.
Forest Ecology and Management, 262, 1507–1513.
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1475
Lawrence, W.T.J. (1996) Plants: the food base. The Food Web of a Tropical
Rain Forest (eds D.P. Reagan & R.B. Waide), pp. 17–51. University of Chi-
cago Press, Chicago, IL, USA.
Legendre, P. (1993) Spatial autocorrelation: trouble or new paradigm? Ecology,
74, 1659–1673.
Legendre, P. & Anderson, M.J. (1999) Distance-based redundancy analysis:
testing multispecies responses in multifactorial ecological experiments. Eco-
logical Monographs, 69, 1–24.
Legendre, P. & Fortin, M.J. (1989) Spatial pattern and ecological analysis.
Vegetatio, 80, 107–138.
Legendre, P. & Gallagher, E.D. (2001) Ecologically meaningful transformations
for ordination of species data. Oecologia, 129, 271–280.
Legendre, P., Mi, X., Ren, H., Ma, K., Yu, M., Sun, I.-F. & He, F. (2009) Par-
titioning beta diversity in a subtropical broad-leaved forest of China. Ecol-
ogy, 90, 663–674.
Legendre, P., Borcard, D., Blanchet, G. & Dray, S. (2010) PCNM: PCNM spatial
eigenfunction and principal coordinate analyses. R package version 2.1/r82.
Lin, G., Stralberg, D., Gong, G., Huang, Z., Ye, W. & Wu, L. (2013) Separat-
ing the effects of environment and space on tree species distribution: from
population to community. PLoS ONE, 8, e56171.
Lindenmayer, D.B., Laurance, W.F. & Franklin, J.F. (2012) Global decline in
large old trees. Science, 338, 1305–1306.
Losos, E.C. & Leigh, E.G. (2004) Tropical Forest Diversity and Dynamism:
Findings from a Large-Scale Plot Network. University of Chicago Press,
Chicago, IL, USA.
Lugo, A.E. & Helmer, E. (2004) Emerging forests on abandoned land: Puerto
Rico’s new forests. Forest Ecology and Management, 190, 145–161.
Martins, K.G., Marques, M.C.M., dos Santos, E. & Marques, R. (2015) Effects
of soil conditions on the diversity of tropical forests across a successional
gradient. Forest Ecology and Management., 349, 4–11.
Mather, A.S. (1992) The forest transition. Area, 24, 367–379.
McCune, B., Grace, J. B. & Urban, D. L. (2002) Analysis of Ecological Com-
munities. MjM Software Design Gleneden Beach, OR, USA.
McCune, B. & Mefford, M.J. (2011) PC-ORD 6: Multivariate Analysis of Eco-
logical Data. MjM Software Design, Gleneden Beach, OR, USA.
Meyfroidt, P. & Lambin, E.F. (2009) Forest transition in Vietnam and displace-
ment of deforestation abroad. Proceedings of the National Academy of
Sciences, 106, 16139–16144.
Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Simpson, G., Solymos, P.,
Stevens, M. & Wagner, H. (2008) The vegan package version 2.2-1. http://
CRAN.R-project.org/package=vegan.
Pacala, S.W. & Rees, M. (1998) Models suggesting field experiments to test
two hypotheses explaining successional diversity. The American Naturalist,
152, 729–737.
Paoli, G.D., Curran, L.M. & Zak, D.R. (2006) Soil nutrients and beta diversity
in the Bornean Dipterocarpaceae: evidence for niche partitioning by tropical
rain forest trees. Journal of Ecology, 94, 157–170.
Peres-Neto, P.R., Legendre, P., Dray, S. & Borcard, D. (2006) Variation parti-
tioning of species data matrices: estimation and comparison of fractions.
Ecology, 87, 2614–2625.
Punchi-Manage, R., Getzin, S., Wiegand, T., Kanagaraj, R., Savitri Gunatilleke,
C., Nimal Gunatilleke, I., Wiegand, K. & Huth, A. (2013) Effects of topogra-
phy on structuring local species assemblages in a Sri Lankan mixed diptero-
carp forest. Journal of Ecology, 101, 149–160.
Quinn, G.P. & Keough, M.J. (2002) Experimental Design and Data Analysis
for Biologists. Cambridge University Press, Cambridge, UK.
Ram
ırez, A. & Melendez-Colom, E. (2003). Meteorological Summary for El
Verde Field Station: 1975–2003 (ed I.T.E.S. U.P.R.), Institute for Tropical
Ecosystem Studies, University of Puerto Rico, San Juan, PR. Available at
http://luq.lternet.edu/data (accessed September 2015).
Ricklefs, R.E. (1977) Environmental heterogeneity and plant species diversity:
a hypothesis. American Naturalist, 111, 376–381.
Riley, S.J. (1999) A Terrain Ruggedness Index That Quantifies Topographic
Heterogeneity. Intermountain Journal of Sciences, 5, 23–27.
Rudel, T.K., Bates, D. & Machinguiashi, R. (2002) A tropical forest transition?
Agricultural change, out-migration, and secondary forests in the Ecuadorian
Amazon. Annals of the Association of American Geographers, 92, 87–102.
Scatena, F. & Larsen, M. (1991) Physical aspects of hurricane Hugo in Puerto
Rico. Biotropica, 23, 317–323.
Soil Survey Soil Staff (1995) Order I Soil Survey of the Luquillo Long-Term
Ecological Research grid, Puerto Rico. USDA, Natural Resources Conserva-
tion Services, Lincoln, NE, USA.
Sousa, W.P. (1984) The role of disturbance in natural communities. Annual
Review of Ecology and Systematics, 15, 353–391.
Tateno, M. & Chapin, F. III (1997) The logic of carbon and nitrogen interac-
tions in terrestrial ecosystems. American Naturalist, 149, 723–744.
Terborgh, J. (1992) Diversity and the Tropical Rain Forest. Scientific American
Library, New York, NY, USA.
Thessler, S., Ruokolainen, K., Tuomisto, H. & Tomppo, E. (2005) Mapping
gradual landscape-scale floristic changes in Amazonian primary rain forests
by combining ordination and remote sensing. Global Ecology and Biogeog-
raphy, 14, 315–325.
Thompson, J., Lugo, A.E. & Thomlinson, J. (2007) Land use history, hurricane
disturbance, and the fate of introduced species in a subtropical wet forest in
Puerto Rico. Plant Ecology, 192, 289–301.
Thompson, J., Brokaw, N., Zimmerman, J.K., Waide, R.B., Everham, E.M. III,
Lodge, D.J., Taylor, C.M., Garc
ıa-Montiel, D. & Fluet, M. (2002) Land use
history, environment, and tree composition in a tropical forest. Ecological
Applications, 12, 1344–1363.
Thompson, J., Brokaw, N., Zimmerman, J.K., Waide, R., Everham, E.M. III &
Schaefer, D.A. (2004) Luquillo Forest Dynamics Plot, Puerto Rico. Tropical
Forest Diversity and Dynamism: Findings from a Large-Scale plot network
(eds E.C. Losos & E.G. Leigh), pp. 540–550. Chicago University Press, Chi-
cago, IL, USA.
Tylianakis, J.M., Rand, T.A., Kahmen, A., Klein, A.-M., Buchmann, N., Per-
ner, J. & Tscharntke, T. (2008) Resource heterogeneity moderates the biodi-
versity-function relationship in real world ecosystems. PLoS Biology, 6,
e122.
Uriarte, M., Canham, C.D., Thompson, J., Zimmerman, J.K., Murphy, L.,
Sabat, A.M., Fetcher, N. & Haines, B.L. (2009) Natural disturbance and
human land use as determinants of tropical forest dynamics: results from a
forest simulator. Ecological Monographs, 79, 423–443.
Uriarte, M., Clark, J.S., Zimmerman, J.K., Comita, L.S., Forero-Monta~na, J. &
Thompson, J. (2012) Multidimensional trade-offs in species responses to dis-
turbance: implications for diversity in a subtropical forest. Ecology, 93, 191–
205.
Wadsworth, F. (1951) Ordenaci
on forestal en las monta~nas de Luquillo. Forest
management in the Luquillo mountains. Caribbean Forester (Puerto Rico),
12, 115–132.
Weaver, P.L. (2000) Environmental gradients affect forest structure in Puerto
Rico’s luquillo mountains. Interciencia, 25, 254–259.
Weaver, P.L. (2012) The Luquillo mountains: forest resources and their
history. General Technical Report-International Institute of Tropical Forestry,
USDA Forest Service.
White, P.S. (1979) Pattern, process, and natural disturbance in vegetation. The
Botanical Review, 45, 229–299.
Wilson, S. (2000) Heterogeneity, diversity and scale in plant communities.
The Ecological Consequences of Environmental Heterogeneity (eds. M.J.
Hutchings, E.A. John & A.J.A. Stewart), pp. 53–69. Blackwell Science,
Oxford, UK.
Wright, J.S. (2002) Plant diversity in tropical forests: a review of mechanisms
of species coexistence. Oecologia, 130, 1–14.
Wright, S.J. (2005) Tropical forests in a changing environment. Trends in Ecol-
ogy & Evolution, 20, 553–560.
Wright, S.J., Kitajima, K., Kraft, N.J., Reich, P.B., Wright, I.J., Bunker, D.E.,
Condit, R., Dalling, J.W., Davies, S.J. & D
ıaz, S. (2010) Functional traits
and the growth-mortality trade-off in tropical trees. Ecology, 91, 3664–
3674.
Zemunik, G., Turner, B.L., Lambers, H. & Lalibert
e, E. (2016) Increasing plant
species diversity and extreme species turnover accompany declining soil fer-
tility along a long-term chronosequence in a biodiversity hotspot. Journal of
Ecology, 104, 792–805.
Zimmerman, J.K., Everham, E.M. III, Waide, R.B., Lodge, D.J., Taylor, C.M.
& Brokaw, N.V. (1994) Responses of tree species to hurricane winds in sub-
tropical wet forest in Puerto Rico: implications for tropical tree life histories.
Journal of Ecology, 82, 911–922.
Zimmerman, J.K., Aide, T.M., Rosario, M., Serrano, M. & Herrera, L. (1995)
Effects of land management and a recent hurricane on forest structure and
composition in the Luquillo Experimental Forest, Puerto Rico. Forest Ecol-
ogy and Management, 77, 65–76.
Zimmerman, J.K., Willig, M.R., Walker, L.R. & Silver, W.L. (1996) Introduc-
tion: disturbance and Caribbean ecosystems. Biotropica, 28, 414–423.
Zimmermann, B., Papritz, A. & Elsenbeer, H. (2010) Asymmetric response to
disturbance and recovery: Changes of soil permeability under forest–pasture–
forest transitions. Geoderma, 159, 209–215.
Received 22 October 2015; accepted 13 May 2016
Handling Editor: Gabriela Bielefeld Nardoto
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
1476 J. A. Hogan et al.
Supporting Information
Additional Supporting Information may be found in the online ver-
sion of this article:
Figure S1. Variance Partitioning Barplots: Results from distance-
based Redundancy Analysis (db-RDA) models shown in barplot form
for the entire (a), the high land-use intensity area (b), and the low
land-use intensity area if the Luquillo Forest Dynamics Plot (LFDP).
Bar Fractions correspond to the variance portioning diagram shown in
Figs 3–5.
Appendix S1. Literate Statistical Document outlining the exploration
of beta-diversity variability within the LFDP.
Appendix S2. Literate Statistical Document showing the transforma-
tion of response variables, forward selection of predictor variables,
and db-RDA.
Appendix S3. Literate Statistical Document containing the PERMANOVA
results shown in Table 1.
© 2016 The Authors. Journal of Ecology © 2016 British Ecological Society, Journal of Ecology, 104, 1466–1477
Land-use legacies and the abiotic environment 1477