No evidence that boron influences tree species distributions in
lowland tropical forests of Panama
Benjamin L. Turner1, Paul-Camilo Zalamea1,2, Richard Condit1, Klaus Winter1, S. Joseph Wright1 and
James W. Dalling1,2
1Smithsonian Tropical Research Institute, Apartado 0843-03092, Balboa, Ancon, Panama; 2Department of Plant Biology, University of Illinois, Urbana, IL 61801, USA
Author for correspondence:
Benjamin L. Turner
Tel: +1 507 212 8171
Email: TurnerBL@si.edu
Received: 8 July 2016
Accepted: 6 October 2016
New Phytologist (2016)
doi: 10.1111/nph.14322
Key words: Barro Colorado Island, boron (B),
ecosystem budget, hot-water extraction,
Mehlich-III extraction, nutrient cycling,
species distribution, tropical forest.
Summary

 It was recently proposed that boron might be the most important nutrient structuring tree
species distributions in tropical forests. Here we combine observational and experimental
studies to test this hypothesis for lowland tropical forests of Panama.

 Plant-available boron is uniformly low in tropical forest soils of Panama and is not signifi-
cantly associated with any of the > 500 species in a regional network of forest dynamics plots.
Experimental manipulation of boron supply to seedlings of three tropical tree species revealed
no evidence of boron deficiency or toxicity at concentrations likely to occur in tropical forest
soils. Foliar boron did not correlate with soil boron along a local scale gradient of boron avail-
ability.

 Fifteen years of boron addition to a tropical forest increased plant-available boron by 70%
but did not significantly change tree productivity or boron concentrations in live leaves, wood
or leaf litter. The annual input of boron in rainfall accounts for a considerable proportion of
the boron in annual litterfall and is similar to the pool of plant-available boron in the soil, and
is therefore sufficient to preclude boron deficiency.

 We conclude that boron does not influence tree species distributions in Panama and
presumably elsewhere in the lowland tropics.
Introduction
It was proposed recently that boron (B) might be the most
important nutrient influencing tree species distributions in tropi-
cal forest on Barro Colorado Island (BCI), Panama (Steidinger,
2015). This was based on a niche-breadth analysis using data on
extractable nutrients and tree community composition from the
BCI large forest dynamics plot, in which B (along with
potassium, K) was found previously to have the strongest effect
on community structure of many measured soil nutrients (John
et al., 2007). The argument for the importance of B is based on
the widely held assumption that B deficiency and toxicity occur
over a relatively narrow range of soil B concentrations compared
with other nutrients (Gupta et al., 1985). However, there is little
evidence in support of this assumption, which has been chal-
lenged on the basis of results for a number of crop plants and cul-
tivars (Chapman et al., 1997).
Boron is an essential plant micronutrient and is the most com-
monly deficient micronutrient for crops (Shorrocks, 1997).
Boron is known to constrain tree growth in plantation forestry in
Scandinavia, New Zealand and Southeast Asia once macronutri-
ent limitation (principally nitrogen, N) has been corrected by the
addition of fertilizer (Hunter et al., 1990; H€ogberg et al., 2006;
Dell et al., 2008). There are also reports suggesting that B can be
the primary nutrient limiting growth in eucalyptus plantations in
some parts of China (Dell et al., 2008), while temporary B defi-
ciency was reported for deciduous forests in Quebec, Canada,
primarily for rapidly growing seedlings on sandy soils (Bernier &
Brazeau, 1988). By contrast, B toxicity is common only in semi-
arid environments, particularly in irrigated agriculture. For exam-
ple, B toxicity in Australia is confined to regions with low annual
rainfall (< 550 mm) in Western Australia, South Australia and
Victoria, particularly on strongly alkaline (pH > 8) soils with
sodic clay subsoils, which are poorly leached and have B concen-
trations > 12 mg B kg1 (Nable et al., 1997). Overall, there is lit-
tle evidence of B deficiency or toxicity in natural forests in the
tropics or elsewhere.
Here we examine the hypothesis that soil B influences species
distributions in lowland tropical forests. We combine manipula-
tive experiments on tropical tree seedlings with analysis of species
responses to soil B availability across a regional-scale network of
forest census plots and a local scale gradient in B availability on
BCI. We also present a B budget for a well-studied site in
Panama, present results of a long-term micronutrient addition
experiment and suggest an explanation for the absence of B
effects on tropical tree distributions in the region. We conclude
that soil B does not influence the distribution of lowland tropical
tree species in Panama.
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016) 1
www.newphytologist.com
Research
Materials and Methods
Soil-extractable B determination
We quantified plant-available soil B in soils from a network of 79
lowland tropical forest census sites in Panama (Pyke et al., 2001;
Condit et al., 2004, 2013; Engelbrecht et al., 2007; Turner &
Engelbrecht, 2011). The sites vary from inventory transects to
plots of various sizes, including 409 40 m, 1 ha, 6 ha and 50 ha
(BCI). The sites span a strong rainfall gradient, from c. 1800 to
> 4000 mm yr1 (Engelbrecht et al., 2007). Soils include a vari-
ety of taxonomic orders (Oxisols, Ultisols, Alfisols and Incepti-
sols) and vary widely in chemical properties, including pH
(3.3–7.0), organic carbon (2–10%) and readily exchangeable soil
phosphorus (P) (< 0.1 to > 20 mg P kg1). For each site, a soil
sample consisted of a composite of either five cores (inventory
transects and 409 40 m plots), 13 cores (1 ha forest dynamics
plots) or 25 cores (a 6 ha plot on the Caribbean coast and a 50 ha
plot on BCI). Cores were taken from the surface soil (0–10 cm
depth), because this zone integrates the nutrient cycle and con-
tains the majority of the extractable nutrients and fine roots.
We extracted B from soils using two procedures: hot water
extraction and Mehlich-III extraction. Hot water soluble B
(Berger & Truog, 1939, 1944; Gupta, 1967) is the most widely
used measure of plant-available soil B (Keren, 1996). We used
hot 0.01M CaCl2 instead of deionized water (Jeffrey & McCal-
lum, 1988), because this procedure yields similar concentrations
to the hot water procedure but reduces interference from sus-
pended clay in the extracts, which is important in the analysis of
the high clay soils of Panama.
Briefly, air-dried soils sieved to < 2 mm were mixed with
0.01M CaCl2 in a 1 : 2 soil to solution ratio and placed in a
shaking water bath at 85°C. After 20 min, samples were cen-
trifuged (8000 g, 10 min) and the supernatant was decanted.
Boron was determined in the extracts by inductively coupled-
plasma optical-emission spectrometry (ICP–OES) (Optima
7300DV; Perkin Elmer Inc., Shelton, CT, USA).
We also used Mehlich-III solution to determine extractable
soil B (Mehlich, 1984). This procedure is a commonly used mul-
ti-element extraction to quantify plant-available nutrients,
including B, and was used in the original analysis on BCI (John
et al., 2007). It includes water-soluble B, as well as additional B
held on soil sorption sites. Briefly, air-dried soils were shaken for
5 min in Mehlich-III solution (0.2 N acetic acid, 0.25 M
NH4NO3, 15 mM NH4F, 13 mM HNO3, and 1 mM EDTA)
in a 1 : 10 soil to solution ratio. The samples were centrifuged
(8000 g, 10 min) and the supernatant was decanted. Boron was
determined by ICP–OES as detailed in the previous paragraph.
All extractions and preparation of standards for the two proce-
dures were conducted in plastic to avoid B contamination from
borosilicate glassware.
The most sensitive wavelength for B determination by
ICP–OES at 249.772 nm suffers from a major interference by
iron (Fe) (Sah & Brown, 1997; Taber, 2004; Turner et al.,
2016). This is a particular problem for Mehlich-III extracts of
tropical forest soils, which contain considerable Fe concentra-
tions, but does not affect 0.01M CaCl2 extracts, which con-
tain insufficient Fe to interfere in B detection (Turner et al.,
2016). Here we used the most sensitive line (249.772 nm) to
determine B in 0.01M CaCl2 extracts and the line without
Fe interference (208.597 nm) to determine B in Mehlich-III
extracts and total element digests.
Regional and local scale species distributions in relation to
soil B
We incorporated plant-available soil B measurements as
described in the previous section into a hierarchical Bayesian
model to quantify regional-scale species distributions in response
to soil B. The model was described in detail previously (Condit
et al., 2013) and included data on the distributions of > 500 tree
species that occur in a network of 72 forest census sites in
Panama, dry season soil moisture deficit, soil exchangeable cal-
cium (Ca) and K, readily exchangeable P determined by extrac-
tion with anion-exchange resins (resin P), the micronutrients Fe
and zinc (Zn), extractable inorganic N, and the potential toxin
exchangeable aluminum (Al). We reran the original model except
that we replaced inorganic N (the nutrient with the weakest influ-
ence on species distributions) with either hot-CaCl2 extractable B
or Mehlich-III extractable B. Significant associations between tree
species distributions and soil nutrients are indicated by effect sizes
– the first order parameter of the logistic model describing the
distribution of a species in relation to a given nutrient. Positive
effect sizes indicate that the species occurs predominantly at high
concentrations of the nutrient, while negative effect sizes indicate
that the species occurs predominantly at low concentrations of
the nutrient.
At the local scale, Steidinger (2015) proposed that species dis-
tributions respond to soil B across a B availability gradient on
BCI. To test this, we measured B in leaves of tree species that
occur across the B gradient on BCI, and compared these concen-
trations with estimated B availability from kriged Mehlich-III
extractable B concentrations (updated values corrected for Fe
interference reported in Turner et al., 2016). We sampled shade
leaves from 10 individuals of each of six species: Chrysophyllum
argenteum Jacq., Guarea guidonia (L.) Sleumer, Hirtella triandra
Sw., Protium panamense (Rose) I.M. Johnst., Tetragastris
panamensis (Engl.) Kuntze and Trichilia tuberculata (Triana &
Planch.) C. DC. Trees were selected for sampling based on K-
means clustering (Hartigan & Wong, 1979) to maximize varia-
tion in kriged estimates of soil base cation concentrations deter-
mined from Mehlich-III extractions (John et al., 2007). This
yielded a parallel variation in extractable B, given that B was
strongly correlated with base cations on the 50 ha plot (Pearson
r > 0.7; see Fig. 1 and table S4 in John et al., 2007). We used the
Poisson cluster method test to determine the association between
tree species distribution and soil B (John et al., 2007). Foliar B
was determined by nitric acid digestion as described below, and
compared with Mehlich-III extractable B concentrations using
linear regression.
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist2
Growth responses of tropical tree seedlings to B
We grew seedlings of three tropical tree species underneath a glass
rain shelter in a mixture of quartz sand and soil. The soil was
obtained from a site under tropical forest on the Santa Rita
Ridge, one of the most infertile sites in central Panama. We
selected species that differed in their distributional affinity for B
based on the results of the Poisson cluster model implemented by
John et al. (2007). Initially, we grew Hura crepitans L. in full
nutrient solution (Johnson et al., 1957) with B concentrations of
0, 1, 4, 10, and 40 times the standard concentration (i.e. 0, 12.5,
50, 125 and 500 lM B) with nutrient solution added once per
week. As the plants showed no sign of B toxicity (see Results sec-
tion) we grew two additional species, Ochroma pyramidale (Cav.
ex Lam.) Urb. and Pentagonia macrophylla Benth., with up to
100 and 150 times the standard B concentration (i.e. 1250 and
1875 lM B), with nutrient solution added twice per week. Small
amounts of water loss by evaporation were replaced every 2 d
using tap water that contained a trace concentration of B equiva-
lent to that in rainfall (< 1 lM B). For the first experiment
(H. crepitans) we used a 50 : 50 sand and soil mixture, while in
the second experiment (O. pyramidale and P. macrophylla) we
used a 95 : 5 sand and soil mixture to ensure an extremely low B
concentration in the zero-B treatment. Thus, all plants received B
from the substrate and water additions.
Seedlings were grown for different periods depending on
growth rate until they attained similar biomass, with growth
period ranging from 27 d for O. pyramidale to 70 d for
P. macrophylla. The seedlings of each species were harvested for
all treatments when the ratio of plant biomass to pot volume was
c. 1 g : 1 l in the 12.5 lM B treatment. Leaves, stems and roots
were separated, dried for 3 d at 60°C, weighed and ground. All
species form arbuscular mycorrhizas and acquire their symbiotic
fungi rapidly after germination. We have verified this in previous
experiments, although we did not measure mycorrhizal coloniza-
tion in this experiment.
Relative growth rate (RGR; mg g1 d1) was calculated
according to the following equation: RGR = ((LogeWf –LogeWi)/
(Dt)), where Wf is the final dry mass, Wi is the initial dry mass
and Dt is the duration of the experiment. Leaf mass fraction
(LMF; leaf mass per unit whole plant mass; g g1), net assimila-
tion rate (NAR; biomass increment per unit leaf area;
g m2 d1), leaf area ratio (LAR; leaf area per unit whole plant
mass; cm2 g1), specific leaf area (SLA; leaf area per unit leaf
mass; cm2 g1) and root mass fraction (RMF, root mass per unit
whole plant mass; g g1) were calculated from harvest data. NAR
was calculated according to the following equation:
NAR = ((WfWi)9 (LogeAf – LogeAi))/((Af Ai)9 (Dt)), where
Wf and Wi are the final and initial dry mass (g), respectively, Af
and Ai are the final and initial leaf area (m
2), respectively, and Dt
is the duration of the experiment (d).
Foliar B was determined by digestion in concentrated nitric
acid under pressure in polytetrafluoroethylene vessels, with detec-
tion by ICP–OES spectrometry at 208.957 nm (although there
was negligible Fe interference at the low concentrations present
in the leaf tissue). Foliar B recoveries from the certified reference
materials NIST 1515 (apple leaves) and NIST 1547 (peach
leaves) were > 95% of the certified values.
Boron addition experiment in a lowland tropical forest
We quantified responses in trees and soil to 15 yr of experimental
B addition at a site in lowland tropical forest on Gigante Penin-
sula, close to BCI. The long-term nutrient addition experiment is
described in detail elsewhere (Yavitt et al., 2009; Wright et al.,
2011; Turner et al., 2015). A micronutrient treatment that
includes B is applied to four 409 40 m (1600 m2) plots, and
four similar plots serve as untreated controls. Since 1998, the four
micronutrient plots have received annual additions of dolomitic
limestone (230 kg ha1) and a soluble trace element mix
(25 kg ha1; Scott’s Miracle Grow Company, Marysville, OH,
USA) that contained 1.35% B, equivalent to 0.3375 kg B ha1 or
33.75 mg Bm2. This is a similar amount to the annual B return
in litter fall at this site (see ‘Boron budget’). To compare the con-
trol and micronutrient plots, we used a one-way ANOVA to
compare differences in plant-available B determined by hot-
CaCl2 extraction for soils from 0 to 5 cm depth sampled in
November 2012 (log-transformed values), and a one-way
ANOVA to compare treatment effects on B in fine roots. We
used a factorial two-way ANOVA with treatment (control vs
micronutrient) and species as main effects to evaluate differences
in B concentrations in wood and live leaves for three species
(Alseis blackiana, Heisteria concinna and T. panamensis) sampled
in 2013. We used a main effects ANOVA (no interaction) with
Fig. 1 Concentrations of soil boron (B) extracted by hot 0.01M CaCl2 and
Mehlich-III solution in soils from 79 sites under lowland tropical forests in
Panama. Each soil was a composite of multiple replicate cores of the
surface soil (0–10 cm depth) at each site. The solid line is the 1 : 1 line. The
dashed line is the regression between hot-CaCl2 extractable B and
Mehlich-III extractable B, defined by the equation [Mehlich
B] = 0.2699 [CaCl2-B] + 0.625; R
2 = 0.008, P = 0.44.
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 3
treatment and season as main effects to evaluate B concentrations
in litter fall during the 2012 wet season and the 2013 dry season.
Boron budget
We quantified B stocks in trees and soil using data from the four
control plots in the nutrient addition experiment on Gigante
Peninsula. We measured B in wood and leaves of the three
species that occur in all plots in the experiment (A. blackiana,
H. concinna, T. panamensis). Samples were digested in concen-
trated nitric acid under pressure with detection by ICP spectrom-
etry as described under ‘Growth responses of tropical tree
seedlings to B’. We used the average concentrations in wood and
leaves of the three species to estimate total B stocks in the forest,
using above-ground biomass estimates derived from allometric
equations involving tree diameter and height (Chave et al.,
2015). This assumes that the three-species means are
representative of the whole-plot means involving all species. We
calculated total soil B in samples taken to 1 m depth in the soil
profile (four cores per plot) by multiplying total B determined by
nitric acid digestion and soil bulk density. In the same manner,
we calculated stocks of plant-available B determined by hot-
CaCl2 extraction in the top 1 m of the profile and the surface
20 cm of the profile, given that the majority of the fine roots
occur in the latter horizon (Yavitt & Wright, 2001). Finally, we
estimated B inputs in rainfall by multiplying the annual rainfall
on BCI (2600 mm; Windsor, 1990) by the global mean B con-
centration in marine rain of 10 2.3 lg l1 (Park & Schlesinger,
2002). The value for marine rain is higher than for continental
rain (mean 4.89 lg l1), but is most likely to resemble the con-
centration in rainfall on BCI given its proximity to the Caribbean
Sea (BCI is c. 25 km south of the coastline). The mean value for
marine rain is greater than the median concentration (6.6 lg l1).
However, the mean concentration is similar to measurements of
rainfall B on the northern Gulf Coast of Florida, USA, which
were 8.3 lg l1 in summer and 11.9 lg l1 in winter (Martens &
Harriss, 1976). Our estimate of B inputs to the forest does not
include dry deposition, for which values are poorly constrained
(Park & Schlesinger, 2002) but potentially important (Cividini
et al., 2010). We did not measure B leaching, but assume that it
is similar to B inputs in rainfall, assuming that the B cycle is in
equilibrium in the forest.
Results
Soil B concentrations in lowland tropical forests of Panama
Across a large number of soils in central Panama under lowland
tropical forests but spanning a range of soil properties, plant-
available B concentrations determined by extraction in hot CaCl2
ranged between 0.04 and 0.53 mg B kg1 (mean
0.22 mg B kg1). These values are in the range considered to rep-
resent B deficiency in agriculture and none were at levels at which
B toxicity is typically expected to occur (> 5 mg B kg1) (Nable
et al., 1997). Mehlich-III extractable B concentrations were
greater than hot CaCl2 extractable B, ranging between 0.21 and
1.64 mg B kg1 (mean 0.68 mg B kg1). The two measures of
extractable B were not significantly correlated (P = 0.44; Fig. 1).
Fig. 2 Relationships between soil properties and plant-available soil boron (B) determined by extraction in hot 0.01M CaCl2 (a–d) and Mehlich-III solution
(e–h). All measures except pH were log-transformed before analysis. P, phosphorus; C, carbon.
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist4
Boron extracted in hot water was correlated negatively with
soil pH in water and CaCl2 (e.g. r =0.39 for pH in CaCl2,
P < 0.001; Fig. 2a), resin P (r =0.36, P = 0.001; Fig. 2b) and
extractable base cations (r =0.42, P < 0.001; Fig. 2c), and cor-
related positively with soil organic carbon (r = 0.36, P = 0.001;
Fig. 2d). By contrast, Mehlich-III extractable B was correlated
positively with pH (e.g. r = 0.61 for pH in CaCl2, P < 0.001;
Fig. 2e), resin P (r = 0.48, P < 0.001; Fig. 2f), base cations
(r = 0.58, P < 0.001; Fig. 2g) and organic carbon (r = 0.25,
P = 0.028; Fig. 2h). All measures except pH were log-
transformed before analysis. Mehlich-III extractable Fe and K
were not correlated significantly with hot water B, while
Mehlich-III extractable Al and Fe were not correlated signifi-
cantly with Mehlich-III extractable B (Supporting Information,
Table S1).
Although the two measures of plant-available B were not
significantly correlated, forward stepwise regression predicted
Mehlich-III extractable B from hot-CaCl2 extractable B and
soil pH (measured in 0.01 M CaCl2) according to the
following model: [Mehlich B] = 1.288 0.2659 [CaCl2 B] +
0.272 0.0309 [pH] 0.912 0.177 (F2,76 = 42.7, R2 = 0.53,
P < 0.001). Inclusion of organic carbon, Mehlich-III extractable
Al or Fe, total exchangeable bases, or resin P did not improve the
model (P > 0.2).
For BCI, Mehlich-III extractable B concentrations used by Stei-
dinger (2015) were between 0.23 and 2.86mg B kg1 (John et al.,
2007). These measurements were made by ICP–OES using a B
line that suffers from interference by Fe (Taber, 2004). Correcting
this using an empirical equation derived from 230 soils under low-
land tropical forest in Panama (i.e. comparing the difference in
Mehlich-III extractable B measured at two lines with and without
Fe interference, against Mehlich-III extractable Fe) allowed us to
correct the reported B values for BCI (Turner et al., 2016). These
corrected Mehlich-III extractable B concentrations for the BCI
50 ha plot ranged between 0.08 and 2.09mg B kg1 (mean
0.58 0.009mg B kg1). The kriged values using the original and
revised data were well correlated (Turner et al., 2016), indicating
that the previous statistical analyses do not require re-evaluation
(John et al., 2007; Steidinger, 2015). As in the cross-Isthmus plots,
Mehlich-III extractable B on the BCI 50 ha plot was correlated
B
Fig. 3 Histograms of individual species
responses to eight environmental factors,
including plant-available boron (B)
determined by extraction in hot 0.01M
CaCl2. The horizontal axis is the effect size,
which is defined as the first-order parameter
of the logistic model and approximates the
change in occurrence probability of a species
across the range of each factor, relative to its
mean occurrence. The shaded portions of
each curve highlight species with strong
responses (effect sizes greater than 0.5),
which indicate species that occur
predominantly at one end of the gradient.
These first-order effects do not include
modal responses. The curves are the fitted
hyper-distributions of effect sizes with fitted
standard deviation, while the points are the
observed number of species within bins of
0.25, including only species with at least 10
occurrences. Moisture, dry-season moisture;
Al, aluminum; P, resin phosphorus; Fe, iron;
Ca, calcium; Zn, zinc; K, potassium.
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 5
positively with pH, base cations (Ca, K, magnesium (Mg)) and
micronutrients (manganese, Zn) (for both the original and the cor-
rected data).
Regional and local scale species distributions in relation to
soil B
At the regional scale, the inclusion of B had a negligible impact
on modeled tree species distributions. Hot-CaCl2 extractable B,
widely used as a measure of soil B availability to plants, was not
significantly associated with the distribution of any species
(Fig. 3). Similarly, Mehlich-B made a near-negligible contribu-
tion to the model: there were fewer species with significant
responses to Mehlich-B (eight of 271 species with > 10 individu-
als in plots) than to all nutrients apart from N (Fig. S1). Of the
various soil nutrients included in the model, the majority of the
species responded significantly to resin P, as in the original model
(Condit et al., 2013). Overall, species associated with low P
tended to show weak negative responses to Mehlich-B, while
species associated with high P tended to show weak positive
responses to Mehlich-B. It therefore appears that the few signifi-
cant Mehlich-B responses are simply byproducts of responses to
resin P. Given that none of the species were associated with hot-
CaCl2 extractable B, and that this B pool does not correlate
strongly with resin P, we conclude that B does not influence
regional species distributions.
At the local scale, the distributions of three of the six species
that we studied on the BCI 50 ha plot were significantly associ-
ated with high concentrations of soil B, while three species
showed no significant association (Table S2). However, none of
the six species showed a significant positive correlation between
foliar B and Mehlich-III extractable soil B across the plot
(Table S2), while one species, G. guidonia, showed a significant
negative correlation. The range in Mehlich-III extractable B var-
ied between 11- and 41-fold for the six species. Across these
ranges of soil B, mean foliar B ranged from
0.06 0.01 mg B g1 for C. argentea to 0.23 0.11 mg B g1
for H. triandra (Table S2).
Growth responses of tropical tree seedlings to B availability
We found little evidence that B influences the growth rates of
seedlings of three tropical trees species at anything other than
extremely high B concentrations. There was no evidence of B
deficiency at extremely low B availability, including in the 0 lM
B treatment (i.e. with no B added to the nutrient solution).
Indeed, there were no significant differences in any measure of
plant performance between the 0 and 12.5 lM B treatments
(Table 1), suggesting that even the small concentrations of B pre-
sent in the sand–soil mixture (with Mehlich-III extractable soil B
below the detection limit; 0.01 mg B kg1), the nutrient solution
(as contaminants in the other nutrient preparations) and the
small volume of tap water used to replace moisture loss by evapo-
ration provided sufficient B for plant growth. By contrast, there
was clear evidence of B toxicity at extremely high B
(≥ 1250 lMB), as indicated by visual leaf damage and reduced
plant size (Fig. 4), a marked increase in foliar B and a decline in
RGR (Table 1).
Table 1 Responses of seedlings of three tropical tree species to variation in boron (B) availability
Boron
treatment
Foliar
B (lg B g1)
Relative growth
rate (RGR,
mg g1 d1)
Leaf mass
fraction
(LMF, g g1)
Net assimilation
rate (NAR,
g m2 d1)
Lear area
ratio (LAR,
cm2 g1)
Specific leaf
area (SLA,
cm2 g1)
Root mass
fraction
(RMF g g1)
Hura crepitans
09 24.3 3.2a 55.3 3.7a 0.27 0.03a 6.3 0.5a 63.7 2.6a 234 25a 0.22 0.02a
19 17.8 6.3b 54.5 2.6a 0.27 0.03a 6.3 0.4a 62.5 6.1a 234 36a 0.23 0.04a
49 39.8 8.7c 54.2 2.9a 0.25 0.01a 6.3 0.4a 62.5 3.5a 251 15a 0.25 0.03a
109 35.8 7.7c 51.9 6.8a 0.28 0.03a 6.0 1.0a 62.0 4.7a 222 24a 0.23 0.03a
409 102.9 20.5d 52.8 4.2a 0.22 0.02b 6.1 0.5a 61.3 5.0a 276 27b 0.24 0.04a
Ochroma pyramidale
09 20.9 4.4a 153.3 8.8a 0.54 0.03a 7.4 0.7a 217.4 13.6a 403 35a 0.33 0.03a
19 26.3 2.9b 159.0 8.6a 0.57 0.03a 8.1 0.9a 201.9 24.3a 356 52ab 0.29 0.04a
109 72.3 11.9c 159.4 6.3a 0.57 0.02a 8.4 0.5a 192.5 9.6ab 337 24b 0.30 0.03a
409 288.0 43.2d 146.6 6.2a 0.60 0.01ab 7.0 0.5ab 218.4 8.6a 366 20ab 0.25 0.02a
1009 487.0 82.3e 122.1 11.4b 0.63 0.03b 6.1 0.7b 206.8 11.6a 326 18b 0.19 0.04b
1509 530.1 168.1f 87.4 15.3c 0.63 0.09b 4.8 1.0c 187.0 31.3b 295 31c 0.18 0.12b
Pentagonia macrophylla
09 11.3 10.4a 66.9 3.9a 0.58 0.05a 5.2 0.4a 127.0 14.2a 219 22a 0.31 0.05a
19 28.5 5.5b 66.5 2.6a 0.53 0.10ab 5.4 1.0a 124.7 28.5a 235 32a 0.38 0.11a
109 221.8 31.8c 64.0 4.6a 0.63 0.04a 4.8 0.7a 137.1 23.4a 218 27a 0.28 0.04ab
409 1511.4 955.8d 62.7 5.6a 0.66 0.02a 4.9 0.8a 129.2 21.9a 196 30a 0.22 0.02b
1009 1674.8 482.7d 43.2 11.8b 0.48 0.10b 5.1 1.4a 71.8 12.4b 152 26b 0.29 0.09ab
Values are mean SD of at least six replicate plants per treatment. Boron treatments refer to the B concentration (12.5 lM) in the standard nutrient
solution (Johnson et al., 1957). Differences among treatments were tested using generalized linear mixed-effects models for each species and response
variable separately. All dependent variables were logarithmically transformed before analysis and results were considered statistically significant at P < 0.05.
Differences among treatments for each species are denoted by characters after the mean values. We coded each seedling as random and B treatment as a
fixed effect, and then used a Student’s t test to determine differences among treatments.
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist6
Foliar B in the 12.5 lM B treatment was similar for the three
species studied (17.8–28.5 lg B g1) , suggesting an approximate
optimum foliar B concentration for tropical tree seedlings. Com-
pared with these values, foliar B in the 0 lM B treatment was
lower than the 12.5 lM B treatment for O. pyramidale and
P. macrophylla, but higher for H. crepitans. However, foliar B
increased dramatically as solution B increased further, particu-
larly beyond the 125 lMB treatment, reaching 530 lg B g1 in
O. pyramidale and 1675 lg B g1 in P. macrophylla.
For H. crepitans, the maximum B treatment (50 lMB) yielded
the highest foliar B (103 lg B g1) and affected LMF and SLA,
but did not significantly change RGR, NAR, LAR or RMF
(Table 1). For O. pyramidale and P. macrophylla there were clear
signs of leaf chlorosis at ≥ 50 lM B (Fig. 4). In addition to the
marked increases in foliar B, there were corresponding reductions
in RGR and SLA at ≥ 1250 lM B for O. pyramidale and
P. macrophylla (Table 1). There were also significant changes in
LAR, LMF and RMF at high B for both species, but NAR
declined in O. pyramidale only at the highest B treatment
(1875 lM B).
Soil and tree response to long-term B addition
Fifteen years of B addition as part of a micronutrient treatment
increased plant-available B in soil significantly (F1,6 = 21.5,
P = 0.0035; Fig. S2). In the upper 5 cm of soil, hot-CaCl2
extractable B increased from 0.113 0.021 mg B kg1 in control
plots to 0.192 0.024 mg B kg1 in the micronutrient treatment
– a significant 70% increase. The greater plant-available B
increased fine-root B significantly (control 0.017 0.04
mg B g1, micronutrient 0.024 0.004 mg B g1; F1,6 = 6.4,
P = 0.045). However, there were no significant changes in B con-
centrations in wood, leaves or litter fall, although concentrations
in the micronutrient treatment were slightly greater in each case.
For wood B, neither the micronutrient main effect (control
0.036 0.013 mg B g1, micronutrient 0.040 0.010 mg
B g1; F1,15 = 0.267, P = 0.613), the species main effect
(F2,15 = 2.58, P = 0.109) nor the treatment9 species interaction
(F2,15 = 0.423, P = 0.633) was significant. For live leaves, neither
the micronutrient main effect (control 0.060 0.015 mg B g1,
micronutrient 0.064 0.015 mg B g1; F1,16 = 0.588,
P = 0.454) nor the treatment9 species interaction (F2,16 = 1.03,
P = 0.378) was significant, but the species main effect was
strongly significant (F2,16 = 7.57, P = 0.0049). For litter fall, nei-
ther the micronutrient main effect (control
0.064 0.013 mg B g1, micronutrient 0.068 0.021 mg
B g1; F1,11 = 0.176, P = 0.683) nor the seasonal effect (F1,11 =
0.396, P = 0.542) was significant. There were no significant dif-
ferences in the mass of litter fall or trunk growth rates in the
micronutrient treatment (P > 0.1).
Boron budget
A simple B budget for Gigante Peninsula illustrates the likely
significance of rainfall in supplying sufficient B for plant
growth (Table 2). Assuming marine-influenced rainfall con-
tains 10 lg B l1, the global marine rainfall concentration
(Park & Schlesinger, 2002), then the annual rainfall on
Gigante Peninsula (a long-term average of 2600 mm on
nearby BCI) provides c. 26 mg Bm2 yr1. In comparison,
the profile-weighted hot-CaCl2 extractable B concentration in
soil on Gigante is 0.062 mg B kg1, equivalent to
70.0 mg Bm2 in the top 1 m of soil. However, the majority
of the fine roots in this forest occur in the upper 20 cm of
the soil profile (Yavitt & Wright, 2001), where the plant-
available B stock is 18.2 mg B m2. The input of B in rainfall
is therefore 37% of the plant-available B pool to 1 m depth
in the soil, but c. 50% greater than the plant-available B in
the main rooting zone. Using the median B concentration in
rainfall (6.6 lg B l1), the B input is of similar magnitude to
Fig. 4 Plant responses to increasing boron (B) concentration for two species,Ochroma pyramidale (upper panel) and Pentagonia macrophylla (lower
panel) receiving an otherwise full nutrient solution (Johnson et al., 1957). The standard solution B concentration was 12.5 lM. Boron concentrations in the
nutrient solution increased from the left (no added B) to right (150 times the standard nutrient solution B forO. pyramidale or 100 times for
P. macrophylla). The scale bar in the lower left is 10 cm long.
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 7
the soil soluble B pool in the rooting zone. Total soil B, by
contrast, is 21 225 mg Bm2 to 1 m depth in the profile, or
300 times greater than the soluble pool.
Foliar B varies markedly among the three species studied on
Gigante, with concentrations from 0.051 mg B g1 in
H. concinna (Erythropalaceae) to 0.070 mg B g1 in the medium-
sized tree A. blackiana (Rubiaceae) (Dalling et al., 2001), yielding
an average concentration of 0.060 mg B g1 for the three tree
species. Multiplying this by leaf mass calculated from an allomet-
ric equation based on trunk diameter (Chave et al., 2008) yielded
81.0 mg Bm2 in leaves – about three times the amount supplied
annually in rainfall and of similar magnitude to the soluble B
pool in the soil. By contrast, wood of the three species contained
on average 0.037 mg B g1, yielding 604.4 mg B m2 – more
than seven times the B in leaves. Assuming that coarse roots (cal-
culated as 20% of above-ground biomass, minus fine roots) con-
tain a similar B concentration to wood, they contribute
103.5 mg Bm2, while fine roots contain an additional
8.0 mg Bm2. Wood therefore contains 76% of the total plant B
(i.e. the sum of B in wood, roots and leaves; 796.9 mg B m2).
Total plant B is 27 times smaller than total soil B.
Finally, total litter fall in plots that did not receive P (added P
caused an increase in litter fall) across the 2006–2007 annual
cycle was 1100 g m2, of which 630 g m2 was leaf litter (Turner
et al., 2015). Leaf litter B varied seasonally, presumably reflecting
reduced transpiration and B uptake during the dry season
(Table 2). Two-thirds of the annual leaf litter fall occurs in the
dry season, yielding a B turnover in annual leaf litter fall of
38.2 mg Bm2. The amount of B in litter fall is therefore c. 1.5
times the amount in rainfall, 55% of the amount in the soil sol-
uble pool to 1 m depth in the profile, and twice the amount in
the soluble pool in the main rooting zone to 0.2 m depth.
Discussion
We found no evidence that B influences the distribution or per-
formance of tree species in lowland tropical rain forests of
Panama. This included evidence from measurements of plant-
available B in soil and its influence on regional species distribu-
tions, growth experiments with tropical tree seedlings in which B
treatments exceeded the range of concentrations found in low-
land tropical forests, foliar B along a local scale soil B gradient on
Table 2 Stocks of boron (B) in rainfall, plant biomass and soil in lowland temperate forest on Gigante Peninsula, part of the Barro Colorado Nature
Monument, Panama
Pool Boron concentration Total mass Boron stocka (mg Bm2)
Rainfall Bb 10 lg B l1 2600mm yr1 26.0 (17.1)
Total soil Bc 18.8 9.2mg B kg1 (profile weighted) 1129 kgm2 21 225
Soil soluble Bd 0.062 0.037mg B kg1 (to 1 m)
0.113 0.025mg B kg1 (to 0.2 m)
1129 kgm2 (to 1m)
163 kgm2 (to 0.2m)
70.0 (to 1 m)
18.4 (to 0.2m)
Foliar Be Alseis blackiana = 0.070 0.011mg B g1
Heisteria concinna = 0.051 0.021mg B g1
Tetragastris panamensis = 0.059 0.008mg B g1
Mean: 0.060mg B g1
1350 gm2 81.0
Wood Bf Alseis blackiana = 0.044 0.016mg B g1
Heisteria concinna = 0.034 0.009mg B g1
Tetragastris panamensis = 0.033 0.016mg B g1
Mean: 0.037mg B g1
16 335 gm2 604.4
Fine root Bg 0.017 0.004mg B g1 470 111 gm2 8.0
Coarse root Bh 0.037mg B g1 2797 gm2 103.5
Leaf litter falli 0.059 0.013mg B g1 (dry season)
0.067 0.014mg B g1 (wet season
630 gm2 yr1 38.2
Plant B (total) 796.9
All masses and B stocks are calculated on an area basis (m2).
aStocks are calculated in mg Bm2 of land surface (multiply by 10 to convert to g B ha1) by multiplying total mass/volume of a pool with its B
concentration.
bThe average marine rainfall B concentration (Park & Schlesinger, 2002). The B input assuming the median concentration in marine rainfall (6.6 lg B l1) is
given in parentheses.
cDetermined by nitric acid digestion and calculated for the profile to 1 m in four control plots.
dDetermined by hot-CaCl2 extraction and calculated for the profile to 1 m in four control plots.
eLeaf mass calculated from tree diameters based on an allometric equation (Chave et al., 2008) using individuals ≥ 5 cm diameter at breast height (dbh) (the
threshold used to develop the allometric relationship) for all species in the plots. Values are species means SD for shade leaves of four individual trees in
the control plots.
fMean values of wood cores to 10 cm into the trunk for trees in four control plots (n = 4 individuals for each species).
gFine root mass is the mean SD of roots < 2mm diameter to 1 m depth in the soil profile in the four control plots in the fertilization experiment. Fine root
B was determined in roots from the 0–5 cm horizon in the four control plots.
hCoarse root mass was calculated as 20% of the above-ground biomass (Chave et al., 2015) and corrected for fine root mass. The B concentration was
assumed to be the same as wood.
iValues are from litter sampled during the dry and wet seasons from traps in each of the 32 plots in the N–P–K factorial nutrient addition experiment. Total
litter fall on Gigante is c. 1100 gm2, with c. 60% falling as leaves (Wright et al., 2011; Turner et al., 2015). The B stock is calculated on the basis of two-
thirds of the total leaf litter falling in the dry season.
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist8
BCI, and stand-level forest experiments that raised plant-available
B by 70% with no effect on tissue B, litter production or tree
growth.
Soil B concentrations in tropical forests
Measurements of plant-available B by two commonly used proce-
dures in tropical forest soils spanning a range of taxonomic
orders, pH and fertility status indicate uniformly low B through-
out the region. Soluble soil B concentrations determined by
extraction in hot 0.01 M CaCl2, the most widely used measure of
plant-available B (Keren, 1996), were < 0.6 mg B kg1 and there-
fore typical for strongly weathered soils (Shorrocks, 1997).
Although diagnosis of B status from soil tests is difficult (Nable
et al., 1997), these values are in the range considered to indicate
potential B deficiency in agriculture (Gupta et al., 1985; Shor-
rocks, 1997) and are far below levels at which B toxicity occurs in
most crop plants (> 5–10 mg B kg1) (Nable et al., 1997). None
of the soils contained soluble B concentrations in the range con-
sidered adequate for crop production (i.e. 1–4 mg B kg1). How-
ever, soils with high clay concentrations, which characterize the
majority of the soils in this study (Turner & Engelbrecht, 2011),
can sorb B and maintain it in plant-available forms despite acidic
conditions (Shorrocks, 1997). This might contribute to the
maintenance of B availability in Panama despite low concentra-
tions of plant-available B.
There is little comparable data on soluble B in tropical forests
other than BCI, although Mehlich-III extractable B was unde-
tectable in a lowland tropical forest at Yasuni National Park in
Ecuador and a montane forest plot at La Planada, Colombia
(John et al., 2007). On BCI, Mehlich-III extractable B concentra-
tions (after correction for iron interference) were up to
2 mg B kg1 (Turner et al., 2016), but plant-available B deter-
mined by Mehlich-III extraction is greater than by extraction in
hot water (Fig. 1). Interpretation of potential toxicity based on
Mehlich-III extractable B should therefore be adjusted accord-
ingly, because B sorbed to soil surfaces, some of which is included
in the Mehlich-III extraction, does not contribute to B toxicity
(Goldberg, 1997). Our measurements of plant-available B in a
wide range of soils suggest that B deficiency is possible, but it
seems unlikely that B toxicity occurs in lowland tropical forests
of Panama.
Regional and local scale species distributions in relation to
soil B
At the regional scale, we found no evidence that tree species dis-
tributions are influenced by soil B, including both hot 0.01 M
CaCl2 and Mehlich-III extractable B. For hot-CaCl2 extractable
B, the most widely used measure of soil B availability, none of
the species were associated significantly with soil B. A small num-
ber of species were related to Mehlich-B, but these relationships
were weak and presumably an artifact of the positive correlation
between Mehlich-B and resin P, because the latter explains the
distribution of c. 60% of the tree species in the region (Condit
et al., 2013). This further suggests that the apparent influence of
B on species distributions at the local scale on BCI (Steidinger,
2015) is an artifact of the strong correlations between B and other
nutrients there, particularly base cations (John et al., 2007). Two
of the species in our growth experiment, H. crepitans and
P. macrophylla, have significant positive associations with
Mehlich-B on BCI, while O. pyramidale has a significant negative
association with B (J. W. Dalling & R. John, unpublished
results). However, based on their responses to experimental B
treatments, it seems unlikely that the range of plant-available B
concentrations is sufficiently wide to influence the distribution of
these species on the plot.
At the local scale, we tested the hypothesis that soil B drives
species distributions by determining foliar B in six species that
span the gradient in Mehlich-III extractable B on BCI. We
assumed that an effect of soil B availability on tree performance
would be reflected in correlations between soil B and foliar B
(Nable et al., 1997; Shorrocks, 1997). Although foliar B was rela-
tively high for some species, the absence of significant positive
correlations between foliar B and soil B for any species, despite
up to 41-fold variation in Mehlich-III extractable B, provides
additional evidence that B does not drive local scale species distri-
butions in this tropical forest.
It is noticeable that the results of the regional scale
Bayesian approach differ markedly from the results of the
local scale niche-breadth analysis on the 50 ha plot. The
absence of significant species associations with B in the
regional scale plot network probably reflects the much larger
study area reducing the importance of dispersal limitation,
and the much larger range of soil chemical properties reduc-
ing the importance of covariance between B and base cations.
This means that although the niche-breadth analysis is a valid
approach for hypothesis generation, causality must be inferred
with caution. For this reason, John et al. (2007) gave relatively
little emphasis to the large niche breadth for B, relative to
other cations, on the 50 ha plot.
Growth responses of tropical tree seedlings to B availability
Strongly weathered soils typically contain low concentrations of
plant-available B, yet we found no evidence of B deficiency from
growth parameters for three species of tropical trees, despite
including a treatment with a solution that did not contain B.
Foliar B in the lowest B treatment was in the range considered
suitable for crop plants (20–100 lg B g1) (Shorrocks, 1997;
Marschner, 2006) and did not drop below the c. 10 lg B g1
threshold below which B deficiency typically occurs (Shorrocks,
1997; Dell et al., 2008). This indicates that tropical tree species
in general can tolerate relatively low B availability.
It is possible that B deficiency might manifest in ways other
than vegetative growth parameters determined here, by affecting
any of the proposed roles for B in plant metabolism, including
sugar transport and carbohydrate metabolism, cell wall synthesis
and structure, lignification, RNA metabolism, respiration and
membrane integrity (Marschner, 2006). Boron deficiency can
also inhibit root elongation and influence seed viability, fruit pro-
duction and P uptake. Although we could not assess seed or fruit
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 9
production, the lowest B treatment did not significantly affect
foliar P (data not shown) or change the root to shoot ratio. Fur-
thermore, B appears to be involved in the development of N-
fixing nodules in at least some legumes (Bolanos et al., 1994),
although we did not include any of the many species of tropical
Fabaceae in our study.
Boron toxicity was induced only at extremely high B concen-
trations, as indicated by chlorosis at leaf margins (Nable et al.,
1997; Marschner, 2006) that coincided with reductions in
growth rates. Foliar B concentrations in all three species in high
B treatments exceeded those considered to represent potentially
toxic concentrations in crop plants (> 100 lg B g1) (Nable
et al., 1997). Symptoms of toxicity occurred at ≥ 500 lM B
which, assuming 50% soil moisture, is equivalent to a hot-water
extractable B concentration of 2.7 mg B kg1. This is > 10 times
the mean water extractable B in the soils studied here and more
than five times the maximum observed concentration (and there-
fore many more times less than the equivalent Mehlich-III
extractable B; Fig. 1).
Although we studied only three of the many hundreds of
species present in tropical forests in Panama, the absence of
seedling responses to the lowest B treatment and a reduction in
growth rates coinciding with the appearance of toxicity symp-
toms at ≥ 500 lM B suggest that tropical trees are unlikely to suf-
fer either deficiency or toxicity of B in nature. These results are
similar to those reported for crop plants, in which optimum
yields were maintained across a wide range of solution B concen-
trations from < 1 to > 100 lM B (Chapman et al., 1997). Cou-
pled with the low soluble B concentrations in soils, it seems
unlikely that B deficiency and toxicity span a narrow range of B
concentrations for tropical trees, nor that plants can accumulate
B to toxic levels in lowland tropical forests.
Forest responses to experimental B addition
Fifteen years of B addition to lowland tropical forest on Gigante
Peninsula increased plant-available B in soil and fine-root B, but
did not significantly change B concentrations in leaves, wood or
litter fall. Similarly, there were no significant changes in stem
growth rates or litter fall, or in the soil microbial biomass (Turner
& Wright, 2014). However, a previous study reported that litter
decomposition was greater in the micronutrient treatment
(Kaspari et al., 2008). As the micronutrient treatment includes
the base cations Ca and Mg, as well as a number of essential plant
micronutrients, the role of B in the litter decomposition response
is unknown. Although the micronutrient treatment contained a
number of different nutrients in addition to B, we interpret the
absence of significant changes in the trees in this treatment,
despite a 70% increase in plant-available soil B, as additional evi-
dence that B availability does not influence vegetation in the
region.
Boron budget for a lowland tropical forest
The global B cycle is driven primarily by a large atmospheric flux
derived from sea salt aerosols, with a smaller anthropogenic
contribution from combustion of biomass and coal (Park & Sch-
lesinger, 2002; Schlesinger & Vengosh, 2016). For a well-studied
site in central Panama, close to BCI, we estimate that atmo-
spheric B inputs are sufficient to sustain the B demand of the for-
est. A study of a temperate forest revealed a considerable amount
of B in throughfall precipitation (Cividini et al., 2010), suggest-
ing that dry deposition represents an important additional B
input to the forest. The annual input of B in rainfall, assuming
the mean B concentration in marine rainfall, is equal to about
one-third of the B in live leaves, about two-thirds of the B in
annual litter fall, and exceeds the standing pool of soluble B in
the soil rooting zone. Using the median B concentration in rain-
fall, the B input is of similar magnitude to the soil soluble B pool.
We therefore assume that the small soluble B pool is replenished
annually by rainfall and litter decomposition. The foliar B pool is
dwarfed by B in wood, as is typical for forests worldwide (Lehto
et al., 2004), while wood B is many times smaller than the total B
in the soil. Indeed, most of the B in the ecosystem is in the total
soil B pool, presumably as tourmaline, one of the most common
and highly resistant cyclosilicates (Keren, 1996). Together with
the results of the micronutrient addition experiment, the B bud-
get indicates that it is unlikely that trees suffer B deficiency at this
site. This might not apply to continental locations such as the
Amazon basin, where rainfall B inputs are likely to be smaller
than in Panama (mean continental rainfall B = 4.89 lg l1) (Park
& Schlesinger, 2002). However, such regions are typically
extremely poor in P and base cations (e.g. in Amazonia; Quesada
et al., 2010), which reduces the likelihood that B limitation will
occur even with low B inputs in rainfall.
Boron is similar to P in terms of its distribution in the tropical
forest soil studied here: soluble concentrations are < 1 mg kg1,
while total concentrations are primarily contained in recalcitrant
forms that are not directly available to plants (compare Table 2
with values for total and soluble P on Gigante Peninsula) (Turner
et al., 2013, 2015). However, comparison of B in rainfall and leaf
litter fall provides support for our assertion that B is unlikely to
be of importance in structuring the vegetation in lowland tropical
forests. While rainfall provides an amount of B of comparable
magnitude to the annual B return in litter fall, rainfall P is an
order of magnitude less than rainfall B (soluble reactive P in BCI
rainfall is typically ≤ 1 lg P l1) and accounts for < 1% of the
annual litter fall P (319 mg Pm2 in the 2006–2007 annual
cycle) (Turner et al., 2015). More importantly, live leaves contain
c. 20 times more P than B, and the return of P in litter fall is
approximately an order of magnitude greater than for B. In other
words, despite similar availability in the soil, the annual plant
requirement for P is considerably greater than for B.
Conclusions
Widespread low concentrations of plant-available B in lowland
tropical forest soils and the absence of associations of tropical
tree species to soil B at the regional scale provide strong evi-
dence that B availability does not influence the distribution of
tropical tree species in Panama. Instead, the significant
responses of many tree species to moisture and readily
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist10
exchangeable soil P demonstrate that these two resources are
the primary drivers of species distributions in tropical forests.
The absence of a response of three species of seedlings to
extremely low B and of mature forest to long term experi-
mental B addition suggests that rainfall provides sufficient B
for optimal growth. We conclude that B availability does not
influence the distributions or productivity of tree species in
Panama and probably elsewhere in the lowland tropics.
Acknowledgements
We thank Aleksandra Bielnicka and Dayana Agudo for labora-
tory support, Omar Hernandez and Rufino Gonzalez for main-
taining the Gigante fertilization experiment, and Jorge Aranda,
Carolina Sarmiento, Carolyn Delevich, Jorge Salas and Jacinto
Perez for support with growth experiments at the Smithsonian
Tropical Research Institute Santa Cruz Research Facility in Gam-
boa, Panama.
Author contributions
B.L.T., P-C.Z., K.W. and J.W.D. designed the study. B.L.T., P-
C.Z., K.W., R.C., S.J.W. and J.W.D. collected data. B.L.T., P-
C.Z., K.W., R.C., S.J.W. and J.W.D. analyzed data. B.L.T.
wrote the manuscript with input from all authors.
References
Berger KC, Truog E. 1939. Boron determination in soils and plants using the
quinalizarin reaction. Industrial and Engineering Chemistry 11: 540–545.
Berger KC, Truog E. 1944. Boron tests and determination for soils and plants.
Soil Science 57: 25–36.
Bernier B, Brazeau M. 1988. An occurrence of boron deficiency in the deciuous
forest of the Quebec Appalachians and the St. Lawrence Lowlands. Canadian
Journal of Forest Research 18: 1652–1655.
Bolanos L, Esteban E, de Lorenzo C, Fernandez-Pascual M, de Felipe MR,
Garate A, Bonilla I. 1994. Essentiality of boron for symbiotic dinitrogen
fixation in pea (Pisum sativum) rhizobium nodules. Plant Physiology 104: 85–
90.
Chapman VJ, Edwards DG, Blamey FPC, Asher CJ. 1997. Challenging the
dogma of a narrow supply range between bdeficiency and toxicity of boron. In:
Bell RW, Rerkasem B, eds. Boron in soils and plants. Dordrecht, the
Netherlands: Kluwer Academic, 151–155.
Chave J, Olivier J, Bongers F, Cha^telet P, Forget P-M, van der Meer P, Norden
N, Riera B, Charles-Dominique P. 2008. Above-ground biomass and
productivity in a rain forest of eastern South America. Journal of Tropical
Ecology 24: 355–366.
Chave J, Rejou-Mechain M, Burquez A, Chidumayo E, Colgan MS, Delitti
WBC, Duque A, Eid T, Fearnside PM, Goodman RC et al. 2015. Improved
allometric models to estimate the aboveground biomass of tropical trees. Global
Change Biology 20: 3177–3190.
Cividini D, Lemarchand D, Chabaux F, Boutin R, Pierret M-C. 2010. From
biological to lithological control of the B geochemical cycle in a forest
watershed (Strengbach, Vosges). Geochimica et Cosmochimica Acta 74: 3143–
3163.
Condit R, Aguilar S, Hernandez A, Perez R, Lao S, Angehr G, Hubbell
SP, Foster RB. 2004. Tropical forest dynamics across a rainfall gradient
and the impact of an El Ni~no dry season. Journal of Tropical Ecology 20:
51–72.
Condit R, Engelbrecht BMJ, Pino D, Perez R, Turner BL. 2013. Species
distributions in response to individual soil nutrients and seasonal drought
across a community of tropical trees. Proceedings of the National Academy of
Sciences, USA 110: 5064–5068.
Dalling JW, Winter K, Nason JD, Hubbell SP, Murawski DA, Hamrick JL.
2001. The unusual life history of Alseis blackiana: a shade-persistent pioneer
tree? Ecology 82: 933–945.
Dell B, Hardy G, Burgess T. 2008.Health and nutrition of plantation eucalypts
in Asia. Southern Forests: A Journal of Forest Science 70: 131–138.
Engelbrecht BMJ, Comita LS, Condit R, Kursar TA, Tyree MT, Turner BL,
Hubbell S. 2007. Drought sensitivity shapes species distribution patterns in
tropical forests. Nature 447: 80–82.
Goldberg S. 1997. Reactions of boron with soils. Plant and Soil 193: 35–48.
Gupta UC. 1967. A simplified method for determining hot water-soluble boron
in podzol soils. Soil Science 103: 424–428.
Gupta UC, Jame YW, Campbell CA, Leyshon AJ, Nicholaichuk W. 1985.
Boron toxicity and deficiency: a review. Canadian Journal of Soil Science 65:
381–409.
Hartigan JA, Wong MA. 1979. A K-means clustering algorithm. Applied Statistics
28: 100–108.
H€ogberg P, Fan H, Quist M, Binkley DAN, Tamm CO. 2006. Tree growth and
soil acidification in response to 30 years of experimental nitrogen loading on
boreal forest. Global Change Biology 12: 489–499.
Hunter IR, Will GM, Skinner MF. 1990. A strategy for the correction of boron
deficiency in radiata pine plantations in New Zealand. Forest Ecology and
Management 37: 77–82.
Jeffrey AJ, McCallum LE. 1988. Investigation of a hot 0.01M CaCl2 soil boron
extraction procedure followed by ICP-AES analysis. Communications in Soil
Science and Plant Analysis 19: 663–673.
John R, Dalling JW, Harms KE, Yavitt JB, Stallard RF, Mirabello M, Hubbell
SP, Valencia R, Navarrete H, Vallejo M et al. 2007. Soil nutrients influence
spatial distributions of tropical tree species. Proceedings of the National Academy
of Sciences, USA 104: 864–869.
Johnson CM, Stout PR, Broyer TC, Carlton AB. 1957. Comparative chlorine
requirements of different plant species. Plant and Soil 8: 337–353.
Kaspari M, Garcia MN, Harms KE, Santana M, Wright SJ, Yavitt JB. 2008.
Multiple nutrients limit litterfall and decomposition in a tropical forest. Ecology
Letters 11: 35–43.
Keren R. 1996. Boron. In: Sparks DL, Page AL, Helmke PA, Loeppert RH,
Soltanpour PN, Tabatabai MA, Johnston CT, Sumner ME, eds.Methods of soil
analysis, part 3–chemical methods. Madison, WI, USA: Soil Science Society of
America, 603–626.
Lehto T, R€ais€anen M, Lavola A, Julkunen-Tiitto R, Aphalo PJ. 2004. Boron
mobility in deciduous forest trees in relation to their polyols. New Phytologist
163: 333–339.
Marschner H. 2006.Mineral nutrition of higher plants. San Diego, CA, USA:
Academic Press.
Martens CS, Harriss RC. 1976. Boron in coastal North Florida rainfall. Journal
of Geophysical Research 81: 6371–6375.
Mehlich A. 1984.Mehlich 3 soil test extractant: a modification of Mehlich 2
extractant. Communications in Soil Science and Plant Analysis 15: 1409–
1416.
Nable RO, Ba~nuelos GS, Paull JG. 1997. Boron toxicity. Plant and Soil 193:
181–198.
Park H, Schlesinger WH. 2002. Global biogeochemical cycle of boron. Global
Biogeochemical Cycles 16: 20-21–20-11.
Pyke CR, Condit R, Aguilar S, Lao S. 2001. Floristic composition across a
climatic gradient in a neotropical lowland forest. Journal of Vegetation Science
12: 553–566.
Quesada CA, Lloyd J, Schwarz M, Pati~no S, Baker TR, Czimczik C, Fyllas NM,
Martinelli L, Nardoto GB, Schmerler J et al. 2010. Variations in chemical and
physical properties of Amazon forest soils in relation to their genesis.
Biogeosciences 7: 1515–1541.
Sah RN, Brown PH. 1997. Boron determination – a review of analytical
methods.Microchemical Journal 56: 285–304.
Schlesinger WH, Vengosh A. 2016. Global boron cycle in the anthropocene.
Global Biogeochemical Cycles 30: 219–230.
Shorrocks V. 1997. The occurrence and correction of boron deficiency. Plant and
Soil 193: 121–148.
No claim to original US government works
New Phytologist
 2016 New Phytologist Trust
New Phytologist (2016)
www.newphytologist.com
New
Phytologist Research 11
Steidinger B. 2015.Qualitative differences in tree species distributions along soil
chemical gradients give clues to the mechanisms of specialization: why boron
may be the most important soil nutrient at Barro Colorado Island. New
Phytologist 206: 895–899.
Taber HG. 2004. Boron analysis of Mehlich No. 3 extractant with modified
inductive coupled argon plasma techniques to eliminate iron interference.
Communications in Soil Science and Plant Analysis 35: 2957–2963.
Turner BL, Bielnicka A, Dalling JW, Wolf JA. 2016. Interference by iron in the
determination of boron by ICP-OES spectrometry in Mehlich-III extracts and
total element digests of tropical forest soils. Communications in Soil Science and
Plant Analysis. doi: 10.1080/00103624.2016.1228952.
Turner BL, Engelbrecht BMJ. 2011. Soil organic phosphorus in lowland tropical
rain forests. Biogeochemistry 103: 297–315.
Turner BL, Wright SJ. 2014. The response of microbial biomass and hydrolytic
enzyme activities to a decade of nitrogen, phosphorus, and potassium addition
in a lowland tropical rain forest. Biogeochemistry 117: 115–130.
Turner BL, Yavitt JB, Harms KE, Garcia M, Wright SJ. 2015. Seasonal changes
in soil organic matter after a decade of nutrient addition in a lowland tropical
forest. Biogeochemistry 123: 221–235.
Turner BL, Yavitt JB, Harms KE, Garcia MN, Romero TE, Wright SJ. 2013.
Seasonal changes and treatment effects on soil inorganic nutrients following a
decade of fertilization in a lowland tropical forest. Soil Science Society of America
Journal 77: 1357–1369.
Windsor DM. 1990. Climate and moisture variability in a tropical forest: long
term records from Barro Colorado Island, Panama. Smithsonian Contributions
to Earth Science 29: 1–145.
Wright SJ, Yavitt JB, Wurzburger N, Turner BL, Tanner EVJ, Sayer EJ,
Santiago LS, Kaspari M, Hedin LO, Harms KE et al. 2011. Potassium,
phosphorus or nitrogen limit root allocation, tree growth and litter production
in a lowland tropical forest. Ecology 92: 1616–1625.
Yavitt JB, Harms KE, Garcia MN, Wright SJ, He F, Mirabello MJ. 2009.
Spatial heterogeneity of soil chemical properties in a lowland tropical moist
forest, Panama. Australian Journal of Soil Research 47: 674–687.
Yavitt JB, Wright SJ. 2001. Drought and irrigation effects on fine root dynamics
in a tropical moist forest, Panama. Biotropica 33: 421–434.
Supporting Information
Additional Supporting Information may be found online in the
Supporting Information tab for this article:
Fig. S1 Histograms of individual species responses to eight envi-
ronmental factors, including soil boron (B) determined by extrac-
tion in Mehlich-III solution.
Fig. S2 Change in plant-available boron determined by hot-
CaCl2 extraction in surface soil after 15 yr of micronutrient addi-
tion to lowland tropical forest on Gigante Peninsula, Panama.
Table S1 Pearson product-moment correlations between soil
properties and plant-available boron concentrations
Table S2 The relationship between foliar boron (B) and esti-
mated Mehlich-III extractable B in soil for six tree species grow-
ing on the Barro Colorado Island 50 ha forest dynamics plot in
Panama
Please note: Wiley Blackwell are not responsible for the content
or functionality of any Supporting Information supplied by the
authors. Any queries (other than missing material) should be
directed to the New Phytologist Central Office.
New Phytologist (2016) No claim to original US government works
New Phytologist
 2016 New Phytologist Trustwww.newphytologist.com
Research
New
Phytologist12