Arthropod Abundance and Diversity in a Lowland Tropical Forest Floor in Panama: The
Role of Habitat Space vs. Nutrient Concentrations
Emma J. Sayer1,2,3, Laura M. E. Sutcliffe1,4, Rebecca I. C. Ross1,5, and Edmund V. J. Tanner1
1 Department of Plant Sciences, Downing Street, Cambridge, CB2 3EA, UK
2 Smithsonian Tropical Research Institute, PO Box 0843-03092, Balboa, Ancon, Panama, Republic of Panama
ABSTRACT
Tropical forest floor characteristics such as depth and nutrient concentrations are highly heterogeneous even over small spatial scales and it is unclear how these
differences contribute to patchiness in forest floor arthropod abundance and diversity. In a lowland tropical forest in Panama we experimentally increased litter
standing crop by removing litter from five plots (L ) and adding it to five other plots (L1); we had five control plots. After 32mo of treatments we investigated how
arthropod abundance and diversity were related to differences in forest floor physical (mass, depth, water content) and chemical properties (pH, nutrient concen-
trations). Forest floor mass and total arthropod abundance were greater in L1 plots compared with controls. There were no treatment differences in nutrient con-
centrations, pH or water content of the organic horizons. Over all plots, the mass of the fermentation horizon (Oe) was greater than the litter horizon (Oi); arthropod
diversity and biomass were also greater in the Oe horizon but nutrient concentrations tended to be higher in the Oi horizon. Arthropod abundance was best explained
by forest floor mass, while arthropod diversity was best explained by phosphorus, calcium and sodium concentrations in the Oi horizon and by phosphorus concen-
trations in the Oe horizon. Differences in arthropod community composition between treatments and horizons correlated with phosphorus concentration and dry mass
of the forest floor. We conclude that at a local scale, arthropod abundance is related to forest floor mass (habitat space), while arthropod diversity is related to forest
floor nutrient concentrations (habitat quality).
Abstract in Spanish is available at http://www.blackwell-synergy.com/loi/btp
Key words: arthropod community composition; Barro Colorado Nature Monument; habitat quality; litter addition; litter fauna; Oe horizon; Oi horizon.
AN ESTIMATED THREE QUARTERS of the total animal biomass of a
tropical rain forest is comprised of small, soil- and litter-dwelling
arthropods (Wilson 1990). Soil and litter fauna contribute greatly
to decomposition dynamics ( Janzen 1987, Cou?teaux et al. 1995)
and nutrient mineralization (Swift et al. 1979, Andersen & Swift
1983), and the importance of soil fauna to decomposition and nu-
trient dynamics is thought to be disproportionately higher in trop-
ical than temperate forests (Heneghan et al. 1999, Gonza?lez &
Seastedt 2001, Yang et al. 2007), because the activity of soil and
litter organisms is not greatly constrained by climate variability
(Lavelle et al. 1993). In addition, soil and litter arthropods repre-
sent an important food resource for other forest animals (Kattan
et al. 2006).
The forest floor is highly heterogeneous over short distances
(Milton & Kaspari 2007) and small-scale changes in the chemical
and physical properties of the forest floor (e.g., depth, microclimate,
nutrient concentrations) will exert a range of direct and indirect
effects on the fauna it supports. Although relatively few arthropod
species feed directly on leaf litter, it is an important substrate for
fungi and bacteria (Swift & Anderson 1989), which in turn are a
major source of invertebrate nutrition (Killham 1994, Lodge
1996). The forest floor represents a habitat for both predators and
prey; litter accumulation provides more structural complexity and
creates a habitat for a more diverse range of insects (Bultman &
Uetz 1984, Barberena-Arias & Aide 2003). Nevertheless, litter mass
alone does not always explain differences in the number of arthro-
pods present at a site, nor will it affect all trophic or taxonomic
groups in the same way: microbivores, for example, are likely to be
limited by nutrients (Sterner & Elser 2002, Milton & Kaspari
2007), while predators are more likely to respond to prey density
and changes in habitat structure (Uetz 1979). The accumulation of
organic matter may also have negative effects on arthropod popu-
lations; compaction can reduce habitat space and make the forest
floor unsuitable for many taxa (Levings & Windsor 1984) and phe-
nolic compounds from decomposing litter could act as a deterrent
to arthropods (Sayer 2006).
The nutritional quality of organic matter is key to determining
decomposition rates (Enriquez et al. 1993) and it has been sug-
gested that the nutrients driving decomposition will also affect the
arthropods involved in the decomposition process and their preda-
tors (Kaspari et al. 2007, Milton & Kaspari 2007). A recent study
found strong evidence for phosphorus limitation of forest litter
fauna at a landscape scale in lowland tropical forest (McGlynn et al.
2007) but a fertilization experiment in disturbed tropical forest
showed little effect of nutrient availability on arthropod abundance
at smaller scales (Yang et al. 2007) and the increased habitat space
provided by thicker litter may ultimately be more important in de-
termining arthropod abundance and diversity than nutrient avail-
ability (Gill 1969). Experimental and observational studies of the
relationships between arthropod abundance and the quantity of
Received 5 November 2008; revision accepted 1 July 2009.
3Corresponding author; e-mail: ejs44@cam.ac.uk
4Current address: Department of Vegetation Analysis and Phytodiversity,
Albrecht von Haller Institute of Plant Sciences, University of Go?ttingen, 37073
Go?ttingen, Germany.
5Current address: Department of Plant Sciences, University of Oxford, South
Parks Road, Oxford, OX1 3RB, UK.
BIOTROPICA 42(2): 194?200 2010 10.1111/j.1744-7429.2009.00576.x
194 r 2009 The Author(s)
Journal compilationr 2009 by The Association for Tropical Biology and Conservation
organic matter on the forest floor, with a few notable exceptions
(e.g., Burghouts et al. 1992, Yang et al. 2007), have been limited to
temperate forests (e.g., David et al. 1991, Reynolds et al. 2003). We
hypothesized that small-scale differences in both habitat space
and habitat quality will be important predictors of arthropod com-
munity composition in the forest floor. To test this, we used a
large-scale litter manipulation experiment in a lowland tropical
semi-evergreen forest to relate forest floor characteristics to arthro-
pod abundance and diversity to determine: (1) whether differences
in forest floor mass are related to arthropod abundance and diver-
sity; (2) whether the nutrients thought to limit decomposition also
explain differences in arthropod communities in the forest floor;
and (3) whether differences in the characteristics of distinct organic
horizons are related to arthropod abundance and diversity.
METHODS
SITE DESCRIPTION.?The study was carried out within an ongoing
long-term litter manipulation experiment (Sayer et al. 2006a, b,
2007). The forest under study is an old growth lowland tropical
forest, located on the Gigante Peninsula (91060 N, 791540 W) of
the Barro Colorado Nature Monument in Panama, Central Amer-
ica. Nearby Barro Colorado Island (ca 5 km from the study site)
receives a mean annual rainfall of 2600mm and has an average
temperature of 271C (Leigh 1999). There is a strong dry season
from January to April with a median rainfall of o 100mm/mo
(Leigh & Wright 1990). Fifteen 45 45m plots were established
within a 40 ha area (500 800m) of old growth forest in 2000. In
2001 all 15 plots were trenched to a depth of 0.5m to minimize
lateral nutrient and water movement via the root/mycorrhizal net-
work; the sides of the trenches were double-lined with plastic and
the trenches were backfilled. Starting in January 2003, the litter
(including branches  100mm diam.) in five plots was raked up
once a month, resulting in low, but not entirely absent, litter stand-
ing crop (L plots). The removed litter was immediately spread
on five further plots, approximately doubling the monthly litterfall
(L1 plots); five plots were left as controls (CT plots).
SAMPLING.?Litter samples were collected at five randomly chosen
sampling sites in each of the CT and L1 plots over a 1-wk period in
September 2005. An equal number of plots per treatment were
sampled on each collection date. At each sampling site, the depth of
the forest floor was measured using a rule-marked knife, a PVC
tube of 20 cm inner diam. and 12 cm height was placed on the for-
est floor and the sample inside the tube was separated from the for-
est floor by cutting along the inside wall of the tube. The organic
horizons of the forest floor were defined according to Brady (1990)
and collected separately: the litter layer (Oi horizon), defined as
freshly fallen leaves in the early stages of decomposition, was re-
moved from the tube by hand, placed in a preweighed, labeled cloth
bag and immediately weighed to the nearest 0.1 g. The depth of the
fermentation layer (Oe horizon), defined as denser, fragmented but
identifiable organic matter in a more advanced stage of decompo-
sition, was then measured and the sample collected as described
above; there was no apparent humus layer (Oa horizon) in the study
forest. Immediately upon returning from the field, the samples were
transferred to Berlese funnels lined with 4mm wire mesh and ar-
thropods were extracted for 24 h, by which time the samples had
dried out. The forest floor samples from the funnels were then
oven-dried to constant weight at 601C to determine water content
and finely ground for nutrient analysis. Phosphorus and cation
concentrations were determined after acid digestion by radial view
inductively coupled plasma?optical emission spectrometry, total
nitrogen was determined by complete combustion gas chromatog-
raphy by Waite Analytical Services, Adelaide, Australia. The pH of
the Oi and Oe horizons was determined on dried, ground subsam-
ples in deionized water (1:3 ratio).
All fauna samples were preserved in 70 percent alcohol until
identification. The animals were identified under a stereomicro-
scope following McGavin (2000) and body length was measured to
the nearest millimeter. Most animals were grouped to order level,
but in some cases higher and lower taxonomic levels were used; ho-
lometabolic larvae were combined into one miscellaneous group.
The animals collected also included Annelids and Gastropods but
as they were only 2 percent of the total we henceforth use only the
terms arthropod abundance and arthropod diversity in the interests
of succinctness.
DATA ANALYSIS.?Total arthropod abundance (number of individ-
uals), average body length and Simpson?s Index of Diversity
(1D) were calculated for each plot and horizon.
Differences in species composition between treatments and
horizons and their relation to forest floor characteristics were ana-
lyzed with nonmetric multidimensional scaling (NMMDS) using
the Soerensen distance measure in PC-ORD 4.1 for Windows
(MjM Software, Gleneden Beach, OR, U.S.A.).
Comparisons between treatments and organic horizons were
made for arthropod abundance, body length, diversity and for the
properties of the organic layers (mass, depth, water and nutrient
content). The paired data of the forest floor layers (N = 10) were
analyzed with Wilcoxon signed-rank tests. Comparisons between
treatments (N = 5 per treatment) for each organic layer separately
were made using Mann?Whitney U tests. We did not apply a cor-
rection of a values for multiple tests, as the tests we performed can-
not be considered completely independent and a correction would
have resulted in a higher probability of Type II errors (Moran 2003,
Gotelli & Ellison 2004).
We explored the relationships between forest floor character-
istics and total arthropod abundance and diversity in the organic
horizons using stepwise linear regression. To avoid data mining, we
considered only those predictor variables that differed between
treatments or horizons or were important in determining commu-
nity composition in our NMMDS analysis (litter depth, litter mass,
P, Ca, Na and Mg). Of these, litter depth and litter mass were
strongly correlated, and as litter mass differed between both treat-
ments and horizons and was an important predictor in the
NMMDS analysis, we excluded litter depth. We thus considered
five predictor variables with a sample size of N = 10 each; variables
were chosen using the stepwise procedure with Po 0.05 to enter or
P4 0.1 to remove (Sokal & Rohlf 1995). All statistical analyses
Arthropods in a Tropical Forest Floor 195
except NMMDS were performed in SPSS 13.0 for Macintosh
(SPSS Inc., Chicago, IL, U.S.A.).
RESULTS
FOREST FLOOR CHARACTERISTICS.?Forest floor depth was 72 percent
greater in the L1 plots (38mm) than the CT plots (22mm;
P = 0.028) but the depth of the individual organic layers did not
differ between treatments. Forest floor mass was 65 percent greater in
the L1 plots (2148 g/m2) than the controls (1301 g/m2; P = 0.047)
and the mass of the Oi and Oe horizons were 51 and 68 percent
greater in the L1 plots compared with the controls (P = 0.047 and
0.076, respectively; Table 1). Depth was strongly correlated with
mass for the forest floor and Oe horizon but not for the Oi horizon.
There were no treatment differences in water content, pH or nutrient
concentrations of the organic horizons (Table 1).
Over all plots (N = 10), the organic horizons differed from
each other in several characteristics; the mass of the Oi horizon
(970 g/m2) was less than the mass of the Oe horizon (2946 g/m2;
P = 0.005) and the water content of the Oi horizon (69.6%) was
greater than the Oe horizon (60.7%; P = 0.005), while the con-
centrations of Ca, Mg, Na and K were higher in the Oi than the Oe
horizon (P = 0.041, 0.008, 0.007 and 0.005, respectively).
LITTER FAUNA.?A total of 2077 individuals were identified; 20
groups were identified to order, 2 groups to class, and 1 group to
subphylum. Hymenoptera was the order with the most individuals
(32 percent of the total number of individuals collected), 99 per-
cent of which were ants. Acarina was the second most abundant
order (14 percent of all individuals collected), and larvae made up
11 percent of all individuals collected.
The mean number of arthropods per plot was higher in the
L1 plots (274 individuals) compared with the controls (141 indi-
viduals; P = 0.047), and the number of individuals in the Oe ho-
rizon was also greater in the L1 plots than the controls (P = 0.032)
but arthropod body length and Simpson?s Index of Biodiversity did
not differ between treatments (Table 2).
NMMDS showed clear differences in arthropod community
composition between treatments (Fig. 1) and these differences were
related to forest floor depth and phosphorus concentration. A two-
dimensional solution was sufficient to achieve low stress values and
the correlations between arthropod groups and axes showed that
Arachnida, Acari and larvae contributed most to the first axis, while
Isoptera and Diptera contributed most to the second axis. Of these,
Arachnida and larvae tended to be more abundant in the L1 plots
(Table 3), while the number of Isoptera in the CT plots was 27
times higher than in the litter addition treatment (Table 3) but their
distribution was very patchy.
The Oi and Oe organic horizons differed from each other with
respect to arthropod community composition (Fig. 1) but not
abundance. A total of 1172 individuals were collected from the Oi
horizon and 905 individuals were collected in the Oe horizon. Over
all plots, arthropod density was greater in the Oi horizon compared
with the Oe horizon (P = 0.005; Table 2) but arthropod diversity
and body length were greater in the Oe horizon (P = 0.005 and
0.007, respectively). Coleoptera and Diplopoda tended to be more
abundant in the Oe than the Oi horizon (Table 3), while Diptera
and Hymenoptera were found in greater numbers in the Oi horizon
(Table 3).
RELATIONSHIPS BETWEEN ARTHROPODS AND FOREST FLOOR CHARACTER-
ISTICS.?Multiple regression analysis showed that no combination
of the forest floor characteristics that differed between organic ho-
rizons or treatments could adequately explain arthropod abundance
in the Oi horizon. However, the diversity of arthropods in the Oi
horizon was best explained by the model including phosphorus,
calcium and sodium concentrations (R2 = 0.649, P = 0.025), of
which calcium concentration was the most important explanatory
variable. The abundance of arthropods in the Oe horizon was best
explained by a model including dry mass of the Oe horizon, mag-
nesium and sodium concentrations, which accounted for 69.8 per-
cent of the variance (P = 0.016) and the mass of the Oe horizon was
the most important explanatory variable in the model. Arthropod
diversity in the Oe horizon was best explained by phosphorus con-
centration and dry mass; the total variance explained by the model
TABLE1. Characteristics of the Oe and Oi organic horizons of the forest floor in control plots and litter addition treatments in a lowland tropical forest in Panama,
Central America; numbers in parentheses are SE of means for N= 5.
Control Litter addition
Oi Oe Oi Oe
Dry mass (g/m2) 194 ( 24) 1107 ( 214) 294 ( 25) 1854 ( 301)
Depth (mm) 11 ( 2.6) 11 ( 1.5) 15 ( 2.0) 23 ( 5.7)
Water content (% fresh weight) 68.2 ( 1.21) 62.5 ( 1.15) 70.9 ( 1.09) 58.9 ( 2.18)
N (% dry weight) 1.4 ( 0.10) 1.4 ( 0.09) 1.6 ( 0.05) 1.3 ( 0.03)
P (mg/kg) 406 ( 29.9) 458 ( 19.3) 446 ( 32.3) 452 ( 14.6)
K (mg/kg) 1462 ( 235) 696 ( 39.7) 1696 ( 242) 664 ( 60.1)
Ca (mg/kg) 17,300 ( 1307) 16,060 ( 2516) 17,400 ( 943) 11,880 ( 424)
Mg (mg/kg) 3040 ( 322) 2188 ( 143) 3100 ( 241) 1926 ( 34.8)
Na (mg/kg) 115 ( 17.2) 88.7 ( 3.13) 134 ( 19.2) 80.6 ( 11.5)
196 Sayer, Sutcliffe, Ross, and Tanner
was 47.6 percent (P = 0.043), but phosphorus concentration alone
explained 42.6 percent (P = 0.024).
DISCUSSION
The depth and mass of the forest floor was increased by our litter
addition treatments, but pH, water content and nutrient con-
centrations remained unaffected. Thus, the litter addition treat-
ment allowed us to study arthropod communities over a wide
range of forest floor depths without affecting forest floor nutrient
concentrations.
The number of arthropods collected in this study probably
underestimates the actual number of arthropods in the forest floor
for two reasons: firstly, the size of the sampling frames used
(0.0625m2) likely excluded many larger arthropods and resulted
in low abundances of, for example, millipedes, centipedes and large
beetles; secondly, sampling efficiency of Berlese traps varies greatly
between arthropod groups, with low sampling efficiency (o 50 per-
cent) for Acari, Annelidae, Collembola, Diplura, Diptera and Le-
pidoptera (Levings & Windsor 1982). Nevertheless, Acari,
Collembola and Diptera were abundant in our samples and the rel-
ative values are reliable for comparing treatments and horizons in
this study. Our identification to order level was considered suffi-
cient, as assemblage differences at higher levels have been shown to
reliably reflect those at species levels (Williams & Gaston 1994,
Balmford et al. 1996).
The greater abundance of arthropods in the L1 plots can be
attributed to the greater mass and depth of the organic horizons in
this treatment, as nutrient concentrations, water content and pH
did not differ from the controls. There were no differences between
treatments when arthropod abundance was calculated as number of
individuals per unit litter mass (Table 2), rather than unit area,
which is a further indication that the treatment differences were
solely due to differences in the mass of organic matter. Studies in
tropical forests in Malaysia and Puerto Rico have shown that an
increase in litter quantity may be more important than an increase
in litter quality in determining arthropod abundance (Burghouts
et al. 1992, Yang et al. 2007) because the increased structural com-
plexity with greater litter depth provides more habitat space (Uetz
1979, Barberena-Arias & Aide 2003) and a greater number of ref-
uges from predators (Pearse 1943). Other studies in tropical forests
have found no strong relationship between arthropod numbers and
litter mass or depth, but this can be attributed to seasonal differ-
ences or sampling methods. For example, arthropod abundance
is poorly related to litter mass during the dry season, when litterfall
is high but arthropod abundance low due to lack of moisture
(Levings & Windsor 1982, 1984) and the number of arthropods
caught in pitfall traps does not always correlate well with litter
depth (McGlynn et al. 2007) as the deeper the litter around the
trap, the greater the structural complexity and the less litter animals
are likely to walk into pitfall traps (Greenslade 1964).
Although it is conceivable that soil fauna is transferred with the
litter to the L1 plots in our study, it is unlikely that the numbers of
animals added to the L1 plots with the litter each month has any
great bearing on our results. The monthly accumulation of litter in
the L plots between raking events at the time of the study was
low (2.2 g/m2/d; E. J. Sayer & E. V. J. Tanner, unpubl. data) and
the sparse litter cover in combination with repeated disturbance in
TABLE2. Mean arthropod abundance (number of individuals), density (individuals per gram organic matter), Simpson?s Index of Diversity, and size (body length) in the Oi
and Oe organic horizons in control plots and litter addition treatments in a lowland tropical forest in Panama, Central America; numbers in parentheses are SE of
means for N= 5; different upper case superscript letters within a row indicate significant differences between treatments and lower case superscript letters indicate
significant differences between horizons within a treatment at Po 0.05.
Control Litter addition
Oi Oe Oi Oe
Abundance (No. ind./m2) 1360 ( 242) 899 ( 94)A 2384 ( 658) 1993 ( 416)B
Density (No. ind./g) 6.98 ( 0.7)a 0.93 ( 0.19)b 9.36 ( 3.3)a 1.08 ( 0.19)b
Simpson?s Index of Diversity (1D) 0.69 ( 0.06)a 0.80 ( 0.01)b 0.73 ( 0.03)a 0.79 ( 0.02)b
Body length (mm) 2.17 ( 0.13)a 3.14 ( 0.17)b 2.41 ( 0.13) 2.96 ( 0.15)
FIGURE1. Nonmetric multidimensional scaling plot based on the Soerensen
distances in arthropod community composition between control plots (circles)
and litter addition plots (triangles) and the Oi horizon (closed symbols) and Oe
horizon (open symbols) in a lowland tropical forest in Panama, Central America;
arrows show the forest floor characteristics correlated with arthropod commu-
nity composition.
Arthropods in a Tropical Forest Floor 197
the L plots is not likely to invite colonization by a sizable com-
munity of litter fauna, especially of larger animals. In addition,
more mobile animals are liable to abandon the litter during litter
raking and transport. While it is possible that smaller arthropods
such as Collembola and Acari are added to the L1 plots with the
litter, a previous study using litter bags showed that the abundance
of mesoarthropods in the L plots is greatly reduced (Sayer et al.
2006b). Thus the numbers of animals added with the litter is ex-
pected to be small compared with the overall effect of accumulated
litter after 3 yr of litter addition treatments.
We had predicted reduced numbers of arthropods in the Oe
horizon due to greater compaction of the organic matter and lower
nutrient levels. Contrary to expectations, arthropod abundance did
not differ between organic horizons and the body size of arthropods
was greater in the Oe than in the Oi horizon. Arthropod abundance
was not related to the compactness (calculated as dry weight per
1mm depth) of either of the organic horizons and the greater mass
of the Oe horizon probably provided more habitat space for larger
arthropods such as Coleoptera and Diplopoda, which were more
abundant in this horizon.
The greater forest floor mass in the L1 plots did not affect any
one particular trophic group; we found higher numbers of detritivores
(e.g., Diplopoda), microbivores (e.g., Collembola) and predators (e.g.,
Arachnida, in particular Pseudoscorpionida) in the L1 plots. Detritiv-
ores and microbivores are likely to be attracted to the greater resource
availability in patches with greater forest floor mass, while predators
are more likely to respond to the greater availability of prey in thicker
litter (Uetz 1979, Chen & Wise 1999, Milton & Kaspari 2007); this
is demonstrated in our study by the concurrent increase in abundance
of both Collembola and Pseudoscorpions, which are known predators
of Collembola (Chen & Wise 1999), in the L1 plots.
The quality of forest floor organic matter played a lesser role in
determining arthropod abundance than did forest floor mass but
nutrient concentrations were important in explaining differences in
diversity. Phosphorus was an important factor in explaining arthro-
pod diversity in both organic horizons and NMMDS analysis dem-
onstrated that phosphorus concentration was strongly related to
arthropod community composition. Phosphorus is thought to be
the main limiting nutrient for decomposition in the study area
(Kaspari et al. 2007), as concentrations in the soil are low (Cavelier
1992, Sayer et al. 2006a), and decomposition studies have shown
net accumulation of phosphorus in the early stages of decomposi-
tion (Sayer et al. 2006b). A fertilization experiment in the study
forest showed that phosphorus stimulated the growth of cellulose-
decomposing microbes resulting in greater microbivore biomass in
phosphorus fertilized plots (Kaspari et al. 2007) and phosphorus
was also shown to limit litter fauna at the landscape scale in tropical
rain forest in Costa Rica (McGlynn et al. 2007).
Despite the high calcium concentrations in the soil and litter
in the study forest (Cavelier 1992), small-scale differences in cal-
cium concentrations remained important in explaining the diversity
of arthropods in the Oi horizon as many arthropods utilize large
amounts of calcium in their exoskeletons (Cromack et al. 1977).
We expected the abundance of fungal-feeding arthropods to
correlate with potassium concentrations as fungi actively accumu-
late potassium during the decomposition process and greater po-
tassium concentrations in the litter can be associated with higher
fungal biomass (Sayer et al. 2006b). In contrast, we found no rela-
tionship between potassium concentrations and the abundance or
diversity of arthropods in this study, even though potassium is also
thought to limit decomposition in this forest (Kaspari et al. 2007).
Higher sodium concentrations were related to the abundance
of arthropods in the Oe horizon and the diversity of arthropods in
the Oi horizon. Sodium is a highly mobile nutrient but many de-
composer fungi accumulate sodium in their fruiting bodies, which
may attract litter arthropods (Schowalter 2006). Numerous inver-
tebrates have high assimilation rates for sodium (Reichle et al.
TABLE3. Total numbers of arthropods in samples taken from the Oi and Oe
horizons of control and litter addition plots in a lowland tropical forest
in Panama, Central America.
Taxa
Control Litter addition
Oi Oe Total Oi Oe Total
Annelida 10 7 17 4 15 19
Arachnida 70 53 123 145 118 263
Araneae 20 8 28 15 17 32
Acarina 45 40 85 93 87 180
Opiliones 0 0 0 2 1 3
Pseudoscorpionida 5 5 10 35 13 48
Collembola 24 9 33 38 38 76
Entomobridae 2 1 3 0 0 0
Isotomidae 9 2 11 22 21 43
Onychiuridae 6 5 11 14 14 28
Smithinuridae 7 1 8 2 3 5
Crustacea 6 5 11 14 22 36
Isopoda 6 5 11 14 22 36
Entognatha 2 7 9 1 20 21
Protura 0 0 0 0 2 2
Diplura 2 7 9 1 18 19
Gastropoda 2 8 10 1 3 4
Insecta 273 124 397 463 272 735
Orthoptera 3 1 4 2 4 6
Blattaria 1 4 5 5 2 7
Coleoptera 18 28 46 26 61 87
Staphilinidae 1 0 1 2 0 1
Diptera 45 23 68 35 13 48
Hemiptera 8 1 9 6 9 15
Hymenoptera 40 60 100 384 182 566
Isoptera 157 5 162 5 1 6
Lepidoptera 1 2 3 0 0 0
Myriapoda 6 16 22 18 50 68
Diplopoda 6 16 22 18 45 63
Chilopoda 0 0 0 0 5 5
Symphyla 0 0 0 1 4 5
Symphylan 0 0 0 1 4 5
Holometab. larvae 32 53 85 62 81 143
Total 425 282 707 747 623 1370
198 Sayer, Sutcliffe, Ross, and Tanner
1969) and arthropod tissues can contain concentrations of sodium
that are much higher than those found in leaves or litter (Cromack
et al. 1977). Thus, sodium may be a limiting nutrient for some
arthropod groups in the study forest.
CONCLUSIONS
Our results demonstrate that, on a local scale, differences in the
numbers of arthropods in the forest floor can be explained mostly
by variation in the mass of the organic horizons, which suggests that
available habitat space is the most important predictor of arthropod
abundance. In contrast, arthropod diversity was strongly related to
small-scale differences in the concentrations of phosphorus, sodium
and calcium, suggesting that habitat quality is more important in
maintaining diversity. Thus, local-scale differences in arthropod
community composition are explained by the influence of habitat
space on the number of individuals and the influence of habitat
quality on taxonomic diversity.
ACKNOWLEDGMENTS
We thank W. Foster for help with insect identification, M. Kaspari
for helpful comments and the use of the Berlese funnels and two
anonymous reviewers for their constructive comments on the manu-
script. This study was funded by the Mellon Foundation, and the
Smithsonian Tropical Research Institute, E. J. S. was supported by
the Gates Cambridge Trust, a Worts? Travelling Scholarship, and a
Cambridge Philosophical Society Travel Grant. R. I. C. R. was sup-
ported by a studentship from the Gatsby Charitable Foundation.
The fieldwork was carried out at the Smithsonian Tropical Research
Institute in Panama, to whom we are very grateful.
LITERATURE CITED
ANDERSEN, J. M., AND M. J. SWIFT. 1983. Decomposition in tropical forests. In
S. L. Sutton, T. C. Whitmore, and A. C. Chadwick (Eds.). Tropical rain
forest: Ecology and management, pp. 287?309. Blackwell Scientific,
Oxford, UK.
BALMFORD, A., H. JAYASURIYA, AND M. GREEN. 1996. Using higher-taxon rich-
ness as a surrogate for species richness: 1. Regional tests. Proc. R. Soc.
Biol. Sci. Ser. B 263: 1267?1274.
BARBERENA-ARIAS, M. F., AND T. M. AIDE. 2003. Variation in species and trophic
composition of insect communities in Puerto Rico. Biotropica 34:
357?367.
BRADY, N. C. 1990. The nature and properties of soils (10th Edition). Prentice
Hall, Englewood Cliffs, New Jersey.
BULTMAN, T. L., AND G. W. UETZ. 1984. Effect of structure and nutritional
quality of litter on abundances of litter-dwelling arthropods. Am. Nat.
111: 165?172.
BURGHOUTS, T., G. ERNSTING, G. KORTHALS, AND T. DE VRIES. 1992. Litterfall,
leaf litter decomposition and litter invertebrates in primary and selec-
tively logged dipterocarp forest in Sabah, Malaysia. Philos. Trans. R.
Soc. Lond., Ser. B: Biol. Sci. 335: 407?416.
CAVELIER, J. 1992. Fine-root biomass and soil properties in a semideciduous and
a lower montane rain forest in Panama. Plant Soil 142: 187?201.
CHEN, B., AND D. H. WISE. 1999. Bottom-up limitation of predaceous arthro-
pods in a detritus-based terrestrial food web. Ecology 80: 761?772.
COU?TEAUX, M., P. BOTTNER, AND B. BERG. 1995. Litter decomposition, climate
and litter quality. Trends Ecol. Evol. 10: 63?66.
CROMACK, K., P. SOLLINS, R. L. TODD, D. A. CROSSLEY, W. M. FENDER, R.
FOGEL, AND A. W. TODD. 1977. Soil microorganism?arthropod interac-
tions: fungi as major calcium and sodium sources. In W. J. Mattson
(Ed.). The role of arthropods in forest ecosystems, pp. 78?84. Springer,
New York, New York.
DAVID, J. F., J. F. PONGE, P. ARPIN, AND G. VANNIER. 1991. Reactions of the
macrofauna of a forest mull to experimental perturbations of litter sup-
ply. Oikos 61: 316?326.
ENRIQUEZ, S., C. DUARTE, AND K. SAND-JENSEN. 1993. Patterns in decomposi-
tion rates among photosynthetic organisms: The importance of detritus
C:N:P content. Oecologia 94: 457?471.
GILL, R. W. 1969. Soil microarthropod abundance following old-field litter
manipulation. Ecology 50: 805?816.
GONZA?LEZ, G., AND T. R. SEASTEDT. 2001. Soil fauna and plant litter decompo-
sition in tropical and subalpine forests. Ecology 82: 955?964.
GOTELLI, N. J., AND A. M. ELLISON. 2004. A primer of ecological statistics.
Sinauer Associates Inc., Sunderland, Massachusetts.
GREENSLADE, P. J. M. 1964. Pitfall trapping as a method for studying popula-
tions of Carabidae (Coleoptera). J. Anim. Ecol. 33: 301?310.
HENEGHAN, L., D. C. COLEMAN, X. ZOU, D. A. CROSSLEY, AND B. L. HAINES. 1999.
Soil microarthropod contributions to decomposition dynamics: Tropical-
temperate comparisons of a single substrate. Ecology 80: 1873?1882.
JANZEN, D. H. 1987. Insect diversity of a Costa Rican dry forest?Why keep it
and how? Biol. J. Linn. Soc. 30: 343?356.
KASPARI, M., M. N. GARCIA, K. E. HARMS, M. SANTANA, S. J. WRIGHT, AND J. B.
YAVITT. 2007. Multiple nutrients limit litterfall and decomposition in
a tropical forest. Ecol. Lett. 10: 1?9.
KATTAN, G. H., D. CORREA, F. ESCOBAR, AND C. MEDINA. 2006. Leaf-litter
arthropods in restored forests in the Columbian Andes: A com-
parison between secondary forest and tree plantations. Restor. Ecol. 14:
95?102.
KILLHAM, K. 1994. Soil ecology. Cambridge University Press, Cambridge, UK.
LAVELLE, P., E. BLANCHART, A. MARTIN, S. MARTIN, A. SPAIN, F. TOUTAIN,
I. BAROIS, AND R. SCHAEFER. 1993. A hierarchical model of decomposi-
tion in terrestrial ecosystems: Application to soils of the humid tropics.
Biotropica 25: 130?150.
LEIGH, E. G. 1999. Tropical forest ecology. Oxford University Press, Oxford,
UK.
LEIGH, E. G., AND S. J. WRIGHT. 1990. Barro Colorado Island and tropical bio-
logy. In A. H. Gentry (Ed.). Four Neotropical rainforests, pp. 28?47.
Yale University Press, New Haven, Connecticut.
LEVINGS, S. C., AND D. M. WINDSOR. 1982. Seasonal and annual variation in
litter arthropod populations. In E. G. Leigh, A. S. Rand, and D. M.
Windsor (Eds.). The ecology of a tropical rain forest, pp. 355?387.
Smithsonian Institution, Washington, DC.
LEVINGS, S. C., AND D. M. WINDSOR. 1984. Litter moisture content as a deter-
minant of litter arthropod distribution and abundance during the dry
season on Barro Colorado Island, Panama. Biotropica 16: 125?131.
LODGE, J. 1996. Microorganisms. In D. Reagan and R. Waide (Eds.). The food
web of a tropical rain forest, pp. 53?108. The University of Chicago
Press, Chicago, Illinois.
MCGAVIN, G. C. 2000. Insects, spiders and other terrestrial arthropods. Dorling
Kindersley, London, UK.
MCGLYNN, T. P., D. J. SALINAS, R. R. DUNN, T. E. WOOD, D. LAWRENCE, AND
D. A. CLARK. 2007. Phosphorus limits tropical rain forest litter fauna.
Biotropica 39: 50?53.
MILTON, Y., AND M. KASPARI. 2007. Bottom-up and top-down regulation of
decomposition in a tropical forest. Oecologia 153: 163?172.
MORAN, M. D. 2003. Arguments for rejecting the sequential Bonferroni in
ecological studies. Oikos 100: 403?405.
PEARSE, A. S. 1943. Effects of burning-over and raking-off litter on certain
animals in the Duke Forest. Am. Nat. 29: 406?424.
REICHLE, D., M. H. SHANKS, AND D. A. CROSSLEY. 1969. Calcium, potassium
and sodium content of forest floor arthropods. Ann. Entomol. Soc. Am.
62: 57?62.
Arthropods in a Tropical Forest Floor 199
REYNOLDS, B. C., D. A. CROSSLEY, AND M. D. HUNTER. 2003. Response of soil
invertebrates to forest canopy inputs along a productivity gradient.
Pedobiologia 47: 127?139.
SAYER, E. J. 2006. Using experimental manipulation to assess the roles of leaf
litter in the functioning of forest ecosystems. Biol. Rev. 81: 1?31.
SAYER, E. J., E. V. J. TANNER, AND A. W. CHEESMAN. 2006a. Increased litterfall
changes fine root distribution in a moist tropical forest. Plant Soil 281:
5?13.
SAYER, E. J., E. V. J. TANNER, AND A. L. LACEY. 2006b. Effects of litter manip-
ulation on early-stage decomposition and meso-arthropod abundance in
a tropical moist forest. For. Ecol. Manage. 229: 285?293.
SAYER, E. J., E. V. J. TANNER, AND J. S. POWERS. 2007. Increased litterfall in
tropical forests boosts the transfer of soil CO2 to the atmosphere. PLOS
One 2: e1299.
SCHOWALTER, T. D. 2006. Insect ecology: An ecosystem approach (2nd Edition).
Academic Press, London, UK.
SOKAL, R. R., AND F. J. ROHLF. 1995. Biometry: The principals and practice of
statistics in biological research (3rd Edition). W. H. Freeman and Co.,
New York.
STERNER, R. W., AND J. J. ELSER. 2002. Ecological stoichiometry: The biology of
elements from molecules to the biosphere. Princeton University Press,
Princeton, New Jersey.
SWIFT, M. J., AND J. M. ANDERSON. 1989. Decomposition. In H. Lieth and
M. J. A. Werger (Eds.). Tropical rain forest ecosystems: Biogeo-
graphical and ecological studies, pp. 547?569. Elsevier, Amsterdam,
The Netherlands.
SWIFT, M. J., O. W. HEAL, AND J. M. ANDERSON. 1979. Decomposition in ter-
restrial ecosystems. Blackwell Scientific, Oxford, UK.
UETZ, G. W. 1979. The influence of variation in litter habitats on spider com-
munities. Oecologia 40: 29?42.
WILLIAMS, P., AND K. GASTON. 1994. Measuring more of biodiversity?Can
higher-taxon richness predict wholesale species richness? Biol. Conserv.
67: 211?217.
WILSON, E. O. 1990. Success and dominance in ecosystems: The case of the
social insects. Ecology Institute, Oldendorf/Luhe, Germany.
YANG, X., M. WARREN, AND X. ZOU. 2007. Fertilization responses of soil litter
fauna and litter quantity, quality and turnover in low and high elevation
forests of Puerto Rico. Appl. Soil Ecol. 37: 63?71.
200 Sayer, Sutcliffe, Ross, and Tanner