Patterns in tropical leaf litter and implications for angiosperm paleobotany

15 Pages·2002·1.06 MB·English

REVIEW
OF
PALAEOBOTANY
AND
PALYNOLOGY
ELSEVIER Review of Palaeobotany and Palynology 18 ( 1994 ) 99 311
Patterns in tropical leaf litter and implications for
angiosperm paleobotany
Robyn J. Burnham
Museum oPaleontoh)gy, l~%iversity q/'Mich~an, Ann Arbor, M148109-1079, USA
(Received March 8. 1993: revised and accepted July ,31 1993)
Abstract
One hectare of undisturbed Amazonian forest, containing about 571 species of trees larger than l0 cm diameter at
breast height, was studied to determine the relationship between high-richness forest and the autochthonous litter
produced by the forest. Litter samples contained up to 52 species, of which one-third represented epiphytes, vines,
and lianas.
These modern leaf litter studies from southeast Amazonian Peru indicate that reconstructions of ancient high-
diversity forests are possible using autochthonous leaf litter deposits. In comparison to temperate litter samples,
however, more sampling must be done to recreate fairly simple descriptors of ancient communities such as species
richness and heterogeneity. Samples must be large, relatively closely spaced, and maintained as distinct collecting
localities to retrieve the maximum amount of data from rich, angiosperm-dominated localities. There are many
advantages justifying more intensive collections. For example, biomass contribution of major life-form categories in
the source forest is reflected in leaf litter accumulating under tropical forest canopies. Tropical forests, because of
their extreme heterogeneity, also can provide the opportunity to reconstruct individual species characteristics from
litter signatures. The relative rarity of most species creates distinct leaf shadows from which the canopy breadth and
volume of many individuals can be estimated. The principles derived from modern tropical litter studies can be
applied to existing fossil collections; however, their power lies with those collections originating from autochthonous
assemblages, for which spatial control during collecting has been maintained, and time averaging has been kept to a
minimum. These reflections of community structure available from the leaf litter provide a means for paleobiologists
to contribute significantly to the study of community evolution and stability.
1. Introduction et al., 1993; Burnham, in press). The characteristics
of such leaf deposits are:
(1) they are accumulations of plant litter that
There are certain types of fossil plant deposits
have not been transported far from their point of
for which indications of fine-scale ecological data
abscission from the source plants (autochthonous
can be retrieved. These are exceptional deposits, or parautochthonous);
but not uncommon in the fossil record. (2) they show fine detail in shape and venation,
Angiosperm-dominated deposits of this type have such that identification of a large proportion of
been recognized throughout western North the leaves is possible;
America and range in age from Late Cretaceous (3) they are exposed over large areas, such that
through to the Recent (Cross and Taggert, 1982: lateral variation in the composition of the bed can
Wing, 1984; Boy& 1985; Johnson, 1992; Wing be determined: and
0034-6667/94/$7.00 ~ 1994 Elsevier Science B.V. All rights reserved
SSDI )034-6667( 94)E0086-K
001 L~.R "nei,~eR/mahnruB q/Palaeobotan), dna r.g~kon'llaP 18 (1994) 99 311
(4) they show indications that sedimentation was It is the purpose of this paper to present investi-
rapid, with complete burial of the plant litter in a gations on modern leaf litter accumulations in a
single event or a few, closely spaced, event-deposits. moist tropical forest for the purpose of extracting
The coincidence of all these features in a fossil guidelines for the collection and analysis of high-
plant deposit indicates that the potential for extrac- diversity fossil plant deposits. These guidelines
tion of high-quality ecological data is possible. should accommodate both the systematic and the
These same deposits offer some of the best possi- ecological interests of paleobotanists. Analogous
bilities for complete systematic studies. Although interpretations of ancient litter accumulations can
the ecological importance of plant "lagerst/itten" be drawn safely only to those deposits that have
is appreciated by the majority of paleobotanists, the characteristics outlined above.
systematic concerns commonly take the front seat
in collection efforts whereas the taphonomic and
ecologic context of the plants may be given only 2. Site description dna methods
a cursory analysis, if any, in the ensuing mono-
graphic treatment. There has not been, until 2.1 Site descriptio,
recently, much attention paid to the methods by
which fossil plants should be collected and ana- Research was carried out at the Pakitsa Guard
lyzed if the highest-quality ecological data are to Station of the Parque Nacional del Manu in the
be extracted, Department of Madre de Dios, Peru (Fig. 1 ). The
Sampling methods for fossil assemblages have Pakitsa site (11°54'S, 71:22'W) is located at
been investigated directly using a variety of census- approximately 340 m elevation on the Rio Manu,
ing methods (reviewed by Scott, 1977; Spicer, a white water river that drains the Peruvian high-
1988 ), but the meaning of quantitative counts (e.g. lands from an elevation of 2000 m and joins the
Spicer and Hill, 1979) has been open to various Madre de Dios at 320 m elevation. At Pakitsa, the
interpretations. Innovations in fossil plant sam- Rio Manu is a meandering river with a high
pling for ecological analysis must follow the guide- suspended load. Cut-off meander lakes are
lines derived from studies in modern sedimentary common, locally known as eochas. Mean annual
environments on the formation of the assemblages. temperature at Cocha Cashu Biological Station,
Currently several researchers are approaching the
processes of plant fossilization in modern environ-
ments from a variety of angles (Burnham, 1989;
Gastaldo, 1989, 1992; Burnham et al., 1992;
Greenwood, 1992). The present research is focused
on one of the first steps in the fossilization process,
that of primary litter accumulation, for two ~J
reasons. First, because most plant parts enter the , .c)--_~
fossil record through this first step, a thorough
understanding of the process is essential to any
plant taphonomic interpretation. Second, in-place
litter accumulations are not uncommon in the
fossil record and these provide some of the best
possible evidence on original community structure.
iR 0
Therefore, a complete understanding of the way
in which the source forest is reflected in the pri-
mary, untransported litter is essential. It is through Madre de Dios "-..~"" w;9G
these attempts at understanding the individual
.giF .1 Location of department of Madre ed Dios, Peril )tesni(
steps in the fossilization process that the means dna location of study area at Pakitsa near eht oiR Madre ed
for efficiently sampling fossil outcrops will emerge. .soiD ehT Parque lanoicaN led Manu si detacidni yb yvaw .senil
.R ,J Burnham~Review q Palaeobotany and Palvnoloey 18 (1994) 99 311
101
located 61 km to the northwest, is 24.1'C, with were made by Dr. Robin B. Foster, Maps were
temperature excursions reaching lows of 15C and verified by the author during each of three subse-
highs of 32"C possible in almost every month of quent field seasons, with minor modifications made
the year (Kalliola et al., 1987). Annual precipita- to the original maps. Thus, for the forest to which
tion at Cocha Cashu is 2080 ram, which is distrib- the litter is compared, a high-quality description
uted primarily during the months of November is available for all canopy and subcanopy tree
through May. Humidity is generally high, During species. This well-characterized forest hectare is
the dry months it is not unusual to receive less referred to hereafter as the source forest. To aid
than 01 cm of rain, although yearly rainfall pat- litter identifications, samples of leaves from each
terns are extremely variable (Terborgh, 1990). species were collected from identified trees on the
In general, the forests at Pakitsa and along the hectare as well as naturally dried leaves from forest
Rio Manu are heterogeneous in aspect, due to litter of each species.
varying water tables, soil types and forest ages
(Terborgh, 1983). This study is centered on a 2.3 Litter collection and quant(fication
single hectare of forest (known as Zone 2) which
supports about 571 tree species larger than 10 cm Litter was collected from the forest floor in 31
diameter at breast height (dbh), but it is notable 0.5 x 0.5 m sample quadrats during October 1990.
that a second hectare, only 2 km away shares only Litter sampling was carried out only in the central
34 (20%) tree species. The hectare of forest 40 x 40 m of the hectare, leaving a mapped buffer
reported here occupies a relatively low, cut-bank zone of 30 m around the outside of the sampling
position along the Rio Manu and can be classified area (Fig. 2). This ensured that very few individuals
as a moist tropical forest originating on a rich could be represented in the litter samples that were
alluvial soil. Canopy dominants belong to the not accounted for in the source forest. All identifi-
Bombacaceae, Leguminosae, Moraceae, Tiliaceae, able litter fragments were collected from the forest
Sapotaceae, and Rubiaceae. The forest is com- floor, including identifiable leaf rachises, fruits and
posed of only a few tree species that remain leafless seeds. Only the leaf fraction of the litter is used
for longer than 2 months. Many of the trees lose here for comparison with the standing forest. Litter
their leaves for a short period of time (2 4 weeks) was dried first in the field in mesh net bags and
during the dry season, and others (classically then returned to the USA for identification, com-
termed "evergreens") simultaneously produce new plete drying and weighing. Foliage material was
leaves and shed old ones during the dry season. separated to species, counted, dried to a constant
The forest contains at least 6 species of palms with weight at 70~C, and each species weighed indepen-
a dbh larger than 01 cm. Three species of palms
are very abundant, making the family the most
abundant in terms of stem number on the
mapped hectare.
j __ LITTER
2.2 Forest mapping 60 SAMPLE
i 2O METER
~ i ~RADIUS
The forest hectare reported upon here was 40
mapped by the BIOLAT (Smithsonian Institution) ~
40 METER
team working at Pakitsa during September 20 --RADIUS
October 1988. All trees larger than 01 cm dbh were
measured, identified and mapped. Lianas were 0 0 20 40 60 80 100
METERS
included only if they met the diameter limit. Data
.giF .2 Litter sample noitcelloc sites ni Zone 2 hectare, Pakitsa
were compiled into map form with identities of
Guard Station, Manu National Park, lireP ( × marks 5.0 × 5.0
each tree, diameter at breast height and exact m litter collection sites). Circles show owt examples of forest
position of each tree plotted. Tree identifications areas compared to litter .selpmas
201 R.J. Burnham~Review q Palaeobotany and Palvnology 81 ( 19942 99 311
demly. Unidentifiable leaves were given a morpho- Using the most simple comparison, that of single
type letter and an example of each morphotype was samples to the source forest, an average of 11.4%
saved for later identification. The "unknown" taxa of the source forest tree species are found in a
are not included in any of the considerations pre- single sample (range 8--15%). This can be com-
sented here, unless explicitly stated. pared to the results reported by Burnham (1993)
on temperate forests in which single litter samples
2.4 Comparative methods contained an average "fo about 70% of the species
from source forests. Because these values are
The leaf litter and source forest were compared derived from the comparison of only 0.25 m 2 of
using both qualitative (presence-absence lists) and litter to a hectare of diverse source forest, a more
quantitative (biomass) comparisons. Presence- appropriate comparison, particularly in tropical
absence comparisons use species lists from litter forests, is made between a single sample and the
and source forest and only identified litter taxa are source species within a limited radius of source
considered here (Appendix 1). Comparisons of forest. The radius that maximizes the proportion
species numbers in three habit classes use all taxa of litter species accounted for in the source forest
for which habit class can be reasonably inferred, and vice versa is a radius of 12.5 m. The average
regardless of the ability to identify the leaf. Litter proportion of species in a 12.5 m radius source
distribution of individual species are compared forest represented in single litter samples is
using leaf numbers, which is appropriate for single- (40.3%). This can be compared to the results
species comparisons. reported by Burnham (1993) on temperate forests
Rank-order comparisons were made on the basis in which single litter samples contained 80 90%
of proportion stem basal area and proportion litter of the species in a 51 m forest radius.
mass. These two attributes were chosen because ( 1 ) An even more reasonable comparison of pres-
basal area more closely approximates species ence absence lists is between several litter samples
importance than does stem number in high- and a specified area of source forest. Two such
diversity forests, and (2) litter mass circumvents comparisons are presented in Fig. 3, in which
the problem of numerical over-representation of combined species lists from the five central samples
compound leaves by leaflets, created by leaf number were compared to forest areas with (a) 51 and (b)
comparisons. The best proxy for leaf mass in fossil 20 m radii. Approximately 70% of the species
assemblages is leaf area (Burnham et al., 1992) present in the source forest are represented in the
which is most accurately calculated by counting litter species lists from five combined samples.
number of leaves or leaflets and multiplying by leaf Further, if all thirteen samples derived from the
area of a complete specimen (leaf or leaflet). central 40 x 40 m area are combined and compared
to the entire surrounding forest hectare, the pro-
portion of source species represented in the litter
3. Results samples is 43% (Fig. 3c).
In all these various areas of source forest, it is
.3 l Species presence absence comparisons significant to note that 80 90% of the woody
biomass (as defined by cross-sectional area of tree
The average number of leaves or leaflets counted trunks) is accounted for by the species repre-
per sample was 426 (range 236 671). Average sented in the corresponding list for litter samples
number of taxa identifiable to species was 25 per (Fig. 3). Thus, the ecological dominants are well
sample (range 20- 33). An average of 01 additional, represented.
distinct taxa that were not assignable to known
species was also found in each sample. Familiarity 3.2 Missing taxa
with the forest and evidence from leaf morphology
indicate that most of these "unknowns" are Missing taxa are informative because these are
canopy vines. the taxa that would be "unseen" or misinterpreted
R.J. weiveR~mahnruB ynatnhnwkaP'fo dna yeolonylaP 8l ~ )4991 99 311 301
FIVE LITTER SAMPLES COMPARED TO :
a b
SURROUNDING 15 M RADIUS SURROUNDING 20 M RADIUS
RETTIL ~ O R E S T
S 1 .PPS 8 .PP5 1s .PPs
SPECIES IN COMMON = 12 (72% OF FOREST SPP.) SPECIES IN COMMON = 27 (6g% OF FOREST SPP.)
REPRESENT 92.2% OF STEM BIOMASS REPRESENT 89.3% OF STEM BIOMASS
c 13 LITTER SAMPLES COMPARED TO
SURROUNDING FOREST HECTARE
SPECIES IN COMMON = 76 (43% OF FOREST SPP.)
REPRESENT 81.2% OF STEM BIOMASS
Fig. .3 Presence absence comparisons between litter and source forest species lists. Horizontal bars represent species l\mnd ni litter
but not in forest area. Vertical bars represent species found ni forest area but not ni litter samples. (a) Combined litter species list
morF 5 samples in central 01 × 01 m area compared to a 51 m radius source forest. (b) Combined litter species list from 5 samples
ni Rartnec 01 × 01 m area compared to a 02 m radius source forest. (c) Combined litter species list from 31 samples ni central 04 × 04
m area compared to a one hectare source forest.
if found as a single isolated leaf in a fossil recon- radius forest. Only one species has a dbh over
struction. The patterns of missing taxa are similar 20 cm. Of the group of twelve species, half have
in all cases studied. They will be illustrated here very large leaves (macrophylls, e.g. Guslavia hexa-
by using the comparison of 5 combined litter petalu, Coccoloha h'hmanii, Sloanea ./)'agrans).
samples to a 20 m radius source forest (Fig. 3b). Although it is possible to find leaves of these
The taxa that are present in the litter samples but species underneath the canopies of the trees, they
not represented in the 20 m radii of source forest are few in number. The production of few large
(Fig. 3b, horizontal lines) belong primarily to two leaves predisposes a tree to underrepresentation in
categories. First, 41% of the "missing" source the litter and may severely bias climatic reconstruc-
species are very large trees outside the 20 m radius tion based on foliar physiognomy. The other half
forest. Second, 25% of the "missing" source species of this group have leaves of mesophyll size and
are identifiable canopy vines and lianas (par- their underrepresentation can only be attributed
ticularly members of the Bignoniaceae and to a small canopy volume.
Menispermaceae). Small trees, herbs and epiphytes
make up about 9% each of the litter species 3.3 L(fe-form representation
"missing" from the source forest list. The propor-
tion of canopy vines/lianas is clearly underrepre- Setting a lower limit to tree diameter mapped
sented by these statistics which do not include in the forest has the potential to cause errors in
unidentified taxa, many of which are also the comparisons of litter and forest because the
vines/lianas. smaller-diameter species below the mapped size
On the other hand, there are trees in the desig- limits (i.e. < 10 cm dbh) may be represented in the
nated 20 m radius not encountered in the litter litter but not in the source forest. Although this is
(Fig. 3b, vertical lines). The twelve source tree a possibility, it appears to have little influence on
species not represented in the litter are all small these data. Instead, the majority of the species that
individuals of rare species. Only 3 species are are found in litter samples but not accounted for
represented by more than one tree in this 20 m by the source forest are one of two classes: either
104 R.J. Burnham~Review /~{ Palaeobotany and Palynology 18 (1994) 99 311
large trees beyond the forest area sampled or contribute 11% of the stem basal area of the
vines/lianas. The low biomass produced by small hectare. There is no doubt that they are quite
dbh trees has a proportionately minor influence important to the ecology of the hectare.
on the litter. Interestingly, however, palms are represented in
Fig. 4 shows the proportion of species in the almost every litter sample by only a very small
litter accounted for by each of three categories: amount of degraded foliage material (usually one
trees, vines/lianas, and herbs/epiphytes. These tab- or rarely two species), which represents only 3%
ulations include all "unknowns" to which life form of the combined biomass of all samples. These
could be assigned with a reasonable degree of observations parallel the observations on palms
certainty. Between 31 and 32% of the species of made by Gastaldo and Huc (1992) and Gastaldo
each sample are vines/lianas. It is important to (1992) from the Mahakam Delta.
note that although there is a high diversity of
epiphytes, herbs, and small trees in the forest at
Pakitsa, their total biomass is low (as gauged by 3,4 Individual species litter shadows
cross-sectional area and by photosynthetic area)
and these life-forms are virtually unrepresented in Litter shadows of trees of four species are shown
the litter. This indicates that the litter is more a in Fig. 5. Leaves were counted in a 0.5 x0.5 m
reflection of the species and life forms that play area at every meter along transects made in three
the largest role in total photosynthetic area, than radial directions away from the trunk of each tree.
it is a list of all species present in the area. Total The species illustrated were chosen because of their
numbers of these life forms has not been deter- relative (or absolute) rarity on the hectare and the
mined in the source forest, but Foster (1990) range of stem diameters encompassed by the group
reported the following proportions for life forms of species. Canopy edge and estimated canopy
in combined Manu floodplain forest floras: trees, height of each tree is noted.
28%; shrubs, 28%; lianas, 18°/,,; herbs, 14%; epi- There is a significant drop in abundance of
phytes, 12%. leaves at the canopy edge in all cases. This indicates
A significant deviation from the generalization two things: (1) leaves are falling, for the most
made above is in the proportion of palms that are part, out of the canopy directly to the forest floor
represented in the litter. Palms are represented by and experiencing little post-abscission redistribu-
6 tree-sized species in the source hectare. They are tion on the forest floor, and (2) for isolated
the dominant family by number of stems and individuals of a species, the canopy edge may be
discernible from autochthonous litter deposits.
HABIT SUMMARY OF SPECIES IN LEAF LITTER Note that in all cases, some leaves are distributed
MANU NATIONAL PARK, PAKITSA, PERU beyond the canopy edge.
50. Although the measurement of individual canopy
height in the source forest was purely by estimation
40
~, 35 and is likely to be inaccurate particularly for the
30
taller trees, there is a strong suggestion that the
25
O
canopy height and the furthest absolute distance
leaves are distributed from the trunk of the tree
m
are related linearly. Leaves tend to be distributed
0. I 2 3 4 5 6 7 8 9 10 11 12 13 to a maximum distance roughly equal to the
SAMPLE NUMBER
canopy height. The usefulness of this relationship
~ERT
x~z~BS/EPXPmrn~s L~.~ is limited to those fossil assemblages in which
no traumatic or post-abscission transport has
.giF .4 Proportion of litter seiceps represented ni three -efil
form .seirogetac Litter seiceps edulcni lla taxa recovered from occurred. Confirmation of the results of the
litter selpmas for which life form si known with elbanosaer shadow distributions on additional species from
.ytniatrec Cocha Cashu Biological Station is underway but
R.J. Burnham~Review o/Palaeobotany and Pulvnoh)gy 81 (1994) 99 311 105
I
I.
140 . Pouteria nitida 111 Lunania parvifona
z2o~ A dbb - l.Om
_ _ ~- dbh = 0.1m
I00 . i L ~ GE //~ YPONAC
5~
60 4-
\
I° + _I +
0 0 .... , 7-', 0
0 5 I0 IS 20 25 30 35 40 2 4 6 8 I0 12 14 16
d
l c" Poulsenia armata I so I ~izyphus cinnamomum
100
o
to I
m
. . . . . . . . , 0 , ,
0 5 I0 15 20 25 30 35 0 5 I0 15 20 25 30
DISTANCE FROM TRUNK OF TREE (M)
Fig. .5 Litter shadows for individual trees of four species. Shadows were made by counting lla leaves of the target species ni a
5.0 × 5.0 m area at 1 m intervals along three transects which originated at the tree bole. Tree bole position si at 0 ni lla cases. Tree
canopy extent si indicated. Approximate tree heights are airetuoP ,aditin 04 :m ainanuL ,ailb/(vrap 5.6 m: ainelsuoP armata, 04 ;m
Zizyphus +mumomannic 21 .m Boxed numbers indicate the three compass directions per tree along which leaves were counted on the
three transects.
preliminary results strongly support the patterns of the same individual of Poulsenia armata is
presented in Fig. 5. shown in both figures (Figs. 5c and 6c). In Fig. 6c,
The preceding litter shadows were generated by the dramatic drop in leaf abundance at the canopy
a very intensive sampling regime (1 m intervals) edge is not plotted because abundances are inter-
focused on a single individual tree in each case. If polated between the thirteen litter collection points
reconstructions are to be made of many trees of of the hectare (i.e. sampling is not dense enough
many species from fossil samples, it is more likely to pick up the drop). Note particularly that Fig. 6c
that generalized sampling will be carried out in a interpolates 25 30 leaves at a distance of 20 m
manner similar to the distribution of the 13 0.25 from the trunk. In fact, as Fig. 5c shows, there are
m 2 samples made from litter accumulations for at most 12 leaves at a distance of 20 m from the
this study. Thus, four tree species were targeted trunk. This indicates that the density of sampling
from those 31 litter samples and the distribution of fossils must be on the appropriate scale to pick
of leaves of these species is shown in Fig. 6. up litter abundance drops.
Isographic contours are shown, created by interpo- If fossil sampling can be extended at least to a
lating betweeen the thirteen sampling sites within 40 m transect, finer resolution of the canopy
the central 40 x 40 m area. Canopy breadth and attributes can be made. To correlate the litter
trunk diameter of the trees are indicated. Only pattern with the exact position of the trunk of
leaves of large canopy trees (30-40 m in height) the tree, it is necessary to bisect the canopy
were selected for this figure because the number shadow completely (note particularly Fig. 6d for
of leaves from any small tree is very low except Spondias mombin). To estimate canopy breadth,
right under the tree canopy. the largest drop in abundance of leaves can be
Attributes of canopy size and shape can be read directly from the litter pattern and would
inferred from litter patterns from the samples, but correlate with the edge of the canopy. To estimate
these inferences are made stronger by knowledge approximate canopy height, the maximum distri-
of the results from the finer-scale sampling on the bution of leaves must be known. Overlapping
transects shown in Fig. 5. For example, the shadow canopies of conspecifics will eliminate the possi-
601 R.J. weiveR~mahnruB o/'Palaeobotanv dna y~o/onvlaP 1~, (1994) 99 311

so b.
I00
°° 1
so '~ m ~h
BIRTSID 1 i I
40-
20- - ~ "
Copa~fera reticulata Dipteryx alata
C. d.
001
A
.44 m dbh • .15 m dbh
so- .38 m dbh
~r O.IS m dbh
o ~~ LITTER
NOITUBIRTSID
° ~ ~ NOI~I~IRLgW
oo- .e"
-O4
f' mgh "" ~°kkk~ 04
1,5 i
I
i I
~,,% s .51 m dbh
.22mdbh .89 m dbh
02 ~ oo,
Poulsenia armata Spondias mombin
CANOPY EXTENT 40 x 40 m AREA FROM TREE OF NAMED SPECIES
WHICH --- 13 LITTER SHOWING DIAMETER AT BREAST
COLLECTIONS WERE MADE HEIGHT (DBH)
, ,
Fig. .6 Litter abundance ni 31 litter samples for 4 targeted species. Number of leaves per sample si shown as lsographic contour
Tree bole position(s) are shown as well as canopy extent.
bility of extracting such autecological data from an area of twice that of the estimated canop5
the fossil record. height.
Autecological data can be retrieved from diverse
fossil assemblages if the sampling regime is suffi- 3.5 Rank-order representation oibrest dominants
ciently dense and carried out over a sufficiently in litter
large area. A rule of thumb indicated for individual
attributes would be that sampling density should Rank-order comparisons of source forest and
be on the order of one sample at least every 7 m, litter were made over a wide variety of combined
and that the sampling should be carried out over samples and forest areas. A trade-off exists in
R.J. Burnham~Review o/' Palaeobotany and Palvnology 18 (1994) 99 311 701
selection of a source forest area between ( 1 ) includ- disappointing when compared to similar results of
ing all tree taxa represented in litter, and (2) a rank-order comparison in a temperate litter study
extending the source area far enough to accommo- (Burnham et al., 1992). However, it is important
date the first parameter such that trees, unrepre- to recognize that 12 of the 25 most important
sented in litter, also are included. The rank-order source forest taxa are at least represented in the
comparison that maximizes the similarity between litter. Thus, there is confirmation of the results
dried litter mass representation and forest stem from the presence-absence data that the most
diameter representation is a comparison between important taxa are present in the litter. Of the five
source forest with radius of 40 m and all 31 litter most important taxa in the forest, four are repre-
samples (determined by Spearman rank-order cor- sented among the l0 most important taxa in the
relation values). This includes a mapped forest litter, and this one exception is a palm, which
area beyond the litter sample area of a minimum are poorly represented in general (see above).
of 21 m (see Fig. 2). Similarly, of the five most important litter taxa,
The 25 most important taxa from all 31 litter all are among the ten most important taxa in the
samples and the 25 most important taxa from the source forest. Thus, although a direct reading of
source forest in a 40 m radius have been ranked taxon importance cannot be made from litter mass,
in order of abundance in Fig. 7. The rank-order general statements about dominance in the source
importance of forest taxa is based on their propor- forest are recoverable from quantitative studies.
tion of total stem biomass as measured in cross-
sectional diameter. Rank-order importance of litter
taxa is determined on the basis of their proportion 4. Discussion
of total leaf litter mass (dried to 70°C). Taxa are
listed in Fig. 7 in order of litter abundance to From the results of the litter source forest com-
facilitate comparison of species rank order between parisons presented here it is clear that there is not
forest and litter. a single answer to the question, "does tropical leaf
The rank of source forest and litter species do litter reflect its source forest?". Instead the question
not follow the same order and the results are must be decomposed first into autecological vs.
( 18
11114~ 1 1 1 ill
lllJl u~r~x
0 2~ IglS"~ MASS
a lmuuJi I
2 lllII-.o.
• ,~z~__ ~_
,otllil l Illlll
B~AL ~
o IIIII lilt iL.,
llllHl .ll ill i il ,..h ....
oii,,Ii, !11 . I.-I_lIIiiIil,i',t,,
SPECIES ACRONYMS
Pig. .7 Rank-order of the 25 most important litter and tree species as determined from litter mass and basal area, respectively. Both
graphs are ordered here by rank order of mass of litter species. All 31 litter samples are used: source forest radius 40 m.
801 R.J. weiveR~mahnruB ~ ynatoboealaP dna yg)honylaP 8l (1994) 99 311
community questions. Autecological attributes quality data on the canopy extent, as read from
include habit, canopy characteristics and distribu- abundance drops. In addition, a complete transect
tions, while community attributes include species starting at the trunk position of a fossil tree and
richness, heterogeneity, composition of life forms, extending beyond the distribution of leaves of that
and absolute or rank-order importance. For each species will retrieve a rough estimate of canopy
of these attributes, the data presented here provide height. These attributes can be derived for ever-
an indication of the trust that we can place in the green or deciduous trees, providing that diversity
reconstructions based on autochthonous litter. is high and conspecific canopy overlap is minimal.
4.1 Autecological attributes 4.2 Community attributes
Plant habit may be one of the most difficult Total species richness clearly is an attribute that
attributes to reconstruct using litter patterns only. almost every paleobotanical floristic study tries to
This is because plant habit alone does not provide reconstruct. Indications from this study hectare
a unique litter distribution pattern. For example,
(Burnham, 1993) and previous tropical and tem-
the litter shadow of canopy lianas may have a
perate forest research (Greenwood, 1991; Burnham
similar form to the litter shadow of subcanopy
et al., 1992) are that rules for reconstruction
tree species (~20 cm dbh) for which individuals
among forest types are not equivalent. Once it is
are distributed patchily. In both cases a diffuse
determined that a heterogeneous high richness
distribution of leaves would be expected over the
forest is under study, the reconstruction of richness
sampling area because of the moderate number of
will likely be a simple multiplicative factor, depend-
leaves produced at any one place. Under a dense
ing on the number of samples made. Based on this
sampling regime, the tree should exhibit higher
research, I recommend combining species lists from
local abundances of leaves, that reflect actual
5 samples of 350 500 leaves placed at intervals of
crown positions and could be used to distinguish
51 m and multiplying by 3 3.5 to reconstruct
tree from liana. However, local abundances in
hectare-based species richness. The richness of a
leaves of liana species may be expected as well.
forest area of 20 m radius is roughly equal to the
Litter distribution, in combination with at least
richness of these five combined samples. It is
some knowledge of morphology and systematic
important to note that although species richness is
affinities, is most likely to reconstruct plant habit
roughly equal in this last case, the actual species
from the fossil record.
represented in the litter may not be those that are
Reconstruction of plant distribution is slightly
canopy trees larger than 10cm dbh (i.e. some
more encouraging than plant habit. If individuals
lianas may replace smaller trees in the list). These
of species are isolated, sampling over large areas
correction factors assume canopy structure compa-
will allow reconstruction of crown positions.
rable to that of today.
However, plant size will affect the interpretation
Forest heterogeneity also is reflected in forest
since a cluster of subcanopy conspecifics may have
a litter shadow very similar in form to a large litter. If heterogeneity is defined as the variability
isolated individual of a canopy tree. In this case in the spatial distribution of species on a hectare,
we should expect absolute abundance (or mass) of it should be one of the easiest attributes to
the canopy individual to be higher than the subca- reconstruct from litter because sample to sample
nopy group because foliage area represented in the heterogeneity is very reflective of the variability in
litter dramatically increases once a tree is > 50 cm the spatial distribution of species. Note that in a
dbh (pers. observation). temperate forest, 2 samples shared 85 90% of their
Canopy characteristics of individual tree species species whereas in this tropical Peruvian hectare,
are the attributes that seem to have the greatest samples share 30-55% of their species (Burnham,
potential for reconstruction from quantitative litter 1993). Two contiguous 20 x 20 m source forest
studies of tropical ecosystems. Sampling along plots from the Peruvian hectare share only 5-6
transects or over areal exposures can provide high- source species, a similarity of only 30%.