This item is only available as the following downloads:
REGENERATION OF COMMERCIAL TREE SPECIES FOLLOWING CONTROLLED
BURNS IN A TROPICAL DRY FOREST IN EASTERN BOLIVIA
DEBORAH K. KENNARD
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Dedicated to my parents,
Margaret Kennard and Robert Kennard,
and to my husband,
I sincerely thank everyone who contributed in some way to this dissertation. First, I
would like to thank the members of my committee. Jack Putz, my advisor throughout my years at
the University of Florida, has been very influential in my interest in forest ecology and
management. I thank him for the opportunities he has provided, his time as a teacher and mentor,
and his diligence as an editor. His enthusiasm for ecology will influence me for years to come.
Henry Gholz was a tremendous help with the project design and generously provided the use of
his lab for resin analyses. Kaoru Kitajima patiently helped me with statistics and was instrumental
in the development of the seedling chapter. Insightful suggestions by Earl Stone aided me
throughout every aspect of the project; I was repeatedly inspired by his immense knowledge and
experience. Finally, I would like to thank George Tanner who provided much-needed
encouragement at critical times.
This dissertation would not have been possible without the generous financial and
logistical support of BOLFOR (Proyecto de Manejo Forestal Sostenible). Several members of
BOLFOR were crucial in its completion. First, I would like to thank Todd Fredericksen, the
Forest Ecologist at BOLFOR, for his help conducting the controlled burns, statistical and technical
advice, ready supply of coca leaves, and humor. T. Fredericksen, J. Nittler, and W. Cordero
provided administrative support in Bolivia. I gratefully acknowledge the following people who
generously volunteered their time to assist with fieldwork: J. McDaniel, L. MacDonald, J.
Chuviru, T. Fredricksen, N. Fredricksen, J. Lincona, J. Justiniano, A. Ademar, K. Gould, F.
Fatima, K. Hueberger, B. Flores, M. Toledo, L. Anderson, B. Mostacedo, and the Aberdeen
students. I wish to acknowledge T. Killeen and the herbarium at the Museo de Noel Kempf
Mercado for use of the data collected in the 1995 inventory of Las Trancas. M. Toledo kindly
assisted with plant identification. My husband and I are very grateful to Todd and Nell
Fredericksen for their hospitality in Santa Cruz.
Numerous Chiquitano community members assisted throughout the 18 months of this
project; I will be forever grateful for and impressed by their hard labors. In particular, I would
like to thank Don Juan "Loco" Pesoa for being instrumental in the installation of the treatments;
Don Juan Faldin for sharing his knowledge of plants and their local uses; and, Don Lucas
Salvatierra for his assistance in locating and measuring abandoned agricultural fields. I would
also like to thank the many members of San Lorenzo who welcomed me into their community
during my time away from El Campamento de Las Trancas.
At the University of Florida, I would like to thank J. Bartos at the Analytical Research
Lab for analysis of soil samples. In the lab of H. Gholz, I would like to thank D. Noletti and K.
Clark for their valuable assistance with the resin extractions. I was supported by a teaching
assistantship offered by the Department of Botany and the Department of Biological Sciences.
Comments of several friends greatly improved the final draft: G. Blate, K. Gould, B. Ostertag, T.
Fredericksen, and J. McDaniel. Most importantly, I would like to thank my traveling companion,
field-assistant, Spanish teacher, motorcycle driver, toilet-digger, wasp-magnet, translator, and
husband, Josh McDaniel.
TABLE OF CONTENTS
A B S T R A C T ............................................................................................................................. vii
Introdu action ....................................................................................................................... 1
Conservation and Management of Tropical Dry Forests................................................... 2
Management of Tropical Dry Forests in Eastern Bolivia..................................................... 3
The Role of Historic Disturbance Regimes in Forest Management................................... 4
The Potential of Prescribed Burning in the Management of Bolivian Dry Forests................. 7
Scope of D issertation ....................................................................................................... 8
2 STUDY SITE AND TREATMENT DESCRIPTIONS
In trod auction ..................................................................................................................... 11
S tu d y S ite .............................................................................................................. ......... 13
M methods ......................... ....... ................ ......... ... ............... ..... ... 18
Initial T reatm ent R results ................................................................................................ 24
3 EFFECTS OF CANOPY GAP FORMATION, PLANT REMOVAL, AND CONTROLLED
BURNS OF HIGH AND LOW INTENSITIES ON SOIL CHEMICAL AND
Introdu action ............................................................................................ ........................ 2 9
M eth od s ..................................................................... .................................................... 3 1
R e su lts ....... ....................................................... .............................................................. 3 8
D discussion ................................................................................................ .................... 52
C o n clu sio n s ....................................................... ............................................................. 7 1
4 EFFECTS OF CANOPY GAP FORMATION, PLANT REMOVAL, AND CONTROLLED
BURNS OF HIGH AND LOW INTENSITIES ON EARLY REGENERATION OF
COMMERCIAL TREE SPECIES
Introduction ................................................................................................. ................. 72
M eth od s ....................................................... .................................................................. 7 5
Results ....................................................................................... ..................................... 79
Discussion ........................................................................................................................ 96
Im plications for m anagement................................................................................ .......... 111
5 EFFECTS OF CANOPY GAP FORMATION, PLANT REMOVAL, AND CONTROLLED
BURNS OF HIGH AND LOW INTENSITIES ON A DRY FOREST PLANT
Introduction ................................................................................................................... 113
M methods ........................................................................................................................ 115
Results ......................................................................................................................... 119
Discussion ..................................................................................................................... 132
Conclusions ................................................................................................................... 147
6 COMMERCIAL TREE SPECIES REGENERATION FOLLOWING AGRICULTURAL
ABANDONMENT IN BOLIVIAN DRY FORESTS
Introduction ................................................................................................................... 149
M methods ........................................................................................................................ 150
Results ......................................................................................................................... 153
Discussion ................................................................................. .................................... 165
Implications: Management potential of secondary forest in Lomerio............................... 172
7 SUMMARY AND CONCLUSIONS
Sum m ary of Study Results............................................................................................. 174
Im plications for M anagem ent......................................................................................... 175
APPEN DIX ............................................................................................................................ 181
REFEREN CES....................................................................................................................... 187
BIOG RAPHICA L SKETCH ............................................................................ ..... ....... 206
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of Doctor of Philosophy
REGENERATION OF COMMERCIAL TREE SPECIES FOLLOWING CONTROLLED
BURNS IN A TROPICAL DRY FOREST IN EASTERN BOLIVIA
Deborah K. Kennard
Chairman: Francis E. Putz
Major Department: Botany
Low levels of disturbance associated with selective logging may be insufficient for the
establishment of many Bolivian dry forest timber species, the majority of which are shade-
intolerant. To examine the ecological potential of prescribed burning as a silvicultural tool, I
compared the effects of canopy opening, plant removal, and controlled bums of high and low
intensities on 1) soil properties; 2) establishment, growth, and survival of commercial tree species;
3) and, plant community structure and composition. To describe commercial tree regeneration over
longer time scales, I characterized tree population structures in abandoned slash-and-bum fields
ranging in age from 1-50 years, and compared these to a mature forest stand.
Both high- and low-intensity bums caused a dramatic but temporary increase in soil
nutrients. High-intensity bums altered several soil physical properties, whereas low-intensity bums
had little effect. Plant removal and canopy opening had little effect on soil chemical and physical
Three responses to gap treatments were observed among commercial tree species. 1)
Shade-intolerant species regenerating from seed were most successful following high-intensity
bums. 2) Shade-tolerant species were most abundant in treatments where survival of their
advanced regeneration was most likely (gap control and plant removal). Some of these species had
the ability to survive controlled bums by sprouting. 3) Individuals of root sprouting species were
most abundant following plant removal and low-intensity bums.
Sprouts dominated regeneration following canopy opening, plant removal, and low-
intensity bums. In contrast, seedlings dominated following high-intensity bums. High-intensity
bums shifted species composition relative to the less disturbed treatments.
Regeneration of shade-intolerant timber species was most abundant in young slash-and-
bum fallows. Similar tree population structures in older slash-and-burn fallows and the mature
forest stand suggests that the mature forest likely formed following a large-scale disturbance.
Although prescribed burning enhanced the regeneration of shade-intolerant timber species,
it is not likely to become a forest management tool in Bolivia in the near future due to economic
and political factors. Managing secondary forests in Bolivia would provide an alternative to
current attempts to regenerate these species after selective harvesting of mature forest.
Eastern Bolivia contains some of the largest and most diverse tracts of tropical dry forest
in Latin America. Natural forest management for timber, if profitable, is one means of
discouraging conversion of these forests to competing land uses. However, insufficient
regeneration of many commercial timber species presently poses an ecological barrier to sustained
timber yield, prompting forest managers to explore additional silvicultural methods to enhance
regeneration of these species. The low levels of disturbance associated with highly selective
logging may be insufficient for the establishment of many dry forest timber species, the majority of
which are shade-intolerant and likely require moderately intense disturbances for their
establishment. Fire, of both natural and anthropogenic origins, has likely been a pervasive
influence on tropical dry forests, and therefore, prescribed burning may be an effective silvicultural
tool to enhance regeneration of timber species following selective logging.
In this dissertation, I present the results of studies that examined commercial tree
regeneration following disturbances of various intensities in a dry forest in lowland Bolivia,
including harvesting gap formation, controlled burns of high and low intensity, and slash-and-bumrn
agriculture. My goal in carrying out these studies was to determine the regeneration requirements
of these commercial tree species, as well as to examine the effects of potential silvicultural
treatments on forest soils and community structure and composition.
Conservation and Management of Tropical Dry Forests
Tropical dry forests comprise approximately 42% of tropical forest land, more than either
moist or wet tropical forests (Murphy and Lugo 1986). Tropical dry forests also have supported
higher human population densities than wetter tropical forests for centuries (Murphy and Lugo
1986) and, as a result, have suffered more degradation and deforestation (Mooney et al. 1995,
Murphy and Lugo 1995). Efforts to slow conversion rates of dry tropical forest have been
negligible (Mooney et al. 1995). For example, in 1988, less than 2% of the original dry forest on
the Pacific coast of Central America remained intact and less than 0.1% had conservation status
(Janzen 1988). Consequently, tropical dry forests are considered by some ecologists as the most
threatened of the major tropical forest types (Janzen 1988).
Given the extensive use of tropical dry forests by rural people, their strict preservation
may not be a realistic conservation goal. As Johnson and Carbarle (1993) note, most developing
tropical countries rarely have the luxury of opting for forest preservation over forest exploitation.
Consequently, in most tropical countries, conversion of forested land continues to increase while
the establishment of protected areas remains low (FAO 1999). Consensus is emerging among
ecologists that protected areas, due to their small number and size, cannot effectively conserve the
majority of tropical species (Hansen et al. 1991, Heinrich 1995, Bawa and Seidler 1998).
Mounting concern over global declines of biodiversity has prompted many ecologists to look
outside of parks and nature preserves to semi-natural areas that may help maintain, or at least slow
the loss of, biodiversity (Sayer and Wegge 1992, Chazdon 1998).
Natural forest management, the sustainable production of timber from natural forest areas,
has been proposed as a means of maintaining forest value, thereby deterring land owners from
clearing forested land for other more profitable and destructive land uses (Poore et al 1989,
Johnson and Carbarle 1993, Maser 1994, but see Rice etal. 1997). Although definitions of
natural forest management vary, they usually encompass two ideas: first, a sustained yield of forest
products, and second, achieving this sustained yield through means that maintain other
environmental services, such as biodiversity, soil quality, and hydrology (reviewed by Johnson and
Carbarle 1993). By maintaining forests in a semi-natural state, natural forest management is
viewed by some as a critical means of maintaining biodiversity (Hansen et al. 1991, Sayer and
Wegge 1992, Frumhoff 1995, Heinrich 1995, Dickinson et al. 1996, Putz et al. in press, but see
Bawa and Siedler 1998), particularly in regions where forests are in danger of conversion. And,
despite the fact that few modem examples of economically viable natural forest management
projects exist (e.g., Panayotou and Ashton 1992, Johnson and Carbarle 1993, Rice et al. 1997,
Bawa and Siedler 1998, Bowles et al. 1998, but see Leslie 1987), the promotion of sustainable
forest management has become a mainstay in international strategies for the protection of tropical
forests (Bawa and Siedler 1998, Haworth and Cousell 1999). As Haworth and Cousell (1999: 62)
explain "this approach has often been justified on the grounds that it is the result of a difficult
choice between accepting, on the one hand, the inevitability of continued commercial logging of
natural forests, which will cause some damage to the ecosystem, or, on the other hand, facing the
complete loss of the forest to other causes."
Management of Tropical Dry Forests in Eastern Bolivia
The Chiquitania region in eastern Bolivia contains one of the largest and most diverse
tropical dry forests in the neotropics (Gentry 1993, Killeen et aL 1998). Although there is
currently 150,000 to 200,000 km2 of relatively intact forest in Chiquitania, Dinerstein et al. (1995)
identified this area as one of the most endangered ecosystems in the neotropics. Deforestation in
the alluvial soils near the city of Santa Cruz is in excess of 80,000 ha year' (Killeen et aL. 1998).
This conservation threat comes largely from large-scale industrial agriculture, but other economic
activities, such as cattle ranching, contribute to this rapid conversion of forested land. These
trends mimic past events in Argentina, Paraguay, and Eastern Brazil where similar dry forests have
been deforested and fragmented over the past two decades (Killeen et al. 1998).
Due to recent forest policy changes in Bolivia, natural forest management may now be a
practical means of controlling deforestation in Chiquitania. In 1996, a new forestry law was
passed that requires, among other features, management plans for all Bolivian forests (Nittler and
Nash 1999). Bolivian logging companies now operate with management plans on an estimated 5.7
million hectares of forest and a total of 660,000 hectares of Bolivian forest has been certified as
sustainably managed (Nittler and Nash 1999).
The Lomerio Community Forest, located in the center of Chiquitania, was the first
Bolivian forest to be certified. Its 60,800 hectares are owned and managed by 27 communities of
the Chiquitano indigenous people. The Chiquitanos have been managing their forests for 19 timber
species, 5 of which are classified as highly valuable (Table 1-1). Acquiring and maintaining
adequate regeneration of commercial tree species, a challenge faced by all natural forest managers,
is particularly apparent in Lomerio. For example, seedlings and saplings of 12 of the 19
commercial species are rare in forest understories (Table 1-1). A lack of seed sources due to
previous over-harvesting may account for the scarcity of regeneration among highly valued timber
species. However, poor regeneration plagues most of the tree species that have only recently been
harvested (Fredricksen 1999). Apparently, the current harvesting and silvicultural techniques
employed in Lomerio do not create conditions appropriate for the regeneration of these species. A
better understanding of the regeneration requirements of these commercial tree species is critical,
as continued regeneration failures will undoubtedly compromise the long-term sustainable
management of these forests.
The Role of Historic Disturbance Regimes in Forest Management
It is often assumed that forest management is more compatible with long-term
sustainability if timber harvesting and silvicultural techniques are designed to mimic historic
disturbance regimes (e.g., Pickett and White 1985, Oliver and Larson 1996, Attiwill 1994a,
1994b). Although this assumption has rarely been tested, ecologists argue that replacing harvested
Table 1-1. Table modified from Pinard et al. 1999 that reports characteristics of 19
commercial tree species of the dry forests of Lomerio. Species were matched to a
general silvicultural system (even- or uneven-aged) based on their regeneration
requirements. Timber value is based on market value in 1999.
Timber Managment Adult Sapling Shade
value system rarity rarity tolerance
1 Amburana cearensis high even 3 3 1
2 Anadenanthera colubrina low even 1 2 1
3 Aspidosperma cylindrocarpon low uneven 2 3 2
4 Aspidosperma rigidum low uneven 1 2 2
5 Astronium urundueva low even 2 3 1
6 Caesalpinia pluviosa low uneven 1 2 2
7 Cariniana estrellensis low uneven 2 2 3
8 Cedrelafissilis high even 3 3 1
9 Centrolobium microchaete high even 1 3 1
10 Copaifera chodatiana low uneven 2 3 3
11 Cordia alliodora high even 3 2 1
12 Hymenea courbaril low even 3 3 1
13 Machaerium scleroxylon high uneven 1 2 3
14 Phyllostylon rhamnoides low uneven 2 2 3
15 Platymiscium ulei low even 3 3 1
16 Schinopsis brasilensis low even 2 3 1
17 Spondias mombin low even 2 3 1
18 Tabebuia impetiginosa low even 1 3 1
19 Tabebuia serratifolia low even 3 3 1
Shade tolerance: 1 = high light only, large gaps; 2 = partial shade, small gaps;
3 = partial or full shade, understory.
Adult rarity (> 20 cm dbh): 1 common (> 5 ha-'); 2 = intermediate (1-5 ha'); 3 = rare (< 1 ha").
Sapling rarity (5-10 cm dbh): 1 common (>20 ha'); 2 = intermediate (5-20 ha'-); 3 = rare (<5 ha').
Even: even-aged management system with group selection.
Uneven: uneven-aged management system with single-tree selection.
trees without irreversibly damaging the residual forest is more likely to occur under conditions
similar to those that formed the original stand (Uhl et al. 1990). The selective cutting systems used
in many tropical forests are often justified on models of gap-phase regeneration in unharvested
forests (e.g., Whitmore 1989, Hartshorn 1989, Gomez-Pompa and Burley 1991).
Gap-phase regeneration, however, is not the most appropriate model for tropical dry
forests. Evidence suggests that single tree-fall gaps are smaller and less frequent in tropical dry
forests than in moist or wet forests (Dickinson 1998). Rather, very large gaps caused by
catastrophic disturbances more likely govern dry forest dynamics. In Central America for
example, mahogany (Swietenia macrophylla) has been noted to regenerate in even-aged stands
after hurricanes and fires (Lamb 1966, Snook 1996). The low level of disturbance created during
highly selective logging appears to be a poor replicate of this disturbance regime, and possibly for
this reason, natural regeneration of mahogany is scarce in most selectively logged areas (Verissimo
et al. 1995, Gullison et al. 1996, Whitman et al. 1997).
In contrast to Central America, the agents of large scale disturbance have not been a topic
of frequent study in Bolivian dry forests (but see Pinard and Huffminan 1997). As hurricanes are
absent in this landlocked country, it is likely that forest fires (both natural and anthropogenic) have
likely been the most pervasive influence on Bolivian forests. Natural fires have historically
influenced vast areas of Amazonian forest (Clark and Uhl 1987), particularly in dry or deciduous
forests where dry fuels may favor lightning fires (Middleton et al. 1997). In fact, most radiometric
dates of charcoal found throughout the Amazon correspond with the expansion of dry forests
during the dry glacial epochs (Saldariagga et al. 1986, Goldammer 1993, Prado and Gibbs 1993).
As is typical in most areas of the tropics, humans likely have been the most common
agents of forest fires in Bolivia. Although most tropical fires are set intentionally by humans for
the purposes of forest conversion, traditional slash-and-bumrn agriculture, or grazing land
management, many of these intentionally set fires escape (Uhl and Buschbacher 1985, Sarre and
Goldhanmmer 1996, Holdsworth and Uhl 1997, Cochrane et al. 1999). Consequently, human-
caused fires presently contribute more to tropical fire regimes than natural fires (Fearnside 1990,
Goldammer 1993. Cochrane and Schultze 1998, Nepstad et al. 1998, 1999, Goldammer 1999).
And, it is likely this was true historically as well, as human population densities in South America
have recovered only in this century to densities present before Europeans arrived (Denevan 1976).
Recent evidence from Bolivia reveals the susceptibility of seasonally dry forests to escaped
human-ignited fires. Over 1 million hectares of Bolivian dry forests burned during a severe dry
season in 1994 (Pinard et al. 1999), and over 3 million hectares burned in one month in 1999 (T.
Fredericksen, personal communication). Evidence also suggests that dry forests are damaged less
by wildfire than moister forest types (Mostacedo et al. 1999), which may also be indicative of the
pervasive role fire has played in the formation of these dry forests.
The Potential of Prescribed Burning in the Management of Bolivian Dry Forests
Although most guidelines for natural forest management focus on ways of reducing
damage to residual stands (Heinrich 1995, Pinard and Putz 1996, Haworth 1999), low-impact
selective logging may not be a sustainable management strategy in dry forests because of the low
levels of disturbance associated with this harvesting technique. In Lomerio for example, roads and
skid trails covered only 2-4% of logged sites and felling gaps were generally only 40-70 nm2ha'
after harvesting operations (Camacho 1996). Likely, this damage does not create sufficient canopy
opening for the regeneration of commercial tree species, 12 of 19 of which were classified as
having shade intolerant regeneration (Table 1-1).
Due to the pervasive influence fire has likely had on the formation and maintenance of
seasonally dry forests in Bolivia, prescribed bums are a promising silvicultural tool for managed
dry forests. Prescribed bums produce several effects that will likely increase regeneration of
shade-intolerant tree species, including vegetation removal, mineral soil exposure, and nutrient
release (Hungerford et al. 1990, Bond and van Wilgen 1996). The use of prescribed burning in
tropical forest management is not a new idea. Ground fires were used as early as the mid-1 800s to
enhance teak (Tectona grandis) regeneration in deciduous forests of South-east Asia (Dawkins and
Philip 1998). Tropical forest managers have recognized the benefits of prescribed bums for
several shade-intolerant timber species in addition to teak, such as sal (Shorea robusta) and several
pine species (Pinus; Goldammer 1994, Rodriquez 1996).
In South America however, the use of prescribed burning to enhance tree regeneration in
broad-leaf forests is rarely practiced. If fire is addressed in forest management policies, it is
primarily in the context of fire prevention or exclusion from protected areas (e.g., Reis 1996, New
Forestry Law in BOLFOR 1997). Particularly in Bolivia, the techniques of prescribed burning are
not well developed and the effects of prescribed bums on dry forest structure and function are not
Scope of Dissertation
The overall objective of this dissertation is to examine the ecological potential of
prescribed burning for the management of seasonally dry forests in eastern Bolivia. To be a viable
management strategy for the certified forests of Lomerio, prescribed bums must enhance
regeneration of commercial tree species without causing irreversible damage to the residual forest.
The negative effects of prescribed burns are likely to increase with increasing fire intensity
(reviewed in Chapter 2). Therefore, in this dissertation, I compare the effects of harvesting gap
formation, and controlled bums of high and low intensities on commercial tree regeneration, forest
soils, and plant diversity.
The dissertation contains seven chapters. In the second chapter, I briefly review the effects
of fire intensity on plants and soils, introduce the study forest, and describe four treatments that
form the basis of Chapters 3, 4, and 5. The treatments represent the following four disturbance
intensities: harvesting gap formation, above-ground biomass removal, and, controlled bums of low
and high intensity.
In the third chapter, I examine changes in soil physical and chemical properties following
the four treatments. I address the mechanisms underlying these changes by examining
experimentally the separate effects of heat and ash on soil properties. I also discuss how each of
these treatments, through their effects on soil properties, may influence tree seedling growth.
Commercial tree establishment, growth, and survival in each of the four treatments is
evaluated in Chapter 4. As species' responses to disturbance often vary among regeneration
guilds, I discuss the effects of each treatment by illustrating how they affect each guild differently.
I relate these results to the different management strategies that are appropriate for different
The effects of silvicultural treatments are primarily aimed at enhancing regeneration of
commercial tree species. Yet the impacts of these treatments on the remaining plant community are
also of concern, particularly in Lomerio where the local indigenous population depends on the
forest for a variety of other uses. In Chapter 5, I examine the response of the plant community to
the four treatments, focusing on changes in the dominance of species, life forms, and regeneration
modes (seedlings or sprouts) among treatments. I discuss these patterns in relation to their
importance for commercial tree regeneration.
The studies presented in Chapters 2 though 5 represent patterns of regeneration over an 18
month period following the treatments. In the sixth chapter, I examine patterns of regeneration
following bums over longer time scales using a chronosequence of secondary forests. I
characterize tree population structures, stand structure, and species richness in abandoned slash-
and-bum fields of 12 different ages, ranging from 1 to 50 years. Comparing these secondary
forests to a nearby mature forest stand, I discuss the possibility that the dominance of shade-
intolerant trees in this region may be the legacy of slash-and-bumrn agriculture.
In the final chapter, I summarize the chapters and discuss the ecological potential of
prescribed bums for the management of Lomerio forests. I discuss how prescribed burning might
fit into the current idea of natural forest management in the tropics. I also raise several questions
of the economical and political constraints to implementing controlled bums on a management
scale in Bolivia.
STUDY SITE AND TREATMENT DESCRIPTIONS
Forest disturbances vary widely in their type, intensity, frequency, and scale (Pickett and
White 1985). Despite this variation, disturbances, by definition, hold their most important
character in common: they reduce the dominance of a site by established individuals and create
openings for colonization and growth by new individuals (Canham and Marks 1985). As such,
disturbances are the primary catalyst of forest stand dynamics (Oliver and Larson 1996).
After most forest disturbances, there is a temporary increase the availability of light,
water, and nutrients. There are at least two mechanisms by which forest disturbance may increase
the availability of these resources. The first is the reduction in rates of uptake or use of resources
due to the loss of plant biomass. This effect is most apparent in the enhancement of light levels in
canopy openings (Chazdon and Fletcher 1984) and increased soil moisture in gaps (Vitousek and
Denslow 1987). Disturbances may also increase resource availability indirectly by altering rates
and pathways of nutrient cycling. For example, increased soil moisture and temperatures following
large-scale windthrow may temporarily increase nutrient availability by increasing the rate of
decomposition of soil organic matter (Bormann and Likens 1979).
Fire is an increasingly common disturbance in tropical forests (e.g., Goldamrnmer 1993,
Bond and van Wilgen 1996, Fearnside 1990, Sarre and Goldamnmer 1996). A feature of fire that
may set it apart from other disturbances is its effect below-ground. Fire acts as a rapid
decomposer, returning some nutrients from above-ground biomass to soil more rapidly than other
disturbances (e.g., Humphreys and Craig 1981, Hungerford et al. 1990, Neary etal. 1999). The
usually vigorous growth of seedlings in burned areas is often attributed to fertilization by deposited
ash and increased mineralization due to soil heating (DeBano et al. 1977). Yet, removal of above-
ground biomass can also be far more complete after fires than after other disturbances, such as
canopy gap formation. As such, it is likely a combination of above- and below-ground effects that
make fire a promising management tool for tree species with shade-intolerant regeneration
(Hungerford et al. 1990, Bond and van Wilgen 1996).
The effects of forest fires on above- and below-ground processes may vary widely
depending on their intensity (Moreno and Oechel 1994, Bond and van Wilgen 1996). For example,
low-intensity fires may have a positive effect on regeneration by increasing available soil nutrients
(DeBano et al. 1977, Wright and Bailey 1982), and stimulating flowering (Whelan 1994,
LeMaitre and Brown 1992), resprouting (Zedler et al. 1983, Moreno and Oechel 1994), and
germination of buried seeds (Bradstock and Auld 1995, Schimmel and Granstrom 1996, Enwright
et al. 1997). In contrast, high-intensity fires may be detrimental to regeneration by volatilizing
nutrients (Wright and Bailey 1982), altering soil properties such as texture, cation-exchange
capacity, and water holding capacity (DeBano et al. 1977, Hungerford et al. 1990), killing buried
seeds (Schinmmel and Granstrom 1996), killing species that would otherwise resprout (Moreno and
Oechel 1994), and damaging or killing potential seed trees. Although there is generally a positive
relationship between the size or intensity of disturbance and the subsequent availability of
resources for plant growth (Canham and Marks 1985) this pattern may not apply for fires of
extreme severity. Regeneration on sites of low-intensity fires may be enhanced while areas of high-
intensity fire may be very slow to recover.
I designed an experiment to address the relative importance of canopy opening, above-
ground biomass removal, and controlled bums of high and low intensity on early patterns of
tropical dry forest regeneration. Using these experiments, I examined soil physical and chemical
properties (Chapter 3), establishment and growth of commercial tree seedlings (Chapter 4), and
changes in the plant community (Chapter 5), comparing each response to conditions in undisturbed
forest understories. In this chapter, I introduce the study site and describe the treatments.
The studies described in this dissertation were conducted in the seasonally dry forests of
Chiquitania, a geographic region in the eastern lowlands of Bolivia located in the Province of Nuflo
de Chavez, Department of Santa Cruz (1645'S, 6145'W; Figure 2-1). Chiquitania is situated in
a transition zone between the humid forests on the southern rim of the Amazon basin and the thorn
scrub formations of the Gran Chaco. The natural vegetation is classified as tropical dry forest
(sensu Holdridge et al. 1967).
The regional climate is characterized by pronounced seasonality with a strong dry season
that corresponds to the austral winter (Figure 2-2). Most of the canopy trees are seasonally
deciduous, shedding their leaves from June to September. The mean annual temperature at
Concepcion is 24.3C with temperatures that vary between 3 (July) and 38. 1C (October, Killeen
et al. 1990). The mean annual precipitation is 1129 mm and interannual variability is large, with
lows having reached 500 mm and highs 1717 mm per year (Killeen et al. 1990). The landscape is
dominated by low hills composed of granite, gneiss, and metamorphic rocks of Precambrian origin
(Geobold 1981 in Killeen et al. 1990) punctuated by exposed granitic outcrops (inselbergs). The
soils of the area are classified as Inceptisols (suborder: Tropepts, group: Ustropepts) and Oxisols
(suborder:Ustox, group: Eutrusox; Ippore 1996). Elevation varies between 400 and 600 m a.s.l.
Canopies of mature forest range from 12-18 m tall and are dominated by trees of the
Leguminosae (60% of total basal area of trees >10 cm dbh); trees in the families Bignoniaceae,
Anacardiaceae, and Bombacaceae are also abundant (Killeen et al. 1998). Understory trees are
mostly represented by the families Sapindaceae and Myrtaceae. A spiny ground bromeliad,
Pseudananas sagenarius, is distributed over approximately 80% of the forest and occurs in
clumps up to 2000 m2 (MacDonald et al. 1998).
S . Lomerio
Figure 2-1. Location of the study site in the seasonally dry forests ofChiquitania, a geographic
region in the eastern lowlands of Bolivia located in the Province of Nuflo de Chavez, Department
of Santa Cruz (16"45'S, 6145'W). In the enlarged area, points mark the 28 communities in the
political region of Lomerio.
'93 '94 '95 '96
Figure 2-2. Mean monthly temperature and monthly precipitation at Concepcion,
Santa Cruz (167' S, 6202' W, 490 m a.s.l.), located approximately 100 km from
Chiquitania is so named for the Chiquitano indigenous people, the largest of the lowland
indigenous groups in Bolivia, with a population of around 72,500. Lomerio, where this study was
conducted, is a political region within Chiquitania made up of 27 Chiquitano communities with a
total population of around 5,000. The Chiquitanos of Lomerio have been managing their forests
for timber since 1982 with technical and financial support from several international institutions.
BOLFOR (Proyecto de Manejo Forestal Sostenible), a sustainable forest management project with
USAID funding, began working in Lomerio in 1992. The objective of the current management
plan for the forests of Lomerio is to produce timber sustainably while minimizing negative impacts
on other biological and physical resources in the forest (Pinard et al. 1999). Forestry operations
of the Chiquitano communities were certified as sustainable by the SmartWood Program of
Rainforest Alliance in 1995.
The particular forest in which I worked is owned by the Chiquitano community of Las
Trancas and situated approximately 12 km northeast of this village. The Las Trancas forest
contains 400 ha management blocks, Las Trancas '94 and Las Trancas '95, so named for the year
in which forest inventories were conducted (Figure 2-3). Las Trancas '94 was selectively logged in
July-September of 1995. On average, 3-10 m3 ha' (2-5 trees ha1) of timber were extracted from 6
species. Damage to the residual stand was slight, with 6% of the residual trees damaged and 2-4%
of the area covered by roads and extraction routes (Camacho 1996). Two logging methods were
employed in Las Trancas "95. In 1996, approximately 75% of the area was selectively logged. In
1997, the remaining 25% of the area was selectively harvested in strips (40 m x 200 m, each
separated by an unharvested area 60 m wide). All commercially valuable trees were harvested
from these strips at a harvesting intensity of 4.4 m3 ha1. Log extraction routes (skid trails) entered
each logging strip 100 m from the north and 100 m from the south.
Las Trancas '94
Las Trancas '95
/ -scale: 1 km
scale: 1 km
Figure 2-3. The '94 and '95 management blocks of the Las Trancas community owned forest.
The enlarged section of Las Trancas '95 represents the trial area for the strip shelter wood system.
The squares within this enlarged area mark the 16 blocks (20 x 20 m gap areas) depicted in Figure
Location of felling gaps
The studies presented in Chapter 3, 4, and 5 were conducted in the selectively harvested
strips in Las Trancas '95. These strips were the only area of logging activity during the dry season
of 1997 and therefore all newly created felling gaps were located in these strips. In June of 1997, I
located 16 felling gaps for study (Figure 2-3). Gap selection was restricted by the following
criteria: canopy gap area between 200-600 m2, slopes no greater than 15, less than 20% rock
outcrops, no trees > 40 cm DBH within gap area, and not located in the path of skid trails.
I located and marked the center of each gap where the midpoints of two perpendicular
transects intersected, the first running the length of the longest axis. Each gap was divided into four
10 x 10 min plots by cardinal axes from the center point (Figure 2-4). Half-meter wide paths around
the perimeter of the gap and along axes were cut by machete. Existing gap area was enlarged to a
uniform 20 m x 20 m area by cutting all vegetation >2 min tall (sensu Brokaw 1985a) by machete or
chainsaw. Because this forest is a timber management area, commercial tree species > 20 cm
DBH located within the 20 m x 20 m gap area were left uncut (this occurred in only 4 of 16 gaps
and standing trees did not exceed 25 cm DBH).
One of four treatments was randomly assigned to each 10 x 10 m plot within each block:
1) high-intensity bum; 2) low-intensity bum; 3) plant and coarse debris removal (hereafter referred
to as plant removal); and, 4) canopy gap with vegetation > 2 min tall cut (the gap control). Other
than cutting all vegetation > 2 min tall, vegetation and woody debris in the gap control was not
manipulated. In the plant removal and low-intensity burn treatments, all vegetation was cut at or
near the soil surface and everything >2.5 cm diameter (>: 100 hour fuels) was removed and
distributed as evenly as possible in the high-intensity bum treatment. Tree trunks and large
diameter branches were sawn into smaller sections so that they could be moved more easily and dry
I I I
- 15-20 m
Figure 2-4. A single block consisting of a 20 x 20 m felling gap and an adjacent forest plot. Each
felling gap was equally partitioned into four 10 x 10 m treatment plots. Each treatment plot was
randomly assigned one of four treatments: gap control, plant removal, low intensity or high
intensity bum. Within each treatment plot, 2 permanent vegetation sampling subplots (2 x 2 each)
were located near the gap center and 2 additional subplots near the gap edge. All soil sampling
was conducted outside of these permanent sampling plots. Two permanent subplots were located
approximately 15-20 m from the edge of each gap in undisturbed forest. Soil sampling in the
forest was conducted outside but within 5 m of these permanent forest subplots. Hereafter, the
400 m2 felling gap and adjacent forest site are "blocks," the 100 m2 treatment areas are
"plots," and the 4 m2 vegetation sampling areas are "subplots."
rapidly. Pseudananas sagenarius and cacti were not added to the high-intensity bum treatment,
but instead were removed from the block altogether because of their low flammability. Therefore,
after fuels were manipulated and before prescribed bums, the plant removal and low-intensity bum
treatments had similar amounts of litter and woody debris and no above-ground vegetation. The
high-intensity bum treatment plots had roughly 3 times its original fuel load. Slash was left for 5
rainless weeks to dry before prescribed bums were conducted.
Fuel loads. Pre-bum fuel loads in the low- and high-intensity bum treatments were
measured in randomly located 0.25 m2 circular subplots, with 2 subplots sampled per plot x
intensity treatment (2 fuel plots x 2 bum treatments x 16 plots). All fuel within each subplot was
removed, divided into fuel diameter size classes (live herbs, <6 mm, 6-25 mm, 25-75 mm, and >75
mm) and weighed in the field. Composite subsamples of each fuel size class were taken from the
field, oven dried to constant weight, and used to calculate the wet-to-dry weight conversion factors.
The diameter and length of trunks and large diameter branches that could not be weighed in the
field were measured in order to estimate volumes. Wood densities available from BOLFOR were
used for volume to mass conversions.
Timing of burns. Although little is known about the historic fire regime in these forests,
seasonal patterns in rainfall and relative humidity make wildfires most likely at the end of the dry
season when fuels are dry and lightning strikes most common. The indigenous Chiquitano
population traditionally bum their agricultural fields and cattle pastures at the end of the dry
season as well, shortly before the onset of rains. Predictably, most escaped fires occur during this
season. Because one of the objectives of my experimental bums was to enhance seedling
establishment of commercial tree species, I planned a burning date in late August or early
September, at the end of the dry season and before peak seed fall of most commercial trees
Fire breaks. All fuel was removed from a 1 m wide fire break around low-intensity bum
treatment plots. Fire breaks around high-intensity bum treatment plots were 1 to 2 m wide, wider
where danger of fire escape was perceived to be higher. Standing dead trees near firebreaks were
felled and ladder fuels such as liana tangles were removed. On the day of burns, firebreaks were
raked free of newly fallen leaves.
Prescribed burns. Prescribed bums were conducted from August 29 to September 1,
1997, near the end of the 5 month dry season (Table 2-1). Each day, the earliest bums were
started at 10:00 a.m. and the last bums by 3:00 p.m. Temperature at 10:00 a.m. over the four day
period varied from 34 to 36.4 C and relative humidity varied from 29-38 %. Winds were variable,
but usually calm in the morning with convectional wind gusts of up to 11 km/hr in the afternoon.
A circular ignition technique was used for both bum treatments. A spot fire was lit with a
drip torch in the plot center, then the perimeter was lit starting with the downwind side. The center
fires created convection which drew the ring fire on the borders inward. In the low-intensity bum
treatment plots, ring fires often did not carry to the center, therefore spot fires were ignited where
A minimum of 5 people conducted the bums over the 4 day period. At least one person
with a backpack water sprayer remained at each fire until fires near the borders were extinguished.
Fires near firebreaks or standing dead trees were extinguished before burning crews returned to
camp. Each fire was checked again after dark and the following morning to extinguish any
potentially dangerous smoldering areas.
Maximum soil temperatures, fire intensities and completeness of burns. Maximum soil
temperature and an index of fire intensity were measured in two locations in each bum plot, near
the two subplots where fuel loads were measured. Maximum soil temperature was measured at 0
and 3 cm depth using temperature indicating paints (Tempilaq , Tempil Division. Air Liquide
Table 2-1. Climatic conditions at 10:00 a.m. the morning of high and low intensity
bums for 16 experimental blocks.
Block Date Ambient Relative Wind speed
burned temperature (C) humidity (%) (km/hr)
1 29-Aug 36 29 0 with gusts
19 29-Aug 36 29 0 with gusts
20 29-Aug 36 29 0 with gusts
4 30-Aug 34 34 11
6 30-Aug 34 34 11
8 30-Aug 34 34 11
21 30-Aug 34 34 11
22 30-Aug 34 34 11
2 31-Aug 34 37 4
7 31-Aug 34 37 4
9 31-Aug 34 37 4
11 01-Sep 35 38 0 with gusts
14 01-Sep 35 38 0 with gusts
17 01-Sep 35 38 0 with gusts
18 01-Sep 35 38 0 with gusts
America Corporation, South Plainfield, New Jersey, USA). Paints of 24 different melting points
ranging from 66 to 1093 C were applied as narrow bands on 2 x 30 cm steel strips. At each
location, one painted steel strip was buried at 3 cm soil depth and another placed flat on the soil
surface directly above it. Soil temperatures were measured to a greater depth in one block. Here,
an additional 3 sets of 4 painted strips were placed at 0, 1, 3, and 7 cm depths. After fires, the
highest indicated melting point was recorded.
Fire intensity was estimated by Beaufait's (1966) technique which calculates total energy
output from the amount of water vaporized from cans during burns as:
total energy output = [(80 cal/g water) x (g water)] + [(540 cal/g water) x (g water)]
Where 80 cal arc needed to raise each gram of water from 20" C to the boiling point and 540 cal
are needed vaporize each gram of water (latent heat of vaporization). Two tin cans per bum were
used, each placed on the soil surface of fuel load subplots. Depth of water was measured
immediately before each bum and within 24 hours after. To account for the amount of water lost
due to evaporation, 2 cans were placed in the center of an unburned gap and the amount of water
evaporated within 24-hrs measured.
Soil moisture, which influences heat movement through soil, was measured several hours
before burns. Soil samples from 0-5 and 5-10 cm depths were collected from each plot, weighed,
oven dried to a constant weight, and moisture content expressed as % of soil dry weight.
The week following bums, completeness-of-bum was estimated visually as the percent
Establishment of permanent vegetation plots
Three weeks following bums, 4 permanent subplots (2 x 2 m each) were established in
each treatment plot, 2 located near the gap center and 2 located near the gap edge (Figure 2-4).
Two additional subplots were established at random points in undisturbed forest 15-20 m from the
edge of each gap. These permanent subplots were used for sampling commercial tree seedling
establishment (Chapter 4) and vegetative cover (Chapter 5). One plot of each pair was used for a
seeding treatment described in Chapter 4.
Treatment effects on canopy cover and microhabitat
Soil temperature to 3 cm depth was measured at a center and edge subplot of each
treatment as well as the forest subplots with a soil thermometer 3 and 6 months following bums.
Percent canopy cover was measured with a spherical densiometer above each gap center, gap edge,
and forest plot 3 months following bums. Litter depth (cm) and percent cover by debris 2-20 cm
and >20 cm diameter were estimated visually for each of the permanent 4 m2 subplots 6 weeks
following bums. Results were analyzed using an analysis of variance, with treatment as a fixed
effect and block as a random effect, followed by Tukey's post-hoc comparisons.
Initial Treatment Results
Pre-burn fuel loads
Pre-burn fuel loads in high-intensity bum treatment subplots ranged from 10.8 to 82.8
kg/m2 and averaged 48 4.9 kg/m2 (mean 1 standard error; Figure 2-5). Almost half of this mass
was comprised of fuels >7.5 cm diameter. Fuel loads in the low-intensity burn treatment subplots
ranged from 0.8 to 4 kg/m2 and averaged 2.2 2.3 kg/m2. Sixty-six percent of the fuel mass in
low-intensity plots was fine fuel, <6 mm diameter.
High-intensity burns. Completeness of high-intensity bums was variable, but the
majority of bums consumed all but the thickest (> 20 cm diameter) branches and trunks. Flame
heights ranged from 1.5 to 5 m. Fire intensities ranged from 152 to 3795 kcal and averaged 1627
241 kcal (n = 15). Temperature at the soil surface during high-intensity burns averaged 704 42
C (n = 16). The highest temperature measured was 927C. Temperature at 3 cm depth averaged
227 27 C (n = 16). Where maximum temperature was measured at additional depths of 1 and 7
4000 Fire intensity (kcal) 1000
400 000--- JA -- io
soil surface 3 cm depth
Figure 2-5. Pre-bum fuel loads, fire intensities, and maximum temperatures at the soil surface
and 3 cm depth during high and low intensity bums. Box plots show medians (center line),
25th and 75th percentiles (top and bottom lines), 10th and 90th percentiles (top and bottom
whiskers), and points greater than the 90th percentile and less than the 19th percentile (dots).
cm, temperatures averaged 871 C at the soil surface, 358C at 1 cm depth, 218 C at 3 cm depth,
and 135 C at 7 cm depth (n = 2). Although visible flames were extinguished by nightfall, some
logs continued to smolder for several days: fire intensities and soil temperatures under these logs
were likely greater than measured values.
Low-intensity burns. In general, completeness of low-intensity bums was more variable
than high-intensity bums. Flame heights were low, ranging from 10 to 50 cm. Fire intensity
ranged from 22 to 68 kcal and averaged 41 3 kcal (n = 15). Temperatures at the soil surface
averaged 225 33 C (n = 12); the highest temperature measured was 413 C. Elevated
temperatures at 3 cm were only detected in 2 of 16 plots; these averaged 107 7 C (n = 2). Soil
moisture on the day of bums was low and did not differ between the high- and low-intensity bum
plots (0-5 cm depth: P = 0.94, 5-10 cm depth: P = 0.23). Therefore, differences among the 2 bum
treatments in heat conductivity due to soil moisture were likely negligible and are hereafter ignored.
Treatment effects on microhabitat
Treatments had significant effects on the amount of soil exposed, mid-day ambient soil
temperature, litter depth, and area covered by woody debris (Figure 2-6). Canopy cover above
forest plots was 78%, higher than canopy cover above all 4 gap treatments, which averaged 22%
(P < 0.001). Although canopy cover above gap-center and gap-edge plots was not significantly
different (P = 0.6), soil temperatures in gap centers were higher than near gap edges (P < 0.001).
A maximum temperature of 43 C was recorded in the center of one high-intensity bum treatment 3
months following bums. After 6 months, soil temperatures at gap centers and edges were not
different (P = 0.52) and only soil temperatures in the high-intensity bum treatment were
significantly higher than the other treatments (P < 0.001).
High-intensity bums removed all litter and deposited a layer of ash ranging from 0-14 cm
depth (4.8 0.2 cm, n = 16). Not all woody material was consumed in the high-intensity bums;
Litter depth d
P < 0.001
Debris 2-20 cm
PK< 0.001 C
High Low Plant
intensity intensity removal
Figure 2-6. Litter depth, percent cover of debris 2-20 cm and > 20 cm diameter,
and mid-day soil temperature in four gap treatment plots and forest plots 6 weeks
after bums. Treatments with the same letter are not significantly different.
Debris > 20 cm
the remaining woody debris covered approximately 12% of the subplots areas. In the low-intensity
bum treatment, an average of 76% of the subplot areas burned to some degree. Burning was not
complete even within these areas as only an average of 30% of the subplot areas had soil exposed.
Small woody debris remained on approximately 3% of the area of low-intensity bum subplots;
most large woody debris had been removed before burning.
Leaf litter or small woody debris covered all of the plant removal treatment subplots, with
no bare soil exposed. As with the low-intensity bum treatment, most large woody debris was
removed. Gap controls were characterized by deep leaf litter (2.9 0.2 cm, n = 16) and small and
large woody debris covering an average of 25% of the subplot areas. Only 20% of the gap control
subplot areas were devoid of either woody material or surviving plants. Forest understories had
the deepest leaf litter (4.0 0.3 cm, n = 16), but small and large woody debris combined covered
an average of only 7% of the subplot areas.
EFFECTS OF CANOPY GAP FORMATION, PLANT REMOVAL,
AND CONTROLLED BURNS OF HIGH AND LOW INTENSITIES ON
SOIL CHEMICAL AND PHYSICAL PROPERTIES
Fire is a rapid decomposer; it compresses the oxidative processes of organic matter decay
into a short time span (Wright and Bailey 1982). The result is a nutrient pulse much larger than
from the normal decomposition of woody debris and litter, at least for the first few months
following fires (Bond and van Wilgen 1996). As such, controlled burns may benefit tree seedling
growth more than unburned treatments, particularly since the timing of nutrient pulses following
fire coincides with maximum light availability. After intense fires, however, the advantages of
increased nutrient availability may be offset by degraded soil structure. Thus, the benefit of
controlled bums for tree seedling growth may ultimately depend on fire intensity. In this chapter, I
examine both soil nutrient availability and soil physical properties following canopy opening, plant
removal, and controlled burns of high- and low-intensity.
There are three primary mechanisms of increased nutrient availability following fire:
nutrients added to the soil as ash; heating of soil organic matter; and, increased rates of biological
mineralization following fire due to increases in soil pH, temperature, and moisture, as well as due
to a reduction in C:N ratios (Wright and Bailey 1982, Pritchett and Fisher 1987). The degree of
increase in nutrient availability following fires depends largely on fire intensity. Most studies of
low to moderately intense fires report increases in available nutrients (reviews by Dunn et aL. 1977,
DeBano et al. 1977, Wells et aL. 1979, Humphreys and Craig 1981, Wright and Bailey 1982,
Hungerford et al 1990, Neary et al. 1999). In contrast, intense fires may cause a net loss of
nutrients (DeBano et al. 1977, Giovannini et al. 1990).
Due to its low temperature of volatilization (200 C; Weast 1988), nitrogen loss is linked
with the consumption of organic matter (e.g., Dunn et al. 1977). Where fuels are completely
consumed and the surface layer of soil organic matter is destroyed, loss of nitrogen through
volatilization can be substantial (e.g., Nye and Greenland 1964, Ewel et al. 1981). Volatilization
of phosphorus and cations are usually minor due to the high volatilization temperatures of these
minerals (>760 C; Weast 1988), however, their loss from severely burned sites may be caused by
surface erosion, leaching, or transport of ash (Wright and Bailey 1982).
Intense burns may also have detrimental effects on soil physical properties by consuming
soil organic matter. Soil organic matter holds sand, silt, and clay particles into aggregates,
therefore a loss of soil organic matter results in a loss of soil structure. Severe fires may also
permanently alter soil texture by fusing clay particles into sand-sized particles (Dymess and
Youngberg 1957, Ulery and Graham 1993). By altering soil structure and texture, severe fires can
increase soil bulk density (DeByle 1981), and reduce soil porosity, water infiltration rates, and
water holding capacity (e.g., Wells et al. 1979). Intense bums may also induce the formation of a
water repellent soil layer by forcing hydrophobic substances in litter downward through the soil
profile (DeBano 1969), reducing water infiltration rates as a consequence (DeBano 1971).
The changes in chemical and physical soil properties caused by fire potentially have
important consequences on tree seedling growth (Johnson 1919). Increased nutrient availability
after fire may benefit plant growth if nutrients are limiting prior to burning (e.g., Hungerford et al.
1990). On the other hand, seedling growth in intensely burned soils may be slowed due to high pH
and toxic levels of minerals (Giovannini et al. 1990). Altered soil physical properties, such as soil
strength, bulk density, and water infiltration rates, may also impair plant growth. Plant uptake of
nutrients and water is slowed in structurally degraded soils through the combined effects of lower
soil moisture and lower soil porosity (Nye and Tinker 1977). Mechanical impedance of root
growth caused by increased bulk density and soil strength (Gerard et al. 1982) also slows nutrient
and water uptake.
In this chapter, I focus on the below-ground effects of the treatments described in Chapter
2. My objectives were to: 1) compare the effects of canopy gap formation, plant removal, and
controlled bums of high and low intensities on soil nutrient availability, soil physical properties,
and fine root mass; 2) compare the relative importance of soil heating and ash-fertilization on soil
nutrient availability; and, 3) discuss how these treatment-induced changes in soil properties
influence tree seedling growth.
The studies presented in this chapter were conducted in the treatment and forest plots
described in Chapter 2 (Figure 2-4). All soil sampling was done within the 100 m2 plots but
outside of the 4 m2 subplots. Forest sampling was done within 5 m of the forest subplots.
Mass and chemical characteristics of ash deposited in high-intensity burn plots
Ash mass deposited in high-intensity bum plots was estimated by collecting and weighing
all ash on the soil surface in a 1 m2 area, replicated in three high-intensity bum plots (n = 3). To
characterize variability in the amount, ash depth was measured in 10 randomly located points in
each high-intensity bum plot (n = 16). Composite ash samples were then collected from each plot
and used to measure pH and nutrient concentrations. Ash pH was determined as for soil pH,
described below. To determine nutrient concentrations, 0.5 g of ash was heated in 10 mL of 1 M
HNO3 and then resolubilized in 10 mL of 1 M HC1. Extracts were then analyzed for phosphorus.
potassium, calcium, and magnesium at the Analytical Research Laboratory at the University of
Florida with an ICAP Spectrometer (Thermo-Jarrell Ash Corporation, Franklin, MA).
Soil samples from 0-8 cm depth were collected 2, 6, 9, 12 and 18 months after bums and
from 8-20 cm depth after 9, 12, and 18 months. These samples were used to assess moisture
contents, pH, organic matter, and extractable elements. In each treatment and forest plot (n = 16
blocks), 4 samples were taken from randomly selected sites with a 10 cm diameter cylindrical
corer. The 4 samples from each treatment were mixed thoroughly in the field and a -300 g
composite subsample bagged (Anderson and Ingram 1993). In one block, the 3-month samples
were bagged separately, rather than composite, to examine intra-treatment variability. Subsample
soil volume was unknown, therefore bulk density and fine root mass were sampled separately as
Soil pH and air-dry moisture content
The pH of fresh soil was determined by adding 50 ml of distilled water to 20 g of soil and
stirring for 10 minutes (Anderson and Ingram 1993). The mixture then stood for 30 minutes and
pH of the supernatant was measured with a hand-held meter (Oakton pHTestr 3). Soil samples
were then weighed, air-dried to a constant weight, and reweighed to calculate air-dry moisture
content. Air-dried samples were passed through a 2 mm sieve, bagged, and stored in a cool dry
area until transported back to the Analytical Research Lab at the University of Florida for
Soil chemical analyses
Phosphorus, potassium, calcium, and magnesium were extracted with Mehlich-I solution:
0.05 M HCI and 0.0125 M H2SO4 (Hanlon et al. 1994). Extracts were then analyzed by ICAP
spectroscopy. Soil organic matter content was analyzed using the Walkley-Black dichromate
methodology (Hanlon et al. 1994). A subset of soil samples was tested for total nitrogen using an
elemental analyzer (Carloerba NCS 2500). Twenty-four samples from all treatments and sampling
periods from the top 8 cm of soil were selected to represent the full range of organic matter content.
Resin-available nitrogen and phosphorus
Resin-available nitrogen (NI--N and NO3"-N) and phosphorus (PO4 3-P) in each
treatment were estimated by burying anion and cation exchange resin bags at 5 cm depth. Resin
bags were prepared by enclosing 5.0 g (moist weight) of either anion exchange resins (Sigma-
Dowex) or cation exchange resins (Fisher Scientific) in bags of nylon stocking material sewn
closed with nylon thread. Before burial, resin bags were hydrated overnight with dionized water.
Four bags of each resin type were buried per treatment plot (4 bags x 2 resin types x 5 treatments x
16 blocks). Three rotations of resins were buried, each for approximately 3 months. Two
rotations included the first and second rainy seasons following bums (November 1997-January
1998 and December 1998-February 1999, respectively). The middle rotation covered the transition
from the first rainy season to the first dry season following burns (May-July 1998). After removal
from the field, resin bags were placed separately in clean plastic bags and kept cool (refrigerated
when possible) until transported to the University of Florida for analysis. For each resin type, the
4 bags per plot were pooled and 12 g extracted in 120 ml of 2 M KCI for 24 hours. Extractions
were analyzed for amonium-N and nitrate-N using automated spectrophotometry (Flow IV Ion
Analyzer, AlpKem (0-I-Analytical), College Station, TX). Extracts from anion exchange resins
were diluted to 1M KCI and analyzed for PO4 3-P using the atomic emission spectrometric method
(Thermo-Jarrell Ash Corp. Franklin, MA).
Soil nutrient concentrations, organic matter, pH, moisture content, and resin-available N
and P were analyzed using an ANOVA with repeated measures. Treatment was a fixed effect and
block a random effect in each model. Soil properties were log transformed for analyses when not
normally distributed, but all values presented in the text are non-transformed. Where a significant
time x treatment interaction was found, variables were analyzed separately by month. Statistically
significant differences (P < 0.05) were further analyzed with Tukey's HSD multiple comparisons.
In order to describe variation within plots, 4 soil samples per treatment in one block (block
4) were analyzed separately for extractable nutrients and organic matter. Also, 4 resin bags per
treatment in one block were extracted and analyzed separately for resin-available nitrogen and
phosphorus. Coefficients of variation (Sokal and Rohlf 1981) were calculated to compare variation
of soil sampled within the same 100 m2 plot and among the 16 different plots.
Fine root mass
Fine root mass (roots < 2 mm diameter) was compared among treatments 12 months
following bums in a reduced sample of 10 blocks (n = 10). Soil cores were extracted with a
cylindrical tube (5 cm inside diameter, 7 cm deep) from 3 randomly located points in each
treatment and forest plot. Fine roots were sorted from samples, dried, and weighed. Fine root
mass (live and dead combined) was compared among treatments using an ANOVA followed by a
Tukey's HSD post-hoc test.
Soil bulk density
Soil bulk density (air-dry) was estimated 6 and 12 months following bums in a reduced
sample of 10 blocks (n = 10). Three samples in each treatment and forest plot were collected using
metal cans (5 cm inside diameter, 7 cm deep, 137 cm3). Samples were air-dried to constant weight
and bulk density calculated as:
bulk density (g/cm3) = g air-dried soil / 137 cm3
Differences among treatments were tested using an ANOVA on square-transformed values of bulk
density with treatment and month as fixed effects and blocks as a random effect, followed by
Tukey's HSD multiple comparisons.
Compressive soil strength was estimated with a pocket penetrometer (Forestry Suppliers)
at 2, 6, 9, and 12 months following bums. Soil strength readings were taken at 4 randomly
selected points in each treatment and forest plot from all 16 blocks (n = 16). Soil strength was
analyzed using ANOVA with repeated measures as described above for soil chemical properties.
Water infiltration rates were estimated in a reduced sample of 4 blocks 8 months following
bums (n = 4). The technique used here was a modified version of the single ring method (Anderson
and Ingrain 1993). Although double ring methods provide better estimates of infiltration rates
because they compensate for lateral flow, a single ring method was chosen for this study because it
used less water (which had to be transported 27 km). In each gap treatment and forest site, a point
was randomly located and cleared of surface litter. A graduated PCV cylinder (10 cm diameter, 25
cm length) was inserted vertically into the soil 10 cm deep and soil pressed around the base to
minimize water leakage. The cylinder was filled with water to 10 cm and timed until the water
level dropped to 5 cm. This process was repeated three times. Infiltration rates were calculated
separately for each repetition (i.e., the first, second, and third 5 cm increments of water which
correspond to 5, 10, and 15 ml cm-2) as the volume flux of water flowing into the soil profile per
unit surface area (Hillel 1982) and expressed as ml cm-2 sec'. Log transformed infiltration rates
were compared among treatments using an ANOVA with treatments and repetitions (i.e., each 5
cm increment) as fixed effects and blocks as random effects.
Soil wettability was estimated using a modification of the water drop penetration time
method (WDPT; Letey 1969) in a reduced sample of 7 blocks 8 months following burns (n = 7).
In each gap treatment and forest site, four 20 x 20 cm areas were randomly located and cleared of
surface litter. Five drops of water were placed on the soil surface with a dropper and the time
recorded when all 5 drops were completely absorbed. This was repeated at 1, 2, and 3 cm soil
depth by scraping surface soil away with a machete. Soil wettability (log transformed seconds)
was compared among treatments using an ANOVA with treatments and soil depth as fixed effects
and blocks as random effects.
Comparative effects of soil heating and ash addition on soil chemical properties
To compare the effects of soil heating and ash addition on soil chemical properties, I
carried out a 2 x 3 factorial experiment with two levels of ash addition (no ash and ash added) with
three levels of soil heating (no heat, low-intensity heat, and high-intensity heat). The first trial was
conducted in the field in Las Trancas '95. The second trial was conducted in the BOLFOR
greenhouse in the city of Santa Cruz. Anadenanthera colubrina served as a bioassay in the field
experiment. Because Anadenanthera did not fruit in 1998, Caesalpinia pluviosa was used as a
bioassay in the greenhouse study.
Field study. I utilized the plant removal treatment plots described in Chapter 2 for the field
study, conducted at the end of the dry season in October 1997. The design is a complete
randomized block; each plant removal plot was considered as a block (n = 12). In each block, six
1 m2 plots were located in the area between the gap center and edge permanent subplots (Figure 2-
4). Competing vegetation had been cleared from the larger treatment plots the month before, but
some regrowth had already occurred. Therefore the 1 m2 plots were cleared again of any
vegetation and raked of surface litter to expose the soil surface. Each plot was randomly assigned
a treatment combination of soil heating (no heat, low-intensity heat, or high-intensity-heat) and ash
(no ash or ash added). Treatments were applied to a 50 x 50 cm area in the center of the 1 m2 plots,
creating a 25 cm buffer along the edge. Heat was applied using a propane blow torch.
Temperature of the flame was measured with Tempil heat sensitive paints. In the low-intensity
heat treatment, a flame of 150-250 C was applied to the soil surface of the treatment area for 5
minutes. In the high-intensity heat treatment, a flame of 500-800 C was applied to the soil
surface of the treatment area for 20 minutes. The torch required constant adjustment to maintain a
similar flame, therefore temperatures varied within a treatment. I am confident, however, that
temperatures ranges did not overlap between the high- and low-intensity treatments. After heat
treatments, approximately 500 g of ash collected from high-intensity bum plots was distributed as
evenly as possible to plots assigned the ash treatment. The week following treatments, 10 seeds of
Anadenanthera colubrina were placed in each plot and checked for germination after 4 days.
After 2 weeks, most seeds were found to have been removed or eaten and were therefore not
Soil samples 0-8 cm depth were collected 3 weeks following treatments. Soil pH, and
phosphorus, potassium, calcium, magnesium, and organic matter concentrations were analyzed
using the methods described above. Resin-available nitrogen (NH4-N and N03-N) was measured
using anion and cation exchange resin bags. One bag of each type was buried at 5 cm depth in
each plot for 85 days (October 31 January 24). Analysis of resins follows that described for the
Greenhouse study. The greenhouse study was conducted at the end of the dry season in
1998. Soil used for this trial was collected from Las Trancas '95 to a depth of 10 cm. Soil was
passed through an 8 mm sieve, mixed well, and divided into three equal batches. Each batch was
assigned a soil heating treatment (no heat, low-intensity heat, and high-intensity heat). In the low-
intensity heat treatment, soil was heated in aluminum pots in a conventional oven at 100-150 C for
a total of 10 minutes (mixing after 5 minutes). Soil in the high-intensity treatment was oven heated
at -200 C for 40 minutes (mixing after 20 minutes) then spread 1 cm deep on a metal tray and
heated with a blow torch for 5 minutes at a temperature of 500-800 C. Oven temperature and
torch temperature were both measured using Tempil heat sensitive paints. One composite soil
sample from each heating treatment (control, low-intensity, and high-intensity) was analyzed for
phosphorus, potassium, calcium, magnesium, and organic matter concentrations. Soil from each
heating treatment was used to fill 24 plastic planting containers (7 x 25 cm). Twelve planting
containers per heating treatment were then selected for the ash addition treatment (15 g of ash
added to the soil surface) and the remaining 12 containers served as controls (n = 12). Two seeds
of Caesalpinia pluviosa were placed in each planting container, watered daily, and seedling height
to the terminal bud was measured after 4 months. Resin-available nitrogen (NH4+-N and NO03-N)
was measured in each treatment combination using additional planting containers. Anion and
cation exchange resin bags were buried at 5 cm depth in 3 containers of each treatment
combination (n = 3) and watered daily for 22 days. Resins were extracted and analyzed using the
methods described above.
Mass and chemical characteristics of ash deposited in high-intensity burn plots
Variability in amount of deposited ash was high and depths ranged from 0-14 cm (4.8
0.3 cm, x S.E., n = 16). Ash mass deposited by high-intensity bums averaged 1.5 0.6 kg/m2
(n = 3). Using this value and measured concentrations of individual elements in ash (Figure 3-1)
indicates an average nutrient deposition of 524 g/m2 of Ca, 26 g/m2 of Mg, 83 g/m2 of K, and 7.7
g/m2 of P. Ash samples had an average pH of 10.7 0.1 (n = 16).
Treatment effects on soil nutrients
High-intensity burns significantly increased P, Mg, K, and Ca in the top 8 cm of soil, but
the magnitude and its change over time varied by nutrient (Figure 3-2). These increases were also
detected at 8-20 cm for all elements except Mg (Figure 3-3). Low-intensity burns also
significantly increased P, Mg, K, and Ca in the top 8 cm of soil, although increases were smaller
than in high-intensity bum plots, did not persist as long, and were not detected at 8-20 cm. Plant
removal and gap control treatments had no detectable effect on P, Mg, K, and Ca at either soil
depth. Results of statistical analyses are summarized in Tables 3-1 and 3-2.
(mg/g ash) 100
Figure 3-1. Box plot diagrams of concentrations of Ca, Mg, K, and P in ash sampled
from high intensity burn plots (n = 9). Box plots show medians (center line), 25th and
75th percentiles (top and bottom lines), 10th and 90th percentiles (top and bottom
whiskers), and observations lying outside of the 10th to 90th percentiles (dots).
40 S ---
20 -- -- - -
0--- --------------^~ i ~~--- -i --
I I t I I-- -= ~ :~
200 -]----------..-..--- --.....
100 -i ,
800 0-- High intensity burn
--- Low intensity burn
600 --- Plant removal
-,- Gap control
400 - Forest
Months after treatment
Figure 3-2. Extractable soil concentrations of P, Mg, Ca, and K in soil samples
(0- 8 cm depth) in four gap treatments and forest sites at 5 sampling times
over an 18 month period following bums (bars = S.E.).
~ ~~... .....-........-......... .-4
9 12 18
Months after treatment
Figure 3-3. Extractable P, Mg, K, and Ca in soil sampled from 8-20 cm in four gap
treatments and forest sites at 9. 12, and 18 months following bums. Y-axis scales are
identical to those in Figure 3-2 for soils sampled from 0-8 cm depth (bars = S.E.).
-*- High intensity bum
-0- Low intensity bum
--v- Plant removal
-V- Gap control
- - -- - -
Table 3-1. Results of ANOVAs of soil nutrients, organic matter, water content, and
soil pH of soil sampled 0-8 cm in four gap treatments and forest plots at 5 times
following bums. All variables were log transformed. Where a significant time *
treatment interaction was found, variables were analyzed separately by month.
Treatments with different letters are significantly different at P < 0.05.
No interaction time treatment
Variable Factor F
Post-hoc test results
P Month high low remove control forest
Magnesium Treatment 23.3 <0.001
Time 8.7 < 0.001
Treatment 70.1 < 0.001
Time 3.0 0.026
Water content Treatment 3.9 0.008
Time 390 <0001
Significant time treatment interaction
Variable Month F P
77.4 < 0.001
7.8 < 0.001
167.5 < 0.001
58.7 < 0.001
37.2 < 0.001
7.3 < 0.001
12.6 < 0.001
28.8 < 0.001
16.9 < 0.001
3 a b c c c
6 a b bc c c
9 a b bc c c
12 a b c c c
18 a b b b b
3 a b c c c
6 a b bc c bc
9 a b b b b
12 a b b b b
18 a a a a a
3 ab a a a b
6 ab ab a ab b
9 a a a a b
12 a a a a a
18 a a a a a
Post-hoc test results
high low remove control forest
a b c c c
a bc c bc b
a bc c b bc
a b b ab a
ab b b ab a
a b c c c
a a b b b
a b c c c
a b c c c
a b c c c
a b b b b
a b b b b
a b b ab ab
a b ab ab b
ab ab b ab a
b b b
a b c
b c c
b c c
Table 3-2. Results of ANOVAs of soil nutrients, organic matter, water content, and
soil pH in the 8-20 cm depth of soil of four gap treatments and forest plots at 3 times
following bums. All variables were log transformed. Where a significant time *
treatment interaction was found, variables were analyzed separately by month.
Treatments with different letters are significantly different at P < 0.05.
No interaction time treatment
Post-hoc test results
P Month high low remove control forest
Phosphorus Treatment 105.4 < 0.001
59.4 < 0.001
Magnesium Treatment 2.1 0.09
Treatment 10.5 < 0.001
Water content Treatment 5.6 0.001
Treatment 70.9 <0.001
Time 58.9 < 0.001
Significant time treatment interaction
9 34.6 < 0.001
12 17.5 < 0.001
18 9.5 <0.001
9 a b bc bc bc
12 a b bc bc bc
18 a b b b b
9 a b b b b
12 a ab b b b
18 a b ab b b
9 a a a a b
18 ab ab a ab b
9 a b bc c bc
12 a b c c c
18 a b c c c
Post-hoc test results
high low removal control forest
a b c bc bc
a b b b b
a b b b ab
Organic matter 9
2.7 < 0.001
a ab ab b ab
b b a c c
Total soil N was strongly related to soil carbon (R2= 0.93; Figure 3-4) thus, patterns of
total N differences among treatments are expected to follow those for soil organic matter.
Both high- and low-intensity bum treatments significantly increased resin-available NH-
N, N03'-N, and P04-3-P during the first rainy season following bums (Table 3-3, Figure 3-5). This
pulse decreased after the first rainy season. Other than an increase in N03-N in plant removal
treatments during the first rainy season, the remaining treatments had little effect on NI--N, N03-
N, and PO3-P availability.
Coefficients of variation (CV) calculated for soil nutrients and organic matter within one
plot and among the 16 plots for each treatment are displayed in Table 3-4. A pattern emerged that
in the burned treatments, soil nutrients and organic matter were more variable within the one plot
than among all 16 plots. The opposite pattern was true for the plant removal, gap control, and
forest plots. Variation was in general greater among the 16 different plots than within the one plot.
Treatment effects on soil pH, soil organic matter content, and soil water content
Soil pH after high-intensity bums at 0-8 and 8-20 cm was higher than in all other
treatments throughout the 18 month sampling period (Figure 3-6 and 3-7). Soil pH in high-
intensity burn plots was 2 pH units higher than forest soils 2 months following bums. In low-
intensity bum treatments, pH was higher than in the remaining treatments at both depths at all
sampling periods. The plant removal and gap control treatments had little effect on soil pH.
High-intensity burn treatments significantly lowered soil organic matter; 2 months
following high-intensity burns soil organic matter in the top 8 cm of soil was approximately 72%
that of forest soils. By 18 months soil organic matter recovered to levels comparable to the
remaining treatments. Differences among the remaining treatments were small and varied
throughout the sampling periods.
0.7 R2 = 0.93
,-- 0.6 -
" 0.4 -0
H 0.2 -
0 1 2 3 4 5 6 7
Total soil carbon (%)
Figure 3-4. Total soil nitrogen and soil carbon in soil sampled 0-8 cm depth.
Soil samples used for this analyses were chosen from among all treatments
and all times since bums (2, 6, 9, 12, and 18 months) to obtain as wide a range
as possible for carbon content.
Table 3-3. Results of ANOVAs of resin exchangable NH4, NO3, and P04 in 4 gap treatment
and forest plots measured at 3 time periods following bums. All variables had a significant
time*treatment interaction and therefore all were analyzed separately by month. All variables
were log transformed prior to analyses.
at P< 0.05.
Treatments with different letters are significantly
Analyzed by time period
NH4-N first wet season
first dry season
second wet season
N03-N first wet season
first dry season
second wet season
P04-P first wet season
first dry season
second wet season
Post-hoc test results
high low removal control forest
a b c c c
a a ab be c
a ab ab b b
a a b b b
a bc c b ab
a bc ab ab ab
> 0.012 -
ir 0.010 -
-0 0.008 -
b 0.006 -
0 .0 0 4 .. .. ..... ....... . . .
0 0.002 --- -
1 0.6 -
'SP 0.4 -
Z 0.2 -" -
0. o-0- High intensity burn
c--G Low intensity bum
0o.3 -- --- Plant removal
o -V-- Gap control
S0.2 *.. Forest
first rainy season first dry season second rainy season
Figure 3-5. Resin-available ammonium, nitrate, and phosphate determined from exchange resins
buried in soil at 5 cm depth during 3 periods following treatments. Resins were buried for
approximately 3 months during each period. Time since bum of sampling periods were:
1st rainy season (2-5 mo), 1'tdry season (8-11 mo), and 2nd rainy season (15-18 mo; bars = S.E.).
Table 3-4. Coefficients of variations (C V) of soil cations, organic matter, and resin-available nitrogen sampled within a
100 m2 plot ("within") and among the 16 different plots ("among") of each treatment. CV within plots was calculated
from 4 samples taken from each treatment in one block. CV among plots was calculated from the composite samples
taken from each treatment of 16 blocks. All calculations were performed with data taken 3 months following bums.
High intensity burn Low intensity burn Plant removal Gap control Forest
within among within among within among within among within among
Ca 39 18 42 35 28 37 15 53 36 46
Mg 40 36 53 32 19 22 7 31 17 35
K 47 21 53 30 34 31 21 49 17 39
P 84 39 66 38 8 27 23 29 14 54
OM 33 24 37 20 24 25 8 35 24 19
N03-N 27 36 25 71 49 58 75 51 48 51
NH4-N 49 160 54 39 41 114 53 18 41 20
ash depth 46 22
a CV = (standard deviation / mean) 100 (Sokal and Rohlf 1981)
* = coefficient of variation greater within a single treatment plot than among all treatment plots.
4 5 1 -. High intensity bum
--- -0- Low intensity burn
_,_ -- Plant removal
1.0 .-v- Gap control
7M a ^--* Forest
I '5' ^==^^ ^ ----------
7 7-""" .
6-~ ~ -a---- _.....
0 8 ...... ........... ..
0 I-I I--
( 1 -
2 6 9 12 18
Months after treatment
Figure 3-6. Soil pH, air-dry water content, and organic matter content measured in soil (0-8 cm depth)
in the four gap treatments and forest sites at 5 sampling times over an 18 month period following bums.
Soil strength was measured at the soil surface with a soil penetrometer at 4 sampling periods over a 12
month period following bums. Water content is expressed as a percentage of air-dry weight (bars = S.E.).
1 ^.-_ ~ -_--
Months after treatment
Figure 3-7. Soil pH, air-dry water content, and organic matter content in soil sampled
from 8-20 cm depth in four gap treatments and forest sites at 9, 12, and 18 months
following bums. Y-axis scales are identical to those in Figure 3-6 for soil sampled from
0-8 cm depth (bars = S.E.).
-- High intensity burn
--0- Low intensity burn
-V Plant removal
--v- Gap control
-. ..a Forest
Although significant differences in soil water content were detected among treatments,
differences were not large and patterns were not consistent over the sampling period. The forest
plots had the lowest soil water content during the first 9 months, but this difference diminished
after 12 months. Larger differences were due to seasonal changes in soil water content, predictably
with the highest water contents in the rainy season of 1999, and the lowest water contents in the
drn season of September 1998.
Treatment effects on fine root mass
Fine root mass 18 months following bums was significantly higher in forest plots than in
high- and low-intensity bum treatments (F = 4.4, P = 0.005, n = 10). Fine root mass (live and
dead combined) in the top 7 cm of soil in was: forest 1.8 0.3 kg/m2, gap control 1.3 0.2 kg/m2,
plant removal 1.3 0.2 kg/m2, low-intensity bum 0.9 0.1 kg/m2, and high-intensity bum 0.7
Treatment effects on soil physical properties
Soil bulk density and soil strength. Bulk density in high-intensity bum treatments was
significantly higher than in forest plots after 6 and 12 months (F = 3.1, P = 0.02, n = 10). There
were no significant differences among the remaining treatments. Bulk density averaged 1.3 0.05
g cm-2 in high-intensity bum treatments and 1.2 0.02 g cm"2 in forest plots after 12 months.
Soil strength in high-intensity bum treatments increased during the first 12 months
following burns (Figure 3-6, Table 3-1). Although at 3 months, soil strength was lowest in high-
intensity bum treatments, it was the highest in this treatment by 9 months. Soil strength in the other
treatments also increased during the first 12 months, but to a less extent than in high-intensity bum
Water infiltration and soil wettability. Water infiltration rates were significantly lower
in the high-intensity bums than in the remaining treatments (F = 31, P < 0.001, Figure 3-8). Soil
wettability significantly differed among treatments (F = 4.6, P = 0.002) and a slight, but non-
significant, difference was found in the wettability of different soil depths (F = 2.4, P = 0.07).
Surface soils in all treatments except the high-intensity bum tended to repel water (Figure 3-9). In
the high-intensity bum treatment, a slightly water-repellent layer was detected at 2-3 cm depths.
Effects of soil heating and ash addition on soil chemical properties
Field study. Due to high seed predation of A. colubrina, I only report results of soil
analyses. Ash addition significantly increased soil concentrations of P, K, Ca, and Mg as well as
soil pH (Table 3-5; Figure 3-10). Ash addition lowered soil organic matter, but did not affect
available resin-available NH4-N and N03-N. Heated soil had lower concentrations of Mg, but did
not have significantly different concentrations of P, K, Ca, organic matter, or resin-available
NH4-N and N03-N.
Greenhouse study. Both soil heating and ash addition decreased growth of C pluviosa
seedlings (Table 3-6; Figure 3-11). Intensely heated soil had higher levels of resin-available NW4-
N than moderately heated or control soil. Resin-available NO3'-N was not detectable in intensely
heated soil, but it was significantly higher in moderately heated soil than control soil. Ash addition
did not affect resin-available NH4-N and N03-N.
Effects of high- and low-intensity burns on soil chemical properties
Controlled bums significantly affected on all soil chemical properties examined (soil pH,
soil organic matter, resin-available N and P, and Mehlich-extractable P, Ca, K, and Mg). These
changes, attributable to soil heating and/or ash deposited during bums, were greater after high-
intensity bums than low-intensity bums. High fuel loads combined with relatively complete bums
in the high-intensity bum treatment resulted in an average ash depth of 4.8 cm. Maximum
temperatures reached during high-intensity bums averaged 704 C at the soil surface and 227 C at
3 cm depth (Chapter 2). Little ash was deposited after low-intensity bums, due mostly to the lower
--0- High intensity burn
-0- Low intensity bum
0.16 Plant removal
-* ,-- Gap control
N' i**" Forest
S 0.08 ....... ......
5 10 15
Cummulative volume of water (ml cm-2)
Figure 3-8. Water infiltration rates of soil in four gap treatments and forest plots.
Infiltration was measured as the time required for the first 5 ml of a 10 ml column
of water to infiltrate soil.
.g Ts/ i ---
" --0-^ ^ high intensity bum
2-- low intensity bum
S/ plant removal
S-/ 7- gap control
/ U- forest
0 5 10 15 20 25
Time for total absorption (s)
Figure 3-9. Soil wettability at the soil surface soil and at 1, 2, and 3 cm depths in four gap
treatments and forest plots. X-axis refers to the time to total abortion of 5 drops of
water applied to the soil surface with a dropper.
Table 3-5. Results of ANOVAs of a field experiment examining the separate effects of
ash addition and soil heating on soil properties. Two levels of ash (no ash, 500 g ash) and
3 levels of soil heating (no heat, low intensity heat, and high intensity heat) were applied
to 0.25 m2 subplots in the field and soil sampled after 3 weeks. In the low intensity heat
treatment, a flame of 150-250C was applied to the soil surface of the treatment area for
5 minutes with a propane torch. In the high intensity heat treatment, a flame of 500-800C
was applied to the soil surface of the treatment area for 20 minutes. Ammonium and nitrate
were measured using anion and cation exchange resins buried at a depth of 5 cm for 3
months (n = 12).
* log transformed prior to analyses
control ash added
control ash added
z 0.0 -
E low heat
- high heat
control ash added
control ash added
control ash added
control ash added
control ash added
Figure 3-10. Results of field study of soil that had received combinations of heating (no heat, low
intensity heat, and high intensity heat) and ash addition (no ash, ash added). All soil was sampled
from 0-8 cm depth, 3 weeks after treatments. Resin available NH+ and N03 were measured
using anion and cation exchange resins buried to a depth of 5 cm for 3 months (n=12; bars=S.E.).
control ash added
Table 3-6. Results of ANOVAs of a greenhouse experiment examining the separate effects o
ash addition and soil heating on available nitrogen and seedling growth of Caesalpinia
pluviosa. Two levels of ash (no ash, 15 g ash added) and 3 levels of soil heating (no heat, lov
intensity heat, and high intensity heat) were applied to soil used to fill planting bags. In the lo,
intensity heat treatment, soil was heated in a conventional oven at 150-200 C for 10 minutes.
In the high intensity\ heat treatment, soil was heated in a conventional oven at -250 C for 40
minutes and fired with a propane torch for 5 minutes at a flame temperature of 500-800 C.
Ammonium and nitrate weremeasured using anion and cation exchange resins buried at a depth
of 5 cm for 3 months. Seedling height was measured 4 months after planting (n = 12).
Ash Heat Ash heat
Variable df F P F P F P
NH4-N 5 6.2 0..023 291.0 < 0.001 1.8 0.20
N03-N 5 1.7 0.20 23.3 < 0.001 0.6 0.58
Seedling height 4.8 0.012**
*Significant interaction for seedling height therefore analyzed separately by heat treatment.
Heat treatment F P tallest
Seedling height no heat 11.4 0.003 w/o ash
low heat 11.7 0.002 w/oash
high heat 0.15 0.7
control ash added
Figure 3-11. Results of greenhouse study of Caesalpinia pluviosa seedlings planted in
soil that been heated (no heat, low intensity heat, and high intensity heat) and had ash
added (no ash, 15 g ash added). Seedling height was measured 4 months after planting
(n = 12). Ammonium and nitrate were measured using anion and cation exchange resins
buried in individual bags and watered for 22 days (n = 3). Bars are S.E.
fuel loads but also to incomplete combustion of these fuels. Maximum temperatures
reached during low-intensity bums averaged 2250 C at the soil surface; elevated temperatures at 3
cm depth were mostly undetectable.
Soil pH. Increased soil pH is a general effect forest fires (e.g., DeBano et al. 1977,
Wright and Bailey 1982, Kutiel et al. 1990, Hungerford et al. 1990, Stromgaard 1992, Neary et al.
1999). High concentrations of basic cations in ash deposited following fires (e.g., Ca, K, Mg, Na)
is the major mechanism of increased soil pH (Kutiel et al. 1990, Stromgaard 1992). Although soil
heating may also increase pH by releasing basic cations from soil organic matter (Giovannini et al.
1990), results of the soil heating and ash addition field study (Figure 3-11) revealed that ash
addition significantly increased soil pH while soil heating had only a slight but non-significant
Soil organic matter. High-intensity bums caused a net loss of organic matter in surface
soils, a predictable consequence of intense fire (e.g., DeBano et. al. 1977, Hungerford et al. 1990,
Neary et al. 1999). Several studies conducted in the tropics have found decreased soil organic
matter following slash burning (Amazonia: Uhl and Jordan 1984, Mackensen et al. 1996,
Australia: Rab 1996). Experimental studies have shown that soil organic matter loss is a direct
effect of soil heating (e.g., Hosking 1938, in Humphreys and Craig 1981, Giovannini et al. 1990),
with distillation of volatile organic compounds occurring between soil temperatures of 100-300 C
and near complete loss of soil organic matter at temperatures >450 C. Soil organic matter
contents following high-intensity bums (Figure 3-6), averaged over 0-8 cm, likely do not reflect
larger losses in soil organic matter that occurred in the first several centimeters. Consumption of
soil organic matter was probably complete at the soil surface during the high-intensity bums,
where maximum soil temperatures averaged 683C. Due the sharp decrease in soil temperature
with depth (Chapter 2), organic matter consumption was probably negligible below the top several
cm of soil and not detected at 8-20 cm depth.
Average surface temperatures during low-intensity bums (160 C) were not hot enough for
the consumption of soil organic matter, hence average soil organic matter contents of low-intensity
bum plots were not significantly lower than those of forest soils. In fact, average soil organic
matter contents after low-intensity bums were higher than those of adjacent forest soils. Increases
in soil organic matter have been shown to occur during light to moderate bums (e.g., Hungerford et
al. 1990) due to the incorporation ofunbumrned or partially burned slash fragments into soil. For
example, Stromgaard (1992) attributed increased soil carbon following slash burning in miombo
woodlands to charcoal accumulation or small organic particles washed in from ash.
Soil organic matter in high-intensity bum plots recovered to levels higher than those in
forest plots within 18 months following bums. High daytime soil temperature in burned plots and
high soil moisture within gap treatments may have contributed to this rapid recovery by increasing
decomposition rates. Though generally rapid, recovery of soil organic matter following slash and
bum vary among and within tropical forests. For example, at the same site in the Venezuelan
Amazon, Montagnini and Buschbacher (1989) reported recovery of soil organic matter within 6
months, while Uhl and Jordan (1984) reported that recovery required 5 years. Other than
differences in climate and site productivity, fire intensity and land use following burns also affect
recovery of soil organic matter, thus making comparisons among studies difficult.
Total soil nitrogen. Total soil N was linearly related to soil organic carbon, hence the
greatest declines in soil N occurred in high-intensity bum plots where organic matter in the top 8
cm of soil decreased an average of 28% from adjacent forest soils. Similarly, total soil N was
reported to decrease following slash and bum of tropical forest in Costa Rica (Ewel et al. 1981)
and in the Venezuelan Amazon (Uhl and Jordan 1984). As with soil organic matter, this average
(28%) underestimates N losses from the top several centimeters of soil or from more intensely
burned patches; N losses reached 84% in the top 8 cm of soil and likely approached 100% in
scorched surface soils. Comparably high N losses were reported in chaparral soil heated to 500 C
(80% N loss; Dunn and DeBano 1977) and Mediterranean soils heated to 600 C (86% N loss;
Kutiel et al. 1990).
Losses of total soil N during low-intensity burns were negligible as indicated by low losses
of soil organic matter. In fact, slightly higher soil organic matter contents in low-intensity bum
plots relative to adjacent forest soil suggest that total soil N increased after this treatment. This is
likely due to the mixing of slash fragments into surface soils. Increases in total N in surface soils
has also been found following slash burning in the tropical forests (Montagnini and Buschbacher
1989, Stromgaard 1992) and temperate forests (Gholz et al. 1985).
Resin-available nitrogen. In contrast to decreases in total soil N, amounts of resin-
available nitrogen (NH4-N and NO03-N) increased after bums of high-intensity. These findings
agree with those of Matson et al.(1987) and Montagnini and Buschbacher (1989), who also
reported increases of ammonium and nitrate following slash burning of tropical forest in Costa
Rica and Venezuela, respectively. In the present study, low-intensity bums increased resin-
available N levels as well, although to a lesser degree than high-intensity bums. Temperate zone
studies have also noted that increases in inorganic N are dependent on fire intensity (Dunn and
DeBano 1977, Giovannini etal. 1990, Kutiel et al. 1990, Rice 1993, Weston and Attiwill 1996,
McMurtrie and Dewar 1997). Dunn and DeBano (1977) demonstrated that the greatest increases
in ammonium and nitrate for chaparral soils occurred at soil temperatures up to 300 C, due to the
mineralization of organic N. At soil temperatures of>500 C, inorganic N decreases due to
volatilization (Dunn and DeBano 1977). Similar results were reported in soil heating studies
conducted by Giovannini et al. (1990).
Increased ammonium availability following bums may be enhanced by soil microbial
death, which occurs at temperatures as low as 50-121 C (Neary et al. 1999). Soil microbial
death was likely substantial in high-intensity bum plots and may have occurred within small well-
burned patches in low-intensity bum plots as well. In their study of nitrogen transformations
following slash and bum on volcanic soils in Costa Rica, Matson et al. (1987) found that the
amount of nitrogen that disappeared from microbial biomass after bums was similar to the
concurrent increase in apparent net nitrogen mineralization.
Matson et al. (1987) also attributed increased nitrate concentrations to enhanced
nitrification rates. Burning generally creates favorable conditions for nitrification, such as raised
pH values and base saturation (Pritchett and Fisher 1987). Increased nitrification rates were also
reported following slash burning of Venezuelan forests (Montagnini and Buschbacher 1989). In
contrast, other studies have shown that nitrification rates are reduced by fire, due to a decreased
biomass ofnitrifiers (e.g., Dunn and DeBano 1977, Stromgaard 1992). Reduced nitrification rates
would cause an accumulation of ammonium, which is less subject to leaching than nitrate. This
effect may explain why nitrate was undetectable in the intensely heated soil of the greenhouse study
while levels of ammonium were very high (Figure 3-12).
Elevated concentrations of ammonium and nitrate following high and low-intensity bums
were short-lived in my study. Within 8 months of bums (after the first rainy season), inorganic N
in burned plots declined to levels found in adjacent forest. This result is similar to the rate of
decline of inorganic N following slash burning of a wet forest in Costa Rica (Matson et al. 1987).
Phosphorus. High- and low-intensity bums increased both extractable P and resin-
available PO43-P. Inorganic P additions in ash likely contributed to these increases. In the soil
heating and ash addition field study (Figure 3-11), ash addition significantly increased extractable
P, while soil heating had little effect. Similarly, Stromgaard (1992) attributed increases in
extractable P after slash burning of miombo woodlands to ash deposition, and Rice (1993) found
that soil P043-P concentrations in Californian chaparral following fire were correlated with ash
depth but not fire intensity.
Though possibly less important than ash deposition, soil heating can increase extractable P
by mineralizing organic P, as may have occurred with resin-available N. Giovannini et al. (1990)
found an increase in inorganic P accompanied by an equivalent decrease in organic P in soil
samples heated to 4600 C; at temperatures >460C, all organic P was destroyed and only inorganic
P remained. Soil heating may have been comparatively more important in low-intensity bum
treatments, where relatively little ash was deposited. Other studies have reported increases in
inorganic P following fires of low-intensity. For example, in a study of soil nutrient levels
associated with shifting agriculture in the Asian tropics, Andriessse and Koopmans (1984) found
available P increased almost 300% after heating to 200 C, which they attributed to mineralization
of organic P.
Cations. High-intensity bums significantly increased soil concentrations of extractable
cations (Ca, K, and Mg). Low-intensity burns also increased cation concentrations, although not
as dramatically. Similar to P, results of the soil heating and ash addition field study suggest that
increases in extractable cation concentrations by burning is mostly due to ash deposition.
Significant increases in extractable Ca, Mg, and K after fire in miombo woodland (Stromgaard
1992) and Brazilian cerrado (Coutinho 1990) were also attributed to ash deposition.
As with P, soil heating may also increase extractable Ca, K, and Mg through
mineralization of organic forms (Giovannini et al. 1990). Extractable Ca and Mg peaked in soil
heated to 200 C and declined at higher temperatures; extractable K peaked in soil heated to 700 C
(Giovannini et al. 1990). Results of the soil heating and ash addition study conform with this
pattern; soil heating significantly decreased extractable Ca and Mg concentrations and had no
effect on extractable K concentration.
Decreases in cation concentrations over the 18 month post-bum period are similar to those
following bums in temperate forests (e.g., DeBano et al. 1977, DeRonde 1990, Kutiel and Shaviv
1992, Hemrnandez el al. 1997) and tropical forests (e.g., Uhl and Jordan 1984, Coutinho 1990,
Stromgaard 1992, Mackensen 1996). The order of decrease (K> Mg > Ca) corresponds with
cations' mobility and susceptibility to leaching. In high-intensity bum plots, plant uptake was
probably not important in cation decreases during the first year, as plant cover remained less than
25% (Chapter 5). Plant uptake may have been more important during the second year, as
vegetative cover reached 60% after 18 months.
Variation of soil nutrients within and among plots. Based on the comparison of one
plot. soil nutrients and organic matter in burned treatments appeared to be more variable within
plots than among plots of the same treatment (Table 3-4). This pattern suggests that natural
variation in soil nutrients was increased due to heterogeneity of bums. The opposite pattern was
true for the plant removal, gap control, and forest plots. Variation in soil properties was in general
greater among the 16 different plots than within the same 100 nm2 plot, suggesting variations in soil
fertility in the absence of fire are expressed at larger scales.
Actual variation in soil properties after high-intensity bums may have been greater than
reflected by random sampling. For example, nutrient and organic matter contents of severely
scorched soil differed greatly from averages of high-intensity bum plots. Mineral concentrations of
such scorched soil sampled to 5 cm (with percentages of high-intensity bum plot averages) were:
6415 mg/kg Ca (133%); 197 mg/kg Mg (56%); 71 mg/kg K (11%); 0.25 mg/kg P (0.3%); and
0.1% organic matter (4%). These extremely scorched soils were not common (< 1% high-intensity
bum plots), but potentially affect plant colonization by providing microsites different from less
intensely burned areas.
Increased soil heterogeneity after bums has been observed by other authors. Christensen
(1985) noted that soil nutrient concentration in chaparral is considerably more variable after fire
than before, due to local variation in fire intensity and the uneven distribution of ash.
Heterogeneity in soil nutrients potentially has important consequences for colonizing plants. For
example, Rice (1993) observed that even small scale patterns in fire intensity and ash distribution
were reflected in later establishment of chaparral shrubs.
Changes in fine root mass following burns
Lower fine root biomass after high-intensity bums was likely due to a combination of rapid
decomposition of dead roots as well as direct oxidation by fire. Experimental studies have shown
that fine roots are desiccated or killed at soil temperatures of 48-54 C (Neary et al. 1999).
Temperatures in high-intensity bum plots (61-399 C at 3 cm depth) were not only well above this
range, but were likely high enough in places to completely oxidize fine roots. Even during low-
intensity burns, temperatures in surface soils (160 C average at soil surface) were sufficiently high
to kill fine roots. Fine root mortality during bums was likely greater than death of larger roots, not
only because of their small size, but also due to their concentration near the soil surface.
Fine root mortality potentially has important effects on soil fertility, as their decomposition
may increase soil nutrient concentrations. Nutrient input from roots has been hypothesized to be
an important pathway for nutrient cycling, particularly in tropical dry forests, due to their larger
store of biomass below-ground (Martinez-Yrizar 1996, Jaramnillo and Sanford 1995). Decreased
fine live-root mass is also expected to contribute to higher water and nutrient availability due to
reduced uptake. However, average soil moisture contents in high-intensity bum plots were not
significantly different than those in other gap treatments. Possibly, lower water uptake was offset
by decreased water holding capacity of intensely bum soil caused by a loss of soil organic matter.
Effects of high- and low-intensity burns on soil physical properties
Changes in soil strength, bulk density, and water infiltration rates in high-intensity bum
plots were substantial. The decrease in soil organic matter in high-intensity bum plots likely
influenced these observed changes in soil physical properties. Organic matter influences soil
structure through aggregate formation; a decrease in organic matter decreases total porosity,
particularly macro-pore spaces (> 0.6 mm). The increase in surface soil strength during the first
year following high-intensity bums was likely due to the settling of soil minerals and ash into
spaces left void by organic matter and fine roots. This settling of soil particles would also
contribute to higher soil bulk densities.
Decreased macro-pore space would also contribute to the lower infiltration rates observed
in high-intensity bum plots. However, these lowered infiltration rates caused by high-intensity
bums did not result in any observable surface runoff. The lowest infiltration rate recorded in a
high-intensity burn plot was 5 times faster than the rate needed to absorb a 5 cmhr" rainfall (0.002
cm3cm2sec"). Further, most plots were located on level ground, therefore if a rain event exceeded
the soil's rate of infiltration, the accrued water would not run-off.
Increased wettability of surface soils after high-intensity bums conforms to studies that
report soil temperatures >288 C destroy water-repellent layers (Neary et al. 1999). Soil
temperatures of 176-288 C reportedly form water-repellent layers (Neary et al. 1999), explaining
the presence of a slightly water-repellent layer at 2-3 cm depth. However, the decreased wettability
of this soil layer does explain the lower infiltration rates in high-intensity burn treatments, as
surface soils in the remaining treatments had similar wettability properties. Possibly, a more
water-repellent layer was formed deeper than 3 cm in high-intensity bum plots, but was undetected
due to the sampling strategy.
Soil strength, bulk density, water infiltration, and water repellency of low-intensity bum
plots were not different from those in the unburned treatments. Again, this pattern may reflect the
influence of organic matter on soil physical characteristics; the lower temperatures during low-
intensity burns (mean 120 C), did not decrease soil organic matter.
Potential effects of high- and low-intensity burns on tree seedling growth
Soil heating and ash addition significantly affected Caesalpinia seedling growth, although
in a manner opposite than expected; ash addition decreased seedling growth. This result suggests
that the quantity of ash added to soil may have been at toxic levels for this species. Also, tree
seedlings were shorter in soil heated at both low and high intensities. This result only partially
corresponds with a similar study by Giovannini et al. (1990), which examined the effects of soil
heating on wheat (Triticum aestivum) seedling growth. They found that while soil heated to 170
C had no effect on plant growth, soil heated to 220 C and 460 C increased seedling height and
biomass, whereas soil heated to 700 C and 900 C had detrimental effects on seedling growth.
The authors attributed increased growth in moderately heated soil to greater ammonium and
available phosphorus concentrations. Lowered growth in intensely heated soil was attributed to the
sharp increase in soil pH and release of Ca and K to toxic levels, as seedlings in this treatment of
their study showed symptoms of nutritional disorder.
In my greenhouse study, decreased seedling growth in intensely heated soil may be the
result of degraded soil structure or toxic levels of cations. However, it is unclear why seedling
growth was impaired in the lightly heated soil. It is important to consider that although seedling
heights significantly differed among treatments, maximum height differences were only 3 cm. This
slight difference after 4 months of growth may not be biologically significant. Perhaps the effects
of soil heating and ash addition on seedling growth would have been more apparent if a shade-
intolerant species had been used as a bioassay. Caesalpinia is partially shade-tolerant and
exhibited slow growth rates in the field as well (Chapter 4).
Despite the potentially negative effects of increased bulk density and soil strength, lowered
infiltration rates, and possibly toxic effects of cations on plant growth, seedling heights of shade-
intolerant species were greatest in high-intensity bum plots (Chapter 4). This increased growth in
intensely burned soils may be due to several factors. Initially, soil strength in high-intensity burn
plots was the lowest of all treatments, therefore early colonizing seedlings should not have
experienced mechanical impedance of root growth. Secondly, nutrient concentrations were highest
in high-intensity burn plots which may have offset decreased movement of nutrients through the
soil. Also, toxic levels of cations may only have been a factor in small areas of high ash deposition
or severely scorched soils; seedlings may not have been able to establish in these small areas and
therefore the effects on growth were not observed. Most importantly, the density of plants
colonizing high-intensity bum plots was low (Chapters 5), so that established tree seedlings likely
benefited from reduced competition for soil water and nutrients.
Effects of plant removal and canopy gap formation on soil chemical and physical properties
Soil moisture content was higher in all of the gap treatments than forest plots for the first 9
months following bums. Higher soil moisture within tree fall gaps than under adjacent forest is a
pattern repeatedly found in tropical forest studies (e.g., Vitousek and Denslow 1986) and has been
attributed to decreased transpiration within gaps due to less vegetation. The difference in soil
moisture content between forest and gap plots diminished over the first year as the amount of
vegetation in gaps increased.
Plant removal and gap control treatments did not significantly change soil chemical or
physical properties from those in adjacent intact forest. Although it is hypothesized that the
increased soil temperatures, moisture, and litter depth in tree fall gaps will increase nutrient
availability (i.e. Bazzaz 1980), conclusive evidence to suggest this is true has not been reported.
For example, in a study of natural tree fall gaps in lowland moist forest in Costa Rica, Vitousek
and Denslow (1986) found that nitrogen mineralization did not increase in tree fall gaps and slight
phosphorus increases were not significant. The only difference they detected was within gap
microhabitats; the root throw zone had significantly less N and P than the crown zone. Luizao et
al. (1998) found similar results in a study of artificial gaps ranging in size from 40-2500 m2 in
Brazilian rain forest. No differences in microbial biomass, soil respiration, and nitrogen
mineralization or nitrification were found between gap and forest sites.
Most of the variation observed within the plant removal, gap control, and forest plots over
time was due to seasonal changes. Soil moisture content varied predictably with changes in rainfall
and N03 availability declined slightly during the dry season. This observation agrees with the few
studies of nutrient cycling conducted in tropical dry forests which have shown that nitrification
rates are highest during the rainy season and lowest at the end of the dry season (Singh et al. 1989,
Garcia-Mendez et al. 1991, see also Smith et al. 1998).
Longer term effects of controlled burns on soil properties
The duration of this study limits its conclusions to only short-term treatment effects. A
similar study conducted in Las Trancas in 1995 (Stanley 1995) however, reveals slightly longer
term effects of burning on these forest soils. Although not identical, the treatments applied in this
earlier study were comparable to those used in mine: a gap control; gap vegetation slashed and
removed; gap vegetation slashed and burned; and gap enlarged by 30%, vegetation slashed, and
burned (Stanley 1995). Fires in the enlarged gaps were likely more intense than fires in gaps that
were not enlarged, due to a greater amount of fuel. I measured soil nutrient concentrations and soil
pH at 0-8 and 8-20 cm depths in April 1998, 3 years following experimental bums and found no
significant differences among gap treatments at either depth (Figure 3-12). The soil property with
the most distinct trend was P concentration (P = 0.11), followed by Ca concentration (P = 0.19),
soil pH (P= 0.22), K concentration (P = 0.27), Mg concentration (P = 0.32), and organic matter
content (P = 0.35).
These results suggest that soil chemical changes following bums are relatively short-lived.
However, the lack of significant results may have been more indicative of the large variation found
within treatments rather than the lack of variation found among treatments. Felling gaps included a
diverse array of habitats, such as rock outcrops and stream side areas. Importantly, there was no
indication of declining soil nutrient concentrations in the cleared and burned areas of this pilot
study after 3 years. A widely held notion about the recovery of tropical ecosystems following
disturbance is that severe nutrient losses following deforestation limit forest regeneration (e.g.,
Allen 1985, Buchbacher et al. 1988). Clearly, longer-term sampling of the burned plots is needed
before conclusions can be drawn about the long-term sustainability of severely burned soils at this
* 0 0
5 Ui H C A
0 w1J -. /Ih O
Mg (mg kg")
k/i 0 k/i 0 kI 0
0 0 0
S. 4 00 - oo
0 00 00 0~ 0~ 0% 0
Ca (mg kg1)
--, M ,
0 0 0 0
P (mg kg1") K (mg kg"')
The experimental bums had mixed effects on physical, chemical, and biological soil
properties. High-intensity bums increased levels of available nitrogen and phosphorus, but these
pulses quickly decreased. Ash deposited after high-intensity bums increased cation concentrations,
but these pulses declined over time as well. Loss of organic matter during high-intensity bums
likely altered surface soil structure. This effect was apparent in increased bulk density and soil
strength, and decreased infiltration.
Low-intensity bums increased cation concentrations and available forms of N and P. Soil
structural changes were not as marked as after high-intensity bums, because little soil organic
matter was consumed. Therefore, soil chemical properties can be altered during low-intensity fires,
but changes in soil physical properties may only occur during intense fires. Plant removal, gap
control, and forest plots had no significant effects on soil chemical or physical properties.
Greenhouse studies suggested that loss of soil structure caused by soil heating and toxic
levels of cations created by large quantities of ash may hinder tree seedling growth. However,
these potentially negative effects were not apparent in the field. Shade-intolerant seedlings
established in high-intensity bum plots grew faster than those in all other treatments. Evidently, the
increase in soil nutrients caused by high-intensity burns was not offset by altered soil structure.
Importantly, the choice of tree species used as bioassays in these studies likely affected the results.
An important feature of the increase in resource availability produced by bums is the
transient nature of the increases. Therefore, the first plants to become established after bums
should benefit most from greater resource availability. Bums also greatly increased soil resource
heterogeneity. Therefore, successful establishment and vigorous growth of plants after bums may
be greatly influenced by chance, i.e., seeds dispersed into patches of well burned soil will exhibit
higher growth than those dispersed into unburned soil or severely scorched soil.
EFFECTS OF CANOPY GAP FORMATION, PLANT REMOVAL, AND
CONTROLLED BURNS OF HIGH AND LOW INTENSITIES ON
EARLY REGENERATION OF COMMERCIAL TREE SPECIES
The most basic principle of sustainable forest management is that rates of timber
harvesting should not exceed the rate at which timber volume accumulates (e.g., Johnson and
Carbarle 1993, Dawkins and Philip 1998). This criterion requires sufficient regeneration of
harvested species and, in many neotropical forests, poses the greatest barrier to sustainable forest
management (e.g., Wyatt-Smith 1987, Verrisimo 1995, Gullison et al. 1996). There are several
silvicultural means of improving poor regeneration, ranging from intensive techniques such as
prescribed burning, to less intensive techniques such as selective harvesting without further
treatment. Choice of technique must be knowledgeably based on the natural regeneration
requirements of the target species. Presently, lack of information of the autoecology of many
harvested tree species is one of the largest deterrents to sustainable timber production (Bazzaz
1990, Bazzaz and Pickett 1980, Fox 1976, Gomez-Pompa and Burley 1991, Hall 1996, Whitmore
For many decades, seedling ecology was a minor part of tropical forestry (Hall 1996);
more recently, tropical seedling ecology has become the focus of much ecological research (e.g.,
Garwood 1983, Augspurger 1984a, 1984b, Swaine 1996, Kitajima 1996, Fenner and Kitajima
1998). The seedling stage is critical to regeneration as most mortality occurs early in the life of a
tree (Lieberman 1996, Li et al. 1996). Understanding this life stage is also important for forest
managers as species vary widely in their seedling ecology (Hall 1996), arid the division of tree
species into regeneration guilds is based on seed germination and seedling establishment
requirements. The most well defined group are species that require high light conditions for seed
germination and seedling establishment and often colonize following disturbances (i.e., early-
successional, Bazazz 1979; pioneer or secondary, Budowski 1965; or shade-intolerant, Swaine and
Whitmore 1988). At the other end of the spectrum are species that can germinate and persist in the
low light of forest understories (i.e., late-successional, Bazazz 1979; climax or primary, Budowski
1965; or shade-tolerant, Swaine and Whitmore 1988).
Tree species with different regeneration strategies require different silvicultural treatments
to enhance their regeneration. For example, many shade-tolerant species have advanced
regeneration in forest understories (Brokaw 1985b, Hartshorn 1989), therefore management
techniques would mostly entail ensuring the survival of this regeneration during harvesting and
enhancing its growth to mature stages. Enhancing regeneration of shade-intolerant species that do
not have seedling banks in forest understories involves creating sites suitable for seed germination
and seedling establishment and promoting safe arrival of seeds to these sites (Dickinson 1998).
In the seasonally dry forests of Bolivia, commercial tree species are represented in both
shade-tolerant and shade-intolerant groups (Guzman 1997). Consequently, a mixed-management
system was proposed for the community-owned forests of Lomerio (Pinard et al. 1999). Single-
tree selection was recommended to enhance regeneration of shade-tolerant trees. Group selection,
harvesting groups of trees to foster the development of even-aged patches, was suggested to
improve regeneration of shade-intolerant species.
Prescribed burning of logging gaps has also been suggested as an additional treatment for
the management of shade-intolerant species (Stanley 1995). Prescribed bums may enhance
seedling establishment and growth of shade-intolerant species in a number of ways, including
removing litter, reducing logging slash, and slowing vine proliferation (Stanley 1999). Abundant
regeneration of shade-intolerant species after wildfire in Lomerio (Mostacedo et al 1999, Gould et
al. 1999) lends further support to the promise of prescribed bums as a silvicultural tool for the
management of dry forests in Bolivia.
Although the pioneer-climax dichotomy provides an often useful paradigm for ecologists,
regeneration strategies of many tree species fall between the two extremes of completely shade-
intolerant or shade-tolerant (Augspurger 1984b, Condit et al. 1996). In fact, most rain forest tree
species are both shade-tolerant and gap-dependent, meaning they have the ability to persist in a
seedling bank in forest understories but require canopy opening to reach maturity (Hartshorn
1989). And, as there is a continuum of species regeneration strategies, there is also a gradient of
disturbance intensities among potential silvicultural treatments. The interaction of individual
species' biology and silvicultural treatments of varying intensities is inherently complex. Before
management techniques can be prescribed on a large scale, the effects of these techniques on all of
the species in question need to be examined. For example, what are the effects of more intense
disturbances, such as high-intensity fires, on the advance regeneration shade-tolerant trees? Or,
what is the minimum disturbance level required for the regeneration of shade-intolerant trees?
In this chapter, I examine the range of responses of commercial timber tree species to
experimental canopy opening, above-ground biomass removal, and controlled bums of high and
low intensities and compare these responses to those in forest understories. My objective was to
determine which gap treatment provided the best conditions for the establishment and growth of
each species. I also address the question of whether regeneration of each of these species is limited
by seed dispersal or by sites suitable for their establishment and growth. I approached these
questions by studying the effects of gap treatments on seed germination, seed predation, vegetative
regeneration, and seedling and sprout growth and mortality. Finally, I discuss results in relation to
silvicultural management options for each species.
I conducted this study in the four gap treatments (high-intensity burn, low-intensity burn,
plant removal, and gap control) applied to the 400 m2 blocks described in Chapter 2. Trees were
sampled in the paired 4 m2 subplots positioned near the gap center and gap edge of each 100 m2
treatment plot as well as the paired forest subplots (see Figure 2-4, Chapter 2).
Seed addition treatment
One randomly selected subplot (4 m2) of each pair was assigned a seed addition treatment.
Seeds of 5 commercial timber species were used in this treatment: Anadenanthera colubrna,
Astronium urundueva, Centrolobium microchate, Copaifera chodatiana, and Schinopsis
brasiliensis (hereafter referred to by genera; Table 4-1). I chose these species based on their
commercial importance and seed availability in 1997. Seeds of all species were collected in the '95
coupe of Las Trancas from the forest floor and stored in cloth bags under a shelter until sown in
plots. I removed all seeds that appeared to be damaged by predators or fungi. Twenty seeds each
(= 5 seeds/m2) of Anadenanthera, Astronium, and Copaifera, and 10 seeds each (= 2.5 seeds/m2)
of Centrolobium and Schinopsis were placed in each plot on the litter or soil surface, depending on
the soil surface conditions in each treatment (Chapter 2).
Viability of collected seeds
Viability of seeds collected for the seed addition treatment was determined in germination
trials conducted over a 60 day period (November-January) in Santa Cruz. One-hundred seeds of
each of the 5 species were planted in 5 trays (20 seeds per tray) of a 50:50 mix of sand and soil.
Trays were placed in a location receiving morning shade and afternoon sun and were watered each
morning. Newly germinated seeds were counted and removed daily.
Table 4-1. Characteristics of tree species used in seed addition treatment or those species with sufficient regeneration for statistical
analysis. All but Acosmium are commercial timber species. Ecological classifiaction of shade tolerance of regeneration taken from
Pinardet al. 1998.
Species name Family tol. Dispersal unit Fruit characteristics Seed size
Acosmium cardenasii Papilinoideae 3 wind seed legume, 10 x 15 mm, 1 mm thick
Anadenanthera colubrina Mimosoideae 1 gravity seed legume, 10-25 cm long 12 x 10 mm, 1 mm thick
Aspidosperma rigidum Apocynaceae 3 wind seed pod, 6 x 5 cm 25 x 20 mm, 1 mm thick
Astronium urundeuva Anacardiaceae 1 wind fruit small dried drupe, calyx to 1 cm 3 mm diam.
Caesalpinia pluviosa Caesalpinaceae 2 gravity seed legume, 10-15 cm long 10 x 10 mm 2-3 mm thick
Centrolobium microcheate Papilinoideae 1 wind fruit samara, 8-10 cm long x 3-4 cm wide 12 x 2 mm
Copaifera chodatiana Caesalpinaceae 3 animal seed dry pod, 2 x 3 cm seeds with oily aril 10 x 5 mm, 4 mm thick
Schinopsis brasiliensis Anacardiaceae 1 wind fruit samara, 15-20 mm long x 5 mm wide 10 x 3 mm
Shade tolerance: 1 = shade intolerant; 2 = partially shade-tolerant; 3 = shade-tolerant
Density, height, and relative height growth rate measurements
Seedlings have been variously defined as individuals still dependent on seed reserves (e.g.,
Garwood 1996), to individuals up to 2.7 m tall (Whitmore 1996). In this study, I did not use size
or physiology as a defining character, rather I define seedlings as individuals originating from seed
as opposed to those regenerating as sprouts. I measured seedling and/or sprout density and height
in each 4 m2 subplot (both seeded and unseeded) at 1.5, 3, 6, 9, 12, and 18 months after bums. At
each sampling period, all commercial species within subplots (of both seeded and unseeded
species) were identified as sprouts or seedlings, tagged, and height to the apical meristem
measured. Sprouts originating from the stem or root collar were easily identified because scars
were visible. Root sprouts (root suckers) were more difficult to identify but were recognizable
because the first leaves generally differed from the first true leaves of seedlings. Relative height
growth rates (hereafter referred to as RGR) was calculated as:
RGR = [In (height t2) In (heightti)] / (t2 t)
where tj and t2 are two measurement periods. Seedlings of Anadenanthera were extremely
abundant in 1997, therefore a maximum of 3 randomly selected individuals per subplot were
tagged for height measurements and the remaining individuals counted. Additionally, a maximum
of 3 randomly selected individuals of Acosmium cardenasii were tagged in each plot and the
remaining individuals counted. Although Acosmium is not commercially valuable due to its
susceptibility to heart rot, I included it because it has the most abundant tree regeneration of any
canopy tree species in Las Trancas '95. Many of the Anadenanthera seedlings had been browsed
between the 6 and 9 month assessments, therefore I noted presence or absence of browsing of
Seedling densities of most tree species could not be normalized, therefore seedling densities
were analyzed using Kruskall-Wallis non-parametric tests (SPSS 1997). Separate tests were run
for each species by month, testing for the effects of the seeding treatment and gap treatments on
densities. For species that regenerated from both seeds and sprouts, these regeneration modes were
analyzed separately. Square-root transformed densities of Acosmium and Anadenanthera were
normally distributed, therefore densities of these species were compared using repeated measures
ANOVAs. For Anadenanthera, the seeding treatment and gap treatments were factors in the
ANOVA model. For Acosmium, gap treatments and regeneration mode were used as factors, as
this species was not used in the seeding treatment but regenerated from both seeds and sprouts.
Seedling height and RGR were analyzed separately for each species using repeated
measures ANOVAs with gap treatments and, for the species that sprouted, regeneration mode as
factors. The effects of location within gap on seedling height and RGR was tested for
Anadenanthera; other species were not sufficiently abundant for this test. Also, the effect of
treatment on proportion of Anadenanthera seedlings browsed was tested with an ANOVA. Blocks
were random effects in each of the above models.
Effects of treatments on seed predation
A seed predation study was conducted to compare rates of seed predation among
treatments and at varying distances from gap centers. Two species were chosen for this study:
Centrolobium and Copaifera. In each treatment, 2 seeds of each species were placed at each of
five stations 1, 3, 5, 7, and 9 m from the gap center. Seeds were inspected after 2 and 9 weeks for
removal or signs of predation. In this study, I assumed that removal indicated predation and
therefore use "predation" as the sum of removed seeds and damaged seeds. The effects of
treatment and distance from gap center on seed predation were tested using Kruskal-Wallis tests on
proportions of seeds remaining and undamaged after 9 weeks.
Effects of the low-intensity burn treatment on seed germination
To assess the effect of low-intensity bums on seed germination of 3 commercial tree
species, 5 seeds each of Copaifera chodatiana, Centrolobium microchaete, and Schinopsis
brasiliensis were placed on leaf litter in low-bum plots just prior to controlled bums. After bums,
remaining seeds were retrieved and transported to the nearby community of San Lorenzo for
germination trials. Seeds from burned plots were placed in plastic trays with sand and watered
daily for 2 months. To detect if germination was related to degree of bum damage, seeds were
inspected before germination trials and assigned a damage score. Seeds of Centrolobium and
Schinopsis, which are protected by a dry husk, had 6 damage categories relating to the degree of
damage to the fruit: 0 (no damage) to 5 (fruit completely burned). Copaifera, which is dispersed
with only a fleshy aril, had 3 damage categories relating to the degree of visible damage to the seed.
The percentage of seeds germinating within each damage category were compared using a Chi-
squared test for independence.
Seed viability of species used in seed addition treatment
Seeds of Copaifera demonstrated the highest viability in greenhouse germination trials,
followed by Anadenanthera, Astronium, Schinopsis, and Centrolobium (Figure 4-1). Seeds of
Anadenanthera and Astronium germinated most rapidly; of their viable seed, 80% and 83%,
respectively, germinated within 4 days.
Effects of treatments on seed predation
Seed predation was uncommon in gaps and there were no differences among treatments
(Copaifera P = 1.00; Centrolobium P = 0.30) or distance from gap edge (Copaifera P = 0.91;
Centrolobium P = 0.73). Overall, 17% of Copaifera and 5% of Centrolobium seeds were
removed or had evidence of predation over the 9 week observation period.
80 ---80------------------- 80
""^ ^^^^_^^ ~,t-~~ -'"
S60 60 -- Centrol
.,vv A v vi i .............A Acosmiu
0 10 20 30 40 50 60
Days after planting
Figure 4-1. Greenhouse seed germination trial of the 5 commercial tree species used in the seeding treatment. Results
represent the cumulative percentages of 100 seeds germinated each day over a 60 day germination trial.
Effects of low-intensity burns on seed germination
Low-intensity bums either decreased or did not affect seed germination of Copaifera,
Centrolobium, and Schinopsis. Copaifera seeds placed in low-intensity bums had an overall
germination rate of 35%, less than half the germination rate of seeds used in the seed viability
germination trial (88%). Percent germination of Copaifera seeds was dependent on the degree of
bum damage (X2 = 9.5, P < 0.05). Germination rates of Schinopsis seeds placed in low-intensity
burns were also lower than unburned seeds (4% compared with 26%), however numbers were too
low for analysis. Centrolobium seeds placed in low-intensity bum plots had germination rates of
4%, only slightly lower than the germination of unburned seeds (5%).
Treatment effects on seedling densities of commercial tree species
Effects of seed addition treatment. Seed addition significantly increased seedling
densities of Centrolobium and Copaifera but not those of Anadenanthera or Astronium (Tables 4-
2, 4-3). Only 9 individuals of Schinopsis were recorded in all subplots; of these individuals, 6
were in seeded plots.
Effects of gap treatments. Patterns of commercial tree density among treatments and
over the 18 month sampling period varied among species. For simplicity, I have displayed
commercial species that had similar density patterns over the 18 months together as groups in
Figure 4-2. Significant differences among treatments are reported separately for each species in
Tables 4-3 and 4-4 and treated in more detail in following sections.
Density of the first group (Anadenanthera and Astronium), all true seedlings, peaked
within 3 months after burns and declined thereafter. Density of the second group (Copaijera.
Aspidosperma, and Caesalpinia) differed according to regeneration mode. While seedling density
of this group gradually increased throughout the 18 month observation period, sprout density
remained fairly constant after the first 3 months. The third group consists of one species
(Centrolobium) that regenerated predominately as root sprouts. Seedlings of Centrolobium, which
Table 4-2. Summary of seed fall and seedling densities of 7 commerical tree species and a non-commerical tree species (Acosmium cardenasii).
Seed fall refers to seedfall before or after the prescribed bum treatments (August 30-September 2, 1997). Significance values for the seeding
treatment are only given for seeded species. Due to the delayed germination of several species, the effects of the seedling treatment was not
detected until 12 months following treatments. For these species, densities and significance values are reported for 12 months. For species
not used in the seeding treatment, total density (seedlings and sprouts combined) at the end of the study (18 months) is reported with the
exception of the 1st year cohort of Caesalpinia, and Acosmium, whose densities are reported for 12 months. Pre-treatment density is based
on a pre-logging inventory conducted in 1995 in the same logging coupe before logging activities (Killeen et al. 1998).
Seed fall Seed addition Seedling density ______
density of unseeded seeded Months Sig. of Pre-treatment
before or after added seeds density density following seeding density
Species burns (mi2) (m"2) (m-2) treatment treatment (m-2)
Caesalpinnia (1s1 year cohort'
Caesalpinnia (2nd year cohort
*** P< 0.001, ** 0.001 < P < 0.01, 0.01 < P < 0.05, NS = not significant
'Numbers too low for statistical analysis, justification described in text.
b Likely seedlings germinated in 1995, the year of the census, as no seedlings of these species > 1-yr old were found in forest plots in this study.
Table 4-3. Statistical analyses testing for the effect of the seed addition treatment.
Seedling densities ofAstronium. Centrolobium, and Copaifera could not be
normalized, therefore the effects of the seeding treatment was tested with a Kruskall
Wallis non-parametric test for each month. Square root transformed seedling densities
of Anadenanthera were normally distributed, therefore a repeated measures ANOVA
was used to test the effect of the seeding treatment and gap treatments for this species.
Because a significant time*treatment interaction was found for Anadenanthera an
ANOVA was used to analyze each month separately.
Astronium Copaifera Centrolobium
Month X2 P X2 P X2 P
3 0.9 0.34 1.1 0.29 1.0 0.32
6 1.3 0.25 13.4 0.00 6.2 0.01
9 0.7 0.40 20.0 0.00 15.2 0.00
12 1.0 0.31 26.0 0.00 15.2 0.00
18 0.7 0.40 23.7 0.00 3.0 0.08
seeding treatment gap treatments
Month F P F P
3 3.4 0.07 6.1 0.000
6 3.5 0.06 2.9 0.03
9 3.3 0.07 2.8 0.03
12 2.7 0.10 3.8 0.01
18 1.8 0.17 5.1 0.001
Group 1. seedlings 0 high intensity burn
--0- low intensity burn
- plant removal
--v- gap control
." ".... f forest
"//7 "" .-..i ........... z. . --.
ff '-- ----- -- --- :- ..s ...........i
Group 2. seedlings
.. . . . . . . . . . . . .;
Group 2. sprouts
4-. i .......... ............ i .. .. ..........................
1.5mo 3 mo 6mo 9mo 12 mo 18 mo
Figure 4-2. Densities of commercial species over the 18 month sampling period following
experimental burns. Species are grouped according to similar regeneration strategies.
Group 1 (Anadenanthera and Astronium) were found predominately as seedlings, Group 2
(Aspidosperma, Copaifera, and Caesalpinia) were found as both seedlings and sprouts,
and Group 3 (Centrolobium) was found predominately as root sprouts (bars = S.E.; n = 16).
were rare, established slowly over the first year and many died by 18 months. In contrast, density
of Centrolobium root sprouts remained relatively constant throughout the sampling period.
The prolonged seedling establishment and sprouting of species in these last 2 groups
restricted the calculations of RGR to the later part of the 18 month sampling period. For these
species, I calculated RGRs from no earlier than 6 or 9 months due to the low seedling and sprout
densities at 3 months. This method of calculating RGR may have limited detection of treatment
differences. If the most rapid growth occurs during the first several weeks following germination
or sprouting, then differences in RGRs measured after this initial growth spurt may not be
detectable. For this reason, seemingly contradictory results were obtained for some species in the
following results sections (i.e., Aspidosperma and Copaifera), where seedling heights were
significantly different among treatments whereas RGRs were not.
Treatment effects on seedling densities, heights, RGRs, and survival of seeded species
Anadenanthera. Seedlings of Anadenanthera were the most abundant of all commercial
tree species with an average density throughout the treatments of 1.1 0.4 seedlings/m2 (mean 1
S.E.) 18 months following treatments. All Anadenanthera seedlings in these analyses are from the
1997 cohort. Seed production of Anadenanthera in 1997 was larger than most years according to
locals. I did not encounter any seedlings > 1 yr-old in subplots, and due to extremely low seed
production in 1998, 1 also did not encounter Anadenanthera seedlings germinating in 1998.
Although Anadenanthera has the ability to coppice (pers. obs.), I did not encounter any sprouts in
the 4 m2 subplots.
Three months after bums, Anadenanthera seedling density was highest in forest plots (F =
6.1, P < 0.001; Figure 4-3). Density declined in forest plots due to high mortality and by 12
months, seedling densities were highest in the plant removal treatment (F = 3.8, P = 0.02).
Anadenanthera seedling density was lowest in gap controls throughout the study.
f 2 -
12- -j-- -
3 mo 6-m- 9 MS 12 in 18- Mn
0 1. ..____ "* ** *. . .... ..."^
3 mo 6mo 9 mo 12 mo 18 mo
100 -4-- high intensity
--80 0- low intensity
u- plant removal
60 ---v- gap control-
S. ... ... .... .. .
6 mo 9 mo 12 mo 18 mo
0.000 ":- "
0-6 mo 6-12 mo 12-18 mo
Figure 4-3. A. Seedling density, B. percent seedling survival, C. seedling height, and
D. seedling relative height growth rates for Anadenanthera colubrina in the four gap
treatments and forest plot. All graphs follow the legend shown in graph C (bars= SE..
Height of Anadenanthera seedlings averaged 49 5 cm after 18 months. The tallest
seedling measured was 3 m and was found in a high-intensity bum plot. Seedlings in high and low-
intensity bum treatments were significantly taller than seedlings in the gap control or forest
understory; seedling height in the plant removal treatment was intermediate (F = 15.4, P < 0.001).
Correspondingly, differences in RGRs among treatments were significant and patterns followed
those for height (F= 18.4, P < 0.001). Anadenanthera seedlings were taller and had higher RGRs
in gap centers than near gap edges (F = 21.8, P = < 0.001; F = 16.0, P = < 0.001). After 18
months, seedlings in gap centers averaged 25 cm taller than seedlings near gap edges.
A mean of 21% (at 9 months) and 12% (at 12 months) of Anadenanthera seedlings were
browsed (tracks suggest by brocket deer, Mazama sp.). Seedlings in high-intensity bum treatments
suffered the highest rates of browsing while seedlings in the forest experienced the lowest rates (F
= 11.16, P< 0.001).
Astronium. Pattern of Astronium seedling density was strongly related to disturbance
intensity (Table 4-4). Throughout the 18 month sampling period, the highest seedling densities
were found in the high-intensity bum treatment plots followed by the low-intensity bum and plant
removal treatments (Figure 4-4). Mortality rates in these treatments was moderately high. Only 2
Astronium individuals were found in all 16 gap control plots (but these died during the 18 month
sampling period) and no Astronium seedlings were found in forest plots. I encountered only one
sprout of Astronium in the permanent subplots (a low-intensity bum treatment) and did not include
it min these analyses.
Height of Astronium seedlings averaged 110 14 cm after 18 months, the tallest mean
seedling height among species. Although heights and RGRs of Astronium seedlings were not
significantly different among treatments (F = 4.0, P = 0.11; F = 1.0, P = 0.40, respectively), there
was a distinct trend of taller Astronium seedlings with increasing disturbance intensity. Mean
Density at 18 months
0. Astronium urundueva*
., Centrolobium microchaete*
03 Caesalpinnia pluviosa
0 (pre-'98 cohorts)
Oi ~ w l j^;; ifW "''"'-Aa
.3] Copaifera chod&
Height at 18 months
1 2 3 4 5
1 2 3 4 5
Figure 4-4. Densities and heights of seedlings (black bars) and sprouts (grey bars) of 6 species
18 months following treatments (* = seeded species). Treatment codes along the x-axis are:
1= high intensity bum, 2 = low intensity bum, 3 = plant removal, 4 = gap control, 5 = forest
(bars = S.E.; n = 16).
m i -_ i
B ,i^ -n "f*
Table 4-4. Statistical analyses testing for differences among gap treatments and forest understory plots. Seedling densities of
Astronium, Aspidosperma, Caesalpinia, Centrolobium, and Copaifera could not be normalized, therefore the effect of gap
treatments was tested using a Kruskall Wallis non-parametric test for each month. Seedlings and sprouts were also analyzed
separately for these species (if applicable). Square-root transformed densities of Acosmium were normally distributed, therefore
a repeated measures ANOVA was used to test of the effect of the gap treatments and regeneration mode (seedling or sprout) for
this species. A significant regeneration mode*treatment was found, therefore seedlings and sprouts were then analyzed separately.
Astronium Aspidosperma Caesalpinia Centrolobium Copaifera
seedlings onl> seedlings sprouts seedlings sprouts seedlings root suckers seedlings sprouts
Mo X2 P X2 P X2 P X2 P X2 P X2 P X2 P X2 P X2 P
3 25.1 <0.001 9.9 0.04 7.7 0.10 21.2 0 8.6 0.07 4.0 0.41 11.1 0.03 17.5 0.002 6.2 0.18
6 34.7 <0.00 1 9.9 0.04 6.7 0.16 14.6 0.006 8.6 0.07 16.2 0.003 11.2 0.02 12.8 0.01 6.2 0.18
9 37.6 <0.001 4.0 0.40 12.1 0.02 11.6 0.021 6.8 0.15 11.8 0.02 23.2 <0.001 12.3 0.02 5.2 0.26
12 37.0 <0.001 5.5 0.24 12.0 0.02 11.6 0.021 5.6 0.23 5.6 0.24 18.5 0.001 10.4 0.03 6.1 0.19
18 37.5 <0.001 7.5 0.11 12.1 0.02 4.3 0.37 5.6 0.23 2.0 0.73 21.6 <0.00 11.5 0.02 6.1 0.20
Acosmium Acosmium seedlings sprouts
Source of variation df F P Source of variation df F P F P
Block 15 2.2 0.01 Block 1 1.1 0.40 1.2 0.29
Regeneration mode 1 0.0 0.93 Treatment 15 12.1 <.001 11.3 <.001
Treatment 4 3.6 0.01 Error 4
Mode Treatment 4 4.0 0.01
height of Astronium seedlings in high-intensity bum treatments was 150 cm, more than twice the
height of seedlings in plant removal treatments (65 cm). Also, the tallest Astronium seedling (4 m)
was found in a high-intensity bum plot.
Copaifera. Copaifera regenerated from both seeds and sprouts, although the overall
density of seedlings was more than 10 times higher than sprouts (Table 4-5). Copaifera seedlings
were most abundant in gap control plots; sprout density did not differ among treatments (Table 4-
4, Figure 4-4). Mortality was low throughout the study period, particularly for sprouts (Table 4-6).
Sprouts of Copaifera were taller than seedlings (F = 8.4, P = 0.01), although their RGRs
did not differ (F = 1.8, P = 0.20). Height and RGRs did not differ among treatments (F = 0.3, P =
0.47; F = 0.3, P = 0.84, respectively).
Centrolobium. Centrolobium regenerated both from seeds and root sprouts. Density of
root sprouts was higher than seedling density at 6 months (Table 4-5). Apparently none of the
seedlings arose from naturally dispersed seeds; natural regeneration of this species was composed
entirely of root sprouts. At 3 months, Centrolobium root sprouts were most abundant in the plant
removal treatment; from 6 to 18 months they were most abundant in the low-intensity burn
treatment (Table 4-4, Figure 4-2). At 6 months, Centrolobium seedlings were most abundant in
the high-intensity burn treatment; at 9 months they were most abundant in the burned and plant
removal treatments. Seedling mortality was greater than root sprout mortality (Table 4-6).
Centrolobium sprouts averaged 267 cm tall after 18 months, more than 7 times the mean
height of Centrolobium seedlings (37 cm; F = 39.2, P = 0.003; Figure 4-4). Similarly, RGRs of
Centrolobium sprouts were also higher than those of seedlings (F = 13.6, P = 0.01). No
differences in heights at 18 months or RGRs were detected among treatments (F = 1.0, P = 0.49; F
= 0.1, P = 0.97. respectively).
Schinopsis brasilensis. A total of 9 Schinopsis seedlings were recorded in all subplots,
although the maximum number at any one sampling period was 7 (Table 4-7). After 18 months, 6
Table 4-5. Statistical analyses comparing seedling and sprout densities ofAspidosperma,
Caesalpinia, Centrolobium, and Copaifera. Seedling and sprout densities of these
species could not be normalized, therefore a Kruskall-Wallis non-parametric test was
used to analyze each month.
Copaifera Centrolobium Aspidosperma Caesalpinnia
Month X2 P X2 P X2 P X2 P
3 5.6 0.02 3.7 0.06 7.7 0.006 0.13 0.72
6 22.2 0.00 5.9 0.02 11.2 0.001 0.32 0.57
9 27.6 0.00 2.6 0.11 10.2 0.001 0.04 0.84
12 31.5 0.00 2.2 0.13 15.2 0.000 0.001 0.98
18 35.9 0.00 3.1 0.08 10.2 0.001 15.6 0.00
Table 4-6. Killing power, a parameter similar to mortality rate, for 5 commercial tree species in the four gap treatments and forest
understory plots. Killing power (k) is calculated separately for seedlings and sprouts of each species, except for Astronium, for which
no sprouts were found. Killing power is calculated as (logloax loglo0ak+), where a, represents the number of individuals in the
first cohort following treatments, and a,+I is the number of these individuals surviving into the next year (Begon and Mortimer 1981).
The numbers of individuals that died during this period are listed in columns "#."
high intensity low intensity plant gap control forest All individuals
Species bum bum removal ____________
k # k # k # k # k # k #
Aspidosperma seedlings 0.20 3 0.30 1 0.30 1 0.60 3 0.15 2 0.25 10
resprouts 0.00 0 0.22 9 0.05 1 0.48 4 1 0.17 15
Caesalpinia seedlings 0.00 0 0.00 0 0.07 1 0.22 2 0.00 0 0.09 3
resprouts 0.00 0 0.11 2 0.00 0 0.06 1 ______ 0.05 3
Centrolobium seedlings 0.11 2 0.12 4 0.06 1 1 2 0.14 10
resprouts 0.18 2 0.00 0 0.11 2 0.12 1 _______ 0.07 5
Copaifera seedlings 0.09 3 0.10 2 0.09 6 0.06 4 0.13 5 0.09 20
resprouts 0.00 0 0.00 0 0.08 1 0.00 0 0.00 0 0.02 1
Astronium seedlings 0.11 10 0.13 9 0.18 9 1 0.14 29
* = where a,+I was equal to 0 (all individuals died), k equals infinity
xml version 1.0 standalone yes
ErrorText remove streak mark
remove streak mark